{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9998557067962099, "global_step": 5197, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.282051282051282e-07, "loss": 1.1039, "step": 1 }, { "epoch": 0.0, "learning_rate": 2.564102564102564e-07, "loss": 1.1237, "step": 2 }, { "epoch": 0.0, "learning_rate": 3.846153846153847e-07, "loss": 1.0527, "step": 3 }, { "epoch": 0.0, "learning_rate": 5.128205128205128e-07, "loss": 1.0217, "step": 4 }, { "epoch": 0.0, "learning_rate": 6.41025641025641e-07, "loss": 0.8716, "step": 5 }, { "epoch": 0.0, "learning_rate": 7.692307692307694e-07, "loss": 1.1155, "step": 6 }, { "epoch": 0.0, "learning_rate": 8.974358974358975e-07, "loss": 1.0593, "step": 7 }, { "epoch": 0.0, "learning_rate": 1.0256410256410257e-06, "loss": 0.7892, "step": 8 }, { "epoch": 0.0, "learning_rate": 1.153846153846154e-06, "loss": 1.0414, "step": 9 }, { "epoch": 0.0, "learning_rate": 1.282051282051282e-06, "loss": 0.753, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.4102564102564104e-06, "loss": 1.0238, "step": 11 }, { "epoch": 0.0, "learning_rate": 1.5384615384615387e-06, "loss": 0.9556, "step": 12 }, { "epoch": 0.0, "learning_rate": 1.6666666666666667e-06, "loss": 0.8112, "step": 13 }, { "epoch": 0.0, "learning_rate": 1.794871794871795e-06, "loss": 0.6639, "step": 14 }, { "epoch": 0.0, "learning_rate": 1.9230769230769234e-06, "loss": 0.8358, "step": 15 }, { "epoch": 0.0, "learning_rate": 2.0512820512820513e-06, "loss": 0.8033, "step": 16 }, { "epoch": 0.0, "learning_rate": 2.1794871794871797e-06, "loss": 0.7926, "step": 17 }, { "epoch": 0.0, "learning_rate": 2.307692307692308e-06, "loss": 0.7502, "step": 18 }, { "epoch": 0.0, "learning_rate": 2.435897435897436e-06, "loss": 0.7932, "step": 19 }, { "epoch": 0.0, "learning_rate": 2.564102564102564e-06, "loss": 0.7898, "step": 20 }, { "epoch": 0.0, "learning_rate": 2.6923076923076923e-06, "loss": 0.5822, "step": 21 }, { "epoch": 0.0, "learning_rate": 2.8205128205128207e-06, "loss": 0.7556, "step": 22 }, { "epoch": 0.0, "learning_rate": 2.948717948717949e-06, "loss": 0.5668, "step": 23 }, { "epoch": 0.0, "learning_rate": 3.0769230769230774e-06, "loss": 0.7231, "step": 24 }, { "epoch": 0.0, "learning_rate": 3.205128205128206e-06, "loss": 0.7115, "step": 25 }, { "epoch": 0.01, "learning_rate": 3.3333333333333333e-06, "loss": 0.7071, "step": 26 }, { "epoch": 0.01, "learning_rate": 3.4615384615384617e-06, "loss": 0.7524, "step": 27 }, { "epoch": 0.01, "learning_rate": 3.58974358974359e-06, "loss": 0.6968, "step": 28 }, { "epoch": 0.01, "learning_rate": 3.7179487179487184e-06, "loss": 0.6838, "step": 29 }, { "epoch": 0.01, "learning_rate": 3.846153846153847e-06, "loss": 0.5848, "step": 30 }, { "epoch": 0.01, "learning_rate": 3.974358974358974e-06, "loss": 0.5729, "step": 31 }, { "epoch": 0.01, "learning_rate": 4.102564102564103e-06, "loss": 0.6658, "step": 32 }, { "epoch": 0.01, "learning_rate": 4.230769230769231e-06, "loss": 0.7242, "step": 33 }, { "epoch": 0.01, "learning_rate": 4.358974358974359e-06, "loss": 0.7274, "step": 34 }, { "epoch": 0.01, "learning_rate": 4.487179487179488e-06, "loss": 0.6313, "step": 35 }, { "epoch": 0.01, "learning_rate": 4.615384615384616e-06, "loss": 0.5373, "step": 36 }, { "epoch": 0.01, "learning_rate": 4.743589743589744e-06, "loss": 0.6631, "step": 37 }, { "epoch": 0.01, "learning_rate": 4.871794871794872e-06, "loss": 0.6629, "step": 38 }, { "epoch": 0.01, "learning_rate": 5e-06, "loss": 0.6717, "step": 39 }, { "epoch": 0.01, "learning_rate": 5.128205128205128e-06, "loss": 0.5422, "step": 40 }, { "epoch": 0.01, "learning_rate": 5.256410256410257e-06, "loss": 0.5194, "step": 41 }, { "epoch": 0.01, "learning_rate": 5.384615384615385e-06, "loss": 0.6561, "step": 42 }, { "epoch": 0.01, "learning_rate": 5.512820512820514e-06, "loss": 0.7092, "step": 43 }, { "epoch": 0.01, "learning_rate": 5.641025641025641e-06, "loss": 0.5319, "step": 44 }, { "epoch": 0.01, "learning_rate": 5.769230769230769e-06, "loss": 0.3902, "step": 45 }, { "epoch": 0.01, "learning_rate": 5.897435897435898e-06, "loss": 0.6643, "step": 46 }, { "epoch": 0.01, "learning_rate": 6.025641025641026e-06, "loss": 0.6466, "step": 47 }, { "epoch": 0.01, "learning_rate": 6.153846153846155e-06, "loss": 0.591, "step": 48 }, { "epoch": 0.01, "learning_rate": 6.282051282051282e-06, "loss": 0.604, "step": 49 }, { "epoch": 0.01, "learning_rate": 6.410256410256412e-06, "loss": 0.638, "step": 50 }, { "epoch": 0.01, "learning_rate": 6.538461538461539e-06, "loss": 0.6499, "step": 51 }, { "epoch": 0.01, "learning_rate": 6.666666666666667e-06, "loss": 0.6562, "step": 52 }, { "epoch": 0.01, "learning_rate": 6.794871794871796e-06, "loss": 0.623, "step": 53 }, { "epoch": 0.01, "learning_rate": 6.923076923076923e-06, "loss": 0.6432, "step": 54 }, { "epoch": 0.01, "learning_rate": 7.051282051282053e-06, "loss": 0.6171, "step": 55 }, { "epoch": 0.01, "learning_rate": 7.17948717948718e-06, "loss": 0.6126, "step": 56 }, { "epoch": 0.01, "learning_rate": 7.307692307692308e-06, "loss": 0.5024, "step": 57 }, { "epoch": 0.01, "learning_rate": 7.435897435897437e-06, "loss": 0.6594, "step": 58 }, { "epoch": 0.01, "learning_rate": 7.564102564102564e-06, "loss": 0.6426, "step": 59 }, { "epoch": 0.01, "learning_rate": 7.692307692307694e-06, "loss": 0.6438, "step": 60 }, { "epoch": 0.01, "learning_rate": 7.820512820512822e-06, "loss": 0.622, "step": 61 }, { "epoch": 0.01, "learning_rate": 7.948717948717949e-06, "loss": 0.5002, "step": 62 }, { "epoch": 0.01, "learning_rate": 8.076923076923077e-06, "loss": 0.6268, "step": 63 }, { "epoch": 0.01, "learning_rate": 8.205128205128205e-06, "loss": 0.5889, "step": 64 }, { "epoch": 0.01, "learning_rate": 8.333333333333334e-06, "loss": 0.4988, "step": 65 }, { "epoch": 0.01, "learning_rate": 8.461538461538462e-06, "loss": 0.6332, "step": 66 }, { "epoch": 0.01, "learning_rate": 8.58974358974359e-06, "loss": 0.6125, "step": 67 }, { "epoch": 0.01, "learning_rate": 8.717948717948719e-06, "loss": 0.5097, "step": 68 }, { "epoch": 0.01, "learning_rate": 8.846153846153847e-06, "loss": 0.633, "step": 69 }, { "epoch": 0.01, "learning_rate": 8.974358974358976e-06, "loss": 0.6446, "step": 70 }, { "epoch": 0.01, "learning_rate": 9.102564102564104e-06, "loss": 0.6222, "step": 71 }, { "epoch": 0.01, "learning_rate": 9.230769230769232e-06, "loss": 0.6086, "step": 72 }, { "epoch": 0.01, "learning_rate": 9.358974358974359e-06, "loss": 0.5191, "step": 73 }, { "epoch": 0.01, "learning_rate": 9.487179487179487e-06, "loss": 0.5129, "step": 74 }, { "epoch": 0.01, "learning_rate": 9.615384615384616e-06, "loss": 0.5915, "step": 75 }, { "epoch": 0.01, "learning_rate": 9.743589743589744e-06, "loss": 0.6364, "step": 76 }, { "epoch": 0.01, "learning_rate": 9.871794871794872e-06, "loss": 0.5893, "step": 77 }, { "epoch": 0.02, "learning_rate": 1e-05, "loss": 0.6201, "step": 78 }, { "epoch": 0.02, "learning_rate": 1.012820512820513e-05, "loss": 0.5947, "step": 79 }, { "epoch": 0.02, "learning_rate": 1.0256410256410256e-05, "loss": 0.6284, "step": 80 }, { "epoch": 0.02, "learning_rate": 1.0384615384615386e-05, "loss": 0.6011, "step": 81 }, { "epoch": 0.02, "learning_rate": 1.0512820512820514e-05, "loss": 0.5936, "step": 82 }, { "epoch": 0.02, "learning_rate": 1.0641025641025643e-05, "loss": 0.5968, "step": 83 }, { "epoch": 0.02, "learning_rate": 1.076923076923077e-05, "loss": 0.6223, "step": 84 }, { "epoch": 0.02, "learning_rate": 1.0897435897435898e-05, "loss": 0.5827, "step": 85 }, { "epoch": 0.02, "learning_rate": 1.1025641025641028e-05, "loss": 0.6114, "step": 86 }, { "epoch": 0.02, "learning_rate": 1.1153846153846154e-05, "loss": 0.6446, "step": 87 }, { "epoch": 0.02, "learning_rate": 1.1282051282051283e-05, "loss": 0.4986, "step": 88 }, { "epoch": 0.02, "learning_rate": 1.1410256410256411e-05, "loss": 0.5722, "step": 89 }, { "epoch": 0.02, "learning_rate": 1.1538461538461538e-05, "loss": 0.5127, "step": 90 }, { "epoch": 0.02, "learning_rate": 1.1666666666666668e-05, "loss": 0.5057, "step": 91 }, { "epoch": 0.02, "learning_rate": 1.1794871794871796e-05, "loss": 0.6006, "step": 92 }, { "epoch": 0.02, "learning_rate": 1.1923076923076925e-05, "loss": 0.6228, "step": 93 }, { "epoch": 0.02, "learning_rate": 1.2051282051282051e-05, "loss": 0.576, "step": 94 }, { "epoch": 0.02, "learning_rate": 1.217948717948718e-05, "loss": 0.6169, "step": 95 }, { "epoch": 0.02, "learning_rate": 1.230769230769231e-05, "loss": 0.5918, "step": 96 }, { "epoch": 0.02, "learning_rate": 1.2435897435897436e-05, "loss": 0.5952, "step": 97 }, { "epoch": 0.02, "learning_rate": 1.2564102564102565e-05, "loss": 0.4686, "step": 98 }, { "epoch": 0.02, "learning_rate": 1.2692307692307693e-05, "loss": 0.6285, "step": 99 }, { "epoch": 0.02, "learning_rate": 1.2820512820512823e-05, "loss": 0.6011, "step": 100 }, { "epoch": 0.02, "learning_rate": 1.294871794871795e-05, "loss": 0.6035, "step": 101 }, { "epoch": 0.02, "learning_rate": 1.3076923076923078e-05, "loss": 0.6514, "step": 102 }, { "epoch": 0.02, "learning_rate": 1.3205128205128207e-05, "loss": 0.6507, "step": 103 }, { "epoch": 0.02, "learning_rate": 1.3333333333333333e-05, "loss": 0.6144, "step": 104 }, { "epoch": 0.02, "learning_rate": 1.3461538461538463e-05, "loss": 0.5667, "step": 105 }, { "epoch": 0.02, "learning_rate": 1.3589743589743592e-05, "loss": 0.6026, "step": 106 }, { "epoch": 0.02, "learning_rate": 1.3717948717948718e-05, "loss": 0.6429, "step": 107 }, { "epoch": 0.02, "learning_rate": 1.3846153846153847e-05, "loss": 0.5884, "step": 108 }, { "epoch": 0.02, "learning_rate": 1.3974358974358975e-05, "loss": 0.6302, "step": 109 }, { "epoch": 0.02, "learning_rate": 1.4102564102564105e-05, "loss": 0.6092, "step": 110 }, { "epoch": 0.02, "learning_rate": 1.4230769230769232e-05, "loss": 0.6191, "step": 111 }, { "epoch": 0.02, "learning_rate": 1.435897435897436e-05, "loss": 0.6067, "step": 112 }, { "epoch": 0.02, "learning_rate": 1.4487179487179489e-05, "loss": 0.6221, "step": 113 }, { "epoch": 0.02, "learning_rate": 1.4615384615384615e-05, "loss": 0.5081, "step": 114 }, { "epoch": 0.02, "learning_rate": 1.4743589743589745e-05, "loss": 0.6363, "step": 115 }, { "epoch": 0.02, "learning_rate": 1.4871794871794874e-05, "loss": 0.6111, "step": 116 }, { "epoch": 0.02, "learning_rate": 1.5000000000000002e-05, "loss": 0.6315, "step": 117 }, { "epoch": 0.02, "learning_rate": 1.5128205128205129e-05, "loss": 0.6325, "step": 118 }, { "epoch": 0.02, "learning_rate": 1.5256410256410257e-05, "loss": 0.5571, "step": 119 }, { "epoch": 0.02, "learning_rate": 1.5384615384615387e-05, "loss": 0.6002, "step": 120 }, { "epoch": 0.02, "learning_rate": 1.5512820512820516e-05, "loss": 0.6023, "step": 121 }, { "epoch": 0.02, "learning_rate": 1.5641025641025644e-05, "loss": 0.6247, "step": 122 }, { "epoch": 0.02, "learning_rate": 1.576923076923077e-05, "loss": 0.6115, "step": 123 }, { "epoch": 0.02, "learning_rate": 1.5897435897435897e-05, "loss": 0.5965, "step": 124 }, { "epoch": 0.02, "learning_rate": 1.602564102564103e-05, "loss": 0.6062, "step": 125 }, { "epoch": 0.02, "learning_rate": 1.6153846153846154e-05, "loss": 0.541, "step": 126 }, { "epoch": 0.02, "learning_rate": 1.6282051282051282e-05, "loss": 0.6142, "step": 127 }, { "epoch": 0.02, "learning_rate": 1.641025641025641e-05, "loss": 0.6134, "step": 128 }, { "epoch": 0.02, "learning_rate": 1.653846153846154e-05, "loss": 0.6083, "step": 129 }, { "epoch": 0.03, "learning_rate": 1.6666666666666667e-05, "loss": 0.5786, "step": 130 }, { "epoch": 0.03, "learning_rate": 1.6794871794871796e-05, "loss": 0.6216, "step": 131 }, { "epoch": 0.03, "learning_rate": 1.6923076923076924e-05, "loss": 0.4867, "step": 132 }, { "epoch": 0.03, "learning_rate": 1.7051282051282053e-05, "loss": 0.5718, "step": 133 }, { "epoch": 0.03, "learning_rate": 1.717948717948718e-05, "loss": 0.5937, "step": 134 }, { "epoch": 0.03, "learning_rate": 1.730769230769231e-05, "loss": 0.4925, "step": 135 }, { "epoch": 0.03, "learning_rate": 1.7435897435897438e-05, "loss": 0.6035, "step": 136 }, { "epoch": 0.03, "learning_rate": 1.7564102564102566e-05, "loss": 0.6061, "step": 137 }, { "epoch": 0.03, "learning_rate": 1.7692307692307694e-05, "loss": 0.6432, "step": 138 }, { "epoch": 0.03, "learning_rate": 1.7820512820512823e-05, "loss": 0.6016, "step": 139 }, { "epoch": 0.03, "learning_rate": 1.794871794871795e-05, "loss": 0.615, "step": 140 }, { "epoch": 0.03, "learning_rate": 1.807692307692308e-05, "loss": 0.4851, "step": 141 }, { "epoch": 0.03, "learning_rate": 1.8205128205128208e-05, "loss": 0.65, "step": 142 }, { "epoch": 0.03, "learning_rate": 1.8333333333333333e-05, "loss": 0.612, "step": 143 }, { "epoch": 0.03, "learning_rate": 1.8461538461538465e-05, "loss": 0.5784, "step": 144 }, { "epoch": 0.03, "learning_rate": 1.8589743589743593e-05, "loss": 0.6089, "step": 145 }, { "epoch": 0.03, "learning_rate": 1.8717948717948718e-05, "loss": 0.6302, "step": 146 }, { "epoch": 0.03, "learning_rate": 1.8846153846153846e-05, "loss": 0.5974, "step": 147 }, { "epoch": 0.03, "learning_rate": 1.8974358974358975e-05, "loss": 0.4927, "step": 148 }, { "epoch": 0.03, "learning_rate": 1.9102564102564106e-05, "loss": 0.5972, "step": 149 }, { "epoch": 0.03, "learning_rate": 1.923076923076923e-05, "loss": 0.6264, "step": 150 }, { "epoch": 0.03, "learning_rate": 1.935897435897436e-05, "loss": 0.6079, "step": 151 }, { "epoch": 0.03, "learning_rate": 1.9487179487179488e-05, "loss": 0.4705, "step": 152 }, { "epoch": 0.03, "learning_rate": 1.9615384615384617e-05, "loss": 0.5561, "step": 153 }, { "epoch": 0.03, "learning_rate": 1.9743589743589745e-05, "loss": 0.6436, "step": 154 }, { "epoch": 0.03, "learning_rate": 1.9871794871794873e-05, "loss": 0.5758, "step": 155 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 0.6044, "step": 156 }, { "epoch": 0.03, "learning_rate": 1.9999998058057616e-05, "loss": 0.6054, "step": 157 }, { "epoch": 0.03, "learning_rate": 1.9999992232231216e-05, "loss": 0.4884, "step": 158 }, { "epoch": 0.03, "learning_rate": 1.999998252252306e-05, "loss": 0.6004, "step": 159 }, { "epoch": 0.03, "learning_rate": 1.9999968928936924e-05, "loss": 0.633, "step": 160 }, { "epoch": 0.03, "learning_rate": 1.999995145147809e-05, "loss": 0.6269, "step": 161 }, { "epoch": 0.03, "learning_rate": 1.9999930090153335e-05, "loss": 0.6158, "step": 162 }, { "epoch": 0.03, "learning_rate": 1.9999904844970963e-05, "loss": 0.4688, "step": 163 }, { "epoch": 0.03, "learning_rate": 1.999987571594078e-05, "loss": 0.612, "step": 164 }, { "epoch": 0.03, "learning_rate": 1.99998427030741e-05, "loss": 0.5913, "step": 165 }, { "epoch": 0.03, "learning_rate": 1.999980580638374e-05, "loss": 0.6232, "step": 166 }, { "epoch": 0.03, "learning_rate": 1.999976502588403e-05, "loss": 0.6068, "step": 167 }, { "epoch": 0.03, "learning_rate": 1.9999720361590812e-05, "loss": 0.6197, "step": 168 }, { "epoch": 0.03, "learning_rate": 1.9999671813521435e-05, "loss": 0.5996, "step": 169 }, { "epoch": 0.03, "learning_rate": 1.999961938169475e-05, "loss": 0.6298, "step": 170 }, { "epoch": 0.03, "learning_rate": 1.9999563066131124e-05, "loss": 0.5608, "step": 171 }, { "epoch": 0.03, "learning_rate": 1.9999502866852427e-05, "loss": 0.6151, "step": 172 }, { "epoch": 0.03, "learning_rate": 1.999943878388204e-05, "loss": 0.6256, "step": 173 }, { "epoch": 0.03, "learning_rate": 1.9999370817244853e-05, "loss": 0.4757, "step": 174 }, { "epoch": 0.03, "learning_rate": 1.9999298966967264e-05, "loss": 0.6342, "step": 175 }, { "epoch": 0.03, "learning_rate": 1.9999223233077178e-05, "loss": 0.6021, "step": 176 }, { "epoch": 0.03, "learning_rate": 1.999914361560401e-05, "loss": 0.5816, "step": 177 }, { "epoch": 0.03, "learning_rate": 1.9999060114578682e-05, "loss": 0.6236, "step": 178 }, { "epoch": 0.03, "learning_rate": 1.9998972730033624e-05, "loss": 0.4725, "step": 179 }, { "epoch": 0.03, "learning_rate": 1.9998881462002778e-05, "loss": 0.6249, "step": 180 }, { "epoch": 0.03, "learning_rate": 1.9998786310521585e-05, "loss": 0.6119, "step": 181 }, { "epoch": 0.04, "learning_rate": 1.9998687275627008e-05, "loss": 0.4973, "step": 182 }, { "epoch": 0.04, "learning_rate": 1.9998584357357503e-05, "loss": 0.6198, "step": 183 }, { "epoch": 0.04, "learning_rate": 1.9998477555753054e-05, "loss": 0.6437, "step": 184 }, { "epoch": 0.04, "learning_rate": 1.9998366870855134e-05, "loss": 0.5718, "step": 185 }, { "epoch": 0.04, "learning_rate": 1.999825230270673e-05, "loss": 0.6031, "step": 186 }, { "epoch": 0.04, "learning_rate": 1.9998133851352342e-05, "loss": 0.6024, "step": 187 }, { "epoch": 0.04, "learning_rate": 1.9998011516837974e-05, "loss": 0.5739, "step": 188 }, { "epoch": 0.04, "learning_rate": 1.999788529921114e-05, "loss": 0.5818, "step": 189 }, { "epoch": 0.04, "learning_rate": 1.999775519852086e-05, "loss": 0.5311, "step": 190 }, { "epoch": 0.04, "learning_rate": 1.999762121481767e-05, "loss": 0.4965, "step": 191 }, { "epoch": 0.04, "learning_rate": 1.99974833481536e-05, "loss": 0.6536, "step": 192 }, { "epoch": 0.04, "learning_rate": 1.9997341598582197e-05, "loss": 0.5123, "step": 193 }, { "epoch": 0.04, "learning_rate": 1.9997195966158518e-05, "loss": 0.6307, "step": 194 }, { "epoch": 0.04, "learning_rate": 1.9997046450939122e-05, "loss": 0.614, "step": 195 }, { "epoch": 0.04, "learning_rate": 1.9996893052982083e-05, "loss": 0.5974, "step": 196 }, { "epoch": 0.04, "learning_rate": 1.9996735772346973e-05, "loss": 0.608, "step": 197 }, { "epoch": 0.04, "learning_rate": 1.9996574609094887e-05, "loss": 0.5804, "step": 198 }, { "epoch": 0.04, "learning_rate": 1.9996409563288404e-05, "loss": 0.6363, "step": 199 }, { "epoch": 0.04, "learning_rate": 1.9996240634991645e-05, "loss": 0.6034, "step": 200 }, { "epoch": 0.04, "learning_rate": 1.9996067824270204e-05, "loss": 0.4865, "step": 201 }, { "epoch": 0.04, "learning_rate": 1.999589113119121e-05, "loss": 0.4811, "step": 202 }, { "epoch": 0.04, "learning_rate": 1.9995710555823277e-05, "loss": 0.6286, "step": 203 }, { "epoch": 0.04, "learning_rate": 1.999552609823655e-05, "loss": 0.6005, "step": 204 }, { "epoch": 0.04, "learning_rate": 1.999533775850266e-05, "loss": 0.5232, "step": 205 }, { "epoch": 0.04, "learning_rate": 1.9995145536694764e-05, "loss": 0.6021, "step": 206 }, { "epoch": 0.04, "learning_rate": 1.9994949432887512e-05, "loss": 0.6234, "step": 207 }, { "epoch": 0.04, "learning_rate": 1.999474944715708e-05, "loss": 0.6109, "step": 208 }, { "epoch": 0.04, "learning_rate": 1.9994545579581125e-05, "loss": 0.5965, "step": 209 }, { "epoch": 0.04, "learning_rate": 1.9994337830238836e-05, "loss": 0.6173, "step": 210 }, { "epoch": 0.04, "learning_rate": 1.9994126199210897e-05, "loss": 0.6324, "step": 211 }, { "epoch": 0.04, "learning_rate": 1.999391068657951e-05, "loss": 0.6279, "step": 212 }, { "epoch": 0.04, "learning_rate": 1.9993691292428364e-05, "loss": 0.6119, "step": 213 }, { "epoch": 0.04, "learning_rate": 1.9993468016842684e-05, "loss": 0.6218, "step": 214 }, { "epoch": 0.04, "learning_rate": 1.999324085990918e-05, "loss": 0.6117, "step": 215 }, { "epoch": 0.04, "learning_rate": 1.9993009821716076e-05, "loss": 0.6275, "step": 216 }, { "epoch": 0.04, "learning_rate": 1.9992774902353104e-05, "loss": 0.6227, "step": 217 }, { "epoch": 0.04, "learning_rate": 1.999253610191151e-05, "loss": 0.5995, "step": 218 }, { "epoch": 0.04, "learning_rate": 1.999229342048404e-05, "loss": 0.6074, "step": 219 }, { "epoch": 0.04, "learning_rate": 1.9992046858164942e-05, "loss": 0.4834, "step": 220 }, { "epoch": 0.04, "learning_rate": 1.999179641504999e-05, "loss": 0.6343, "step": 221 }, { "epoch": 0.04, "learning_rate": 1.9991542091236438e-05, "loss": 0.6212, "step": 222 }, { "epoch": 0.04, "learning_rate": 1.9991283886823075e-05, "loss": 0.5663, "step": 223 }, { "epoch": 0.04, "learning_rate": 1.9991021801910177e-05, "loss": 0.4747, "step": 224 }, { "epoch": 0.04, "learning_rate": 1.999075583659954e-05, "loss": 0.5667, "step": 225 }, { "epoch": 0.04, "learning_rate": 1.999048599099446e-05, "loss": 0.5899, "step": 226 }, { "epoch": 0.04, "learning_rate": 1.9990212265199738e-05, "loss": 0.6272, "step": 227 }, { "epoch": 0.04, "learning_rate": 1.998993465932169e-05, "loss": 0.6068, "step": 228 }, { "epoch": 0.04, "learning_rate": 1.9989653173468137e-05, "loss": 0.5798, "step": 229 }, { "epoch": 0.04, "learning_rate": 1.99893678077484e-05, "loss": 0.6158, "step": 230 }, { "epoch": 0.04, "learning_rate": 1.9989078562273313e-05, "loss": 0.6221, "step": 231 }, { "epoch": 0.04, "learning_rate": 1.9988785437155222e-05, "loss": 0.4644, "step": 232 }, { "epoch": 0.04, "learning_rate": 1.9988488432507963e-05, "loss": 0.6098, "step": 233 }, { "epoch": 0.05, "learning_rate": 1.9988187548446895e-05, "loss": 0.6078, "step": 234 }, { "epoch": 0.05, "learning_rate": 1.998788278508888e-05, "loss": 0.484, "step": 235 }, { "epoch": 0.05, "learning_rate": 1.9987574142552274e-05, "loss": 0.6002, "step": 236 }, { "epoch": 0.05, "learning_rate": 1.9987261620956964e-05, "loss": 0.6406, "step": 237 }, { "epoch": 0.05, "learning_rate": 1.9986945220424326e-05, "loss": 0.4974, "step": 238 }, { "epoch": 0.05, "learning_rate": 1.998662494107724e-05, "loss": 0.4546, "step": 239 }, { "epoch": 0.05, "learning_rate": 1.99863007830401e-05, "loss": 0.5792, "step": 240 }, { "epoch": 0.05, "learning_rate": 1.9985972746438815e-05, "loss": 0.6074, "step": 241 }, { "epoch": 0.05, "learning_rate": 1.9985640831400778e-05, "loss": 0.6183, "step": 242 }, { "epoch": 0.05, "learning_rate": 1.998530503805491e-05, "loss": 0.6341, "step": 243 }, { "epoch": 0.05, "learning_rate": 1.9984965366531624e-05, "loss": 0.631, "step": 244 }, { "epoch": 0.05, "learning_rate": 1.9984621816962843e-05, "loss": 0.6225, "step": 245 }, { "epoch": 0.05, "learning_rate": 1.9984274389482005e-05, "loss": 0.6089, "step": 246 }, { "epoch": 0.05, "learning_rate": 1.9983923084224047e-05, "loss": 0.4597, "step": 247 }, { "epoch": 0.05, "learning_rate": 1.9983567901325404e-05, "loss": 0.6577, "step": 248 }, { "epoch": 0.05, "learning_rate": 1.9983208840924028e-05, "loss": 0.6513, "step": 249 }, { "epoch": 0.05, "learning_rate": 1.998284590315937e-05, "loss": 0.6115, "step": 250 }, { "epoch": 0.05, "learning_rate": 1.9982479088172403e-05, "loss": 0.6194, "step": 251 }, { "epoch": 0.05, "learning_rate": 1.9982108396105584e-05, "loss": 0.6128, "step": 252 }, { "epoch": 0.05, "learning_rate": 1.9981733827102884e-05, "loss": 0.6001, "step": 253 }, { "epoch": 0.05, "learning_rate": 1.998135538130979e-05, "loss": 0.5933, "step": 254 }, { "epoch": 0.05, "learning_rate": 1.998097305887328e-05, "loss": 0.5337, "step": 255 }, { "epoch": 0.05, "learning_rate": 1.9980586859941846e-05, "loss": 0.6333, "step": 256 }, { "epoch": 0.05, "learning_rate": 1.998019678466548e-05, "loss": 0.4688, "step": 257 }, { "epoch": 0.05, "learning_rate": 1.997980283319568e-05, "loss": 0.6491, "step": 258 }, { "epoch": 0.05, "learning_rate": 1.9979405005685466e-05, "loss": 0.6028, "step": 259 }, { "epoch": 0.05, "learning_rate": 1.9979003302289336e-05, "loss": 0.6161, "step": 260 }, { "epoch": 0.05, "learning_rate": 1.997859772316331e-05, "loss": 0.5836, "step": 261 }, { "epoch": 0.05, "learning_rate": 1.9978188268464912e-05, "loss": 0.6171, "step": 262 }, { "epoch": 0.05, "learning_rate": 1.997777493835317e-05, "loss": 0.5919, "step": 263 }, { "epoch": 0.05, "learning_rate": 1.9977357732988616e-05, "loss": 0.5708, "step": 264 }, { "epoch": 0.05, "learning_rate": 1.9976936652533288e-05, "loss": 0.6049, "step": 265 }, { "epoch": 0.05, "learning_rate": 1.997651169715073e-05, "loss": 0.4702, "step": 266 }, { "epoch": 0.05, "learning_rate": 1.9976082867005985e-05, "loss": 0.6062, "step": 267 }, { "epoch": 0.05, "learning_rate": 1.997565016226561e-05, "loss": 0.5052, "step": 268 }, { "epoch": 0.05, "learning_rate": 1.997521358309766e-05, "loss": 0.6174, "step": 269 }, { "epoch": 0.05, "learning_rate": 1.99747731296717e-05, "loss": 0.3902, "step": 270 }, { "epoch": 0.05, "learning_rate": 1.9974328802158798e-05, "loss": 0.6053, "step": 271 }, { "epoch": 0.05, "learning_rate": 1.997388060073152e-05, "loss": 0.599, "step": 272 }, { "epoch": 0.05, "learning_rate": 1.9973428525563948e-05, "loss": 0.6074, "step": 273 }, { "epoch": 0.05, "learning_rate": 1.9972972576831656e-05, "loss": 0.4901, "step": 274 }, { "epoch": 0.05, "learning_rate": 1.9972512754711738e-05, "loss": 0.5916, "step": 275 }, { "epoch": 0.05, "learning_rate": 1.997204905938278e-05, "loss": 0.6232, "step": 276 }, { "epoch": 0.05, "learning_rate": 1.9971581491024873e-05, "loss": 0.551, "step": 277 }, { "epoch": 0.05, "learning_rate": 1.997111004981962e-05, "loss": 0.4665, "step": 278 }, { "epoch": 0.05, "learning_rate": 1.9970634735950117e-05, "loss": 0.6021, "step": 279 }, { "epoch": 0.05, "learning_rate": 1.9970155549600978e-05, "loss": 0.6071, "step": 280 }, { "epoch": 0.05, "learning_rate": 1.9969672490958304e-05, "loss": 0.4864, "step": 281 }, { "epoch": 0.05, "learning_rate": 1.996918556020972e-05, "loss": 0.5991, "step": 282 }, { "epoch": 0.05, "learning_rate": 1.996869475754434e-05, "loss": 0.4816, "step": 283 }, { "epoch": 0.05, "learning_rate": 1.9968200083152784e-05, "loss": 0.6337, "step": 284 }, { "epoch": 0.05, "learning_rate": 1.9967701537227175e-05, "loss": 0.5832, "step": 285 }, { "epoch": 0.06, "learning_rate": 1.996719911996115e-05, "loss": 0.5863, "step": 286 }, { "epoch": 0.06, "learning_rate": 1.996669283154984e-05, "loss": 0.6151, "step": 287 }, { "epoch": 0.06, "learning_rate": 1.996618267218988e-05, "loss": 0.6574, "step": 288 }, { "epoch": 0.06, "learning_rate": 1.996566864207941e-05, "loss": 0.5815, "step": 289 }, { "epoch": 0.06, "learning_rate": 1.9965150741418072e-05, "loss": 0.614, "step": 290 }, { "epoch": 0.06, "learning_rate": 1.9964628970407018e-05, "loss": 0.6379, "step": 291 }, { "epoch": 0.06, "learning_rate": 1.9964103329248892e-05, "loss": 0.6008, "step": 292 }, { "epoch": 0.06, "learning_rate": 1.996357381814785e-05, "loss": 0.6018, "step": 293 }, { "epoch": 0.06, "learning_rate": 1.996304043730955e-05, "loss": 0.6046, "step": 294 }, { "epoch": 0.06, "learning_rate": 1.9962503186941143e-05, "loss": 0.4834, "step": 295 }, { "epoch": 0.06, "learning_rate": 1.9961962067251298e-05, "loss": 0.5222, "step": 296 }, { "epoch": 0.06, "learning_rate": 1.9961417078450177e-05, "loss": 0.5846, "step": 297 }, { "epoch": 0.06, "learning_rate": 1.996086822074945e-05, "loss": 0.5962, "step": 298 }, { "epoch": 0.06, "learning_rate": 1.9960315494362286e-05, "loss": 0.6214, "step": 299 }, { "epoch": 0.06, "learning_rate": 1.9959758899503355e-05, "loss": 0.5794, "step": 300 }, { "epoch": 0.06, "learning_rate": 1.995919843638883e-05, "loss": 0.641, "step": 301 }, { "epoch": 0.06, "learning_rate": 1.9958634105236395e-05, "loss": 0.4615, "step": 302 }, { "epoch": 0.06, "learning_rate": 1.9958065906265228e-05, "loss": 0.4722, "step": 303 }, { "epoch": 0.06, "learning_rate": 1.9957493839696013e-05, "loss": 0.486, "step": 304 }, { "epoch": 0.06, "learning_rate": 1.9956917905750926e-05, "loss": 0.5895, "step": 305 }, { "epoch": 0.06, "learning_rate": 1.995633810465366e-05, "loss": 0.5914, "step": 306 }, { "epoch": 0.06, "learning_rate": 1.99557544366294e-05, "loss": 0.5941, "step": 307 }, { "epoch": 0.06, "learning_rate": 1.9955166901904838e-05, "loss": 0.5236, "step": 308 }, { "epoch": 0.06, "learning_rate": 1.9954575500708164e-05, "loss": 0.6119, "step": 309 }, { "epoch": 0.06, "learning_rate": 1.995398023326907e-05, "loss": 0.6041, "step": 310 }, { "epoch": 0.06, "learning_rate": 1.9953381099818756e-05, "loss": 0.6068, "step": 311 }, { "epoch": 0.06, "learning_rate": 1.9952778100589912e-05, "loss": 0.5968, "step": 312 }, { "epoch": 0.06, "learning_rate": 1.9952171235816747e-05, "loss": 0.6198, "step": 313 }, { "epoch": 0.06, "learning_rate": 1.9951560505734948e-05, "loss": 0.4951, "step": 314 }, { "epoch": 0.06, "learning_rate": 1.9950945910581718e-05, "loss": 0.6198, "step": 315 }, { "epoch": 0.06, "learning_rate": 1.9950327450595766e-05, "loss": 0.3923, "step": 316 }, { "epoch": 0.06, "learning_rate": 1.9949705126017286e-05, "loss": 0.5229, "step": 317 }, { "epoch": 0.06, "learning_rate": 1.9949078937087988e-05, "loss": 0.376, "step": 318 }, { "epoch": 0.06, "learning_rate": 1.994844888405107e-05, "loss": 0.6154, "step": 319 }, { "epoch": 0.06, "learning_rate": 1.9947814967151246e-05, "loss": 0.4958, "step": 320 }, { "epoch": 0.06, "learning_rate": 1.9947177186634716e-05, "loss": 0.6561, "step": 321 }, { "epoch": 0.06, "learning_rate": 1.9946535542749187e-05, "loss": 0.5708, "step": 322 }, { "epoch": 0.06, "learning_rate": 1.9945890035743866e-05, "loss": 0.5869, "step": 323 }, { "epoch": 0.06, "learning_rate": 1.9945240665869465e-05, "loss": 0.5845, "step": 324 }, { "epoch": 0.06, "learning_rate": 1.9944587433378187e-05, "loss": 0.6454, "step": 325 }, { "epoch": 0.06, "learning_rate": 1.994393033852374e-05, "loss": 0.6096, "step": 326 }, { "epoch": 0.06, "learning_rate": 1.9943269381561334e-05, "loss": 0.6124, "step": 327 }, { "epoch": 0.06, "learning_rate": 1.994260456274768e-05, "loss": 0.594, "step": 328 }, { "epoch": 0.06, "learning_rate": 1.9941935882340976e-05, "loss": 0.6032, "step": 329 }, { "epoch": 0.06, "learning_rate": 1.994126334060094e-05, "loss": 0.5403, "step": 330 }, { "epoch": 0.06, "learning_rate": 1.994058693778878e-05, "loss": 0.6185, "step": 331 }, { "epoch": 0.06, "learning_rate": 1.9939906674167192e-05, "loss": 0.5903, "step": 332 }, { "epoch": 0.06, "learning_rate": 1.993922255000039e-05, "loss": 0.6377, "step": 333 }, { "epoch": 0.06, "learning_rate": 1.993853456555408e-05, "loss": 0.6234, "step": 334 }, { "epoch": 0.06, "learning_rate": 1.9937842721095468e-05, "loss": 0.6505, "step": 335 }, { "epoch": 0.06, "learning_rate": 1.9937147016893257e-05, "loss": 0.5746, "step": 336 }, { "epoch": 0.06, "learning_rate": 1.9936447453217646e-05, "loss": 0.5948, "step": 337 }, { "epoch": 0.07, "learning_rate": 1.9935744030340347e-05, "loss": 0.5365, "step": 338 }, { "epoch": 0.07, "learning_rate": 1.9935036748534555e-05, "loss": 0.5128, "step": 339 }, { "epoch": 0.07, "learning_rate": 1.993432560807497e-05, "loss": 0.6312, "step": 340 }, { "epoch": 0.07, "learning_rate": 1.993361060923779e-05, "loss": 0.5732, "step": 341 }, { "epoch": 0.07, "learning_rate": 1.9932891752300717e-05, "loss": 0.3878, "step": 342 }, { "epoch": 0.07, "learning_rate": 1.9932169037542947e-05, "loss": 0.5061, "step": 343 }, { "epoch": 0.07, "learning_rate": 1.9931442465245164e-05, "loss": 0.6012, "step": 344 }, { "epoch": 0.07, "learning_rate": 1.9930712035689576e-05, "loss": 0.6054, "step": 345 }, { "epoch": 0.07, "learning_rate": 1.992997774915986e-05, "loss": 0.6418, "step": 346 }, { "epoch": 0.07, "learning_rate": 1.992923960594121e-05, "loss": 0.5778, "step": 347 }, { "epoch": 0.07, "learning_rate": 1.9928497606320308e-05, "loss": 0.4857, "step": 348 }, { "epoch": 0.07, "learning_rate": 1.992775175058535e-05, "loss": 0.475, "step": 349 }, { "epoch": 0.07, "learning_rate": 1.9927002039026002e-05, "loss": 0.5203, "step": 350 }, { "epoch": 0.07, "learning_rate": 1.9926248471933453e-05, "loss": 0.6302, "step": 351 }, { "epoch": 0.07, "learning_rate": 1.9925491049600382e-05, "loss": 0.6122, "step": 352 }, { "epoch": 0.07, "learning_rate": 1.9924729772320953e-05, "loss": 0.5622, "step": 353 }, { "epoch": 0.07, "learning_rate": 1.9923964640390846e-05, "loss": 0.6148, "step": 354 }, { "epoch": 0.07, "learning_rate": 1.9923195654107227e-05, "loss": 0.589, "step": 355 }, { "epoch": 0.07, "learning_rate": 1.992242281376876e-05, "loss": 0.5258, "step": 356 }, { "epoch": 0.07, "learning_rate": 1.9921646119675606e-05, "loss": 0.5029, "step": 357 }, { "epoch": 0.07, "learning_rate": 1.9920865572129426e-05, "loss": 0.6769, "step": 358 }, { "epoch": 0.07, "learning_rate": 1.9920081171433377e-05, "loss": 0.6153, "step": 359 }, { "epoch": 0.07, "learning_rate": 1.991929291789211e-05, "loss": 0.6189, "step": 360 }, { "epoch": 0.07, "learning_rate": 1.9918500811811778e-05, "loss": 0.6231, "step": 361 }, { "epoch": 0.07, "learning_rate": 1.991770485350002e-05, "loss": 0.5996, "step": 362 }, { "epoch": 0.07, "learning_rate": 1.991690504326597e-05, "loss": 0.49, "step": 363 }, { "epoch": 0.07, "learning_rate": 1.9916101381420285e-05, "loss": 0.618, "step": 364 }, { "epoch": 0.07, "learning_rate": 1.9915293868275083e-05, "loss": 0.64, "step": 365 }, { "epoch": 0.07, "learning_rate": 1.9914482504143996e-05, "loss": 0.6203, "step": 366 }, { "epoch": 0.07, "learning_rate": 1.9913667289342147e-05, "loss": 0.5983, "step": 367 }, { "epoch": 0.07, "learning_rate": 1.991284822418616e-05, "loss": 0.5752, "step": 368 }, { "epoch": 0.07, "learning_rate": 1.9912025308994146e-05, "loss": 0.4788, "step": 369 }, { "epoch": 0.07, "learning_rate": 1.9911198544085723e-05, "loss": 0.5837, "step": 370 }, { "epoch": 0.07, "learning_rate": 1.991036792978199e-05, "loss": 0.6469, "step": 371 }, { "epoch": 0.07, "learning_rate": 1.990953346640555e-05, "loss": 0.6228, "step": 372 }, { "epoch": 0.07, "learning_rate": 1.9908695154280496e-05, "loss": 0.6208, "step": 373 }, { "epoch": 0.07, "learning_rate": 1.9907852993732425e-05, "loss": 0.5914, "step": 374 }, { "epoch": 0.07, "learning_rate": 1.990700698508842e-05, "loss": 0.4966, "step": 375 }, { "epoch": 0.07, "learning_rate": 1.990615712867706e-05, "loss": 0.6559, "step": 376 }, { "epoch": 0.07, "learning_rate": 1.9905303424828418e-05, "loss": 0.6233, "step": 377 }, { "epoch": 0.07, "learning_rate": 1.9904445873874068e-05, "loss": 0.6389, "step": 378 }, { "epoch": 0.07, "learning_rate": 1.9903584476147066e-05, "loss": 0.6107, "step": 379 }, { "epoch": 0.07, "learning_rate": 1.9902719231981975e-05, "loss": 0.6474, "step": 380 }, { "epoch": 0.07, "learning_rate": 1.9901850141714843e-05, "loss": 0.5936, "step": 381 }, { "epoch": 0.07, "learning_rate": 1.9900977205683213e-05, "loss": 0.5962, "step": 382 }, { "epoch": 0.07, "learning_rate": 1.9900100424226124e-05, "loss": 0.475, "step": 383 }, { "epoch": 0.07, "learning_rate": 1.9899219797684113e-05, "loss": 0.5205, "step": 384 }, { "epoch": 0.07, "learning_rate": 1.98983353263992e-05, "loss": 0.6281, "step": 385 }, { "epoch": 0.07, "learning_rate": 1.9897447010714905e-05, "loss": 0.611, "step": 386 }, { "epoch": 0.07, "learning_rate": 1.989655485097624e-05, "loss": 0.5874, "step": 387 }, { "epoch": 0.07, "learning_rate": 1.989565884752971e-05, "loss": 0.6147, "step": 388 }, { "epoch": 0.07, "learning_rate": 1.9894759000723308e-05, "loss": 0.6487, "step": 389 }, { "epoch": 0.08, "learning_rate": 1.9893855310906526e-05, "loss": 0.5749, "step": 390 }, { "epoch": 0.08, "learning_rate": 1.9892947778430352e-05, "loss": 0.5919, "step": 391 }, { "epoch": 0.08, "learning_rate": 1.9892036403647256e-05, "loss": 0.5996, "step": 392 }, { "epoch": 0.08, "learning_rate": 1.9891121186911207e-05, "loss": 0.4937, "step": 393 }, { "epoch": 0.08, "learning_rate": 1.9890202128577664e-05, "loss": 0.5969, "step": 394 }, { "epoch": 0.08, "learning_rate": 1.988927922900358e-05, "loss": 0.5271, "step": 395 }, { "epoch": 0.08, "learning_rate": 1.9888352488547397e-05, "loss": 0.5849, "step": 396 }, { "epoch": 0.08, "learning_rate": 1.988742190756905e-05, "loss": 0.5207, "step": 397 }, { "epoch": 0.08, "learning_rate": 1.9886487486429966e-05, "loss": 0.6153, "step": 398 }, { "epoch": 0.08, "learning_rate": 1.9885549225493064e-05, "loss": 0.627, "step": 399 }, { "epoch": 0.08, "learning_rate": 1.9884607125122753e-05, "loss": 0.494, "step": 400 }, { "epoch": 0.08, "learning_rate": 1.988366118568494e-05, "loss": 0.4993, "step": 401 }, { "epoch": 0.08, "learning_rate": 1.988271140754701e-05, "loss": 0.6213, "step": 402 }, { "epoch": 0.08, "learning_rate": 1.9881757791077848e-05, "loss": 0.5462, "step": 403 }, { "epoch": 0.08, "learning_rate": 1.9880800336647825e-05, "loss": 0.5806, "step": 404 }, { "epoch": 0.08, "learning_rate": 1.987983904462881e-05, "loss": 0.5671, "step": 405 }, { "epoch": 0.08, "learning_rate": 1.9878873915394154e-05, "loss": 0.4681, "step": 406 }, { "epoch": 0.08, "learning_rate": 1.9877904949318704e-05, "loss": 0.6251, "step": 407 }, { "epoch": 0.08, "learning_rate": 1.9876932146778796e-05, "loss": 0.5881, "step": 408 }, { "epoch": 0.08, "learning_rate": 1.9875955508152254e-05, "loss": 0.4791, "step": 409 }, { "epoch": 0.08, "learning_rate": 1.987497503381839e-05, "loss": 0.6384, "step": 410 }, { "epoch": 0.08, "learning_rate": 1.9873990724158014e-05, "loss": 0.6137, "step": 411 }, { "epoch": 0.08, "learning_rate": 1.987300257955342e-05, "loss": 0.5916, "step": 412 }, { "epoch": 0.08, "learning_rate": 1.987201060038839e-05, "loss": 0.6055, "step": 413 }, { "epoch": 0.08, "learning_rate": 1.9871014787048197e-05, "loss": 0.603, "step": 414 }, { "epoch": 0.08, "learning_rate": 1.9870015139919606e-05, "loss": 0.5995, "step": 415 }, { "epoch": 0.08, "learning_rate": 1.9869011659390866e-05, "loss": 0.573, "step": 416 }, { "epoch": 0.08, "learning_rate": 1.9868004345851716e-05, "loss": 0.4806, "step": 417 }, { "epoch": 0.08, "learning_rate": 1.9866993199693393e-05, "loss": 0.4987, "step": 418 }, { "epoch": 0.08, "learning_rate": 1.98659782213086e-05, "loss": 0.5028, "step": 419 }, { "epoch": 0.08, "learning_rate": 1.986495941109156e-05, "loss": 0.5118, "step": 420 }, { "epoch": 0.08, "learning_rate": 1.9863936769437956e-05, "loss": 0.5789, "step": 421 }, { "epoch": 0.08, "learning_rate": 1.986291029674497e-05, "loss": 0.6371, "step": 422 }, { "epoch": 0.08, "learning_rate": 1.986187999341128e-05, "loss": 0.6321, "step": 423 }, { "epoch": 0.08, "learning_rate": 1.9860845859837034e-05, "loss": 0.6285, "step": 424 }, { "epoch": 0.08, "learning_rate": 1.985980789642388e-05, "loss": 0.5901, "step": 425 }, { "epoch": 0.08, "learning_rate": 1.985876610357496e-05, "loss": 0.5876, "step": 426 }, { "epoch": 0.08, "learning_rate": 1.9857720481694887e-05, "loss": 0.6116, "step": 427 }, { "epoch": 0.08, "learning_rate": 1.9856671031189765e-05, "loss": 0.6342, "step": 428 }, { "epoch": 0.08, "learning_rate": 1.98556177524672e-05, "loss": 0.5871, "step": 429 }, { "epoch": 0.08, "learning_rate": 1.9854560645936262e-05, "loss": 0.5118, "step": 430 }, { "epoch": 0.08, "learning_rate": 1.9853499712007523e-05, "loss": 0.4796, "step": 431 }, { "epoch": 0.08, "learning_rate": 1.9852434951093035e-05, "loss": 0.6498, "step": 432 }, { "epoch": 0.08, "learning_rate": 1.985136636360635e-05, "loss": 0.6357, "step": 433 }, { "epoch": 0.08, "learning_rate": 1.985029394996248e-05, "loss": 0.5874, "step": 434 }, { "epoch": 0.08, "learning_rate": 1.9849217710577945e-05, "loss": 0.4977, "step": 435 }, { "epoch": 0.08, "learning_rate": 1.9848137645870745e-05, "loss": 0.4615, "step": 436 }, { "epoch": 0.08, "learning_rate": 1.9847053756260363e-05, "loss": 0.5631, "step": 437 }, { "epoch": 0.08, "learning_rate": 1.984596604216777e-05, "loss": 0.6138, "step": 438 }, { "epoch": 0.08, "learning_rate": 1.984487450401542e-05, "loss": 0.6022, "step": 439 }, { "epoch": 0.08, "learning_rate": 1.9843779142227258e-05, "loss": 0.5982, "step": 440 }, { "epoch": 0.08, "learning_rate": 1.9842679957228706e-05, "loss": 0.6411, "step": 441 }, { "epoch": 0.09, "learning_rate": 1.9841576949446675e-05, "loss": 0.5634, "step": 442 }, { "epoch": 0.09, "learning_rate": 1.984047011930956e-05, "loss": 0.5013, "step": 443 }, { "epoch": 0.09, "learning_rate": 1.9839359467247243e-05, "loss": 0.5979, "step": 444 }, { "epoch": 0.09, "learning_rate": 1.983824499369109e-05, "loss": 0.5351, "step": 445 }, { "epoch": 0.09, "learning_rate": 1.9837126699073948e-05, "loss": 0.4439, "step": 446 }, { "epoch": 0.09, "learning_rate": 1.9836004583830146e-05, "loss": 0.4928, "step": 447 }, { "epoch": 0.09, "learning_rate": 1.9834878648395507e-05, "loss": 0.5401, "step": 448 }, { "epoch": 0.09, "learning_rate": 1.9833748893207326e-05, "loss": 0.5912, "step": 449 }, { "epoch": 0.09, "learning_rate": 1.9832615318704388e-05, "loss": 0.4824, "step": 450 }, { "epoch": 0.09, "learning_rate": 1.9831477925326962e-05, "loss": 0.4997, "step": 451 }, { "epoch": 0.09, "learning_rate": 1.98303367135168e-05, "loss": 0.5011, "step": 452 }, { "epoch": 0.09, "learning_rate": 1.9829191683717133e-05, "loss": 0.5668, "step": 453 }, { "epoch": 0.09, "learning_rate": 1.9828042836372677e-05, "loss": 0.623, "step": 454 }, { "epoch": 0.09, "learning_rate": 1.9826890171929634e-05, "loss": 0.6117, "step": 455 }, { "epoch": 0.09, "learning_rate": 1.982573369083568e-05, "loss": 0.4957, "step": 456 }, { "epoch": 0.09, "learning_rate": 1.9824573393539984e-05, "loss": 0.661, "step": 457 }, { "epoch": 0.09, "learning_rate": 1.982340928049319e-05, "loss": 0.495, "step": 458 }, { "epoch": 0.09, "learning_rate": 1.9822241352147426e-05, "loss": 0.5778, "step": 459 }, { "epoch": 0.09, "learning_rate": 1.9821069608956307e-05, "loss": 0.6402, "step": 460 }, { "epoch": 0.09, "learning_rate": 1.9819894051374917e-05, "loss": 0.6132, "step": 461 }, { "epoch": 0.09, "learning_rate": 1.981871467985983e-05, "loss": 0.6185, "step": 462 }, { "epoch": 0.09, "learning_rate": 1.9817531494869105e-05, "loss": 0.5992, "step": 463 }, { "epoch": 0.09, "learning_rate": 1.9816344496862272e-05, "loss": 0.6134, "step": 464 }, { "epoch": 0.09, "learning_rate": 1.9815153686300352e-05, "loss": 0.601, "step": 465 }, { "epoch": 0.09, "learning_rate": 1.981395906364584e-05, "loss": 0.5726, "step": 466 }, { "epoch": 0.09, "learning_rate": 1.9812760629362714e-05, "loss": 0.6403, "step": 467 }, { "epoch": 0.09, "learning_rate": 1.981155838391643e-05, "loss": 0.4809, "step": 468 }, { "epoch": 0.09, "learning_rate": 1.9810352327773935e-05, "loss": 0.5605, "step": 469 }, { "epoch": 0.09, "learning_rate": 1.9809142461403635e-05, "loss": 0.4473, "step": 470 }, { "epoch": 0.09, "learning_rate": 1.9807928785275433e-05, "loss": 0.5579, "step": 471 }, { "epoch": 0.09, "learning_rate": 1.980671129986071e-05, "loss": 0.5642, "step": 472 }, { "epoch": 0.09, "learning_rate": 1.9805490005632323e-05, "loss": 0.6157, "step": 473 }, { "epoch": 0.09, "learning_rate": 1.98042649030646e-05, "loss": 0.6151, "step": 474 }, { "epoch": 0.09, "learning_rate": 1.9803035992633366e-05, "loss": 0.6302, "step": 475 }, { "epoch": 0.09, "learning_rate": 1.9801803274815915e-05, "loss": 0.5968, "step": 476 }, { "epoch": 0.09, "learning_rate": 1.9800566750091018e-05, "loss": 0.494, "step": 477 }, { "epoch": 0.09, "learning_rate": 1.9799326418938924e-05, "loss": 0.5291, "step": 478 }, { "epoch": 0.09, "learning_rate": 1.979808228184137e-05, "loss": 0.5061, "step": 479 }, { "epoch": 0.09, "learning_rate": 1.9796834339281557e-05, "loss": 0.5812, "step": 480 }, { "epoch": 0.09, "learning_rate": 1.979558259174418e-05, "loss": 0.4581, "step": 481 }, { "epoch": 0.09, "learning_rate": 1.9794327039715395e-05, "loss": 0.4843, "step": 482 }, { "epoch": 0.09, "learning_rate": 1.979306768368285e-05, "loss": 0.6024, "step": 483 }, { "epoch": 0.09, "learning_rate": 1.9791804524135663e-05, "loss": 0.6036, "step": 484 }, { "epoch": 0.09, "learning_rate": 1.979053756156443e-05, "loss": 0.5723, "step": 485 }, { "epoch": 0.09, "learning_rate": 1.9789266796461222e-05, "loss": 0.478, "step": 486 }, { "epoch": 0.09, "learning_rate": 1.9787992229319594e-05, "loss": 0.5939, "step": 487 }, { "epoch": 0.09, "learning_rate": 1.978671386063457e-05, "loss": 0.5945, "step": 488 }, { "epoch": 0.09, "learning_rate": 1.978543169090265e-05, "loss": 0.3971, "step": 489 }, { "epoch": 0.09, "learning_rate": 1.9784145720621827e-05, "loss": 0.5978, "step": 490 }, { "epoch": 0.09, "learning_rate": 1.9782855950291542e-05, "loss": 0.6137, "step": 491 }, { "epoch": 0.09, "learning_rate": 1.978156238041274e-05, "loss": 0.5426, "step": 492 }, { "epoch": 0.09, "learning_rate": 1.9780265011487822e-05, "loss": 0.6145, "step": 493 }, { "epoch": 0.1, "learning_rate": 1.9778963844020668e-05, "loss": 0.6534, "step": 494 }, { "epoch": 0.1, "learning_rate": 1.977765887851664e-05, "loss": 0.5005, "step": 495 }, { "epoch": 0.1, "learning_rate": 1.977635011548257e-05, "loss": 0.5138, "step": 496 }, { "epoch": 0.1, "learning_rate": 1.9775037555426772e-05, "loss": 0.5311, "step": 497 }, { "epoch": 0.1, "learning_rate": 1.9773721198859024e-05, "loss": 0.4397, "step": 498 }, { "epoch": 0.1, "learning_rate": 1.9772401046290584e-05, "loss": 0.6476, "step": 499 }, { "epoch": 0.1, "learning_rate": 1.9771077098234187e-05, "loss": 0.5818, "step": 500 }, { "epoch": 0.1, "learning_rate": 1.9769749355204034e-05, "loss": 0.6118, "step": 501 }, { "epoch": 0.1, "learning_rate": 1.976841781771581e-05, "loss": 0.4957, "step": 502 }, { "epoch": 0.1, "learning_rate": 1.9767082486286667e-05, "loss": 0.3703, "step": 503 }, { "epoch": 0.1, "learning_rate": 1.9765743361435234e-05, "loss": 0.571, "step": 504 }, { "epoch": 0.1, "learning_rate": 1.9764400443681607e-05, "loss": 0.5526, "step": 505 }, { "epoch": 0.1, "learning_rate": 1.9763053733547367e-05, "loss": 0.5633, "step": 506 }, { "epoch": 0.1, "learning_rate": 1.976170323155555e-05, "loss": 0.6204, "step": 507 }, { "epoch": 0.1, "learning_rate": 1.976034893823069e-05, "loss": 0.6255, "step": 508 }, { "epoch": 0.1, "learning_rate": 1.975899085409876e-05, "loss": 0.6225, "step": 509 }, { "epoch": 0.1, "learning_rate": 1.9757628979687247e-05, "loss": 0.6133, "step": 510 }, { "epoch": 0.1, "learning_rate": 1.975626331552507e-05, "loss": 0.579, "step": 511 }, { "epoch": 0.1, "learning_rate": 1.9754893862142643e-05, "loss": 0.5967, "step": 512 }, { "epoch": 0.1, "learning_rate": 1.9753520620071846e-05, "loss": 0.4807, "step": 513 }, { "epoch": 0.1, "learning_rate": 1.9752143589846027e-05, "loss": 0.5926, "step": 514 }, { "epoch": 0.1, "learning_rate": 1.9750762772000014e-05, "loss": 0.4806, "step": 515 }, { "epoch": 0.1, "learning_rate": 1.9749378167070097e-05, "loss": 0.5065, "step": 516 }, { "epoch": 0.1, "learning_rate": 1.9747989775594044e-05, "loss": 0.5799, "step": 517 }, { "epoch": 0.1, "learning_rate": 1.974659759811109e-05, "loss": 0.5847, "step": 518 }, { "epoch": 0.1, "learning_rate": 1.9745201635161938e-05, "loss": 0.5079, "step": 519 }, { "epoch": 0.1, "learning_rate": 1.9743801887288762e-05, "loss": 0.5517, "step": 520 }, { "epoch": 0.1, "learning_rate": 1.9742398355035212e-05, "loss": 0.6365, "step": 521 }, { "epoch": 0.1, "learning_rate": 1.9740991038946404e-05, "loss": 0.6222, "step": 522 }, { "epoch": 0.1, "learning_rate": 1.973957993956892e-05, "loss": 0.4981, "step": 523 }, { "epoch": 0.1, "learning_rate": 1.9738165057450817e-05, "loss": 0.6079, "step": 524 }, { "epoch": 0.1, "learning_rate": 1.9736746393141617e-05, "loss": 0.6281, "step": 525 }, { "epoch": 0.1, "learning_rate": 1.9735323947192317e-05, "loss": 0.5965, "step": 526 }, { "epoch": 0.1, "learning_rate": 1.9733897720155377e-05, "loss": 0.5914, "step": 527 }, { "epoch": 0.1, "learning_rate": 1.9732467712584723e-05, "loss": 0.6392, "step": 528 }, { "epoch": 0.1, "learning_rate": 1.973103392503576e-05, "loss": 0.4657, "step": 529 }, { "epoch": 0.1, "learning_rate": 1.9729596358065347e-05, "loss": 0.5172, "step": 530 }, { "epoch": 0.1, "learning_rate": 1.9728155012231825e-05, "loss": 0.5946, "step": 531 }, { "epoch": 0.1, "learning_rate": 1.9726709888094994e-05, "loss": 0.6144, "step": 532 }, { "epoch": 0.1, "learning_rate": 1.972526098621612e-05, "loss": 0.597, "step": 533 }, { "epoch": 0.1, "learning_rate": 1.972380830715795e-05, "loss": 0.4893, "step": 534 }, { "epoch": 0.1, "learning_rate": 1.9722351851484677e-05, "loss": 0.5092, "step": 535 }, { "epoch": 0.1, "learning_rate": 1.9720891619761974e-05, "loss": 0.5737, "step": 536 }, { "epoch": 0.1, "learning_rate": 1.9719427612556982e-05, "loss": 0.5458, "step": 537 }, { "epoch": 0.1, "learning_rate": 1.9717959830438302e-05, "loss": 0.6011, "step": 538 }, { "epoch": 0.1, "learning_rate": 1.9716488273976006e-05, "loss": 0.6226, "step": 539 }, { "epoch": 0.1, "learning_rate": 1.971501294374162e-05, "loss": 0.6248, "step": 540 }, { "epoch": 0.1, "learning_rate": 1.971353384030816e-05, "loss": 0.5617, "step": 541 }, { "epoch": 0.1, "learning_rate": 1.9712050964250083e-05, "loss": 0.5899, "step": 542 }, { "epoch": 0.1, "learning_rate": 1.9710564316143323e-05, "loss": 0.5792, "step": 543 }, { "epoch": 0.1, "learning_rate": 1.9709073896565276e-05, "loss": 0.6081, "step": 544 }, { "epoch": 0.1, "learning_rate": 1.9707579706094807e-05, "loss": 0.6118, "step": 545 }, { "epoch": 0.11, "learning_rate": 1.970608174531224e-05, "loss": 0.6033, "step": 546 }, { "epoch": 0.11, "learning_rate": 1.970458001479937e-05, "loss": 0.5155, "step": 547 }, { "epoch": 0.11, "learning_rate": 1.9703074515139445e-05, "loss": 0.5736, "step": 548 }, { "epoch": 0.11, "learning_rate": 1.9701565246917184e-05, "loss": 0.5979, "step": 549 }, { "epoch": 0.11, "learning_rate": 1.9700052210718775e-05, "loss": 0.6469, "step": 550 }, { "epoch": 0.11, "learning_rate": 1.969853540713186e-05, "loss": 0.4911, "step": 551 }, { "epoch": 0.11, "learning_rate": 1.9697014836745552e-05, "loss": 0.6319, "step": 552 }, { "epoch": 0.11, "learning_rate": 1.9695490500150418e-05, "loss": 0.5939, "step": 553 }, { "epoch": 0.11, "learning_rate": 1.9693962397938495e-05, "loss": 0.6478, "step": 554 }, { "epoch": 0.11, "learning_rate": 1.9692430530703282e-05, "loss": 0.6048, "step": 555 }, { "epoch": 0.11, "learning_rate": 1.9690894899039735e-05, "loss": 0.6606, "step": 556 }, { "epoch": 0.11, "learning_rate": 1.9689355503544277e-05, "loss": 0.5969, "step": 557 }, { "epoch": 0.11, "learning_rate": 1.968781234481479e-05, "loss": 0.6042, "step": 558 }, { "epoch": 0.11, "learning_rate": 1.9686265423450624e-05, "loss": 0.5928, "step": 559 }, { "epoch": 0.11, "learning_rate": 1.9684714740052584e-05, "loss": 0.5926, "step": 560 }, { "epoch": 0.11, "learning_rate": 1.9683160295222934e-05, "loss": 0.5957, "step": 561 }, { "epoch": 0.11, "learning_rate": 1.9681602089565403e-05, "loss": 0.6123, "step": 562 }, { "epoch": 0.11, "learning_rate": 1.968004012368518e-05, "loss": 0.6087, "step": 563 }, { "epoch": 0.11, "learning_rate": 1.967847439818892e-05, "loss": 0.6115, "step": 564 }, { "epoch": 0.11, "learning_rate": 1.9676904913684725e-05, "loss": 0.5915, "step": 565 }, { "epoch": 0.11, "learning_rate": 1.967533167078217e-05, "loss": 0.5076, "step": 566 }, { "epoch": 0.11, "learning_rate": 1.9673754670092283e-05, "loss": 0.5951, "step": 567 }, { "epoch": 0.11, "learning_rate": 1.9672173912227556e-05, "loss": 0.5982, "step": 568 }, { "epoch": 0.11, "learning_rate": 1.967058939780193e-05, "loss": 0.5437, "step": 569 }, { "epoch": 0.11, "learning_rate": 1.966900112743082e-05, "loss": 0.3865, "step": 570 }, { "epoch": 0.11, "learning_rate": 1.966740910173108e-05, "loss": 0.5772, "step": 571 }, { "epoch": 0.11, "learning_rate": 1.9665813321321054e-05, "loss": 0.5866, "step": 572 }, { "epoch": 0.11, "learning_rate": 1.9664213786820502e-05, "loss": 0.6333, "step": 573 }, { "epoch": 0.11, "learning_rate": 1.9662610498850684e-05, "loss": 0.5951, "step": 574 }, { "epoch": 0.11, "learning_rate": 1.9661003458034288e-05, "loss": 0.647, "step": 575 }, { "epoch": 0.11, "learning_rate": 1.965939266499547e-05, "loss": 0.4921, "step": 576 }, { "epoch": 0.11, "learning_rate": 1.9657778120359848e-05, "loss": 0.5723, "step": 577 }, { "epoch": 0.11, "learning_rate": 1.965615982475449e-05, "loss": 0.5312, "step": 578 }, { "epoch": 0.11, "learning_rate": 1.9654537778807924e-05, "loss": 0.5879, "step": 579 }, { "epoch": 0.11, "learning_rate": 1.9652911983150135e-05, "loss": 0.6428, "step": 580 }, { "epoch": 0.11, "learning_rate": 1.965128243841256e-05, "loss": 0.6262, "step": 581 }, { "epoch": 0.11, "learning_rate": 1.96496491452281e-05, "loss": 0.6041, "step": 582 }, { "epoch": 0.11, "learning_rate": 1.9648012104231106e-05, "loss": 0.614, "step": 583 }, { "epoch": 0.11, "learning_rate": 1.964637131605738e-05, "loss": 0.601, "step": 584 }, { "epoch": 0.11, "learning_rate": 1.9644726781344197e-05, "loss": 0.605, "step": 585 }, { "epoch": 0.11, "learning_rate": 1.964307850073026e-05, "loss": 0.586, "step": 586 }, { "epoch": 0.11, "learning_rate": 1.964142647485576e-05, "loss": 0.599, "step": 587 }, { "epoch": 0.11, "learning_rate": 1.9639770704362305e-05, "loss": 0.504, "step": 588 }, { "epoch": 0.11, "learning_rate": 1.9638111189892994e-05, "loss": 0.6156, "step": 589 }, { "epoch": 0.11, "learning_rate": 1.9636447932092354e-05, "loss": 0.5924, "step": 590 }, { "epoch": 0.11, "learning_rate": 1.963478093160638e-05, "loss": 0.5715, "step": 591 }, { "epoch": 0.11, "learning_rate": 1.9633110189082515e-05, "loss": 0.6452, "step": 592 }, { "epoch": 0.11, "learning_rate": 1.963143570516965e-05, "loss": 0.5065, "step": 593 }, { "epoch": 0.11, "learning_rate": 1.9629757480518144e-05, "loss": 0.6115, "step": 594 }, { "epoch": 0.11, "learning_rate": 1.9628075515779796e-05, "loss": 0.4903, "step": 595 }, { "epoch": 0.11, "learning_rate": 1.962638981160786e-05, "loss": 0.4943, "step": 596 }, { "epoch": 0.11, "learning_rate": 1.9624700368657045e-05, "loss": 0.6398, "step": 597 }, { "epoch": 0.12, "learning_rate": 1.9623007187583518e-05, "loss": 0.573, "step": 598 }, { "epoch": 0.12, "learning_rate": 1.962131026904488e-05, "loss": 0.6646, "step": 599 }, { "epoch": 0.12, "learning_rate": 1.96196096137002e-05, "loss": 0.508, "step": 600 }, { "epoch": 0.12, "learning_rate": 1.9617905222209998e-05, "loss": 0.6266, "step": 601 }, { "epoch": 0.12, "learning_rate": 1.961619709523623e-05, "loss": 0.5725, "step": 602 }, { "epoch": 0.12, "learning_rate": 1.9614485233442316e-05, "loss": 0.462, "step": 603 }, { "epoch": 0.12, "learning_rate": 1.961276963749313e-05, "loss": 0.6235, "step": 604 }, { "epoch": 0.12, "learning_rate": 1.9611050308054982e-05, "loss": 0.6281, "step": 605 }, { "epoch": 0.12, "learning_rate": 1.9609327245795642e-05, "loss": 0.5798, "step": 606 }, { "epoch": 0.12, "learning_rate": 1.9607600451384327e-05, "loss": 0.5845, "step": 607 }, { "epoch": 0.12, "learning_rate": 1.960586992549171e-05, "loss": 0.6309, "step": 608 }, { "epoch": 0.12, "learning_rate": 1.9604135668789897e-05, "loss": 0.3719, "step": 609 }, { "epoch": 0.12, "learning_rate": 1.9602397681952462e-05, "loss": 0.5967, "step": 610 }, { "epoch": 0.12, "learning_rate": 1.9600655965654413e-05, "loss": 0.6195, "step": 611 }, { "epoch": 0.12, "learning_rate": 1.959891052057222e-05, "loss": 0.673, "step": 612 }, { "epoch": 0.12, "learning_rate": 1.9597161347383783e-05, "loss": 0.5794, "step": 613 }, { "epoch": 0.12, "learning_rate": 1.959540844676847e-05, "loss": 0.6027, "step": 614 }, { "epoch": 0.12, "learning_rate": 1.9593651819407084e-05, "loss": 0.6263, "step": 615 }, { "epoch": 0.12, "learning_rate": 1.959189146598188e-05, "loss": 0.6216, "step": 616 }, { "epoch": 0.12, "learning_rate": 1.9590127387176556e-05, "loss": 0.6225, "step": 617 }, { "epoch": 0.12, "learning_rate": 1.9588359583676263e-05, "loss": 0.5884, "step": 618 }, { "epoch": 0.12, "learning_rate": 1.9586588056167595e-05, "loss": 0.4702, "step": 619 }, { "epoch": 0.12, "learning_rate": 1.958481280533859e-05, "loss": 0.6404, "step": 620 }, { "epoch": 0.12, "learning_rate": 1.958303383187874e-05, "loss": 0.4723, "step": 621 }, { "epoch": 0.12, "learning_rate": 1.9581251136478974e-05, "loss": 0.5093, "step": 622 }, { "epoch": 0.12, "learning_rate": 1.9579464719831668e-05, "loss": 0.5964, "step": 623 }, { "epoch": 0.12, "learning_rate": 1.9577674582630653e-05, "loss": 0.6376, "step": 624 }, { "epoch": 0.12, "learning_rate": 1.957588072557119e-05, "loss": 0.6478, "step": 625 }, { "epoch": 0.12, "learning_rate": 1.957408314935e-05, "loss": 0.5982, "step": 626 }, { "epoch": 0.12, "learning_rate": 1.9572281854665233e-05, "loss": 0.5906, "step": 627 }, { "epoch": 0.12, "learning_rate": 1.95704768422165e-05, "loss": 0.6043, "step": 628 }, { "epoch": 0.12, "learning_rate": 1.956866811270484e-05, "loss": 0.6169, "step": 629 }, { "epoch": 0.12, "learning_rate": 1.9566855666832743e-05, "loss": 0.6039, "step": 630 }, { "epoch": 0.12, "learning_rate": 1.9565039505304145e-05, "loss": 0.5912, "step": 631 }, { "epoch": 0.12, "learning_rate": 1.956321962882442e-05, "loss": 0.5666, "step": 632 }, { "epoch": 0.12, "learning_rate": 1.956139603810039e-05, "loss": 0.6197, "step": 633 }, { "epoch": 0.12, "learning_rate": 1.9559568733840317e-05, "loss": 0.6081, "step": 634 }, { "epoch": 0.12, "learning_rate": 1.9557737716753898e-05, "loss": 0.6464, "step": 635 }, { "epoch": 0.12, "learning_rate": 1.9555902987552283e-05, "loss": 0.6173, "step": 636 }, { "epoch": 0.12, "learning_rate": 1.9554064546948064e-05, "loss": 0.6193, "step": 637 }, { "epoch": 0.12, "learning_rate": 1.9552222395655262e-05, "loss": 0.6043, "step": 638 }, { "epoch": 0.12, "learning_rate": 1.9550376534389355e-05, "loss": 0.6245, "step": 639 }, { "epoch": 0.12, "learning_rate": 1.9548526963867253e-05, "loss": 0.5792, "step": 640 }, { "epoch": 0.12, "learning_rate": 1.9546673684807303e-05, "loss": 0.5245, "step": 641 }, { "epoch": 0.12, "learning_rate": 1.95448166979293e-05, "loss": 0.6322, "step": 642 }, { "epoch": 0.12, "learning_rate": 1.9542956003954477e-05, "loss": 0.604, "step": 643 }, { "epoch": 0.12, "learning_rate": 1.9541091603605508e-05, "loss": 0.458, "step": 644 }, { "epoch": 0.12, "learning_rate": 1.95392234976065e-05, "loss": 0.4597, "step": 645 }, { "epoch": 0.12, "learning_rate": 1.9537351686683003e-05, "loss": 0.6145, "step": 646 }, { "epoch": 0.12, "learning_rate": 1.9535476171562014e-05, "loss": 0.5713, "step": 647 }, { "epoch": 0.12, "learning_rate": 1.9533596952971955e-05, "loss": 0.5754, "step": 648 }, { "epoch": 0.12, "learning_rate": 1.9531714031642698e-05, "loss": 0.6039, "step": 649 }, { "epoch": 0.13, "learning_rate": 1.9529827408305542e-05, "loss": 0.6423, "step": 650 }, { "epoch": 0.13, "learning_rate": 1.9527937083693233e-05, "loss": 0.6337, "step": 651 }, { "epoch": 0.13, "learning_rate": 1.952604305853995e-05, "loss": 0.6464, "step": 652 }, { "epoch": 0.13, "learning_rate": 1.9524145333581315e-05, "loss": 0.4934, "step": 653 }, { "epoch": 0.13, "learning_rate": 1.9522243909554375e-05, "loss": 0.6039, "step": 654 }, { "epoch": 0.13, "learning_rate": 1.952033878719763e-05, "loss": 0.3708, "step": 655 }, { "epoch": 0.13, "learning_rate": 1.9518429967251e-05, "loss": 0.4473, "step": 656 }, { "epoch": 0.13, "learning_rate": 1.9516517450455853e-05, "loss": 0.6005, "step": 657 }, { "epoch": 0.13, "learning_rate": 1.951460123755499e-05, "loss": 0.6039, "step": 658 }, { "epoch": 0.13, "learning_rate": 1.9512681329292635e-05, "loss": 0.4929, "step": 659 }, { "epoch": 0.13, "learning_rate": 1.951075772641447e-05, "loss": 0.4877, "step": 660 }, { "epoch": 0.13, "learning_rate": 1.95088304296676e-05, "loss": 0.3703, "step": 661 }, { "epoch": 0.13, "learning_rate": 1.950689943980056e-05, "loss": 0.6491, "step": 662 }, { "epoch": 0.13, "learning_rate": 1.9504964757563322e-05, "loss": 0.6325, "step": 663 }, { "epoch": 0.13, "learning_rate": 1.95030263837073e-05, "loss": 0.5779, "step": 664 }, { "epoch": 0.13, "learning_rate": 1.9501084318985335e-05, "loss": 0.572, "step": 665 }, { "epoch": 0.13, "learning_rate": 1.94991385641517e-05, "loss": 0.6068, "step": 666 }, { "epoch": 0.13, "learning_rate": 1.9497189119962105e-05, "loss": 0.6224, "step": 667 }, { "epoch": 0.13, "learning_rate": 1.9495235987173693e-05, "loss": 0.4497, "step": 668 }, { "epoch": 0.13, "learning_rate": 1.949327916654504e-05, "loss": 0.6261, "step": 669 }, { "epoch": 0.13, "learning_rate": 1.949131865883614e-05, "loss": 0.6622, "step": 670 }, { "epoch": 0.13, "learning_rate": 1.948935446480845e-05, "loss": 0.6111, "step": 671 }, { "epoch": 0.13, "learning_rate": 1.948738658522483e-05, "loss": 0.495, "step": 672 }, { "epoch": 0.13, "learning_rate": 1.9485415020849583e-05, "loss": 0.5725, "step": 673 }, { "epoch": 0.13, "learning_rate": 1.9483439772448444e-05, "loss": 0.5933, "step": 674 }, { "epoch": 0.13, "learning_rate": 1.9481460840788573e-05, "loss": 0.6064, "step": 675 }, { "epoch": 0.13, "learning_rate": 1.9479478226638565e-05, "loss": 0.6082, "step": 676 }, { "epoch": 0.13, "learning_rate": 1.947749193076845e-05, "loss": 0.5967, "step": 677 }, { "epoch": 0.13, "learning_rate": 1.9475501953949674e-05, "loss": 0.6157, "step": 678 }, { "epoch": 0.13, "learning_rate": 1.9473508296955126e-05, "loss": 0.5464, "step": 679 }, { "epoch": 0.13, "learning_rate": 1.9471510960559122e-05, "loss": 0.6051, "step": 680 }, { "epoch": 0.13, "learning_rate": 1.9469509945537395e-05, "loss": 0.6344, "step": 681 }, { "epoch": 0.13, "learning_rate": 1.9467505252667126e-05, "loss": 0.6227, "step": 682 }, { "epoch": 0.13, "learning_rate": 1.9465496882726913e-05, "loss": 0.572, "step": 683 }, { "epoch": 0.13, "learning_rate": 1.946348483649678e-05, "loss": 0.6334, "step": 684 }, { "epoch": 0.13, "learning_rate": 1.9461469114758184e-05, "loss": 0.6136, "step": 685 }, { "epoch": 0.13, "learning_rate": 1.9459449718294008e-05, "loss": 0.5024, "step": 686 }, { "epoch": 0.13, "learning_rate": 1.945742664788856e-05, "loss": 0.5835, "step": 687 }, { "epoch": 0.13, "learning_rate": 1.9455399904327585e-05, "loss": 0.4859, "step": 688 }, { "epoch": 0.13, "learning_rate": 1.945336948839824e-05, "loss": 0.6013, "step": 689 }, { "epoch": 0.13, "learning_rate": 1.9451335400889114e-05, "loss": 0.607, "step": 690 }, { "epoch": 0.13, "learning_rate": 1.944929764259023e-05, "loss": 0.635, "step": 691 }, { "epoch": 0.13, "learning_rate": 1.9447256214293026e-05, "loss": 0.6068, "step": 692 }, { "epoch": 0.13, "learning_rate": 1.9445211116790365e-05, "loss": 0.6094, "step": 693 }, { "epoch": 0.13, "learning_rate": 1.9443162350876544e-05, "loss": 0.6228, "step": 694 }, { "epoch": 0.13, "learning_rate": 1.944110991734728e-05, "loss": 0.6341, "step": 695 }, { "epoch": 0.13, "learning_rate": 1.9439053816999717e-05, "loss": 0.3718, "step": 696 }, { "epoch": 0.13, "learning_rate": 1.9436994050632415e-05, "loss": 0.6154, "step": 697 }, { "epoch": 0.13, "learning_rate": 1.9434930619045367e-05, "loss": 0.6372, "step": 698 }, { "epoch": 0.13, "learning_rate": 1.9432863523039986e-05, "loss": 0.4365, "step": 699 }, { "epoch": 0.13, "learning_rate": 1.9430792763419105e-05, "loss": 0.6608, "step": 700 }, { "epoch": 0.13, "learning_rate": 1.942871834098699e-05, "loss": 0.6238, "step": 701 }, { "epoch": 0.14, "learning_rate": 1.9426640256549313e-05, "loss": 0.5803, "step": 702 }, { "epoch": 0.14, "learning_rate": 1.9424558510913186e-05, "loss": 0.4646, "step": 703 }, { "epoch": 0.14, "learning_rate": 1.9422473104887133e-05, "loss": 0.6285, "step": 704 }, { "epoch": 0.14, "learning_rate": 1.9420384039281103e-05, "loss": 0.6361, "step": 705 }, { "epoch": 0.14, "learning_rate": 1.941829131490646e-05, "loss": 0.6583, "step": 706 }, { "epoch": 0.14, "learning_rate": 1.9416194932576e-05, "loss": 0.5842, "step": 707 }, { "epoch": 0.14, "learning_rate": 1.941409489310393e-05, "loss": 0.6151, "step": 708 }, { "epoch": 0.14, "learning_rate": 1.9411991197305878e-05, "loss": 0.5734, "step": 709 }, { "epoch": 0.14, "learning_rate": 1.9409883845998905e-05, "loss": 0.6294, "step": 710 }, { "epoch": 0.14, "learning_rate": 1.9407772840001473e-05, "loss": 0.6176, "step": 711 }, { "epoch": 0.14, "learning_rate": 1.9405658180133477e-05, "loss": 0.6312, "step": 712 }, { "epoch": 0.14, "learning_rate": 1.9403539867216226e-05, "loss": 0.5934, "step": 713 }, { "epoch": 0.14, "learning_rate": 1.9401417902072447e-05, "loss": 0.6296, "step": 714 }, { "epoch": 0.14, "learning_rate": 1.9399292285526286e-05, "loss": 0.4853, "step": 715 }, { "epoch": 0.14, "learning_rate": 1.939716301840331e-05, "loss": 0.4877, "step": 716 }, { "epoch": 0.14, "learning_rate": 1.9395030101530504e-05, "loss": 0.6256, "step": 717 }, { "epoch": 0.14, "learning_rate": 1.939289353573626e-05, "loss": 0.4756, "step": 718 }, { "epoch": 0.14, "learning_rate": 1.9390753321850404e-05, "loss": 0.6149, "step": 719 }, { "epoch": 0.14, "learning_rate": 1.938860946070417e-05, "loss": 0.5882, "step": 720 }, { "epoch": 0.14, "learning_rate": 1.93864619531302e-05, "loss": 0.6172, "step": 721 }, { "epoch": 0.14, "learning_rate": 1.9384310799962575e-05, "loss": 0.6584, "step": 722 }, { "epoch": 0.14, "learning_rate": 1.9382156002036764e-05, "loss": 0.6273, "step": 723 }, { "epoch": 0.14, "learning_rate": 1.9379997560189677e-05, "loss": 0.5135, "step": 724 }, { "epoch": 0.14, "learning_rate": 1.937783547525962e-05, "loss": 0.6272, "step": 725 }, { "epoch": 0.14, "learning_rate": 1.9375669748086326e-05, "loss": 0.482, "step": 726 }, { "epoch": 0.14, "learning_rate": 1.937350037951094e-05, "loss": 0.6136, "step": 727 }, { "epoch": 0.14, "learning_rate": 1.9371327370376018e-05, "loss": 0.6486, "step": 728 }, { "epoch": 0.14, "learning_rate": 1.936915072152553e-05, "loss": 0.4997, "step": 729 }, { "epoch": 0.14, "learning_rate": 1.936697043380486e-05, "loss": 0.6419, "step": 730 }, { "epoch": 0.14, "learning_rate": 1.936478650806081e-05, "loss": 0.5908, "step": 731 }, { "epoch": 0.14, "learning_rate": 1.936259894514159e-05, "loss": 0.5711, "step": 732 }, { "epoch": 0.14, "learning_rate": 1.9360407745896828e-05, "loss": 0.609, "step": 733 }, { "epoch": 0.14, "learning_rate": 1.9358212911177556e-05, "loss": 0.6603, "step": 734 }, { "epoch": 0.14, "learning_rate": 1.935601444183622e-05, "loss": 0.6704, "step": 735 }, { "epoch": 0.14, "learning_rate": 1.935381233872669e-05, "loss": 0.5975, "step": 736 }, { "epoch": 0.14, "learning_rate": 1.935160660270423e-05, "loss": 0.6558, "step": 737 }, { "epoch": 0.14, "learning_rate": 1.934939723462552e-05, "loss": 0.6132, "step": 738 }, { "epoch": 0.14, "learning_rate": 1.9347184235348663e-05, "loss": 0.6071, "step": 739 }, { "epoch": 0.14, "learning_rate": 1.9344967605733154e-05, "loss": 0.6531, "step": 740 }, { "epoch": 0.14, "learning_rate": 1.934274734663991e-05, "loss": 0.5949, "step": 741 }, { "epoch": 0.14, "learning_rate": 1.934052345893125e-05, "loss": 0.5939, "step": 742 }, { "epoch": 0.14, "learning_rate": 1.9338295943470915e-05, "loss": 0.5818, "step": 743 }, { "epoch": 0.14, "learning_rate": 1.9336064801124034e-05, "loss": 0.6256, "step": 744 }, { "epoch": 0.14, "learning_rate": 1.933383003275717e-05, "loss": 0.6196, "step": 745 }, { "epoch": 0.14, "learning_rate": 1.933159163923827e-05, "loss": 0.4667, "step": 746 }, { "epoch": 0.14, "learning_rate": 1.9329349621436708e-05, "loss": 0.6249, "step": 747 }, { "epoch": 0.14, "learning_rate": 1.9327103980223255e-05, "loss": 0.6114, "step": 748 }, { "epoch": 0.14, "learning_rate": 1.932485471647009e-05, "loss": 0.6213, "step": 749 }, { "epoch": 0.14, "learning_rate": 1.9322601831050804e-05, "loss": 0.5749, "step": 750 }, { "epoch": 0.14, "learning_rate": 1.9320345324840396e-05, "loss": 0.5437, "step": 751 }, { "epoch": 0.14, "learning_rate": 1.9318085198715257e-05, "loss": 0.4443, "step": 752 }, { "epoch": 0.14, "learning_rate": 1.93158214535532e-05, "loss": 0.5933, "step": 753 }, { "epoch": 0.15, "learning_rate": 1.9313554090233436e-05, "loss": 0.6155, "step": 754 }, { "epoch": 0.15, "learning_rate": 1.9311283109636586e-05, "loss": 0.6243, "step": 755 }, { "epoch": 0.15, "learning_rate": 1.9309008512644668e-05, "loss": 0.5643, "step": 756 }, { "epoch": 0.15, "learning_rate": 1.930673030014111e-05, "loss": 0.4994, "step": 757 }, { "epoch": 0.15, "learning_rate": 1.930444847301075e-05, "loss": 0.6055, "step": 758 }, { "epoch": 0.15, "learning_rate": 1.9302163032139813e-05, "loss": 0.6391, "step": 759 }, { "epoch": 0.15, "learning_rate": 1.9299873978415947e-05, "loss": 0.5795, "step": 760 }, { "epoch": 0.15, "learning_rate": 1.9297581312728187e-05, "loss": 0.6128, "step": 761 }, { "epoch": 0.15, "learning_rate": 1.929528503596698e-05, "loss": 0.5817, "step": 762 }, { "epoch": 0.15, "learning_rate": 1.929298514902418e-05, "loss": 0.4837, "step": 763 }, { "epoch": 0.15, "learning_rate": 1.929068165279303e-05, "loss": 0.5028, "step": 764 }, { "epoch": 0.15, "learning_rate": 1.928837454816818e-05, "loss": 0.5026, "step": 765 }, { "epoch": 0.15, "learning_rate": 1.9286063836045687e-05, "loss": 0.5168, "step": 766 }, { "epoch": 0.15, "learning_rate": 1.9283749517323e-05, "loss": 0.5966, "step": 767 }, { "epoch": 0.15, "learning_rate": 1.928143159289898e-05, "loss": 0.6108, "step": 768 }, { "epoch": 0.15, "learning_rate": 1.927911006367388e-05, "loss": 0.6094, "step": 769 }, { "epoch": 0.15, "learning_rate": 1.927678493054935e-05, "loss": 0.5891, "step": 770 }, { "epoch": 0.15, "learning_rate": 1.9274456194428454e-05, "loss": 0.5774, "step": 771 }, { "epoch": 0.15, "learning_rate": 1.9272123856215643e-05, "loss": 0.5005, "step": 772 }, { "epoch": 0.15, "learning_rate": 1.9269787916816764e-05, "loss": 0.5932, "step": 773 }, { "epoch": 0.15, "learning_rate": 1.9267448377139074e-05, "loss": 0.6176, "step": 774 }, { "epoch": 0.15, "learning_rate": 1.9265105238091227e-05, "loss": 0.5729, "step": 775 }, { "epoch": 0.15, "learning_rate": 1.9262758500583265e-05, "loss": 0.5993, "step": 776 }, { "epoch": 0.15, "learning_rate": 1.9260408165526638e-05, "loss": 0.5767, "step": 777 }, { "epoch": 0.15, "learning_rate": 1.9258054233834184e-05, "loss": 0.6426, "step": 778 }, { "epoch": 0.15, "learning_rate": 1.9255696706420147e-05, "loss": 0.531, "step": 779 }, { "epoch": 0.15, "learning_rate": 1.9253335584200165e-05, "loss": 0.5182, "step": 780 }, { "epoch": 0.15, "learning_rate": 1.9250970868091268e-05, "loss": 0.6413, "step": 781 }, { "epoch": 0.15, "learning_rate": 1.924860255901188e-05, "loss": 0.4984, "step": 782 }, { "epoch": 0.15, "learning_rate": 1.9246230657881834e-05, "loss": 0.6305, "step": 783 }, { "epoch": 0.15, "learning_rate": 1.9243855165622345e-05, "loss": 0.6033, "step": 784 }, { "epoch": 0.15, "learning_rate": 1.9241476083156026e-05, "loss": 0.5802, "step": 785 }, { "epoch": 0.15, "learning_rate": 1.9239093411406885e-05, "loss": 0.6134, "step": 786 }, { "epoch": 0.15, "learning_rate": 1.9236707151300326e-05, "loss": 0.5954, "step": 787 }, { "epoch": 0.15, "learning_rate": 1.9234317303763145e-05, "loss": 0.618, "step": 788 }, { "epoch": 0.15, "learning_rate": 1.9231923869723528e-05, "loss": 0.5941, "step": 789 }, { "epoch": 0.15, "learning_rate": 1.922952685011106e-05, "loss": 0.5575, "step": 790 }, { "epoch": 0.15, "learning_rate": 1.9227126245856716e-05, "loss": 0.5868, "step": 791 }, { "epoch": 0.15, "learning_rate": 1.922472205789286e-05, "loss": 0.6114, "step": 792 }, { "epoch": 0.15, "learning_rate": 1.9222314287153255e-05, "loss": 0.4945, "step": 793 }, { "epoch": 0.15, "learning_rate": 1.9219902934573048e-05, "loss": 0.5732, "step": 794 }, { "epoch": 0.15, "learning_rate": 1.9217488001088784e-05, "loss": 0.5746, "step": 795 }, { "epoch": 0.15, "learning_rate": 1.9215069487638396e-05, "loss": 0.5973, "step": 796 }, { "epoch": 0.15, "learning_rate": 1.92126473951612e-05, "loss": 0.5995, "step": 797 }, { "epoch": 0.15, "learning_rate": 1.921022172459791e-05, "loss": 0.6107, "step": 798 }, { "epoch": 0.15, "learning_rate": 1.920779247689064e-05, "loss": 0.5928, "step": 799 }, { "epoch": 0.15, "learning_rate": 1.9205359652982868e-05, "loss": 0.6103, "step": 800 }, { "epoch": 0.15, "learning_rate": 1.9202923253819482e-05, "loss": 0.5716, "step": 801 }, { "epoch": 0.15, "learning_rate": 1.920048328034675e-05, "loss": 0.6365, "step": 802 }, { "epoch": 0.15, "learning_rate": 1.9198039733512326e-05, "loss": 0.6206, "step": 803 }, { "epoch": 0.15, "learning_rate": 1.9195592614265262e-05, "loss": 0.5577, "step": 804 }, { "epoch": 0.15, "learning_rate": 1.9193141923555984e-05, "loss": 0.5112, "step": 805 }, { "epoch": 0.16, "learning_rate": 1.919068766233632e-05, "loss": 0.5051, "step": 806 }, { "epoch": 0.16, "learning_rate": 1.9188229831559468e-05, "loss": 0.4899, "step": 807 }, { "epoch": 0.16, "learning_rate": 1.9185768432180026e-05, "loss": 0.4938, "step": 808 }, { "epoch": 0.16, "learning_rate": 1.9183303465153972e-05, "loss": 0.6064, "step": 809 }, { "epoch": 0.16, "learning_rate": 1.9180834931438673e-05, "loss": 0.617, "step": 810 }, { "epoch": 0.16, "learning_rate": 1.917836283199288e-05, "loss": 0.6176, "step": 811 }, { "epoch": 0.16, "learning_rate": 1.917588716777672e-05, "loss": 0.6273, "step": 812 }, { "epoch": 0.16, "learning_rate": 1.917340793975172e-05, "loss": 0.6949, "step": 813 }, { "epoch": 0.16, "learning_rate": 1.917092514888078e-05, "loss": 0.609, "step": 814 }, { "epoch": 0.16, "learning_rate": 1.9168438796128193e-05, "loss": 0.6052, "step": 815 }, { "epoch": 0.16, "learning_rate": 1.9165948882459623e-05, "loss": 0.5856, "step": 816 }, { "epoch": 0.16, "learning_rate": 1.9163455408842123e-05, "loss": 0.624, "step": 817 }, { "epoch": 0.16, "learning_rate": 1.9160958376244138e-05, "loss": 0.6028, "step": 818 }, { "epoch": 0.16, "learning_rate": 1.9158457785635478e-05, "loss": 0.615, "step": 819 }, { "epoch": 0.16, "learning_rate": 1.915595363798735e-05, "loss": 0.6053, "step": 820 }, { "epoch": 0.16, "learning_rate": 1.915344593427233e-05, "loss": 0.651, "step": 821 }, { "epoch": 0.16, "learning_rate": 1.9150934675464384e-05, "loss": 0.6097, "step": 822 }, { "epoch": 0.16, "learning_rate": 1.9148419862538858e-05, "loss": 0.5057, "step": 823 }, { "epoch": 0.16, "learning_rate": 1.9145901496472474e-05, "loss": 0.5345, "step": 824 }, { "epoch": 0.16, "learning_rate": 1.9143379578243335e-05, "loss": 0.6237, "step": 825 }, { "epoch": 0.16, "learning_rate": 1.914085410883093e-05, "loss": 0.6118, "step": 826 }, { "epoch": 0.16, "learning_rate": 1.9138325089216118e-05, "loss": 0.48, "step": 827 }, { "epoch": 0.16, "learning_rate": 1.913579252038114e-05, "loss": 0.4791, "step": 828 }, { "epoch": 0.16, "learning_rate": 1.9133256403309627e-05, "loss": 0.6573, "step": 829 }, { "epoch": 0.16, "learning_rate": 1.913071673898656e-05, "loss": 0.4709, "step": 830 }, { "epoch": 0.16, "learning_rate": 1.912817352839833e-05, "loss": 0.6151, "step": 831 }, { "epoch": 0.16, "learning_rate": 1.9125626772532683e-05, "loss": 0.6443, "step": 832 }, { "epoch": 0.16, "learning_rate": 1.9123076472378753e-05, "loss": 0.6092, "step": 833 }, { "epoch": 0.16, "learning_rate": 1.9120522628927047e-05, "loss": 0.6602, "step": 834 }, { "epoch": 0.16, "learning_rate": 1.9117965243169444e-05, "loss": 0.5988, "step": 835 }, { "epoch": 0.16, "learning_rate": 1.9115404316099212e-05, "loss": 0.59, "step": 836 }, { "epoch": 0.16, "learning_rate": 1.9112839848710978e-05, "loss": 0.5136, "step": 837 }, { "epoch": 0.16, "learning_rate": 1.9110271842000755e-05, "loss": 0.4454, "step": 838 }, { "epoch": 0.16, "learning_rate": 1.9107700296965926e-05, "loss": 0.5907, "step": 839 }, { "epoch": 0.16, "learning_rate": 1.910512521460525e-05, "loss": 0.6368, "step": 840 }, { "epoch": 0.16, "learning_rate": 1.9102546595918857e-05, "loss": 0.6242, "step": 841 }, { "epoch": 0.16, "learning_rate": 1.9099964441908258e-05, "loss": 0.6345, "step": 842 }, { "epoch": 0.16, "learning_rate": 1.9097378753576327e-05, "loss": 0.3973, "step": 843 }, { "epoch": 0.16, "learning_rate": 1.9094789531927315e-05, "loss": 0.6371, "step": 844 }, { "epoch": 0.16, "learning_rate": 1.909219677796685e-05, "loss": 0.5126, "step": 845 }, { "epoch": 0.16, "learning_rate": 1.9089600492701926e-05, "loss": 0.6008, "step": 846 }, { "epoch": 0.16, "learning_rate": 1.908700067714091e-05, "loss": 0.5865, "step": 847 }, { "epoch": 0.16, "learning_rate": 1.9084397332293537e-05, "loss": 0.6266, "step": 848 }, { "epoch": 0.16, "learning_rate": 1.9081790459170926e-05, "loss": 0.6219, "step": 849 }, { "epoch": 0.16, "learning_rate": 1.9079180058785547e-05, "loss": 0.6166, "step": 850 }, { "epoch": 0.16, "learning_rate": 1.9076566132151255e-05, "loss": 0.5325, "step": 851 }, { "epoch": 0.16, "learning_rate": 1.907394868028326e-05, "loss": 0.6127, "step": 852 }, { "epoch": 0.16, "learning_rate": 1.9071327704198163e-05, "loss": 0.5849, "step": 853 }, { "epoch": 0.16, "learning_rate": 1.906870320491391e-05, "loss": 0.6496, "step": 854 }, { "epoch": 0.16, "learning_rate": 1.9066075183449835e-05, "loss": 0.4904, "step": 855 }, { "epoch": 0.16, "learning_rate": 1.9063443640826624e-05, "loss": 0.617, "step": 856 }, { "epoch": 0.16, "learning_rate": 1.906080857806634e-05, "loss": 0.6311, "step": 857 }, { "epoch": 0.17, "learning_rate": 1.905816999619242e-05, "loss": 0.6222, "step": 858 }, { "epoch": 0.17, "learning_rate": 1.9055527896229642e-05, "loss": 0.5991, "step": 859 }, { "epoch": 0.17, "learning_rate": 1.905288227920418e-05, "loss": 0.6122, "step": 860 }, { "epoch": 0.17, "learning_rate": 1.9050233146143554e-05, "loss": 0.6345, "step": 861 }, { "epoch": 0.17, "learning_rate": 1.9047580498076663e-05, "loss": 0.6096, "step": 862 }, { "epoch": 0.17, "learning_rate": 1.904492433603376e-05, "loss": 0.6033, "step": 863 }, { "epoch": 0.17, "learning_rate": 1.904226466104647e-05, "loss": 0.6251, "step": 864 }, { "epoch": 0.17, "learning_rate": 1.903960147414778e-05, "loss": 0.6169, "step": 865 }, { "epoch": 0.17, "learning_rate": 1.903693477637204e-05, "loss": 0.616, "step": 866 }, { "epoch": 0.17, "learning_rate": 1.9034264568754967e-05, "loss": 0.6023, "step": 867 }, { "epoch": 0.17, "learning_rate": 1.9031590852333637e-05, "loss": 0.3777, "step": 868 }, { "epoch": 0.17, "learning_rate": 1.9028913628146487e-05, "loss": 0.5841, "step": 869 }, { "epoch": 0.17, "learning_rate": 1.902623289723333e-05, "loss": 0.5974, "step": 870 }, { "epoch": 0.17, "learning_rate": 1.902354866063532e-05, "loss": 0.6127, "step": 871 }, { "epoch": 0.17, "learning_rate": 1.9020860919394992e-05, "loss": 0.645, "step": 872 }, { "epoch": 0.17, "learning_rate": 1.9018169674556228e-05, "loss": 0.6572, "step": 873 }, { "epoch": 0.17, "learning_rate": 1.901547492716428e-05, "loss": 0.6281, "step": 874 }, { "epoch": 0.17, "learning_rate": 1.9012776678265756e-05, "loss": 0.6117, "step": 875 }, { "epoch": 0.17, "learning_rate": 1.9010074928908624e-05, "loss": 0.6437, "step": 876 }, { "epoch": 0.17, "learning_rate": 1.900736968014221e-05, "loss": 0.6113, "step": 877 }, { "epoch": 0.17, "learning_rate": 1.9004660933017208e-05, "loss": 0.6205, "step": 878 }, { "epoch": 0.17, "learning_rate": 1.900194868858566e-05, "loss": 0.5913, "step": 879 }, { "epoch": 0.17, "learning_rate": 1.8999232947900968e-05, "loss": 0.5757, "step": 880 }, { "epoch": 0.17, "learning_rate": 1.89965137120179e-05, "loss": 0.5952, "step": 881 }, { "epoch": 0.17, "learning_rate": 1.899379098199257e-05, "loss": 0.5612, "step": 882 }, { "epoch": 0.17, "learning_rate": 1.899106475888246e-05, "loss": 0.6717, "step": 883 }, { "epoch": 0.17, "learning_rate": 1.89883350437464e-05, "loss": 0.6272, "step": 884 }, { "epoch": 0.17, "learning_rate": 1.8985601837644586e-05, "loss": 0.6362, "step": 885 }, { "epoch": 0.17, "learning_rate": 1.8982865141638557e-05, "loss": 0.6068, "step": 886 }, { "epoch": 0.17, "learning_rate": 1.8980124956791216e-05, "loss": 0.6344, "step": 887 }, { "epoch": 0.17, "learning_rate": 1.8977381284166818e-05, "loss": 0.3807, "step": 888 }, { "epoch": 0.17, "learning_rate": 1.897463412483098e-05, "loss": 0.556, "step": 889 }, { "epoch": 0.17, "learning_rate": 1.897188347985066e-05, "loss": 0.5902, "step": 890 }, { "epoch": 0.17, "learning_rate": 1.896912935029418e-05, "loss": 0.6487, "step": 891 }, { "epoch": 0.17, "learning_rate": 1.896637173723121e-05, "loss": 0.5013, "step": 892 }, { "epoch": 0.17, "learning_rate": 1.8963610641732777e-05, "loss": 0.5942, "step": 893 }, { "epoch": 0.17, "learning_rate": 1.8960846064871257e-05, "loss": 0.5964, "step": 894 }, { "epoch": 0.17, "learning_rate": 1.8958078007720387e-05, "loss": 0.5053, "step": 895 }, { "epoch": 0.17, "learning_rate": 1.895530647135524e-05, "loss": 0.5962, "step": 896 }, { "epoch": 0.17, "learning_rate": 1.8952531456852248e-05, "loss": 0.5292, "step": 897 }, { "epoch": 0.17, "learning_rate": 1.8949752965289197e-05, "loss": 0.573, "step": 898 }, { "epoch": 0.17, "learning_rate": 1.894697099774523e-05, "loss": 0.6428, "step": 899 }, { "epoch": 0.17, "learning_rate": 1.894418555530082e-05, "loss": 0.6034, "step": 900 }, { "epoch": 0.17, "learning_rate": 1.89413966390378e-05, "loss": 0.5969, "step": 901 }, { "epoch": 0.17, "learning_rate": 1.8938604250039362e-05, "loss": 0.6397, "step": 902 }, { "epoch": 0.17, "learning_rate": 1.8935808389390032e-05, "loss": 0.5663, "step": 903 }, { "epoch": 0.17, "learning_rate": 1.893300905817569e-05, "loss": 0.6192, "step": 904 }, { "epoch": 0.17, "learning_rate": 1.8930206257483566e-05, "loss": 0.6296, "step": 905 }, { "epoch": 0.17, "learning_rate": 1.8927399988402233e-05, "loss": 0.4469, "step": 906 }, { "epoch": 0.17, "learning_rate": 1.8924590252021614e-05, "loss": 0.6079, "step": 907 }, { "epoch": 0.17, "learning_rate": 1.8921777049432985e-05, "loss": 0.6296, "step": 908 }, { "epoch": 0.17, "learning_rate": 1.8918960381728947e-05, "loss": 0.6203, "step": 909 }, { "epoch": 0.18, "learning_rate": 1.8916140250003475e-05, "loss": 0.4735, "step": 910 }, { "epoch": 0.18, "learning_rate": 1.891331665535187e-05, "loss": 0.4725, "step": 911 }, { "epoch": 0.18, "learning_rate": 1.8910489598870784e-05, "loss": 0.6129, "step": 912 }, { "epoch": 0.18, "learning_rate": 1.8907659081658214e-05, "loss": 0.6343, "step": 913 }, { "epoch": 0.18, "learning_rate": 1.8904825104813497e-05, "loss": 0.4888, "step": 914 }, { "epoch": 0.18, "learning_rate": 1.8901987669437322e-05, "loss": 0.6146, "step": 915 }, { "epoch": 0.18, "learning_rate": 1.889914677663171e-05, "loss": 0.5636, "step": 916 }, { "epoch": 0.18, "learning_rate": 1.8896302427500042e-05, "loss": 0.6066, "step": 917 }, { "epoch": 0.18, "learning_rate": 1.8893454623147017e-05, "loss": 0.6292, "step": 918 }, { "epoch": 0.18, "learning_rate": 1.88906033646787e-05, "loss": 0.6048, "step": 919 }, { "epoch": 0.18, "learning_rate": 1.8887748653202478e-05, "loss": 0.4697, "step": 920 }, { "epoch": 0.18, "learning_rate": 1.8884890489827097e-05, "loss": 0.6233, "step": 921 }, { "epoch": 0.18, "learning_rate": 1.8882028875662627e-05, "loss": 0.6244, "step": 922 }, { "epoch": 0.18, "learning_rate": 1.8879163811820493e-05, "loss": 0.5139, "step": 923 }, { "epoch": 0.18, "learning_rate": 1.8876295299413445e-05, "loss": 0.6297, "step": 924 }, { "epoch": 0.18, "learning_rate": 1.8873423339555584e-05, "loss": 0.501, "step": 925 }, { "epoch": 0.18, "learning_rate": 1.8870547933362352e-05, "loss": 0.4981, "step": 926 }, { "epoch": 0.18, "learning_rate": 1.886766908195051e-05, "loss": 0.6119, "step": 927 }, { "epoch": 0.18, "learning_rate": 1.8864786786438187e-05, "loss": 0.6105, "step": 928 }, { "epoch": 0.18, "learning_rate": 1.886190104794482e-05, "loss": 0.5863, "step": 929 }, { "epoch": 0.18, "learning_rate": 1.8859011867591203e-05, "loss": 0.6269, "step": 930 }, { "epoch": 0.18, "learning_rate": 1.885611924649946e-05, "loss": 0.5992, "step": 931 }, { "epoch": 0.18, "learning_rate": 1.885322318579305e-05, "loss": 0.6303, "step": 932 }, { "epoch": 0.18, "learning_rate": 1.8850323686596766e-05, "loss": 0.6343, "step": 933 }, { "epoch": 0.18, "learning_rate": 1.8847420750036748e-05, "loss": 0.5828, "step": 934 }, { "epoch": 0.18, "learning_rate": 1.884451437724046e-05, "loss": 0.6338, "step": 935 }, { "epoch": 0.18, "learning_rate": 1.8841604569336702e-05, "loss": 0.51, "step": 936 }, { "epoch": 0.18, "learning_rate": 1.883869132745561e-05, "loss": 0.6349, "step": 937 }, { "epoch": 0.18, "learning_rate": 1.883577465272866e-05, "loss": 0.6372, "step": 938 }, { "epoch": 0.18, "learning_rate": 1.8832854546288642e-05, "loss": 0.5182, "step": 939 }, { "epoch": 0.18, "learning_rate": 1.8829931009269707e-05, "loss": 0.5919, "step": 940 }, { "epoch": 0.18, "learning_rate": 1.882700404280731e-05, "loss": 0.6097, "step": 941 }, { "epoch": 0.18, "learning_rate": 1.8824073648038258e-05, "loss": 0.6446, "step": 942 }, { "epoch": 0.18, "learning_rate": 1.882113982610068e-05, "loss": 0.384, "step": 943 }, { "epoch": 0.18, "learning_rate": 1.881820257813404e-05, "loss": 0.5831, "step": 944 }, { "epoch": 0.18, "learning_rate": 1.8815261905279133e-05, "loss": 0.6267, "step": 945 }, { "epoch": 0.18, "learning_rate": 1.8812317808678075e-05, "loss": 0.5901, "step": 946 }, { "epoch": 0.18, "learning_rate": 1.8809370289474327e-05, "loss": 0.3862, "step": 947 }, { "epoch": 0.18, "learning_rate": 1.8806419348812673e-05, "loss": 0.5717, "step": 948 }, { "epoch": 0.18, "learning_rate": 1.8803464987839217e-05, "loss": 0.6302, "step": 949 }, { "epoch": 0.18, "learning_rate": 1.88005072077014e-05, "loss": 0.636, "step": 950 }, { "epoch": 0.18, "learning_rate": 1.8797546009547996e-05, "loss": 0.6262, "step": 951 }, { "epoch": 0.18, "learning_rate": 1.879458139452909e-05, "loss": 0.5896, "step": 952 }, { "epoch": 0.18, "learning_rate": 1.8791613363796118e-05, "loss": 0.6227, "step": 953 }, { "epoch": 0.18, "learning_rate": 1.8788641918501817e-05, "loss": 0.3828, "step": 954 }, { "epoch": 0.18, "learning_rate": 1.8785667059800264e-05, "loss": 0.5907, "step": 955 }, { "epoch": 0.18, "learning_rate": 1.8782688788846865e-05, "loss": 0.6347, "step": 956 }, { "epoch": 0.18, "learning_rate": 1.877970710679834e-05, "loss": 0.5753, "step": 957 }, { "epoch": 0.18, "learning_rate": 1.877672201481275e-05, "loss": 0.6271, "step": 958 }, { "epoch": 0.18, "learning_rate": 1.877373351404946e-05, "loss": 0.605, "step": 959 }, { "epoch": 0.18, "learning_rate": 1.8770741605669173e-05, "loss": 0.6484, "step": 960 }, { "epoch": 0.18, "learning_rate": 1.876774629083391e-05, "loss": 0.6364, "step": 961 }, { "epoch": 0.19, "learning_rate": 1.8764747570707017e-05, "loss": 0.5981, "step": 962 }, { "epoch": 0.19, "learning_rate": 1.8761745446453167e-05, "loss": 0.641, "step": 963 }, { "epoch": 0.19, "learning_rate": 1.875873991923835e-05, "loss": 0.6159, "step": 964 }, { "epoch": 0.19, "learning_rate": 1.875573099022987e-05, "loss": 0.605, "step": 965 }, { "epoch": 0.19, "learning_rate": 1.8752718660596367e-05, "loss": 0.6173, "step": 966 }, { "epoch": 0.19, "learning_rate": 1.8749702931507797e-05, "loss": 0.5897, "step": 967 }, { "epoch": 0.19, "learning_rate": 1.874668380413543e-05, "loss": 0.6125, "step": 968 }, { "epoch": 0.19, "learning_rate": 1.8743661279651856e-05, "loss": 0.6285, "step": 969 }, { "epoch": 0.19, "learning_rate": 1.8740635359231e-05, "loss": 0.5088, "step": 970 }, { "epoch": 0.19, "learning_rate": 1.8737606044048086e-05, "loss": 0.6206, "step": 971 }, { "epoch": 0.19, "learning_rate": 1.873457333527967e-05, "loss": 0.5948, "step": 972 }, { "epoch": 0.19, "learning_rate": 1.873153723410362e-05, "loss": 0.6318, "step": 973 }, { "epoch": 0.19, "learning_rate": 1.8728497741699115e-05, "loss": 0.3984, "step": 974 }, { "epoch": 0.19, "learning_rate": 1.872545485924667e-05, "loss": 0.6291, "step": 975 }, { "epoch": 0.19, "learning_rate": 1.8722408587928104e-05, "loss": 0.479, "step": 976 }, { "epoch": 0.19, "learning_rate": 1.8719358928926546e-05, "loss": 0.6401, "step": 977 }, { "epoch": 0.19, "learning_rate": 1.8716305883426456e-05, "loss": 0.6012, "step": 978 }, { "epoch": 0.19, "learning_rate": 1.87132494526136e-05, "loss": 0.4983, "step": 979 }, { "epoch": 0.19, "learning_rate": 1.8710189637675055e-05, "loss": 0.5748, "step": 980 }, { "epoch": 0.19, "learning_rate": 1.8707126439799225e-05, "loss": 0.5956, "step": 981 }, { "epoch": 0.19, "learning_rate": 1.870405986017582e-05, "loss": 0.6095, "step": 982 }, { "epoch": 0.19, "learning_rate": 1.8700989899995857e-05, "loss": 0.5659, "step": 983 }, { "epoch": 0.19, "learning_rate": 1.8697916560451682e-05, "loss": 0.6461, "step": 984 }, { "epoch": 0.19, "learning_rate": 1.869483984273694e-05, "loss": 0.6293, "step": 985 }, { "epoch": 0.19, "learning_rate": 1.8691759748046594e-05, "loss": 0.6009, "step": 986 }, { "epoch": 0.19, "learning_rate": 1.8688676277576916e-05, "loss": 0.5271, "step": 987 }, { "epoch": 0.19, "learning_rate": 1.868558943252549e-05, "loss": 0.5324, "step": 988 }, { "epoch": 0.19, "learning_rate": 1.868249921409122e-05, "loss": 0.552, "step": 989 }, { "epoch": 0.19, "learning_rate": 1.8679405623474294e-05, "loss": 0.4921, "step": 990 }, { "epoch": 0.19, "learning_rate": 1.8676308661876242e-05, "loss": 0.5784, "step": 991 }, { "epoch": 0.19, "learning_rate": 1.8673208330499884e-05, "loss": 0.5109, "step": 992 }, { "epoch": 0.19, "learning_rate": 1.8670104630549348e-05, "loss": 0.6255, "step": 993 }, { "epoch": 0.19, "learning_rate": 1.866699756323008e-05, "loss": 0.605, "step": 994 }, { "epoch": 0.19, "learning_rate": 1.866388712974883e-05, "loss": 0.5924, "step": 995 }, { "epoch": 0.19, "learning_rate": 1.866077333131365e-05, "loss": 0.5779, "step": 996 }, { "epoch": 0.19, "learning_rate": 1.8657656169133908e-05, "loss": 0.6373, "step": 997 }, { "epoch": 0.19, "learning_rate": 1.8654535644420276e-05, "loss": 0.5981, "step": 998 }, { "epoch": 0.19, "learning_rate": 1.8651411758384718e-05, "loss": 0.6248, "step": 999 }, { "epoch": 0.19, "learning_rate": 1.8648284512240527e-05, "loss": 0.5633, "step": 1000 }, { "epoch": 0.19, "learning_rate": 1.8645153907202285e-05, "loss": 0.6179, "step": 1001 }, { "epoch": 0.19, "learning_rate": 1.8642019944485884e-05, "loss": 0.6131, "step": 1002 }, { "epoch": 0.19, "learning_rate": 1.863888262530852e-05, "loss": 0.5853, "step": 1003 }, { "epoch": 0.19, "learning_rate": 1.863574195088869e-05, "loss": 0.601, "step": 1004 }, { "epoch": 0.19, "learning_rate": 1.8632597922446195e-05, "loss": 0.5724, "step": 1005 }, { "epoch": 0.19, "learning_rate": 1.8629450541202142e-05, "loss": 0.6565, "step": 1006 }, { "epoch": 0.19, "learning_rate": 1.8626299808378933e-05, "loss": 0.5945, "step": 1007 }, { "epoch": 0.19, "learning_rate": 1.862314572520028e-05, "loss": 0.6091, "step": 1008 }, { "epoch": 0.19, "learning_rate": 1.861998829289119e-05, "loss": 0.6166, "step": 1009 }, { "epoch": 0.19, "learning_rate": 1.861682751267798e-05, "loss": 0.4609, "step": 1010 }, { "epoch": 0.19, "learning_rate": 1.861366338578825e-05, "loss": 0.5887, "step": 1011 }, { "epoch": 0.19, "learning_rate": 1.8610495913450922e-05, "loss": 0.5964, "step": 1012 }, { "epoch": 0.19, "learning_rate": 1.8607325096896197e-05, "loss": 0.4812, "step": 1013 }, { "epoch": 0.2, "learning_rate": 1.8604150937355588e-05, "loss": 0.6052, "step": 1014 }, { "epoch": 0.2, "learning_rate": 1.86009734360619e-05, "loss": 0.5254, "step": 1015 }, { "epoch": 0.2, "learning_rate": 1.8597792594249237e-05, "loss": 0.5944, "step": 1016 }, { "epoch": 0.2, "learning_rate": 1.8594608413153e-05, "loss": 0.6203, "step": 1017 }, { "epoch": 0.2, "learning_rate": 1.8591420894009897e-05, "loss": 0.5403, "step": 1018 }, { "epoch": 0.2, "learning_rate": 1.8588230038057913e-05, "loss": 0.5871, "step": 1019 }, { "epoch": 0.2, "learning_rate": 1.8585035846536347e-05, "loss": 0.462, "step": 1020 }, { "epoch": 0.2, "learning_rate": 1.8581838320685782e-05, "loss": 0.5752, "step": 1021 }, { "epoch": 0.2, "learning_rate": 1.8578637461748105e-05, "loss": 0.6053, "step": 1022 }, { "epoch": 0.2, "learning_rate": 1.857543327096649e-05, "loss": 0.612, "step": 1023 }, { "epoch": 0.2, "learning_rate": 1.85722257495854e-05, "loss": 0.587, "step": 1024 }, { "epoch": 0.2, "learning_rate": 1.856901489885061e-05, "loss": 0.5763, "step": 1025 }, { "epoch": 0.2, "learning_rate": 1.856580072000918e-05, "loss": 0.626, "step": 1026 }, { "epoch": 0.2, "learning_rate": 1.8562583214309447e-05, "loss": 0.5283, "step": 1027 }, { "epoch": 0.2, "learning_rate": 1.855936238300106e-05, "loss": 0.6448, "step": 1028 }, { "epoch": 0.2, "learning_rate": 1.8556138227334957e-05, "loss": 0.6084, "step": 1029 }, { "epoch": 0.2, "learning_rate": 1.855291074856336e-05, "loss": 0.6427, "step": 1030 }, { "epoch": 0.2, "learning_rate": 1.8549679947939778e-05, "loss": 0.6143, "step": 1031 }, { "epoch": 0.2, "learning_rate": 1.8546445826719023e-05, "loss": 0.6284, "step": 1032 }, { "epoch": 0.2, "learning_rate": 1.8543208386157195e-05, "loss": 0.5812, "step": 1033 }, { "epoch": 0.2, "learning_rate": 1.853996762751167e-05, "loss": 0.5964, "step": 1034 }, { "epoch": 0.2, "learning_rate": 1.8536723552041124e-05, "loss": 0.5479, "step": 1035 }, { "epoch": 0.2, "learning_rate": 1.853347616100552e-05, "loss": 0.6162, "step": 1036 }, { "epoch": 0.2, "learning_rate": 1.8530225455666103e-05, "loss": 0.5764, "step": 1037 }, { "epoch": 0.2, "learning_rate": 1.8526971437285416e-05, "loss": 0.6142, "step": 1038 }, { "epoch": 0.2, "learning_rate": 1.8523714107127278e-05, "loss": 0.5648, "step": 1039 }, { "epoch": 0.2, "learning_rate": 1.8520453466456797e-05, "loss": 0.6092, "step": 1040 }, { "epoch": 0.2, "learning_rate": 1.8517189516540376e-05, "loss": 0.6338, "step": 1041 }, { "epoch": 0.2, "learning_rate": 1.8513922258645687e-05, "loss": 0.5615, "step": 1042 }, { "epoch": 0.2, "learning_rate": 1.8510651694041702e-05, "loss": 0.5913, "step": 1043 }, { "epoch": 0.2, "learning_rate": 1.8507377823998664e-05, "loss": 0.591, "step": 1044 }, { "epoch": 0.2, "learning_rate": 1.850410064978811e-05, "loss": 0.6029, "step": 1045 }, { "epoch": 0.2, "learning_rate": 1.8500820172682858e-05, "loss": 0.487, "step": 1046 }, { "epoch": 0.2, "learning_rate": 1.8497536393957005e-05, "loss": 0.5564, "step": 1047 }, { "epoch": 0.2, "learning_rate": 1.8494249314885932e-05, "loss": 0.6125, "step": 1048 }, { "epoch": 0.2, "learning_rate": 1.8490958936746304e-05, "loss": 0.6283, "step": 1049 }, { "epoch": 0.2, "learning_rate": 1.848766526081607e-05, "loss": 0.5961, "step": 1050 }, { "epoch": 0.2, "learning_rate": 1.8484368288374452e-05, "loss": 0.6038, "step": 1051 }, { "epoch": 0.2, "learning_rate": 1.8481068020701954e-05, "loss": 0.582, "step": 1052 }, { "epoch": 0.2, "learning_rate": 1.8477764459080364e-05, "loss": 0.6137, "step": 1053 }, { "epoch": 0.2, "learning_rate": 1.8474457604792746e-05, "loss": 0.606, "step": 1054 }, { "epoch": 0.2, "learning_rate": 1.8471147459123447e-05, "loss": 0.6441, "step": 1055 }, { "epoch": 0.2, "learning_rate": 1.8467834023358088e-05, "loss": 0.5873, "step": 1056 }, { "epoch": 0.2, "learning_rate": 1.846451729878357e-05, "loss": 0.5815, "step": 1057 }, { "epoch": 0.2, "learning_rate": 1.846119728668807e-05, "loss": 0.6006, "step": 1058 }, { "epoch": 0.2, "learning_rate": 1.845787398836104e-05, "loss": 0.5105, "step": 1059 }, { "epoch": 0.2, "learning_rate": 1.8454547405093212e-05, "loss": 0.5709, "step": 1060 }, { "epoch": 0.2, "learning_rate": 1.8451217538176597e-05, "loss": 0.5897, "step": 1061 }, { "epoch": 0.2, "learning_rate": 1.844788438890447e-05, "loss": 0.4907, "step": 1062 }, { "epoch": 0.2, "learning_rate": 1.8444547958571396e-05, "loss": 0.6078, "step": 1063 }, { "epoch": 0.2, "learning_rate": 1.84412082484732e-05, "loss": 0.6204, "step": 1064 }, { "epoch": 0.2, "learning_rate": 1.8437865259906987e-05, "loss": 0.5526, "step": 1065 }, { "epoch": 0.21, "learning_rate": 1.8434518994171136e-05, "loss": 0.6074, "step": 1066 }, { "epoch": 0.21, "learning_rate": 1.84311694525653e-05, "loss": 0.5783, "step": 1067 }, { "epoch": 0.21, "learning_rate": 1.84278166363904e-05, "loss": 0.5879, "step": 1068 }, { "epoch": 0.21, "learning_rate": 1.8424460546948632e-05, "loss": 0.5917, "step": 1069 }, { "epoch": 0.21, "learning_rate": 1.8421101185543463e-05, "loss": 0.4708, "step": 1070 }, { "epoch": 0.21, "learning_rate": 1.841773855347963e-05, "loss": 0.6002, "step": 1071 }, { "epoch": 0.21, "learning_rate": 1.841437265206314e-05, "loss": 0.6262, "step": 1072 }, { "epoch": 0.21, "learning_rate": 1.841100348260127e-05, "loss": 0.6314, "step": 1073 }, { "epoch": 0.21, "learning_rate": 1.840763104640257e-05, "loss": 0.5686, "step": 1074 }, { "epoch": 0.21, "learning_rate": 1.8404255344776853e-05, "loss": 0.5092, "step": 1075 }, { "epoch": 0.21, "learning_rate": 1.84008763790352e-05, "loss": 0.6319, "step": 1076 }, { "epoch": 0.21, "learning_rate": 1.8397494150489963e-05, "loss": 0.6186, "step": 1077 }, { "epoch": 0.21, "learning_rate": 1.8394108660454766e-05, "loss": 0.6295, "step": 1078 }, { "epoch": 0.21, "learning_rate": 1.8390719910244487e-05, "loss": 0.5211, "step": 1079 }, { "epoch": 0.21, "learning_rate": 1.8387327901175286e-05, "loss": 0.5773, "step": 1080 }, { "epoch": 0.21, "learning_rate": 1.838393263456457e-05, "loss": 0.5005, "step": 1081 }, { "epoch": 0.21, "learning_rate": 1.838053411173103e-05, "loss": 0.5855, "step": 1082 }, { "epoch": 0.21, "learning_rate": 1.8377132333994606e-05, "loss": 0.5799, "step": 1083 }, { "epoch": 0.21, "learning_rate": 1.837372730267652e-05, "loss": 0.5628, "step": 1084 }, { "epoch": 0.21, "learning_rate": 1.8370319019099236e-05, "loss": 0.6203, "step": 1085 }, { "epoch": 0.21, "learning_rate": 1.8366907484586497e-05, "loss": 0.5873, "step": 1086 }, { "epoch": 0.21, "learning_rate": 1.83634927004633e-05, "loss": 0.5086, "step": 1087 }, { "epoch": 0.21, "learning_rate": 1.8360074668055915e-05, "loss": 0.6231, "step": 1088 }, { "epoch": 0.21, "learning_rate": 1.8356653388691857e-05, "loss": 0.574, "step": 1089 }, { "epoch": 0.21, "learning_rate": 1.8353228863699922e-05, "loss": 0.5898, "step": 1090 }, { "epoch": 0.21, "learning_rate": 1.8349801094410148e-05, "loss": 0.5734, "step": 1091 }, { "epoch": 0.21, "learning_rate": 1.8346370082153843e-05, "loss": 0.6089, "step": 1092 }, { "epoch": 0.21, "learning_rate": 1.8342935828263574e-05, "loss": 0.403, "step": 1093 }, { "epoch": 0.21, "learning_rate": 1.8339498334073166e-05, "loss": 0.6322, "step": 1094 }, { "epoch": 0.21, "learning_rate": 1.83360576009177e-05, "loss": 0.6149, "step": 1095 }, { "epoch": 0.21, "learning_rate": 1.833261363013352e-05, "loss": 0.611, "step": 1096 }, { "epoch": 0.21, "learning_rate": 1.832916642305822e-05, "loss": 0.6251, "step": 1097 }, { "epoch": 0.21, "learning_rate": 1.832571598103066e-05, "loss": 0.5996, "step": 1098 }, { "epoch": 0.21, "learning_rate": 1.8322262305390948e-05, "loss": 0.4989, "step": 1099 }, { "epoch": 0.21, "learning_rate": 1.8318805397480455e-05, "loss": 0.6285, "step": 1100 }, { "epoch": 0.21, "learning_rate": 1.8315345258641802e-05, "loss": 0.5632, "step": 1101 }, { "epoch": 0.21, "learning_rate": 1.8311881890218873e-05, "loss": 0.589, "step": 1102 }, { "epoch": 0.21, "learning_rate": 1.830841529355679e-05, "loss": 0.6151, "step": 1103 }, { "epoch": 0.21, "learning_rate": 1.8304945470001948e-05, "loss": 0.5759, "step": 1104 }, { "epoch": 0.21, "learning_rate": 1.8301472420901985e-05, "loss": 0.5952, "step": 1105 }, { "epoch": 0.21, "learning_rate": 1.8297996147605787e-05, "loss": 0.6141, "step": 1106 }, { "epoch": 0.21, "learning_rate": 1.829451665146351e-05, "loss": 0.6456, "step": 1107 }, { "epoch": 0.21, "learning_rate": 1.8291033933826535e-05, "loss": 0.6345, "step": 1108 }, { "epoch": 0.21, "learning_rate": 1.8287547996047523e-05, "loss": 0.6018, "step": 1109 }, { "epoch": 0.21, "learning_rate": 1.8284058839480363e-05, "loss": 0.6349, "step": 1110 }, { "epoch": 0.21, "learning_rate": 1.8280566465480206e-05, "loss": 0.4168, "step": 1111 }, { "epoch": 0.21, "learning_rate": 1.8277070875403455e-05, "loss": 0.5082, "step": 1112 }, { "epoch": 0.21, "learning_rate": 1.8273572070607756e-05, "loss": 0.5869, "step": 1113 }, { "epoch": 0.21, "learning_rate": 1.8270070052451995e-05, "loss": 0.4878, "step": 1114 }, { "epoch": 0.21, "learning_rate": 1.8266564822296323e-05, "loss": 0.6082, "step": 1115 }, { "epoch": 0.21, "learning_rate": 1.826305638150213e-05, "loss": 0.396, "step": 1116 }, { "epoch": 0.21, "learning_rate": 1.825954473143205e-05, "loss": 0.6525, "step": 1117 }, { "epoch": 0.22, "learning_rate": 1.8256029873449976e-05, "loss": 0.5285, "step": 1118 }, { "epoch": 0.22, "learning_rate": 1.825251180892103e-05, "loss": 0.6072, "step": 1119 }, { "epoch": 0.22, "learning_rate": 1.8248990539211596e-05, "loss": 0.6111, "step": 1120 }, { "epoch": 0.22, "learning_rate": 1.8245466065689282e-05, "loss": 0.6174, "step": 1121 }, { "epoch": 0.22, "learning_rate": 1.824193838972297e-05, "loss": 0.6152, "step": 1122 }, { "epoch": 0.22, "learning_rate": 1.823840751268275e-05, "loss": 0.5755, "step": 1123 }, { "epoch": 0.22, "learning_rate": 1.8234873435939987e-05, "loss": 0.5717, "step": 1124 }, { "epoch": 0.22, "learning_rate": 1.8231336160867275e-05, "loss": 0.6416, "step": 1125 }, { "epoch": 0.22, "learning_rate": 1.8227795688838446e-05, "loss": 0.6412, "step": 1126 }, { "epoch": 0.22, "learning_rate": 1.822425202122858e-05, "loss": 0.5966, "step": 1127 }, { "epoch": 0.22, "learning_rate": 1.8220705159413996e-05, "loss": 0.6008, "step": 1128 }, { "epoch": 0.22, "learning_rate": 1.8217155104772256e-05, "loss": 0.592, "step": 1129 }, { "epoch": 0.22, "learning_rate": 1.8213601858682158e-05, "loss": 0.6129, "step": 1130 }, { "epoch": 0.22, "learning_rate": 1.8210045422523744e-05, "loss": 0.5755, "step": 1131 }, { "epoch": 0.22, "learning_rate": 1.8206485797678294e-05, "loss": 0.6107, "step": 1132 }, { "epoch": 0.22, "learning_rate": 1.820292298552832e-05, "loss": 0.6109, "step": 1133 }, { "epoch": 0.22, "learning_rate": 1.819935698745759e-05, "loss": 0.6098, "step": 1134 }, { "epoch": 0.22, "learning_rate": 1.8195787804851076e-05, "loss": 0.627, "step": 1135 }, { "epoch": 0.22, "learning_rate": 1.8192215439095025e-05, "loss": 0.5856, "step": 1136 }, { "epoch": 0.22, "learning_rate": 1.8188639891576893e-05, "loss": 0.6042, "step": 1137 }, { "epoch": 0.22, "learning_rate": 1.8185061163685386e-05, "loss": 0.6252, "step": 1138 }, { "epoch": 0.22, "learning_rate": 1.818147925681044e-05, "loss": 0.6078, "step": 1139 }, { "epoch": 0.22, "learning_rate": 1.8177894172343227e-05, "loss": 0.6128, "step": 1140 }, { "epoch": 0.22, "learning_rate": 1.817430591167615e-05, "loss": 0.5775, "step": 1141 }, { "epoch": 0.22, "learning_rate": 1.8170714476202848e-05, "loss": 0.4895, "step": 1142 }, { "epoch": 0.22, "learning_rate": 1.8167119867318197e-05, "loss": 0.5875, "step": 1143 }, { "epoch": 0.22, "learning_rate": 1.81635220864183e-05, "loss": 0.5885, "step": 1144 }, { "epoch": 0.22, "learning_rate": 1.8159921134900486e-05, "loss": 0.5101, "step": 1145 }, { "epoch": 0.22, "learning_rate": 1.8156317014163337e-05, "loss": 0.5975, "step": 1146 }, { "epoch": 0.22, "learning_rate": 1.8152709725606642e-05, "loss": 0.6096, "step": 1147 }, { "epoch": 0.22, "learning_rate": 1.8149099270631434e-05, "loss": 0.6449, "step": 1148 }, { "epoch": 0.22, "learning_rate": 1.8145485650639973e-05, "loss": 0.5831, "step": 1149 }, { "epoch": 0.22, "learning_rate": 1.8141868867035745e-05, "loss": 0.598, "step": 1150 }, { "epoch": 0.22, "learning_rate": 1.8138248921223465e-05, "loss": 0.6287, "step": 1151 }, { "epoch": 0.22, "learning_rate": 1.8134625814609084e-05, "loss": 0.5381, "step": 1152 }, { "epoch": 0.22, "learning_rate": 1.8130999548599767e-05, "loss": 0.5873, "step": 1153 }, { "epoch": 0.22, "learning_rate": 1.8127370124603927e-05, "loss": 0.6196, "step": 1154 }, { "epoch": 0.22, "learning_rate": 1.8123737544031178e-05, "loss": 0.5802, "step": 1155 }, { "epoch": 0.22, "learning_rate": 1.8120101808292373e-05, "loss": 0.6262, "step": 1156 }, { "epoch": 0.22, "learning_rate": 1.81164629187996e-05, "loss": 0.5678, "step": 1157 }, { "epoch": 0.22, "learning_rate": 1.811282087696615e-05, "loss": 0.6204, "step": 1158 }, { "epoch": 0.22, "learning_rate": 1.8109175684206558e-05, "loss": 0.61, "step": 1159 }, { "epoch": 0.22, "learning_rate": 1.8105527341936574e-05, "loss": 0.5706, "step": 1160 }, { "epoch": 0.22, "learning_rate": 1.810187585157317e-05, "loss": 0.5505, "step": 1161 }, { "epoch": 0.22, "learning_rate": 1.8098221214534543e-05, "loss": 0.6232, "step": 1162 }, { "epoch": 0.22, "learning_rate": 1.8094563432240107e-05, "loss": 0.6546, "step": 1163 }, { "epoch": 0.22, "learning_rate": 1.8090902506110513e-05, "loss": 0.6451, "step": 1164 }, { "epoch": 0.22, "learning_rate": 1.8087238437567614e-05, "loss": 0.5076, "step": 1165 }, { "epoch": 0.22, "learning_rate": 1.8083571228034498e-05, "loss": 0.6508, "step": 1166 }, { "epoch": 0.22, "learning_rate": 1.807990087893546e-05, "loss": 0.6126, "step": 1167 }, { "epoch": 0.22, "learning_rate": 1.807622739169603e-05, "loss": 0.62, "step": 1168 }, { "epoch": 0.22, "learning_rate": 1.807255076774294e-05, "loss": 0.5911, "step": 1169 }, { "epoch": 0.23, "learning_rate": 1.8068871008504153e-05, "loss": 0.5634, "step": 1170 }, { "epoch": 0.23, "learning_rate": 1.8065188115408844e-05, "loss": 0.6097, "step": 1171 }, { "epoch": 0.23, "learning_rate": 1.8061502089887406e-05, "loss": 0.6297, "step": 1172 }, { "epoch": 0.23, "learning_rate": 1.805781293337145e-05, "loss": 0.3849, "step": 1173 }, { "epoch": 0.23, "learning_rate": 1.8054120647293798e-05, "loss": 0.6117, "step": 1174 }, { "epoch": 0.23, "learning_rate": 1.8050425233088496e-05, "loss": 0.4955, "step": 1175 }, { "epoch": 0.23, "learning_rate": 1.80467266921908e-05, "loss": 0.6098, "step": 1176 }, { "epoch": 0.23, "learning_rate": 1.8043025026037178e-05, "loss": 0.6192, "step": 1177 }, { "epoch": 0.23, "learning_rate": 1.8039320236065314e-05, "loss": 0.5928, "step": 1178 }, { "epoch": 0.23, "learning_rate": 1.803561232371411e-05, "loss": 0.581, "step": 1179 }, { "epoch": 0.23, "learning_rate": 1.803190129042367e-05, "loss": 0.6068, "step": 1180 }, { "epoch": 0.23, "learning_rate": 1.8028187137635325e-05, "loss": 0.5904, "step": 1181 }, { "epoch": 0.23, "learning_rate": 1.8024469866791602e-05, "loss": 0.5154, "step": 1182 }, { "epoch": 0.23, "learning_rate": 1.802074947933625e-05, "loss": 0.5843, "step": 1183 }, { "epoch": 0.23, "learning_rate": 1.801702597671422e-05, "loss": 0.6134, "step": 1184 }, { "epoch": 0.23, "learning_rate": 1.8013299360371685e-05, "loss": 0.5833, "step": 1185 }, { "epoch": 0.23, "learning_rate": 1.8009569631756013e-05, "loss": 0.6377, "step": 1186 }, { "epoch": 0.23, "learning_rate": 1.8005836792315793e-05, "loss": 0.5994, "step": 1187 }, { "epoch": 0.23, "learning_rate": 1.800210084350081e-05, "loss": 0.6086, "step": 1188 }, { "epoch": 0.23, "learning_rate": 1.799836178676207e-05, "loss": 0.5501, "step": 1189 }, { "epoch": 0.23, "learning_rate": 1.799461962355178e-05, "loss": 0.5769, "step": 1190 }, { "epoch": 0.23, "learning_rate": 1.7990874355323345e-05, "loss": 0.602, "step": 1191 }, { "epoch": 0.23, "learning_rate": 1.7987125983531393e-05, "loss": 0.506, "step": 1192 }, { "epoch": 0.23, "learning_rate": 1.7983374509631742e-05, "loss": 0.6282, "step": 1193 }, { "epoch": 0.23, "learning_rate": 1.7979619935081424e-05, "loss": 0.5136, "step": 1194 }, { "epoch": 0.23, "learning_rate": 1.797586226133867e-05, "loss": 0.6271, "step": 1195 }, { "epoch": 0.23, "learning_rate": 1.7972101489862924e-05, "loss": 0.5292, "step": 1196 }, { "epoch": 0.23, "learning_rate": 1.7968337622114824e-05, "loss": 0.5994, "step": 1197 }, { "epoch": 0.23, "learning_rate": 1.7964570659556206e-05, "loss": 0.6097, "step": 1198 }, { "epoch": 0.23, "learning_rate": 1.796080060365012e-05, "loss": 0.4786, "step": 1199 }, { "epoch": 0.23, "learning_rate": 1.7957027455860815e-05, "loss": 0.5971, "step": 1200 }, { "epoch": 0.23, "learning_rate": 1.795325121765373e-05, "loss": 0.6288, "step": 1201 }, { "epoch": 0.23, "learning_rate": 1.794947189049552e-05, "loss": 0.6583, "step": 1202 }, { "epoch": 0.23, "learning_rate": 1.7945689475854033e-05, "loss": 0.6229, "step": 1203 }, { "epoch": 0.23, "learning_rate": 1.7941903975198305e-05, "loss": 0.5069, "step": 1204 }, { "epoch": 0.23, "learning_rate": 1.7938115389998595e-05, "loss": 0.5679, "step": 1205 }, { "epoch": 0.23, "learning_rate": 1.7934323721726334e-05, "loss": 0.5967, "step": 1206 }, { "epoch": 0.23, "learning_rate": 1.7930528971854166e-05, "loss": 0.6393, "step": 1207 }, { "epoch": 0.23, "learning_rate": 1.792673114185593e-05, "loss": 0.468, "step": 1208 }, { "epoch": 0.23, "learning_rate": 1.7922930233206656e-05, "loss": 0.643, "step": 1209 }, { "epoch": 0.23, "learning_rate": 1.7919126247382576e-05, "loss": 0.6183, "step": 1210 }, { "epoch": 0.23, "learning_rate": 1.791531918586112e-05, "loss": 0.6069, "step": 1211 }, { "epoch": 0.23, "learning_rate": 1.7911509050120892e-05, "loss": 0.5753, "step": 1212 }, { "epoch": 0.23, "learning_rate": 1.7907695841641716e-05, "loss": 0.5099, "step": 1213 }, { "epoch": 0.23, "learning_rate": 1.7903879561904597e-05, "loss": 0.6366, "step": 1214 }, { "epoch": 0.23, "learning_rate": 1.790006021239173e-05, "loss": 0.6257, "step": 1215 }, { "epoch": 0.23, "learning_rate": 1.789623779458651e-05, "loss": 0.6633, "step": 1216 }, { "epoch": 0.23, "learning_rate": 1.789241230997352e-05, "loss": 0.576, "step": 1217 }, { "epoch": 0.23, "learning_rate": 1.7888583760038534e-05, "loss": 0.6154, "step": 1218 }, { "epoch": 0.23, "learning_rate": 1.7884752146268513e-05, "loss": 0.6087, "step": 1219 }, { "epoch": 0.23, "learning_rate": 1.7880917470151614e-05, "loss": 0.493, "step": 1220 }, { "epoch": 0.23, "learning_rate": 1.7877079733177185e-05, "loss": 0.6189, "step": 1221 }, { "epoch": 0.24, "learning_rate": 1.7873238936835754e-05, "loss": 0.5956, "step": 1222 }, { "epoch": 0.24, "learning_rate": 1.786939508261904e-05, "loss": 0.6039, "step": 1223 }, { "epoch": 0.24, "learning_rate": 1.786554817201996e-05, "loss": 0.6285, "step": 1224 }, { "epoch": 0.24, "learning_rate": 1.78616982065326e-05, "loss": 0.6226, "step": 1225 }, { "epoch": 0.24, "learning_rate": 1.785784518765225e-05, "loss": 0.6022, "step": 1226 }, { "epoch": 0.24, "learning_rate": 1.7853989116875373e-05, "loss": 0.6466, "step": 1227 }, { "epoch": 0.24, "learning_rate": 1.7850129995699626e-05, "loss": 0.4938, "step": 1228 }, { "epoch": 0.24, "learning_rate": 1.7846267825623843e-05, "loss": 0.6278, "step": 1229 }, { "epoch": 0.24, "learning_rate": 1.7842402608148053e-05, "loss": 0.6075, "step": 1230 }, { "epoch": 0.24, "learning_rate": 1.7838534344773453e-05, "loss": 0.6334, "step": 1231 }, { "epoch": 0.24, "learning_rate": 1.7834663037002444e-05, "loss": 0.6154, "step": 1232 }, { "epoch": 0.24, "learning_rate": 1.7830788686338586e-05, "loss": 0.5594, "step": 1233 }, { "epoch": 0.24, "learning_rate": 1.7826911294286636e-05, "loss": 0.6198, "step": 1234 }, { "epoch": 0.24, "learning_rate": 1.782303086235253e-05, "loss": 0.6099, "step": 1235 }, { "epoch": 0.24, "learning_rate": 1.781914739204338e-05, "loss": 0.5989, "step": 1236 }, { "epoch": 0.24, "learning_rate": 1.7815260884867486e-05, "loss": 0.642, "step": 1237 }, { "epoch": 0.24, "learning_rate": 1.781137134233432e-05, "loss": 0.5874, "step": 1238 }, { "epoch": 0.24, "learning_rate": 1.7807478765954532e-05, "loss": 0.5717, "step": 1239 }, { "epoch": 0.24, "learning_rate": 1.7803583157239958e-05, "loss": 0.4234, "step": 1240 }, { "epoch": 0.24, "learning_rate": 1.7799684517703605e-05, "loss": 0.5837, "step": 1241 }, { "epoch": 0.24, "learning_rate": 1.779578284885966e-05, "loss": 0.6271, "step": 1242 }, { "epoch": 0.24, "learning_rate": 1.779187815222349e-05, "loss": 0.5993, "step": 1243 }, { "epoch": 0.24, "learning_rate": 1.778797042931163e-05, "loss": 0.6107, "step": 1244 }, { "epoch": 0.24, "learning_rate": 1.7784059681641798e-05, "loss": 0.6167, "step": 1245 }, { "epoch": 0.24, "learning_rate": 1.778014591073288e-05, "loss": 0.5803, "step": 1246 }, { "epoch": 0.24, "learning_rate": 1.777622911810494e-05, "loss": 0.5001, "step": 1247 }, { "epoch": 0.24, "learning_rate": 1.777230930527922e-05, "loss": 0.6094, "step": 1248 }, { "epoch": 0.24, "learning_rate": 1.7768386473778124e-05, "loss": 0.5135, "step": 1249 }, { "epoch": 0.24, "learning_rate": 1.7764460625125236e-05, "loss": 0.6059, "step": 1250 }, { "epoch": 0.24, "learning_rate": 1.776053176084531e-05, "loss": 0.5902, "step": 1251 }, { "epoch": 0.24, "learning_rate": 1.7756599882464274e-05, "loss": 0.4779, "step": 1252 }, { "epoch": 0.24, "learning_rate": 1.7752664991509224e-05, "loss": 0.5059, "step": 1253 }, { "epoch": 0.24, "learning_rate": 1.7748727089508423e-05, "loss": 0.576, "step": 1254 }, { "epoch": 0.24, "learning_rate": 1.7744786177991307e-05, "loss": 0.5612, "step": 1255 }, { "epoch": 0.24, "learning_rate": 1.774084225848849e-05, "loss": 0.6164, "step": 1256 }, { "epoch": 0.24, "learning_rate": 1.773689533253173e-05, "loss": 0.4862, "step": 1257 }, { "epoch": 0.24, "learning_rate": 1.7732945401653978e-05, "loss": 0.4726, "step": 1258 }, { "epoch": 0.24, "learning_rate": 1.7728992467389342e-05, "loss": 0.4859, "step": 1259 }, { "epoch": 0.24, "learning_rate": 1.7725036531273087e-05, "loss": 0.6178, "step": 1260 }, { "epoch": 0.24, "learning_rate": 1.7721077594841663e-05, "loss": 0.6089, "step": 1261 }, { "epoch": 0.24, "learning_rate": 1.771711565963267e-05, "loss": 0.475, "step": 1262 }, { "epoch": 0.24, "learning_rate": 1.7713150727184878e-05, "loss": 0.6435, "step": 1263 }, { "epoch": 0.24, "learning_rate": 1.770918279903822e-05, "loss": 0.6088, "step": 1264 }, { "epoch": 0.24, "learning_rate": 1.77052118767338e-05, "loss": 0.5059, "step": 1265 }, { "epoch": 0.24, "learning_rate": 1.7701237961813874e-05, "loss": 0.4651, "step": 1266 }, { "epoch": 0.24, "learning_rate": 1.7697261055821864e-05, "loss": 0.5999, "step": 1267 }, { "epoch": 0.24, "learning_rate": 1.7693281160302354e-05, "loss": 0.6253, "step": 1268 }, { "epoch": 0.24, "learning_rate": 1.7689298276801095e-05, "loss": 0.6076, "step": 1269 }, { "epoch": 0.24, "learning_rate": 1.7685312406864986e-05, "loss": 0.5796, "step": 1270 }, { "epoch": 0.24, "learning_rate": 1.7681323552042094e-05, "loss": 0.5973, "step": 1271 }, { "epoch": 0.24, "learning_rate": 1.767733171388165e-05, "loss": 0.5847, "step": 1272 }, { "epoch": 0.24, "learning_rate": 1.7673336893934033e-05, "loss": 0.5989, "step": 1273 }, { "epoch": 0.25, "learning_rate": 1.7669339093750786e-05, "loss": 0.4762, "step": 1274 }, { "epoch": 0.25, "learning_rate": 1.766533831488461e-05, "loss": 0.5274, "step": 1275 }, { "epoch": 0.25, "learning_rate": 1.7661334558889357e-05, "loss": 0.6334, "step": 1276 }, { "epoch": 0.25, "learning_rate": 1.7657327827320046e-05, "loss": 0.6338, "step": 1277 }, { "epoch": 0.25, "learning_rate": 1.765331812173284e-05, "loss": 0.6124, "step": 1278 }, { "epoch": 0.25, "learning_rate": 1.7649305443685068e-05, "loss": 0.5636, "step": 1279 }, { "epoch": 0.25, "learning_rate": 1.76452897947352e-05, "loss": 0.5712, "step": 1280 }, { "epoch": 0.25, "learning_rate": 1.7641271176442876e-05, "loss": 0.5849, "step": 1281 }, { "epoch": 0.25, "learning_rate": 1.7637249590368878e-05, "loss": 0.4919, "step": 1282 }, { "epoch": 0.25, "learning_rate": 1.763322503807514e-05, "loss": 0.6, "step": 1283 }, { "epoch": 0.25, "learning_rate": 1.7629197521124758e-05, "loss": 0.6034, "step": 1284 }, { "epoch": 0.25, "learning_rate": 1.7625167041081967e-05, "loss": 0.399, "step": 1285 }, { "epoch": 0.25, "learning_rate": 1.7621133599512163e-05, "loss": 0.5765, "step": 1286 }, { "epoch": 0.25, "learning_rate": 1.761709719798189e-05, "loss": 0.6485, "step": 1287 }, { "epoch": 0.25, "learning_rate": 1.761305783805883e-05, "loss": 0.6207, "step": 1288 }, { "epoch": 0.25, "learning_rate": 1.7609015521311836e-05, "loss": 0.5545, "step": 1289 }, { "epoch": 0.25, "learning_rate": 1.7604970249310893e-05, "loss": 0.4442, "step": 1290 }, { "epoch": 0.25, "learning_rate": 1.7600922023627137e-05, "loss": 0.5844, "step": 1291 }, { "epoch": 0.25, "learning_rate": 1.759687084583285e-05, "loss": 0.6283, "step": 1292 }, { "epoch": 0.25, "learning_rate": 1.759281671750147e-05, "loss": 0.6001, "step": 1293 }, { "epoch": 0.25, "learning_rate": 1.7588759640207564e-05, "loss": 0.6054, "step": 1294 }, { "epoch": 0.25, "learning_rate": 1.7584699615526857e-05, "loss": 0.5075, "step": 1295 }, { "epoch": 0.25, "learning_rate": 1.7580636645036224e-05, "loss": 0.3899, "step": 1296 }, { "epoch": 0.25, "learning_rate": 1.757657073031367e-05, "loss": 0.64, "step": 1297 }, { "epoch": 0.25, "learning_rate": 1.7572501872938343e-05, "loss": 0.5638, "step": 1298 }, { "epoch": 0.25, "learning_rate": 1.756843007449055e-05, "loss": 0.5475, "step": 1299 }, { "epoch": 0.25, "learning_rate": 1.7564355336551727e-05, "loss": 0.5613, "step": 1300 }, { "epoch": 0.25, "learning_rate": 1.7560277660704455e-05, "loss": 0.6186, "step": 1301 }, { "epoch": 0.25, "learning_rate": 1.755619704853246e-05, "loss": 0.365, "step": 1302 }, { "epoch": 0.25, "learning_rate": 1.7552113501620595e-05, "loss": 0.6248, "step": 1303 }, { "epoch": 0.25, "learning_rate": 1.7548027021554874e-05, "loss": 0.486, "step": 1304 }, { "epoch": 0.25, "learning_rate": 1.754393760992243e-05, "loss": 0.5866, "step": 1305 }, { "epoch": 0.25, "learning_rate": 1.7539845268311548e-05, "loss": 0.5653, "step": 1306 }, { "epoch": 0.25, "learning_rate": 1.7535749998311645e-05, "loss": 0.5503, "step": 1307 }, { "epoch": 0.25, "learning_rate": 1.753165180151328e-05, "loss": 0.5822, "step": 1308 }, { "epoch": 0.25, "learning_rate": 1.752755067950814e-05, "loss": 0.5231, "step": 1309 }, { "epoch": 0.25, "learning_rate": 1.752344663388906e-05, "loss": 0.5832, "step": 1310 }, { "epoch": 0.25, "learning_rate": 1.7519339666249997e-05, "loss": 0.6066, "step": 1311 }, { "epoch": 0.25, "learning_rate": 1.7515229778186052e-05, "loss": 0.6414, "step": 1312 }, { "epoch": 0.25, "learning_rate": 1.7511116971293463e-05, "loss": 0.5668, "step": 1313 }, { "epoch": 0.25, "learning_rate": 1.7507001247169587e-05, "loss": 0.6291, "step": 1314 }, { "epoch": 0.25, "learning_rate": 1.7502882607412933e-05, "loss": 0.4887, "step": 1315 }, { "epoch": 0.25, "learning_rate": 1.749876105362313e-05, "loss": 0.5766, "step": 1316 }, { "epoch": 0.25, "learning_rate": 1.7494636587400942e-05, "loss": 0.5817, "step": 1317 }, { "epoch": 0.25, "learning_rate": 1.749050921034826e-05, "loss": 0.4057, "step": 1318 }, { "epoch": 0.25, "learning_rate": 1.7486378924068123e-05, "loss": 0.5929, "step": 1319 }, { "epoch": 0.25, "learning_rate": 1.748224573016467e-05, "loss": 0.5872, "step": 1320 }, { "epoch": 0.25, "learning_rate": 1.7478109630243195e-05, "loss": 0.6594, "step": 1321 }, { "epoch": 0.25, "learning_rate": 1.747397062591011e-05, "loss": 0.6167, "step": 1322 }, { "epoch": 0.25, "learning_rate": 1.746982871877296e-05, "loss": 0.5894, "step": 1323 }, { "epoch": 0.25, "learning_rate": 1.7465683910440405e-05, "loss": 0.599, "step": 1324 }, { "epoch": 0.25, "learning_rate": 1.7461536202522248e-05, "loss": 0.5934, "step": 1325 }, { "epoch": 0.26, "learning_rate": 1.745738559662941e-05, "loss": 0.496, "step": 1326 }, { "epoch": 0.26, "learning_rate": 1.7453232094373936e-05, "loss": 0.6092, "step": 1327 }, { "epoch": 0.26, "learning_rate": 1.7449075697369005e-05, "loss": 0.6374, "step": 1328 }, { "epoch": 0.26, "learning_rate": 1.7444916407228904e-05, "loss": 0.6122, "step": 1329 }, { "epoch": 0.26, "learning_rate": 1.744075422556906e-05, "loss": 0.3674, "step": 1330 }, { "epoch": 0.26, "learning_rate": 1.7436589154006014e-05, "loss": 0.6155, "step": 1331 }, { "epoch": 0.26, "learning_rate": 1.743242119415743e-05, "loss": 0.584, "step": 1332 }, { "epoch": 0.26, "learning_rate": 1.7428250347642102e-05, "loss": 0.6188, "step": 1333 }, { "epoch": 0.26, "learning_rate": 1.7424076616079933e-05, "loss": 0.6124, "step": 1334 }, { "epoch": 0.26, "learning_rate": 1.7419900001091953e-05, "loss": 0.5824, "step": 1335 }, { "epoch": 0.26, "learning_rate": 1.7415720504300314e-05, "loss": 0.6159, "step": 1336 }, { "epoch": 0.26, "learning_rate": 1.741153812732828e-05, "loss": 0.6463, "step": 1337 }, { "epoch": 0.26, "learning_rate": 1.7407352871800246e-05, "loss": 0.5474, "step": 1338 }, { "epoch": 0.26, "learning_rate": 1.7403164739341708e-05, "loss": 0.4406, "step": 1339 }, { "epoch": 0.26, "learning_rate": 1.739897373157929e-05, "loss": 0.6068, "step": 1340 }, { "epoch": 0.26, "learning_rate": 1.7394779850140736e-05, "loss": 0.6015, "step": 1341 }, { "epoch": 0.26, "learning_rate": 1.7390583096654895e-05, "loss": 0.5982, "step": 1342 }, { "epoch": 0.26, "learning_rate": 1.7386383472751745e-05, "loss": 0.6192, "step": 1343 }, { "epoch": 0.26, "learning_rate": 1.7382180980062365e-05, "loss": 0.6113, "step": 1344 }, { "epoch": 0.26, "learning_rate": 1.7377975620218954e-05, "loss": 0.606, "step": 1345 }, { "epoch": 0.26, "learning_rate": 1.7373767394854836e-05, "loss": 0.5714, "step": 1346 }, { "epoch": 0.26, "learning_rate": 1.7369556305604422e-05, "loss": 0.6112, "step": 1347 }, { "epoch": 0.26, "learning_rate": 1.736534235410326e-05, "loss": 0.5229, "step": 1348 }, { "epoch": 0.26, "learning_rate": 1.7361125541988e-05, "loss": 0.3519, "step": 1349 }, { "epoch": 0.26, "learning_rate": 1.7356905870896407e-05, "loss": 0.5976, "step": 1350 }, { "epoch": 0.26, "learning_rate": 1.735268334246734e-05, "loss": 0.6096, "step": 1351 }, { "epoch": 0.26, "learning_rate": 1.7348457958340792e-05, "loss": 0.5735, "step": 1352 }, { "epoch": 0.26, "learning_rate": 1.7344229720157846e-05, "loss": 0.5959, "step": 1353 }, { "epoch": 0.26, "learning_rate": 1.7339998629560705e-05, "loss": 0.6034, "step": 1354 }, { "epoch": 0.26, "learning_rate": 1.7335764688192676e-05, "loss": 0.6013, "step": 1355 }, { "epoch": 0.26, "learning_rate": 1.733152789769817e-05, "loss": 0.5731, "step": 1356 }, { "epoch": 0.26, "learning_rate": 1.7327288259722714e-05, "loss": 0.5889, "step": 1357 }, { "epoch": 0.26, "learning_rate": 1.7323045775912927e-05, "loss": 0.6445, "step": 1358 }, { "epoch": 0.26, "learning_rate": 1.7318800447916543e-05, "loss": 0.5788, "step": 1359 }, { "epoch": 0.26, "learning_rate": 1.7314552277382403e-05, "loss": 0.5847, "step": 1360 }, { "epoch": 0.26, "learning_rate": 1.7310301265960446e-05, "loss": 0.5897, "step": 1361 }, { "epoch": 0.26, "learning_rate": 1.7306047415301706e-05, "loss": 0.6369, "step": 1362 }, { "epoch": 0.26, "learning_rate": 1.7301790727058344e-05, "loss": 0.489, "step": 1363 }, { "epoch": 0.26, "learning_rate": 1.7297531202883598e-05, "loss": 0.4463, "step": 1364 }, { "epoch": 0.26, "learning_rate": 1.7293268844431826e-05, "loss": 0.4726, "step": 1365 }, { "epoch": 0.26, "learning_rate": 1.7289003653358472e-05, "loss": 0.5062, "step": 1366 }, { "epoch": 0.26, "learning_rate": 1.7284735631320093e-05, "loss": 0.535, "step": 1367 }, { "epoch": 0.26, "learning_rate": 1.7280464779974335e-05, "loss": 0.4573, "step": 1368 }, { "epoch": 0.26, "learning_rate": 1.7276191100979952e-05, "loss": 0.6107, "step": 1369 }, { "epoch": 0.26, "learning_rate": 1.7271914595996784e-05, "loss": 0.5876, "step": 1370 }, { "epoch": 0.26, "learning_rate": 1.7267635266685782e-05, "loss": 0.6018, "step": 1371 }, { "epoch": 0.26, "learning_rate": 1.7263353114708993e-05, "loss": 0.598, "step": 1372 }, { "epoch": 0.26, "learning_rate": 1.7259068141729542e-05, "loss": 0.6546, "step": 1373 }, { "epoch": 0.26, "learning_rate": 1.7254780349411677e-05, "loss": 0.5234, "step": 1374 }, { "epoch": 0.26, "learning_rate": 1.7250489739420718e-05, "loss": 0.5783, "step": 1375 }, { "epoch": 0.26, "learning_rate": 1.7246196313423095e-05, "loss": 0.5926, "step": 1376 }, { "epoch": 0.26, "learning_rate": 1.7241900073086318e-05, "loss": 0.6209, "step": 1377 }, { "epoch": 0.27, "learning_rate": 1.7237601020079003e-05, "loss": 0.5863, "step": 1378 }, { "epoch": 0.27, "learning_rate": 1.7233299156070852e-05, "loss": 0.4042, "step": 1379 }, { "epoch": 0.27, "learning_rate": 1.7228994482732653e-05, "loss": 0.6168, "step": 1380 }, { "epoch": 0.27, "learning_rate": 1.72246870017363e-05, "loss": 0.5749, "step": 1381 }, { "epoch": 0.27, "learning_rate": 1.7220376714754766e-05, "loss": 0.5823, "step": 1382 }, { "epoch": 0.27, "learning_rate": 1.7216063623462112e-05, "loss": 0.5592, "step": 1383 }, { "epoch": 0.27, "learning_rate": 1.7211747729533504e-05, "loss": 0.4572, "step": 1384 }, { "epoch": 0.27, "learning_rate": 1.7207429034645176e-05, "loss": 0.6461, "step": 1385 }, { "epoch": 0.27, "learning_rate": 1.720310754047446e-05, "loss": 0.6518, "step": 1386 }, { "epoch": 0.27, "learning_rate": 1.719878324869978e-05, "loss": 0.6017, "step": 1387 }, { "epoch": 0.27, "learning_rate": 1.7194456161000634e-05, "loss": 0.553, "step": 1388 }, { "epoch": 0.27, "learning_rate": 1.719012627905762e-05, "loss": 0.6718, "step": 1389 }, { "epoch": 0.27, "learning_rate": 1.718579360455241e-05, "loss": 0.633, "step": 1390 }, { "epoch": 0.27, "learning_rate": 1.7181458139167767e-05, "loss": 0.5888, "step": 1391 }, { "epoch": 0.27, "learning_rate": 1.7177119884587536e-05, "loss": 0.6476, "step": 1392 }, { "epoch": 0.27, "learning_rate": 1.717277884249664e-05, "loss": 0.5807, "step": 1393 }, { "epoch": 0.27, "learning_rate": 1.716843501458109e-05, "loss": 0.6199, "step": 1394 }, { "epoch": 0.27, "learning_rate": 1.716408840252799e-05, "loss": 0.6284, "step": 1395 }, { "epoch": 0.27, "learning_rate": 1.7159739008025503e-05, "loss": 0.5443, "step": 1396 }, { "epoch": 0.27, "learning_rate": 1.7155386832762892e-05, "loss": 0.6076, "step": 1397 }, { "epoch": 0.27, "learning_rate": 1.715103187843048e-05, "loss": 0.668, "step": 1398 }, { "epoch": 0.27, "learning_rate": 1.7146674146719688e-05, "loss": 0.5065, "step": 1399 }, { "epoch": 0.27, "learning_rate": 1.7142313639323012e-05, "loss": 0.5638, "step": 1400 }, { "epoch": 0.27, "learning_rate": 1.7137950357934017e-05, "loss": 0.6286, "step": 1401 }, { "epoch": 0.27, "learning_rate": 1.7133584304247354e-05, "loss": 0.6123, "step": 1402 }, { "epoch": 0.27, "learning_rate": 1.7129215479958747e-05, "loss": 0.623, "step": 1403 }, { "epoch": 0.27, "learning_rate": 1.7124843886765e-05, "loss": 0.5825, "step": 1404 }, { "epoch": 0.27, "learning_rate": 1.712046952636398e-05, "loss": 0.5746, "step": 1405 }, { "epoch": 0.27, "learning_rate": 1.7116092400454655e-05, "loss": 0.5647, "step": 1406 }, { "epoch": 0.27, "learning_rate": 1.7111712510737035e-05, "loss": 0.6313, "step": 1407 }, { "epoch": 0.27, "learning_rate": 1.7107329858912226e-05, "loss": 0.5814, "step": 1408 }, { "epoch": 0.27, "learning_rate": 1.7102944446682393e-05, "loss": 0.5693, "step": 1409 }, { "epoch": 0.27, "learning_rate": 1.709855627575079e-05, "loss": 0.6459, "step": 1410 }, { "epoch": 0.27, "learning_rate": 1.7094165347821724e-05, "loss": 0.5808, "step": 1411 }, { "epoch": 0.27, "learning_rate": 1.7089771664600584e-05, "loss": 0.6413, "step": 1412 }, { "epoch": 0.27, "learning_rate": 1.708537522779382e-05, "loss": 0.6453, "step": 1413 }, { "epoch": 0.27, "learning_rate": 1.7080976039108964e-05, "loss": 0.58, "step": 1414 }, { "epoch": 0.27, "learning_rate": 1.7076574100254614e-05, "loss": 0.5137, "step": 1415 }, { "epoch": 0.27, "learning_rate": 1.707216941294042e-05, "loss": 0.6115, "step": 1416 }, { "epoch": 0.27, "learning_rate": 1.706776197887712e-05, "loss": 0.6165, "step": 1417 }, { "epoch": 0.27, "learning_rate": 1.7063351799776514e-05, "loss": 0.5199, "step": 1418 }, { "epoch": 0.27, "learning_rate": 1.7058938877351456e-05, "loss": 0.6387, "step": 1419 }, { "epoch": 0.27, "learning_rate": 1.705452321331588e-05, "loss": 0.5515, "step": 1420 }, { "epoch": 0.27, "learning_rate": 1.7050104809384774e-05, "loss": 0.5204, "step": 1421 }, { "epoch": 0.27, "learning_rate": 1.70456836672742e-05, "loss": 0.614, "step": 1422 }, { "epoch": 0.27, "learning_rate": 1.704125978870128e-05, "loss": 0.5952, "step": 1423 }, { "epoch": 0.27, "learning_rate": 1.7036833175384192e-05, "loss": 0.6419, "step": 1424 }, { "epoch": 0.27, "learning_rate": 1.7032403829042182e-05, "loss": 0.6032, "step": 1425 }, { "epoch": 0.27, "learning_rate": 1.7027971751395563e-05, "loss": 0.6368, "step": 1426 }, { "epoch": 0.27, "learning_rate": 1.7023536944165697e-05, "loss": 0.6277, "step": 1427 }, { "epoch": 0.27, "learning_rate": 1.7019099409075014e-05, "loss": 0.6452, "step": 1428 }, { "epoch": 0.27, "learning_rate": 1.7014659147847005e-05, "loss": 0.6096, "step": 1429 }, { "epoch": 0.28, "learning_rate": 1.701021616220621e-05, "loss": 0.4622, "step": 1430 }, { "epoch": 0.28, "learning_rate": 1.7005770453878234e-05, "loss": 0.6234, "step": 1431 }, { "epoch": 0.28, "learning_rate": 1.7001322024589742e-05, "loss": 0.6239, "step": 1432 }, { "epoch": 0.28, "learning_rate": 1.6996870876068455e-05, "loss": 0.5456, "step": 1433 }, { "epoch": 0.28, "learning_rate": 1.6992417010043144e-05, "loss": 0.4767, "step": 1434 }, { "epoch": 0.28, "learning_rate": 1.6987960428243637e-05, "loss": 0.4946, "step": 1435 }, { "epoch": 0.28, "learning_rate": 1.6983501132400825e-05, "loss": 0.5915, "step": 1436 }, { "epoch": 0.28, "learning_rate": 1.6979039124246643e-05, "loss": 0.5604, "step": 1437 }, { "epoch": 0.28, "learning_rate": 1.6974574405514083e-05, "loss": 0.6236, "step": 1438 }, { "epoch": 0.28, "learning_rate": 1.6970106977937192e-05, "loss": 0.4114, "step": 1439 }, { "epoch": 0.28, "learning_rate": 1.696563684325107e-05, "loss": 0.6074, "step": 1440 }, { "epoch": 0.28, "learning_rate": 1.6961164003191862e-05, "loss": 0.6158, "step": 1441 }, { "epoch": 0.28, "learning_rate": 1.6956688459496767e-05, "loss": 0.499, "step": 1442 }, { "epoch": 0.28, "learning_rate": 1.695221021390404e-05, "loss": 0.6104, "step": 1443 }, { "epoch": 0.28, "learning_rate": 1.6947729268152972e-05, "loss": 0.6309, "step": 1444 }, { "epoch": 0.28, "learning_rate": 1.6943245623983918e-05, "loss": 0.6237, "step": 1445 }, { "epoch": 0.28, "learning_rate": 1.6938759283138268e-05, "loss": 0.5218, "step": 1446 }, { "epoch": 0.28, "learning_rate": 1.693427024735847e-05, "loss": 0.6081, "step": 1447 }, { "epoch": 0.28, "learning_rate": 1.692977851838801e-05, "loss": 0.6204, "step": 1448 }, { "epoch": 0.28, "learning_rate": 1.6925284097971427e-05, "loss": 0.5849, "step": 1449 }, { "epoch": 0.28, "learning_rate": 1.6920786987854296e-05, "loss": 0.5798, "step": 1450 }, { "epoch": 0.28, "learning_rate": 1.691628718978325e-05, "loss": 0.6241, "step": 1451 }, { "epoch": 0.28, "learning_rate": 1.691178470550596e-05, "loss": 0.5998, "step": 1452 }, { "epoch": 0.28, "learning_rate": 1.6907279536771127e-05, "loss": 0.5402, "step": 1453 }, { "epoch": 0.28, "learning_rate": 1.6902771685328524e-05, "loss": 0.5879, "step": 1454 }, { "epoch": 0.28, "learning_rate": 1.6898261152928933e-05, "loss": 0.5614, "step": 1455 }, { "epoch": 0.28, "learning_rate": 1.6893747941324197e-05, "loss": 0.5583, "step": 1456 }, { "epoch": 0.28, "learning_rate": 1.6889232052267203e-05, "loss": 0.5937, "step": 1457 }, { "epoch": 0.28, "learning_rate": 1.688471348751186e-05, "loss": 0.6552, "step": 1458 }, { "epoch": 0.28, "learning_rate": 1.688019224881313e-05, "loss": 0.6181, "step": 1459 }, { "epoch": 0.28, "learning_rate": 1.6875668337927014e-05, "loss": 0.5816, "step": 1460 }, { "epoch": 0.28, "learning_rate": 1.6871141756610544e-05, "loss": 0.616, "step": 1461 }, { "epoch": 0.28, "learning_rate": 1.6866612506621788e-05, "loss": 0.5727, "step": 1462 }, { "epoch": 0.28, "learning_rate": 1.6862080589719863e-05, "loss": 0.4935, "step": 1463 }, { "epoch": 0.28, "learning_rate": 1.6857546007664908e-05, "loss": 0.5192, "step": 1464 }, { "epoch": 0.28, "learning_rate": 1.6853008762218103e-05, "loss": 0.6015, "step": 1465 }, { "epoch": 0.28, "learning_rate": 1.684846885514166e-05, "loss": 0.586, "step": 1466 }, { "epoch": 0.28, "learning_rate": 1.6843926288198828e-05, "loss": 0.5654, "step": 1467 }, { "epoch": 0.28, "learning_rate": 1.683938106315389e-05, "loss": 0.452, "step": 1468 }, { "epoch": 0.28, "learning_rate": 1.683483318177216e-05, "loss": 0.6189, "step": 1469 }, { "epoch": 0.28, "learning_rate": 1.6830282645819974e-05, "loss": 0.6018, "step": 1470 }, { "epoch": 0.28, "learning_rate": 1.6825729457064718e-05, "loss": 0.5854, "step": 1471 }, { "epoch": 0.28, "learning_rate": 1.6821173617274793e-05, "loss": 0.6313, "step": 1472 }, { "epoch": 0.28, "learning_rate": 1.6816615128219635e-05, "loss": 0.484, "step": 1473 }, { "epoch": 0.28, "learning_rate": 1.681205399166971e-05, "loss": 0.6389, "step": 1474 }, { "epoch": 0.28, "learning_rate": 1.6807490209396506e-05, "loss": 0.5928, "step": 1475 }, { "epoch": 0.28, "learning_rate": 1.6802923783172553e-05, "loss": 0.6631, "step": 1476 }, { "epoch": 0.28, "learning_rate": 1.679835471477139e-05, "loss": 0.5692, "step": 1477 }, { "epoch": 0.28, "learning_rate": 1.6793783005967593e-05, "loss": 0.4867, "step": 1478 }, { "epoch": 0.28, "learning_rate": 1.678920865853676e-05, "loss": 0.6248, "step": 1479 }, { "epoch": 0.28, "learning_rate": 1.678463167425552e-05, "loss": 0.5887, "step": 1480 }, { "epoch": 0.28, "learning_rate": 1.6780052054901512e-05, "loss": 0.5838, "step": 1481 }, { "epoch": 0.29, "learning_rate": 1.6775469802253416e-05, "loss": 0.6479, "step": 1482 }, { "epoch": 0.29, "learning_rate": 1.6770884918090923e-05, "loss": 0.6079, "step": 1483 }, { "epoch": 0.29, "learning_rate": 1.6766297404194745e-05, "loss": 0.6336, "step": 1484 }, { "epoch": 0.29, "learning_rate": 1.6761707262346624e-05, "loss": 0.5894, "step": 1485 }, { "epoch": 0.29, "learning_rate": 1.675711449432932e-05, "loss": 0.5762, "step": 1486 }, { "epoch": 0.29, "learning_rate": 1.6752519101926606e-05, "loss": 0.6227, "step": 1487 }, { "epoch": 0.29, "learning_rate": 1.6747921086923284e-05, "loss": 0.6298, "step": 1488 }, { "epoch": 0.29, "learning_rate": 1.674332045110517e-05, "loss": 0.5768, "step": 1489 }, { "epoch": 0.29, "learning_rate": 1.6738717196259092e-05, "loss": 0.6172, "step": 1490 }, { "epoch": 0.29, "learning_rate": 1.673411132417291e-05, "loss": 0.6276, "step": 1491 }, { "epoch": 0.29, "learning_rate": 1.672950283663548e-05, "loss": 0.4996, "step": 1492 }, { "epoch": 0.29, "learning_rate": 1.6724891735436697e-05, "loss": 0.5663, "step": 1493 }, { "epoch": 0.29, "learning_rate": 1.6720278022367453e-05, "loss": 0.5307, "step": 1494 }, { "epoch": 0.29, "learning_rate": 1.6715661699219664e-05, "loss": 0.5702, "step": 1495 }, { "epoch": 0.29, "learning_rate": 1.6711042767786257e-05, "loss": 0.521, "step": 1496 }, { "epoch": 0.29, "learning_rate": 1.6706421229861168e-05, "loss": 0.5366, "step": 1497 }, { "epoch": 0.29, "learning_rate": 1.6701797087239354e-05, "loss": 0.4114, "step": 1498 }, { "epoch": 0.29, "learning_rate": 1.6697170341716772e-05, "loss": 0.5022, "step": 1499 }, { "epoch": 0.29, "learning_rate": 1.6692540995090403e-05, "loss": 0.4914, "step": 1500 }, { "epoch": 0.29, "learning_rate": 1.668790904915823e-05, "loss": 0.612, "step": 1501 }, { "epoch": 0.29, "learning_rate": 1.6683274505719248e-05, "loss": 0.5987, "step": 1502 }, { "epoch": 0.29, "learning_rate": 1.6678637366573455e-05, "loss": 0.5994, "step": 1503 }, { "epoch": 0.29, "learning_rate": 1.667399763352187e-05, "loss": 0.6506, "step": 1504 }, { "epoch": 0.29, "learning_rate": 1.666935530836651e-05, "loss": 0.566, "step": 1505 }, { "epoch": 0.29, "learning_rate": 1.6664710392910396e-05, "loss": 0.5759, "step": 1506 }, { "epoch": 0.29, "learning_rate": 1.6660062888957564e-05, "loss": 0.581, "step": 1507 }, { "epoch": 0.29, "learning_rate": 1.665541279831305e-05, "loss": 0.584, "step": 1508 }, { "epoch": 0.29, "learning_rate": 1.6650760122782898e-05, "loss": 0.516, "step": 1509 }, { "epoch": 0.29, "learning_rate": 1.6646104864174147e-05, "loss": 0.6029, "step": 1510 }, { "epoch": 0.29, "learning_rate": 1.664144702429485e-05, "loss": 0.6226, "step": 1511 }, { "epoch": 0.29, "learning_rate": 1.663678660495406e-05, "loss": 0.5758, "step": 1512 }, { "epoch": 0.29, "learning_rate": 1.663212360796183e-05, "loss": 0.5809, "step": 1513 }, { "epoch": 0.29, "learning_rate": 1.662745803512921e-05, "loss": 0.6053, "step": 1514 }, { "epoch": 0.29, "learning_rate": 1.662278988826826e-05, "loss": 0.5824, "step": 1515 }, { "epoch": 0.29, "learning_rate": 1.6618119169192027e-05, "loss": 0.6018, "step": 1516 }, { "epoch": 0.29, "learning_rate": 1.661344587971457e-05, "loss": 0.5214, "step": 1517 }, { "epoch": 0.29, "learning_rate": 1.6608770021650945e-05, "loss": 0.6274, "step": 1518 }, { "epoch": 0.29, "learning_rate": 1.6604091596817193e-05, "loss": 0.6264, "step": 1519 }, { "epoch": 0.29, "learning_rate": 1.6599410607030363e-05, "loss": 0.607, "step": 1520 }, { "epoch": 0.29, "learning_rate": 1.6594727054108498e-05, "loss": 0.5958, "step": 1521 }, { "epoch": 0.29, "learning_rate": 1.659004093987064e-05, "loss": 0.6688, "step": 1522 }, { "epoch": 0.29, "learning_rate": 1.6585352266136814e-05, "loss": 0.596, "step": 1523 }, { "epoch": 0.29, "learning_rate": 1.6580661034728055e-05, "loss": 0.5134, "step": 1524 }, { "epoch": 0.29, "learning_rate": 1.6575967247466376e-05, "loss": 0.5873, "step": 1525 }, { "epoch": 0.29, "learning_rate": 1.657127090617479e-05, "loss": 0.4782, "step": 1526 }, { "epoch": 0.29, "learning_rate": 1.656657201267731e-05, "loss": 0.5676, "step": 1527 }, { "epoch": 0.29, "learning_rate": 1.6561870568798927e-05, "loss": 0.582, "step": 1528 }, { "epoch": 0.29, "learning_rate": 1.655716657636562e-05, "loss": 0.6334, "step": 1529 }, { "epoch": 0.29, "learning_rate": 1.6552460037204382e-05, "loss": 0.6369, "step": 1530 }, { "epoch": 0.29, "learning_rate": 1.6547750953143168e-05, "loss": 0.4945, "step": 1531 }, { "epoch": 0.29, "learning_rate": 1.654303932601093e-05, "loss": 0.628, "step": 1532 }, { "epoch": 0.29, "learning_rate": 1.6538325157637614e-05, "loss": 0.4886, "step": 1533 }, { "epoch": 0.3, "learning_rate": 1.653360844985415e-05, "loss": 0.5451, "step": 1534 }, { "epoch": 0.3, "learning_rate": 1.652888920449245e-05, "loss": 0.6197, "step": 1535 }, { "epoch": 0.3, "learning_rate": 1.6524167423385414e-05, "loss": 0.4773, "step": 1536 }, { "epoch": 0.3, "learning_rate": 1.651944310836693e-05, "loss": 0.4891, "step": 1537 }, { "epoch": 0.3, "learning_rate": 1.6514716261271866e-05, "loss": 0.5981, "step": 1538 }, { "epoch": 0.3, "learning_rate": 1.6509986883936073e-05, "loss": 0.5856, "step": 1539 }, { "epoch": 0.3, "learning_rate": 1.650525497819639e-05, "loss": 0.5722, "step": 1540 }, { "epoch": 0.3, "learning_rate": 1.6500520545890634e-05, "loss": 0.5314, "step": 1541 }, { "epoch": 0.3, "learning_rate": 1.6495783588857605e-05, "loss": 0.4878, "step": 1542 }, { "epoch": 0.3, "learning_rate": 1.649104410893708e-05, "loss": 0.6134, "step": 1543 }, { "epoch": 0.3, "learning_rate": 1.648630210796982e-05, "loss": 0.5916, "step": 1544 }, { "epoch": 0.3, "learning_rate": 1.6481557587797562e-05, "loss": 0.6168, "step": 1545 }, { "epoch": 0.3, "learning_rate": 1.6476810550263023e-05, "loss": 0.5743, "step": 1546 }, { "epoch": 0.3, "learning_rate": 1.6472060997209898e-05, "loss": 0.6232, "step": 1547 }, { "epoch": 0.3, "learning_rate": 1.6467308930482863e-05, "loss": 0.6037, "step": 1548 }, { "epoch": 0.3, "learning_rate": 1.6462554351927558e-05, "loss": 0.639, "step": 1549 }, { "epoch": 0.3, "learning_rate": 1.6457797263390613e-05, "loss": 0.6227, "step": 1550 }, { "epoch": 0.3, "learning_rate": 1.6453037666719624e-05, "loss": 0.4829, "step": 1551 }, { "epoch": 0.3, "learning_rate": 1.6448275563763162e-05, "loss": 0.6144, "step": 1552 }, { "epoch": 0.3, "learning_rate": 1.644351095637078e-05, "loss": 0.6376, "step": 1553 }, { "epoch": 0.3, "learning_rate": 1.6438743846392987e-05, "loss": 0.4711, "step": 1554 }, { "epoch": 0.3, "learning_rate": 1.6433974235681274e-05, "loss": 0.6402, "step": 1555 }, { "epoch": 0.3, "learning_rate": 1.6429202126088112e-05, "loss": 0.6447, "step": 1556 }, { "epoch": 0.3, "learning_rate": 1.6424427519466925e-05, "loss": 0.5127, "step": 1557 }, { "epoch": 0.3, "learning_rate": 1.641965041767212e-05, "loss": 0.5774, "step": 1558 }, { "epoch": 0.3, "learning_rate": 1.6414870822559064e-05, "loss": 0.6577, "step": 1559 }, { "epoch": 0.3, "learning_rate": 1.6410088735984103e-05, "loss": 0.5614, "step": 1560 }, { "epoch": 0.3, "learning_rate": 1.6405304159804534e-05, "loss": 0.5093, "step": 1561 }, { "epoch": 0.3, "learning_rate": 1.6400517095878644e-05, "loss": 0.473, "step": 1562 }, { "epoch": 0.3, "learning_rate": 1.6395727546065665e-05, "loss": 0.4844, "step": 1563 }, { "epoch": 0.3, "learning_rate": 1.6390935512225806e-05, "loss": 0.569, "step": 1564 }, { "epoch": 0.3, "learning_rate": 1.6386140996220232e-05, "loss": 0.4814, "step": 1565 }, { "epoch": 0.3, "learning_rate": 1.6381343999911088e-05, "loss": 0.633, "step": 1566 }, { "epoch": 0.3, "learning_rate": 1.6376544525161463e-05, "loss": 0.6025, "step": 1567 }, { "epoch": 0.3, "learning_rate": 1.6371742573835426e-05, "loss": 0.654, "step": 1568 }, { "epoch": 0.3, "learning_rate": 1.636693814779799e-05, "loss": 0.6463, "step": 1569 }, { "epoch": 0.3, "learning_rate": 1.6362131248915145e-05, "loss": 0.6596, "step": 1570 }, { "epoch": 0.3, "learning_rate": 1.6357321879053833e-05, "loss": 0.5599, "step": 1571 }, { "epoch": 0.3, "learning_rate": 1.6352510040081962e-05, "loss": 0.5961, "step": 1572 }, { "epoch": 0.3, "learning_rate": 1.634769573386839e-05, "loss": 0.5937, "step": 1573 }, { "epoch": 0.3, "learning_rate": 1.634287896228294e-05, "loss": 0.5856, "step": 1574 }, { "epoch": 0.3, "learning_rate": 1.6338059727196386e-05, "loss": 0.6194, "step": 1575 }, { "epoch": 0.3, "learning_rate": 1.6333238030480473e-05, "loss": 0.6273, "step": 1576 }, { "epoch": 0.3, "learning_rate": 1.6328413874007884e-05, "loss": 0.5231, "step": 1577 }, { "epoch": 0.3, "learning_rate": 1.6323587259652267e-05, "loss": 0.6551, "step": 1578 }, { "epoch": 0.3, "learning_rate": 1.6318758189288227e-05, "loss": 0.6097, "step": 1579 }, { "epoch": 0.3, "learning_rate": 1.6313926664791316e-05, "loss": 0.6002, "step": 1580 }, { "epoch": 0.3, "learning_rate": 1.6309092688038047e-05, "loss": 0.5854, "step": 1581 }, { "epoch": 0.3, "learning_rate": 1.6304256260905872e-05, "loss": 0.5753, "step": 1582 }, { "epoch": 0.3, "learning_rate": 1.6299417385273216e-05, "loss": 0.586, "step": 1583 }, { "epoch": 0.3, "learning_rate": 1.629457606301943e-05, "loss": 0.3562, "step": 1584 }, { "epoch": 0.3, "learning_rate": 1.6289732296024837e-05, "loss": 0.6097, "step": 1585 }, { "epoch": 0.31, "learning_rate": 1.6284886086170697e-05, "loss": 0.5894, "step": 1586 }, { "epoch": 0.31, "learning_rate": 1.628003743533922e-05, "loss": 0.5914, "step": 1587 }, { "epoch": 0.31, "learning_rate": 1.6275186345413566e-05, "loss": 0.4832, "step": 1588 }, { "epoch": 0.31, "learning_rate": 1.627033281827785e-05, "loss": 0.6071, "step": 1589 }, { "epoch": 0.31, "learning_rate": 1.6265476855817116e-05, "loss": 0.588, "step": 1590 }, { "epoch": 0.31, "learning_rate": 1.6260618459917366e-05, "loss": 0.6105, "step": 1591 }, { "epoch": 0.31, "learning_rate": 1.6255757632465553e-05, "loss": 0.6288, "step": 1592 }, { "epoch": 0.31, "learning_rate": 1.625089437534956e-05, "loss": 0.6461, "step": 1593 }, { "epoch": 0.31, "learning_rate": 1.624602869045822e-05, "loss": 0.5939, "step": 1594 }, { "epoch": 0.31, "learning_rate": 1.624116057968131e-05, "loss": 0.6131, "step": 1595 }, { "epoch": 0.31, "learning_rate": 1.6236290044909543e-05, "loss": 0.6023, "step": 1596 }, { "epoch": 0.31, "learning_rate": 1.6231417088034585e-05, "loss": 0.4826, "step": 1597 }, { "epoch": 0.31, "learning_rate": 1.622654171094904e-05, "loss": 0.6214, "step": 1598 }, { "epoch": 0.31, "learning_rate": 1.6221663915546437e-05, "loss": 0.5567, "step": 1599 }, { "epoch": 0.31, "learning_rate": 1.6216783703721265e-05, "loss": 0.6368, "step": 1600 }, { "epoch": 0.31, "learning_rate": 1.6211901077368937e-05, "loss": 0.5795, "step": 1601 }, { "epoch": 0.31, "learning_rate": 1.620701603838581e-05, "loss": 0.6355, "step": 1602 }, { "epoch": 0.31, "learning_rate": 1.6202128588669177e-05, "loss": 0.6137, "step": 1603 }, { "epoch": 0.31, "learning_rate": 1.619723873011727e-05, "loss": 0.5656, "step": 1604 }, { "epoch": 0.31, "learning_rate": 1.6192346464629247e-05, "loss": 0.5823, "step": 1605 }, { "epoch": 0.31, "learning_rate": 1.6187451794105212e-05, "loss": 0.6493, "step": 1606 }, { "epoch": 0.31, "learning_rate": 1.61825547204462e-05, "loss": 0.5739, "step": 1607 }, { "epoch": 0.31, "learning_rate": 1.6177655245554177e-05, "loss": 0.5819, "step": 1608 }, { "epoch": 0.31, "learning_rate": 1.617275337133204e-05, "loss": 0.5621, "step": 1609 }, { "epoch": 0.31, "learning_rate": 1.6167849099683623e-05, "loss": 0.5726, "step": 1610 }, { "epoch": 0.31, "learning_rate": 1.6162942432513687e-05, "loss": 0.6014, "step": 1611 }, { "epoch": 0.31, "learning_rate": 1.6158033371727924e-05, "loss": 0.6333, "step": 1612 }, { "epoch": 0.31, "learning_rate": 1.6153121919232962e-05, "loss": 0.6052, "step": 1613 }, { "epoch": 0.31, "learning_rate": 1.614820807693635e-05, "loss": 0.5003, "step": 1614 }, { "epoch": 0.31, "learning_rate": 1.6143291846746563e-05, "loss": 0.6452, "step": 1615 }, { "epoch": 0.31, "learning_rate": 1.613837323057301e-05, "loss": 0.5925, "step": 1616 }, { "epoch": 0.31, "learning_rate": 1.6133452230326035e-05, "loss": 0.5166, "step": 1617 }, { "epoch": 0.31, "learning_rate": 1.6128528847916883e-05, "loss": 0.6068, "step": 1618 }, { "epoch": 0.31, "learning_rate": 1.6123603085257746e-05, "loss": 0.4927, "step": 1619 }, { "epoch": 0.31, "learning_rate": 1.6118674944261732e-05, "loss": 0.6259, "step": 1620 }, { "epoch": 0.31, "learning_rate": 1.6113744426842882e-05, "loss": 0.5423, "step": 1621 }, { "epoch": 0.31, "learning_rate": 1.6108811534916137e-05, "loss": 0.6395, "step": 1622 }, { "epoch": 0.31, "learning_rate": 1.6103876270397387e-05, "loss": 0.5913, "step": 1623 }, { "epoch": 0.31, "learning_rate": 1.609893863520343e-05, "loss": 0.5767, "step": 1624 }, { "epoch": 0.31, "learning_rate": 1.609399863125198e-05, "loss": 0.5503, "step": 1625 }, { "epoch": 0.31, "learning_rate": 1.6089056260461687e-05, "loss": 0.4853, "step": 1626 }, { "epoch": 0.31, "learning_rate": 1.6084111524752107e-05, "loss": 0.5952, "step": 1627 }, { "epoch": 0.31, "learning_rate": 1.607916442604372e-05, "loss": 0.5206, "step": 1628 }, { "epoch": 0.31, "learning_rate": 1.6074214966257914e-05, "loss": 0.6259, "step": 1629 }, { "epoch": 0.31, "learning_rate": 1.6069263147317015e-05, "loss": 0.5935, "step": 1630 }, { "epoch": 0.31, "learning_rate": 1.6064308971144236e-05, "loss": 0.5703, "step": 1631 }, { "epoch": 0.31, "learning_rate": 1.605935243966374e-05, "loss": 0.4955, "step": 1632 }, { "epoch": 0.31, "learning_rate": 1.6054393554800574e-05, "loss": 0.5735, "step": 1633 }, { "epoch": 0.31, "learning_rate": 1.604943231848072e-05, "loss": 0.5809, "step": 1634 }, { "epoch": 0.31, "learning_rate": 1.604446873263106e-05, "loss": 0.5781, "step": 1635 }, { "epoch": 0.31, "learning_rate": 1.6039502799179394e-05, "loss": 0.6453, "step": 1636 }, { "epoch": 0.31, "learning_rate": 1.6034534520054435e-05, "loss": 0.5043, "step": 1637 }, { "epoch": 0.32, "learning_rate": 1.60295638971858e-05, "loss": 0.4838, "step": 1638 }, { "epoch": 0.32, "learning_rate": 1.602459093250403e-05, "loss": 0.4726, "step": 1639 }, { "epoch": 0.32, "learning_rate": 1.601961562794056e-05, "loss": 0.6036, "step": 1640 }, { "epoch": 0.32, "learning_rate": 1.601463798542775e-05, "loss": 0.6294, "step": 1641 }, { "epoch": 0.32, "learning_rate": 1.6009658006898848e-05, "loss": 0.505, "step": 1642 }, { "epoch": 0.32, "learning_rate": 1.600467569428803e-05, "loss": 0.5503, "step": 1643 }, { "epoch": 0.32, "learning_rate": 1.599969104953036e-05, "loss": 0.6315, "step": 1644 }, { "epoch": 0.32, "learning_rate": 1.599470407456182e-05, "loss": 0.5936, "step": 1645 }, { "epoch": 0.32, "learning_rate": 1.5989714771319297e-05, "loss": 0.5923, "step": 1646 }, { "epoch": 0.32, "learning_rate": 1.5984723141740578e-05, "loss": 0.6055, "step": 1647 }, { "epoch": 0.32, "learning_rate": 1.597972918776435e-05, "loss": 0.5811, "step": 1648 }, { "epoch": 0.32, "learning_rate": 1.5974732911330208e-05, "loss": 0.5104, "step": 1649 }, { "epoch": 0.32, "learning_rate": 1.5969734314378654e-05, "loss": 0.6073, "step": 1650 }, { "epoch": 0.32, "learning_rate": 1.5964733398851078e-05, "loss": 0.6045, "step": 1651 }, { "epoch": 0.32, "learning_rate": 1.5959730166689783e-05, "loss": 0.6232, "step": 1652 }, { "epoch": 0.32, "learning_rate": 1.5954724619837966e-05, "loss": 0.5923, "step": 1653 }, { "epoch": 0.32, "learning_rate": 1.5949716760239722e-05, "loss": 0.6344, "step": 1654 }, { "epoch": 0.32, "learning_rate": 1.5944706589840046e-05, "loss": 0.4089, "step": 1655 }, { "epoch": 0.32, "learning_rate": 1.5939694110584833e-05, "loss": 0.59, "step": 1656 }, { "epoch": 0.32, "learning_rate": 1.593467932442087e-05, "loss": 0.665, "step": 1657 }, { "epoch": 0.32, "learning_rate": 1.5929662233295846e-05, "loss": 0.6291, "step": 1658 }, { "epoch": 0.32, "learning_rate": 1.5924642839158334e-05, "loss": 0.6072, "step": 1659 }, { "epoch": 0.32, "learning_rate": 1.591962114395781e-05, "loss": 0.6543, "step": 1660 }, { "epoch": 0.32, "learning_rate": 1.5914597149644654e-05, "loss": 0.6407, "step": 1661 }, { "epoch": 0.32, "learning_rate": 1.5909570858170115e-05, "loss": 0.6437, "step": 1662 }, { "epoch": 0.32, "learning_rate": 1.5904542271486346e-05, "loss": 0.5934, "step": 1663 }, { "epoch": 0.32, "learning_rate": 1.5899511391546403e-05, "loss": 0.6187, "step": 1664 }, { "epoch": 0.32, "learning_rate": 1.5894478220304215e-05, "loss": 0.5996, "step": 1665 }, { "epoch": 0.32, "learning_rate": 1.5889442759714603e-05, "loss": 0.6129, "step": 1666 }, { "epoch": 0.32, "learning_rate": 1.5884405011733294e-05, "loss": 0.6333, "step": 1667 }, { "epoch": 0.32, "learning_rate": 1.587936497831688e-05, "loss": 0.5991, "step": 1668 }, { "epoch": 0.32, "learning_rate": 1.5874322661422856e-05, "loss": 0.5281, "step": 1669 }, { "epoch": 0.32, "learning_rate": 1.5869278063009602e-05, "loss": 0.5699, "step": 1670 }, { "epoch": 0.32, "learning_rate": 1.586423118503638e-05, "loss": 0.5954, "step": 1671 }, { "epoch": 0.32, "learning_rate": 1.585918202946334e-05, "loss": 0.62, "step": 1672 }, { "epoch": 0.32, "learning_rate": 1.5854130598251514e-05, "loss": 0.6273, "step": 1673 }, { "epoch": 0.32, "learning_rate": 1.5849076893362822e-05, "loss": 0.5879, "step": 1674 }, { "epoch": 0.32, "learning_rate": 1.584402091676006e-05, "loss": 0.4601, "step": 1675 }, { "epoch": 0.32, "learning_rate": 1.5838962670406918e-05, "loss": 0.589, "step": 1676 }, { "epoch": 0.32, "learning_rate": 1.5833902156267956e-05, "loss": 0.5958, "step": 1677 }, { "epoch": 0.32, "learning_rate": 1.582883937630862e-05, "loss": 0.5976, "step": 1678 }, { "epoch": 0.32, "learning_rate": 1.5823774332495236e-05, "loss": 0.5342, "step": 1679 }, { "epoch": 0.32, "learning_rate": 1.581870702679501e-05, "loss": 0.5172, "step": 1680 }, { "epoch": 0.32, "learning_rate": 1.581363746117602e-05, "loss": 0.5654, "step": 1681 }, { "epoch": 0.32, "learning_rate": 1.580856563760724e-05, "loss": 0.5832, "step": 1682 }, { "epoch": 0.32, "learning_rate": 1.5803491558058486e-05, "loss": 0.5082, "step": 1683 }, { "epoch": 0.32, "learning_rate": 1.579841522450049e-05, "loss": 0.6413, "step": 1684 }, { "epoch": 0.32, "learning_rate": 1.5793336638904838e-05, "loss": 0.6402, "step": 1685 }, { "epoch": 0.32, "learning_rate": 1.578825580324399e-05, "loss": 0.6154, "step": 1686 }, { "epoch": 0.32, "learning_rate": 1.5783172719491288e-05, "loss": 0.5432, "step": 1687 }, { "epoch": 0.32, "learning_rate": 1.577808738962094e-05, "loss": 0.4845, "step": 1688 }, { "epoch": 0.32, "learning_rate": 1.577299981560803e-05, "loss": 0.6011, "step": 1689 }, { "epoch": 0.33, "learning_rate": 1.5767909999428513e-05, "loss": 0.5841, "step": 1690 }, { "epoch": 0.33, "learning_rate": 1.576281794305922e-05, "loss": 0.6016, "step": 1691 }, { "epoch": 0.33, "learning_rate": 1.575772364847784e-05, "loss": 0.4952, "step": 1692 }, { "epoch": 0.33, "learning_rate": 1.575262711766294e-05, "loss": 0.6216, "step": 1693 }, { "epoch": 0.33, "learning_rate": 1.5747528352593956e-05, "loss": 0.5037, "step": 1694 }, { "epoch": 0.33, "learning_rate": 1.574242735525119e-05, "loss": 0.6281, "step": 1695 }, { "epoch": 0.33, "learning_rate": 1.5737324127615808e-05, "loss": 0.6143, "step": 1696 }, { "epoch": 0.33, "learning_rate": 1.5732218671669847e-05, "loss": 0.553, "step": 1697 }, { "epoch": 0.33, "learning_rate": 1.5727110989396205e-05, "loss": 0.6324, "step": 1698 }, { "epoch": 0.33, "learning_rate": 1.5722001082778645e-05, "loss": 0.5967, "step": 1699 }, { "epoch": 0.33, "learning_rate": 1.5716888953801805e-05, "loss": 0.5952, "step": 1700 }, { "epoch": 0.33, "learning_rate": 1.5711774604451168e-05, "loss": 0.5851, "step": 1701 }, { "epoch": 0.33, "learning_rate": 1.5706658036713093e-05, "loss": 0.4622, "step": 1702 }, { "epoch": 0.33, "learning_rate": 1.5701539252574795e-05, "loss": 0.5971, "step": 1703 }, { "epoch": 0.33, "learning_rate": 1.5696418254024344e-05, "loss": 0.6349, "step": 1704 }, { "epoch": 0.33, "learning_rate": 1.569129504305069e-05, "loss": 0.4699, "step": 1705 }, { "epoch": 0.33, "learning_rate": 1.568616962164362e-05, "loss": 0.6168, "step": 1706 }, { "epoch": 0.33, "learning_rate": 1.5681041991793788e-05, "loss": 0.6252, "step": 1707 }, { "epoch": 0.33, "learning_rate": 1.567591215549271e-05, "loss": 0.6421, "step": 1708 }, { "epoch": 0.33, "learning_rate": 1.567078011473276e-05, "loss": 0.603, "step": 1709 }, { "epoch": 0.33, "learning_rate": 1.5665645871507152e-05, "loss": 0.5162, "step": 1710 }, { "epoch": 0.33, "learning_rate": 1.5660509427809973e-05, "loss": 0.5777, "step": 1711 }, { "epoch": 0.33, "learning_rate": 1.565537078563616e-05, "loss": 0.6129, "step": 1712 }, { "epoch": 0.33, "learning_rate": 1.56502299469815e-05, "loss": 0.6277, "step": 1713 }, { "epoch": 0.33, "learning_rate": 1.564508691384264e-05, "loss": 0.5898, "step": 1714 }, { "epoch": 0.33, "learning_rate": 1.5639941688217063e-05, "loss": 0.5659, "step": 1715 }, { "epoch": 0.33, "learning_rate": 1.5634794272103126e-05, "loss": 0.5464, "step": 1716 }, { "epoch": 0.33, "learning_rate": 1.562964466750003e-05, "loss": 0.61, "step": 1717 }, { "epoch": 0.33, "learning_rate": 1.562449287640781e-05, "loss": 0.4671, "step": 1718 }, { "epoch": 0.33, "learning_rate": 1.5619338900827368e-05, "loss": 0.56, "step": 1719 }, { "epoch": 0.33, "learning_rate": 1.5614182742760448e-05, "loss": 0.5843, "step": 1720 }, { "epoch": 0.33, "learning_rate": 1.5609024404209643e-05, "loss": 0.6278, "step": 1721 }, { "epoch": 0.33, "learning_rate": 1.5603863887178393e-05, "loss": 0.5752, "step": 1722 }, { "epoch": 0.33, "learning_rate": 1.5598701193670983e-05, "loss": 0.571, "step": 1723 }, { "epoch": 0.33, "learning_rate": 1.559353632569254e-05, "loss": 0.6157, "step": 1724 }, { "epoch": 0.33, "learning_rate": 1.5588369285249048e-05, "loss": 0.6042, "step": 1725 }, { "epoch": 0.33, "learning_rate": 1.5583200074347318e-05, "loss": 0.4994, "step": 1726 }, { "epoch": 0.33, "learning_rate": 1.557802869499501e-05, "loss": 0.6056, "step": 1727 }, { "epoch": 0.33, "learning_rate": 1.5572855149200637e-05, "loss": 0.5046, "step": 1728 }, { "epoch": 0.33, "learning_rate": 1.5567679438973543e-05, "loss": 0.516, "step": 1729 }, { "epoch": 0.33, "learning_rate": 1.5562501566323906e-05, "loss": 0.6132, "step": 1730 }, { "epoch": 0.33, "learning_rate": 1.555732153326276e-05, "loss": 0.6423, "step": 1731 }, { "epoch": 0.33, "learning_rate": 1.5552139341801965e-05, "loss": 0.5209, "step": 1732 }, { "epoch": 0.33, "learning_rate": 1.554695499395423e-05, "loss": 0.4956, "step": 1733 }, { "epoch": 0.33, "learning_rate": 1.5541768491733092e-05, "loss": 0.5868, "step": 1734 }, { "epoch": 0.33, "learning_rate": 1.5536579837152927e-05, "loss": 0.5946, "step": 1735 }, { "epoch": 0.33, "learning_rate": 1.5531389032228955e-05, "loss": 0.601, "step": 1736 }, { "epoch": 0.33, "learning_rate": 1.552619607897722e-05, "loss": 0.5818, "step": 1737 }, { "epoch": 0.33, "learning_rate": 1.55210009794146e-05, "loss": 0.6118, "step": 1738 }, { "epoch": 0.33, "learning_rate": 1.5515803735558827e-05, "loss": 0.5951, "step": 1739 }, { "epoch": 0.33, "learning_rate": 1.5510604349428438e-05, "loss": 0.507, "step": 1740 }, { "epoch": 0.33, "learning_rate": 1.550540282304282e-05, "loss": 0.466, "step": 1741 }, { "epoch": 0.34, "learning_rate": 1.550019915842218e-05, "loss": 0.6295, "step": 1742 }, { "epoch": 0.34, "learning_rate": 1.549499335758757e-05, "loss": 0.522, "step": 1743 }, { "epoch": 0.34, "learning_rate": 1.548978542256086e-05, "loss": 0.596, "step": 1744 }, { "epoch": 0.34, "learning_rate": 1.5484575355364744e-05, "loss": 0.6252, "step": 1745 }, { "epoch": 0.34, "learning_rate": 1.5479363158022763e-05, "loss": 0.4779, "step": 1746 }, { "epoch": 0.34, "learning_rate": 1.547414883255927e-05, "loss": 0.6369, "step": 1747 }, { "epoch": 0.34, "learning_rate": 1.546893238099945e-05, "loss": 0.5956, "step": 1748 }, { "epoch": 0.34, "learning_rate": 1.5463713805369312e-05, "loss": 0.6343, "step": 1749 }, { "epoch": 0.34, "learning_rate": 1.5458493107695688e-05, "loss": 0.6158, "step": 1750 }, { "epoch": 0.34, "learning_rate": 1.5453270290006237e-05, "loss": 0.6257, "step": 1751 }, { "epoch": 0.34, "learning_rate": 1.544804535432945e-05, "loss": 0.5585, "step": 1752 }, { "epoch": 0.34, "learning_rate": 1.544281830269462e-05, "loss": 0.5521, "step": 1753 }, { "epoch": 0.34, "learning_rate": 1.5437589137131882e-05, "loss": 0.5364, "step": 1754 }, { "epoch": 0.34, "learning_rate": 1.5432357859672177e-05, "loss": 0.6273, "step": 1755 }, { "epoch": 0.34, "learning_rate": 1.542712447234728e-05, "loss": 0.5936, "step": 1756 }, { "epoch": 0.34, "learning_rate": 1.542188897718977e-05, "loss": 0.5854, "step": 1757 }, { "epoch": 0.34, "learning_rate": 1.5416651376233062e-05, "loss": 0.5099, "step": 1758 }, { "epoch": 0.34, "learning_rate": 1.5411411671511376e-05, "loss": 0.5859, "step": 1759 }, { "epoch": 0.34, "learning_rate": 1.5406169865059747e-05, "loss": 0.6191, "step": 1760 }, { "epoch": 0.34, "learning_rate": 1.5400925958914045e-05, "loss": 0.5927, "step": 1761 }, { "epoch": 0.34, "learning_rate": 1.5395679955110927e-05, "loss": 0.6194, "step": 1762 }, { "epoch": 0.34, "learning_rate": 1.53904318556879e-05, "loss": 0.6295, "step": 1763 }, { "epoch": 0.34, "learning_rate": 1.5385181662683244e-05, "loss": 0.5037, "step": 1764 }, { "epoch": 0.34, "learning_rate": 1.5379929378136088e-05, "loss": 0.6069, "step": 1765 }, { "epoch": 0.34, "learning_rate": 1.5374675004086353e-05, "loss": 0.5143, "step": 1766 }, { "epoch": 0.34, "learning_rate": 1.5369418542574782e-05, "loss": 0.5648, "step": 1767 }, { "epoch": 0.34, "learning_rate": 1.536415999564292e-05, "loss": 0.5058, "step": 1768 }, { "epoch": 0.34, "learning_rate": 1.5358899365333123e-05, "loss": 0.6293, "step": 1769 }, { "epoch": 0.34, "learning_rate": 1.5353636653688563e-05, "loss": 0.5878, "step": 1770 }, { "epoch": 0.34, "learning_rate": 1.534837186275322e-05, "loss": 0.6273, "step": 1771 }, { "epoch": 0.34, "learning_rate": 1.5343104994571877e-05, "loss": 0.5981, "step": 1772 }, { "epoch": 0.34, "learning_rate": 1.533783605119012e-05, "loss": 0.5284, "step": 1773 }, { "epoch": 0.34, "learning_rate": 1.5332565034654344e-05, "loss": 0.5765, "step": 1774 }, { "epoch": 0.34, "learning_rate": 1.5327291947011763e-05, "loss": 0.6187, "step": 1775 }, { "epoch": 0.34, "learning_rate": 1.5322016790310373e-05, "loss": 0.6406, "step": 1776 }, { "epoch": 0.34, "learning_rate": 1.5316739566598985e-05, "loss": 0.4958, "step": 1777 }, { "epoch": 0.34, "learning_rate": 1.531146027792722e-05, "loss": 0.6119, "step": 1778 }, { "epoch": 0.34, "learning_rate": 1.530617892634548e-05, "loss": 0.6264, "step": 1779 }, { "epoch": 0.34, "learning_rate": 1.5300895513904993e-05, "loss": 0.6291, "step": 1780 }, { "epoch": 0.34, "learning_rate": 1.529561004265777e-05, "loss": 0.5378, "step": 1781 }, { "epoch": 0.34, "learning_rate": 1.5290322514656624e-05, "loss": 0.5906, "step": 1782 }, { "epoch": 0.34, "learning_rate": 1.5285032931955177e-05, "loss": 0.4789, "step": 1783 }, { "epoch": 0.34, "learning_rate": 1.527974129660784e-05, "loss": 0.641, "step": 1784 }, { "epoch": 0.34, "learning_rate": 1.527444761066982e-05, "loss": 0.4967, "step": 1785 }, { "epoch": 0.34, "learning_rate": 1.5269151876197127e-05, "loss": 0.5991, "step": 1786 }, { "epoch": 0.34, "learning_rate": 1.5263854095246557e-05, "loss": 0.5, "step": 1787 }, { "epoch": 0.34, "learning_rate": 1.5258554269875716e-05, "loss": 0.6009, "step": 1788 }, { "epoch": 0.34, "learning_rate": 1.5253252402142989e-05, "loss": 0.5801, "step": 1789 }, { "epoch": 0.34, "learning_rate": 1.5247948494107566e-05, "loss": 0.5793, "step": 1790 }, { "epoch": 0.34, "learning_rate": 1.5242642547829416e-05, "loss": 0.6315, "step": 1791 }, { "epoch": 0.34, "learning_rate": 1.523733456536931e-05, "loss": 0.5914, "step": 1792 }, { "epoch": 0.34, "learning_rate": 1.5232024548788813e-05, "loss": 0.5515, "step": 1793 }, { "epoch": 0.35, "learning_rate": 1.5226712500150267e-05, "loss": 0.6003, "step": 1794 }, { "epoch": 0.35, "learning_rate": 1.5221398421516816e-05, "loss": 0.6169, "step": 1795 }, { "epoch": 0.35, "learning_rate": 1.5216082314952383e-05, "loss": 0.632, "step": 1796 }, { "epoch": 0.35, "learning_rate": 1.521076418252168e-05, "loss": 0.6407, "step": 1797 }, { "epoch": 0.35, "learning_rate": 1.5205444026290218e-05, "loss": 0.6176, "step": 1798 }, { "epoch": 0.35, "learning_rate": 1.5200121848324276e-05, "loss": 0.3946, "step": 1799 }, { "epoch": 0.35, "learning_rate": 1.5194797650690926e-05, "loss": 0.5907, "step": 1800 }, { "epoch": 0.35, "learning_rate": 1.5189471435458032e-05, "loss": 0.5206, "step": 1801 }, { "epoch": 0.35, "learning_rate": 1.5184143204694231e-05, "loss": 0.6176, "step": 1802 }, { "epoch": 0.35, "learning_rate": 1.5178812960468945e-05, "loss": 0.5883, "step": 1803 }, { "epoch": 0.35, "learning_rate": 1.5173480704852379e-05, "loss": 0.4785, "step": 1804 }, { "epoch": 0.35, "learning_rate": 1.5168146439915525e-05, "loss": 0.5812, "step": 1805 }, { "epoch": 0.35, "learning_rate": 1.5162810167730144e-05, "loss": 0.5996, "step": 1806 }, { "epoch": 0.35, "learning_rate": 1.5157471890368785e-05, "loss": 0.6242, "step": 1807 }, { "epoch": 0.35, "learning_rate": 1.5152131609904773e-05, "loss": 0.6321, "step": 1808 }, { "epoch": 0.35, "learning_rate": 1.5146789328412213e-05, "loss": 0.5547, "step": 1809 }, { "epoch": 0.35, "learning_rate": 1.5141445047965984e-05, "loss": 0.5934, "step": 1810 }, { "epoch": 0.35, "learning_rate": 1.5136098770641741e-05, "loss": 0.6124, "step": 1811 }, { "epoch": 0.35, "learning_rate": 1.513075049851592e-05, "loss": 0.5883, "step": 1812 }, { "epoch": 0.35, "learning_rate": 1.5125400233665728e-05, "loss": 0.6182, "step": 1813 }, { "epoch": 0.35, "learning_rate": 1.5120047978169146e-05, "loss": 0.6105, "step": 1814 }, { "epoch": 0.35, "learning_rate": 1.5114693734104926e-05, "loss": 0.6065, "step": 1815 }, { "epoch": 0.35, "learning_rate": 1.5109337503552594e-05, "loss": 0.5692, "step": 1816 }, { "epoch": 0.35, "learning_rate": 1.5103979288592454e-05, "loss": 0.3977, "step": 1817 }, { "epoch": 0.35, "learning_rate": 1.5098619091305571e-05, "loss": 0.5836, "step": 1818 }, { "epoch": 0.35, "learning_rate": 1.5093256913773786e-05, "loss": 0.6449, "step": 1819 }, { "epoch": 0.35, "learning_rate": 1.50878927580797e-05, "loss": 0.5729, "step": 1820 }, { "epoch": 0.35, "learning_rate": 1.5082526626306698e-05, "loss": 0.4759, "step": 1821 }, { "epoch": 0.35, "learning_rate": 1.5077158520538921e-05, "loss": 0.6138, "step": 1822 }, { "epoch": 0.35, "learning_rate": 1.5071788442861277e-05, "loss": 0.6029, "step": 1823 }, { "epoch": 0.35, "learning_rate": 1.5066416395359444e-05, "loss": 0.608, "step": 1824 }, { "epoch": 0.35, "learning_rate": 1.5061042380119864e-05, "loss": 0.6101, "step": 1825 }, { "epoch": 0.35, "learning_rate": 1.5055666399229743e-05, "loss": 0.5832, "step": 1826 }, { "epoch": 0.35, "learning_rate": 1.5050288454777047e-05, "loss": 0.6351, "step": 1827 }, { "epoch": 0.35, "learning_rate": 1.504490854885051e-05, "loss": 0.6225, "step": 1828 }, { "epoch": 0.35, "learning_rate": 1.5039526683539627e-05, "loss": 0.5968, "step": 1829 }, { "epoch": 0.35, "learning_rate": 1.5034142860934649e-05, "loss": 0.5753, "step": 1830 }, { "epoch": 0.35, "learning_rate": 1.5028757083126594e-05, "loss": 0.6072, "step": 1831 }, { "epoch": 0.35, "learning_rate": 1.5023369352207229e-05, "loss": 0.632, "step": 1832 }, { "epoch": 0.35, "learning_rate": 1.5017979670269096e-05, "loss": 0.5836, "step": 1833 }, { "epoch": 0.35, "learning_rate": 1.501258803940548e-05, "loss": 0.5988, "step": 1834 }, { "epoch": 0.35, "learning_rate": 1.500719446171043e-05, "loss": 0.5062, "step": 1835 }, { "epoch": 0.35, "learning_rate": 1.500179893927875e-05, "loss": 0.5995, "step": 1836 }, { "epoch": 0.35, "learning_rate": 1.4996401474205997e-05, "loss": 0.6143, "step": 1837 }, { "epoch": 0.35, "learning_rate": 1.4991002068588484e-05, "loss": 0.6355, "step": 1838 }, { "epoch": 0.35, "learning_rate": 1.4985600724523282e-05, "loss": 0.6332, "step": 1839 }, { "epoch": 0.35, "learning_rate": 1.4980197444108205e-05, "loss": 0.5922, "step": 1840 }, { "epoch": 0.35, "learning_rate": 1.4974792229441826e-05, "loss": 0.4872, "step": 1841 }, { "epoch": 0.35, "learning_rate": 1.4969385082623473e-05, "loss": 0.6313, "step": 1842 }, { "epoch": 0.35, "learning_rate": 1.4963976005753216e-05, "loss": 0.491, "step": 1843 }, { "epoch": 0.35, "learning_rate": 1.4958565000931877e-05, "loss": 0.5797, "step": 1844 }, { "epoch": 0.35, "learning_rate": 1.4953152070261027e-05, "loss": 0.5793, "step": 1845 }, { "epoch": 0.36, "learning_rate": 1.494773721584299e-05, "loss": 0.5934, "step": 1846 }, { "epoch": 0.36, "learning_rate": 1.4942320439780833e-05, "loss": 0.58, "step": 1847 }, { "epoch": 0.36, "learning_rate": 1.4936901744178367e-05, "loss": 0.6284, "step": 1848 }, { "epoch": 0.36, "learning_rate": 1.4931481131140149e-05, "loss": 0.4289, "step": 1849 }, { "epoch": 0.36, "learning_rate": 1.4926058602771484e-05, "loss": 0.6051, "step": 1850 }, { "epoch": 0.36, "learning_rate": 1.4920634161178424e-05, "loss": 0.4833, "step": 1851 }, { "epoch": 0.36, "learning_rate": 1.4915207808467756e-05, "loss": 0.6005, "step": 1852 }, { "epoch": 0.36, "learning_rate": 1.4909779546747011e-05, "loss": 0.6315, "step": 1853 }, { "epoch": 0.36, "learning_rate": 1.4904349378124467e-05, "loss": 0.6041, "step": 1854 }, { "epoch": 0.36, "learning_rate": 1.489891730470914e-05, "loss": 0.6285, "step": 1855 }, { "epoch": 0.36, "learning_rate": 1.4893483328610778e-05, "loss": 0.4697, "step": 1856 }, { "epoch": 0.36, "learning_rate": 1.488804745193988e-05, "loss": 0.5621, "step": 1857 }, { "epoch": 0.36, "learning_rate": 1.4882609676807675e-05, "loss": 0.6012, "step": 1858 }, { "epoch": 0.36, "learning_rate": 1.4877170005326136e-05, "loss": 0.6058, "step": 1859 }, { "epoch": 0.36, "learning_rate": 1.4871728439607967e-05, "loss": 0.611, "step": 1860 }, { "epoch": 0.36, "learning_rate": 1.4866284981766607e-05, "loss": 0.6233, "step": 1861 }, { "epoch": 0.36, "learning_rate": 1.4860839633916236e-05, "loss": 0.5108, "step": 1862 }, { "epoch": 0.36, "learning_rate": 1.4855392398171762e-05, "loss": 0.581, "step": 1863 }, { "epoch": 0.36, "learning_rate": 1.484994327664883e-05, "loss": 0.6545, "step": 1864 }, { "epoch": 0.36, "learning_rate": 1.4844492271463814e-05, "loss": 0.5016, "step": 1865 }, { "epoch": 0.36, "learning_rate": 1.4839039384733821e-05, "loss": 0.5825, "step": 1866 }, { "epoch": 0.36, "learning_rate": 1.4833584618576695e-05, "loss": 0.4971, "step": 1867 }, { "epoch": 0.36, "learning_rate": 1.4828127975111e-05, "loss": 0.6199, "step": 1868 }, { "epoch": 0.36, "learning_rate": 1.4822669456456031e-05, "loss": 0.6246, "step": 1869 }, { "epoch": 0.36, "learning_rate": 1.4817209064731819e-05, "loss": 0.5985, "step": 1870 }, { "epoch": 0.36, "learning_rate": 1.4811746802059115e-05, "loss": 0.6233, "step": 1871 }, { "epoch": 0.36, "learning_rate": 1.48062826705594e-05, "loss": 0.6014, "step": 1872 }, { "epoch": 0.36, "learning_rate": 1.4800816672354876e-05, "loss": 0.6355, "step": 1873 }, { "epoch": 0.36, "learning_rate": 1.4795348809568477e-05, "loss": 0.6126, "step": 1874 }, { "epoch": 0.36, "learning_rate": 1.4789879084323858e-05, "loss": 0.6193, "step": 1875 }, { "epoch": 0.36, "learning_rate": 1.4784407498745394e-05, "loss": 0.6237, "step": 1876 }, { "epoch": 0.36, "learning_rate": 1.477893405495819e-05, "loss": 0.5912, "step": 1877 }, { "epoch": 0.36, "learning_rate": 1.4773458755088068e-05, "loss": 0.5787, "step": 1878 }, { "epoch": 0.36, "learning_rate": 1.4767981601261567e-05, "loss": 0.6118, "step": 1879 }, { "epoch": 0.36, "learning_rate": 1.4762502595605957e-05, "loss": 0.6293, "step": 1880 }, { "epoch": 0.36, "learning_rate": 1.4757021740249213e-05, "loss": 0.6226, "step": 1881 }, { "epoch": 0.36, "learning_rate": 1.4751539037320044e-05, "loss": 0.5508, "step": 1882 }, { "epoch": 0.36, "learning_rate": 1.4746054488947863e-05, "loss": 0.6134, "step": 1883 }, { "epoch": 0.36, "learning_rate": 1.4740568097262811e-05, "loss": 0.5856, "step": 1884 }, { "epoch": 0.36, "learning_rate": 1.473507986439573e-05, "loss": 0.5984, "step": 1885 }, { "epoch": 0.36, "learning_rate": 1.4729589792478193e-05, "loss": 0.561, "step": 1886 }, { "epoch": 0.36, "learning_rate": 1.4724097883642482e-05, "loss": 0.5734, "step": 1887 }, { "epoch": 0.36, "learning_rate": 1.4718604140021588e-05, "loss": 0.5474, "step": 1888 }, { "epoch": 0.36, "learning_rate": 1.471310856374922e-05, "loss": 0.5293, "step": 1889 }, { "epoch": 0.36, "learning_rate": 1.470761115695979e-05, "loss": 0.6006, "step": 1890 }, { "epoch": 0.36, "learning_rate": 1.4702111921788437e-05, "loss": 0.5559, "step": 1891 }, { "epoch": 0.36, "learning_rate": 1.4696610860370997e-05, "loss": 0.6232, "step": 1892 }, { "epoch": 0.36, "learning_rate": 1.4691107974844015e-05, "loss": 0.5943, "step": 1893 }, { "epoch": 0.36, "learning_rate": 1.468560326734475e-05, "loss": 0.4889, "step": 1894 }, { "epoch": 0.36, "learning_rate": 1.4680096740011172e-05, "loss": 0.5628, "step": 1895 }, { "epoch": 0.36, "learning_rate": 1.4674588394981948e-05, "loss": 0.5891, "step": 1896 }, { "epoch": 0.36, "learning_rate": 1.4669078234396454e-05, "loss": 0.5033, "step": 1897 }, { "epoch": 0.37, "learning_rate": 1.4663566260394775e-05, "loss": 0.6059, "step": 1898 }, { "epoch": 0.37, "learning_rate": 1.4658052475117704e-05, "loss": 0.5954, "step": 1899 }, { "epoch": 0.37, "learning_rate": 1.4652536880706723e-05, "loss": 0.5569, "step": 1900 }, { "epoch": 0.37, "learning_rate": 1.4647019479304028e-05, "loss": 0.5986, "step": 1901 }, { "epoch": 0.37, "learning_rate": 1.4641500273052516e-05, "loss": 0.6173, "step": 1902 }, { "epoch": 0.37, "learning_rate": 1.463597926409578e-05, "loss": 0.6183, "step": 1903 }, { "epoch": 0.37, "learning_rate": 1.4630456454578122e-05, "loss": 0.4625, "step": 1904 }, { "epoch": 0.37, "learning_rate": 1.462493184664453e-05, "loss": 0.4833, "step": 1905 }, { "epoch": 0.37, "learning_rate": 1.4619405442440702e-05, "loss": 0.5183, "step": 1906 }, { "epoch": 0.37, "learning_rate": 1.4613877244113033e-05, "loss": 0.5666, "step": 1907 }, { "epoch": 0.37, "learning_rate": 1.4608347253808605e-05, "loss": 0.5646, "step": 1908 }, { "epoch": 0.37, "learning_rate": 1.460281547367521e-05, "loss": 0.5646, "step": 1909 }, { "epoch": 0.37, "learning_rate": 1.4597281905861318e-05, "loss": 0.5963, "step": 1910 }, { "epoch": 0.37, "learning_rate": 1.4591746552516109e-05, "loss": 0.5642, "step": 1911 }, { "epoch": 0.37, "learning_rate": 1.4586209415789452e-05, "loss": 0.6166, "step": 1912 }, { "epoch": 0.37, "learning_rate": 1.4580670497831904e-05, "loss": 0.3855, "step": 1913 }, { "epoch": 0.37, "learning_rate": 1.4575129800794718e-05, "loss": 0.6039, "step": 1914 }, { "epoch": 0.37, "learning_rate": 1.4569587326829834e-05, "loss": 0.5943, "step": 1915 }, { "epoch": 0.37, "learning_rate": 1.4564043078089891e-05, "loss": 0.6069, "step": 1916 }, { "epoch": 0.37, "learning_rate": 1.4558497056728205e-05, "loss": 0.5628, "step": 1917 }, { "epoch": 0.37, "learning_rate": 1.4552949264898795e-05, "loss": 0.643, "step": 1918 }, { "epoch": 0.37, "learning_rate": 1.4547399704756348e-05, "loss": 0.6169, "step": 1919 }, { "epoch": 0.37, "learning_rate": 1.4541848378456255e-05, "loss": 0.4639, "step": 1920 }, { "epoch": 0.37, "learning_rate": 1.4536295288154594e-05, "loss": 0.6035, "step": 1921 }, { "epoch": 0.37, "learning_rate": 1.4530740436008111e-05, "loss": 0.609, "step": 1922 }, { "epoch": 0.37, "learning_rate": 1.452518382417425e-05, "loss": 0.5905, "step": 1923 }, { "epoch": 0.37, "learning_rate": 1.4519625454811135e-05, "loss": 0.5213, "step": 1924 }, { "epoch": 0.37, "learning_rate": 1.4514065330077575e-05, "loss": 0.5809, "step": 1925 }, { "epoch": 0.37, "learning_rate": 1.4508503452133053e-05, "loss": 0.5894, "step": 1926 }, { "epoch": 0.37, "learning_rate": 1.4502939823137744e-05, "loss": 0.6169, "step": 1927 }, { "epoch": 0.37, "learning_rate": 1.4497374445252496e-05, "loss": 0.4696, "step": 1928 }, { "epoch": 0.37, "learning_rate": 1.4491807320638835e-05, "loss": 0.5902, "step": 1929 }, { "epoch": 0.37, "learning_rate": 1.4486238451458972e-05, "loss": 0.5905, "step": 1930 }, { "epoch": 0.37, "learning_rate": 1.4480667839875786e-05, "loss": 0.6194, "step": 1931 }, { "epoch": 0.37, "learning_rate": 1.4475095488052843e-05, "loss": 0.5847, "step": 1932 }, { "epoch": 0.37, "learning_rate": 1.4469521398154381e-05, "loss": 0.5057, "step": 1933 }, { "epoch": 0.37, "learning_rate": 1.4463945572345308e-05, "loss": 0.5849, "step": 1934 }, { "epoch": 0.37, "learning_rate": 1.4458368012791213e-05, "loss": 0.6089, "step": 1935 }, { "epoch": 0.37, "learning_rate": 1.4452788721658355e-05, "loss": 0.6563, "step": 1936 }, { "epoch": 0.37, "learning_rate": 1.4447207701113669e-05, "loss": 0.603, "step": 1937 }, { "epoch": 0.37, "learning_rate": 1.4441624953324755e-05, "loss": 0.6078, "step": 1938 }, { "epoch": 0.37, "learning_rate": 1.4436040480459891e-05, "loss": 0.4972, "step": 1939 }, { "epoch": 0.37, "learning_rate": 1.443045428468802e-05, "loss": 0.5984, "step": 1940 }, { "epoch": 0.37, "learning_rate": 1.4424866368178761e-05, "loss": 0.6062, "step": 1941 }, { "epoch": 0.37, "learning_rate": 1.441927673310239e-05, "loss": 0.4771, "step": 1942 }, { "epoch": 0.37, "learning_rate": 1.4413685381629855e-05, "loss": 0.6887, "step": 1943 }, { "epoch": 0.37, "learning_rate": 1.440809231593278e-05, "loss": 0.5801, "step": 1944 }, { "epoch": 0.37, "learning_rate": 1.4402497538183444e-05, "loss": 0.5891, "step": 1945 }, { "epoch": 0.37, "learning_rate": 1.4396901050554794e-05, "loss": 0.6109, "step": 1946 }, { "epoch": 0.37, "learning_rate": 1.4391302855220442e-05, "loss": 0.5911, "step": 1947 }, { "epoch": 0.37, "learning_rate": 1.4385702954354662e-05, "loss": 0.5842, "step": 1948 }, { "epoch": 0.37, "learning_rate": 1.438010135013239e-05, "loss": 0.4536, "step": 1949 }, { "epoch": 0.38, "learning_rate": 1.4374498044729225e-05, "loss": 0.6228, "step": 1950 }, { "epoch": 0.38, "learning_rate": 1.4368893040321428e-05, "loss": 0.6195, "step": 1951 }, { "epoch": 0.38, "learning_rate": 1.4363286339085915e-05, "loss": 0.5899, "step": 1952 }, { "epoch": 0.38, "learning_rate": 1.435767794320027e-05, "loss": 0.5999, "step": 1953 }, { "epoch": 0.38, "learning_rate": 1.4352067854842724e-05, "loss": 0.4937, "step": 1954 }, { "epoch": 0.38, "learning_rate": 1.434645607619217e-05, "loss": 0.6153, "step": 1955 }, { "epoch": 0.38, "learning_rate": 1.434084260942816e-05, "loss": 0.5899, "step": 1956 }, { "epoch": 0.38, "learning_rate": 1.4335227456730902e-05, "loss": 0.6136, "step": 1957 }, { "epoch": 0.38, "learning_rate": 1.4329610620281253e-05, "loss": 0.6092, "step": 1958 }, { "epoch": 0.38, "learning_rate": 1.4323992102260733e-05, "loss": 0.6033, "step": 1959 }, { "epoch": 0.38, "learning_rate": 1.4318371904851502e-05, "loss": 0.4632, "step": 1960 }, { "epoch": 0.38, "learning_rate": 1.4312750030236382e-05, "loss": 0.514, "step": 1961 }, { "epoch": 0.38, "learning_rate": 1.4307126480598852e-05, "loss": 0.5293, "step": 1962 }, { "epoch": 0.38, "learning_rate": 1.4301501258123024e-05, "loss": 0.6144, "step": 1963 }, { "epoch": 0.38, "learning_rate": 1.4295874364993672e-05, "loss": 0.6005, "step": 1964 }, { "epoch": 0.38, "learning_rate": 1.4290245803396221e-05, "loss": 0.6102, "step": 1965 }, { "epoch": 0.38, "learning_rate": 1.4284615575516737e-05, "loss": 0.5154, "step": 1966 }, { "epoch": 0.38, "learning_rate": 1.4278983683541934e-05, "loss": 0.6091, "step": 1967 }, { "epoch": 0.38, "learning_rate": 1.4273350129659173e-05, "loss": 0.5924, "step": 1968 }, { "epoch": 0.38, "learning_rate": 1.4267714916056465e-05, "loss": 0.451, "step": 1969 }, { "epoch": 0.38, "learning_rate": 1.426207804492246e-05, "loss": 0.6132, "step": 1970 }, { "epoch": 0.38, "learning_rate": 1.4256439518446456e-05, "loss": 0.6005, "step": 1971 }, { "epoch": 0.38, "learning_rate": 1.4250799338818388e-05, "loss": 0.4671, "step": 1972 }, { "epoch": 0.38, "learning_rate": 1.424515750822884e-05, "loss": 0.5012, "step": 1973 }, { "epoch": 0.38, "learning_rate": 1.4239514028869032e-05, "loss": 0.5893, "step": 1974 }, { "epoch": 0.38, "learning_rate": 1.4233868902930827e-05, "loss": 0.5393, "step": 1975 }, { "epoch": 0.38, "learning_rate": 1.4228222132606729e-05, "loss": 0.5727, "step": 1976 }, { "epoch": 0.38, "learning_rate": 1.4222573720089874e-05, "loss": 0.6037, "step": 1977 }, { "epoch": 0.38, "learning_rate": 1.4216923667574042e-05, "loss": 0.5068, "step": 1978 }, { "epoch": 0.38, "learning_rate": 1.4211271977253653e-05, "loss": 0.5777, "step": 1979 }, { "epoch": 0.38, "learning_rate": 1.4205618651323753e-05, "loss": 0.6057, "step": 1980 }, { "epoch": 0.38, "learning_rate": 1.4199963691980027e-05, "loss": 0.5165, "step": 1981 }, { "epoch": 0.38, "learning_rate": 1.4194307101418805e-05, "loss": 0.5629, "step": 1982 }, { "epoch": 0.38, "learning_rate": 1.4188648881837033e-05, "loss": 0.6328, "step": 1983 }, { "epoch": 0.38, "learning_rate": 1.4182989035432299e-05, "loss": 0.6079, "step": 1984 }, { "epoch": 0.38, "learning_rate": 1.4177327564402825e-05, "loss": 0.6444, "step": 1985 }, { "epoch": 0.38, "learning_rate": 1.4171664470947464e-05, "loss": 0.5892, "step": 1986 }, { "epoch": 0.38, "learning_rate": 1.416599975726569e-05, "loss": 0.5793, "step": 1987 }, { "epoch": 0.38, "learning_rate": 1.4160333425557616e-05, "loss": 0.617, "step": 1988 }, { "epoch": 0.38, "learning_rate": 1.4154665478023977e-05, "loss": 0.6123, "step": 1989 }, { "epoch": 0.38, "learning_rate": 1.4148995916866139e-05, "loss": 0.6057, "step": 1990 }, { "epoch": 0.38, "learning_rate": 1.41433247442861e-05, "loss": 0.5774, "step": 1991 }, { "epoch": 0.38, "learning_rate": 1.4137651962486472e-05, "loss": 0.4875, "step": 1992 }, { "epoch": 0.38, "learning_rate": 1.4131977573670499e-05, "loss": 0.6166, "step": 1993 }, { "epoch": 0.38, "learning_rate": 1.412630158004205e-05, "loss": 0.5233, "step": 1994 }, { "epoch": 0.38, "learning_rate": 1.4120623983805617e-05, "loss": 0.6294, "step": 1995 }, { "epoch": 0.38, "learning_rate": 1.4114944787166307e-05, "loss": 0.6032, "step": 1996 }, { "epoch": 0.38, "learning_rate": 1.4109263992329858e-05, "loss": 0.5955, "step": 1997 }, { "epoch": 0.38, "learning_rate": 1.4103581601502629e-05, "loss": 0.5937, "step": 1998 }, { "epoch": 0.38, "learning_rate": 1.409789761689159e-05, "loss": 0.6006, "step": 1999 }, { "epoch": 0.38, "learning_rate": 1.4092212040704336e-05, "loss": 0.6208, "step": 2000 }, { "epoch": 0.38, "learning_rate": 1.408652487514908e-05, "loss": 0.5951, "step": 2001 }, { "epoch": 0.39, "learning_rate": 1.408083612243465e-05, "loss": 0.6328, "step": 2002 }, { "epoch": 0.39, "learning_rate": 1.4075145784770496e-05, "loss": 0.582, "step": 2003 }, { "epoch": 0.39, "learning_rate": 1.4069453864366678e-05, "loss": 0.6166, "step": 2004 }, { "epoch": 0.39, "learning_rate": 1.4063760363433867e-05, "loss": 0.5732, "step": 2005 }, { "epoch": 0.39, "learning_rate": 1.405806528418336e-05, "loss": 0.5836, "step": 2006 }, { "epoch": 0.39, "learning_rate": 1.4052368628827057e-05, "loss": 0.6451, "step": 2007 }, { "epoch": 0.39, "learning_rate": 1.4046670399577478e-05, "loss": 0.5899, "step": 2008 }, { "epoch": 0.39, "learning_rate": 1.4040970598647742e-05, "loss": 0.5828, "step": 2009 }, { "epoch": 0.39, "learning_rate": 1.4035269228251589e-05, "loss": 0.6072, "step": 2010 }, { "epoch": 0.39, "learning_rate": 1.4029566290603368e-05, "loss": 0.6688, "step": 2011 }, { "epoch": 0.39, "learning_rate": 1.4023861787918031e-05, "loss": 0.4727, "step": 2012 }, { "epoch": 0.39, "learning_rate": 1.4018155722411144e-05, "loss": 0.6179, "step": 2013 }, { "epoch": 0.39, "learning_rate": 1.4012448096298874e-05, "loss": 0.5538, "step": 2014 }, { "epoch": 0.39, "learning_rate": 1.4006738911798001e-05, "loss": 0.4996, "step": 2015 }, { "epoch": 0.39, "learning_rate": 1.40010281711259e-05, "loss": 0.4413, "step": 2016 }, { "epoch": 0.39, "learning_rate": 1.3995315876500565e-05, "loss": 0.6251, "step": 2017 }, { "epoch": 0.39, "learning_rate": 1.3989602030140581e-05, "loss": 0.5783, "step": 2018 }, { "epoch": 0.39, "learning_rate": 1.398388663426514e-05, "loss": 0.5161, "step": 2019 }, { "epoch": 0.39, "learning_rate": 1.3978169691094037e-05, "loss": 0.587, "step": 2020 }, { "epoch": 0.39, "learning_rate": 1.3972451202847665e-05, "loss": 0.5791, "step": 2021 }, { "epoch": 0.39, "learning_rate": 1.3966731171747024e-05, "loss": 0.635, "step": 2022 }, { "epoch": 0.39, "learning_rate": 1.3961009600013702e-05, "loss": 0.608, "step": 2023 }, { "epoch": 0.39, "learning_rate": 1.3955286489869894e-05, "loss": 0.5732, "step": 2024 }, { "epoch": 0.39, "learning_rate": 1.394956184353839e-05, "loss": 0.6325, "step": 2025 }, { "epoch": 0.39, "learning_rate": 1.3943835663242577e-05, "loss": 0.6383, "step": 2026 }, { "epoch": 0.39, "learning_rate": 1.3938107951206438e-05, "loss": 0.659, "step": 2027 }, { "epoch": 0.39, "learning_rate": 1.3932378709654548e-05, "loss": 0.6109, "step": 2028 }, { "epoch": 0.39, "learning_rate": 1.3926647940812081e-05, "loss": 0.5734, "step": 2029 }, { "epoch": 0.39, "learning_rate": 1.39209156469048e-05, "loss": 0.5409, "step": 2030 }, { "epoch": 0.39, "learning_rate": 1.3915181830159061e-05, "loss": 0.5936, "step": 2031 }, { "epoch": 0.39, "learning_rate": 1.3909446492801819e-05, "loss": 0.5565, "step": 2032 }, { "epoch": 0.39, "learning_rate": 1.3903709637060605e-05, "loss": 0.5406, "step": 2033 }, { "epoch": 0.39, "learning_rate": 1.3897971265163546e-05, "loss": 0.5481, "step": 2034 }, { "epoch": 0.39, "learning_rate": 1.3892231379339369e-05, "loss": 0.5274, "step": 2035 }, { "epoch": 0.39, "learning_rate": 1.3886489981817375e-05, "loss": 0.438, "step": 2036 }, { "epoch": 0.39, "learning_rate": 1.3880747074827454e-05, "loss": 0.5232, "step": 2037 }, { "epoch": 0.39, "learning_rate": 1.3875002660600085e-05, "loss": 0.5956, "step": 2038 }, { "epoch": 0.39, "learning_rate": 1.386925674136634e-05, "loss": 0.6308, "step": 2039 }, { "epoch": 0.39, "learning_rate": 1.3863509319357857e-05, "loss": 0.5663, "step": 2040 }, { "epoch": 0.39, "learning_rate": 1.3857760396806876e-05, "loss": 0.5784, "step": 2041 }, { "epoch": 0.39, "learning_rate": 1.3852009975946209e-05, "loss": 0.5569, "step": 2042 }, { "epoch": 0.39, "learning_rate": 1.3846258059009252e-05, "loss": 0.5738, "step": 2043 }, { "epoch": 0.39, "learning_rate": 1.384050464822999e-05, "loss": 0.5781, "step": 2044 }, { "epoch": 0.39, "learning_rate": 1.383474974584297e-05, "loss": 0.5843, "step": 2045 }, { "epoch": 0.39, "learning_rate": 1.3828993354083342e-05, "loss": 0.6127, "step": 2046 }, { "epoch": 0.39, "learning_rate": 1.3823235475186816e-05, "loss": 0.4764, "step": 2047 }, { "epoch": 0.39, "learning_rate": 1.3817476111389685e-05, "loss": 0.5795, "step": 2048 }, { "epoch": 0.39, "learning_rate": 1.3811715264928824e-05, "loss": 0.5851, "step": 2049 }, { "epoch": 0.39, "learning_rate": 1.3805952938041674e-05, "loss": 0.4802, "step": 2050 }, { "epoch": 0.39, "learning_rate": 1.3800189132966257e-05, "loss": 0.5534, "step": 2051 }, { "epoch": 0.39, "learning_rate": 1.3794423851941174e-05, "loss": 0.615, "step": 2052 }, { "epoch": 0.39, "learning_rate": 1.378865709720559e-05, "loss": 0.4809, "step": 2053 }, { "epoch": 0.4, "learning_rate": 1.3782888870999245e-05, "loss": 0.5721, "step": 2054 }, { "epoch": 0.4, "learning_rate": 1.377711917556245e-05, "loss": 0.6092, "step": 2055 }, { "epoch": 0.4, "learning_rate": 1.3771348013136096e-05, "loss": 0.5854, "step": 2056 }, { "epoch": 0.4, "learning_rate": 1.3765575385961627e-05, "loss": 0.5865, "step": 2057 }, { "epoch": 0.4, "learning_rate": 1.3759801296281072e-05, "loss": 0.5083, "step": 2058 }, { "epoch": 0.4, "learning_rate": 1.3754025746337014e-05, "loss": 0.5019, "step": 2059 }, { "epoch": 0.4, "learning_rate": 1.3748248738372616e-05, "loss": 0.5134, "step": 2060 }, { "epoch": 0.4, "learning_rate": 1.3742470274631599e-05, "loss": 0.6007, "step": 2061 }, { "epoch": 0.4, "learning_rate": 1.3736690357358253e-05, "loss": 0.61, "step": 2062 }, { "epoch": 0.4, "learning_rate": 1.3730908988797427e-05, "loss": 0.5211, "step": 2063 }, { "epoch": 0.4, "learning_rate": 1.3725126171194543e-05, "loss": 0.5645, "step": 2064 }, { "epoch": 0.4, "learning_rate": 1.371934190679558e-05, "loss": 0.587, "step": 2065 }, { "epoch": 0.4, "learning_rate": 1.3713556197847076e-05, "loss": 0.4963, "step": 2066 }, { "epoch": 0.4, "learning_rate": 1.3707769046596136e-05, "loss": 0.6115, "step": 2067 }, { "epoch": 0.4, "learning_rate": 1.3701980455290425e-05, "loss": 0.5577, "step": 2068 }, { "epoch": 0.4, "learning_rate": 1.3696190426178162e-05, "loss": 0.6413, "step": 2069 }, { "epoch": 0.4, "learning_rate": 1.3690398961508128e-05, "loss": 0.612, "step": 2070 }, { "epoch": 0.4, "learning_rate": 1.3684606063529662e-05, "loss": 0.6343, "step": 2071 }, { "epoch": 0.4, "learning_rate": 1.3678811734492659e-05, "loss": 0.5992, "step": 2072 }, { "epoch": 0.4, "learning_rate": 1.367301597664757e-05, "loss": 0.5734, "step": 2073 }, { "epoch": 0.4, "learning_rate": 1.36672187922454e-05, "loss": 0.5979, "step": 2074 }, { "epoch": 0.4, "learning_rate": 1.3661420183537705e-05, "loss": 0.6276, "step": 2075 }, { "epoch": 0.4, "learning_rate": 1.3655620152776605e-05, "loss": 0.5093, "step": 2076 }, { "epoch": 0.4, "learning_rate": 1.364981870221476e-05, "loss": 0.5849, "step": 2077 }, { "epoch": 0.4, "learning_rate": 1.364401583410539e-05, "loss": 0.6009, "step": 2078 }, { "epoch": 0.4, "learning_rate": 1.3638211550702256e-05, "loss": 0.6012, "step": 2079 }, { "epoch": 0.4, "learning_rate": 1.363240585425968e-05, "loss": 0.6208, "step": 2080 }, { "epoch": 0.4, "learning_rate": 1.362659874703253e-05, "loss": 0.4978, "step": 2081 }, { "epoch": 0.4, "learning_rate": 1.3620790231276213e-05, "loss": 0.6685, "step": 2082 }, { "epoch": 0.4, "learning_rate": 1.3614980309246692e-05, "loss": 0.6367, "step": 2083 }, { "epoch": 0.4, "learning_rate": 1.3609168983200474e-05, "loss": 0.6097, "step": 2084 }, { "epoch": 0.4, "learning_rate": 1.3603356255394613e-05, "loss": 0.6068, "step": 2085 }, { "epoch": 0.4, "learning_rate": 1.3597542128086702e-05, "loss": 0.5196, "step": 2086 }, { "epoch": 0.4, "learning_rate": 1.3591726603534885e-05, "loss": 0.5661, "step": 2087 }, { "epoch": 0.4, "learning_rate": 1.3585909683997842e-05, "loss": 0.6163, "step": 2088 }, { "epoch": 0.4, "learning_rate": 1.3580091371734798e-05, "loss": 0.6112, "step": 2089 }, { "epoch": 0.4, "learning_rate": 1.357427166900552e-05, "loss": 0.5904, "step": 2090 }, { "epoch": 0.4, "learning_rate": 1.3568450578070309e-05, "loss": 0.4913, "step": 2091 }, { "epoch": 0.4, "learning_rate": 1.3562628101190015e-05, "loss": 0.6174, "step": 2092 }, { "epoch": 0.4, "learning_rate": 1.3556804240626019e-05, "loss": 0.5044, "step": 2093 }, { "epoch": 0.4, "learning_rate": 1.3550978998640241e-05, "loss": 0.5722, "step": 2094 }, { "epoch": 0.4, "learning_rate": 1.3545152377495136e-05, "loss": 0.5025, "step": 2095 }, { "epoch": 0.4, "learning_rate": 1.3539324379453698e-05, "loss": 0.5091, "step": 2096 }, { "epoch": 0.4, "learning_rate": 1.3533495006779455e-05, "loss": 0.599, "step": 2097 }, { "epoch": 0.4, "learning_rate": 1.3527664261736471e-05, "loss": 0.4715, "step": 2098 }, { "epoch": 0.4, "learning_rate": 1.3521832146589335e-05, "loss": 0.5622, "step": 2099 }, { "epoch": 0.4, "learning_rate": 1.3515998663603174e-05, "loss": 0.487, "step": 2100 }, { "epoch": 0.4, "learning_rate": 1.3510163815043647e-05, "loss": 0.5878, "step": 2101 }, { "epoch": 0.4, "learning_rate": 1.3504327603176943e-05, "loss": 0.5867, "step": 2102 }, { "epoch": 0.4, "learning_rate": 1.3498490030269782e-05, "loss": 0.607, "step": 2103 }, { "epoch": 0.4, "learning_rate": 1.3492651098589398e-05, "loss": 0.5761, "step": 2104 }, { "epoch": 0.4, "learning_rate": 1.3486810810403578e-05, "loss": 0.6227, "step": 2105 }, { "epoch": 0.41, "learning_rate": 1.348096916798062e-05, "loss": 0.5914, "step": 2106 }, { "epoch": 0.41, "learning_rate": 1.3475126173589343e-05, "loss": 0.6163, "step": 2107 }, { "epoch": 0.41, "learning_rate": 1.3469281829499107e-05, "loss": 0.6058, "step": 2108 }, { "epoch": 0.41, "learning_rate": 1.3463436137979786e-05, "loss": 0.5878, "step": 2109 }, { "epoch": 0.41, "learning_rate": 1.3457589101301776e-05, "loss": 0.6209, "step": 2110 }, { "epoch": 0.41, "learning_rate": 1.3451740721736005e-05, "loss": 0.6055, "step": 2111 }, { "epoch": 0.41, "learning_rate": 1.3445891001553905e-05, "loss": 0.5632, "step": 2112 }, { "epoch": 0.41, "learning_rate": 1.3440039943027452e-05, "loss": 0.5811, "step": 2113 }, { "epoch": 0.41, "learning_rate": 1.3434187548429126e-05, "loss": 0.5315, "step": 2114 }, { "epoch": 0.41, "learning_rate": 1.3428333820031922e-05, "loss": 0.5242, "step": 2115 }, { "epoch": 0.41, "learning_rate": 1.3422478760109371e-05, "loss": 0.6139, "step": 2116 }, { "epoch": 0.41, "learning_rate": 1.3416622370935507e-05, "loss": 0.5024, "step": 2117 }, { "epoch": 0.41, "learning_rate": 1.3410764654784885e-05, "loss": 0.6538, "step": 2118 }, { "epoch": 0.41, "learning_rate": 1.3404905613932573e-05, "loss": 0.5028, "step": 2119 }, { "epoch": 0.41, "learning_rate": 1.3399045250654152e-05, "loss": 0.6171, "step": 2120 }, { "epoch": 0.41, "learning_rate": 1.3393183567225724e-05, "loss": 0.5979, "step": 2121 }, { "epoch": 0.41, "learning_rate": 1.3387320565923901e-05, "loss": 0.6037, "step": 2122 }, { "epoch": 0.41, "learning_rate": 1.33814562490258e-05, "loss": 0.5935, "step": 2123 }, { "epoch": 0.41, "learning_rate": 1.3375590618809056e-05, "loss": 0.5264, "step": 2124 }, { "epoch": 0.41, "learning_rate": 1.3369723677551813e-05, "loss": 0.5214, "step": 2125 }, { "epoch": 0.41, "learning_rate": 1.3363855427532724e-05, "loss": 0.6236, "step": 2126 }, { "epoch": 0.41, "learning_rate": 1.3357985871030948e-05, "loss": 0.4362, "step": 2127 }, { "epoch": 0.41, "learning_rate": 1.3352115010326155e-05, "loss": 0.598, "step": 2128 }, { "epoch": 0.41, "learning_rate": 1.3346242847698516e-05, "loss": 0.5185, "step": 2129 }, { "epoch": 0.41, "learning_rate": 1.3340369385428713e-05, "loss": 0.6038, "step": 2130 }, { "epoch": 0.41, "learning_rate": 1.3334494625797936e-05, "loss": 0.5708, "step": 2131 }, { "epoch": 0.41, "learning_rate": 1.3328618571087867e-05, "loss": 0.5913, "step": 2132 }, { "epoch": 0.41, "learning_rate": 1.33227412235807e-05, "loss": 0.6017, "step": 2133 }, { "epoch": 0.41, "learning_rate": 1.3316862585559132e-05, "loss": 0.5951, "step": 2134 }, { "epoch": 0.41, "learning_rate": 1.3310982659306352e-05, "loss": 0.5594, "step": 2135 }, { "epoch": 0.41, "learning_rate": 1.3305101447106064e-05, "loss": 0.6098, "step": 2136 }, { "epoch": 0.41, "learning_rate": 1.3299218951242456e-05, "loss": 0.6039, "step": 2137 }, { "epoch": 0.41, "learning_rate": 1.3293335174000226e-05, "loss": 0.6117, "step": 2138 }, { "epoch": 0.41, "learning_rate": 1.328745011766456e-05, "loss": 0.587, "step": 2139 }, { "epoch": 0.41, "learning_rate": 1.3281563784521154e-05, "loss": 0.4867, "step": 2140 }, { "epoch": 0.41, "learning_rate": 1.3275676176856185e-05, "loss": 0.5958, "step": 2141 }, { "epoch": 0.41, "learning_rate": 1.3269787296956333e-05, "loss": 0.5645, "step": 2142 }, { "epoch": 0.41, "learning_rate": 1.3263897147108778e-05, "loss": 0.6294, "step": 2143 }, { "epoch": 0.41, "learning_rate": 1.3258005729601178e-05, "loss": 0.5879, "step": 2144 }, { "epoch": 0.41, "learning_rate": 1.3252113046721692e-05, "loss": 0.6317, "step": 2145 }, { "epoch": 0.41, "learning_rate": 1.3246219100758974e-05, "loss": 0.6305, "step": 2146 }, { "epoch": 0.41, "learning_rate": 1.3240323894002166e-05, "loss": 0.5858, "step": 2147 }, { "epoch": 0.41, "learning_rate": 1.3234427428740895e-05, "loss": 0.5937, "step": 2148 }, { "epoch": 0.41, "learning_rate": 1.3228529707265279e-05, "loss": 0.5176, "step": 2149 }, { "epoch": 0.41, "learning_rate": 1.322263073186593e-05, "loss": 0.6094, "step": 2150 }, { "epoch": 0.41, "learning_rate": 1.3216730504833938e-05, "loss": 0.4976, "step": 2151 }, { "epoch": 0.41, "learning_rate": 1.3210829028460883e-05, "loss": 0.5732, "step": 2152 }, { "epoch": 0.41, "learning_rate": 1.3204926305038832e-05, "loss": 0.6546, "step": 2153 }, { "epoch": 0.41, "learning_rate": 1.3199022336860335e-05, "loss": 0.5947, "step": 2154 }, { "epoch": 0.41, "learning_rate": 1.3193117126218425e-05, "loss": 0.5946, "step": 2155 }, { "epoch": 0.41, "learning_rate": 1.3187210675406617e-05, "loss": 0.629, "step": 2156 }, { "epoch": 0.41, "learning_rate": 1.318130298671891e-05, "loss": 0.6539, "step": 2157 }, { "epoch": 0.42, "learning_rate": 1.3175394062449777e-05, "loss": 0.5644, "step": 2158 }, { "epoch": 0.42, "learning_rate": 1.3169483904894185e-05, "loss": 0.5713, "step": 2159 }, { "epoch": 0.42, "learning_rate": 1.3163572516347565e-05, "loss": 0.5768, "step": 2160 }, { "epoch": 0.42, "learning_rate": 1.3157659899105835e-05, "loss": 0.6063, "step": 2161 }, { "epoch": 0.42, "learning_rate": 1.315174605546538e-05, "loss": 0.648, "step": 2162 }, { "epoch": 0.42, "learning_rate": 1.3145830987723081e-05, "loss": 0.5017, "step": 2163 }, { "epoch": 0.42, "learning_rate": 1.3139914698176273e-05, "loss": 0.6373, "step": 2164 }, { "epoch": 0.42, "learning_rate": 1.3133997189122777e-05, "loss": 0.6109, "step": 2165 }, { "epoch": 0.42, "learning_rate": 1.3128078462860887e-05, "loss": 0.465, "step": 2166 }, { "epoch": 0.42, "learning_rate": 1.3122158521689367e-05, "loss": 0.5528, "step": 2167 }, { "epoch": 0.42, "learning_rate": 1.3116237367907454e-05, "loss": 0.5919, "step": 2168 }, { "epoch": 0.42, "learning_rate": 1.3110315003814855e-05, "loss": 0.6213, "step": 2169 }, { "epoch": 0.42, "learning_rate": 1.3104391431711748e-05, "loss": 0.5875, "step": 2170 }, { "epoch": 0.42, "learning_rate": 1.309846665389878e-05, "loss": 0.6033, "step": 2171 }, { "epoch": 0.42, "learning_rate": 1.309254067267707e-05, "loss": 0.6, "step": 2172 }, { "epoch": 0.42, "learning_rate": 1.3086613490348198e-05, "loss": 0.5804, "step": 2173 }, { "epoch": 0.42, "learning_rate": 1.3080685109214208e-05, "loss": 0.6181, "step": 2174 }, { "epoch": 0.42, "learning_rate": 1.3074755531577628e-05, "loss": 0.5747, "step": 2175 }, { "epoch": 0.42, "learning_rate": 1.3068824759741428e-05, "loss": 0.3578, "step": 2176 }, { "epoch": 0.42, "learning_rate": 1.306289279600905e-05, "loss": 0.5216, "step": 2177 }, { "epoch": 0.42, "learning_rate": 1.3056959642684404e-05, "loss": 0.4459, "step": 2178 }, { "epoch": 0.42, "learning_rate": 1.305102530207186e-05, "loss": 0.6026, "step": 2179 }, { "epoch": 0.42, "learning_rate": 1.3045089776476246e-05, "loss": 0.5562, "step": 2180 }, { "epoch": 0.42, "learning_rate": 1.3039153068202853e-05, "loss": 0.5717, "step": 2181 }, { "epoch": 0.42, "learning_rate": 1.3033215179557424e-05, "loss": 0.5883, "step": 2182 }, { "epoch": 0.42, "learning_rate": 1.3027276112846172e-05, "loss": 0.5856, "step": 2183 }, { "epoch": 0.42, "learning_rate": 1.3021335870375763e-05, "loss": 0.6042, "step": 2184 }, { "epoch": 0.42, "learning_rate": 1.3015394454453316e-05, "loss": 0.571, "step": 2185 }, { "epoch": 0.42, "learning_rate": 1.3009451867386411e-05, "loss": 0.5675, "step": 2186 }, { "epoch": 0.42, "learning_rate": 1.3003508111483077e-05, "loss": 0.6045, "step": 2187 }, { "epoch": 0.42, "learning_rate": 1.29975631890518e-05, "loss": 0.3814, "step": 2188 }, { "epoch": 0.42, "learning_rate": 1.2991617102401524e-05, "loss": 0.5797, "step": 2189 }, { "epoch": 0.42, "learning_rate": 1.2985669853841635e-05, "loss": 0.6382, "step": 2190 }, { "epoch": 0.42, "learning_rate": 1.297972144568198e-05, "loss": 0.6164, "step": 2191 }, { "epoch": 0.42, "learning_rate": 1.2973771880232853e-05, "loss": 0.6015, "step": 2192 }, { "epoch": 0.42, "learning_rate": 1.2967821159804994e-05, "loss": 0.5095, "step": 2193 }, { "epoch": 0.42, "learning_rate": 1.2961869286709594e-05, "loss": 0.5051, "step": 2194 }, { "epoch": 0.42, "learning_rate": 1.295591626325829e-05, "loss": 0.6078, "step": 2195 }, { "epoch": 0.42, "learning_rate": 1.2949962091763174e-05, "loss": 0.6137, "step": 2196 }, { "epoch": 0.42, "learning_rate": 1.2944006774536773e-05, "loss": 0.5459, "step": 2197 }, { "epoch": 0.42, "learning_rate": 1.2938050313892062e-05, "loss": 0.5668, "step": 2198 }, { "epoch": 0.42, "learning_rate": 1.2932092712142468e-05, "loss": 0.5765, "step": 2199 }, { "epoch": 0.42, "learning_rate": 1.292613397160185e-05, "loss": 0.6046, "step": 2200 }, { "epoch": 0.42, "learning_rate": 1.2920174094584514e-05, "loss": 0.6078, "step": 2201 }, { "epoch": 0.42, "learning_rate": 1.2914213083405211e-05, "loss": 0.5898, "step": 2202 }, { "epoch": 0.42, "learning_rate": 1.2908250940379124e-05, "loss": 0.5923, "step": 2203 }, { "epoch": 0.42, "learning_rate": 1.2902287667821885e-05, "loss": 0.6347, "step": 2204 }, { "epoch": 0.42, "learning_rate": 1.289632326804956e-05, "loss": 0.5796, "step": 2205 }, { "epoch": 0.42, "learning_rate": 1.2890357743378649e-05, "loss": 0.5622, "step": 2206 }, { "epoch": 0.42, "learning_rate": 1.2884391096126098e-05, "loss": 0.5756, "step": 2207 }, { "epoch": 0.42, "learning_rate": 1.2878423328609281e-05, "loss": 0.6553, "step": 2208 }, { "epoch": 0.42, "learning_rate": 1.2872454443146015e-05, "loss": 0.5825, "step": 2209 }, { "epoch": 0.43, "learning_rate": 1.286648444205454e-05, "loss": 0.6418, "step": 2210 }, { "epoch": 0.43, "learning_rate": 1.2860513327653537e-05, "loss": 0.563, "step": 2211 }, { "epoch": 0.43, "learning_rate": 1.2854541102262119e-05, "loss": 0.5752, "step": 2212 }, { "epoch": 0.43, "learning_rate": 1.284856776819983e-05, "loss": 0.5923, "step": 2213 }, { "epoch": 0.43, "learning_rate": 1.2842593327786649e-05, "loss": 0.6476, "step": 2214 }, { "epoch": 0.43, "learning_rate": 1.2836617783342968e-05, "loss": 0.5708, "step": 2215 }, { "epoch": 0.43, "learning_rate": 1.2830641137189628e-05, "loss": 0.5698, "step": 2216 }, { "epoch": 0.43, "learning_rate": 1.282466339164789e-05, "loss": 0.6124, "step": 2217 }, { "epoch": 0.43, "learning_rate": 1.2818684549039437e-05, "loss": 0.5999, "step": 2218 }, { "epoch": 0.43, "learning_rate": 1.2812704611686386e-05, "loss": 0.598, "step": 2219 }, { "epoch": 0.43, "learning_rate": 1.2806723581911274e-05, "loss": 0.5284, "step": 2220 }, { "epoch": 0.43, "learning_rate": 1.2800741462037065e-05, "loss": 0.6123, "step": 2221 }, { "epoch": 0.43, "learning_rate": 1.2794758254387147e-05, "loss": 0.5808, "step": 2222 }, { "epoch": 0.43, "learning_rate": 1.2788773961285323e-05, "loss": 0.5786, "step": 2223 }, { "epoch": 0.43, "learning_rate": 1.2782788585055829e-05, "loss": 0.6143, "step": 2224 }, { "epoch": 0.43, "learning_rate": 1.2776802128023317e-05, "loss": 0.6236, "step": 2225 }, { "epoch": 0.43, "learning_rate": 1.2770814592512853e-05, "loss": 0.5854, "step": 2226 }, { "epoch": 0.43, "learning_rate": 1.2764825980849931e-05, "loss": 0.6239, "step": 2227 }, { "epoch": 0.43, "learning_rate": 1.2758836295360455e-05, "loss": 0.5655, "step": 2228 }, { "epoch": 0.43, "learning_rate": 1.2752845538370752e-05, "loss": 0.5885, "step": 2229 }, { "epoch": 0.43, "learning_rate": 1.2746853712207567e-05, "loss": 0.5514, "step": 2230 }, { "epoch": 0.43, "learning_rate": 1.274086081919805e-05, "loss": 0.5078, "step": 2231 }, { "epoch": 0.43, "learning_rate": 1.273486686166977e-05, "loss": 0.6252, "step": 2232 }, { "epoch": 0.43, "learning_rate": 1.2728871841950719e-05, "loss": 0.6586, "step": 2233 }, { "epoch": 0.43, "learning_rate": 1.2722875762369288e-05, "loss": 0.5949, "step": 2234 }, { "epoch": 0.43, "learning_rate": 1.2716878625254287e-05, "loss": 0.5931, "step": 2235 }, { "epoch": 0.43, "learning_rate": 1.2710880432934934e-05, "loss": 0.4396, "step": 2236 }, { "epoch": 0.43, "learning_rate": 1.270488118774086e-05, "loss": 0.3731, "step": 2237 }, { "epoch": 0.43, "learning_rate": 1.26988808920021e-05, "loss": 0.5721, "step": 2238 }, { "epoch": 0.43, "learning_rate": 1.26928795480491e-05, "loss": 0.6065, "step": 2239 }, { "epoch": 0.43, "learning_rate": 1.2686877158212715e-05, "loss": 0.6609, "step": 2240 }, { "epoch": 0.43, "learning_rate": 1.26808737248242e-05, "loss": 0.6128, "step": 2241 }, { "epoch": 0.43, "learning_rate": 1.2674869250215225e-05, "loss": 0.6478, "step": 2242 }, { "epoch": 0.43, "learning_rate": 1.2668863736717855e-05, "loss": 0.5843, "step": 2243 }, { "epoch": 0.43, "learning_rate": 1.2662857186664558e-05, "loss": 0.6142, "step": 2244 }, { "epoch": 0.43, "learning_rate": 1.2656849602388222e-05, "loss": 0.646, "step": 2245 }, { "epoch": 0.43, "learning_rate": 1.2650840986222111e-05, "loss": 0.6368, "step": 2246 }, { "epoch": 0.43, "learning_rate": 1.2644831340499906e-05, "loss": 0.5928, "step": 2247 }, { "epoch": 0.43, "learning_rate": 1.2638820667555685e-05, "loss": 0.625, "step": 2248 }, { "epoch": 0.43, "learning_rate": 1.2632808969723927e-05, "loss": 0.606, "step": 2249 }, { "epoch": 0.43, "learning_rate": 1.26267962493395e-05, "loss": 0.485, "step": 2250 }, { "epoch": 0.43, "learning_rate": 1.2620782508737678e-05, "loss": 0.591, "step": 2251 }, { "epoch": 0.43, "learning_rate": 1.2614767750254129e-05, "loss": 0.6606, "step": 2252 }, { "epoch": 0.43, "learning_rate": 1.2608751976224916e-05, "loss": 0.579, "step": 2253 }, { "epoch": 0.43, "learning_rate": 1.2602735188986498e-05, "loss": 0.5938, "step": 2254 }, { "epoch": 0.43, "learning_rate": 1.2596717390875721e-05, "loss": 0.4533, "step": 2255 }, { "epoch": 0.43, "learning_rate": 1.2590698584229834e-05, "loss": 0.5894, "step": 2256 }, { "epoch": 0.43, "learning_rate": 1.2584678771386467e-05, "loss": 0.3919, "step": 2257 }, { "epoch": 0.43, "learning_rate": 1.2578657954683651e-05, "loss": 0.5943, "step": 2258 }, { "epoch": 0.43, "learning_rate": 1.2572636136459799e-05, "loss": 0.585, "step": 2259 }, { "epoch": 0.43, "learning_rate": 1.2566613319053713e-05, "loss": 0.6071, "step": 2260 }, { "epoch": 0.43, "learning_rate": 1.2560589504804592e-05, "loss": 0.6201, "step": 2261 }, { "epoch": 0.44, "learning_rate": 1.2554564696052011e-05, "loss": 0.4951, "step": 2262 }, { "epoch": 0.44, "learning_rate": 1.2548538895135942e-05, "loss": 0.5885, "step": 2263 }, { "epoch": 0.44, "learning_rate": 1.254251210439673e-05, "loss": 0.6257, "step": 2264 }, { "epoch": 0.44, "learning_rate": 1.2536484326175114e-05, "loss": 0.6247, "step": 2265 }, { "epoch": 0.44, "learning_rate": 1.2530455562812214e-05, "loss": 0.5855, "step": 2266 }, { "epoch": 0.44, "learning_rate": 1.252442581664953e-05, "loss": 0.6066, "step": 2267 }, { "epoch": 0.44, "learning_rate": 1.2518395090028952e-05, "loss": 0.5854, "step": 2268 }, { "epoch": 0.44, "learning_rate": 1.2512363385292739e-05, "loss": 0.6272, "step": 2269 }, { "epoch": 0.44, "learning_rate": 1.2506330704783533e-05, "loss": 0.6136, "step": 2270 }, { "epoch": 0.44, "learning_rate": 1.2500297050844367e-05, "loss": 0.6047, "step": 2271 }, { "epoch": 0.44, "learning_rate": 1.2494262425818637e-05, "loss": 0.5709, "step": 2272 }, { "epoch": 0.44, "learning_rate": 1.2488226832050116e-05, "loss": 0.6069, "step": 2273 }, { "epoch": 0.44, "learning_rate": 1.2482190271882973e-05, "loss": 0.4847, "step": 2274 }, { "epoch": 0.44, "learning_rate": 1.2476152747661727e-05, "loss": 0.6205, "step": 2275 }, { "epoch": 0.44, "learning_rate": 1.2470114261731288e-05, "loss": 0.5862, "step": 2276 }, { "epoch": 0.44, "learning_rate": 1.246407481643693e-05, "loss": 0.6279, "step": 2277 }, { "epoch": 0.44, "learning_rate": 1.245803441412431e-05, "loss": 0.5822, "step": 2278 }, { "epoch": 0.44, "learning_rate": 1.2451993057139445e-05, "loss": 0.6344, "step": 2279 }, { "epoch": 0.44, "learning_rate": 1.2445950747828732e-05, "loss": 0.6312, "step": 2280 }, { "epoch": 0.44, "learning_rate": 1.2439907488538934e-05, "loss": 0.5788, "step": 2281 }, { "epoch": 0.44, "learning_rate": 1.243386328161718e-05, "loss": 0.6017, "step": 2282 }, { "epoch": 0.44, "learning_rate": 1.2427818129410975e-05, "loss": 0.5812, "step": 2283 }, { "epoch": 0.44, "learning_rate": 1.2421772034268187e-05, "loss": 0.6121, "step": 2284 }, { "epoch": 0.44, "learning_rate": 1.2415724998537042e-05, "loss": 0.4934, "step": 2285 }, { "epoch": 0.44, "learning_rate": 1.2409677024566145e-05, "loss": 0.6432, "step": 2286 }, { "epoch": 0.44, "learning_rate": 1.240362811470446e-05, "loss": 0.6105, "step": 2287 }, { "epoch": 0.44, "learning_rate": 1.2397578271301312e-05, "loss": 0.4952, "step": 2288 }, { "epoch": 0.44, "learning_rate": 1.2391527496706389e-05, "loss": 0.6112, "step": 2289 }, { "epoch": 0.44, "learning_rate": 1.2385475793269744e-05, "loss": 0.599, "step": 2290 }, { "epoch": 0.44, "learning_rate": 1.2379423163341791e-05, "loss": 0.5798, "step": 2291 }, { "epoch": 0.44, "learning_rate": 1.2373369609273299e-05, "loss": 0.5877, "step": 2292 }, { "epoch": 0.44, "learning_rate": 1.2367315133415396e-05, "loss": 0.4022, "step": 2293 }, { "epoch": 0.44, "learning_rate": 1.2361259738119575e-05, "loss": 0.6053, "step": 2294 }, { "epoch": 0.44, "learning_rate": 1.2355203425737683e-05, "loss": 0.596, "step": 2295 }, { "epoch": 0.44, "learning_rate": 1.2349146198621917e-05, "loss": 0.5886, "step": 2296 }, { "epoch": 0.44, "learning_rate": 1.2343088059124839e-05, "loss": 0.6084, "step": 2297 }, { "epoch": 0.44, "learning_rate": 1.2337029009599357e-05, "loss": 0.4843, "step": 2298 }, { "epoch": 0.44, "learning_rate": 1.2330969052398735e-05, "loss": 0.5521, "step": 2299 }, { "epoch": 0.44, "learning_rate": 1.2324908189876597e-05, "loss": 0.5983, "step": 2300 }, { "epoch": 0.44, "learning_rate": 1.2318846424386907e-05, "loss": 0.6191, "step": 2301 }, { "epoch": 0.44, "learning_rate": 1.2312783758283981e-05, "loss": 0.5636, "step": 2302 }, { "epoch": 0.44, "learning_rate": 1.23067201939225e-05, "loss": 0.5896, "step": 2303 }, { "epoch": 0.44, "learning_rate": 1.2300655733657475e-05, "loss": 0.6041, "step": 2304 }, { "epoch": 0.44, "learning_rate": 1.2294590379844268e-05, "loss": 0.6241, "step": 2305 }, { "epoch": 0.44, "learning_rate": 1.2288524134838602e-05, "loss": 0.5443, "step": 2306 }, { "epoch": 0.44, "learning_rate": 1.2282457000996533e-05, "loss": 0.6117, "step": 2307 }, { "epoch": 0.44, "learning_rate": 1.2276388980674465e-05, "loss": 0.6043, "step": 2308 }, { "epoch": 0.44, "learning_rate": 1.227032007622915e-05, "loss": 0.6153, "step": 2309 }, { "epoch": 0.44, "learning_rate": 1.2264250290017675e-05, "loss": 0.5723, "step": 2310 }, { "epoch": 0.44, "learning_rate": 1.2258179624397477e-05, "loss": 0.4797, "step": 2311 }, { "epoch": 0.44, "learning_rate": 1.2252108081726337e-05, "loss": 0.5289, "step": 2312 }, { "epoch": 0.45, "learning_rate": 1.224603566436237e-05, "loss": 0.6021, "step": 2313 }, { "epoch": 0.45, "learning_rate": 1.2239962374664029e-05, "loss": 0.6445, "step": 2314 }, { "epoch": 0.45, "learning_rate": 1.2233888214990113e-05, "loss": 0.5241, "step": 2315 }, { "epoch": 0.45, "learning_rate": 1.2227813187699757e-05, "loss": 0.5687, "step": 2316 }, { "epoch": 0.45, "learning_rate": 1.222173729515243e-05, "loss": 0.6088, "step": 2317 }, { "epoch": 0.45, "learning_rate": 1.2215660539707936e-05, "loss": 0.3985, "step": 2318 }, { "epoch": 0.45, "learning_rate": 1.2209582923726424e-05, "loss": 0.6163, "step": 2319 }, { "epoch": 0.45, "learning_rate": 1.2203504449568361e-05, "loss": 0.5898, "step": 2320 }, { "epoch": 0.45, "learning_rate": 1.2197425119594563e-05, "loss": 0.6022, "step": 2321 }, { "epoch": 0.45, "learning_rate": 1.219134493616617e-05, "loss": 0.5894, "step": 2322 }, { "epoch": 0.45, "learning_rate": 1.2185263901644653e-05, "loss": 0.472, "step": 2323 }, { "epoch": 0.45, "learning_rate": 1.217918201839182e-05, "loss": 0.5963, "step": 2324 }, { "epoch": 0.45, "learning_rate": 1.2173099288769799e-05, "loss": 0.5574, "step": 2325 }, { "epoch": 0.45, "learning_rate": 1.2167015715141057e-05, "loss": 0.48, "step": 2326 }, { "epoch": 0.45, "learning_rate": 1.216093129986838e-05, "loss": 0.6018, "step": 2327 }, { "epoch": 0.45, "learning_rate": 1.2154846045314885e-05, "loss": 0.5056, "step": 2328 }, { "epoch": 0.45, "learning_rate": 1.214875995384402e-05, "loss": 0.5892, "step": 2329 }, { "epoch": 0.45, "learning_rate": 1.214267302781955e-05, "loss": 0.5484, "step": 2330 }, { "epoch": 0.45, "learning_rate": 1.2136585269605558e-05, "loss": 0.5127, "step": 2331 }, { "epoch": 0.45, "learning_rate": 1.2130496681566475e-05, "loss": 0.5929, "step": 2332 }, { "epoch": 0.45, "learning_rate": 1.212440726606703e-05, "loss": 0.5635, "step": 2333 }, { "epoch": 0.45, "learning_rate": 1.211831702547228e-05, "loss": 0.6038, "step": 2334 }, { "epoch": 0.45, "learning_rate": 1.2112225962147605e-05, "loss": 0.5093, "step": 2335 }, { "epoch": 0.45, "learning_rate": 1.210613407845871e-05, "loss": 0.506, "step": 2336 }, { "epoch": 0.45, "learning_rate": 1.2100041376771605e-05, "loss": 0.5891, "step": 2337 }, { "epoch": 0.45, "learning_rate": 1.209394785945263e-05, "loss": 0.6406, "step": 2338 }, { "epoch": 0.45, "learning_rate": 1.2087853528868432e-05, "loss": 0.5662, "step": 2339 }, { "epoch": 0.45, "learning_rate": 1.2081758387385982e-05, "loss": 0.6426, "step": 2340 }, { "epoch": 0.45, "learning_rate": 1.2075662437372567e-05, "loss": 0.5013, "step": 2341 }, { "epoch": 0.45, "learning_rate": 1.2069565681195776e-05, "loss": 0.6042, "step": 2342 }, { "epoch": 0.45, "learning_rate": 1.206346812122352e-05, "loss": 0.6008, "step": 2343 }, { "epoch": 0.45, "learning_rate": 1.2057369759824025e-05, "loss": 0.4889, "step": 2344 }, { "epoch": 0.45, "learning_rate": 1.2051270599365825e-05, "loss": 0.6136, "step": 2345 }, { "epoch": 0.45, "learning_rate": 1.2045170642217756e-05, "loss": 0.621, "step": 2346 }, { "epoch": 0.45, "learning_rate": 1.2039069890748978e-05, "loss": 0.5803, "step": 2347 }, { "epoch": 0.45, "learning_rate": 1.2032968347328952e-05, "loss": 0.4927, "step": 2348 }, { "epoch": 0.45, "learning_rate": 1.2026866014327446e-05, "loss": 0.4749, "step": 2349 }, { "epoch": 0.45, "learning_rate": 1.2020762894114535e-05, "loss": 0.4636, "step": 2350 }, { "epoch": 0.45, "learning_rate": 1.20146589890606e-05, "loss": 0.5833, "step": 2351 }, { "epoch": 0.45, "learning_rate": 1.2008554301536328e-05, "loss": 0.5948, "step": 2352 }, { "epoch": 0.45, "learning_rate": 1.2002448833912712e-05, "loss": 0.5159, "step": 2353 }, { "epoch": 0.45, "learning_rate": 1.1996342588561042e-05, "loss": 0.4897, "step": 2354 }, { "epoch": 0.45, "learning_rate": 1.1990235567852917e-05, "loss": 0.605, "step": 2355 }, { "epoch": 0.45, "learning_rate": 1.1984127774160226e-05, "loss": 0.5697, "step": 2356 }, { "epoch": 0.45, "learning_rate": 1.1978019209855174e-05, "loss": 0.6433, "step": 2357 }, { "epoch": 0.45, "learning_rate": 1.1971909877310253e-05, "loss": 0.5957, "step": 2358 }, { "epoch": 0.45, "learning_rate": 1.1965799778898258e-05, "loss": 0.6361, "step": 2359 }, { "epoch": 0.45, "learning_rate": 1.1959688916992279e-05, "loss": 0.6224, "step": 2360 }, { "epoch": 0.45, "learning_rate": 1.1953577293965707e-05, "loss": 0.4187, "step": 2361 }, { "epoch": 0.45, "learning_rate": 1.1947464912192228e-05, "loss": 0.5968, "step": 2362 }, { "epoch": 0.45, "learning_rate": 1.1941351774045815e-05, "loss": 0.5846, "step": 2363 }, { "epoch": 0.45, "learning_rate": 1.1935237881900743e-05, "loss": 0.5614, "step": 2364 }, { "epoch": 0.46, "learning_rate": 1.1929123238131579e-05, "loss": 0.6423, "step": 2365 }, { "epoch": 0.46, "learning_rate": 1.1923007845113178e-05, "loss": 0.5773, "step": 2366 }, { "epoch": 0.46, "learning_rate": 1.1916891705220689e-05, "loss": 0.447, "step": 2367 }, { "epoch": 0.46, "learning_rate": 1.191077482082955e-05, "loss": 0.5777, "step": 2368 }, { "epoch": 0.46, "learning_rate": 1.1904657194315486e-05, "loss": 0.6207, "step": 2369 }, { "epoch": 0.46, "learning_rate": 1.1898538828054517e-05, "loss": 0.5934, "step": 2370 }, { "epoch": 0.46, "learning_rate": 1.1892419724422946e-05, "loss": 0.495, "step": 2371 }, { "epoch": 0.46, "learning_rate": 1.1886299885797357e-05, "loss": 0.4871, "step": 2372 }, { "epoch": 0.46, "learning_rate": 1.1880179314554629e-05, "loss": 0.4744, "step": 2373 }, { "epoch": 0.46, "learning_rate": 1.1874058013071923e-05, "loss": 0.6174, "step": 2374 }, { "epoch": 0.46, "learning_rate": 1.1867935983726676e-05, "loss": 0.4657, "step": 2375 }, { "epoch": 0.46, "learning_rate": 1.186181322889662e-05, "loss": 0.5892, "step": 2376 }, { "epoch": 0.46, "learning_rate": 1.1855689750959759e-05, "loss": 0.5678, "step": 2377 }, { "epoch": 0.46, "learning_rate": 1.1849565552294379e-05, "loss": 0.5976, "step": 2378 }, { "epoch": 0.46, "learning_rate": 1.1843440635279056e-05, "loss": 0.5893, "step": 2379 }, { "epoch": 0.46, "learning_rate": 1.1837315002292629e-05, "loss": 0.6107, "step": 2380 }, { "epoch": 0.46, "learning_rate": 1.1831188655714225e-05, "loss": 0.554, "step": 2381 }, { "epoch": 0.46, "learning_rate": 1.182506159792325e-05, "loss": 0.6153, "step": 2382 }, { "epoch": 0.46, "learning_rate": 1.1818933831299381e-05, "loss": 0.5035, "step": 2383 }, { "epoch": 0.46, "learning_rate": 1.1812805358222571e-05, "loss": 0.4553, "step": 2384 }, { "epoch": 0.46, "learning_rate": 1.180667618107305e-05, "loss": 0.576, "step": 2385 }, { "epoch": 0.46, "learning_rate": 1.1800546302231317e-05, "loss": 0.6049, "step": 2386 }, { "epoch": 0.46, "learning_rate": 1.1794415724078147e-05, "loss": 0.5805, "step": 2387 }, { "epoch": 0.46, "learning_rate": 1.1788284448994588e-05, "loss": 0.5166, "step": 2388 }, { "epoch": 0.46, "learning_rate": 1.1782152479361956e-05, "loss": 0.5749, "step": 2389 }, { "epoch": 0.46, "learning_rate": 1.1776019817561834e-05, "loss": 0.5899, "step": 2390 }, { "epoch": 0.46, "learning_rate": 1.1769886465976086e-05, "loss": 0.5599, "step": 2391 }, { "epoch": 0.46, "learning_rate": 1.1763752426986823e-05, "loss": 0.5881, "step": 2392 }, { "epoch": 0.46, "learning_rate": 1.1757617702976443e-05, "loss": 0.5956, "step": 2393 }, { "epoch": 0.46, "learning_rate": 1.17514822963276e-05, "loss": 0.5878, "step": 2394 }, { "epoch": 0.46, "learning_rate": 1.1745346209423216e-05, "loss": 0.5918, "step": 2395 }, { "epoch": 0.46, "learning_rate": 1.1739209444646479e-05, "loss": 0.5582, "step": 2396 }, { "epoch": 0.46, "learning_rate": 1.1733072004380827e-05, "loss": 0.5007, "step": 2397 }, { "epoch": 0.46, "learning_rate": 1.1726933891009985e-05, "loss": 0.5708, "step": 2398 }, { "epoch": 0.46, "learning_rate": 1.1720795106917917e-05, "loss": 0.5489, "step": 2399 }, { "epoch": 0.46, "learning_rate": 1.171465565448886e-05, "loss": 0.6066, "step": 2400 }, { "epoch": 0.46, "learning_rate": 1.1708515536107299e-05, "loss": 0.6038, "step": 2401 }, { "epoch": 0.46, "learning_rate": 1.1702374754157998e-05, "loss": 0.5022, "step": 2402 }, { "epoch": 0.46, "learning_rate": 1.1696233311025957e-05, "loss": 0.6661, "step": 2403 }, { "epoch": 0.46, "learning_rate": 1.1690091209096441e-05, "loss": 0.5932, "step": 2404 }, { "epoch": 0.46, "learning_rate": 1.1683948450754976e-05, "loss": 0.602, "step": 2405 }, { "epoch": 0.46, "learning_rate": 1.1677805038387337e-05, "loss": 0.5749, "step": 2406 }, { "epoch": 0.46, "learning_rate": 1.1671660974379554e-05, "loss": 0.6093, "step": 2407 }, { "epoch": 0.46, "learning_rate": 1.1665516261117914e-05, "loss": 0.5437, "step": 2408 }, { "epoch": 0.46, "learning_rate": 1.1659370900988946e-05, "loss": 0.5883, "step": 2409 }, { "epoch": 0.46, "learning_rate": 1.165322489637944e-05, "loss": 0.4977, "step": 2410 }, { "epoch": 0.46, "learning_rate": 1.164707824967644e-05, "loss": 0.6271, "step": 2411 }, { "epoch": 0.46, "learning_rate": 1.1640930963267226e-05, "loss": 0.4974, "step": 2412 }, { "epoch": 0.46, "learning_rate": 1.1634783039539328e-05, "loss": 0.5079, "step": 2413 }, { "epoch": 0.46, "learning_rate": 1.162863448088054e-05, "loss": 0.4929, "step": 2414 }, { "epoch": 0.46, "learning_rate": 1.1622485289678886e-05, "loss": 0.5717, "step": 2415 }, { "epoch": 0.46, "learning_rate": 1.1616335468322641e-05, "loss": 0.6385, "step": 2416 }, { "epoch": 0.47, "learning_rate": 1.1610185019200324e-05, "loss": 0.6055, "step": 2417 }, { "epoch": 0.47, "learning_rate": 1.1604033944700701e-05, "loss": 0.6101, "step": 2418 }, { "epoch": 0.47, "learning_rate": 1.1597882247212776e-05, "loss": 0.5121, "step": 2419 }, { "epoch": 0.47, "learning_rate": 1.15917299291258e-05, "loss": 0.6182, "step": 2420 }, { "epoch": 0.47, "learning_rate": 1.1585576992829261e-05, "loss": 0.5843, "step": 2421 }, { "epoch": 0.47, "learning_rate": 1.1579423440712887e-05, "loss": 0.4731, "step": 2422 }, { "epoch": 0.47, "learning_rate": 1.1573269275166652e-05, "loss": 0.4058, "step": 2423 }, { "epoch": 0.47, "learning_rate": 1.1567114498580758e-05, "loss": 0.495, "step": 2424 }, { "epoch": 0.47, "learning_rate": 1.1560959113345649e-05, "loss": 0.5128, "step": 2425 }, { "epoch": 0.47, "learning_rate": 1.1554803121852005e-05, "loss": 0.5672, "step": 2426 }, { "epoch": 0.47, "learning_rate": 1.1548646526490749e-05, "loss": 0.5722, "step": 2427 }, { "epoch": 0.47, "learning_rate": 1.1542489329653024e-05, "loss": 0.5121, "step": 2428 }, { "epoch": 0.47, "learning_rate": 1.153633153373022e-05, "loss": 0.4776, "step": 2429 }, { "epoch": 0.47, "learning_rate": 1.1530173141113947e-05, "loss": 0.6058, "step": 2430 }, { "epoch": 0.47, "learning_rate": 1.1524014154196063e-05, "loss": 0.5954, "step": 2431 }, { "epoch": 0.47, "learning_rate": 1.1517854575368644e-05, "loss": 0.6298, "step": 2432 }, { "epoch": 0.47, "learning_rate": 1.1511694407023994e-05, "loss": 0.5083, "step": 2433 }, { "epoch": 0.47, "learning_rate": 1.1505533651554654e-05, "loss": 0.4812, "step": 2434 }, { "epoch": 0.47, "learning_rate": 1.1499372311353398e-05, "loss": 0.5971, "step": 2435 }, { "epoch": 0.47, "learning_rate": 1.149321038881321e-05, "loss": 0.5969, "step": 2436 }, { "epoch": 0.47, "learning_rate": 1.1487047886327314e-05, "loss": 0.6015, "step": 2437 }, { "epoch": 0.47, "learning_rate": 1.1480884806289151e-05, "loss": 0.6081, "step": 2438 }, { "epoch": 0.47, "learning_rate": 1.1474721151092397e-05, "loss": 0.5722, "step": 2439 }, { "epoch": 0.47, "learning_rate": 1.1468556923130943e-05, "loss": 0.4567, "step": 2440 }, { "epoch": 0.47, "learning_rate": 1.14623921247989e-05, "loss": 0.5977, "step": 2441 }, { "epoch": 0.47, "learning_rate": 1.1456226758490603e-05, "loss": 0.5931, "step": 2442 }, { "epoch": 0.47, "learning_rate": 1.1450060826600618e-05, "loss": 0.5443, "step": 2443 }, { "epoch": 0.47, "learning_rate": 1.1443894331523718e-05, "loss": 0.585, "step": 2444 }, { "epoch": 0.47, "learning_rate": 1.1437727275654893e-05, "loss": 0.4729, "step": 2445 }, { "epoch": 0.47, "learning_rate": 1.1431559661389362e-05, "loss": 0.6259, "step": 2446 }, { "epoch": 0.47, "learning_rate": 1.1425391491122557e-05, "loss": 0.6184, "step": 2447 }, { "epoch": 0.47, "learning_rate": 1.141922276725012e-05, "loss": 0.5398, "step": 2448 }, { "epoch": 0.47, "learning_rate": 1.1413053492167915e-05, "loss": 0.6178, "step": 2449 }, { "epoch": 0.47, "learning_rate": 1.1406883668272015e-05, "loss": 0.4405, "step": 2450 }, { "epoch": 0.47, "learning_rate": 1.140071329795871e-05, "loss": 0.5802, "step": 2451 }, { "epoch": 0.47, "learning_rate": 1.13945423836245e-05, "loss": 0.5842, "step": 2452 }, { "epoch": 0.47, "learning_rate": 1.1388370927666102e-05, "loss": 0.5678, "step": 2453 }, { "epoch": 0.47, "learning_rate": 1.1382198932480429e-05, "loss": 0.6028, "step": 2454 }, { "epoch": 0.47, "learning_rate": 1.1376026400464616e-05, "loss": 0.5669, "step": 2455 }, { "epoch": 0.47, "learning_rate": 1.136985333401601e-05, "loss": 0.5128, "step": 2456 }, { "epoch": 0.47, "learning_rate": 1.1363679735532151e-05, "loss": 0.5883, "step": 2457 }, { "epoch": 0.47, "learning_rate": 1.1357505607410797e-05, "loss": 0.6134, "step": 2458 }, { "epoch": 0.47, "learning_rate": 1.1351330952049908e-05, "loss": 0.6321, "step": 2459 }, { "epoch": 0.47, "learning_rate": 1.1345155771847646e-05, "loss": 0.5231, "step": 2460 }, { "epoch": 0.47, "learning_rate": 1.1338980069202388e-05, "loss": 0.5844, "step": 2461 }, { "epoch": 0.47, "learning_rate": 1.1332803846512697e-05, "loss": 0.5708, "step": 2462 }, { "epoch": 0.47, "learning_rate": 1.1326627106177348e-05, "loss": 0.5674, "step": 2463 }, { "epoch": 0.47, "learning_rate": 1.132044985059532e-05, "loss": 0.4822, "step": 2464 }, { "epoch": 0.47, "learning_rate": 1.1314272082165785e-05, "loss": 0.5825, "step": 2465 }, { "epoch": 0.47, "learning_rate": 1.1308093803288119e-05, "loss": 0.584, "step": 2466 }, { "epoch": 0.47, "learning_rate": 1.130191501636189e-05, "loss": 0.6121, "step": 2467 }, { "epoch": 0.47, "learning_rate": 1.1295735723786872e-05, "loss": 0.5755, "step": 2468 }, { "epoch": 0.48, "learning_rate": 1.1289555927963032e-05, "loss": 0.5069, "step": 2469 }, { "epoch": 0.48, "learning_rate": 1.1283375631290528e-05, "loss": 0.5803, "step": 2470 }, { "epoch": 0.48, "learning_rate": 1.1277194836169714e-05, "loss": 0.5109, "step": 2471 }, { "epoch": 0.48, "learning_rate": 1.1271013545001144e-05, "loss": 0.5438, "step": 2472 }, { "epoch": 0.48, "learning_rate": 1.1264831760185562e-05, "loss": 0.6031, "step": 2473 }, { "epoch": 0.48, "learning_rate": 1.1258649484123895e-05, "loss": 0.4573, "step": 2474 }, { "epoch": 0.48, "learning_rate": 1.1252466719217274e-05, "loss": 0.6277, "step": 2475 }, { "epoch": 0.48, "learning_rate": 1.1246283467867012e-05, "loss": 0.6192, "step": 2476 }, { "epoch": 0.48, "learning_rate": 1.1240099732474613e-05, "loss": 0.618, "step": 2477 }, { "epoch": 0.48, "learning_rate": 1.1233915515441765e-05, "loss": 0.6302, "step": 2478 }, { "epoch": 0.48, "learning_rate": 1.1227730819170349e-05, "loss": 0.5341, "step": 2479 }, { "epoch": 0.48, "learning_rate": 1.1221545646062431e-05, "loss": 0.5778, "step": 2480 }, { "epoch": 0.48, "learning_rate": 1.121535999852026e-05, "loss": 0.397, "step": 2481 }, { "epoch": 0.48, "learning_rate": 1.1209173878946271e-05, "loss": 0.5841, "step": 2482 }, { "epoch": 0.48, "learning_rate": 1.1202987289743078e-05, "loss": 0.4813, "step": 2483 }, { "epoch": 0.48, "learning_rate": 1.1196800233313488e-05, "loss": 0.6134, "step": 2484 }, { "epoch": 0.48, "learning_rate": 1.1190612712060475e-05, "loss": 0.6038, "step": 2485 }, { "epoch": 0.48, "learning_rate": 1.1184424728387204e-05, "loss": 0.5771, "step": 2486 }, { "epoch": 0.48, "learning_rate": 1.1178236284697017e-05, "loss": 0.5623, "step": 2487 }, { "epoch": 0.48, "learning_rate": 1.1172047383393434e-05, "loss": 0.5843, "step": 2488 }, { "epoch": 0.48, "learning_rate": 1.1165858026880151e-05, "loss": 0.6149, "step": 2489 }, { "epoch": 0.48, "learning_rate": 1.1159668217561048e-05, "loss": 0.5581, "step": 2490 }, { "epoch": 0.48, "learning_rate": 1.115347795784017e-05, "loss": 0.609, "step": 2491 }, { "epoch": 0.48, "learning_rate": 1.1147287250121745e-05, "loss": 0.5712, "step": 2492 }, { "epoch": 0.48, "learning_rate": 1.1141096096810174e-05, "loss": 0.5288, "step": 2493 }, { "epoch": 0.48, "learning_rate": 1.1134904500310029e-05, "loss": 0.5607, "step": 2494 }, { "epoch": 0.48, "learning_rate": 1.1128712463026048e-05, "loss": 0.6152, "step": 2495 }, { "epoch": 0.48, "learning_rate": 1.1122519987363156e-05, "loss": 0.6474, "step": 2496 }, { "epoch": 0.48, "learning_rate": 1.1116327075726436e-05, "loss": 0.5286, "step": 2497 }, { "epoch": 0.48, "learning_rate": 1.1110133730521142e-05, "loss": 0.5531, "step": 2498 }, { "epoch": 0.48, "learning_rate": 1.11039399541527e-05, "loss": 0.6056, "step": 2499 }, { "epoch": 0.48, "learning_rate": 1.10977457490267e-05, "loss": 0.5965, "step": 2500 }, { "epoch": 0.48, "learning_rate": 1.10915511175489e-05, "loss": 0.5849, "step": 2501 }, { "epoch": 0.48, "learning_rate": 1.1085356062125225e-05, "loss": 0.5864, "step": 2502 }, { "epoch": 0.48, "learning_rate": 1.1079160585161759e-05, "loss": 0.603, "step": 2503 }, { "epoch": 0.48, "learning_rate": 1.107296468906476e-05, "loss": 0.5588, "step": 2504 }, { "epoch": 0.48, "learning_rate": 1.106676837624064e-05, "loss": 0.477, "step": 2505 }, { "epoch": 0.48, "learning_rate": 1.1060571649095972e-05, "loss": 0.5988, "step": 2506 }, { "epoch": 0.48, "learning_rate": 1.10543745100375e-05, "loss": 0.6151, "step": 2507 }, { "epoch": 0.48, "learning_rate": 1.1048176961472114e-05, "loss": 0.5407, "step": 2508 }, { "epoch": 0.48, "learning_rate": 1.1041979005806876e-05, "loss": 0.6438, "step": 2509 }, { "epoch": 0.48, "learning_rate": 1.1035780645449001e-05, "loss": 0.6125, "step": 2510 }, { "epoch": 0.48, "learning_rate": 1.1029581882805857e-05, "loss": 0.6637, "step": 2511 }, { "epoch": 0.48, "learning_rate": 1.1023382720284973e-05, "loss": 0.576, "step": 2512 }, { "epoch": 0.48, "learning_rate": 1.1017183160294033e-05, "loss": 0.5783, "step": 2513 }, { "epoch": 0.48, "learning_rate": 1.1010983205240878e-05, "loss": 0.6196, "step": 2514 }, { "epoch": 0.48, "learning_rate": 1.1004782857533488e-05, "loss": 0.553, "step": 2515 }, { "epoch": 0.48, "learning_rate": 1.099858211958002e-05, "loss": 0.477, "step": 2516 }, { "epoch": 0.48, "learning_rate": 1.0992380993788763e-05, "loss": 0.4133, "step": 2517 }, { "epoch": 0.48, "learning_rate": 1.0986179482568162e-05, "loss": 0.6379, "step": 2518 }, { "epoch": 0.48, "learning_rate": 1.0979977588326815e-05, "loss": 0.5856, "step": 2519 }, { "epoch": 0.48, "learning_rate": 1.0973775313473465e-05, "loss": 0.6107, "step": 2520 }, { "epoch": 0.49, "learning_rate": 1.0967572660417001e-05, "loss": 0.4647, "step": 2521 }, { "epoch": 0.49, "learning_rate": 1.0961369631566468e-05, "loss": 0.5221, "step": 2522 }, { "epoch": 0.49, "learning_rate": 1.0955166229331048e-05, "loss": 0.5925, "step": 2523 }, { "epoch": 0.49, "learning_rate": 1.0948962456120068e-05, "loss": 0.5577, "step": 2524 }, { "epoch": 0.49, "learning_rate": 1.0942758314343007e-05, "loss": 0.6241, "step": 2525 }, { "epoch": 0.49, "learning_rate": 1.0936553806409482e-05, "loss": 0.499, "step": 2526 }, { "epoch": 0.49, "learning_rate": 1.0930348934729249e-05, "loss": 0.5947, "step": 2527 }, { "epoch": 0.49, "learning_rate": 1.0924143701712211e-05, "loss": 0.5893, "step": 2528 }, { "epoch": 0.49, "learning_rate": 1.0917938109768404e-05, "loss": 0.6044, "step": 2529 }, { "epoch": 0.49, "learning_rate": 1.0911732161308014e-05, "loss": 0.579, "step": 2530 }, { "epoch": 0.49, "learning_rate": 1.0905525858741364e-05, "loss": 0.62, "step": 2531 }, { "epoch": 0.49, "learning_rate": 1.08993192044789e-05, "loss": 0.626, "step": 2532 }, { "epoch": 0.49, "learning_rate": 1.089311220093122e-05, "loss": 0.6474, "step": 2533 }, { "epoch": 0.49, "learning_rate": 1.0886904850509052e-05, "loss": 0.5569, "step": 2534 }, { "epoch": 0.49, "learning_rate": 1.0880697155623264e-05, "loss": 0.5643, "step": 2535 }, { "epoch": 0.49, "learning_rate": 1.0874489118684846e-05, "loss": 0.6077, "step": 2536 }, { "epoch": 0.49, "learning_rate": 1.086828074210493e-05, "loss": 0.5745, "step": 2537 }, { "epoch": 0.49, "learning_rate": 1.0862072028294777e-05, "loss": 0.5915, "step": 2538 }, { "epoch": 0.49, "learning_rate": 1.0855862979665788e-05, "loss": 0.6271, "step": 2539 }, { "epoch": 0.49, "learning_rate": 1.0849653598629477e-05, "loss": 0.6197, "step": 2540 }, { "epoch": 0.49, "learning_rate": 1.0843443887597495e-05, "loss": 0.5725, "step": 2541 }, { "epoch": 0.49, "learning_rate": 1.0837233848981632e-05, "loss": 0.5768, "step": 2542 }, { "epoch": 0.49, "learning_rate": 1.0831023485193787e-05, "loss": 0.5051, "step": 2543 }, { "epoch": 0.49, "learning_rate": 1.0824812798645997e-05, "loss": 0.6257, "step": 2544 }, { "epoch": 0.49, "learning_rate": 1.0818601791750418e-05, "loss": 0.609, "step": 2545 }, { "epoch": 0.49, "learning_rate": 1.0812390466919337e-05, "loss": 0.4737, "step": 2546 }, { "epoch": 0.49, "learning_rate": 1.0806178826565162e-05, "loss": 0.6103, "step": 2547 }, { "epoch": 0.49, "learning_rate": 1.0799966873100419e-05, "loss": 0.5483, "step": 2548 }, { "epoch": 0.49, "learning_rate": 1.0793754608937758e-05, "loss": 0.5182, "step": 2549 }, { "epoch": 0.49, "learning_rate": 1.0787542036489955e-05, "loss": 0.5691, "step": 2550 }, { "epoch": 0.49, "learning_rate": 1.0781329158169902e-05, "loss": 0.5926, "step": 2551 }, { "epoch": 0.49, "learning_rate": 1.0775115976390607e-05, "loss": 0.5904, "step": 2552 }, { "epoch": 0.49, "learning_rate": 1.0768902493565197e-05, "loss": 0.5859, "step": 2553 }, { "epoch": 0.49, "learning_rate": 1.0762688712106918e-05, "loss": 0.5584, "step": 2554 }, { "epoch": 0.49, "learning_rate": 1.0756474634429133e-05, "loss": 0.5022, "step": 2555 }, { "epoch": 0.49, "learning_rate": 1.0750260262945314e-05, "loss": 0.5738, "step": 2556 }, { "epoch": 0.49, "learning_rate": 1.0744045600069055e-05, "loss": 0.4659, "step": 2557 }, { "epoch": 0.49, "learning_rate": 1.0737830648214063e-05, "loss": 0.5655, "step": 2558 }, { "epoch": 0.49, "learning_rate": 1.0731615409794144e-05, "loss": 0.5996, "step": 2559 }, { "epoch": 0.49, "learning_rate": 1.0725399887223234e-05, "loss": 0.5809, "step": 2560 }, { "epoch": 0.49, "learning_rate": 1.0719184082915364e-05, "loss": 0.5854, "step": 2561 }, { "epoch": 0.49, "learning_rate": 1.0712967999284682e-05, "loss": 0.6577, "step": 2562 }, { "epoch": 0.49, "learning_rate": 1.0706751638745448e-05, "loss": 0.613, "step": 2563 }, { "epoch": 0.49, "learning_rate": 1.0700535003712023e-05, "loss": 0.466, "step": 2564 }, { "epoch": 0.49, "learning_rate": 1.069431809659887e-05, "loss": 0.6041, "step": 2565 }, { "epoch": 0.49, "learning_rate": 1.068810091982057e-05, "loss": 0.5922, "step": 2566 }, { "epoch": 0.49, "learning_rate": 1.0681883475791803e-05, "loss": 0.5014, "step": 2567 }, { "epoch": 0.49, "learning_rate": 1.067566576692735e-05, "loss": 0.6164, "step": 2568 }, { "epoch": 0.49, "learning_rate": 1.0669447795642103e-05, "loss": 0.6356, "step": 2569 }, { "epoch": 0.49, "learning_rate": 1.066322956435104e-05, "loss": 0.5895, "step": 2570 }, { "epoch": 0.49, "learning_rate": 1.065701107546926e-05, "loss": 0.6064, "step": 2571 }, { "epoch": 0.49, "learning_rate": 1.065079233141195e-05, "loss": 0.5981, "step": 2572 }, { "epoch": 0.5, "learning_rate": 1.0644573334594395e-05, "loss": 0.6034, "step": 2573 }, { "epoch": 0.5, "learning_rate": 1.0638354087431986e-05, "loss": 0.4745, "step": 2574 }, { "epoch": 0.5, "learning_rate": 1.0632134592340204e-05, "loss": 0.5358, "step": 2575 }, { "epoch": 0.5, "learning_rate": 1.0625914851734632e-05, "loss": 0.6069, "step": 2576 }, { "epoch": 0.5, "learning_rate": 1.0619694868030943e-05, "loss": 0.5841, "step": 2577 }, { "epoch": 0.5, "learning_rate": 1.0613474643644907e-05, "loss": 0.5802, "step": 2578 }, { "epoch": 0.5, "learning_rate": 1.0607254180992391e-05, "loss": 0.5881, "step": 2579 }, { "epoch": 0.5, "learning_rate": 1.0601033482489346e-05, "loss": 0.6201, "step": 2580 }, { "epoch": 0.5, "learning_rate": 1.0594812550551826e-05, "loss": 0.5087, "step": 2581 }, { "epoch": 0.5, "learning_rate": 1.058859138759596e-05, "loss": 0.5913, "step": 2582 }, { "epoch": 0.5, "learning_rate": 1.0582369996037985e-05, "loss": 0.5632, "step": 2583 }, { "epoch": 0.5, "learning_rate": 1.0576148378294213e-05, "loss": 0.5602, "step": 2584 }, { "epoch": 0.5, "learning_rate": 1.056992653678105e-05, "loss": 0.6184, "step": 2585 }, { "epoch": 0.5, "learning_rate": 1.0563704473914986e-05, "loss": 0.5823, "step": 2586 }, { "epoch": 0.5, "learning_rate": 1.0557482192112603e-05, "loss": 0.5765, "step": 2587 }, { "epoch": 0.5, "learning_rate": 1.0551259693790556e-05, "loss": 0.6062, "step": 2588 }, { "epoch": 0.5, "learning_rate": 1.0545036981365601e-05, "loss": 0.5864, "step": 2589 }, { "epoch": 0.5, "learning_rate": 1.053881405725456e-05, "loss": 0.6052, "step": 2590 }, { "epoch": 0.5, "learning_rate": 1.0532590923874349e-05, "loss": 0.5639, "step": 2591 }, { "epoch": 0.5, "learning_rate": 1.0526367583641958e-05, "loss": 0.5774, "step": 2592 }, { "epoch": 0.5, "learning_rate": 1.0520144038974468e-05, "loss": 0.4503, "step": 2593 }, { "epoch": 0.5, "learning_rate": 1.0513920292289021e-05, "loss": 0.5702, "step": 2594 }, { "epoch": 0.5, "learning_rate": 1.0507696346002857e-05, "loss": 0.6266, "step": 2595 }, { "epoch": 0.5, "learning_rate": 1.0501472202533285e-05, "loss": 0.5339, "step": 2596 }, { "epoch": 0.5, "learning_rate": 1.0495247864297684e-05, "loss": 0.5805, "step": 2597 }, { "epoch": 0.5, "learning_rate": 1.0489023333713522e-05, "loss": 0.551, "step": 2598 }, { "epoch": 0.5, "learning_rate": 1.0482798613198328e-05, "loss": 0.5855, "step": 2599 }, { "epoch": 0.5, "learning_rate": 1.047657370516972e-05, "loss": 0.6049, "step": 2600 }, { "epoch": 0.5, "learning_rate": 1.0470348612045376e-05, "loss": 0.5766, "step": 2601 }, { "epoch": 0.5, "learning_rate": 1.0464123336243049e-05, "loss": 0.4906, "step": 2602 }, { "epoch": 0.5, "learning_rate": 1.0457897880180566e-05, "loss": 0.6338, "step": 2603 }, { "epoch": 0.5, "learning_rate": 1.0451672246275826e-05, "loss": 0.5784, "step": 2604 }, { "epoch": 0.5, "learning_rate": 1.0445446436946788e-05, "loss": 0.5707, "step": 2605 }, { "epoch": 0.5, "learning_rate": 1.0439220454611486e-05, "loss": 0.588, "step": 2606 }, { "epoch": 0.5, "learning_rate": 1.0432994301688021e-05, "loss": 0.6141, "step": 2607 }, { "epoch": 0.5, "learning_rate": 1.0426767980594559e-05, "loss": 0.5141, "step": 2608 }, { "epoch": 0.5, "learning_rate": 1.0420541493749332e-05, "loss": 0.6187, "step": 2609 }, { "epoch": 0.5, "learning_rate": 1.0414314843570634e-05, "loss": 0.5544, "step": 2610 }, { "epoch": 0.5, "learning_rate": 1.0408088032476822e-05, "loss": 0.5116, "step": 2611 }, { "epoch": 0.5, "learning_rate": 1.0401861062886324e-05, "loss": 0.5753, "step": 2612 }, { "epoch": 0.5, "learning_rate": 1.0395633937217622e-05, "loss": 0.5607, "step": 2613 }, { "epoch": 0.5, "learning_rate": 1.0389406657889254e-05, "loss": 0.6185, "step": 2614 }, { "epoch": 0.5, "learning_rate": 1.0383179227319826e-05, "loss": 0.6366, "step": 2615 }, { "epoch": 0.5, "learning_rate": 1.0376951647928007e-05, "loss": 0.3768, "step": 2616 }, { "epoch": 0.5, "learning_rate": 1.0370723922132506e-05, "loss": 0.5876, "step": 2617 }, { "epoch": 0.5, "learning_rate": 1.036449605235211e-05, "loss": 0.4971, "step": 2618 }, { "epoch": 0.5, "learning_rate": 1.0358268041005644e-05, "loss": 0.6079, "step": 2619 }, { "epoch": 0.5, "learning_rate": 1.0352039890511997e-05, "loss": 0.5859, "step": 2620 }, { "epoch": 0.5, "learning_rate": 1.034581160329012e-05, "loss": 0.496, "step": 2621 }, { "epoch": 0.5, "learning_rate": 1.0339583181758997e-05, "loss": 0.6088, "step": 2622 }, { "epoch": 0.5, "learning_rate": 1.033335462833768e-05, "loss": 0.6003, "step": 2623 }, { "epoch": 0.5, "learning_rate": 1.0327125945445265e-05, "loss": 0.611, "step": 2624 }, { "epoch": 0.51, "learning_rate": 1.0320897135500904e-05, "loss": 0.6159, "step": 2625 }, { "epoch": 0.51, "learning_rate": 1.0314668200923791e-05, "loss": 0.6002, "step": 2626 }, { "epoch": 0.51, "learning_rate": 1.0308439144133177e-05, "loss": 0.5994, "step": 2627 }, { "epoch": 0.51, "learning_rate": 1.0302209967548354e-05, "loss": 0.6228, "step": 2628 }, { "epoch": 0.51, "learning_rate": 1.029598067358866e-05, "loss": 0.6203, "step": 2629 }, { "epoch": 0.51, "learning_rate": 1.0289751264673485e-05, "loss": 0.594, "step": 2630 }, { "epoch": 0.51, "learning_rate": 1.0283521743222256e-05, "loss": 0.4744, "step": 2631 }, { "epoch": 0.51, "learning_rate": 1.0277292111654447e-05, "loss": 0.5097, "step": 2632 }, { "epoch": 0.51, "learning_rate": 1.0271062372389582e-05, "loss": 0.6275, "step": 2633 }, { "epoch": 0.51, "learning_rate": 1.0264832527847212e-05, "loss": 0.488, "step": 2634 }, { "epoch": 0.51, "learning_rate": 1.0258602580446941e-05, "loss": 0.626, "step": 2635 }, { "epoch": 0.51, "learning_rate": 1.0252372532608405e-05, "loss": 0.5625, "step": 2636 }, { "epoch": 0.51, "learning_rate": 1.024614238675129e-05, "loss": 0.4786, "step": 2637 }, { "epoch": 0.51, "learning_rate": 1.0239912145295303e-05, "loss": 0.6158, "step": 2638 }, { "epoch": 0.51, "learning_rate": 1.0233681810660207e-05, "loss": 0.586, "step": 2639 }, { "epoch": 0.51, "learning_rate": 1.0227451385265788e-05, "loss": 0.576, "step": 2640 }, { "epoch": 0.51, "learning_rate": 1.022122087153187e-05, "loss": 0.6029, "step": 2641 }, { "epoch": 0.51, "learning_rate": 1.0214990271878319e-05, "loss": 0.641, "step": 2642 }, { "epoch": 0.51, "learning_rate": 1.0208759588725016e-05, "loss": 0.5927, "step": 2643 }, { "epoch": 0.51, "learning_rate": 1.0202528824491899e-05, "loss": 0.5352, "step": 2644 }, { "epoch": 0.51, "learning_rate": 1.0196297981598921e-05, "loss": 0.6066, "step": 2645 }, { "epoch": 0.51, "learning_rate": 1.019006706246607e-05, "loss": 0.6134, "step": 2646 }, { "epoch": 0.51, "learning_rate": 1.018383606951336e-05, "loss": 0.5946, "step": 2647 }, { "epoch": 0.51, "learning_rate": 1.0177605005160837e-05, "loss": 0.4347, "step": 2648 }, { "epoch": 0.51, "learning_rate": 1.0171373871828578e-05, "loss": 0.4897, "step": 2649 }, { "epoch": 0.51, "learning_rate": 1.0165142671936685e-05, "loss": 0.5946, "step": 2650 }, { "epoch": 0.51, "learning_rate": 1.0158911407905279e-05, "loss": 0.6277, "step": 2651 }, { "epoch": 0.51, "learning_rate": 1.0152680082154514e-05, "loss": 0.5932, "step": 2652 }, { "epoch": 0.51, "learning_rate": 1.0146448697104561e-05, "loss": 0.6219, "step": 2653 }, { "epoch": 0.51, "learning_rate": 1.0140217255175626e-05, "loss": 0.5585, "step": 2654 }, { "epoch": 0.51, "learning_rate": 1.013398575878792e-05, "loss": 0.5963, "step": 2655 }, { "epoch": 0.51, "learning_rate": 1.0127754210361694e-05, "loss": 0.5725, "step": 2656 }, { "epoch": 0.51, "learning_rate": 1.0121522612317204e-05, "loss": 0.5666, "step": 2657 }, { "epoch": 0.51, "learning_rate": 1.011529096707473e-05, "loss": 0.4946, "step": 2658 }, { "epoch": 0.51, "learning_rate": 1.0109059277054574e-05, "loss": 0.4841, "step": 2659 }, { "epoch": 0.51, "learning_rate": 1.010282754467705e-05, "loss": 0.5589, "step": 2660 }, { "epoch": 0.51, "learning_rate": 1.0096595772362492e-05, "loss": 0.5104, "step": 2661 }, { "epoch": 0.51, "learning_rate": 1.0090363962531251e-05, "loss": 0.5545, "step": 2662 }, { "epoch": 0.51, "learning_rate": 1.0084132117603689e-05, "loss": 0.5228, "step": 2663 }, { "epoch": 0.51, "learning_rate": 1.0077900240000181e-05, "loss": 0.6012, "step": 2664 }, { "epoch": 0.51, "learning_rate": 1.0071668332141115e-05, "loss": 0.6048, "step": 2665 }, { "epoch": 0.51, "learning_rate": 1.0065436396446899e-05, "loss": 0.6146, "step": 2666 }, { "epoch": 0.51, "learning_rate": 1.0059204435337938e-05, "loss": 0.6003, "step": 2667 }, { "epoch": 0.51, "learning_rate": 1.0052972451234656e-05, "loss": 0.5859, "step": 2668 }, { "epoch": 0.51, "learning_rate": 1.0046740446557485e-05, "loss": 0.4983, "step": 2669 }, { "epoch": 0.51, "learning_rate": 1.0040508423726865e-05, "loss": 0.6109, "step": 2670 }, { "epoch": 0.51, "learning_rate": 1.0034276385163238e-05, "loss": 0.5714, "step": 2671 }, { "epoch": 0.51, "learning_rate": 1.0028044333287056e-05, "loss": 0.575, "step": 2672 }, { "epoch": 0.51, "learning_rate": 1.002181227051878e-05, "loss": 0.5898, "step": 2673 }, { "epoch": 0.51, "learning_rate": 1.0015580199278873e-05, "loss": 0.5573, "step": 2674 }, { "epoch": 0.51, "learning_rate": 1.0009348121987795e-05, "loss": 0.6567, "step": 2675 }, { "epoch": 0.51, "learning_rate": 1.000311604106601e-05, "loss": 0.5041, "step": 2676 }, { "epoch": 0.52, "learning_rate": 9.996883958933993e-06, "loss": 0.4804, "step": 2677 }, { "epoch": 0.52, "learning_rate": 9.99065187801221e-06, "loss": 0.6062, "step": 2678 }, { "epoch": 0.52, "learning_rate": 9.984419800721132e-06, "loss": 0.6158, "step": 2679 }, { "epoch": 0.52, "learning_rate": 9.978187729481218e-06, "loss": 0.5041, "step": 2680 }, { "epoch": 0.52, "learning_rate": 9.971955666712945e-06, "loss": 0.5928, "step": 2681 }, { "epoch": 0.52, "learning_rate": 9.965723614836764e-06, "loss": 0.5204, "step": 2682 }, { "epoch": 0.52, "learning_rate": 9.959491576273139e-06, "loss": 0.5638, "step": 2683 }, { "epoch": 0.52, "learning_rate": 9.95325955344252e-06, "loss": 0.5379, "step": 2684 }, { "epoch": 0.52, "learning_rate": 9.947027548765347e-06, "loss": 0.5854, "step": 2685 }, { "epoch": 0.52, "learning_rate": 9.940795564662064e-06, "loss": 0.5956, "step": 2686 }, { "epoch": 0.52, "learning_rate": 9.934563603553103e-06, "loss": 0.4725, "step": 2687 }, { "epoch": 0.52, "learning_rate": 9.928331667858886e-06, "loss": 0.6085, "step": 2688 }, { "epoch": 0.52, "learning_rate": 9.922099759999822e-06, "loss": 0.5487, "step": 2689 }, { "epoch": 0.52, "learning_rate": 9.915867882396314e-06, "loss": 0.4728, "step": 2690 }, { "epoch": 0.52, "learning_rate": 9.909636037468754e-06, "loss": 0.5817, "step": 2691 }, { "epoch": 0.52, "learning_rate": 9.90340422763751e-06, "loss": 0.595, "step": 2692 }, { "epoch": 0.52, "learning_rate": 9.897172455322953e-06, "loss": 0.5769, "step": 2693 }, { "epoch": 0.52, "learning_rate": 9.890940722945429e-06, "loss": 0.6257, "step": 2694 }, { "epoch": 0.52, "learning_rate": 9.884709032925274e-06, "loss": 0.6033, "step": 2695 }, { "epoch": 0.52, "learning_rate": 9.878477387682801e-06, "loss": 0.5915, "step": 2696 }, { "epoch": 0.52, "learning_rate": 9.872245789638308e-06, "loss": 0.5831, "step": 2697 }, { "epoch": 0.52, "learning_rate": 9.866014241212078e-06, "loss": 0.6171, "step": 2698 }, { "epoch": 0.52, "learning_rate": 9.859782744824376e-06, "loss": 0.4865, "step": 2699 }, { "epoch": 0.52, "learning_rate": 9.85355130289544e-06, "loss": 0.5733, "step": 2700 }, { "epoch": 0.52, "learning_rate": 9.84731991784549e-06, "loss": 0.6176, "step": 2701 }, { "epoch": 0.52, "learning_rate": 9.841088592094726e-06, "loss": 0.574, "step": 2702 }, { "epoch": 0.52, "learning_rate": 9.834857328063316e-06, "loss": 0.6226, "step": 2703 }, { "epoch": 0.52, "learning_rate": 9.828626128171422e-06, "loss": 0.5729, "step": 2704 }, { "epoch": 0.52, "learning_rate": 9.822394994839164e-06, "loss": 0.5482, "step": 2705 }, { "epoch": 0.52, "learning_rate": 9.816163930486643e-06, "loss": 0.5722, "step": 2706 }, { "epoch": 0.52, "learning_rate": 9.809932937533935e-06, "loss": 0.577, "step": 2707 }, { "epoch": 0.52, "learning_rate": 9.803702018401084e-06, "loss": 0.5841, "step": 2708 }, { "epoch": 0.52, "learning_rate": 9.797471175508101e-06, "loss": 0.4638, "step": 2709 }, { "epoch": 0.52, "learning_rate": 9.791240411274982e-06, "loss": 0.5714, "step": 2710 }, { "epoch": 0.52, "learning_rate": 9.785009728121686e-06, "loss": 0.3662, "step": 2711 }, { "epoch": 0.52, "learning_rate": 9.778779128468133e-06, "loss": 0.5679, "step": 2712 }, { "epoch": 0.52, "learning_rate": 9.772548614734217e-06, "loss": 0.5868, "step": 2713 }, { "epoch": 0.52, "learning_rate": 9.766318189339798e-06, "loss": 0.5961, "step": 2714 }, { "epoch": 0.52, "learning_rate": 9.760087854704697e-06, "loss": 0.5785, "step": 2715 }, { "epoch": 0.52, "learning_rate": 9.753857613248714e-06, "loss": 0.5196, "step": 2716 }, { "epoch": 0.52, "learning_rate": 9.747627467391596e-06, "loss": 0.3787, "step": 2717 }, { "epoch": 0.52, "learning_rate": 9.741397419553062e-06, "loss": 0.4706, "step": 2718 }, { "epoch": 0.52, "learning_rate": 9.735167472152793e-06, "loss": 0.5665, "step": 2719 }, { "epoch": 0.52, "learning_rate": 9.728937627610425e-06, "loss": 0.5833, "step": 2720 }, { "epoch": 0.52, "learning_rate": 9.722707888345553e-06, "loss": 0.6065, "step": 2721 }, { "epoch": 0.52, "learning_rate": 9.716478256777749e-06, "loss": 0.5729, "step": 2722 }, { "epoch": 0.52, "learning_rate": 9.710248735326519e-06, "loss": 0.6251, "step": 2723 }, { "epoch": 0.52, "learning_rate": 9.704019326411344e-06, "loss": 0.6075, "step": 2724 }, { "epoch": 0.52, "learning_rate": 9.697790032451651e-06, "loss": 0.5281, "step": 2725 }, { "epoch": 0.52, "learning_rate": 9.691560855866826e-06, "loss": 0.5943, "step": 2726 }, { "epoch": 0.52, "learning_rate": 9.685331799076208e-06, "loss": 0.6204, "step": 2727 }, { "epoch": 0.52, "learning_rate": 9.6791028644991e-06, "loss": 0.6165, "step": 2728 }, { "epoch": 0.53, "learning_rate": 9.672874054554738e-06, "loss": 0.6613, "step": 2729 }, { "epoch": 0.53, "learning_rate": 9.666645371662324e-06, "loss": 0.5024, "step": 2730 }, { "epoch": 0.53, "learning_rate": 9.660416818241007e-06, "loss": 0.6059, "step": 2731 }, { "epoch": 0.53, "learning_rate": 9.654188396709882e-06, "loss": 0.5967, "step": 2732 }, { "epoch": 0.53, "learning_rate": 9.647960109488003e-06, "loss": 0.5988, "step": 2733 }, { "epoch": 0.53, "learning_rate": 9.64173195899436e-06, "loss": 0.5691, "step": 2734 }, { "epoch": 0.53, "learning_rate": 9.635503947647894e-06, "loss": 0.4953, "step": 2735 }, { "epoch": 0.53, "learning_rate": 9.629276077867497e-06, "loss": 0.4903, "step": 2736 }, { "epoch": 0.53, "learning_rate": 9.623048352071998e-06, "loss": 0.5938, "step": 2737 }, { "epoch": 0.53, "learning_rate": 9.616820772680174e-06, "loss": 0.6173, "step": 2738 }, { "epoch": 0.53, "learning_rate": 9.610593342110746e-06, "loss": 0.615, "step": 2739 }, { "epoch": 0.53, "learning_rate": 9.604366062782381e-06, "loss": 0.4643, "step": 2740 }, { "epoch": 0.53, "learning_rate": 9.598138937113677e-06, "loss": 0.5922, "step": 2741 }, { "epoch": 0.53, "learning_rate": 9.59191196752318e-06, "loss": 0.546, "step": 2742 }, { "epoch": 0.53, "learning_rate": 9.58568515642937e-06, "loss": 0.479, "step": 2743 }, { "epoch": 0.53, "learning_rate": 9.579458506250668e-06, "loss": 0.6091, "step": 2744 }, { "epoch": 0.53, "learning_rate": 9.573232019405441e-06, "loss": 0.5513, "step": 2745 }, { "epoch": 0.53, "learning_rate": 9.567005698311982e-06, "loss": 0.5901, "step": 2746 }, { "epoch": 0.53, "learning_rate": 9.560779545388517e-06, "loss": 0.6015, "step": 2747 }, { "epoch": 0.53, "learning_rate": 9.554553563053217e-06, "loss": 0.5641, "step": 2748 }, { "epoch": 0.53, "learning_rate": 9.548327753724181e-06, "loss": 0.4991, "step": 2749 }, { "epoch": 0.53, "learning_rate": 9.542102119819436e-06, "loss": 0.5814, "step": 2750 }, { "epoch": 0.53, "learning_rate": 9.535876663756955e-06, "loss": 0.4769, "step": 2751 }, { "epoch": 0.53, "learning_rate": 9.529651387954628e-06, "loss": 0.5453, "step": 2752 }, { "epoch": 0.53, "learning_rate": 9.523426294830284e-06, "loss": 0.4828, "step": 2753 }, { "epoch": 0.53, "learning_rate": 9.517201386801675e-06, "loss": 0.5692, "step": 2754 }, { "epoch": 0.53, "learning_rate": 9.510976666286484e-06, "loss": 0.5844, "step": 2755 }, { "epoch": 0.53, "learning_rate": 9.504752135702318e-06, "loss": 0.5478, "step": 2756 }, { "epoch": 0.53, "learning_rate": 9.498527797466718e-06, "loss": 0.5411, "step": 2757 }, { "epoch": 0.53, "learning_rate": 9.492303653997146e-06, "loss": 0.5871, "step": 2758 }, { "epoch": 0.53, "learning_rate": 9.48607970771098e-06, "loss": 0.5939, "step": 2759 }, { "epoch": 0.53, "learning_rate": 9.479855961025538e-06, "loss": 0.4979, "step": 2760 }, { "epoch": 0.53, "learning_rate": 9.473632416358045e-06, "loss": 0.5854, "step": 2761 }, { "epoch": 0.53, "learning_rate": 9.467409076125653e-06, "loss": 0.3455, "step": 2762 }, { "epoch": 0.53, "learning_rate": 9.461185942745443e-06, "loss": 0.6125, "step": 2763 }, { "epoch": 0.53, "learning_rate": 9.454963018634402e-06, "loss": 0.5619, "step": 2764 }, { "epoch": 0.53, "learning_rate": 9.448740306209447e-06, "loss": 0.6087, "step": 2765 }, { "epoch": 0.53, "learning_rate": 9.442517807887402e-06, "loss": 0.584, "step": 2766 }, { "epoch": 0.53, "learning_rate": 9.436295526085016e-06, "loss": 0.5583, "step": 2767 }, { "epoch": 0.53, "learning_rate": 9.430073463218952e-06, "loss": 0.5868, "step": 2768 }, { "epoch": 0.53, "learning_rate": 9.423851621705789e-06, "loss": 0.4507, "step": 2769 }, { "epoch": 0.53, "learning_rate": 9.41763000396202e-06, "loss": 0.5673, "step": 2770 }, { "epoch": 0.53, "learning_rate": 9.411408612404043e-06, "loss": 0.5625, "step": 2771 }, { "epoch": 0.53, "learning_rate": 9.40518744944818e-06, "loss": 0.5965, "step": 2772 }, { "epoch": 0.53, "learning_rate": 9.398966517510654e-06, "loss": 0.5897, "step": 2773 }, { "epoch": 0.53, "learning_rate": 9.39274581900761e-06, "loss": 0.6344, "step": 2774 }, { "epoch": 0.53, "learning_rate": 9.386525356355095e-06, "loss": 0.5958, "step": 2775 }, { "epoch": 0.53, "learning_rate": 9.380305131969059e-06, "loss": 0.6356, "step": 2776 }, { "epoch": 0.53, "learning_rate": 9.374085148265372e-06, "loss": 0.5888, "step": 2777 }, { "epoch": 0.53, "learning_rate": 9.3678654076598e-06, "loss": 0.5811, "step": 2778 }, { "epoch": 0.53, "learning_rate": 9.361645912568015e-06, "loss": 0.5851, "step": 2779 }, { "epoch": 0.53, "learning_rate": 9.355426665405607e-06, "loss": 0.5803, "step": 2780 }, { "epoch": 0.54, "learning_rate": 9.349207668588053e-06, "loss": 0.4944, "step": 2781 }, { "epoch": 0.54, "learning_rate": 9.342988924530742e-06, "loss": 0.6142, "step": 2782 }, { "epoch": 0.54, "learning_rate": 9.336770435648963e-06, "loss": 0.5742, "step": 2783 }, { "epoch": 0.54, "learning_rate": 9.330552204357904e-06, "loss": 0.6018, "step": 2784 }, { "epoch": 0.54, "learning_rate": 9.32433423307265e-06, "loss": 0.5921, "step": 2785 }, { "epoch": 0.54, "learning_rate": 9.318116524208198e-06, "loss": 0.5886, "step": 2786 }, { "epoch": 0.54, "learning_rate": 9.311899080179433e-06, "loss": 0.6014, "step": 2787 }, { "epoch": 0.54, "learning_rate": 9.305681903401133e-06, "loss": 0.6024, "step": 2788 }, { "epoch": 0.54, "learning_rate": 9.299464996287984e-06, "loss": 0.5662, "step": 2789 }, { "epoch": 0.54, "learning_rate": 9.293248361254557e-06, "loss": 0.5367, "step": 2790 }, { "epoch": 0.54, "learning_rate": 9.287032000715318e-06, "loss": 0.5677, "step": 2791 }, { "epoch": 0.54, "learning_rate": 9.28081591708464e-06, "loss": 0.5819, "step": 2792 }, { "epoch": 0.54, "learning_rate": 9.27460011277677e-06, "loss": 0.5823, "step": 2793 }, { "epoch": 0.54, "learning_rate": 9.268384590205858e-06, "loss": 0.5958, "step": 2794 }, { "epoch": 0.54, "learning_rate": 9.262169351785944e-06, "loss": 0.5474, "step": 2795 }, { "epoch": 0.54, "learning_rate": 9.255954399930948e-06, "loss": 0.6343, "step": 2796 }, { "epoch": 0.54, "learning_rate": 9.249739737054686e-06, "loss": 0.6277, "step": 2797 }, { "epoch": 0.54, "learning_rate": 9.24352536557087e-06, "loss": 0.4626, "step": 2798 }, { "epoch": 0.54, "learning_rate": 9.237311287893086e-06, "loss": 0.5909, "step": 2799 }, { "epoch": 0.54, "learning_rate": 9.231097506434808e-06, "loss": 0.551, "step": 2800 }, { "epoch": 0.54, "learning_rate": 9.224884023609398e-06, "loss": 0.5566, "step": 2801 }, { "epoch": 0.54, "learning_rate": 9.218670841830098e-06, "loss": 0.6022, "step": 2802 }, { "epoch": 0.54, "learning_rate": 9.212457963510045e-06, "loss": 0.6183, "step": 2803 }, { "epoch": 0.54, "learning_rate": 9.206245391062243e-06, "loss": 0.6052, "step": 2804 }, { "epoch": 0.54, "learning_rate": 9.200033126899585e-06, "loss": 0.5751, "step": 2805 }, { "epoch": 0.54, "learning_rate": 9.193821173434843e-06, "loss": 0.4918, "step": 2806 }, { "epoch": 0.54, "learning_rate": 9.187609533080668e-06, "loss": 0.5651, "step": 2807 }, { "epoch": 0.54, "learning_rate": 9.181398208249583e-06, "loss": 0.6111, "step": 2808 }, { "epoch": 0.54, "learning_rate": 9.175187201354005e-06, "loss": 0.5928, "step": 2809 }, { "epoch": 0.54, "learning_rate": 9.168976514806216e-06, "loss": 0.4847, "step": 2810 }, { "epoch": 0.54, "learning_rate": 9.162766151018372e-06, "loss": 0.6194, "step": 2811 }, { "epoch": 0.54, "learning_rate": 9.156556112402508e-06, "loss": 0.5831, "step": 2812 }, { "epoch": 0.54, "learning_rate": 9.150346401370528e-06, "loss": 0.6201, "step": 2813 }, { "epoch": 0.54, "learning_rate": 9.144137020334214e-06, "loss": 0.4893, "step": 2814 }, { "epoch": 0.54, "learning_rate": 9.137927971705223e-06, "loss": 0.4792, "step": 2815 }, { "epoch": 0.54, "learning_rate": 9.131719257895074e-06, "loss": 0.595, "step": 2816 }, { "epoch": 0.54, "learning_rate": 9.125510881315159e-06, "loss": 0.5606, "step": 2817 }, { "epoch": 0.54, "learning_rate": 9.119302844376741e-06, "loss": 0.6203, "step": 2818 }, { "epoch": 0.54, "learning_rate": 9.113095149490951e-06, "loss": 0.6167, "step": 2819 }, { "epoch": 0.54, "learning_rate": 9.106887799068782e-06, "loss": 0.5527, "step": 2820 }, { "epoch": 0.54, "learning_rate": 9.100680795521104e-06, "loss": 0.6514, "step": 2821 }, { "epoch": 0.54, "learning_rate": 9.09447414125864e-06, "loss": 0.592, "step": 2822 }, { "epoch": 0.54, "learning_rate": 9.088267838691987e-06, "loss": 0.5698, "step": 2823 }, { "epoch": 0.54, "learning_rate": 9.0820618902316e-06, "loss": 0.604, "step": 2824 }, { "epoch": 0.54, "learning_rate": 9.075856298287796e-06, "loss": 0.5927, "step": 2825 }, { "epoch": 0.54, "learning_rate": 9.069651065270753e-06, "loss": 0.6109, "step": 2826 }, { "epoch": 0.54, "learning_rate": 9.06344619359052e-06, "loss": 0.6358, "step": 2827 }, { "epoch": 0.54, "learning_rate": 9.057241685656995e-06, "loss": 0.4962, "step": 2828 }, { "epoch": 0.54, "learning_rate": 9.051037543879933e-06, "loss": 0.604, "step": 2829 }, { "epoch": 0.54, "learning_rate": 9.044833770668957e-06, "loss": 0.5818, "step": 2830 }, { "epoch": 0.54, "learning_rate": 9.038630368433537e-06, "loss": 0.5819, "step": 2831 }, { "epoch": 0.54, "learning_rate": 9.032427339583e-06, "loss": 0.5995, "step": 2832 }, { "epoch": 0.55, "learning_rate": 9.026224686526539e-06, "loss": 0.6004, "step": 2833 }, { "epoch": 0.55, "learning_rate": 9.020022411673186e-06, "loss": 0.5863, "step": 2834 }, { "epoch": 0.55, "learning_rate": 9.013820517431841e-06, "loss": 0.63, "step": 2835 }, { "epoch": 0.55, "learning_rate": 9.00761900621124e-06, "loss": 0.5524, "step": 2836 }, { "epoch": 0.55, "learning_rate": 9.00141788041998e-06, "loss": 0.5922, "step": 2837 }, { "epoch": 0.55, "learning_rate": 8.99521714246651e-06, "loss": 0.5137, "step": 2838 }, { "epoch": 0.55, "learning_rate": 8.989016794759127e-06, "loss": 0.5131, "step": 2839 }, { "epoch": 0.55, "learning_rate": 8.98281683970597e-06, "loss": 0.5574, "step": 2840 }, { "epoch": 0.55, "learning_rate": 8.97661727971503e-06, "loss": 0.5407, "step": 2841 }, { "epoch": 0.55, "learning_rate": 8.970418117194146e-06, "loss": 0.5836, "step": 2842 }, { "epoch": 0.55, "learning_rate": 8.964219354550999e-06, "loss": 0.5267, "step": 2843 }, { "epoch": 0.55, "learning_rate": 8.958020994193124e-06, "loss": 0.5974, "step": 2844 }, { "epoch": 0.55, "learning_rate": 8.951823038527887e-06, "loss": 0.5678, "step": 2845 }, { "epoch": 0.55, "learning_rate": 8.945625489962503e-06, "loss": 0.6051, "step": 2846 }, { "epoch": 0.55, "learning_rate": 8.93942835090403e-06, "loss": 0.6293, "step": 2847 }, { "epoch": 0.55, "learning_rate": 8.933231623759365e-06, "loss": 0.6262, "step": 2848 }, { "epoch": 0.55, "learning_rate": 8.927035310935241e-06, "loss": 0.6284, "step": 2849 }, { "epoch": 0.55, "learning_rate": 8.920839414838243e-06, "loss": 0.5062, "step": 2850 }, { "epoch": 0.55, "learning_rate": 8.914643937874778e-06, "loss": 0.5588, "step": 2851 }, { "epoch": 0.55, "learning_rate": 8.908448882451104e-06, "loss": 0.5601, "step": 2852 }, { "epoch": 0.55, "learning_rate": 8.902254250973306e-06, "loss": 0.5758, "step": 2853 }, { "epoch": 0.55, "learning_rate": 8.896060045847305e-06, "loss": 0.6003, "step": 2854 }, { "epoch": 0.55, "learning_rate": 8.88986626947886e-06, "loss": 0.5854, "step": 2855 }, { "epoch": 0.55, "learning_rate": 8.883672924273566e-06, "loss": 0.4862, "step": 2856 }, { "epoch": 0.55, "learning_rate": 8.877480012636847e-06, "loss": 0.6159, "step": 2857 }, { "epoch": 0.55, "learning_rate": 8.871287536973954e-06, "loss": 0.6328, "step": 2858 }, { "epoch": 0.55, "learning_rate": 8.865095499689978e-06, "loss": 0.5965, "step": 2859 }, { "epoch": 0.55, "learning_rate": 8.85890390318983e-06, "loss": 0.5692, "step": 2860 }, { "epoch": 0.55, "learning_rate": 8.852712749878255e-06, "loss": 0.5875, "step": 2861 }, { "epoch": 0.55, "learning_rate": 8.846522042159833e-06, "loss": 0.4616, "step": 2862 }, { "epoch": 0.55, "learning_rate": 8.840331782438954e-06, "loss": 0.5793, "step": 2863 }, { "epoch": 0.55, "learning_rate": 8.83414197311985e-06, "loss": 0.5967, "step": 2864 }, { "epoch": 0.55, "learning_rate": 8.82795261660657e-06, "loss": 0.5755, "step": 2865 }, { "epoch": 0.55, "learning_rate": 8.821763715302986e-06, "loss": 0.5838, "step": 2866 }, { "epoch": 0.55, "learning_rate": 8.815575271612798e-06, "loss": 0.6019, "step": 2867 }, { "epoch": 0.55, "learning_rate": 8.809387287939528e-06, "loss": 0.4697, "step": 2868 }, { "epoch": 0.55, "learning_rate": 8.803199766686517e-06, "loss": 0.5909, "step": 2869 }, { "epoch": 0.55, "learning_rate": 8.797012710256923e-06, "loss": 0.6273, "step": 2870 }, { "epoch": 0.55, "learning_rate": 8.790826121053732e-06, "loss": 0.3972, "step": 2871 }, { "epoch": 0.55, "learning_rate": 8.784640001479741e-06, "loss": 0.4816, "step": 2872 }, { "epoch": 0.55, "learning_rate": 8.77845435393757e-06, "loss": 0.5018, "step": 2873 }, { "epoch": 0.55, "learning_rate": 8.772269180829653e-06, "loss": 0.466, "step": 2874 }, { "epoch": 0.55, "learning_rate": 8.766084484558237e-06, "loss": 0.5583, "step": 2875 }, { "epoch": 0.55, "learning_rate": 8.759900267525393e-06, "loss": 0.5972, "step": 2876 }, { "epoch": 0.55, "learning_rate": 8.753716532132992e-06, "loss": 0.5656, "step": 2877 }, { "epoch": 0.55, "learning_rate": 8.747533280782725e-06, "loss": 0.6053, "step": 2878 }, { "epoch": 0.55, "learning_rate": 8.741350515876103e-06, "loss": 0.5637, "step": 2879 }, { "epoch": 0.55, "learning_rate": 8.73516823981444e-06, "loss": 0.5526, "step": 2880 }, { "epoch": 0.55, "learning_rate": 8.728986454998858e-06, "loss": 0.5559, "step": 2881 }, { "epoch": 0.55, "learning_rate": 8.72280516383029e-06, "loss": 0.4775, "step": 2882 }, { "epoch": 0.55, "learning_rate": 8.716624368709477e-06, "loss": 0.5998, "step": 2883 }, { "epoch": 0.55, "learning_rate": 8.71044407203697e-06, "loss": 0.5846, "step": 2884 }, { "epoch": 0.56, "learning_rate": 8.70426427621313e-06, "loss": 0.5646, "step": 2885 }, { "epoch": 0.56, "learning_rate": 8.698084983638111e-06, "loss": 0.6088, "step": 2886 }, { "epoch": 0.56, "learning_rate": 8.691906196711884e-06, "loss": 0.5892, "step": 2887 }, { "epoch": 0.56, "learning_rate": 8.685727917834218e-06, "loss": 0.6172, "step": 2888 }, { "epoch": 0.56, "learning_rate": 8.679550149404685e-06, "loss": 0.4541, "step": 2889 }, { "epoch": 0.56, "learning_rate": 8.673372893822653e-06, "loss": 0.5979, "step": 2890 }, { "epoch": 0.56, "learning_rate": 8.667196153487308e-06, "loss": 0.5725, "step": 2891 }, { "epoch": 0.56, "learning_rate": 8.661019930797615e-06, "loss": 0.4897, "step": 2892 }, { "epoch": 0.56, "learning_rate": 8.654844228152355e-06, "loss": 0.5349, "step": 2893 }, { "epoch": 0.56, "learning_rate": 8.648669047950097e-06, "loss": 0.6074, "step": 2894 }, { "epoch": 0.56, "learning_rate": 8.642494392589206e-06, "loss": 0.6079, "step": 2895 }, { "epoch": 0.56, "learning_rate": 8.63632026446785e-06, "loss": 0.5898, "step": 2896 }, { "epoch": 0.56, "learning_rate": 8.630146665983993e-06, "loss": 0.6388, "step": 2897 }, { "epoch": 0.56, "learning_rate": 8.623973599535385e-06, "loss": 0.5991, "step": 2898 }, { "epoch": 0.56, "learning_rate": 8.617801067519575e-06, "loss": 0.5164, "step": 2899 }, { "epoch": 0.56, "learning_rate": 8.611629072333905e-06, "loss": 0.6169, "step": 2900 }, { "epoch": 0.56, "learning_rate": 8.605457616375503e-06, "loss": 0.652, "step": 2901 }, { "epoch": 0.56, "learning_rate": 8.599286702041292e-06, "loss": 0.5787, "step": 2902 }, { "epoch": 0.56, "learning_rate": 8.593116331727987e-06, "loss": 0.6017, "step": 2903 }, { "epoch": 0.56, "learning_rate": 8.586946507832088e-06, "loss": 0.5706, "step": 2904 }, { "epoch": 0.56, "learning_rate": 8.580777232749883e-06, "loss": 0.6102, "step": 2905 }, { "epoch": 0.56, "learning_rate": 8.574608508877448e-06, "loss": 0.3726, "step": 2906 }, { "epoch": 0.56, "learning_rate": 8.568440338610638e-06, "loss": 0.6165, "step": 2907 }, { "epoch": 0.56, "learning_rate": 8.562272724345108e-06, "loss": 0.5592, "step": 2908 }, { "epoch": 0.56, "learning_rate": 8.556105668476287e-06, "loss": 0.6177, "step": 2909 }, { "epoch": 0.56, "learning_rate": 8.549939173399385e-06, "loss": 0.6205, "step": 2910 }, { "epoch": 0.56, "learning_rate": 8.5437732415094e-06, "loss": 0.5413, "step": 2911 }, { "epoch": 0.56, "learning_rate": 8.537607875201106e-06, "loss": 0.4754, "step": 2912 }, { "epoch": 0.56, "learning_rate": 8.531443076869058e-06, "loss": 0.5725, "step": 2913 }, { "epoch": 0.56, "learning_rate": 8.525278848907603e-06, "loss": 0.5793, "step": 2914 }, { "epoch": 0.56, "learning_rate": 8.51911519371085e-06, "loss": 0.6136, "step": 2915 }, { "epoch": 0.56, "learning_rate": 8.512952113672689e-06, "loss": 0.5652, "step": 2916 }, { "epoch": 0.56, "learning_rate": 8.506789611186794e-06, "loss": 0.5411, "step": 2917 }, { "epoch": 0.56, "learning_rate": 8.500627688646607e-06, "loss": 0.5716, "step": 2918 }, { "epoch": 0.56, "learning_rate": 8.494466348445345e-06, "loss": 0.5889, "step": 2919 }, { "epoch": 0.56, "learning_rate": 8.48830559297601e-06, "loss": 0.6046, "step": 2920 }, { "epoch": 0.56, "learning_rate": 8.48214542463136e-06, "loss": 0.4392, "step": 2921 }, { "epoch": 0.56, "learning_rate": 8.475985845803938e-06, "loss": 0.5521, "step": 2922 }, { "epoch": 0.56, "learning_rate": 8.469826858886054e-06, "loss": 0.4905, "step": 2923 }, { "epoch": 0.56, "learning_rate": 8.463668466269785e-06, "loss": 0.5732, "step": 2924 }, { "epoch": 0.56, "learning_rate": 8.457510670346976e-06, "loss": 0.5939, "step": 2925 }, { "epoch": 0.56, "learning_rate": 8.451353473509254e-06, "loss": 0.6274, "step": 2926 }, { "epoch": 0.56, "learning_rate": 8.445196878147997e-06, "loss": 0.5612, "step": 2927 }, { "epoch": 0.56, "learning_rate": 8.439040886654354e-06, "loss": 0.589, "step": 2928 }, { "epoch": 0.56, "learning_rate": 8.432885501419248e-06, "loss": 0.5026, "step": 2929 }, { "epoch": 0.56, "learning_rate": 8.426730724833354e-06, "loss": 0.5771, "step": 2930 }, { "epoch": 0.56, "learning_rate": 8.420576559287112e-06, "loss": 0.6007, "step": 2931 }, { "epoch": 0.56, "learning_rate": 8.414423007170742e-06, "loss": 0.5789, "step": 2932 }, { "epoch": 0.56, "learning_rate": 8.408270070874201e-06, "loss": 0.5745, "step": 2933 }, { "epoch": 0.56, "learning_rate": 8.402117752787225e-06, "loss": 0.597, "step": 2934 }, { "epoch": 0.56, "learning_rate": 8.395966055299302e-06, "loss": 0.5382, "step": 2935 }, { "epoch": 0.56, "learning_rate": 8.389814980799679e-06, "loss": 0.6255, "step": 2936 }, { "epoch": 0.57, "learning_rate": 8.38366453167736e-06, "loss": 0.5665, "step": 2937 }, { "epoch": 0.57, "learning_rate": 8.377514710321117e-06, "loss": 0.5771, "step": 2938 }, { "epoch": 0.57, "learning_rate": 8.371365519119463e-06, "loss": 0.4295, "step": 2939 }, { "epoch": 0.57, "learning_rate": 8.365216960460675e-06, "loss": 0.5804, "step": 2940 }, { "epoch": 0.57, "learning_rate": 8.359069036732781e-06, "loss": 0.5153, "step": 2941 }, { "epoch": 0.57, "learning_rate": 8.352921750323562e-06, "loss": 0.507, "step": 2942 }, { "epoch": 0.57, "learning_rate": 8.346775103620559e-06, "loss": 0.5868, "step": 2943 }, { "epoch": 0.57, "learning_rate": 8.340629099011057e-06, "loss": 0.514, "step": 2944 }, { "epoch": 0.57, "learning_rate": 8.33448373888209e-06, "loss": 0.5868, "step": 2945 }, { "epoch": 0.57, "learning_rate": 8.328339025620449e-06, "loss": 0.4614, "step": 2946 }, { "epoch": 0.57, "learning_rate": 8.322194961612668e-06, "loss": 0.5682, "step": 2947 }, { "epoch": 0.57, "learning_rate": 8.316051549245026e-06, "loss": 0.6091, "step": 2948 }, { "epoch": 0.57, "learning_rate": 8.309908790903562e-06, "loss": 0.5706, "step": 2949 }, { "epoch": 0.57, "learning_rate": 8.303766688974047e-06, "loss": 0.5721, "step": 2950 }, { "epoch": 0.57, "learning_rate": 8.297625245842006e-06, "loss": 0.6095, "step": 2951 }, { "epoch": 0.57, "learning_rate": 8.291484463892703e-06, "loss": 0.4741, "step": 2952 }, { "epoch": 0.57, "learning_rate": 8.285344345511147e-06, "loss": 0.5684, "step": 2953 }, { "epoch": 0.57, "learning_rate": 8.279204893082083e-06, "loss": 0.5613, "step": 2954 }, { "epoch": 0.57, "learning_rate": 8.273066108990017e-06, "loss": 0.5681, "step": 2955 }, { "epoch": 0.57, "learning_rate": 8.266927995619175e-06, "loss": 0.6146, "step": 2956 }, { "epoch": 0.57, "learning_rate": 8.260790555353526e-06, "loss": 0.4875, "step": 2957 }, { "epoch": 0.57, "learning_rate": 8.254653790576787e-06, "loss": 0.5498, "step": 2958 }, { "epoch": 0.57, "learning_rate": 8.248517703672405e-06, "loss": 0.5436, "step": 2959 }, { "epoch": 0.57, "learning_rate": 8.242382297023558e-06, "loss": 0.5972, "step": 2960 }, { "epoch": 0.57, "learning_rate": 8.23624757301318e-06, "loss": 0.5523, "step": 2961 }, { "epoch": 0.57, "learning_rate": 8.230113534023917e-06, "loss": 0.4802, "step": 2962 }, { "epoch": 0.57, "learning_rate": 8.223980182438167e-06, "loss": 0.5538, "step": 2963 }, { "epoch": 0.57, "learning_rate": 8.217847520638049e-06, "loss": 0.5494, "step": 2964 }, { "epoch": 0.57, "learning_rate": 8.211715551005414e-06, "loss": 0.5287, "step": 2965 }, { "epoch": 0.57, "learning_rate": 8.205584275921854e-06, "loss": 0.6475, "step": 2966 }, { "epoch": 0.57, "learning_rate": 8.199453697768686e-06, "loss": 0.596, "step": 2967 }, { "epoch": 0.57, "learning_rate": 8.193323818926955e-06, "loss": 0.5787, "step": 2968 }, { "epoch": 0.57, "learning_rate": 8.187194641777432e-06, "loss": 0.6467, "step": 2969 }, { "epoch": 0.57, "learning_rate": 8.181066168700622e-06, "loss": 0.5872, "step": 2970 }, { "epoch": 0.57, "learning_rate": 8.174938402076754e-06, "loss": 0.5588, "step": 2971 }, { "epoch": 0.57, "learning_rate": 8.168811344285776e-06, "loss": 0.5799, "step": 2972 }, { "epoch": 0.57, "learning_rate": 8.162684997707374e-06, "loss": 0.5638, "step": 2973 }, { "epoch": 0.57, "learning_rate": 8.156559364720947e-06, "loss": 0.5253, "step": 2974 }, { "epoch": 0.57, "learning_rate": 8.150434447705623e-06, "loss": 0.5835, "step": 2975 }, { "epoch": 0.57, "learning_rate": 8.144310249040246e-06, "loss": 0.6154, "step": 2976 }, { "epoch": 0.57, "learning_rate": 8.138186771103382e-06, "loss": 0.5777, "step": 2977 }, { "epoch": 0.57, "learning_rate": 8.132064016273325e-06, "loss": 0.5698, "step": 2978 }, { "epoch": 0.57, "learning_rate": 8.12594198692808e-06, "loss": 0.3987, "step": 2979 }, { "epoch": 0.57, "learning_rate": 8.119820685445373e-06, "loss": 0.6001, "step": 2980 }, { "epoch": 0.57, "learning_rate": 8.113700114202647e-06, "loss": 0.5788, "step": 2981 }, { "epoch": 0.57, "learning_rate": 8.107580275577059e-06, "loss": 0.6039, "step": 2982 }, { "epoch": 0.57, "learning_rate": 8.101461171945483e-06, "loss": 0.6033, "step": 2983 }, { "epoch": 0.57, "learning_rate": 8.095342805684516e-06, "loss": 0.5086, "step": 2984 }, { "epoch": 0.57, "learning_rate": 8.089225179170454e-06, "loss": 0.5617, "step": 2985 }, { "epoch": 0.57, "learning_rate": 8.083108294779313e-06, "loss": 0.5979, "step": 2986 }, { "epoch": 0.57, "learning_rate": 8.076992154886826e-06, "loss": 0.6048, "step": 2987 }, { "epoch": 0.57, "learning_rate": 8.070876761868426e-06, "loss": 0.3836, "step": 2988 }, { "epoch": 0.58, "learning_rate": 8.064762118099258e-06, "loss": 0.5839, "step": 2989 }, { "epoch": 0.58, "learning_rate": 8.058648225954188e-06, "loss": 0.576, "step": 2990 }, { "epoch": 0.58, "learning_rate": 8.052535087807774e-06, "loss": 0.584, "step": 2991 }, { "epoch": 0.58, "learning_rate": 8.046422706034294e-06, "loss": 0.6345, "step": 2992 }, { "epoch": 0.58, "learning_rate": 8.040311083007725e-06, "loss": 0.614, "step": 2993 }, { "epoch": 0.58, "learning_rate": 8.034200221101746e-06, "loss": 0.6165, "step": 2994 }, { "epoch": 0.58, "learning_rate": 8.028090122689747e-06, "loss": 0.6079, "step": 2995 }, { "epoch": 0.58, "learning_rate": 8.021980790144828e-06, "loss": 0.4664, "step": 2996 }, { "epoch": 0.58, "learning_rate": 8.015872225839776e-06, "loss": 0.523, "step": 2997 }, { "epoch": 0.58, "learning_rate": 8.009764432147086e-06, "loss": 0.5964, "step": 2998 }, { "epoch": 0.58, "learning_rate": 8.003657411438961e-06, "loss": 0.5998, "step": 2999 }, { "epoch": 0.58, "learning_rate": 7.997551166087293e-06, "loss": 0.5858, "step": 3000 }, { "epoch": 0.58, "learning_rate": 7.991445698463672e-06, "loss": 0.5772, "step": 3001 }, { "epoch": 0.58, "learning_rate": 7.985341010939402e-06, "loss": 0.5757, "step": 3002 }, { "epoch": 0.58, "learning_rate": 7.979237105885467e-06, "loss": 0.5468, "step": 3003 }, { "epoch": 0.58, "learning_rate": 7.973133985672558e-06, "loss": 0.5229, "step": 3004 }, { "epoch": 0.58, "learning_rate": 7.967031652671051e-06, "loss": 0.5966, "step": 3005 }, { "epoch": 0.58, "learning_rate": 7.960930109251023e-06, "loss": 0.5236, "step": 3006 }, { "epoch": 0.58, "learning_rate": 7.954829357782243e-06, "loss": 0.5719, "step": 3007 }, { "epoch": 0.58, "learning_rate": 7.948729400634178e-06, "loss": 0.5539, "step": 3008 }, { "epoch": 0.58, "learning_rate": 7.942630240175977e-06, "loss": 0.4948, "step": 3009 }, { "epoch": 0.58, "learning_rate": 7.936531878776484e-06, "loss": 0.4765, "step": 3010 }, { "epoch": 0.58, "learning_rate": 7.930434318804229e-06, "loss": 0.5692, "step": 3011 }, { "epoch": 0.58, "learning_rate": 7.924337562627435e-06, "loss": 0.5992, "step": 3012 }, { "epoch": 0.58, "learning_rate": 7.918241612614016e-06, "loss": 0.624, "step": 3013 }, { "epoch": 0.58, "learning_rate": 7.91214647113157e-06, "loss": 0.4995, "step": 3014 }, { "epoch": 0.58, "learning_rate": 7.906052140547373e-06, "loss": 0.5873, "step": 3015 }, { "epoch": 0.58, "learning_rate": 7.899958623228398e-06, "loss": 0.5797, "step": 3016 }, { "epoch": 0.58, "learning_rate": 7.893865921541294e-06, "loss": 0.4932, "step": 3017 }, { "epoch": 0.58, "learning_rate": 7.887774037852395e-06, "loss": 0.5773, "step": 3018 }, { "epoch": 0.58, "learning_rate": 7.881682974527723e-06, "loss": 0.6072, "step": 3019 }, { "epoch": 0.58, "learning_rate": 7.875592733932972e-06, "loss": 0.5846, "step": 3020 }, { "epoch": 0.58, "learning_rate": 7.869503318433529e-06, "loss": 0.6118, "step": 3021 }, { "epoch": 0.58, "learning_rate": 7.863414730394444e-06, "loss": 0.3916, "step": 3022 }, { "epoch": 0.58, "learning_rate": 7.857326972180455e-06, "loss": 0.5717, "step": 3023 }, { "epoch": 0.58, "learning_rate": 7.85124004615598e-06, "loss": 0.5933, "step": 3024 }, { "epoch": 0.58, "learning_rate": 7.845153954685114e-06, "loss": 0.4464, "step": 3025 }, { "epoch": 0.58, "learning_rate": 7.839068700131623e-06, "loss": 0.6107, "step": 3026 }, { "epoch": 0.58, "learning_rate": 7.832984284858946e-06, "loss": 0.5748, "step": 3027 }, { "epoch": 0.58, "learning_rate": 7.826900711230204e-06, "loss": 0.4949, "step": 3028 }, { "epoch": 0.58, "learning_rate": 7.820817981608185e-06, "loss": 0.5617, "step": 3029 }, { "epoch": 0.58, "learning_rate": 7.814736098355348e-06, "loss": 0.5472, "step": 3030 }, { "epoch": 0.58, "learning_rate": 7.808655063833832e-06, "loss": 0.5607, "step": 3031 }, { "epoch": 0.58, "learning_rate": 7.802574880405438e-06, "loss": 0.4566, "step": 3032 }, { "epoch": 0.58, "learning_rate": 7.79649555043164e-06, "loss": 0.6252, "step": 3033 }, { "epoch": 0.58, "learning_rate": 7.790417076273581e-06, "loss": 0.5784, "step": 3034 }, { "epoch": 0.58, "learning_rate": 7.784339460292065e-06, "loss": 0.4496, "step": 3035 }, { "epoch": 0.58, "learning_rate": 7.77826270484757e-06, "loss": 0.5146, "step": 3036 }, { "epoch": 0.58, "learning_rate": 7.772186812300244e-06, "loss": 0.4622, "step": 3037 }, { "epoch": 0.58, "learning_rate": 7.766111785009888e-06, "loss": 0.5709, "step": 3038 }, { "epoch": 0.58, "learning_rate": 7.760037625335973e-06, "loss": 0.494, "step": 3039 }, { "epoch": 0.58, "learning_rate": 7.753964335637634e-06, "loss": 0.5484, "step": 3040 }, { "epoch": 0.59, "learning_rate": 7.747891918273668e-06, "loss": 0.5541, "step": 3041 }, { "epoch": 0.59, "learning_rate": 7.741820375602524e-06, "loss": 0.37, "step": 3042 }, { "epoch": 0.59, "learning_rate": 7.735749709982329e-06, "loss": 0.5754, "step": 3043 }, { "epoch": 0.59, "learning_rate": 7.729679923770855e-06, "loss": 0.5876, "step": 3044 }, { "epoch": 0.59, "learning_rate": 7.723611019325538e-06, "loss": 0.6154, "step": 3045 }, { "epoch": 0.59, "learning_rate": 7.71754299900347e-06, "loss": 0.5954, "step": 3046 }, { "epoch": 0.59, "learning_rate": 7.7114758651614e-06, "loss": 0.4834, "step": 3047 }, { "epoch": 0.59, "learning_rate": 7.705409620155733e-06, "loss": 0.5349, "step": 3048 }, { "epoch": 0.59, "learning_rate": 7.699344266342529e-06, "loss": 0.5441, "step": 3049 }, { "epoch": 0.59, "learning_rate": 7.693279806077504e-06, "loss": 0.4699, "step": 3050 }, { "epoch": 0.59, "learning_rate": 7.68721624171602e-06, "loss": 0.601, "step": 3051 }, { "epoch": 0.59, "learning_rate": 7.681153575613098e-06, "loss": 0.6212, "step": 3052 }, { "epoch": 0.59, "learning_rate": 7.675091810123404e-06, "loss": 0.5741, "step": 3053 }, { "epoch": 0.59, "learning_rate": 7.669030947601265e-06, "loss": 0.617, "step": 3054 }, { "epoch": 0.59, "learning_rate": 7.662970990400647e-06, "loss": 0.466, "step": 3055 }, { "epoch": 0.59, "learning_rate": 7.656911940875163e-06, "loss": 0.5717, "step": 3056 }, { "epoch": 0.59, "learning_rate": 7.650853801378084e-06, "loss": 0.5743, "step": 3057 }, { "epoch": 0.59, "learning_rate": 7.644796574262322e-06, "loss": 0.6562, "step": 3058 }, { "epoch": 0.59, "learning_rate": 7.638740261880423e-06, "loss": 0.5426, "step": 3059 }, { "epoch": 0.59, "learning_rate": 7.632684866584606e-06, "loss": 0.5571, "step": 3060 }, { "epoch": 0.59, "learning_rate": 7.626630390726704e-06, "loss": 0.607, "step": 3061 }, { "epoch": 0.59, "learning_rate": 7.620576836658212e-06, "loss": 0.5713, "step": 3062 }, { "epoch": 0.59, "learning_rate": 7.61452420673026e-06, "loss": 0.5035, "step": 3063 }, { "epoch": 0.59, "learning_rate": 7.608472503293615e-06, "loss": 0.5726, "step": 3064 }, { "epoch": 0.59, "learning_rate": 7.60242172869869e-06, "loss": 0.4673, "step": 3065 }, { "epoch": 0.59, "learning_rate": 7.596371885295542e-06, "loss": 0.4717, "step": 3066 }, { "epoch": 0.59, "learning_rate": 7.590322975433857e-06, "loss": 0.5223, "step": 3067 }, { "epoch": 0.59, "learning_rate": 7.584275001462961e-06, "loss": 0.6203, "step": 3068 }, { "epoch": 0.59, "learning_rate": 7.578227965731819e-06, "loss": 0.537, "step": 3069 }, { "epoch": 0.59, "learning_rate": 7.572181870589028e-06, "loss": 0.6123, "step": 3070 }, { "epoch": 0.59, "learning_rate": 7.566136718382821e-06, "loss": 0.4718, "step": 3071 }, { "epoch": 0.59, "learning_rate": 7.560092511461069e-06, "loss": 0.59, "step": 3072 }, { "epoch": 0.59, "learning_rate": 7.55404925217127e-06, "loss": 0.5926, "step": 3073 }, { "epoch": 0.59, "learning_rate": 7.548006942860557e-06, "loss": 0.4625, "step": 3074 }, { "epoch": 0.59, "learning_rate": 7.541965585875695e-06, "loss": 0.593, "step": 3075 }, { "epoch": 0.59, "learning_rate": 7.535925183563073e-06, "loss": 0.5893, "step": 3076 }, { "epoch": 0.59, "learning_rate": 7.529885738268714e-06, "loss": 0.5577, "step": 3077 }, { "epoch": 0.59, "learning_rate": 7.523847252338274e-06, "loss": 0.6096, "step": 3078 }, { "epoch": 0.59, "learning_rate": 7.51780972811703e-06, "loss": 0.5948, "step": 3079 }, { "epoch": 0.59, "learning_rate": 7.511773167949885e-06, "loss": 0.5317, "step": 3080 }, { "epoch": 0.59, "learning_rate": 7.5057375741813685e-06, "loss": 0.5745, "step": 3081 }, { "epoch": 0.59, "learning_rate": 7.499702949155634e-06, "loss": 0.5922, "step": 3082 }, { "epoch": 0.59, "learning_rate": 7.493669295216467e-06, "loss": 0.4587, "step": 3083 }, { "epoch": 0.59, "learning_rate": 7.487636614707265e-06, "loss": 0.4675, "step": 3084 }, { "epoch": 0.59, "learning_rate": 7.48160490997105e-06, "loss": 0.4753, "step": 3085 }, { "epoch": 0.59, "learning_rate": 7.475574183350471e-06, "loss": 0.5126, "step": 3086 }, { "epoch": 0.59, "learning_rate": 7.46954443718779e-06, "loss": 0.5216, "step": 3087 }, { "epoch": 0.59, "learning_rate": 7.463515673824888e-06, "loss": 0.5119, "step": 3088 }, { "epoch": 0.59, "learning_rate": 7.457487895603273e-06, "loss": 0.5802, "step": 3089 }, { "epoch": 0.59, "learning_rate": 7.451461104864061e-06, "loss": 0.3814, "step": 3090 }, { "epoch": 0.59, "learning_rate": 7.44543530394799e-06, "loss": 0.5859, "step": 3091 }, { "epoch": 0.59, "learning_rate": 7.439410495195411e-06, "loss": 0.5735, "step": 3092 }, { "epoch": 0.6, "learning_rate": 7.433386680946288e-06, "loss": 0.5775, "step": 3093 }, { "epoch": 0.6, "learning_rate": 7.427363863540202e-06, "loss": 0.5673, "step": 3094 }, { "epoch": 0.6, "learning_rate": 7.421342045316351e-06, "loss": 0.6184, "step": 3095 }, { "epoch": 0.6, "learning_rate": 7.415321228613534e-06, "loss": 0.5839, "step": 3096 }, { "epoch": 0.6, "learning_rate": 7.409301415770168e-06, "loss": 0.6046, "step": 3097 }, { "epoch": 0.6, "learning_rate": 7.403282609124281e-06, "loss": 0.5235, "step": 3098 }, { "epoch": 0.6, "learning_rate": 7.397264811013507e-06, "loss": 0.4962, "step": 3099 }, { "epoch": 0.6, "learning_rate": 7.391248023775084e-06, "loss": 0.5837, "step": 3100 }, { "epoch": 0.6, "learning_rate": 7.385232249745873e-06, "loss": 0.5652, "step": 3101 }, { "epoch": 0.6, "learning_rate": 7.379217491262325e-06, "loss": 0.6076, "step": 3102 }, { "epoch": 0.6, "learning_rate": 7.373203750660505e-06, "loss": 0.5486, "step": 3103 }, { "epoch": 0.6, "learning_rate": 7.36719103027608e-06, "loss": 0.5854, "step": 3104 }, { "epoch": 0.6, "learning_rate": 7.361179332444318e-06, "loss": 0.4465, "step": 3105 }, { "epoch": 0.6, "learning_rate": 7.355168659500094e-06, "loss": 0.5987, "step": 3106 }, { "epoch": 0.6, "learning_rate": 7.3491590137778915e-06, "loss": 0.5813, "step": 3107 }, { "epoch": 0.6, "learning_rate": 7.343150397611782e-06, "loss": 0.3702, "step": 3108 }, { "epoch": 0.6, "learning_rate": 7.3371428133354435e-06, "loss": 0.4913, "step": 3109 }, { "epoch": 0.6, "learning_rate": 7.33113626328215e-06, "loss": 0.6017, "step": 3110 }, { "epoch": 0.6, "learning_rate": 7.325130749784781e-06, "loss": 0.6172, "step": 3111 }, { "epoch": 0.6, "learning_rate": 7.3191262751758005e-06, "loss": 0.5986, "step": 3112 }, { "epoch": 0.6, "learning_rate": 7.3131228417872905e-06, "loss": 0.6094, "step": 3113 }, { "epoch": 0.6, "learning_rate": 7.307120451950902e-06, "loss": 0.6198, "step": 3114 }, { "epoch": 0.6, "learning_rate": 7.301119107997905e-06, "loss": 0.6029, "step": 3115 }, { "epoch": 0.6, "learning_rate": 7.295118812259145e-06, "loss": 0.6059, "step": 3116 }, { "epoch": 0.6, "learning_rate": 7.289119567065068e-06, "loss": 0.5598, "step": 3117 }, { "epoch": 0.6, "learning_rate": 7.2831213747457155e-06, "loss": 0.6181, "step": 3118 }, { "epoch": 0.6, "learning_rate": 7.2771242376307125e-06, "loss": 0.4929, "step": 3119 }, { "epoch": 0.6, "learning_rate": 7.271128158049283e-06, "loss": 0.6425, "step": 3120 }, { "epoch": 0.6, "learning_rate": 7.2651331383302326e-06, "loss": 0.5636, "step": 3121 }, { "epoch": 0.6, "learning_rate": 7.2591391808019555e-06, "loss": 0.5516, "step": 3122 }, { "epoch": 0.6, "learning_rate": 7.253146287792434e-06, "loss": 0.566, "step": 3123 }, { "epoch": 0.6, "learning_rate": 7.247154461629248e-06, "loss": 0.6006, "step": 3124 }, { "epoch": 0.6, "learning_rate": 7.241163704639547e-06, "loss": 0.462, "step": 3125 }, { "epoch": 0.6, "learning_rate": 7.235174019150071e-06, "loss": 0.495, "step": 3126 }, { "epoch": 0.6, "learning_rate": 7.229185407487149e-06, "loss": 0.5638, "step": 3127 }, { "epoch": 0.6, "learning_rate": 7.2231978719766884e-06, "loss": 0.5723, "step": 3128 }, { "epoch": 0.6, "learning_rate": 7.217211414944171e-06, "loss": 0.578, "step": 3129 }, { "epoch": 0.6, "learning_rate": 7.2112260387146784e-06, "loss": 0.4585, "step": 3130 }, { "epoch": 0.6, "learning_rate": 7.2052417456128565e-06, "loss": 0.5286, "step": 3131 }, { "epoch": 0.6, "learning_rate": 7.199258537962936e-06, "loss": 0.5078, "step": 3132 }, { "epoch": 0.6, "learning_rate": 7.193276418088729e-06, "loss": 0.6151, "step": 3133 }, { "epoch": 0.6, "learning_rate": 7.187295388313618e-06, "loss": 0.5796, "step": 3134 }, { "epoch": 0.6, "learning_rate": 7.181315450960562e-06, "loss": 0.5706, "step": 3135 }, { "epoch": 0.6, "learning_rate": 7.175336608352113e-06, "loss": 0.5928, "step": 3136 }, { "epoch": 0.6, "learning_rate": 7.169358862810374e-06, "loss": 0.4819, "step": 3137 }, { "epoch": 0.6, "learning_rate": 7.163382216657033e-06, "loss": 0.6006, "step": 3138 }, { "epoch": 0.6, "learning_rate": 7.1574066722133565e-06, "loss": 0.5551, "step": 3139 }, { "epoch": 0.6, "learning_rate": 7.151432231800173e-06, "loss": 0.5686, "step": 3140 }, { "epoch": 0.6, "learning_rate": 7.145458897737882e-06, "loss": 0.5858, "step": 3141 }, { "epoch": 0.6, "learning_rate": 7.139486672346466e-06, "loss": 0.5887, "step": 3142 }, { "epoch": 0.6, "learning_rate": 7.133515557945463e-06, "loss": 0.5545, "step": 3143 }, { "epoch": 0.6, "learning_rate": 7.12754555685399e-06, "loss": 0.5654, "step": 3144 }, { "epoch": 0.61, "learning_rate": 7.121576671390722e-06, "loss": 0.5163, "step": 3145 }, { "epoch": 0.61, "learning_rate": 7.115608903873905e-06, "loss": 0.5697, "step": 3146 }, { "epoch": 0.61, "learning_rate": 7.109642256621353e-06, "loss": 0.381, "step": 3147 }, { "epoch": 0.61, "learning_rate": 7.103676731950443e-06, "loss": 0.6095, "step": 3148 }, { "epoch": 0.61, "learning_rate": 7.0977123321781176e-06, "loss": 0.5685, "step": 3149 }, { "epoch": 0.61, "learning_rate": 7.091749059620881e-06, "loss": 0.6026, "step": 3150 }, { "epoch": 0.61, "learning_rate": 7.0857869165947945e-06, "loss": 0.5732, "step": 3151 }, { "epoch": 0.61, "learning_rate": 7.079825905415491e-06, "loss": 0.4898, "step": 3152 }, { "epoch": 0.61, "learning_rate": 7.073866028398153e-06, "loss": 0.5953, "step": 3153 }, { "epoch": 0.61, "learning_rate": 7.067907287857535e-06, "loss": 0.5489, "step": 3154 }, { "epoch": 0.61, "learning_rate": 7.061949686107938e-06, "loss": 0.4923, "step": 3155 }, { "epoch": 0.61, "learning_rate": 7.0559932254632315e-06, "loss": 0.5915, "step": 3156 }, { "epoch": 0.61, "learning_rate": 7.0500379082368305e-06, "loss": 0.6014, "step": 3157 }, { "epoch": 0.61, "learning_rate": 7.044083736741711e-06, "loss": 0.4707, "step": 3158 }, { "epoch": 0.61, "learning_rate": 7.03813071329041e-06, "loss": 0.5656, "step": 3159 }, { "epoch": 0.61, "learning_rate": 7.032178840195009e-06, "loss": 0.5958, "step": 3160 }, { "epoch": 0.61, "learning_rate": 7.026228119767149e-06, "loss": 0.6111, "step": 3161 }, { "epoch": 0.61, "learning_rate": 7.020278554318023e-06, "loss": 0.5306, "step": 3162 }, { "epoch": 0.61, "learning_rate": 7.014330146158367e-06, "loss": 0.4939, "step": 3163 }, { "epoch": 0.61, "learning_rate": 7.008382897598477e-06, "loss": 0.4689, "step": 3164 }, { "epoch": 0.61, "learning_rate": 7.002436810948201e-06, "loss": 0.589, "step": 3165 }, { "epoch": 0.61, "learning_rate": 6.996491888516927e-06, "loss": 0.5881, "step": 3166 }, { "epoch": 0.61, "learning_rate": 6.990548132613592e-06, "loss": 0.5376, "step": 3167 }, { "epoch": 0.61, "learning_rate": 6.984605545546686e-06, "loss": 0.4606, "step": 3168 }, { "epoch": 0.61, "learning_rate": 6.978664129624241e-06, "loss": 0.6286, "step": 3169 }, { "epoch": 0.61, "learning_rate": 6.972723887153828e-06, "loss": 0.5801, "step": 3170 }, { "epoch": 0.61, "learning_rate": 6.9667848204425785e-06, "loss": 0.5751, "step": 3171 }, { "epoch": 0.61, "learning_rate": 6.960846931797152e-06, "loss": 0.4781, "step": 3172 }, { "epoch": 0.61, "learning_rate": 6.9549102235237565e-06, "loss": 0.6153, "step": 3173 }, { "epoch": 0.61, "learning_rate": 6.948974697928144e-06, "loss": 0.5712, "step": 3174 }, { "epoch": 0.61, "learning_rate": 6.943040357315598e-06, "loss": 0.592, "step": 3175 }, { "epoch": 0.61, "learning_rate": 6.9371072039909515e-06, "loss": 0.4756, "step": 3176 }, { "epoch": 0.61, "learning_rate": 6.931175240258576e-06, "loss": 0.51, "step": 3177 }, { "epoch": 0.61, "learning_rate": 6.9252444684223765e-06, "loss": 0.5869, "step": 3178 }, { "epoch": 0.61, "learning_rate": 6.919314890785793e-06, "loss": 0.5624, "step": 3179 }, { "epoch": 0.61, "learning_rate": 6.913386509651807e-06, "loss": 0.5995, "step": 3180 }, { "epoch": 0.61, "learning_rate": 6.907459327322934e-06, "loss": 0.5872, "step": 3181 }, { "epoch": 0.61, "learning_rate": 6.90153334610122e-06, "loss": 0.5813, "step": 3182 }, { "epoch": 0.61, "learning_rate": 6.895608568288255e-06, "loss": 0.6309, "step": 3183 }, { "epoch": 0.61, "learning_rate": 6.889684996185148e-06, "loss": 0.4576, "step": 3184 }, { "epoch": 0.61, "learning_rate": 6.88376263209255e-06, "loss": 0.5171, "step": 3185 }, { "epoch": 0.61, "learning_rate": 6.877841478310639e-06, "loss": 0.6034, "step": 3186 }, { "epoch": 0.61, "learning_rate": 6.871921537139117e-06, "loss": 0.4544, "step": 3187 }, { "epoch": 0.61, "learning_rate": 6.866002810877224e-06, "loss": 0.5886, "step": 3188 }, { "epoch": 0.61, "learning_rate": 6.860085301823729e-06, "loss": 0.464, "step": 3189 }, { "epoch": 0.61, "learning_rate": 6.854169012276923e-06, "loss": 0.6126, "step": 3190 }, { "epoch": 0.61, "learning_rate": 6.848253944534622e-06, "loss": 0.4323, "step": 3191 }, { "epoch": 0.61, "learning_rate": 6.84234010089417e-06, "loss": 0.5988, "step": 3192 }, { "epoch": 0.61, "learning_rate": 6.836427483652436e-06, "loss": 0.4348, "step": 3193 }, { "epoch": 0.61, "learning_rate": 6.830516095105817e-06, "loss": 0.5385, "step": 3194 }, { "epoch": 0.61, "learning_rate": 6.824605937550224e-06, "loss": 0.5847, "step": 3195 }, { "epoch": 0.61, "learning_rate": 6.818697013281093e-06, "loss": 0.4632, "step": 3196 }, { "epoch": 0.62, "learning_rate": 6.8127893245933864e-06, "loss": 0.5699, "step": 3197 }, { "epoch": 0.62, "learning_rate": 6.806882873781579e-06, "loss": 0.5545, "step": 3198 }, { "epoch": 0.62, "learning_rate": 6.800977663139666e-06, "loss": 0.5569, "step": 3199 }, { "epoch": 0.62, "learning_rate": 6.795073694961171e-06, "loss": 0.4895, "step": 3200 }, { "epoch": 0.62, "learning_rate": 6.789170971539119e-06, "loss": 0.5708, "step": 3201 }, { "epoch": 0.62, "learning_rate": 6.783269495166066e-06, "loss": 0.5915, "step": 3202 }, { "epoch": 0.62, "learning_rate": 6.777369268134076e-06, "loss": 0.5978, "step": 3203 }, { "epoch": 0.62, "learning_rate": 6.771470292734723e-06, "loss": 0.4643, "step": 3204 }, { "epoch": 0.62, "learning_rate": 6.7655725712591055e-06, "loss": 0.4228, "step": 3205 }, { "epoch": 0.62, "learning_rate": 6.759676105997834e-06, "loss": 0.5592, "step": 3206 }, { "epoch": 0.62, "learning_rate": 6.753780899241027e-06, "loss": 0.5734, "step": 3207 }, { "epoch": 0.62, "learning_rate": 6.747886953278311e-06, "loss": 0.5635, "step": 3208 }, { "epoch": 0.62, "learning_rate": 6.741994270398826e-06, "loss": 0.4926, "step": 3209 }, { "epoch": 0.62, "learning_rate": 6.736102852891227e-06, "loss": 0.5418, "step": 3210 }, { "epoch": 0.62, "learning_rate": 6.730212703043666e-06, "loss": 0.5676, "step": 3211 }, { "epoch": 0.62, "learning_rate": 6.7243238231438176e-06, "loss": 0.5877, "step": 3212 }, { "epoch": 0.62, "learning_rate": 6.718436215478849e-06, "loss": 0.5891, "step": 3213 }, { "epoch": 0.62, "learning_rate": 6.712549882335442e-06, "loss": 0.5921, "step": 3214 }, { "epoch": 0.62, "learning_rate": 6.70666482599978e-06, "loss": 0.3759, "step": 3215 }, { "epoch": 0.62, "learning_rate": 6.7007810487575475e-06, "loss": 0.5679, "step": 3216 }, { "epoch": 0.62, "learning_rate": 6.694898552893941e-06, "loss": 0.468, "step": 3217 }, { "epoch": 0.62, "learning_rate": 6.6890173406936485e-06, "loss": 0.6004, "step": 3218 }, { "epoch": 0.62, "learning_rate": 6.683137414440872e-06, "loss": 0.5502, "step": 3219 }, { "epoch": 0.62, "learning_rate": 6.677258776419304e-06, "loss": 0.5813, "step": 3220 }, { "epoch": 0.62, "learning_rate": 6.671381428912138e-06, "loss": 0.5943, "step": 3221 }, { "epoch": 0.62, "learning_rate": 6.66550537420207e-06, "loss": 0.6023, "step": 3222 }, { "epoch": 0.62, "learning_rate": 6.659630614571287e-06, "loss": 0.5703, "step": 3223 }, { "epoch": 0.62, "learning_rate": 6.653757152301488e-06, "loss": 0.5195, "step": 3224 }, { "epoch": 0.62, "learning_rate": 6.647884989673849e-06, "loss": 0.5973, "step": 3225 }, { "epoch": 0.62, "learning_rate": 6.642014128969055e-06, "loss": 0.5653, "step": 3226 }, { "epoch": 0.62, "learning_rate": 6.63614457246728e-06, "loss": 0.6168, "step": 3227 }, { "epoch": 0.62, "learning_rate": 6.630276322448188e-06, "loss": 0.6149, "step": 3228 }, { "epoch": 0.62, "learning_rate": 6.624409381190946e-06, "loss": 0.5465, "step": 3229 }, { "epoch": 0.62, "learning_rate": 6.618543750974202e-06, "loss": 0.5308, "step": 3230 }, { "epoch": 0.62, "learning_rate": 6.6126794340761025e-06, "loss": 0.564, "step": 3231 }, { "epoch": 0.62, "learning_rate": 6.606816432774279e-06, "loss": 0.4716, "step": 3232 }, { "epoch": 0.62, "learning_rate": 6.600954749345851e-06, "loss": 0.5033, "step": 3233 }, { "epoch": 0.62, "learning_rate": 6.595094386067428e-06, "loss": 0.4656, "step": 3234 }, { "epoch": 0.62, "learning_rate": 6.589235345215117e-06, "loss": 0.4839, "step": 3235 }, { "epoch": 0.62, "learning_rate": 6.583377629064494e-06, "loss": 0.6052, "step": 3236 }, { "epoch": 0.62, "learning_rate": 6.5775212398906295e-06, "loss": 0.4698, "step": 3237 }, { "epoch": 0.62, "learning_rate": 6.571666179968079e-06, "loss": 0.645, "step": 3238 }, { "epoch": 0.62, "learning_rate": 6.565812451570881e-06, "loss": 0.6191, "step": 3239 }, { "epoch": 0.62, "learning_rate": 6.5599600569725495e-06, "loss": 0.5745, "step": 3240 }, { "epoch": 0.62, "learning_rate": 6.554108998446096e-06, "loss": 0.4857, "step": 3241 }, { "epoch": 0.62, "learning_rate": 6.548259278263999e-06, "loss": 0.5941, "step": 3242 }, { "epoch": 0.62, "learning_rate": 6.542410898698226e-06, "loss": 0.5544, "step": 3243 }, { "epoch": 0.62, "learning_rate": 6.536563862020218e-06, "loss": 0.4569, "step": 3244 }, { "epoch": 0.62, "learning_rate": 6.530718170500896e-06, "loss": 0.5589, "step": 3245 }, { "epoch": 0.62, "learning_rate": 6.524873826410658e-06, "loss": 0.549, "step": 3246 }, { "epoch": 0.62, "learning_rate": 6.519030832019383e-06, "loss": 0.5954, "step": 3247 }, { "epoch": 0.62, "learning_rate": 6.513189189596422e-06, "loss": 0.6064, "step": 3248 }, { "epoch": 0.63, "learning_rate": 6.507348901410604e-06, "loss": 0.5937, "step": 3249 }, { "epoch": 0.63, "learning_rate": 6.501509969730224e-06, "loss": 0.5556, "step": 3250 }, { "epoch": 0.63, "learning_rate": 6.495672396823061e-06, "loss": 0.5696, "step": 3251 }, { "epoch": 0.63, "learning_rate": 6.489836184956353e-06, "loss": 0.4915, "step": 3252 }, { "epoch": 0.63, "learning_rate": 6.484001336396828e-06, "loss": 0.5832, "step": 3253 }, { "epoch": 0.63, "learning_rate": 6.478167853410668e-06, "loss": 0.6103, "step": 3254 }, { "epoch": 0.63, "learning_rate": 6.472335738263534e-06, "loss": 0.5677, "step": 3255 }, { "epoch": 0.63, "learning_rate": 6.466504993220548e-06, "loss": 0.5673, "step": 3256 }, { "epoch": 0.63, "learning_rate": 6.460675620546305e-06, "loss": 0.4947, "step": 3257 }, { "epoch": 0.63, "learning_rate": 6.454847622504867e-06, "loss": 0.5704, "step": 3258 }, { "epoch": 0.63, "learning_rate": 6.4490210013597635e-06, "loss": 0.5462, "step": 3259 }, { "epoch": 0.63, "learning_rate": 6.4431957593739845e-06, "loss": 0.5021, "step": 3260 }, { "epoch": 0.63, "learning_rate": 6.4373718988099896e-06, "loss": 0.5086, "step": 3261 }, { "epoch": 0.63, "learning_rate": 6.431549421929694e-06, "loss": 0.4948, "step": 3262 }, { "epoch": 0.63, "learning_rate": 6.4257283309944804e-06, "loss": 0.5953, "step": 3263 }, { "epoch": 0.63, "learning_rate": 6.419908628265203e-06, "loss": 0.5692, "step": 3264 }, { "epoch": 0.63, "learning_rate": 6.414090316002161e-06, "loss": 0.5598, "step": 3265 }, { "epoch": 0.63, "learning_rate": 6.4082733964651166e-06, "loss": 0.6275, "step": 3266 }, { "epoch": 0.63, "learning_rate": 6.4024578719133e-06, "loss": 0.5926, "step": 3267 }, { "epoch": 0.63, "learning_rate": 6.396643744605391e-06, "loss": 0.5654, "step": 3268 }, { "epoch": 0.63, "learning_rate": 6.390831016799527e-06, "loss": 0.521, "step": 3269 }, { "epoch": 0.63, "learning_rate": 6.385019690753311e-06, "loss": 0.5749, "step": 3270 }, { "epoch": 0.63, "learning_rate": 6.379209768723791e-06, "loss": 0.5669, "step": 3271 }, { "epoch": 0.63, "learning_rate": 6.373401252967475e-06, "loss": 0.5946, "step": 3272 }, { "epoch": 0.63, "learning_rate": 6.367594145740324e-06, "loss": 0.5719, "step": 3273 }, { "epoch": 0.63, "learning_rate": 6.361788449297748e-06, "loss": 0.4716, "step": 3274 }, { "epoch": 0.63, "learning_rate": 6.355984165894613e-06, "loss": 0.5488, "step": 3275 }, { "epoch": 0.63, "learning_rate": 6.350181297785242e-06, "loss": 0.589, "step": 3276 }, { "epoch": 0.63, "learning_rate": 6.344379847223398e-06, "loss": 0.4908, "step": 3277 }, { "epoch": 0.63, "learning_rate": 6.338579816462298e-06, "loss": 0.5639, "step": 3278 }, { "epoch": 0.63, "learning_rate": 6.332781207754605e-06, "loss": 0.5663, "step": 3279 }, { "epoch": 0.63, "learning_rate": 6.326984023352435e-06, "loss": 0.4736, "step": 3280 }, { "epoch": 0.63, "learning_rate": 6.321188265507342e-06, "loss": 0.6071, "step": 3281 }, { "epoch": 0.63, "learning_rate": 6.31539393647034e-06, "loss": 0.6079, "step": 3282 }, { "epoch": 0.63, "learning_rate": 6.309601038491874e-06, "loss": 0.589, "step": 3283 }, { "epoch": 0.63, "learning_rate": 6.303809573821842e-06, "loss": 0.4763, "step": 3284 }, { "epoch": 0.63, "learning_rate": 6.298019544709579e-06, "loss": 0.6032, "step": 3285 }, { "epoch": 0.63, "learning_rate": 6.292230953403866e-06, "loss": 0.5339, "step": 3286 }, { "epoch": 0.63, "learning_rate": 6.286443802152926e-06, "loss": 0.5892, "step": 3287 }, { "epoch": 0.63, "learning_rate": 6.280658093204422e-06, "loss": 0.5679, "step": 3288 }, { "epoch": 0.63, "learning_rate": 6.274873828805459e-06, "loss": 0.5511, "step": 3289 }, { "epoch": 0.63, "learning_rate": 6.269091011202576e-06, "loss": 0.6179, "step": 3290 }, { "epoch": 0.63, "learning_rate": 6.263309642641751e-06, "loss": 0.5972, "step": 3291 }, { "epoch": 0.63, "learning_rate": 6.257529725368405e-06, "loss": 0.4881, "step": 3292 }, { "epoch": 0.63, "learning_rate": 6.251751261627386e-06, "loss": 0.5803, "step": 3293 }, { "epoch": 0.63, "learning_rate": 6.245974253662988e-06, "loss": 0.5858, "step": 3294 }, { "epoch": 0.63, "learning_rate": 6.240198703718932e-06, "loss": 0.5274, "step": 3295 }, { "epoch": 0.63, "learning_rate": 6.234424614038375e-06, "loss": 0.5744, "step": 3296 }, { "epoch": 0.63, "learning_rate": 6.2286519868639095e-06, "loss": 0.5889, "step": 3297 }, { "epoch": 0.63, "learning_rate": 6.222880824437549e-06, "loss": 0.6064, "step": 3298 }, { "epoch": 0.63, "learning_rate": 6.217111129000759e-06, "loss": 0.3908, "step": 3299 }, { "epoch": 0.63, "learning_rate": 6.211342902794413e-06, "loss": 0.4846, "step": 3300 }, { "epoch": 0.64, "learning_rate": 6.205576148058828e-06, "loss": 0.5681, "step": 3301 }, { "epoch": 0.64, "learning_rate": 6.199810867033745e-06, "loss": 0.5089, "step": 3302 }, { "epoch": 0.64, "learning_rate": 6.19404706195833e-06, "loss": 0.5134, "step": 3303 }, { "epoch": 0.64, "learning_rate": 6.188284735071177e-06, "loss": 0.5664, "step": 3304 }, { "epoch": 0.64, "learning_rate": 6.182523888610316e-06, "loss": 0.5692, "step": 3305 }, { "epoch": 0.64, "learning_rate": 6.176764524813187e-06, "loss": 0.5856, "step": 3306 }, { "epoch": 0.64, "learning_rate": 6.171006645916662e-06, "loss": 0.4891, "step": 3307 }, { "epoch": 0.64, "learning_rate": 6.165250254157032e-06, "loss": 0.5788, "step": 3308 }, { "epoch": 0.64, "learning_rate": 6.159495351770017e-06, "loss": 0.5913, "step": 3309 }, { "epoch": 0.64, "learning_rate": 6.153741940990749e-06, "loss": 0.6087, "step": 3310 }, { "epoch": 0.64, "learning_rate": 6.1479900240537956e-06, "loss": 0.5642, "step": 3311 }, { "epoch": 0.64, "learning_rate": 6.142239603193128e-06, "loss": 0.4731, "step": 3312 }, { "epoch": 0.64, "learning_rate": 6.136490680642146e-06, "loss": 0.4893, "step": 3313 }, { "epoch": 0.64, "learning_rate": 6.130743258633667e-06, "loss": 0.5589, "step": 3314 }, { "epoch": 0.64, "learning_rate": 6.124997339399916e-06, "loss": 0.4526, "step": 3315 }, { "epoch": 0.64, "learning_rate": 6.119252925172549e-06, "loss": 0.5527, "step": 3316 }, { "epoch": 0.64, "learning_rate": 6.113510018182628e-06, "loss": 0.61, "step": 3317 }, { "epoch": 0.64, "learning_rate": 6.107768620660633e-06, "loss": 0.5806, "step": 3318 }, { "epoch": 0.64, "learning_rate": 6.102028734836456e-06, "loss": 0.5878, "step": 3319 }, { "epoch": 0.64, "learning_rate": 6.0962903629394e-06, "loss": 0.5912, "step": 3320 }, { "epoch": 0.64, "learning_rate": 6.090553507198187e-06, "loss": 0.5886, "step": 3321 }, { "epoch": 0.64, "learning_rate": 6.0848181698409384e-06, "loss": 0.485, "step": 3322 }, { "epoch": 0.64, "learning_rate": 6.079084353095202e-06, "loss": 0.6147, "step": 3323 }, { "epoch": 0.64, "learning_rate": 6.07335205918792e-06, "loss": 0.6025, "step": 3324 }, { "epoch": 0.64, "learning_rate": 6.067621290345455e-06, "loss": 0.5634, "step": 3325 }, { "epoch": 0.64, "learning_rate": 6.061892048793568e-06, "loss": 0.469, "step": 3326 }, { "epoch": 0.64, "learning_rate": 6.056164336757426e-06, "loss": 0.5547, "step": 3327 }, { "epoch": 0.64, "learning_rate": 6.050438156461613e-06, "loss": 0.5769, "step": 3328 }, { "epoch": 0.64, "learning_rate": 6.044713510130108e-06, "loss": 0.5427, "step": 3329 }, { "epoch": 0.64, "learning_rate": 6.038990399986302e-06, "loss": 0.5814, "step": 3330 }, { "epoch": 0.64, "learning_rate": 6.03326882825298e-06, "loss": 0.6027, "step": 3331 }, { "epoch": 0.64, "learning_rate": 6.027548797152336e-06, "loss": 0.582, "step": 3332 }, { "epoch": 0.64, "learning_rate": 6.021830308905963e-06, "loss": 0.569, "step": 3333 }, { "epoch": 0.64, "learning_rate": 6.016113365734861e-06, "loss": 0.592, "step": 3334 }, { "epoch": 0.64, "learning_rate": 6.0103979698594215e-06, "loss": 0.5735, "step": 3335 }, { "epoch": 0.64, "learning_rate": 6.004684123499436e-06, "loss": 0.5832, "step": 3336 }, { "epoch": 0.64, "learning_rate": 5.998971828874102e-06, "loss": 0.5724, "step": 3337 }, { "epoch": 0.64, "learning_rate": 5.993261088202005e-06, "loss": 0.6013, "step": 3338 }, { "epoch": 0.64, "learning_rate": 5.987551903701128e-06, "loss": 0.5693, "step": 3339 }, { "epoch": 0.64, "learning_rate": 5.9818442775888595e-06, "loss": 0.4723, "step": 3340 }, { "epoch": 0.64, "learning_rate": 5.97613821208197e-06, "loss": 0.6181, "step": 3341 }, { "epoch": 0.64, "learning_rate": 5.970433709396635e-06, "loss": 0.6204, "step": 3342 }, { "epoch": 0.64, "learning_rate": 5.964730771748415e-06, "loss": 0.5823, "step": 3343 }, { "epoch": 0.64, "learning_rate": 5.959029401352262e-06, "loss": 0.5679, "step": 3344 }, { "epoch": 0.64, "learning_rate": 5.953329600422524e-06, "loss": 0.5735, "step": 3345 }, { "epoch": 0.64, "learning_rate": 5.947631371172943e-06, "loss": 0.4687, "step": 3346 }, { "epoch": 0.64, "learning_rate": 5.941934715816642e-06, "loss": 0.6238, "step": 3347 }, { "epoch": 0.64, "learning_rate": 5.936239636566137e-06, "loss": 0.6219, "step": 3348 }, { "epoch": 0.64, "learning_rate": 5.930546135633327e-06, "loss": 0.583, "step": 3349 }, { "epoch": 0.64, "learning_rate": 5.924854215229509e-06, "loss": 0.4807, "step": 3350 }, { "epoch": 0.64, "learning_rate": 5.919163877565351e-06, "loss": 0.5022, "step": 3351 }, { "epoch": 0.64, "learning_rate": 5.9134751248509236e-06, "loss": 0.6036, "step": 3352 }, { "epoch": 0.65, "learning_rate": 5.9077879592956675e-06, "loss": 0.5914, "step": 3353 }, { "epoch": 0.65, "learning_rate": 5.902102383108415e-06, "loss": 0.5671, "step": 3354 }, { "epoch": 0.65, "learning_rate": 5.896418398497377e-06, "loss": 0.5812, "step": 3355 }, { "epoch": 0.65, "learning_rate": 5.890736007670144e-06, "loss": 0.5497, "step": 3356 }, { "epoch": 0.65, "learning_rate": 5.885055212833696e-06, "loss": 0.6067, "step": 3357 }, { "epoch": 0.65, "learning_rate": 5.879376016194387e-06, "loss": 0.5733, "step": 3358 }, { "epoch": 0.65, "learning_rate": 5.873698419957952e-06, "loss": 0.5736, "step": 3359 }, { "epoch": 0.65, "learning_rate": 5.8680224263295045e-06, "loss": 0.6437, "step": 3360 }, { "epoch": 0.65, "learning_rate": 5.862348037513533e-06, "loss": 0.6151, "step": 3361 }, { "epoch": 0.65, "learning_rate": 5.856675255713905e-06, "loss": 0.5106, "step": 3362 }, { "epoch": 0.65, "learning_rate": 5.851004083133862e-06, "loss": 0.4733, "step": 3363 }, { "epoch": 0.65, "learning_rate": 5.8453345219760275e-06, "loss": 0.5999, "step": 3364 }, { "epoch": 0.65, "learning_rate": 5.839666574442389e-06, "loss": 0.5708, "step": 3365 }, { "epoch": 0.65, "learning_rate": 5.834000242734317e-06, "loss": 0.5162, "step": 3366 }, { "epoch": 0.65, "learning_rate": 5.828335529052541e-06, "loss": 0.6014, "step": 3367 }, { "epoch": 0.65, "learning_rate": 5.822672435597172e-06, "loss": 0.5015, "step": 3368 }, { "epoch": 0.65, "learning_rate": 5.817010964567702e-06, "loss": 0.5951, "step": 3369 }, { "epoch": 0.65, "learning_rate": 5.811351118162969e-06, "loss": 0.5691, "step": 3370 }, { "epoch": 0.65, "learning_rate": 5.805692898581196e-06, "loss": 0.5825, "step": 3371 }, { "epoch": 0.65, "learning_rate": 5.800036308019974e-06, "loss": 0.5433, "step": 3372 }, { "epoch": 0.65, "learning_rate": 5.79438134867625e-06, "loss": 0.5852, "step": 3373 }, { "epoch": 0.65, "learning_rate": 5.788728022746348e-06, "loss": 0.5903, "step": 3374 }, { "epoch": 0.65, "learning_rate": 5.783076332425957e-06, "loss": 0.4983, "step": 3375 }, { "epoch": 0.65, "learning_rate": 5.777426279910125e-06, "loss": 0.559, "step": 3376 }, { "epoch": 0.65, "learning_rate": 5.771777867393275e-06, "loss": 0.5896, "step": 3377 }, { "epoch": 0.65, "learning_rate": 5.766131097069174e-06, "loss": 0.5559, "step": 3378 }, { "epoch": 0.65, "learning_rate": 5.760485971130969e-06, "loss": 0.3716, "step": 3379 }, { "epoch": 0.65, "learning_rate": 5.7548424917711596e-06, "loss": 0.6081, "step": 3380 }, { "epoch": 0.65, "learning_rate": 5.749200661181611e-06, "loss": 0.5556, "step": 3381 }, { "epoch": 0.65, "learning_rate": 5.7435604815535475e-06, "loss": 0.5732, "step": 3382 }, { "epoch": 0.65, "learning_rate": 5.7379219550775415e-06, "loss": 0.5571, "step": 3383 }, { "epoch": 0.65, "learning_rate": 5.732285083943537e-06, "loss": 0.5626, "step": 3384 }, { "epoch": 0.65, "learning_rate": 5.726649870340833e-06, "loss": 0.5798, "step": 3385 }, { "epoch": 0.65, "learning_rate": 5.721016316458068e-06, "loss": 0.4679, "step": 3386 }, { "epoch": 0.65, "learning_rate": 5.715384424483268e-06, "loss": 0.548, "step": 3387 }, { "epoch": 0.65, "learning_rate": 5.709754196603781e-06, "loss": 0.5511, "step": 3388 }, { "epoch": 0.65, "learning_rate": 5.704125635006329e-06, "loss": 0.5729, "step": 3389 }, { "epoch": 0.65, "learning_rate": 5.6984987418769825e-06, "loss": 0.5998, "step": 3390 }, { "epoch": 0.65, "learning_rate": 5.692873519401154e-06, "loss": 0.5913, "step": 3391 }, { "epoch": 0.65, "learning_rate": 5.6872499697636195e-06, "loss": 0.567, "step": 3392 }, { "epoch": 0.65, "learning_rate": 5.681628095148502e-06, "loss": 0.5827, "step": 3393 }, { "epoch": 0.65, "learning_rate": 5.6760078977392706e-06, "loss": 0.5425, "step": 3394 }, { "epoch": 0.65, "learning_rate": 5.67038937971875e-06, "loss": 0.603, "step": 3395 }, { "epoch": 0.65, "learning_rate": 5.664772543269101e-06, "loss": 0.5906, "step": 3396 }, { "epoch": 0.65, "learning_rate": 5.659157390571842e-06, "loss": 0.569, "step": 3397 }, { "epoch": 0.65, "learning_rate": 5.653543923807833e-06, "loss": 0.538, "step": 3398 }, { "epoch": 0.65, "learning_rate": 5.6479321451572785e-06, "loss": 0.3975, "step": 3399 }, { "epoch": 0.65, "learning_rate": 5.642322056799732e-06, "loss": 0.5929, "step": 3400 }, { "epoch": 0.65, "learning_rate": 5.636713660914087e-06, "loss": 0.5947, "step": 3401 }, { "epoch": 0.65, "learning_rate": 5.631106959678575e-06, "loss": 0.5914, "step": 3402 }, { "epoch": 0.65, "learning_rate": 5.625501955270777e-06, "loss": 0.4727, "step": 3403 }, { "epoch": 0.65, "learning_rate": 5.619898649867612e-06, "loss": 0.5818, "step": 3404 }, { "epoch": 0.66, "learning_rate": 5.614297045645339e-06, "loss": 0.5487, "step": 3405 }, { "epoch": 0.66, "learning_rate": 5.6086971447795625e-06, "loss": 0.5758, "step": 3406 }, { "epoch": 0.66, "learning_rate": 5.603098949445209e-06, "loss": 0.5678, "step": 3407 }, { "epoch": 0.66, "learning_rate": 5.597502461816557e-06, "loss": 0.4725, "step": 3408 }, { "epoch": 0.66, "learning_rate": 5.5919076840672215e-06, "loss": 0.606, "step": 3409 }, { "epoch": 0.66, "learning_rate": 5.5863146183701454e-06, "loss": 0.5884, "step": 3410 }, { "epoch": 0.66, "learning_rate": 5.580723266897616e-06, "loss": 0.5766, "step": 3411 }, { "epoch": 0.66, "learning_rate": 5.575133631821243e-06, "loss": 0.6116, "step": 3412 }, { "epoch": 0.66, "learning_rate": 5.5695457153119806e-06, "loss": 0.5775, "step": 3413 }, { "epoch": 0.66, "learning_rate": 5.563959519540114e-06, "loss": 0.6037, "step": 3414 }, { "epoch": 0.66, "learning_rate": 5.558375046675244e-06, "loss": 0.5787, "step": 3415 }, { "epoch": 0.66, "learning_rate": 5.552792298886335e-06, "loss": 0.5476, "step": 3416 }, { "epoch": 0.66, "learning_rate": 5.547211278341646e-06, "loss": 0.5523, "step": 3417 }, { "epoch": 0.66, "learning_rate": 5.541631987208789e-06, "loss": 0.5406, "step": 3418 }, { "epoch": 0.66, "learning_rate": 5.536054427654698e-06, "loss": 0.5667, "step": 3419 }, { "epoch": 0.66, "learning_rate": 5.530478601845624e-06, "loss": 0.5541, "step": 3420 }, { "epoch": 0.66, "learning_rate": 5.52490451194716e-06, "loss": 0.5665, "step": 3421 }, { "epoch": 0.66, "learning_rate": 5.519332160124215e-06, "loss": 0.5528, "step": 3422 }, { "epoch": 0.66, "learning_rate": 5.513761548541032e-06, "loss": 0.3604, "step": 3423 }, { "epoch": 0.66, "learning_rate": 5.508192679361169e-06, "loss": 0.4748, "step": 3424 }, { "epoch": 0.66, "learning_rate": 5.502625554747508e-06, "loss": 0.5781, "step": 3425 }, { "epoch": 0.66, "learning_rate": 5.497060176862259e-06, "loss": 0.594, "step": 3426 }, { "epoch": 0.66, "learning_rate": 5.491496547866948e-06, "loss": 0.604, "step": 3427 }, { "epoch": 0.66, "learning_rate": 5.485934669922428e-06, "loss": 0.5265, "step": 3428 }, { "epoch": 0.66, "learning_rate": 5.480374545188866e-06, "loss": 0.504, "step": 3429 }, { "epoch": 0.66, "learning_rate": 5.474816175825754e-06, "loss": 0.5898, "step": 3430 }, { "epoch": 0.66, "learning_rate": 5.469259563991894e-06, "loss": 0.5842, "step": 3431 }, { "epoch": 0.66, "learning_rate": 5.46370471184541e-06, "loss": 0.4511, "step": 3432 }, { "epoch": 0.66, "learning_rate": 5.458151621543744e-06, "loss": 0.4745, "step": 3433 }, { "epoch": 0.66, "learning_rate": 5.452600295243653e-06, "loss": 0.5775, "step": 3434 }, { "epoch": 0.66, "learning_rate": 5.4470507351012116e-06, "loss": 0.5836, "step": 3435 }, { "epoch": 0.66, "learning_rate": 5.441502943271797e-06, "loss": 0.5973, "step": 3436 }, { "epoch": 0.66, "learning_rate": 5.4359569219101115e-06, "loss": 0.5808, "step": 3437 }, { "epoch": 0.66, "learning_rate": 5.430412673170167e-06, "loss": 0.5438, "step": 3438 }, { "epoch": 0.66, "learning_rate": 5.424870199205283e-06, "loss": 0.5508, "step": 3439 }, { "epoch": 0.66, "learning_rate": 5.4193295021681e-06, "loss": 0.5434, "step": 3440 }, { "epoch": 0.66, "learning_rate": 5.413790584210551e-06, "loss": 0.5671, "step": 3441 }, { "epoch": 0.66, "learning_rate": 5.408253447483892e-06, "loss": 0.5845, "step": 3442 }, { "epoch": 0.66, "learning_rate": 5.402718094138688e-06, "loss": 0.6191, "step": 3443 }, { "epoch": 0.66, "learning_rate": 5.397184526324792e-06, "loss": 0.5328, "step": 3444 }, { "epoch": 0.66, "learning_rate": 5.391652746191398e-06, "loss": 0.4718, "step": 3445 }, { "epoch": 0.66, "learning_rate": 5.38612275588697e-06, "loss": 0.5191, "step": 3446 }, { "epoch": 0.66, "learning_rate": 5.380594557559298e-06, "loss": 0.5949, "step": 3447 }, { "epoch": 0.66, "learning_rate": 5.375068153355474e-06, "loss": 0.5604, "step": 3448 }, { "epoch": 0.66, "learning_rate": 5.369543545421883e-06, "loss": 0.452, "step": 3449 }, { "epoch": 0.66, "learning_rate": 5.364020735904223e-06, "loss": 0.556, "step": 3450 }, { "epoch": 0.66, "learning_rate": 5.358499726947488e-06, "loss": 0.588, "step": 3451 }, { "epoch": 0.66, "learning_rate": 5.352980520695974e-06, "loss": 0.5833, "step": 3452 }, { "epoch": 0.66, "learning_rate": 5.347463119293283e-06, "loss": 0.5167, "step": 3453 }, { "epoch": 0.66, "learning_rate": 5.341947524882301e-06, "loss": 0.5835, "step": 3454 }, { "epoch": 0.66, "learning_rate": 5.336433739605227e-06, "loss": 0.5685, "step": 3455 }, { "epoch": 0.66, "learning_rate": 5.330921765603549e-06, "loss": 0.5551, "step": 3456 }, { "epoch": 0.67, "learning_rate": 5.325411605018056e-06, "loss": 0.4768, "step": 3457 }, { "epoch": 0.67, "learning_rate": 5.31990325998883e-06, "loss": 0.4926, "step": 3458 }, { "epoch": 0.67, "learning_rate": 5.314396732655253e-06, "loss": 0.5502, "step": 3459 }, { "epoch": 0.67, "learning_rate": 5.308892025155989e-06, "loss": 0.5145, "step": 3460 }, { "epoch": 0.67, "learning_rate": 5.303389139629007e-06, "loss": 0.6045, "step": 3461 }, { "epoch": 0.67, "learning_rate": 5.297888078211564e-06, "loss": 0.5434, "step": 3462 }, { "epoch": 0.67, "learning_rate": 5.2923888430402085e-06, "loss": 0.5654, "step": 3463 }, { "epoch": 0.67, "learning_rate": 5.286891436250785e-06, "loss": 0.5727, "step": 3464 }, { "epoch": 0.67, "learning_rate": 5.281395859978414e-06, "loss": 0.5974, "step": 3465 }, { "epoch": 0.67, "learning_rate": 5.2759021163575184e-06, "loss": 0.5411, "step": 3466 }, { "epoch": 0.67, "learning_rate": 5.27041020752181e-06, "loss": 0.5775, "step": 3467 }, { "epoch": 0.67, "learning_rate": 5.26492013560427e-06, "loss": 0.5031, "step": 3468 }, { "epoch": 0.67, "learning_rate": 5.259431902737195e-06, "loss": 0.5613, "step": 3469 }, { "epoch": 0.67, "learning_rate": 5.2539455110521385e-06, "loss": 0.4798, "step": 3470 }, { "epoch": 0.67, "learning_rate": 5.248460962679958e-06, "loss": 0.5783, "step": 3471 }, { "epoch": 0.67, "learning_rate": 5.24297825975079e-06, "loss": 0.5421, "step": 3472 }, { "epoch": 0.67, "learning_rate": 5.237497404394044e-06, "loss": 0.5485, "step": 3473 }, { "epoch": 0.67, "learning_rate": 5.232018398738436e-06, "loss": 0.5883, "step": 3474 }, { "epoch": 0.67, "learning_rate": 5.226541244911936e-06, "loss": 0.5685, "step": 3475 }, { "epoch": 0.67, "learning_rate": 5.221065945041811e-06, "loss": 0.5864, "step": 3476 }, { "epoch": 0.67, "learning_rate": 5.215592501254609e-06, "loss": 0.4663, "step": 3477 }, { "epoch": 0.67, "learning_rate": 5.210120915676147e-06, "loss": 0.5448, "step": 3478 }, { "epoch": 0.67, "learning_rate": 5.2046511904315265e-06, "loss": 0.5895, "step": 3479 }, { "epoch": 0.67, "learning_rate": 5.199183327645128e-06, "loss": 0.3733, "step": 3480 }, { "epoch": 0.67, "learning_rate": 5.193717329440604e-06, "loss": 0.5195, "step": 3481 }, { "epoch": 0.67, "learning_rate": 5.188253197940889e-06, "loss": 0.5554, "step": 3482 }, { "epoch": 0.67, "learning_rate": 5.182790935268185e-06, "loss": 0.5563, "step": 3483 }, { "epoch": 0.67, "learning_rate": 5.177330543543971e-06, "loss": 0.6144, "step": 3484 }, { "epoch": 0.67, "learning_rate": 5.171872024889004e-06, "loss": 0.5621, "step": 3485 }, { "epoch": 0.67, "learning_rate": 5.166415381423306e-06, "loss": 0.5624, "step": 3486 }, { "epoch": 0.67, "learning_rate": 5.160960615266179e-06, "loss": 0.5932, "step": 3487 }, { "epoch": 0.67, "learning_rate": 5.155507728536191e-06, "loss": 0.5267, "step": 3488 }, { "epoch": 0.67, "learning_rate": 5.150056723351173e-06, "loss": 0.5754, "step": 3489 }, { "epoch": 0.67, "learning_rate": 5.14460760182824e-06, "loss": 0.5859, "step": 3490 }, { "epoch": 0.67, "learning_rate": 5.139160366083765e-06, "loss": 0.5877, "step": 3491 }, { "epoch": 0.67, "learning_rate": 5.133715018233393e-06, "loss": 0.5269, "step": 3492 }, { "epoch": 0.67, "learning_rate": 5.128271560392037e-06, "loss": 0.6164, "step": 3493 }, { "epoch": 0.67, "learning_rate": 5.122829994673866e-06, "loss": 0.6118, "step": 3494 }, { "epoch": 0.67, "learning_rate": 5.117390323192326e-06, "loss": 0.4497, "step": 3495 }, { "epoch": 0.67, "learning_rate": 5.111952548060126e-06, "loss": 0.5268, "step": 3496 }, { "epoch": 0.67, "learning_rate": 5.106516671389224e-06, "loss": 0.5303, "step": 3497 }, { "epoch": 0.67, "learning_rate": 5.101082695290866e-06, "loss": 0.4756, "step": 3498 }, { "epoch": 0.67, "learning_rate": 5.0956506218755344e-06, "loss": 0.4793, "step": 3499 }, { "epoch": 0.67, "learning_rate": 5.09022045325299e-06, "loss": 0.5718, "step": 3500 }, { "epoch": 0.67, "learning_rate": 5.0847921915322486e-06, "loss": 0.6266, "step": 3501 }, { "epoch": 0.67, "learning_rate": 5.07936583882158e-06, "loss": 0.5653, "step": 3502 }, { "epoch": 0.67, "learning_rate": 5.073941397228518e-06, "loss": 0.6087, "step": 3503 }, { "epoch": 0.67, "learning_rate": 5.068518868859854e-06, "loss": 0.4806, "step": 3504 }, { "epoch": 0.67, "learning_rate": 5.063098255821637e-06, "loss": 0.5522, "step": 3505 }, { "epoch": 0.67, "learning_rate": 5.0576795602191734e-06, "loss": 0.6085, "step": 3506 }, { "epoch": 0.67, "learning_rate": 5.052262784157014e-06, "loss": 0.56, "step": 3507 }, { "epoch": 0.67, "learning_rate": 5.046847929738971e-06, "loss": 0.5704, "step": 3508 }, { "epoch": 0.68, "learning_rate": 5.041434999068127e-06, "loss": 0.539, "step": 3509 }, { "epoch": 0.68, "learning_rate": 5.036023994246787e-06, "loss": 0.4621, "step": 3510 }, { "epoch": 0.68, "learning_rate": 5.030614917376532e-06, "loss": 0.3678, "step": 3511 }, { "epoch": 0.68, "learning_rate": 5.025207770558176e-06, "loss": 0.5825, "step": 3512 }, { "epoch": 0.68, "learning_rate": 5.0198025558917985e-06, "loss": 0.4786, "step": 3513 }, { "epoch": 0.68, "learning_rate": 5.014399275476721e-06, "loss": 0.5477, "step": 3514 }, { "epoch": 0.68, "learning_rate": 5.008997931411517e-06, "loss": 0.5034, "step": 3515 }, { "epoch": 0.68, "learning_rate": 5.003598525794002e-06, "loss": 0.5354, "step": 3516 }, { "epoch": 0.68, "learning_rate": 4.998201060721253e-06, "loss": 0.5739, "step": 3517 }, { "epoch": 0.68, "learning_rate": 4.992805538289571e-06, "loss": 0.5765, "step": 3518 }, { "epoch": 0.68, "learning_rate": 4.987411960594521e-06, "loss": 0.6145, "step": 3519 }, { "epoch": 0.68, "learning_rate": 4.982020329730904e-06, "loss": 0.5383, "step": 3520 }, { "epoch": 0.68, "learning_rate": 4.976630647792771e-06, "loss": 0.5746, "step": 3521 }, { "epoch": 0.68, "learning_rate": 4.971242916873412e-06, "loss": 0.382, "step": 3522 }, { "epoch": 0.68, "learning_rate": 4.965857139065354e-06, "loss": 0.571, "step": 3523 }, { "epoch": 0.68, "learning_rate": 4.9604733164603755e-06, "loss": 0.5707, "step": 3524 }, { "epoch": 0.68, "learning_rate": 4.955091451149495e-06, "loss": 0.4477, "step": 3525 }, { "epoch": 0.68, "learning_rate": 4.9497115452229535e-06, "loss": 0.5716, "step": 3526 }, { "epoch": 0.68, "learning_rate": 4.9443336007702614e-06, "loss": 0.5895, "step": 3527 }, { "epoch": 0.68, "learning_rate": 4.938957619880138e-06, "loss": 0.6062, "step": 3528 }, { "epoch": 0.68, "learning_rate": 4.9335836046405575e-06, "loss": 0.4762, "step": 3529 }, { "epoch": 0.68, "learning_rate": 4.928211557138728e-06, "loss": 0.5965, "step": 3530 }, { "epoch": 0.68, "learning_rate": 4.922841479461083e-06, "loss": 0.6069, "step": 3531 }, { "epoch": 0.68, "learning_rate": 4.917473373693305e-06, "loss": 0.5758, "step": 3532 }, { "epoch": 0.68, "learning_rate": 4.9121072419203016e-06, "loss": 0.5748, "step": 3533 }, { "epoch": 0.68, "learning_rate": 4.906743086226218e-06, "loss": 0.5505, "step": 3534 }, { "epoch": 0.68, "learning_rate": 4.901380908694434e-06, "loss": 0.6007, "step": 3535 }, { "epoch": 0.68, "learning_rate": 4.8960207114075495e-06, "loss": 0.5767, "step": 3536 }, { "epoch": 0.68, "learning_rate": 4.890662496447407e-06, "loss": 0.3674, "step": 3537 }, { "epoch": 0.68, "learning_rate": 4.8853062658950765e-06, "loss": 0.5651, "step": 3538 }, { "epoch": 0.68, "learning_rate": 4.879952021830856e-06, "loss": 0.4915, "step": 3539 }, { "epoch": 0.68, "learning_rate": 4.874599766334276e-06, "loss": 0.5964, "step": 3540 }, { "epoch": 0.68, "learning_rate": 4.8692495014840825e-06, "loss": 0.5656, "step": 3541 }, { "epoch": 0.68, "learning_rate": 4.863901229358261e-06, "loss": 0.4729, "step": 3542 }, { "epoch": 0.68, "learning_rate": 4.858554952034019e-06, "loss": 0.5379, "step": 3543 }, { "epoch": 0.68, "learning_rate": 4.853210671587789e-06, "loss": 0.5839, "step": 3544 }, { "epoch": 0.68, "learning_rate": 4.847868390095227e-06, "loss": 0.6175, "step": 3545 }, { "epoch": 0.68, "learning_rate": 4.842528109631218e-06, "loss": 0.5332, "step": 3546 }, { "epoch": 0.68, "learning_rate": 4.837189832269858e-06, "loss": 0.5773, "step": 3547 }, { "epoch": 0.68, "learning_rate": 4.8318535600844775e-06, "loss": 0.5898, "step": 3548 }, { "epoch": 0.68, "learning_rate": 4.8265192951476206e-06, "loss": 0.528, "step": 3549 }, { "epoch": 0.68, "learning_rate": 4.8211870395310556e-06, "loss": 0.5746, "step": 3550 }, { "epoch": 0.68, "learning_rate": 4.815856795305772e-06, "loss": 0.6149, "step": 3551 }, { "epoch": 0.68, "learning_rate": 4.81052856454197e-06, "loss": 0.4719, "step": 3552 }, { "epoch": 0.68, "learning_rate": 4.805202349309074e-06, "loss": 0.6219, "step": 3553 }, { "epoch": 0.68, "learning_rate": 4.7998781516757295e-06, "loss": 0.6006, "step": 3554 }, { "epoch": 0.68, "learning_rate": 4.794555973709783e-06, "loss": 0.5984, "step": 3555 }, { "epoch": 0.68, "learning_rate": 4.789235817478322e-06, "loss": 0.5891, "step": 3556 }, { "epoch": 0.68, "learning_rate": 4.783917685047621e-06, "loss": 0.5757, "step": 3557 }, { "epoch": 0.68, "learning_rate": 4.778601578483187e-06, "loss": 0.4448, "step": 3558 }, { "epoch": 0.68, "learning_rate": 4.773287499849737e-06, "loss": 0.6205, "step": 3559 }, { "epoch": 0.68, "learning_rate": 4.767975451211191e-06, "loss": 0.555, "step": 3560 }, { "epoch": 0.69, "learning_rate": 4.762665434630692e-06, "loss": 0.4537, "step": 3561 }, { "epoch": 0.69, "learning_rate": 4.757357452170588e-06, "loss": 0.6018, "step": 3562 }, { "epoch": 0.69, "learning_rate": 4.752051505892438e-06, "loss": 0.5649, "step": 3563 }, { "epoch": 0.69, "learning_rate": 4.746747597857014e-06, "loss": 0.5858, "step": 3564 }, { "epoch": 0.69, "learning_rate": 4.741445730124287e-06, "loss": 0.5577, "step": 3565 }, { "epoch": 0.69, "learning_rate": 4.736145904753445e-06, "loss": 0.5953, "step": 3566 }, { "epoch": 0.69, "learning_rate": 4.730848123802877e-06, "loss": 0.4254, "step": 3567 }, { "epoch": 0.69, "learning_rate": 4.725552389330183e-06, "loss": 0.6045, "step": 3568 }, { "epoch": 0.69, "learning_rate": 4.720258703392161e-06, "loss": 0.4705, "step": 3569 }, { "epoch": 0.69, "learning_rate": 4.714967068044826e-06, "loss": 0.5709, "step": 3570 }, { "epoch": 0.69, "learning_rate": 4.7096774853433765e-06, "loss": 0.5482, "step": 3571 }, { "epoch": 0.69, "learning_rate": 4.704389957342237e-06, "loss": 0.4635, "step": 3572 }, { "epoch": 0.69, "learning_rate": 4.699104486095008e-06, "loss": 0.5745, "step": 3573 }, { "epoch": 0.69, "learning_rate": 4.69382107365452e-06, "loss": 0.5088, "step": 3574 }, { "epoch": 0.69, "learning_rate": 4.6885397220727855e-06, "loss": 0.5959, "step": 3575 }, { "epoch": 0.69, "learning_rate": 4.683260433401016e-06, "loss": 0.4742, "step": 3576 }, { "epoch": 0.69, "learning_rate": 4.677983209689631e-06, "loss": 0.5605, "step": 3577 }, { "epoch": 0.69, "learning_rate": 4.6727080529882394e-06, "loss": 0.6098, "step": 3578 }, { "epoch": 0.69, "learning_rate": 4.667434965345654e-06, "loss": 0.5735, "step": 3579 }, { "epoch": 0.69, "learning_rate": 4.6621639488098856e-06, "loss": 0.5898, "step": 3580 }, { "epoch": 0.69, "learning_rate": 4.656895005428127e-06, "loss": 0.5726, "step": 3581 }, { "epoch": 0.69, "learning_rate": 4.651628137246781e-06, "loss": 0.5943, "step": 3582 }, { "epoch": 0.69, "learning_rate": 4.6463633463114395e-06, "loss": 0.5799, "step": 3583 }, { "epoch": 0.69, "learning_rate": 4.641100634666877e-06, "loss": 0.6094, "step": 3584 }, { "epoch": 0.69, "learning_rate": 4.635840004357086e-06, "loss": 0.5677, "step": 3585 }, { "epoch": 0.69, "learning_rate": 4.630581457425222e-06, "loss": 0.6203, "step": 3586 }, { "epoch": 0.69, "learning_rate": 4.625324995913648e-06, "loss": 0.4643, "step": 3587 }, { "epoch": 0.69, "learning_rate": 4.620070621863917e-06, "loss": 0.5712, "step": 3588 }, { "epoch": 0.69, "learning_rate": 4.614818337316759e-06, "loss": 0.5036, "step": 3589 }, { "epoch": 0.69, "learning_rate": 4.609568144312107e-06, "loss": 0.6089, "step": 3590 }, { "epoch": 0.69, "learning_rate": 4.6043200448890724e-06, "loss": 0.4621, "step": 3591 }, { "epoch": 0.69, "learning_rate": 4.599074041085958e-06, "loss": 0.5798, "step": 3592 }, { "epoch": 0.69, "learning_rate": 4.593830134940256e-06, "loss": 0.499, "step": 3593 }, { "epoch": 0.69, "learning_rate": 4.588588328488629e-06, "loss": 0.6031, "step": 3594 }, { "epoch": 0.69, "learning_rate": 4.5833486237669414e-06, "loss": 0.5455, "step": 3595 }, { "epoch": 0.69, "learning_rate": 4.578111022810231e-06, "loss": 0.6, "step": 3596 }, { "epoch": 0.69, "learning_rate": 4.5728755276527225e-06, "loss": 0.5628, "step": 3597 }, { "epoch": 0.69, "learning_rate": 4.567642140327823e-06, "loss": 0.5214, "step": 3598 }, { "epoch": 0.69, "learning_rate": 4.562410862868123e-06, "loss": 0.5646, "step": 3599 }, { "epoch": 0.69, "learning_rate": 4.557181697305383e-06, "loss": 0.5284, "step": 3600 }, { "epoch": 0.69, "learning_rate": 4.551954645670557e-06, "loss": 0.5347, "step": 3601 }, { "epoch": 0.69, "learning_rate": 4.546729709993762e-06, "loss": 0.4653, "step": 3602 }, { "epoch": 0.69, "learning_rate": 4.541506892304314e-06, "loss": 0.5504, "step": 3603 }, { "epoch": 0.69, "learning_rate": 4.536286194630694e-06, "loss": 0.603, "step": 3604 }, { "epoch": 0.69, "learning_rate": 4.531067619000553e-06, "loss": 0.6013, "step": 3605 }, { "epoch": 0.69, "learning_rate": 4.525851167440731e-06, "loss": 0.5898, "step": 3606 }, { "epoch": 0.69, "learning_rate": 4.52063684197724e-06, "loss": 0.4854, "step": 3607 }, { "epoch": 0.69, "learning_rate": 4.515424644635254e-06, "loss": 0.5837, "step": 3608 }, { "epoch": 0.69, "learning_rate": 4.510214577439146e-06, "loss": 0.6258, "step": 3609 }, { "epoch": 0.69, "learning_rate": 4.5050066424124324e-06, "loss": 0.5893, "step": 3610 }, { "epoch": 0.69, "learning_rate": 4.49980084157782e-06, "loss": 0.5953, "step": 3611 }, { "epoch": 0.69, "learning_rate": 4.494597176957186e-06, "loss": 0.5957, "step": 3612 }, { "epoch": 0.7, "learning_rate": 4.489395650571562e-06, "loss": 0.5993, "step": 3613 }, { "epoch": 0.7, "learning_rate": 4.4841962644411765e-06, "loss": 0.4812, "step": 3614 }, { "epoch": 0.7, "learning_rate": 4.4789990205854e-06, "loss": 0.5803, "step": 3615 }, { "epoch": 0.7, "learning_rate": 4.473803921022784e-06, "loss": 0.4633, "step": 3616 }, { "epoch": 0.7, "learning_rate": 4.468610967771051e-06, "loss": 0.4841, "step": 3617 }, { "epoch": 0.7, "learning_rate": 4.4634201628470766e-06, "loss": 0.5983, "step": 3618 }, { "epoch": 0.7, "learning_rate": 4.458231508266912e-06, "loss": 0.5711, "step": 3619 }, { "epoch": 0.7, "learning_rate": 4.453045006045773e-06, "loss": 0.6068, "step": 3620 }, { "epoch": 0.7, "learning_rate": 4.447860658198035e-06, "loss": 0.5305, "step": 3621 }, { "epoch": 0.7, "learning_rate": 4.442678466737245e-06, "loss": 0.5621, "step": 3622 }, { "epoch": 0.7, "learning_rate": 4.4374984336760975e-06, "loss": 0.5078, "step": 3623 }, { "epoch": 0.7, "learning_rate": 4.432320561026461e-06, "loss": 0.6029, "step": 3624 }, { "epoch": 0.7, "learning_rate": 4.427144850799363e-06, "loss": 0.4803, "step": 3625 }, { "epoch": 0.7, "learning_rate": 4.421971305004989e-06, "loss": 0.5421, "step": 3626 }, { "epoch": 0.7, "learning_rate": 4.416799925652684e-06, "loss": 0.4972, "step": 3627 }, { "epoch": 0.7, "learning_rate": 4.411630714750956e-06, "loss": 0.562, "step": 3628 }, { "epoch": 0.7, "learning_rate": 4.4064636743074605e-06, "loss": 0.5649, "step": 3629 }, { "epoch": 0.7, "learning_rate": 4.40129880632902e-06, "loss": 0.4255, "step": 3630 }, { "epoch": 0.7, "learning_rate": 4.396136112821608e-06, "loss": 0.5191, "step": 3631 }, { "epoch": 0.7, "learning_rate": 4.390975595790358e-06, "loss": 0.5896, "step": 3632 }, { "epoch": 0.7, "learning_rate": 4.385817257239556e-06, "loss": 0.592, "step": 3633 }, { "epoch": 0.7, "learning_rate": 4.380661099172636e-06, "loss": 0.4488, "step": 3634 }, { "epoch": 0.7, "learning_rate": 4.375507123592194e-06, "loss": 0.3635, "step": 3635 }, { "epoch": 0.7, "learning_rate": 4.370355332499977e-06, "loss": 0.4678, "step": 3636 }, { "epoch": 0.7, "learning_rate": 4.365205727896872e-06, "loss": 0.5609, "step": 3637 }, { "epoch": 0.7, "learning_rate": 4.36005831178294e-06, "loss": 0.5719, "step": 3638 }, { "epoch": 0.7, "learning_rate": 4.354913086157367e-06, "loss": 0.5423, "step": 3639 }, { "epoch": 0.7, "learning_rate": 4.349770053018502e-06, "loss": 0.494, "step": 3640 }, { "epoch": 0.7, "learning_rate": 4.344629214363845e-06, "loss": 0.5594, "step": 3641 }, { "epoch": 0.7, "learning_rate": 4.339490572190031e-06, "loss": 0.6058, "step": 3642 }, { "epoch": 0.7, "learning_rate": 4.334354128492851e-06, "loss": 0.5307, "step": 3643 }, { "epoch": 0.7, "learning_rate": 4.329219885267244e-06, "loss": 0.4602, "step": 3644 }, { "epoch": 0.7, "learning_rate": 4.324087844507289e-06, "loss": 0.6015, "step": 3645 }, { "epoch": 0.7, "learning_rate": 4.318958008206214e-06, "loss": 0.3529, "step": 3646 }, { "epoch": 0.7, "learning_rate": 4.313830378356384e-06, "loss": 0.5249, "step": 3647 }, { "epoch": 0.7, "learning_rate": 4.3087049569493136e-06, "loss": 0.4486, "step": 3648 }, { "epoch": 0.7, "learning_rate": 4.303581745975656e-06, "loss": 0.5117, "step": 3649 }, { "epoch": 0.7, "learning_rate": 4.2984607474252084e-06, "loss": 0.4782, "step": 3650 }, { "epoch": 0.7, "learning_rate": 4.293341963286912e-06, "loss": 0.457, "step": 3651 }, { "epoch": 0.7, "learning_rate": 4.288225395548835e-06, "loss": 0.5542, "step": 3652 }, { "epoch": 0.7, "learning_rate": 4.283111046198198e-06, "loss": 0.6218, "step": 3653 }, { "epoch": 0.7, "learning_rate": 4.277998917221354e-06, "loss": 0.5621, "step": 3654 }, { "epoch": 0.7, "learning_rate": 4.272889010603798e-06, "loss": 0.5835, "step": 3655 }, { "epoch": 0.7, "learning_rate": 4.267781328330155e-06, "loss": 0.591, "step": 3656 }, { "epoch": 0.7, "learning_rate": 4.262675872384197e-06, "loss": 0.4926, "step": 3657 }, { "epoch": 0.7, "learning_rate": 4.257572644748813e-06, "loss": 0.59, "step": 3658 }, { "epoch": 0.7, "learning_rate": 4.252471647406045e-06, "loss": 0.606, "step": 3659 }, { "epoch": 0.7, "learning_rate": 4.2473728823370605e-06, "loss": 0.4574, "step": 3660 }, { "epoch": 0.7, "learning_rate": 4.242276351522161e-06, "loss": 0.5866, "step": 3661 }, { "epoch": 0.7, "learning_rate": 4.237182056940784e-06, "loss": 0.5531, "step": 3662 }, { "epoch": 0.7, "learning_rate": 4.232090000571488e-06, "loss": 0.591, "step": 3663 }, { "epoch": 0.7, "learning_rate": 4.2270001843919714e-06, "loss": 0.5853, "step": 3664 }, { "epoch": 0.71, "learning_rate": 4.221912610379065e-06, "loss": 0.571, "step": 3665 }, { "epoch": 0.71, "learning_rate": 4.216827280508712e-06, "loss": 0.6265, "step": 3666 }, { "epoch": 0.71, "learning_rate": 4.211744196756011e-06, "loss": 0.6075, "step": 3667 }, { "epoch": 0.71, "learning_rate": 4.206663361095164e-06, "loss": 0.55, "step": 3668 }, { "epoch": 0.71, "learning_rate": 4.201584775499509e-06, "loss": 0.4784, "step": 3669 }, { "epoch": 0.71, "learning_rate": 4.196508441941516e-06, "loss": 0.5703, "step": 3670 }, { "epoch": 0.71, "learning_rate": 4.191434362392768e-06, "loss": 0.473, "step": 3671 }, { "epoch": 0.71, "learning_rate": 4.186362538823981e-06, "loss": 0.5755, "step": 3672 }, { "epoch": 0.71, "learning_rate": 4.181292973204992e-06, "loss": 0.5909, "step": 3673 }, { "epoch": 0.71, "learning_rate": 4.1762256675047655e-06, "loss": 0.4869, "step": 3674 }, { "epoch": 0.71, "learning_rate": 4.171160623691384e-06, "loss": 0.5837, "step": 3675 }, { "epoch": 0.71, "learning_rate": 4.166097843732048e-06, "loss": 0.575, "step": 3676 }, { "epoch": 0.71, "learning_rate": 4.161037329593085e-06, "loss": 0.5628, "step": 3677 }, { "epoch": 0.71, "learning_rate": 4.155979083239942e-06, "loss": 0.5323, "step": 3678 }, { "epoch": 0.71, "learning_rate": 4.1509231066371815e-06, "loss": 0.5183, "step": 3679 }, { "epoch": 0.71, "learning_rate": 4.1458694017484915e-06, "loss": 0.548, "step": 3680 }, { "epoch": 0.71, "learning_rate": 4.140817970536664e-06, "loss": 0.452, "step": 3681 }, { "epoch": 0.71, "learning_rate": 4.135768814963622e-06, "loss": 0.5437, "step": 3682 }, { "epoch": 0.71, "learning_rate": 4.130721936990399e-06, "loss": 0.6273, "step": 3683 }, { "epoch": 0.71, "learning_rate": 4.1256773385771444e-06, "loss": 0.6151, "step": 3684 }, { "epoch": 0.71, "learning_rate": 4.120635021683122e-06, "loss": 0.5733, "step": 3685 }, { "epoch": 0.71, "learning_rate": 4.115594988266711e-06, "loss": 0.5801, "step": 3686 }, { "epoch": 0.71, "learning_rate": 4.1105572402853976e-06, "loss": 0.5317, "step": 3687 }, { "epoch": 0.71, "learning_rate": 4.1055217796957895e-06, "loss": 0.5717, "step": 3688 }, { "epoch": 0.71, "learning_rate": 4.100488608453599e-06, "loss": 0.5489, "step": 3689 }, { "epoch": 0.71, "learning_rate": 4.095457728513652e-06, "loss": 0.4662, "step": 3690 }, { "epoch": 0.71, "learning_rate": 4.09042914182989e-06, "loss": 0.5702, "step": 3691 }, { "epoch": 0.71, "learning_rate": 4.08540285035535e-06, "loss": 0.5784, "step": 3692 }, { "epoch": 0.71, "learning_rate": 4.0803788560421885e-06, "loss": 0.6227, "step": 3693 }, { "epoch": 0.71, "learning_rate": 4.075357160841671e-06, "loss": 0.5941, "step": 3694 }, { "epoch": 0.71, "learning_rate": 4.070337766704155e-06, "loss": 0.5904, "step": 3695 }, { "epoch": 0.71, "learning_rate": 4.065320675579132e-06, "loss": 0.5559, "step": 3696 }, { "epoch": 0.71, "learning_rate": 4.0603058894151685e-06, "loss": 0.5829, "step": 3697 }, { "epoch": 0.71, "learning_rate": 4.055293410159954e-06, "loss": 0.577, "step": 3698 }, { "epoch": 0.71, "learning_rate": 4.050283239760282e-06, "loss": 0.4638, "step": 3699 }, { "epoch": 0.71, "learning_rate": 4.045275380162038e-06, "loss": 0.548, "step": 3700 }, { "epoch": 0.71, "learning_rate": 4.04026983331022e-06, "loss": 0.5613, "step": 3701 }, { "epoch": 0.71, "learning_rate": 4.035266601148924e-06, "loss": 0.5939, "step": 3702 }, { "epoch": 0.71, "learning_rate": 4.03026568562135e-06, "loss": 0.5699, "step": 3703 }, { "epoch": 0.71, "learning_rate": 4.025267088669797e-06, "loss": 0.5683, "step": 3704 }, { "epoch": 0.71, "learning_rate": 4.020270812235656e-06, "loss": 0.5935, "step": 3705 }, { "epoch": 0.71, "learning_rate": 4.015276858259427e-06, "loss": 0.5266, "step": 3706 }, { "epoch": 0.71, "learning_rate": 4.010285228680705e-06, "loss": 0.5667, "step": 3707 }, { "epoch": 0.71, "learning_rate": 4.005295925438181e-06, "loss": 0.6058, "step": 3708 }, { "epoch": 0.71, "learning_rate": 4.000308950469646e-06, "loss": 0.4285, "step": 3709 }, { "epoch": 0.71, "learning_rate": 3.995324305711976e-06, "loss": 0.5889, "step": 3710 }, { "epoch": 0.71, "learning_rate": 3.990341993101154e-06, "loss": 0.5643, "step": 3711 }, { "epoch": 0.71, "learning_rate": 3.985362014572256e-06, "loss": 0.5722, "step": 3712 }, { "epoch": 0.71, "learning_rate": 3.9803843720594385e-06, "loss": 0.5871, "step": 3713 }, { "epoch": 0.71, "learning_rate": 3.97540906749597e-06, "loss": 0.5952, "step": 3714 }, { "epoch": 0.71, "learning_rate": 3.970436102814203e-06, "loss": 0.558, "step": 3715 }, { "epoch": 0.71, "learning_rate": 3.965465479945569e-06, "loss": 0.4323, "step": 3716 }, { "epoch": 0.72, "learning_rate": 3.9604972008206085e-06, "loss": 0.5892, "step": 3717 }, { "epoch": 0.72, "learning_rate": 3.955531267368942e-06, "loss": 0.604, "step": 3718 }, { "epoch": 0.72, "learning_rate": 3.950567681519279e-06, "loss": 0.5984, "step": 3719 }, { "epoch": 0.72, "learning_rate": 3.945606445199427e-06, "loss": 0.5634, "step": 3720 }, { "epoch": 0.72, "learning_rate": 3.940647560336262e-06, "loss": 0.6039, "step": 3721 }, { "epoch": 0.72, "learning_rate": 3.935691028855763e-06, "loss": 0.6139, "step": 3722 }, { "epoch": 0.72, "learning_rate": 3.930736852682993e-06, "loss": 0.6311, "step": 3723 }, { "epoch": 0.72, "learning_rate": 3.9257850337420856e-06, "loss": 0.4688, "step": 3724 }, { "epoch": 0.72, "learning_rate": 3.920835573956285e-06, "loss": 0.5504, "step": 3725 }, { "epoch": 0.72, "learning_rate": 3.915888475247894e-06, "loss": 0.625, "step": 3726 }, { "epoch": 0.72, "learning_rate": 3.910943739538313e-06, "loss": 0.4755, "step": 3727 }, { "epoch": 0.72, "learning_rate": 3.906001368748023e-06, "loss": 0.5376, "step": 3728 }, { "epoch": 0.72, "learning_rate": 3.901061364796574e-06, "loss": 0.5733, "step": 3729 }, { "epoch": 0.72, "learning_rate": 3.8961237296026155e-06, "loss": 0.5912, "step": 3730 }, { "epoch": 0.72, "learning_rate": 3.891188465083865e-06, "loss": 0.4848, "step": 3731 }, { "epoch": 0.72, "learning_rate": 3.886255573157121e-06, "loss": 0.536, "step": 3732 }, { "epoch": 0.72, "learning_rate": 3.88132505573827e-06, "loss": 0.4687, "step": 3733 }, { "epoch": 0.72, "learning_rate": 3.876396914742258e-06, "loss": 0.5911, "step": 3734 }, { "epoch": 0.72, "learning_rate": 3.871471152083121e-06, "loss": 0.4663, "step": 3735 }, { "epoch": 0.72, "learning_rate": 3.866547769673968e-06, "loss": 0.4918, "step": 3736 }, { "epoch": 0.72, "learning_rate": 3.861626769426988e-06, "loss": 0.5488, "step": 3737 }, { "epoch": 0.72, "learning_rate": 3.8567081532534374e-06, "loss": 0.5832, "step": 3738 }, { "epoch": 0.72, "learning_rate": 3.851791923063655e-06, "loss": 0.5713, "step": 3739 }, { "epoch": 0.72, "learning_rate": 3.846878080767039e-06, "loss": 0.5593, "step": 3740 }, { "epoch": 0.72, "learning_rate": 3.841966628272079e-06, "loss": 0.6114, "step": 3741 }, { "epoch": 0.72, "learning_rate": 3.837057567486314e-06, "loss": 0.5829, "step": 3742 }, { "epoch": 0.72, "learning_rate": 3.832150900316377e-06, "loss": 0.574, "step": 3743 }, { "epoch": 0.72, "learning_rate": 3.827246628667962e-06, "loss": 0.596, "step": 3744 }, { "epoch": 0.72, "learning_rate": 3.822344754445826e-06, "loss": 0.5783, "step": 3745 }, { "epoch": 0.72, "learning_rate": 3.817445279553801e-06, "loss": 0.5526, "step": 3746 }, { "epoch": 0.72, "learning_rate": 3.8125482058947905e-06, "loss": 0.6004, "step": 3747 }, { "epoch": 0.72, "learning_rate": 3.8076535353707523e-06, "loss": 0.5707, "step": 3748 }, { "epoch": 0.72, "learning_rate": 3.8027612698827344e-06, "loss": 0.5481, "step": 3749 }, { "epoch": 0.72, "learning_rate": 3.7978714113308246e-06, "loss": 0.6012, "step": 3750 }, { "epoch": 0.72, "learning_rate": 3.7929839616141917e-06, "loss": 0.562, "step": 3751 }, { "epoch": 0.72, "learning_rate": 3.788098922631067e-06, "loss": 0.6109, "step": 3752 }, { "epoch": 0.72, "learning_rate": 3.7832162962787355e-06, "loss": 0.5608, "step": 3753 }, { "epoch": 0.72, "learning_rate": 3.7783360844535653e-06, "loss": 0.615, "step": 3754 }, { "epoch": 0.72, "learning_rate": 3.773458289050963e-06, "loss": 0.5594, "step": 3755 }, { "epoch": 0.72, "learning_rate": 3.768582911965414e-06, "loss": 0.505, "step": 3756 }, { "epoch": 0.72, "learning_rate": 3.763709955090461e-06, "loss": 0.6, "step": 3757 }, { "epoch": 0.72, "learning_rate": 3.7588394203186963e-06, "loss": 0.5753, "step": 3758 }, { "epoch": 0.72, "learning_rate": 3.753971309541784e-06, "loss": 0.5148, "step": 3759 }, { "epoch": 0.72, "learning_rate": 3.7491056246504433e-06, "loss": 0.5733, "step": 3760 }, { "epoch": 0.72, "learning_rate": 3.7442423675344474e-06, "loss": 0.5702, "step": 3761 }, { "epoch": 0.72, "learning_rate": 3.739381540082635e-06, "loss": 0.5725, "step": 3762 }, { "epoch": 0.72, "learning_rate": 3.7345231441828876e-06, "loss": 0.5504, "step": 3763 }, { "epoch": 0.72, "learning_rate": 3.729667181722154e-06, "loss": 0.4808, "step": 3764 }, { "epoch": 0.72, "learning_rate": 3.7248136545864345e-06, "loss": 0.4509, "step": 3765 }, { "epoch": 0.72, "learning_rate": 3.719962564660783e-06, "loss": 0.4573, "step": 3766 }, { "epoch": 0.72, "learning_rate": 3.7151139138293056e-06, "loss": 0.5775, "step": 3767 }, { "epoch": 0.72, "learning_rate": 3.7102677039751667e-06, "loss": 0.5542, "step": 3768 }, { "epoch": 0.73, "learning_rate": 3.705423936980572e-06, "loss": 0.5881, "step": 3769 }, { "epoch": 0.73, "learning_rate": 3.700582614726791e-06, "loss": 0.4738, "step": 3770 }, { "epoch": 0.73, "learning_rate": 3.6957437390941274e-06, "loss": 0.5418, "step": 3771 }, { "epoch": 0.73, "learning_rate": 3.6909073119619555e-06, "loss": 0.5827, "step": 3772 }, { "epoch": 0.73, "learning_rate": 3.6860733352086866e-06, "loss": 0.592, "step": 3773 }, { "epoch": 0.73, "learning_rate": 3.6812418107117765e-06, "loss": 0.5679, "step": 3774 }, { "epoch": 0.73, "learning_rate": 3.6764127403477347e-06, "loss": 0.5902, "step": 3775 }, { "epoch": 0.73, "learning_rate": 3.6715861259921226e-06, "loss": 0.5969, "step": 3776 }, { "epoch": 0.73, "learning_rate": 3.6667619695195287e-06, "loss": 0.5762, "step": 3777 }, { "epoch": 0.73, "learning_rate": 3.6619402728036157e-06, "loss": 0.4975, "step": 3778 }, { "epoch": 0.73, "learning_rate": 3.657121037717064e-06, "loss": 0.4751, "step": 3779 }, { "epoch": 0.73, "learning_rate": 3.652304266131612e-06, "loss": 0.4558, "step": 3780 }, { "epoch": 0.73, "learning_rate": 3.6474899599180426e-06, "loss": 0.495, "step": 3781 }, { "epoch": 0.73, "learning_rate": 3.642678120946168e-06, "loss": 0.3608, "step": 3782 }, { "epoch": 0.73, "learning_rate": 3.6378687510848576e-06, "loss": 0.5689, "step": 3783 }, { "epoch": 0.73, "learning_rate": 3.6330618522020124e-06, "loss": 0.5686, "step": 3784 }, { "epoch": 0.73, "learning_rate": 3.6282574261645776e-06, "loss": 0.5837, "step": 3785 }, { "epoch": 0.73, "learning_rate": 3.62345547483854e-06, "loss": 0.5696, "step": 3786 }, { "epoch": 0.73, "learning_rate": 3.618656000088916e-06, "loss": 0.5443, "step": 3787 }, { "epoch": 0.73, "learning_rate": 3.6138590037797695e-06, "loss": 0.486, "step": 3788 }, { "epoch": 0.73, "learning_rate": 3.6090644877741986e-06, "loss": 0.5723, "step": 3789 }, { "epoch": 0.73, "learning_rate": 3.6042724539343378e-06, "loss": 0.4791, "step": 3790 }, { "epoch": 0.73, "learning_rate": 3.599482904121361e-06, "loss": 0.5433, "step": 3791 }, { "epoch": 0.73, "learning_rate": 3.594695840195468e-06, "loss": 0.5727, "step": 3792 }, { "epoch": 0.73, "learning_rate": 3.5899112640159017e-06, "loss": 0.4462, "step": 3793 }, { "epoch": 0.73, "learning_rate": 3.585129177440938e-06, "loss": 0.5833, "step": 3794 }, { "epoch": 0.73, "learning_rate": 3.580349582327882e-06, "loss": 0.5784, "step": 3795 }, { "epoch": 0.73, "learning_rate": 3.575572480533076e-06, "loss": 0.5744, "step": 3796 }, { "epoch": 0.73, "learning_rate": 3.570797873911892e-06, "loss": 0.565, "step": 3797 }, { "epoch": 0.73, "learning_rate": 3.566025764318728e-06, "loss": 0.5578, "step": 3798 }, { "epoch": 0.73, "learning_rate": 3.5612561536070213e-06, "loss": 0.5344, "step": 3799 }, { "epoch": 0.73, "learning_rate": 3.5564890436292243e-06, "loss": 0.5851, "step": 3800 }, { "epoch": 0.73, "learning_rate": 3.5517244362368363e-06, "loss": 0.5048, "step": 3801 }, { "epoch": 0.73, "learning_rate": 3.5469623332803795e-06, "loss": 0.5972, "step": 3802 }, { "epoch": 0.73, "learning_rate": 3.5422027366093893e-06, "loss": 0.5189, "step": 3803 }, { "epoch": 0.73, "learning_rate": 3.5374456480724427e-06, "loss": 0.447, "step": 3804 }, { "epoch": 0.73, "learning_rate": 3.532691069517142e-06, "loss": 0.5894, "step": 3805 }, { "epoch": 0.73, "learning_rate": 3.5279390027901004e-06, "loss": 0.5703, "step": 3806 }, { "epoch": 0.73, "learning_rate": 3.5231894497369802e-06, "loss": 0.4658, "step": 3807 }, { "epoch": 0.73, "learning_rate": 3.5184424122024406e-06, "loss": 0.5968, "step": 3808 }, { "epoch": 0.73, "learning_rate": 3.5136978920301822e-06, "loss": 0.5336, "step": 3809 }, { "epoch": 0.73, "learning_rate": 3.508955891062924e-06, "loss": 0.5833, "step": 3810 }, { "epoch": 0.73, "learning_rate": 3.5042164111423983e-06, "loss": 0.5364, "step": 3811 }, { "epoch": 0.73, "learning_rate": 3.4994794541093667e-06, "loss": 0.5467, "step": 3812 }, { "epoch": 0.73, "learning_rate": 3.4947450218036106e-06, "loss": 0.5706, "step": 3813 }, { "epoch": 0.73, "learning_rate": 3.4900131160639283e-06, "loss": 0.5563, "step": 3814 }, { "epoch": 0.73, "learning_rate": 3.485283738728139e-06, "loss": 0.5961, "step": 3815 }, { "epoch": 0.73, "learning_rate": 3.4805568916330747e-06, "loss": 0.4931, "step": 3816 }, { "epoch": 0.73, "learning_rate": 3.4758325766145896e-06, "loss": 0.5801, "step": 3817 }, { "epoch": 0.73, "learning_rate": 3.471110795507554e-06, "loss": 0.457, "step": 3818 }, { "epoch": 0.73, "learning_rate": 3.4663915501458523e-06, "loss": 0.5389, "step": 3819 }, { "epoch": 0.73, "learning_rate": 3.4616748423623893e-06, "loss": 0.5763, "step": 3820 }, { "epoch": 0.74, "learning_rate": 3.4569606739890737e-06, "loss": 0.5068, "step": 3821 }, { "epoch": 0.74, "learning_rate": 3.452249046856836e-06, "loss": 0.5905, "step": 3822 }, { "epoch": 0.74, "learning_rate": 3.4475399627956197e-06, "loss": 0.5742, "step": 3823 }, { "epoch": 0.74, "learning_rate": 3.4428334236343774e-06, "loss": 0.5795, "step": 3824 }, { "epoch": 0.74, "learning_rate": 3.438129431201075e-06, "loss": 0.546, "step": 3825 }, { "epoch": 0.74, "learning_rate": 3.433427987322693e-06, "loss": 0.5602, "step": 3826 }, { "epoch": 0.74, "learning_rate": 3.4287290938252103e-06, "loss": 0.603, "step": 3827 }, { "epoch": 0.74, "learning_rate": 3.424032752533627e-06, "loss": 0.5715, "step": 3828 }, { "epoch": 0.74, "learning_rate": 3.4193389652719478e-06, "loss": 0.6403, "step": 3829 }, { "epoch": 0.74, "learning_rate": 3.4146477338631856e-06, "loss": 0.4554, "step": 3830 }, { "epoch": 0.74, "learning_rate": 3.4099590601293632e-06, "loss": 0.4323, "step": 3831 }, { "epoch": 0.74, "learning_rate": 3.4052729458915024e-06, "loss": 0.5406, "step": 3832 }, { "epoch": 0.74, "learning_rate": 3.4005893929696377e-06, "loss": 0.5551, "step": 3833 }, { "epoch": 0.74, "learning_rate": 3.3959084031828114e-06, "loss": 0.4865, "step": 3834 }, { "epoch": 0.74, "learning_rate": 3.3912299783490567e-06, "loss": 0.5537, "step": 3835 }, { "epoch": 0.74, "learning_rate": 3.3865541202854314e-06, "loss": 0.4565, "step": 3836 }, { "epoch": 0.74, "learning_rate": 3.3818808308079753e-06, "loss": 0.5455, "step": 3837 }, { "epoch": 0.74, "learning_rate": 3.3772101117317437e-06, "loss": 0.481, "step": 3838 }, { "epoch": 0.74, "learning_rate": 3.372541964870795e-06, "loss": 0.5605, "step": 3839 }, { "epoch": 0.74, "learning_rate": 3.367876392038174e-06, "loss": 0.5495, "step": 3840 }, { "epoch": 0.74, "learning_rate": 3.363213395045941e-06, "loss": 0.5965, "step": 3841 }, { "epoch": 0.74, "learning_rate": 3.3585529757051504e-06, "loss": 0.5185, "step": 3842 }, { "epoch": 0.74, "learning_rate": 3.353895135825854e-06, "loss": 0.6373, "step": 3843 }, { "epoch": 0.74, "learning_rate": 3.3492398772171074e-06, "loss": 0.5199, "step": 3844 }, { "epoch": 0.74, "learning_rate": 3.344587201686952e-06, "loss": 0.4744, "step": 3845 }, { "epoch": 0.74, "learning_rate": 3.3399371110424372e-06, "loss": 0.5205, "step": 3846 }, { "epoch": 0.74, "learning_rate": 3.3352896070896057e-06, "loss": 0.5469, "step": 3847 }, { "epoch": 0.74, "learning_rate": 3.330644691633492e-06, "loss": 0.4517, "step": 3848 }, { "epoch": 0.74, "learning_rate": 3.3260023664781326e-06, "loss": 0.5857, "step": 3849 }, { "epoch": 0.74, "learning_rate": 3.321362633426547e-06, "loss": 0.576, "step": 3850 }, { "epoch": 0.74, "learning_rate": 3.316725494280757e-06, "loss": 0.5411, "step": 3851 }, { "epoch": 0.74, "learning_rate": 3.3120909508417754e-06, "loss": 0.4921, "step": 3852 }, { "epoch": 0.74, "learning_rate": 3.307459004909599e-06, "loss": 0.6146, "step": 3853 }, { "epoch": 0.74, "learning_rate": 3.3028296582832285e-06, "loss": 0.4571, "step": 3854 }, { "epoch": 0.74, "learning_rate": 3.2982029127606517e-06, "loss": 0.5703, "step": 3855 }, { "epoch": 0.74, "learning_rate": 3.2935787701388346e-06, "loss": 0.5428, "step": 3856 }, { "epoch": 0.74, "learning_rate": 3.2889572322137454e-06, "loss": 0.5737, "step": 3857 }, { "epoch": 0.74, "learning_rate": 3.2843383007803364e-06, "loss": 0.5612, "step": 3858 }, { "epoch": 0.74, "learning_rate": 3.279721977632546e-06, "loss": 0.4708, "step": 3859 }, { "epoch": 0.74, "learning_rate": 3.275108264563306e-06, "loss": 0.5358, "step": 3860 }, { "epoch": 0.74, "learning_rate": 3.270497163364521e-06, "loss": 0.5336, "step": 3861 }, { "epoch": 0.74, "learning_rate": 3.2658886758270947e-06, "loss": 0.5768, "step": 3862 }, { "epoch": 0.74, "learning_rate": 3.2612828037409116e-06, "loss": 0.6053, "step": 3863 }, { "epoch": 0.74, "learning_rate": 3.256679548894831e-06, "loss": 0.6073, "step": 3864 }, { "epoch": 0.74, "learning_rate": 3.252078913076718e-06, "loss": 0.5405, "step": 3865 }, { "epoch": 0.74, "learning_rate": 3.247480898073395e-06, "loss": 0.45, "step": 3866 }, { "epoch": 0.74, "learning_rate": 3.242885505670681e-06, "loss": 0.6043, "step": 3867 }, { "epoch": 0.74, "learning_rate": 3.238292737653379e-06, "loss": 0.5798, "step": 3868 }, { "epoch": 0.74, "learning_rate": 3.233702595805258e-06, "loss": 0.5457, "step": 3869 }, { "epoch": 0.74, "learning_rate": 3.229115081909082e-06, "loss": 0.5872, "step": 3870 }, { "epoch": 0.74, "learning_rate": 3.224530197746587e-06, "loss": 0.3536, "step": 3871 }, { "epoch": 0.74, "learning_rate": 3.2199479450984892e-06, "loss": 0.4809, "step": 3872 }, { "epoch": 0.75, "learning_rate": 3.2153683257444856e-06, "loss": 0.576, "step": 3873 }, { "epoch": 0.75, "learning_rate": 3.210791341463243e-06, "loss": 0.5581, "step": 3874 }, { "epoch": 0.75, "learning_rate": 3.206216994032411e-06, "loss": 0.6095, "step": 3875 }, { "epoch": 0.75, "learning_rate": 3.2016452852286127e-06, "loss": 0.6029, "step": 3876 }, { "epoch": 0.75, "learning_rate": 3.1970762168274495e-06, "loss": 0.5953, "step": 3877 }, { "epoch": 0.75, "learning_rate": 3.1925097906034962e-06, "loss": 0.5639, "step": 3878 }, { "epoch": 0.75, "learning_rate": 3.187946008330295e-06, "loss": 0.5495, "step": 3879 }, { "epoch": 0.75, "learning_rate": 3.1833848717803674e-06, "loss": 0.5241, "step": 3880 }, { "epoch": 0.75, "learning_rate": 3.178826382725212e-06, "loss": 0.58, "step": 3881 }, { "epoch": 0.75, "learning_rate": 3.1742705429352827e-06, "loss": 0.4837, "step": 3882 }, { "epoch": 0.75, "learning_rate": 3.1697173541800254e-06, "loss": 0.4558, "step": 3883 }, { "epoch": 0.75, "learning_rate": 3.165166818227845e-06, "loss": 0.4366, "step": 3884 }, { "epoch": 0.75, "learning_rate": 3.1606189368461117e-06, "loss": 0.4742, "step": 3885 }, { "epoch": 0.75, "learning_rate": 3.156073711801172e-06, "loss": 0.5065, "step": 3886 }, { "epoch": 0.75, "learning_rate": 3.151531144858344e-06, "loss": 0.6105, "step": 3887 }, { "epoch": 0.75, "learning_rate": 3.146991237781899e-06, "loss": 0.3633, "step": 3888 }, { "epoch": 0.75, "learning_rate": 3.142453992335096e-06, "loss": 0.5819, "step": 3889 }, { "epoch": 0.75, "learning_rate": 3.137919410280139e-06, "loss": 0.5775, "step": 3890 }, { "epoch": 0.75, "learning_rate": 3.1333874933782114e-06, "loss": 0.5552, "step": 3891 }, { "epoch": 0.75, "learning_rate": 3.128858243389461e-06, "loss": 0.5673, "step": 3892 }, { "epoch": 0.75, "learning_rate": 3.124331662072987e-06, "loss": 0.4589, "step": 3893 }, { "epoch": 0.75, "learning_rate": 3.119807751186872e-06, "loss": 0.5517, "step": 3894 }, { "epoch": 0.75, "learning_rate": 3.1152865124881436e-06, "loss": 0.4535, "step": 3895 }, { "epoch": 0.75, "learning_rate": 3.110767947732801e-06, "loss": 0.5858, "step": 3896 }, { "epoch": 0.75, "learning_rate": 3.106252058675806e-06, "loss": 0.4713, "step": 3897 }, { "epoch": 0.75, "learning_rate": 3.101738847071072e-06, "loss": 0.4946, "step": 3898 }, { "epoch": 0.75, "learning_rate": 3.097228314671481e-06, "loss": 0.5809, "step": 3899 }, { "epoch": 0.75, "learning_rate": 3.092720463228872e-06, "loss": 0.5041, "step": 3900 }, { "epoch": 0.75, "learning_rate": 3.0882152944940423e-06, "loss": 0.5589, "step": 3901 }, { "epoch": 0.75, "learning_rate": 3.0837128102167514e-06, "loss": 0.5629, "step": 3902 }, { "epoch": 0.75, "learning_rate": 3.079213012145705e-06, "loss": 0.6092, "step": 3903 }, { "epoch": 0.75, "learning_rate": 3.0747159020285766e-06, "loss": 0.5716, "step": 3904 }, { "epoch": 0.75, "learning_rate": 3.0702214816119925e-06, "loss": 0.5659, "step": 3905 }, { "epoch": 0.75, "learning_rate": 3.065729752641532e-06, "loss": 0.598, "step": 3906 }, { "epoch": 0.75, "learning_rate": 3.0612407168617352e-06, "loss": 0.4524, "step": 3907 }, { "epoch": 0.75, "learning_rate": 3.0567543760160866e-06, "loss": 0.5545, "step": 3908 }, { "epoch": 0.75, "learning_rate": 3.05227073184703e-06, "loss": 0.6232, "step": 3909 }, { "epoch": 0.75, "learning_rate": 3.047789786095967e-06, "loss": 0.5427, "step": 3910 }, { "epoch": 0.75, "learning_rate": 3.0433115405032334e-06, "loss": 0.5824, "step": 3911 }, { "epoch": 0.75, "learning_rate": 3.0388359968081395e-06, "loss": 0.6015, "step": 3912 }, { "epoch": 0.75, "learning_rate": 3.034363156748933e-06, "loss": 0.4707, "step": 3913 }, { "epoch": 0.75, "learning_rate": 3.0298930220628086e-06, "loss": 0.6004, "step": 3914 }, { "epoch": 0.75, "learning_rate": 3.025425594485919e-06, "loss": 0.5447, "step": 3915 }, { "epoch": 0.75, "learning_rate": 3.0209608757533626e-06, "loss": 0.5358, "step": 3916 }, { "epoch": 0.75, "learning_rate": 3.0164988675991768e-06, "loss": 0.5923, "step": 3917 }, { "epoch": 0.75, "learning_rate": 3.0120395717563655e-06, "loss": 0.5776, "step": 3918 }, { "epoch": 0.75, "learning_rate": 3.00758298995686e-06, "loss": 0.526, "step": 3919 }, { "epoch": 0.75, "learning_rate": 3.0031291239315473e-06, "loss": 0.5831, "step": 3920 }, { "epoch": 0.75, "learning_rate": 2.9986779754102613e-06, "loss": 0.4396, "step": 3921 }, { "epoch": 0.75, "learning_rate": 2.9942295461217698e-06, "loss": 0.6225, "step": 3922 }, { "epoch": 0.75, "learning_rate": 2.9897838377937947e-06, "loss": 0.5614, "step": 3923 }, { "epoch": 0.75, "learning_rate": 2.985340852152999e-06, "loss": 0.6001, "step": 3924 }, { "epoch": 0.76, "learning_rate": 2.9809005909249866e-06, "loss": 0.5934, "step": 3925 }, { "epoch": 0.76, "learning_rate": 2.9764630558343064e-06, "loss": 0.5663, "step": 3926 }, { "epoch": 0.76, "learning_rate": 2.9720282486044407e-06, "loss": 0.5371, "step": 3927 }, { "epoch": 0.76, "learning_rate": 2.9675961709578194e-06, "loss": 0.3955, "step": 3928 }, { "epoch": 0.76, "learning_rate": 2.9631668246158105e-06, "loss": 0.5721, "step": 3929 }, { "epoch": 0.76, "learning_rate": 2.958740211298722e-06, "loss": 0.5735, "step": 3930 }, { "epoch": 0.76, "learning_rate": 2.9543163327258016e-06, "loss": 0.6097, "step": 3931 }, { "epoch": 0.76, "learning_rate": 2.949895190615227e-06, "loss": 0.5777, "step": 3932 }, { "epoch": 0.76, "learning_rate": 2.9454767866841225e-06, "loss": 0.5535, "step": 3933 }, { "epoch": 0.76, "learning_rate": 2.941061122648545e-06, "loss": 0.4839, "step": 3934 }, { "epoch": 0.76, "learning_rate": 2.9366482002234874e-06, "loss": 0.5694, "step": 3935 }, { "epoch": 0.76, "learning_rate": 2.932238021122877e-06, "loss": 0.5602, "step": 3936 }, { "epoch": 0.76, "learning_rate": 2.9278305870595814e-06, "loss": 0.4554, "step": 3937 }, { "epoch": 0.76, "learning_rate": 2.92342589974539e-06, "loss": 0.5528, "step": 3938 }, { "epoch": 0.76, "learning_rate": 2.919023960891039e-06, "loss": 0.5089, "step": 3939 }, { "epoch": 0.76, "learning_rate": 2.9146247722061806e-06, "loss": 0.3505, "step": 3940 }, { "epoch": 0.76, "learning_rate": 2.910228335399419e-06, "loss": 0.5988, "step": 3941 }, { "epoch": 0.76, "learning_rate": 2.90583465217828e-06, "loss": 0.5777, "step": 3942 }, { "epoch": 0.76, "learning_rate": 2.9014437242492133e-06, "loss": 0.447, "step": 3943 }, { "epoch": 0.76, "learning_rate": 2.897055553317607e-06, "loss": 0.3599, "step": 3944 }, { "epoch": 0.76, "learning_rate": 2.89267014108778e-06, "loss": 0.5665, "step": 3945 }, { "epoch": 0.76, "learning_rate": 2.8882874892629654e-06, "loss": 0.4608, "step": 3946 }, { "epoch": 0.76, "learning_rate": 2.883907599545348e-06, "loss": 0.5023, "step": 3947 }, { "epoch": 0.76, "learning_rate": 2.8795304736360184e-06, "loss": 0.6077, "step": 3948 }, { "epoch": 0.76, "learning_rate": 2.8751561132350025e-06, "loss": 0.5593, "step": 3949 }, { "epoch": 0.76, "learning_rate": 2.8707845200412567e-06, "loss": 0.5805, "step": 3950 }, { "epoch": 0.76, "learning_rate": 2.866415695752649e-06, "loss": 0.5544, "step": 3951 }, { "epoch": 0.76, "learning_rate": 2.862049642065986e-06, "loss": 0.5679, "step": 3952 }, { "epoch": 0.76, "learning_rate": 2.857686360676991e-06, "loss": 0.5799, "step": 3953 }, { "epoch": 0.76, "learning_rate": 2.853325853280312e-06, "loss": 0.5978, "step": 3954 }, { "epoch": 0.76, "learning_rate": 2.8489681215695242e-06, "loss": 0.5199, "step": 3955 }, { "epoch": 0.76, "learning_rate": 2.8446131672371136e-06, "loss": 0.5802, "step": 3956 }, { "epoch": 0.76, "learning_rate": 2.840260991974497e-06, "loss": 0.4199, "step": 3957 }, { "epoch": 0.76, "learning_rate": 2.83591159747201e-06, "loss": 0.5467, "step": 3958 }, { "epoch": 0.76, "learning_rate": 2.8315649854189066e-06, "loss": 0.6084, "step": 3959 }, { "epoch": 0.76, "learning_rate": 2.8272211575033635e-06, "loss": 0.4501, "step": 3960 }, { "epoch": 0.76, "learning_rate": 2.8228801154124687e-06, "loss": 0.4066, "step": 3961 }, { "epoch": 0.76, "learning_rate": 2.8185418608322344e-06, "loss": 0.4899, "step": 3962 }, { "epoch": 0.76, "learning_rate": 2.814206395447593e-06, "loss": 0.3488, "step": 3963 }, { "epoch": 0.76, "learning_rate": 2.80987372094238e-06, "loss": 0.5711, "step": 3964 }, { "epoch": 0.76, "learning_rate": 2.805543838999364e-06, "loss": 0.5531, "step": 3965 }, { "epoch": 0.76, "learning_rate": 2.801216751300223e-06, "loss": 0.5458, "step": 3966 }, { "epoch": 0.76, "learning_rate": 2.796892459525541e-06, "loss": 0.5847, "step": 3967 }, { "epoch": 0.76, "learning_rate": 2.7925709653548295e-06, "loss": 0.4672, "step": 3968 }, { "epoch": 0.76, "learning_rate": 2.788252270466497e-06, "loss": 0.5918, "step": 3969 }, { "epoch": 0.76, "learning_rate": 2.783936376537886e-06, "loss": 0.3687, "step": 3970 }, { "epoch": 0.76, "learning_rate": 2.7796232852452378e-06, "loss": 0.5596, "step": 3971 }, { "epoch": 0.76, "learning_rate": 2.775312998263703e-06, "loss": 0.5878, "step": 3972 }, { "epoch": 0.76, "learning_rate": 2.771005517267349e-06, "loss": 0.5803, "step": 3973 }, { "epoch": 0.76, "learning_rate": 2.7667008439291552e-06, "loss": 0.5772, "step": 3974 }, { "epoch": 0.76, "learning_rate": 2.762398979920998e-06, "loss": 0.5708, "step": 3975 }, { "epoch": 0.76, "learning_rate": 2.7580999269136854e-06, "loss": 0.5853, "step": 3976 }, { "epoch": 0.77, "learning_rate": 2.7538036865769093e-06, "loss": 0.442, "step": 3977 }, { "epoch": 0.77, "learning_rate": 2.749510260579282e-06, "loss": 0.5654, "step": 3978 }, { "epoch": 0.77, "learning_rate": 2.7452196505883265e-06, "loss": 0.5636, "step": 3979 }, { "epoch": 0.77, "learning_rate": 2.7409318582704594e-06, "loss": 0.3963, "step": 3980 }, { "epoch": 0.77, "learning_rate": 2.736646885291011e-06, "loss": 0.5733, "step": 3981 }, { "epoch": 0.77, "learning_rate": 2.7323647333142176e-06, "loss": 0.5863, "step": 3982 }, { "epoch": 0.77, "learning_rate": 2.728085404003217e-06, "loss": 0.4851, "step": 3983 }, { "epoch": 0.77, "learning_rate": 2.7238088990200538e-06, "loss": 0.5574, "step": 3984 }, { "epoch": 0.77, "learning_rate": 2.7195352200256675e-06, "loss": 0.3829, "step": 3985 }, { "epoch": 0.77, "learning_rate": 2.7152643686799095e-06, "loss": 0.5376, "step": 3986 }, { "epoch": 0.77, "learning_rate": 2.710996346641528e-06, "loss": 0.5444, "step": 3987 }, { "epoch": 0.77, "learning_rate": 2.706731155568175e-06, "loss": 0.5945, "step": 3988 }, { "epoch": 0.77, "learning_rate": 2.7024687971164032e-06, "loss": 0.5506, "step": 3989 }, { "epoch": 0.77, "learning_rate": 2.698209272941659e-06, "loss": 0.5647, "step": 3990 }, { "epoch": 0.77, "learning_rate": 2.693952584698294e-06, "loss": 0.5602, "step": 3991 }, { "epoch": 0.77, "learning_rate": 2.689698734039561e-06, "loss": 0.4001, "step": 3992 }, { "epoch": 0.77, "learning_rate": 2.685447722617597e-06, "loss": 0.3836, "step": 3993 }, { "epoch": 0.77, "learning_rate": 2.6811995520834543e-06, "loss": 0.547, "step": 3994 }, { "epoch": 0.77, "learning_rate": 2.676954224087075e-06, "loss": 0.5447, "step": 3995 }, { "epoch": 0.77, "learning_rate": 2.6727117402772886e-06, "loss": 0.5879, "step": 3996 }, { "epoch": 0.77, "learning_rate": 2.668472102301829e-06, "loss": 0.5882, "step": 3997 }, { "epoch": 0.77, "learning_rate": 2.664235311807327e-06, "loss": 0.4654, "step": 3998 }, { "epoch": 0.77, "learning_rate": 2.6600013704392946e-06, "loss": 0.4034, "step": 3999 }, { "epoch": 0.77, "learning_rate": 2.655770279842157e-06, "loss": 0.5681, "step": 4000 }, { "epoch": 0.77, "learning_rate": 2.651542041659211e-06, "loss": 0.4698, "step": 4001 }, { "epoch": 0.77, "learning_rate": 2.6473166575326603e-06, "loss": 0.5373, "step": 4002 }, { "epoch": 0.77, "learning_rate": 2.6430941291035984e-06, "loss": 0.5766, "step": 4003 }, { "epoch": 0.77, "learning_rate": 2.6388744580119975e-06, "loss": 0.5596, "step": 4004 }, { "epoch": 0.77, "learning_rate": 2.6346576458967397e-06, "loss": 0.5492, "step": 4005 }, { "epoch": 0.77, "learning_rate": 2.630443694395579e-06, "loss": 0.5973, "step": 4006 }, { "epoch": 0.77, "learning_rate": 2.6262326051451683e-06, "loss": 0.5739, "step": 4007 }, { "epoch": 0.77, "learning_rate": 2.6220243797810483e-06, "loss": 0.5701, "step": 4008 }, { "epoch": 0.77, "learning_rate": 2.6178190199376394e-06, "loss": 0.4258, "step": 4009 }, { "epoch": 0.77, "learning_rate": 2.6136165272482596e-06, "loss": 0.4754, "step": 4010 }, { "epoch": 0.77, "learning_rate": 2.6094169033451066e-06, "loss": 0.5915, "step": 4011 }, { "epoch": 0.77, "learning_rate": 2.605220149859267e-06, "loss": 0.5751, "step": 4012 }, { "epoch": 0.77, "learning_rate": 2.6010262684207134e-06, "loss": 0.52, "step": 4013 }, { "epoch": 0.77, "learning_rate": 2.596835260658297e-06, "loss": 0.4826, "step": 4014 }, { "epoch": 0.77, "learning_rate": 2.5926471281997577e-06, "loss": 0.5813, "step": 4015 }, { "epoch": 0.77, "learning_rate": 2.588461872671719e-06, "loss": 0.5047, "step": 4016 }, { "epoch": 0.77, "learning_rate": 2.5842794956996863e-06, "loss": 0.5469, "step": 4017 }, { "epoch": 0.77, "learning_rate": 2.580099998908049e-06, "loss": 0.5691, "step": 4018 }, { "epoch": 0.77, "learning_rate": 2.575923383920069e-06, "loss": 0.5045, "step": 4019 }, { "epoch": 0.77, "learning_rate": 2.5717496523578998e-06, "loss": 0.5857, "step": 4020 }, { "epoch": 0.77, "learning_rate": 2.5675788058425723e-06, "loss": 0.5544, "step": 4021 }, { "epoch": 0.77, "learning_rate": 2.563410845993988e-06, "loss": 0.4816, "step": 4022 }, { "epoch": 0.77, "learning_rate": 2.5592457744309405e-06, "loss": 0.5518, "step": 4023 }, { "epoch": 0.77, "learning_rate": 2.5550835927710982e-06, "loss": 0.5705, "step": 4024 }, { "epoch": 0.77, "learning_rate": 2.5509243026309983e-06, "loss": 0.6034, "step": 4025 }, { "epoch": 0.77, "learning_rate": 2.546767905626063e-06, "loss": 0.568, "step": 4026 }, { "epoch": 0.77, "learning_rate": 2.5426144033705937e-06, "loss": 0.481, "step": 4027 }, { "epoch": 0.77, "learning_rate": 2.5384637974777513e-06, "loss": 0.5849, "step": 4028 }, { "epoch": 0.78, "learning_rate": 2.5343160895595977e-06, "loss": 0.5802, "step": 4029 }, { "epoch": 0.78, "learning_rate": 2.530171281227044e-06, "loss": 0.5827, "step": 4030 }, { "epoch": 0.78, "learning_rate": 2.52602937408989e-06, "loss": 0.5615, "step": 4031 }, { "epoch": 0.78, "learning_rate": 2.5218903697568075e-06, "loss": 0.5583, "step": 4032 }, { "epoch": 0.78, "learning_rate": 2.517754269835332e-06, "loss": 0.4448, "step": 4033 }, { "epoch": 0.78, "learning_rate": 2.5136210759318814e-06, "loss": 0.5485, "step": 4034 }, { "epoch": 0.78, "learning_rate": 2.5094907896517383e-06, "loss": 0.4553, "step": 4035 }, { "epoch": 0.78, "learning_rate": 2.505363412599059e-06, "loss": 0.3517, "step": 4036 }, { "epoch": 0.78, "learning_rate": 2.5012389463768737e-06, "loss": 0.6001, "step": 4037 }, { "epoch": 0.78, "learning_rate": 2.4971173925870694e-06, "loss": 0.5512, "step": 4038 }, { "epoch": 0.78, "learning_rate": 2.4929987528304144e-06, "loss": 0.567, "step": 4039 }, { "epoch": 0.78, "learning_rate": 2.4888830287065414e-06, "loss": 0.506, "step": 4040 }, { "epoch": 0.78, "learning_rate": 2.4847702218139493e-06, "loss": 0.5507, "step": 4041 }, { "epoch": 0.78, "learning_rate": 2.480660333750007e-06, "loss": 0.4663, "step": 4042 }, { "epoch": 0.78, "learning_rate": 2.476553366110944e-06, "loss": 0.5399, "step": 4043 }, { "epoch": 0.78, "learning_rate": 2.4724493204918598e-06, "loss": 0.4629, "step": 4044 }, { "epoch": 0.78, "learning_rate": 2.4683481984867207e-06, "loss": 0.5855, "step": 4045 }, { "epoch": 0.78, "learning_rate": 2.4642500016883532e-06, "loss": 0.5686, "step": 4046 }, { "epoch": 0.78, "learning_rate": 2.4601547316884544e-06, "loss": 0.5651, "step": 4047 }, { "epoch": 0.78, "learning_rate": 2.4560623900775728e-06, "loss": 0.611, "step": 4048 }, { "epoch": 0.78, "learning_rate": 2.45197297844513e-06, "loss": 0.5862, "step": 4049 }, { "epoch": 0.78, "learning_rate": 2.447886498379409e-06, "loss": 0.4897, "step": 4050 }, { "epoch": 0.78, "learning_rate": 2.4438029514675444e-06, "loss": 0.5445, "step": 4051 }, { "epoch": 0.78, "learning_rate": 2.439722339295545e-06, "loss": 0.5643, "step": 4052 }, { "epoch": 0.78, "learning_rate": 2.4356446634482756e-06, "loss": 0.5805, "step": 4053 }, { "epoch": 0.78, "learning_rate": 2.4315699255094516e-06, "loss": 0.5362, "step": 4054 }, { "epoch": 0.78, "learning_rate": 2.427498127061658e-06, "loss": 0.467, "step": 4055 }, { "epoch": 0.78, "learning_rate": 2.423429269686336e-06, "loss": 0.5529, "step": 4056 }, { "epoch": 0.78, "learning_rate": 2.4193633549637765e-06, "loss": 0.5844, "step": 4057 }, { "epoch": 0.78, "learning_rate": 2.4153003844731425e-06, "loss": 0.4737, "step": 4058 }, { "epoch": 0.78, "learning_rate": 2.411240359792438e-06, "loss": 0.5686, "step": 4059 }, { "epoch": 0.78, "learning_rate": 2.407183282498534e-06, "loss": 0.5194, "step": 4060 }, { "epoch": 0.78, "learning_rate": 2.403129154167153e-06, "loss": 0.5781, "step": 4061 }, { "epoch": 0.78, "learning_rate": 2.3990779763728666e-06, "loss": 0.5541, "step": 4062 }, { "epoch": 0.78, "learning_rate": 2.3950297506891084e-06, "loss": 0.5383, "step": 4063 }, { "epoch": 0.78, "learning_rate": 2.390984478688164e-06, "loss": 0.5502, "step": 4064 }, { "epoch": 0.78, "learning_rate": 2.386942161941169e-06, "loss": 0.5984, "step": 4065 }, { "epoch": 0.78, "learning_rate": 2.3829028020181154e-06, "loss": 0.5792, "step": 4066 }, { "epoch": 0.78, "learning_rate": 2.3788664004878405e-06, "loss": 0.4874, "step": 4067 }, { "epoch": 0.78, "learning_rate": 2.374832958918035e-06, "loss": 0.468, "step": 4068 }, { "epoch": 0.78, "learning_rate": 2.3708024788752448e-06, "loss": 0.5715, "step": 4069 }, { "epoch": 0.78, "learning_rate": 2.3667749619248614e-06, "loss": 0.5552, "step": 4070 }, { "epoch": 0.78, "learning_rate": 2.3627504096311273e-06, "loss": 0.5799, "step": 4071 }, { "epoch": 0.78, "learning_rate": 2.3587288235571258e-06, "loss": 0.4592, "step": 4072 }, { "epoch": 0.78, "learning_rate": 2.354710205264801e-06, "loss": 0.5649, "step": 4073 }, { "epoch": 0.78, "learning_rate": 2.350694556314934e-06, "loss": 0.5775, "step": 4074 }, { "epoch": 0.78, "learning_rate": 2.3466818782671597e-06, "loss": 0.564, "step": 4075 }, { "epoch": 0.78, "learning_rate": 2.3426721726799573e-06, "loss": 0.5689, "step": 4076 }, { "epoch": 0.78, "learning_rate": 2.3386654411106446e-06, "loss": 0.6064, "step": 4077 }, { "epoch": 0.78, "learning_rate": 2.3346616851153935e-06, "loss": 0.591, "step": 4078 }, { "epoch": 0.78, "learning_rate": 2.330660906249218e-06, "loss": 0.581, "step": 4079 }, { "epoch": 0.78, "learning_rate": 2.3266631060659685e-06, "loss": 0.5421, "step": 4080 }, { "epoch": 0.79, "learning_rate": 2.32266828611835e-06, "loss": 0.4798, "step": 4081 }, { "epoch": 0.79, "learning_rate": 2.318676447957907e-06, "loss": 0.5534, "step": 4082 }, { "epoch": 0.79, "learning_rate": 2.3146875931350165e-06, "loss": 0.5874, "step": 4083 }, { "epoch": 0.79, "learning_rate": 2.310701723198908e-06, "loss": 0.5801, "step": 4084 }, { "epoch": 0.79, "learning_rate": 2.3067188396976482e-06, "loss": 0.5938, "step": 4085 }, { "epoch": 0.79, "learning_rate": 2.3027389441781368e-06, "loss": 0.5962, "step": 4086 }, { "epoch": 0.79, "learning_rate": 2.2987620381861288e-06, "loss": 0.5685, "step": 4087 }, { "epoch": 0.79, "learning_rate": 2.2947881232662007e-06, "loss": 0.6027, "step": 4088 }, { "epoch": 0.79, "learning_rate": 2.290817200961779e-06, "loss": 0.4542, "step": 4089 }, { "epoch": 0.79, "learning_rate": 2.286849272815126e-06, "loss": 0.5473, "step": 4090 }, { "epoch": 0.79, "learning_rate": 2.2828843403673338e-06, "loss": 0.5465, "step": 4091 }, { "epoch": 0.79, "learning_rate": 2.2789224051583403e-06, "loss": 0.5731, "step": 4092 }, { "epoch": 0.79, "learning_rate": 2.274963468726914e-06, "loss": 0.5675, "step": 4093 }, { "epoch": 0.79, "learning_rate": 2.2710075326106618e-06, "loss": 0.4964, "step": 4094 }, { "epoch": 0.79, "learning_rate": 2.2670545983460245e-06, "loss": 0.5811, "step": 4095 }, { "epoch": 0.79, "learning_rate": 2.263104667468272e-06, "loss": 0.5882, "step": 4096 }, { "epoch": 0.79, "learning_rate": 2.259157741511515e-06, "loss": 0.5858, "step": 4097 }, { "epoch": 0.79, "learning_rate": 2.2552138220086927e-06, "loss": 0.3531, "step": 4098 }, { "epoch": 0.79, "learning_rate": 2.2512729104915787e-06, "loss": 0.5363, "step": 4099 }, { "epoch": 0.79, "learning_rate": 2.2473350084907806e-06, "loss": 0.5567, "step": 4100 }, { "epoch": 0.79, "learning_rate": 2.243400117535729e-06, "loss": 0.5226, "step": 4101 }, { "epoch": 0.79, "learning_rate": 2.2394682391546928e-06, "loss": 0.5299, "step": 4102 }, { "epoch": 0.79, "learning_rate": 2.2355393748747702e-06, "loss": 0.4796, "step": 4103 }, { "epoch": 0.79, "learning_rate": 2.2316135262218787e-06, "loss": 0.5777, "step": 4104 }, { "epoch": 0.79, "learning_rate": 2.227690694720784e-06, "loss": 0.4486, "step": 4105 }, { "epoch": 0.79, "learning_rate": 2.223770881895061e-06, "loss": 0.5704, "step": 4106 }, { "epoch": 0.79, "learning_rate": 2.2198540892671215e-06, "loss": 0.611, "step": 4107 }, { "epoch": 0.79, "learning_rate": 2.215940318358206e-06, "loss": 0.5716, "step": 4108 }, { "epoch": 0.79, "learning_rate": 2.2120295706883698e-06, "loss": 0.5757, "step": 4109 }, { "epoch": 0.79, "learning_rate": 2.2081218477765097e-06, "loss": 0.5591, "step": 4110 }, { "epoch": 0.79, "learning_rate": 2.204217151140342e-06, "loss": 0.6073, "step": 4111 }, { "epoch": 0.79, "learning_rate": 2.200315482296398e-06, "loss": 0.5275, "step": 4112 }, { "epoch": 0.79, "learning_rate": 2.1964168427600462e-06, "loss": 0.6101, "step": 4113 }, { "epoch": 0.79, "learning_rate": 2.1925212340454737e-06, "loss": 0.5555, "step": 4114 }, { "epoch": 0.79, "learning_rate": 2.1886286576656834e-06, "loss": 0.5857, "step": 4115 }, { "epoch": 0.79, "learning_rate": 2.184739115132517e-06, "loss": 0.5736, "step": 4116 }, { "epoch": 0.79, "learning_rate": 2.1808526079566215e-06, "loss": 0.523, "step": 4117 }, { "epoch": 0.79, "learning_rate": 2.1769691376474722e-06, "loss": 0.5827, "step": 4118 }, { "epoch": 0.79, "learning_rate": 2.1730887057133678e-06, "loss": 0.3653, "step": 4119 }, { "epoch": 0.79, "learning_rate": 2.1692113136614177e-06, "loss": 0.5434, "step": 4120 }, { "epoch": 0.79, "learning_rate": 2.1653369629975595e-06, "loss": 0.5471, "step": 4121 }, { "epoch": 0.79, "learning_rate": 2.1614656552265457e-06, "loss": 0.569, "step": 4122 }, { "epoch": 0.79, "learning_rate": 2.157597391851949e-06, "loss": 0.5983, "step": 4123 }, { "epoch": 0.79, "learning_rate": 2.1537321743761587e-06, "loss": 0.5463, "step": 4124 }, { "epoch": 0.79, "learning_rate": 2.1498700043003773e-06, "loss": 0.4792, "step": 4125 }, { "epoch": 0.79, "learning_rate": 2.1460108831246295e-06, "loss": 0.5814, "step": 4126 }, { "epoch": 0.79, "learning_rate": 2.142154812347753e-06, "loss": 0.4586, "step": 4127 }, { "epoch": 0.79, "learning_rate": 2.1383017934674012e-06, "loss": 0.5303, "step": 4128 }, { "epoch": 0.79, "learning_rate": 2.1344518279800454e-06, "loss": 0.5453, "step": 4129 }, { "epoch": 0.79, "learning_rate": 2.130604917380962e-06, "loss": 0.4571, "step": 4130 }, { "epoch": 0.79, "learning_rate": 2.12676106316425e-06, "loss": 0.4716, "step": 4131 }, { "epoch": 0.79, "learning_rate": 2.1229202668228197e-06, "loss": 0.574, "step": 4132 }, { "epoch": 0.8, "learning_rate": 2.1190825298483855e-06, "loss": 0.5602, "step": 4133 }, { "epoch": 0.8, "learning_rate": 2.115247853731488e-06, "loss": 0.597, "step": 4134 }, { "epoch": 0.8, "learning_rate": 2.11141623996147e-06, "loss": 0.4484, "step": 4135 }, { "epoch": 0.8, "learning_rate": 2.107587690026481e-06, "loss": 0.4808, "step": 4136 }, { "epoch": 0.8, "learning_rate": 2.103762205413493e-06, "loss": 0.5082, "step": 4137 }, { "epoch": 0.8, "learning_rate": 2.0999397876082726e-06, "loss": 0.5742, "step": 4138 }, { "epoch": 0.8, "learning_rate": 2.096120438095404e-06, "loss": 0.5751, "step": 4139 }, { "epoch": 0.8, "learning_rate": 2.092304158358286e-06, "loss": 0.554, "step": 4140 }, { "epoch": 0.8, "learning_rate": 2.0884909498791106e-06, "loss": 0.4739, "step": 4141 }, { "epoch": 0.8, "learning_rate": 2.0846808141388852e-06, "loss": 0.5573, "step": 4142 }, { "epoch": 0.8, "learning_rate": 2.080873752617426e-06, "loss": 0.4911, "step": 4143 }, { "epoch": 0.8, "learning_rate": 2.0770697667933436e-06, "loss": 0.5395, "step": 4144 }, { "epoch": 0.8, "learning_rate": 2.073268858144074e-06, "loss": 0.5556, "step": 4145 }, { "epoch": 0.8, "learning_rate": 2.0694710281458372e-06, "loss": 0.3844, "step": 4146 }, { "epoch": 0.8, "learning_rate": 2.0656762782736693e-06, "loss": 0.5535, "step": 4147 }, { "epoch": 0.8, "learning_rate": 2.061884610001411e-06, "loss": 0.5406, "step": 4148 }, { "epoch": 0.8, "learning_rate": 2.0580960248016966e-06, "loss": 0.5314, "step": 4149 }, { "epoch": 0.8, "learning_rate": 2.0543105241459713e-06, "loss": 0.5837, "step": 4150 }, { "epoch": 0.8, "learning_rate": 2.0505281095044804e-06, "loss": 0.5975, "step": 4151 }, { "epoch": 0.8, "learning_rate": 2.0467487823462696e-06, "loss": 0.5797, "step": 4152 }, { "epoch": 0.8, "learning_rate": 2.042972544139189e-06, "loss": 0.5777, "step": 4153 }, { "epoch": 0.8, "learning_rate": 2.039199396349881e-06, "loss": 0.4174, "step": 4154 }, { "epoch": 0.8, "learning_rate": 2.0354293404437963e-06, "loss": 0.6005, "step": 4155 }, { "epoch": 0.8, "learning_rate": 2.0316623778851784e-06, "loss": 0.4734, "step": 4156 }, { "epoch": 0.8, "learning_rate": 2.0278985101370753e-06, "loss": 0.5628, "step": 4157 }, { "epoch": 0.8, "learning_rate": 2.024137738661329e-06, "loss": 0.5634, "step": 4158 }, { "epoch": 0.8, "learning_rate": 2.020380064918579e-06, "loss": 0.5722, "step": 4159 }, { "epoch": 0.8, "learning_rate": 2.01662549036826e-06, "loss": 0.57, "step": 4160 }, { "epoch": 0.8, "learning_rate": 2.0128740164686134e-06, "loss": 0.4456, "step": 4161 }, { "epoch": 0.8, "learning_rate": 2.009125644676656e-06, "loss": 0.5826, "step": 4162 }, { "epoch": 0.8, "learning_rate": 2.0053803764482226e-06, "loss": 0.5807, "step": 4163 }, { "epoch": 0.8, "learning_rate": 2.001638213237932e-06, "loss": 0.5912, "step": 4164 }, { "epoch": 0.8, "learning_rate": 1.997899156499191e-06, "loss": 0.5528, "step": 4165 }, { "epoch": 0.8, "learning_rate": 1.994163207684212e-06, "loss": 0.5882, "step": 4166 }, { "epoch": 0.8, "learning_rate": 1.9904303682439896e-06, "loss": 0.5837, "step": 4167 }, { "epoch": 0.8, "learning_rate": 1.986700639628316e-06, "loss": 0.586, "step": 4168 }, { "epoch": 0.8, "learning_rate": 1.9829740232857807e-06, "loss": 0.4529, "step": 4169 }, { "epoch": 0.8, "learning_rate": 1.9792505206637523e-06, "loss": 0.5802, "step": 4170 }, { "epoch": 0.8, "learning_rate": 1.9755301332083997e-06, "loss": 0.5533, "step": 4171 }, { "epoch": 0.8, "learning_rate": 1.9718128623646792e-06, "loss": 0.5009, "step": 4172 }, { "epoch": 0.8, "learning_rate": 1.9680987095763315e-06, "loss": 0.5439, "step": 4173 }, { "epoch": 0.8, "learning_rate": 1.964387676285894e-06, "loss": 0.438, "step": 4174 }, { "epoch": 0.8, "learning_rate": 1.9606797639346874e-06, "loss": 0.5504, "step": 4175 }, { "epoch": 0.8, "learning_rate": 1.9569749739628243e-06, "loss": 0.4264, "step": 4176 }, { "epoch": 0.8, "learning_rate": 1.9532733078092034e-06, "loss": 0.5694, "step": 4177 }, { "epoch": 0.8, "learning_rate": 1.9495747669115062e-06, "loss": 0.478, "step": 4178 }, { "epoch": 0.8, "learning_rate": 1.9458793527062035e-06, "loss": 0.5156, "step": 4179 }, { "epoch": 0.8, "learning_rate": 1.9421870666285523e-06, "loss": 0.5873, "step": 4180 }, { "epoch": 0.8, "learning_rate": 1.9384979101125944e-06, "loss": 0.619, "step": 4181 }, { "epoch": 0.8, "learning_rate": 1.934811884591159e-06, "loss": 0.5672, "step": 4182 }, { "epoch": 0.8, "learning_rate": 1.93112899149585e-06, "loss": 0.5488, "step": 4183 }, { "epoch": 0.8, "learning_rate": 1.9274492322570616e-06, "loss": 0.5435, "step": 4184 }, { "epoch": 0.81, "learning_rate": 1.923772608303972e-06, "loss": 0.584, "step": 4185 }, { "epoch": 0.81, "learning_rate": 1.9200991210645394e-06, "loss": 0.5907, "step": 4186 }, { "epoch": 0.81, "learning_rate": 1.916428771965506e-06, "loss": 0.5662, "step": 4187 }, { "epoch": 0.81, "learning_rate": 1.912761562432388e-06, "loss": 0.5498, "step": 4188 }, { "epoch": 0.81, "learning_rate": 1.9090974938894902e-06, "loss": 0.5252, "step": 4189 }, { "epoch": 0.81, "learning_rate": 1.9054365677598963e-06, "loss": 0.5553, "step": 4190 }, { "epoch": 0.81, "learning_rate": 1.9017787854654613e-06, "loss": 0.5666, "step": 4191 }, { "epoch": 0.81, "learning_rate": 1.898124148426832e-06, "loss": 0.5899, "step": 4192 }, { "epoch": 0.81, "learning_rate": 1.8944726580634287e-06, "loss": 0.4606, "step": 4193 }, { "epoch": 0.81, "learning_rate": 1.8908243157934424e-06, "loss": 0.5609, "step": 4194 }, { "epoch": 0.81, "learning_rate": 1.8871791230338499e-06, "loss": 0.5551, "step": 4195 }, { "epoch": 0.81, "learning_rate": 1.883537081200404e-06, "loss": 0.5726, "step": 4196 }, { "epoch": 0.81, "learning_rate": 1.8798981917076254e-06, "loss": 0.5466, "step": 4197 }, { "epoch": 0.81, "learning_rate": 1.876262455968826e-06, "loss": 0.5725, "step": 4198 }, { "epoch": 0.81, "learning_rate": 1.872629875396076e-06, "loss": 0.561, "step": 4199 }, { "epoch": 0.81, "learning_rate": 1.8690004514002314e-06, "loss": 0.5562, "step": 4200 }, { "epoch": 0.81, "learning_rate": 1.8653741853909201e-06, "loss": 0.5893, "step": 4201 }, { "epoch": 0.81, "learning_rate": 1.861751078776538e-06, "loss": 0.5582, "step": 4202 }, { "epoch": 0.81, "learning_rate": 1.8581311329642592e-06, "loss": 0.599, "step": 4203 }, { "epoch": 0.81, "learning_rate": 1.8545143493600293e-06, "loss": 0.472, "step": 4204 }, { "epoch": 0.81, "learning_rate": 1.8509007293685666e-06, "loss": 0.5184, "step": 4205 }, { "epoch": 0.81, "learning_rate": 1.8472902743933608e-06, "loss": 0.4742, "step": 4206 }, { "epoch": 0.81, "learning_rate": 1.8436829858366655e-06, "loss": 0.5822, "step": 4207 }, { "epoch": 0.81, "learning_rate": 1.8400788650995137e-06, "loss": 0.5697, "step": 4208 }, { "epoch": 0.81, "learning_rate": 1.8364779135817045e-06, "loss": 0.4936, "step": 4209 }, { "epoch": 0.81, "learning_rate": 1.8328801326818045e-06, "loss": 0.5747, "step": 4210 }, { "epoch": 0.81, "learning_rate": 1.829285523797155e-06, "loss": 0.5562, "step": 4211 }, { "epoch": 0.81, "learning_rate": 1.8256940883238538e-06, "loss": 0.5711, "step": 4212 }, { "epoch": 0.81, "learning_rate": 1.822105827656776e-06, "loss": 0.5608, "step": 4213 }, { "epoch": 0.81, "learning_rate": 1.8185207431895613e-06, "loss": 0.5941, "step": 4214 }, { "epoch": 0.81, "learning_rate": 1.8149388363146148e-06, "loss": 0.4731, "step": 4215 }, { "epoch": 0.81, "learning_rate": 1.8113601084231091e-06, "loss": 0.5846, "step": 4216 }, { "epoch": 0.81, "learning_rate": 1.8077845609049782e-06, "loss": 0.5766, "step": 4217 }, { "epoch": 0.81, "learning_rate": 1.8042121951489254e-06, "loss": 0.5501, "step": 4218 }, { "epoch": 0.81, "learning_rate": 1.800643012542418e-06, "loss": 0.5501, "step": 4219 }, { "epoch": 0.81, "learning_rate": 1.7970770144716777e-06, "loss": 0.5743, "step": 4220 }, { "epoch": 0.81, "learning_rate": 1.7935142023217056e-06, "loss": 0.5613, "step": 4221 }, { "epoch": 0.81, "learning_rate": 1.7899545774762573e-06, "loss": 0.5383, "step": 4222 }, { "epoch": 0.81, "learning_rate": 1.7863981413178433e-06, "loss": 0.5361, "step": 4223 }, { "epoch": 0.81, "learning_rate": 1.7828448952277456e-06, "loss": 0.5572, "step": 4224 }, { "epoch": 0.81, "learning_rate": 1.7792948405860079e-06, "loss": 0.5048, "step": 4225 }, { "epoch": 0.81, "learning_rate": 1.7757479787714217e-06, "loss": 0.4603, "step": 4226 }, { "epoch": 0.81, "learning_rate": 1.7722043111615572e-06, "loss": 0.5849, "step": 4227 }, { "epoch": 0.81, "learning_rate": 1.768663839132727e-06, "loss": 0.5375, "step": 4228 }, { "epoch": 0.81, "learning_rate": 1.7651265640600113e-06, "loss": 0.3658, "step": 4229 }, { "epoch": 0.81, "learning_rate": 1.7615924873172506e-06, "loss": 0.4522, "step": 4230 }, { "epoch": 0.81, "learning_rate": 1.7580616102770353e-06, "loss": 0.5524, "step": 4231 }, { "epoch": 0.81, "learning_rate": 1.754533934310717e-06, "loss": 0.5347, "step": 4232 }, { "epoch": 0.81, "learning_rate": 1.7510094607884075e-06, "loss": 0.5837, "step": 4233 }, { "epoch": 0.81, "learning_rate": 1.7474881910789698e-06, "loss": 0.5924, "step": 4234 }, { "epoch": 0.81, "learning_rate": 1.7439701265500274e-06, "loss": 0.5559, "step": 4235 }, { "epoch": 0.81, "learning_rate": 1.740455268567951e-06, "loss": 0.5692, "step": 4236 }, { "epoch": 0.82, "learning_rate": 1.7369436184978738e-06, "loss": 0.5836, "step": 4237 }, { "epoch": 0.82, "learning_rate": 1.7334351777036807e-06, "loss": 0.5547, "step": 4238 }, { "epoch": 0.82, "learning_rate": 1.729929947548008e-06, "loss": 0.6208, "step": 4239 }, { "epoch": 0.82, "learning_rate": 1.7264279293922503e-06, "loss": 0.5186, "step": 4240 }, { "epoch": 0.82, "learning_rate": 1.7229291245965463e-06, "loss": 0.5142, "step": 4241 }, { "epoch": 0.82, "learning_rate": 1.7194335345197933e-06, "loss": 0.4884, "step": 4242 }, { "epoch": 0.82, "learning_rate": 1.7159411605196407e-06, "loss": 0.5298, "step": 4243 }, { "epoch": 0.82, "learning_rate": 1.7124520039524805e-06, "loss": 0.569, "step": 4244 }, { "epoch": 0.82, "learning_rate": 1.7089660661734685e-06, "loss": 0.6102, "step": 4245 }, { "epoch": 0.82, "learning_rate": 1.7054833485364962e-06, "loss": 0.5992, "step": 4246 }, { "epoch": 0.82, "learning_rate": 1.702003852394214e-06, "loss": 0.3444, "step": 4247 }, { "epoch": 0.82, "learning_rate": 1.6985275790980205e-06, "loss": 0.4988, "step": 4248 }, { "epoch": 0.82, "learning_rate": 1.6950545299980526e-06, "loss": 0.5593, "step": 4249 }, { "epoch": 0.82, "learning_rate": 1.691584706443209e-06, "loss": 0.5659, "step": 4250 }, { "epoch": 0.82, "learning_rate": 1.6881181097811305e-06, "loss": 0.5535, "step": 4251 }, { "epoch": 0.82, "learning_rate": 1.6846547413581981e-06, "loss": 0.5885, "step": 4252 }, { "epoch": 0.82, "learning_rate": 1.6811946025195459e-06, "loss": 0.439, "step": 4253 }, { "epoch": 0.82, "learning_rate": 1.6777376946090552e-06, "loss": 0.5441, "step": 4254 }, { "epoch": 0.82, "learning_rate": 1.674284018969342e-06, "loss": 0.4469, "step": 4255 }, { "epoch": 0.82, "learning_rate": 1.6708335769417827e-06, "loss": 0.5942, "step": 4256 }, { "epoch": 0.82, "learning_rate": 1.667386369866484e-06, "loss": 0.5847, "step": 4257 }, { "epoch": 0.82, "learning_rate": 1.6639423990823011e-06, "loss": 0.4817, "step": 4258 }, { "epoch": 0.82, "learning_rate": 1.660501665926838e-06, "loss": 0.5572, "step": 4259 }, { "epoch": 0.82, "learning_rate": 1.6570641717364277e-06, "loss": 0.5597, "step": 4260 }, { "epoch": 0.82, "learning_rate": 1.653629917846159e-06, "loss": 0.5955, "step": 4261 }, { "epoch": 0.82, "learning_rate": 1.6501989055898537e-06, "loss": 0.5681, "step": 4262 }, { "epoch": 0.82, "learning_rate": 1.6467711363000794e-06, "loss": 0.5703, "step": 4263 }, { "epoch": 0.82, "learning_rate": 1.6433466113081442e-06, "loss": 0.5597, "step": 4264 }, { "epoch": 0.82, "learning_rate": 1.6399253319440888e-06, "loss": 0.5351, "step": 4265 }, { "epoch": 0.82, "learning_rate": 1.6365072995367004e-06, "loss": 0.5671, "step": 4266 }, { "epoch": 0.82, "learning_rate": 1.6330925154135057e-06, "loss": 0.3898, "step": 4267 }, { "epoch": 0.82, "learning_rate": 1.6296809809007652e-06, "loss": 0.4691, "step": 4268 }, { "epoch": 0.82, "learning_rate": 1.6262726973234844e-06, "loss": 0.4778, "step": 4269 }, { "epoch": 0.82, "learning_rate": 1.6228676660053932e-06, "loss": 0.5666, "step": 4270 }, { "epoch": 0.82, "learning_rate": 1.6194658882689718e-06, "loss": 0.5684, "step": 4271 }, { "epoch": 0.82, "learning_rate": 1.6160673654354331e-06, "loss": 0.5965, "step": 4272 }, { "epoch": 0.82, "learning_rate": 1.6126720988247168e-06, "loss": 0.6166, "step": 4273 }, { "epoch": 0.82, "learning_rate": 1.609280089755515e-06, "loss": 0.5553, "step": 4274 }, { "epoch": 0.82, "learning_rate": 1.605891339545237e-06, "loss": 0.5642, "step": 4275 }, { "epoch": 0.82, "learning_rate": 1.6025058495100388e-06, "loss": 0.3866, "step": 4276 }, { "epoch": 0.82, "learning_rate": 1.5991236209648052e-06, "loss": 0.5831, "step": 4277 }, { "epoch": 0.82, "learning_rate": 1.5957446552231526e-06, "loss": 0.5739, "step": 4278 }, { "epoch": 0.82, "learning_rate": 1.5923689535974307e-06, "loss": 0.5456, "step": 4279 }, { "epoch": 0.82, "learning_rate": 1.588996517398731e-06, "loss": 0.5403, "step": 4280 }, { "epoch": 0.82, "learning_rate": 1.5856273479368611e-06, "loss": 0.576, "step": 4281 }, { "epoch": 0.82, "learning_rate": 1.582261446520371e-06, "loss": 0.5886, "step": 4282 }, { "epoch": 0.82, "learning_rate": 1.5788988144565397e-06, "loss": 0.4441, "step": 4283 }, { "epoch": 0.82, "learning_rate": 1.575539453051369e-06, "loss": 0.5592, "step": 4284 }, { "epoch": 0.82, "learning_rate": 1.572183363609603e-06, "loss": 0.6233, "step": 4285 }, { "epoch": 0.82, "learning_rate": 1.568830547434703e-06, "loss": 0.5985, "step": 4286 }, { "epoch": 0.82, "learning_rate": 1.5654810058288661e-06, "loss": 0.5799, "step": 4287 }, { "epoch": 0.82, "learning_rate": 1.5621347400930176e-06, "loss": 0.5772, "step": 4288 }, { "epoch": 0.83, "learning_rate": 1.5587917515268048e-06, "loss": 0.6118, "step": 4289 }, { "epoch": 0.83, "learning_rate": 1.5554520414286067e-06, "loss": 0.6228, "step": 4290 }, { "epoch": 0.83, "learning_rate": 1.5521156110955293e-06, "loss": 0.5487, "step": 4291 }, { "epoch": 0.83, "learning_rate": 1.5487824618234049e-06, "loss": 0.5764, "step": 4292 }, { "epoch": 0.83, "learning_rate": 1.54545259490679e-06, "loss": 0.4878, "step": 4293 }, { "epoch": 0.83, "learning_rate": 1.5421260116389636e-06, "loss": 0.5388, "step": 4294 }, { "epoch": 0.83, "learning_rate": 1.5388027133119343e-06, "loss": 0.5755, "step": 4295 }, { "epoch": 0.83, "learning_rate": 1.535482701216433e-06, "loss": 0.5849, "step": 4296 }, { "epoch": 0.83, "learning_rate": 1.5321659766419129e-06, "loss": 0.5398, "step": 4297 }, { "epoch": 0.83, "learning_rate": 1.5288525408765564e-06, "loss": 0.5714, "step": 4298 }, { "epoch": 0.83, "learning_rate": 1.5255423952072567e-06, "loss": 0.5417, "step": 4299 }, { "epoch": 0.83, "learning_rate": 1.52223554091964e-06, "loss": 0.5804, "step": 4300 }, { "epoch": 0.83, "learning_rate": 1.5189319792980517e-06, "loss": 0.6045, "step": 4301 }, { "epoch": 0.83, "learning_rate": 1.5156317116255515e-06, "loss": 0.6253, "step": 4302 }, { "epoch": 0.83, "learning_rate": 1.5123347391839305e-06, "loss": 0.4781, "step": 4303 }, { "epoch": 0.83, "learning_rate": 1.5090410632536968e-06, "loss": 0.5633, "step": 4304 }, { "epoch": 0.83, "learning_rate": 1.5057506851140701e-06, "loss": 0.5886, "step": 4305 }, { "epoch": 0.83, "learning_rate": 1.5024636060429998e-06, "loss": 0.5779, "step": 4306 }, { "epoch": 0.83, "learning_rate": 1.4991798273171465e-06, "loss": 0.558, "step": 4307 }, { "epoch": 0.83, "learning_rate": 1.49589935021189e-06, "loss": 0.5898, "step": 4308 }, { "epoch": 0.83, "learning_rate": 1.4926221760013393e-06, "loss": 0.5892, "step": 4309 }, { "epoch": 0.83, "learning_rate": 1.4893483059583014e-06, "loss": 0.4455, "step": 4310 }, { "epoch": 0.83, "learning_rate": 1.4860777413543138e-06, "loss": 0.5444, "step": 4311 }, { "epoch": 0.83, "learning_rate": 1.4828104834596268e-06, "loss": 0.4757, "step": 4312 }, { "epoch": 0.83, "learning_rate": 1.4795465335432036e-06, "loss": 0.5438, "step": 4313 }, { "epoch": 0.83, "learning_rate": 1.4762858928727241e-06, "loss": 0.5764, "step": 4314 }, { "epoch": 0.83, "learning_rate": 1.4730285627145858e-06, "loss": 0.5732, "step": 4315 }, { "epoch": 0.83, "learning_rate": 1.4697745443338984e-06, "loss": 0.5928, "step": 4316 }, { "epoch": 0.83, "learning_rate": 1.4665238389944859e-06, "loss": 0.5975, "step": 4317 }, { "epoch": 0.83, "learning_rate": 1.46327644795888e-06, "loss": 0.496, "step": 4318 }, { "epoch": 0.83, "learning_rate": 1.4600323724883337e-06, "loss": 0.5572, "step": 4319 }, { "epoch": 0.83, "learning_rate": 1.4567916138428072e-06, "loss": 0.4583, "step": 4320 }, { "epoch": 0.83, "learning_rate": 1.4535541732809755e-06, "loss": 0.5792, "step": 4321 }, { "epoch": 0.83, "learning_rate": 1.4503200520602245e-06, "loss": 0.5346, "step": 4322 }, { "epoch": 0.83, "learning_rate": 1.4470892514366442e-06, "loss": 0.3638, "step": 4323 }, { "epoch": 0.83, "learning_rate": 1.443861772665044e-06, "loss": 0.5869, "step": 4324 }, { "epoch": 0.83, "learning_rate": 1.4406376169989389e-06, "loss": 0.5466, "step": 4325 }, { "epoch": 0.83, "learning_rate": 1.4374167856905542e-06, "loss": 0.5229, "step": 4326 }, { "epoch": 0.83, "learning_rate": 1.4341992799908255e-06, "loss": 0.4816, "step": 4327 }, { "epoch": 0.83, "learning_rate": 1.4309851011493903e-06, "loss": 0.5699, "step": 4328 }, { "epoch": 0.83, "learning_rate": 1.427774250414601e-06, "loss": 0.5895, "step": 4329 }, { "epoch": 0.83, "learning_rate": 1.4245667290335175e-06, "loss": 0.5433, "step": 4330 }, { "epoch": 0.83, "learning_rate": 1.421362538251897e-06, "loss": 0.5781, "step": 4331 }, { "epoch": 0.83, "learning_rate": 1.4181616793142173e-06, "loss": 0.5957, "step": 4332 }, { "epoch": 0.83, "learning_rate": 1.414964153463655e-06, "loss": 0.5392, "step": 4333 }, { "epoch": 0.83, "learning_rate": 1.4117699619420878e-06, "loss": 0.5807, "step": 4334 }, { "epoch": 0.83, "learning_rate": 1.4085791059901077e-06, "loss": 0.5559, "step": 4335 }, { "epoch": 0.83, "learning_rate": 1.4053915868470013e-06, "loss": 0.5821, "step": 4336 }, { "epoch": 0.83, "learning_rate": 1.402207405750765e-06, "loss": 0.5807, "step": 4337 }, { "epoch": 0.83, "learning_rate": 1.399026563938105e-06, "loss": 0.537, "step": 4338 }, { "epoch": 0.83, "learning_rate": 1.3958490626444154e-06, "loss": 0.5603, "step": 4339 }, { "epoch": 0.83, "learning_rate": 1.3926749031038055e-06, "loss": 0.4745, "step": 4340 }, { "epoch": 0.84, "learning_rate": 1.3895040865490817e-06, "loss": 0.6173, "step": 4341 }, { "epoch": 0.84, "learning_rate": 1.3863366142117506e-06, "loss": 0.4582, "step": 4342 }, { "epoch": 0.84, "learning_rate": 1.383172487322023e-06, "loss": 0.5578, "step": 4343 }, { "epoch": 0.84, "learning_rate": 1.3800117071088104e-06, "loss": 0.5825, "step": 4344 }, { "epoch": 0.84, "learning_rate": 1.3768542747997215e-06, "loss": 0.4022, "step": 4345 }, { "epoch": 0.84, "learning_rate": 1.3737001916210713e-06, "loss": 0.5638, "step": 4346 }, { "epoch": 0.84, "learning_rate": 1.370549458797863e-06, "loss": 0.5345, "step": 4347 }, { "epoch": 0.84, "learning_rate": 1.3674020775538078e-06, "loss": 0.5825, "step": 4348 }, { "epoch": 0.84, "learning_rate": 1.3642580491113122e-06, "loss": 0.583, "step": 4349 }, { "epoch": 0.84, "learning_rate": 1.3611173746914797e-06, "loss": 0.591, "step": 4350 }, { "epoch": 0.84, "learning_rate": 1.3579800555141165e-06, "loss": 0.6177, "step": 4351 }, { "epoch": 0.84, "learning_rate": 1.3548460927977158e-06, "loss": 0.5921, "step": 4352 }, { "epoch": 0.84, "learning_rate": 1.351715487759474e-06, "loss": 0.4674, "step": 4353 }, { "epoch": 0.84, "learning_rate": 1.3485882416152819e-06, "loss": 0.3856, "step": 4354 }, { "epoch": 0.84, "learning_rate": 1.3454643555797276e-06, "loss": 0.4723, "step": 4355 }, { "epoch": 0.84, "learning_rate": 1.3423438308660929e-06, "loss": 0.4939, "step": 4356 }, { "epoch": 0.84, "learning_rate": 1.3392266686863508e-06, "loss": 0.5539, "step": 4357 }, { "epoch": 0.84, "learning_rate": 1.3361128702511716e-06, "loss": 0.5325, "step": 4358 }, { "epoch": 0.84, "learning_rate": 1.3330024367699224e-06, "loss": 0.56, "step": 4359 }, { "epoch": 0.84, "learning_rate": 1.3298953694506522e-06, "loss": 0.603, "step": 4360 }, { "epoch": 0.84, "learning_rate": 1.3267916695001172e-06, "loss": 0.5616, "step": 4361 }, { "epoch": 0.84, "learning_rate": 1.3236913381237592e-06, "loss": 0.5507, "step": 4362 }, { "epoch": 0.84, "learning_rate": 1.3205943765257057e-06, "loss": 0.5813, "step": 4363 }, { "epoch": 0.84, "learning_rate": 1.317500785908783e-06, "loss": 0.5395, "step": 4364 }, { "epoch": 0.84, "learning_rate": 1.31441056747451e-06, "loss": 0.5584, "step": 4365 }, { "epoch": 0.84, "learning_rate": 1.3113237224230836e-06, "loss": 0.5625, "step": 4366 }, { "epoch": 0.84, "learning_rate": 1.3082402519534076e-06, "loss": 0.5863, "step": 4367 }, { "epoch": 0.84, "learning_rate": 1.3051601572630611e-06, "loss": 0.5535, "step": 4368 }, { "epoch": 0.84, "learning_rate": 1.3020834395483195e-06, "loss": 0.4989, "step": 4369 }, { "epoch": 0.84, "learning_rate": 1.2990101000041445e-06, "loss": 0.5868, "step": 4370 }, { "epoch": 0.84, "learning_rate": 1.2959401398241844e-06, "loss": 0.4956, "step": 4371 }, { "epoch": 0.84, "learning_rate": 1.2928735602007768e-06, "loss": 0.5179, "step": 4372 }, { "epoch": 0.84, "learning_rate": 1.2898103623249458e-06, "loss": 0.6018, "step": 4373 }, { "epoch": 0.84, "learning_rate": 1.2867505473864029e-06, "loss": 0.5739, "step": 4374 }, { "epoch": 0.84, "learning_rate": 1.283694116573546e-06, "loss": 0.5941, "step": 4375 }, { "epoch": 0.84, "learning_rate": 1.2806410710734552e-06, "loss": 0.5443, "step": 4376 }, { "epoch": 0.84, "learning_rate": 1.2775914120718992e-06, "loss": 0.4613, "step": 4377 }, { "epoch": 0.84, "learning_rate": 1.2745451407533294e-06, "loss": 0.6001, "step": 4378 }, { "epoch": 0.84, "learning_rate": 1.2715022583008851e-06, "loss": 0.5398, "step": 4379 }, { "epoch": 0.84, "learning_rate": 1.2684627658963865e-06, "loss": 0.5369, "step": 4380 }, { "epoch": 0.84, "learning_rate": 1.265426664720334e-06, "loss": 0.5588, "step": 4381 }, { "epoch": 0.84, "learning_rate": 1.2623939559519161e-06, "loss": 0.553, "step": 4382 }, { "epoch": 0.84, "learning_rate": 1.2593646407690051e-06, "loss": 0.5809, "step": 4383 }, { "epoch": 0.84, "learning_rate": 1.2563387203481447e-06, "loss": 0.5701, "step": 4384 }, { "epoch": 0.84, "learning_rate": 1.2533161958645755e-06, "loss": 0.5725, "step": 4385 }, { "epoch": 0.84, "learning_rate": 1.2502970684922067e-06, "loss": 0.6152, "step": 4386 }, { "epoch": 0.84, "learning_rate": 1.2472813394036344e-06, "loss": 0.5736, "step": 4387 }, { "epoch": 0.84, "learning_rate": 1.2442690097701327e-06, "loss": 0.5814, "step": 4388 }, { "epoch": 0.84, "learning_rate": 1.2412600807616526e-06, "loss": 0.5011, "step": 4389 }, { "epoch": 0.84, "learning_rate": 1.2382545535468316e-06, "loss": 0.5572, "step": 4390 }, { "epoch": 0.84, "learning_rate": 1.2352524292929823e-06, "loss": 0.3628, "step": 4391 }, { "epoch": 0.84, "learning_rate": 1.2322537091660912e-06, "loss": 0.469, "step": 4392 }, { "epoch": 0.85, "learning_rate": 1.22925839433083e-06, "loss": 0.5539, "step": 4393 }, { "epoch": 0.85, "learning_rate": 1.2262664859505434e-06, "loss": 0.4677, "step": 4394 }, { "epoch": 0.85, "learning_rate": 1.2232779851872511e-06, "loss": 0.4675, "step": 4395 }, { "epoch": 0.85, "learning_rate": 1.2202928932016588e-06, "loss": 0.5782, "step": 4396 }, { "epoch": 0.85, "learning_rate": 1.217311211153137e-06, "loss": 0.5725, "step": 4397 }, { "epoch": 0.85, "learning_rate": 1.2143329401997372e-06, "loss": 0.5627, "step": 4398 }, { "epoch": 0.85, "learning_rate": 1.2113580814981884e-06, "loss": 0.5261, "step": 4399 }, { "epoch": 0.85, "learning_rate": 1.2083866362038865e-06, "loss": 0.4515, "step": 4400 }, { "epoch": 0.85, "learning_rate": 1.2054186054709105e-06, "loss": 0.5683, "step": 4401 }, { "epoch": 0.85, "learning_rate": 1.2024539904520072e-06, "loss": 0.581, "step": 4402 }, { "epoch": 0.85, "learning_rate": 1.1994927922985999e-06, "loss": 0.4567, "step": 4403 }, { "epoch": 0.85, "learning_rate": 1.1965350121607866e-06, "loss": 0.6001, "step": 4404 }, { "epoch": 0.85, "learning_rate": 1.1935806511873306e-06, "loss": 0.5599, "step": 4405 }, { "epoch": 0.85, "learning_rate": 1.1906297105256725e-06, "loss": 0.568, "step": 4406 }, { "epoch": 0.85, "learning_rate": 1.187682191321925e-06, "loss": 0.5922, "step": 4407 }, { "epoch": 0.85, "learning_rate": 1.1847380947208697e-06, "loss": 0.572, "step": 4408 }, { "epoch": 0.85, "learning_rate": 1.1817974218659621e-06, "loss": 0.4722, "step": 4409 }, { "epoch": 0.85, "learning_rate": 1.178860173899321e-06, "loss": 0.5374, "step": 4410 }, { "epoch": 0.85, "learning_rate": 1.1759263519617437e-06, "loss": 0.5036, "step": 4411 }, { "epoch": 0.85, "learning_rate": 1.172995957192693e-06, "loss": 0.5881, "step": 4412 }, { "epoch": 0.85, "learning_rate": 1.1700689907302953e-06, "loss": 0.4665, "step": 4413 }, { "epoch": 0.85, "learning_rate": 1.167145453711358e-06, "loss": 0.5909, "step": 4414 }, { "epoch": 0.85, "learning_rate": 1.1642253472713427e-06, "loss": 0.5619, "step": 4415 }, { "epoch": 0.85, "learning_rate": 1.161308672544389e-06, "loss": 0.3737, "step": 4416 }, { "epoch": 0.85, "learning_rate": 1.1583954306633004e-06, "loss": 0.4775, "step": 4417 }, { "epoch": 0.85, "learning_rate": 1.1554856227595435e-06, "loss": 0.5164, "step": 4418 }, { "epoch": 0.85, "learning_rate": 1.1525792499632526e-06, "loss": 0.4443, "step": 4419 }, { "epoch": 0.85, "learning_rate": 1.1496763134032363e-06, "loss": 0.5333, "step": 4420 }, { "epoch": 0.85, "learning_rate": 1.1467768142069546e-06, "loss": 0.5555, "step": 4421 }, { "epoch": 0.85, "learning_rate": 1.1438807535005437e-06, "loss": 0.5505, "step": 4422 }, { "epoch": 0.85, "learning_rate": 1.1409881324088013e-06, "loss": 0.5668, "step": 4423 }, { "epoch": 0.85, "learning_rate": 1.138098952055181e-06, "loss": 0.5392, "step": 4424 }, { "epoch": 0.85, "learning_rate": 1.1352132135618165e-06, "loss": 0.4423, "step": 4425 }, { "epoch": 0.85, "learning_rate": 1.13233091804949e-06, "loss": 0.5638, "step": 4426 }, { "epoch": 0.85, "learning_rate": 1.1294520666376518e-06, "loss": 0.5671, "step": 4427 }, { "epoch": 0.85, "learning_rate": 1.1265766604444172e-06, "loss": 0.6014, "step": 4428 }, { "epoch": 0.85, "learning_rate": 1.1237047005865576e-06, "loss": 0.58, "step": 4429 }, { "epoch": 0.85, "learning_rate": 1.1208361881795116e-06, "loss": 0.5432, "step": 4430 }, { "epoch": 0.85, "learning_rate": 1.1179711243373736e-06, "loss": 0.4436, "step": 4431 }, { "epoch": 0.85, "learning_rate": 1.1151095101729047e-06, "loss": 0.5546, "step": 4432 }, { "epoch": 0.85, "learning_rate": 1.1122513467975237e-06, "loss": 0.5814, "step": 4433 }, { "epoch": 0.85, "learning_rate": 1.1093966353213036e-06, "loss": 0.5608, "step": 4434 }, { "epoch": 0.85, "learning_rate": 1.1065453768529844e-06, "loss": 0.5595, "step": 4435 }, { "epoch": 0.85, "learning_rate": 1.103697572499961e-06, "loss": 0.5312, "step": 4436 }, { "epoch": 0.85, "learning_rate": 1.1008532233682878e-06, "loss": 0.567, "step": 4437 }, { "epoch": 0.85, "learning_rate": 1.0980123305626812e-06, "loss": 0.4153, "step": 4438 }, { "epoch": 0.85, "learning_rate": 1.0951748951865048e-06, "loss": 0.5804, "step": 4439 }, { "epoch": 0.85, "learning_rate": 1.0923409183417887e-06, "loss": 0.5833, "step": 4440 }, { "epoch": 0.85, "learning_rate": 1.0895104011292202e-06, "loss": 0.6102, "step": 4441 }, { "epoch": 0.85, "learning_rate": 1.0866833446481317e-06, "loss": 0.4622, "step": 4442 }, { "epoch": 0.85, "learning_rate": 1.0838597499965276e-06, "loss": 0.5547, "step": 4443 }, { "epoch": 0.85, "learning_rate": 1.0810396182710535e-06, "loss": 0.5644, "step": 4444 }, { "epoch": 0.86, "learning_rate": 1.0782229505670195e-06, "loss": 0.5822, "step": 4445 }, { "epoch": 0.86, "learning_rate": 1.0754097479783876e-06, "loss": 0.5218, "step": 4446 }, { "epoch": 0.86, "learning_rate": 1.0726000115977696e-06, "loss": 0.596, "step": 4447 }, { "epoch": 0.86, "learning_rate": 1.069793742516435e-06, "loss": 0.5723, "step": 4448 }, { "epoch": 0.86, "learning_rate": 1.066990941824312e-06, "loss": 0.5679, "step": 4449 }, { "epoch": 0.86, "learning_rate": 1.0641916106099691e-06, "loss": 0.5953, "step": 4450 }, { "epoch": 0.86, "learning_rate": 1.0613957499606388e-06, "loss": 0.4822, "step": 4451 }, { "epoch": 0.86, "learning_rate": 1.0586033609622004e-06, "loss": 0.5912, "step": 4452 }, { "epoch": 0.86, "learning_rate": 1.0558144446991836e-06, "loss": 0.5944, "step": 4453 }, { "epoch": 0.86, "learning_rate": 1.053029002254773e-06, "loss": 0.5725, "step": 4454 }, { "epoch": 0.86, "learning_rate": 1.0502470347108017e-06, "loss": 0.3453, "step": 4455 }, { "epoch": 0.86, "learning_rate": 1.0474685431477537e-06, "loss": 0.5429, "step": 4456 }, { "epoch": 0.86, "learning_rate": 1.0446935286447657e-06, "loss": 0.5446, "step": 4457 }, { "epoch": 0.86, "learning_rate": 1.0419219922796175e-06, "loss": 0.5567, "step": 4458 }, { "epoch": 0.86, "learning_rate": 1.039153935128744e-06, "loss": 0.584, "step": 4459 }, { "epoch": 0.86, "learning_rate": 1.0363893582672246e-06, "loss": 0.5456, "step": 4460 }, { "epoch": 0.86, "learning_rate": 1.033628262768792e-06, "loss": 0.5486, "step": 4461 }, { "epoch": 0.86, "learning_rate": 1.0308706497058252e-06, "loss": 0.5648, "step": 4462 }, { "epoch": 0.86, "learning_rate": 1.0281165201493437e-06, "loss": 0.5761, "step": 4463 }, { "epoch": 0.86, "learning_rate": 1.0253658751690232e-06, "loss": 0.5558, "step": 4464 }, { "epoch": 0.86, "learning_rate": 1.0226187158331825e-06, "loss": 0.5491, "step": 4465 }, { "epoch": 0.86, "learning_rate": 1.0198750432087855e-06, "loss": 0.5395, "step": 4466 }, { "epoch": 0.86, "learning_rate": 1.017134858361446e-06, "loss": 0.5628, "step": 4467 }, { "epoch": 0.86, "learning_rate": 1.0143981623554155e-06, "loss": 0.5672, "step": 4468 }, { "epoch": 0.86, "learning_rate": 1.0116649562535984e-06, "loss": 0.5975, "step": 4469 }, { "epoch": 0.86, "learning_rate": 1.0089352411175424e-06, "loss": 0.5529, "step": 4470 }, { "epoch": 0.86, "learning_rate": 1.006209018007429e-06, "loss": 0.3809, "step": 4471 }, { "epoch": 0.86, "learning_rate": 1.0034862879821029e-06, "loss": 0.5237, "step": 4472 }, { "epoch": 0.86, "learning_rate": 1.0007670520990331e-06, "loss": 0.5715, "step": 4473 }, { "epoch": 0.86, "learning_rate": 9.98051311414342e-07, "loss": 0.5012, "step": 4474 }, { "epoch": 0.86, "learning_rate": 9.953390669827944e-07, "loss": 0.4454, "step": 4475 }, { "epoch": 0.86, "learning_rate": 9.926303198577913e-07, "loss": 0.4973, "step": 4476 }, { "epoch": 0.86, "learning_rate": 9.899250710913767e-07, "loss": 0.5385, "step": 4477 }, { "epoch": 0.86, "learning_rate": 9.872233217342463e-07, "loss": 0.5435, "step": 4478 }, { "epoch": 0.86, "learning_rate": 9.845250728357214e-07, "loss": 0.6136, "step": 4479 }, { "epoch": 0.86, "learning_rate": 9.818303254437723e-07, "loss": 0.5545, "step": 4480 }, { "epoch": 0.86, "learning_rate": 9.791390806050117e-07, "loss": 0.5842, "step": 4481 }, { "epoch": 0.86, "learning_rate": 9.764513393646812e-07, "loss": 0.5897, "step": 4482 }, { "epoch": 0.86, "learning_rate": 9.737671027666728e-07, "loss": 0.5518, "step": 4483 }, { "epoch": 0.86, "learning_rate": 9.710863718535135e-07, "loss": 0.5788, "step": 4484 }, { "epoch": 0.86, "learning_rate": 9.684091476663659e-07, "loss": 0.537, "step": 4485 }, { "epoch": 0.86, "learning_rate": 9.657354312450363e-07, "loss": 0.5019, "step": 4486 }, { "epoch": 0.86, "learning_rate": 9.630652236279626e-07, "loss": 0.5613, "step": 4487 }, { "epoch": 0.86, "learning_rate": 9.603985258522219e-07, "loss": 0.5357, "step": 4488 }, { "epoch": 0.86, "learning_rate": 9.577353389535315e-07, "loss": 0.5724, "step": 4489 }, { "epoch": 0.86, "learning_rate": 9.550756639662417e-07, "loss": 0.5534, "step": 4490 }, { "epoch": 0.86, "learning_rate": 9.524195019233407e-07, "loss": 0.571, "step": 4491 }, { "epoch": 0.86, "learning_rate": 9.497668538564475e-07, "loss": 0.5244, "step": 4492 }, { "epoch": 0.86, "learning_rate": 9.471177207958238e-07, "loss": 0.5439, "step": 4493 }, { "epoch": 0.86, "learning_rate": 9.444721037703597e-07, "loss": 0.548, "step": 4494 }, { "epoch": 0.86, "learning_rate": 9.418300038075845e-07, "loss": 0.5609, "step": 4495 }, { "epoch": 0.86, "learning_rate": 9.391914219336606e-07, "loss": 0.5295, "step": 4496 }, { "epoch": 0.87, "learning_rate": 9.365563591733784e-07, "loss": 0.536, "step": 4497 }, { "epoch": 0.87, "learning_rate": 9.33924816550168e-07, "loss": 0.561, "step": 4498 }, { "epoch": 0.87, "learning_rate": 9.31296795086093e-07, "loss": 0.5934, "step": 4499 }, { "epoch": 0.87, "learning_rate": 9.286722958018391e-07, "loss": 0.5313, "step": 4500 }, { "epoch": 0.87, "learning_rate": 9.260513197167398e-07, "loss": 0.5901, "step": 4501 }, { "epoch": 0.87, "learning_rate": 9.234338678487509e-07, "loss": 0.5525, "step": 4502 }, { "epoch": 0.87, "learning_rate": 9.208199412144559e-07, "loss": 0.4792, "step": 4503 }, { "epoch": 0.87, "learning_rate": 9.182095408290781e-07, "loss": 0.5703, "step": 4504 }, { "epoch": 0.87, "learning_rate": 9.156026677064633e-07, "loss": 0.577, "step": 4505 }, { "epoch": 0.87, "learning_rate": 9.129993228590917e-07, "loss": 0.5625, "step": 4506 }, { "epoch": 0.87, "learning_rate": 9.103995072980765e-07, "loss": 0.5481, "step": 4507 }, { "epoch": 0.87, "learning_rate": 9.078032220331523e-07, "loss": 0.5431, "step": 4508 }, { "epoch": 0.87, "learning_rate": 9.052104680726859e-07, "loss": 0.5586, "step": 4509 }, { "epoch": 0.87, "learning_rate": 9.026212464236772e-07, "loss": 0.5488, "step": 4510 }, { "epoch": 0.87, "learning_rate": 9.000355580917464e-07, "loss": 0.593, "step": 4511 }, { "epoch": 0.87, "learning_rate": 8.974534040811444e-07, "loss": 0.572, "step": 4512 }, { "epoch": 0.87, "learning_rate": 8.948747853947526e-07, "loss": 0.4637, "step": 4513 }, { "epoch": 0.87, "learning_rate": 8.922997030340752e-07, "loss": 0.4732, "step": 4514 }, { "epoch": 0.87, "learning_rate": 8.897281579992467e-07, "loss": 0.4808, "step": 4515 }, { "epoch": 0.87, "learning_rate": 8.871601512890238e-07, "loss": 0.6018, "step": 4516 }, { "epoch": 0.87, "learning_rate": 8.845956839007897e-07, "loss": 0.6005, "step": 4517 }, { "epoch": 0.87, "learning_rate": 8.820347568305543e-07, "loss": 0.5553, "step": 4518 }, { "epoch": 0.87, "learning_rate": 8.794773710729543e-07, "loss": 0.5672, "step": 4519 }, { "epoch": 0.87, "learning_rate": 8.769235276212496e-07, "loss": 0.6163, "step": 4520 }, { "epoch": 0.87, "learning_rate": 8.743732274673189e-07, "loss": 0.5523, "step": 4521 }, { "epoch": 0.87, "learning_rate": 8.718264716016722e-07, "loss": 0.3486, "step": 4522 }, { "epoch": 0.87, "learning_rate": 8.692832610134428e-07, "loss": 0.588, "step": 4523 }, { "epoch": 0.87, "learning_rate": 8.66743596690377e-07, "loss": 0.4569, "step": 4524 }, { "epoch": 0.87, "learning_rate": 8.642074796188594e-07, "loss": 0.6022, "step": 4525 }, { "epoch": 0.87, "learning_rate": 8.61674910783884e-07, "loss": 0.5855, "step": 4526 }, { "epoch": 0.87, "learning_rate": 8.59145891169072e-07, "loss": 0.5963, "step": 4527 }, { "epoch": 0.87, "learning_rate": 8.566204217566664e-07, "loss": 0.5764, "step": 4528 }, { "epoch": 0.87, "learning_rate": 8.540985035275273e-07, "loss": 0.4975, "step": 4529 }, { "epoch": 0.87, "learning_rate": 8.515801374611432e-07, "loss": 0.4714, "step": 4530 }, { "epoch": 0.87, "learning_rate": 8.490653245356184e-07, "loss": 0.5727, "step": 4531 }, { "epoch": 0.87, "learning_rate": 8.465540657276728e-07, "loss": 0.3425, "step": 4532 }, { "epoch": 0.87, "learning_rate": 8.44046362012656e-07, "loss": 0.5594, "step": 4533 }, { "epoch": 0.87, "learning_rate": 8.415422143645247e-07, "loss": 0.5746, "step": 4534 }, { "epoch": 0.87, "learning_rate": 8.390416237558641e-07, "loss": 0.5426, "step": 4535 }, { "epoch": 0.87, "learning_rate": 8.365445911578785e-07, "loss": 0.5447, "step": 4536 }, { "epoch": 0.87, "learning_rate": 8.340511175403809e-07, "loss": 0.4754, "step": 4537 }, { "epoch": 0.87, "learning_rate": 8.315612038718101e-07, "loss": 0.4463, "step": 4538 }, { "epoch": 0.87, "learning_rate": 8.290748511192214e-07, "loss": 0.5591, "step": 4539 }, { "epoch": 0.87, "learning_rate": 8.265920602482825e-07, "loss": 0.4678, "step": 4540 }, { "epoch": 0.87, "learning_rate": 8.241128322232816e-07, "loss": 0.4646, "step": 4541 }, { "epoch": 0.87, "learning_rate": 8.216371680071244e-07, "loss": 0.5401, "step": 4542 }, { "epoch": 0.87, "learning_rate": 8.191650685613273e-07, "loss": 0.6077, "step": 4543 }, { "epoch": 0.87, "learning_rate": 8.166965348460298e-07, "loss": 0.449, "step": 4544 }, { "epoch": 0.87, "learning_rate": 8.142315678199764e-07, "loss": 0.5707, "step": 4545 }, { "epoch": 0.87, "learning_rate": 8.117701684405343e-07, "loss": 0.4391, "step": 4546 }, { "epoch": 0.87, "learning_rate": 8.093123376636836e-07, "loss": 0.5596, "step": 4547 }, { "epoch": 0.87, "learning_rate": 8.06858076444017e-07, "loss": 0.6021, "step": 4548 }, { "epoch": 0.88, "learning_rate": 8.044073857347423e-07, "loss": 0.5814, "step": 4549 }, { "epoch": 0.88, "learning_rate": 8.019602664876758e-07, "loss": 0.5919, "step": 4550 }, { "epoch": 0.88, "learning_rate": 7.995167196532527e-07, "loss": 0.5819, "step": 4551 }, { "epoch": 0.88, "learning_rate": 7.970767461805218e-07, "loss": 0.5833, "step": 4552 }, { "epoch": 0.88, "learning_rate": 7.946403470171326e-07, "loss": 0.5665, "step": 4553 }, { "epoch": 0.88, "learning_rate": 7.922075231093628e-07, "loss": 0.5687, "step": 4554 }, { "epoch": 0.88, "learning_rate": 7.897782754020889e-07, "loss": 0.5117, "step": 4555 }, { "epoch": 0.88, "learning_rate": 7.873526048388025e-07, "loss": 0.5796, "step": 4556 }, { "epoch": 0.88, "learning_rate": 7.849305123616091e-07, "loss": 0.5557, "step": 4557 }, { "epoch": 0.88, "learning_rate": 7.825119989112173e-07, "loss": 0.5761, "step": 4558 }, { "epoch": 0.88, "learning_rate": 7.800970654269513e-07, "loss": 0.5449, "step": 4559 }, { "epoch": 0.88, "learning_rate": 7.776857128467464e-07, "loss": 0.5662, "step": 4560 }, { "epoch": 0.88, "learning_rate": 7.75277942107141e-07, "loss": 0.5655, "step": 4561 }, { "epoch": 0.88, "learning_rate": 7.728737541432862e-07, "loss": 0.4814, "step": 4562 }, { "epoch": 0.88, "learning_rate": 7.704731498889428e-07, "loss": 0.4618, "step": 4563 }, { "epoch": 0.88, "learning_rate": 7.680761302764727e-07, "loss": 0.4598, "step": 4564 }, { "epoch": 0.88, "learning_rate": 7.65682696236858e-07, "loss": 0.5898, "step": 4565 }, { "epoch": 0.88, "learning_rate": 7.632928486996749e-07, "loss": 0.5405, "step": 4566 }, { "epoch": 0.88, "learning_rate": 7.609065885931155e-07, "loss": 0.5223, "step": 4567 }, { "epoch": 0.88, "learning_rate": 7.585239168439762e-07, "loss": 0.5962, "step": 4568 }, { "epoch": 0.88, "learning_rate": 7.561448343776567e-07, "loss": 0.5791, "step": 4569 }, { "epoch": 0.88, "learning_rate": 7.537693421181658e-07, "loss": 0.5346, "step": 4570 }, { "epoch": 0.88, "learning_rate": 7.513974409881186e-07, "loss": 0.4844, "step": 4571 }, { "epoch": 0.88, "learning_rate": 7.49029131908734e-07, "loss": 0.4344, "step": 4572 }, { "epoch": 0.88, "learning_rate": 7.466644157998371e-07, "loss": 0.6064, "step": 4573 }, { "epoch": 0.88, "learning_rate": 7.443032935798533e-07, "loss": 0.429, "step": 4574 }, { "epoch": 0.88, "learning_rate": 7.419457661658169e-07, "loss": 0.5828, "step": 4575 }, { "epoch": 0.88, "learning_rate": 7.395918344733644e-07, "loss": 0.56, "step": 4576 }, { "epoch": 0.88, "learning_rate": 7.372414994167354e-07, "loss": 0.4562, "step": 4577 }, { "epoch": 0.88, "learning_rate": 7.348947619087754e-07, "loss": 0.418, "step": 4578 }, { "epoch": 0.88, "learning_rate": 7.325516228609264e-07, "loss": 0.5933, "step": 4579 }, { "epoch": 0.88, "learning_rate": 7.302120831832382e-07, "loss": 0.6046, "step": 4580 }, { "epoch": 0.88, "learning_rate": 7.278761437843629e-07, "loss": 0.5592, "step": 4581 }, { "epoch": 0.88, "learning_rate": 7.255438055715469e-07, "loss": 0.5708, "step": 4582 }, { "epoch": 0.88, "learning_rate": 7.232150694506512e-07, "loss": 0.5204, "step": 4583 }, { "epoch": 0.88, "learning_rate": 7.208899363261234e-07, "loss": 0.5434, "step": 4584 }, { "epoch": 0.88, "learning_rate": 7.185684071010224e-07, "loss": 0.5709, "step": 4585 }, { "epoch": 0.88, "learning_rate": 7.162504826770033e-07, "loss": 0.5816, "step": 4586 }, { "epoch": 0.88, "learning_rate": 7.139361639543185e-07, "loss": 0.4274, "step": 4587 }, { "epoch": 0.88, "learning_rate": 7.116254518318222e-07, "loss": 0.5264, "step": 4588 }, { "epoch": 0.88, "learning_rate": 7.093183472069753e-07, "loss": 0.5247, "step": 4589 }, { "epoch": 0.88, "learning_rate": 7.070148509758223e-07, "loss": 0.5002, "step": 4590 }, { "epoch": 0.88, "learning_rate": 7.047149640330197e-07, "loss": 0.6435, "step": 4591 }, { "epoch": 0.88, "learning_rate": 7.024186872718164e-07, "loss": 0.5657, "step": 4592 }, { "epoch": 0.88, "learning_rate": 7.001260215840567e-07, "loss": 0.574, "step": 4593 }, { "epoch": 0.88, "learning_rate": 6.978369678601892e-07, "loss": 0.5781, "step": 4594 }, { "epoch": 0.88, "learning_rate": 6.955515269892533e-07, "loss": 0.5656, "step": 4595 }, { "epoch": 0.88, "learning_rate": 6.932696998588895e-07, "loss": 0.5487, "step": 4596 }, { "epoch": 0.88, "learning_rate": 6.909914873553347e-07, "loss": 0.4806, "step": 4597 }, { "epoch": 0.88, "learning_rate": 6.887168903634178e-07, "loss": 0.4614, "step": 4598 }, { "epoch": 0.88, "learning_rate": 6.864459097665654e-07, "loss": 0.603, "step": 4599 }, { "epoch": 0.88, "learning_rate": 6.84178546446802e-07, "loss": 0.5466, "step": 4600 }, { "epoch": 0.89, "learning_rate": 6.819148012847454e-07, "loss": 0.6173, "step": 4601 }, { "epoch": 0.89, "learning_rate": 6.796546751596089e-07, "loss": 0.5138, "step": 4602 }, { "epoch": 0.89, "learning_rate": 6.77398168949196e-07, "loss": 0.5536, "step": 4603 }, { "epoch": 0.89, "learning_rate": 6.751452835299111e-07, "loss": 0.5398, "step": 4604 }, { "epoch": 0.89, "learning_rate": 6.728960197767475e-07, "loss": 0.5719, "step": 4605 }, { "epoch": 0.89, "learning_rate": 6.706503785632934e-07, "loss": 0.448, "step": 4606 }, { "epoch": 0.89, "learning_rate": 6.68408360761732e-07, "loss": 0.446, "step": 4607 }, { "epoch": 0.89, "learning_rate": 6.661699672428334e-07, "loss": 0.5434, "step": 4608 }, { "epoch": 0.89, "learning_rate": 6.639351988759657e-07, "loss": 0.4508, "step": 4609 }, { "epoch": 0.89, "learning_rate": 6.6170405652909e-07, "loss": 0.4338, "step": 4610 }, { "epoch": 0.89, "learning_rate": 6.594765410687487e-07, "loss": 0.3769, "step": 4611 }, { "epoch": 0.89, "learning_rate": 6.57252653360092e-07, "loss": 0.5549, "step": 4612 }, { "epoch": 0.89, "learning_rate": 6.550323942668469e-07, "loss": 0.5816, "step": 4613 }, { "epoch": 0.89, "learning_rate": 6.528157646513378e-07, "loss": 0.5425, "step": 4614 }, { "epoch": 0.89, "learning_rate": 6.506027653744796e-07, "loss": 0.5696, "step": 4615 }, { "epoch": 0.89, "learning_rate": 6.483933972957734e-07, "loss": 0.5479, "step": 4616 }, { "epoch": 0.89, "learning_rate": 6.461876612733109e-07, "loss": 0.5263, "step": 4617 }, { "epoch": 0.89, "learning_rate": 6.4398555816378e-07, "loss": 0.5251, "step": 4618 }, { "epoch": 0.89, "learning_rate": 6.417870888224476e-07, "loss": 0.5613, "step": 4619 }, { "epoch": 0.89, "learning_rate": 6.395922541031741e-07, "loss": 0.5456, "step": 4620 }, { "epoch": 0.89, "learning_rate": 6.374010548584119e-07, "loss": 0.5259, "step": 4621 }, { "epoch": 0.89, "learning_rate": 6.352134919391928e-07, "loss": 0.5354, "step": 4622 }, { "epoch": 0.89, "learning_rate": 6.330295661951436e-07, "loss": 0.5468, "step": 4623 }, { "epoch": 0.89, "learning_rate": 6.308492784744746e-07, "loss": 0.6044, "step": 4624 }, { "epoch": 0.89, "learning_rate": 6.286726296239854e-07, "loss": 0.5323, "step": 4625 }, { "epoch": 0.89, "learning_rate": 6.264996204890628e-07, "loss": 0.5513, "step": 4626 }, { "epoch": 0.89, "learning_rate": 6.24330251913674e-07, "loss": 0.3647, "step": 4627 }, { "epoch": 0.89, "learning_rate": 6.221645247403807e-07, "loss": 0.6065, "step": 4628 }, { "epoch": 0.89, "learning_rate": 6.200024398103255e-07, "loss": 0.4666, "step": 4629 }, { "epoch": 0.89, "learning_rate": 6.178439979632367e-07, "loss": 0.5672, "step": 4630 }, { "epoch": 0.89, "learning_rate": 6.156892000374293e-07, "loss": 0.5543, "step": 4631 }, { "epoch": 0.89, "learning_rate": 6.135380468698004e-07, "loss": 0.5541, "step": 4632 }, { "epoch": 0.89, "learning_rate": 6.113905392958342e-07, "loss": 0.5215, "step": 4633 }, { "epoch": 0.89, "learning_rate": 6.092466781495976e-07, "loss": 0.5623, "step": 4634 }, { "epoch": 0.89, "learning_rate": 6.071064642637404e-07, "loss": 0.5542, "step": 4635 }, { "epoch": 0.89, "learning_rate": 6.049698984695007e-07, "loss": 0.5913, "step": 4636 }, { "epoch": 0.89, "learning_rate": 6.028369815966917e-07, "loss": 0.5592, "step": 4637 }, { "epoch": 0.89, "learning_rate": 6.007077144737161e-07, "loss": 0.5824, "step": 4638 }, { "epoch": 0.89, "learning_rate": 5.985820979275569e-07, "loss": 0.5553, "step": 4639 }, { "epoch": 0.89, "learning_rate": 5.96460132783776e-07, "loss": 0.5924, "step": 4640 }, { "epoch": 0.89, "learning_rate": 5.943418198665251e-07, "loss": 0.6012, "step": 4641 }, { "epoch": 0.89, "learning_rate": 5.922271599985286e-07, "loss": 0.5906, "step": 4642 }, { "epoch": 0.89, "learning_rate": 5.901161540010969e-07, "loss": 0.6441, "step": 4643 }, { "epoch": 0.89, "learning_rate": 5.880088026941233e-07, "loss": 0.5855, "step": 4644 }, { "epoch": 0.89, "learning_rate": 5.859051068960741e-07, "loss": 0.4514, "step": 4645 }, { "epoch": 0.89, "learning_rate": 5.838050674240025e-07, "loss": 0.5534, "step": 4646 }, { "epoch": 0.89, "learning_rate": 5.817086850935416e-07, "loss": 0.4235, "step": 4647 }, { "epoch": 0.89, "learning_rate": 5.796159607189001e-07, "loss": 0.5525, "step": 4648 }, { "epoch": 0.89, "learning_rate": 5.775268951128676e-07, "loss": 0.576, "step": 4649 }, { "epoch": 0.89, "learning_rate": 5.754414890868154e-07, "loss": 0.5945, "step": 4650 }, { "epoch": 0.89, "learning_rate": 5.733597434506877e-07, "loss": 0.5648, "step": 4651 }, { "epoch": 0.9, "learning_rate": 5.712816590130133e-07, "loss": 0.5635, "step": 4652 }, { "epoch": 0.9, "learning_rate": 5.692072365808954e-07, "loss": 0.5502, "step": 4653 }, { "epoch": 0.9, "learning_rate": 5.671364769600162e-07, "loss": 0.5354, "step": 4654 }, { "epoch": 0.9, "learning_rate": 5.650693809546348e-07, "loss": 0.5617, "step": 4655 }, { "epoch": 0.9, "learning_rate": 5.630059493675866e-07, "loss": 0.576, "step": 4656 }, { "epoch": 0.9, "learning_rate": 5.60946183000285e-07, "loss": 0.5513, "step": 4657 }, { "epoch": 0.9, "learning_rate": 5.5889008265272e-07, "loss": 0.5564, "step": 4658 }, { "epoch": 0.9, "learning_rate": 5.568376491234562e-07, "loss": 0.4713, "step": 4659 }, { "epoch": 0.9, "learning_rate": 5.547888832096382e-07, "loss": 0.5838, "step": 4660 }, { "epoch": 0.9, "learning_rate": 5.527437857069784e-07, "loss": 0.5707, "step": 4661 }, { "epoch": 0.9, "learning_rate": 5.507023574097725e-07, "loss": 0.5745, "step": 4662 }, { "epoch": 0.9, "learning_rate": 5.486645991108875e-07, "loss": 0.5766, "step": 4663 }, { "epoch": 0.9, "learning_rate": 5.466305116017623e-07, "loss": 0.5755, "step": 4664 }, { "epoch": 0.9, "learning_rate": 5.446000956724174e-07, "loss": 0.5495, "step": 4665 }, { "epoch": 0.9, "learning_rate": 5.425733521114396e-07, "loss": 0.6272, "step": 4666 }, { "epoch": 0.9, "learning_rate": 5.405502817059937e-07, "loss": 0.522, "step": 4667 }, { "epoch": 0.9, "learning_rate": 5.385308852418191e-07, "loss": 0.5612, "step": 4668 }, { "epoch": 0.9, "learning_rate": 5.365151635032218e-07, "loss": 0.6074, "step": 4669 }, { "epoch": 0.9, "learning_rate": 5.345031172730875e-07, "loss": 0.6054, "step": 4670 }, { "epoch": 0.9, "learning_rate": 5.324947473328735e-07, "loss": 0.5872, "step": 4671 }, { "epoch": 0.9, "learning_rate": 5.304900544626046e-07, "loss": 0.4563, "step": 4672 }, { "epoch": 0.9, "learning_rate": 5.284890394408826e-07, "loss": 0.5505, "step": 4673 }, { "epoch": 0.9, "learning_rate": 5.264917030448757e-07, "loss": 0.4865, "step": 4674 }, { "epoch": 0.9, "learning_rate": 5.244980460503268e-07, "loss": 0.5933, "step": 4675 }, { "epoch": 0.9, "learning_rate": 5.225080692315532e-07, "loss": 0.5127, "step": 4676 }, { "epoch": 0.9, "learning_rate": 5.205217733614353e-07, "loss": 0.5705, "step": 4677 }, { "epoch": 0.9, "learning_rate": 5.185391592114286e-07, "loss": 0.5716, "step": 4678 }, { "epoch": 0.9, "learning_rate": 5.165602275515592e-07, "loss": 0.568, "step": 4679 }, { "epoch": 0.9, "learning_rate": 5.145849791504187e-07, "loss": 0.5487, "step": 4680 }, { "epoch": 0.9, "learning_rate": 5.126134147751716e-07, "loss": 0.5767, "step": 4681 }, { "epoch": 0.9, "learning_rate": 5.106455351915507e-07, "loss": 0.571, "step": 4682 }, { "epoch": 0.9, "learning_rate": 5.086813411638581e-07, "loss": 0.5618, "step": 4683 }, { "epoch": 0.9, "learning_rate": 5.067208334549656e-07, "loss": 0.4855, "step": 4684 }, { "epoch": 0.9, "learning_rate": 5.047640128263087e-07, "loss": 0.614, "step": 4685 }, { "epoch": 0.9, "learning_rate": 5.028108800378961e-07, "loss": 0.5343, "step": 4686 }, { "epoch": 0.9, "learning_rate": 5.008614358483021e-07, "loss": 0.5721, "step": 4687 }, { "epoch": 0.9, "learning_rate": 4.989156810146667e-07, "loss": 0.5678, "step": 4688 }, { "epoch": 0.9, "learning_rate": 4.969736162927019e-07, "loss": 0.5518, "step": 4689 }, { "epoch": 0.9, "learning_rate": 4.9503524243668e-07, "loss": 0.5822, "step": 4690 }, { "epoch": 0.9, "learning_rate": 4.931005601994432e-07, "loss": 0.6057, "step": 4691 }, { "epoch": 0.9, "learning_rate": 4.911695703324038e-07, "loss": 0.5189, "step": 4692 }, { "epoch": 0.9, "learning_rate": 4.892422735855284e-07, "loss": 0.5434, "step": 4693 }, { "epoch": 0.9, "learning_rate": 4.873186707073663e-07, "loss": 0.6028, "step": 4694 }, { "epoch": 0.9, "learning_rate": 4.853987624450151e-07, "loss": 0.5383, "step": 4695 }, { "epoch": 0.9, "learning_rate": 4.834825495441475e-07, "loss": 0.6133, "step": 4696 }, { "epoch": 0.9, "learning_rate": 4.815700327490014e-07, "loss": 0.4315, "step": 4697 }, { "epoch": 0.9, "learning_rate": 4.796612128023726e-07, "loss": 0.5666, "step": 4698 }, { "epoch": 0.9, "learning_rate": 4.777560904456236e-07, "loss": 0.3713, "step": 4699 }, { "epoch": 0.9, "learning_rate": 4.7585466641868696e-07, "loss": 0.5065, "step": 4700 }, { "epoch": 0.9, "learning_rate": 4.7395694146004976e-07, "loss": 0.5902, "step": 4701 }, { "epoch": 0.9, "learning_rate": 4.7206291630677024e-07, "loss": 0.5894, "step": 4702 }, { "epoch": 0.9, "learning_rate": 4.7017259169446104e-07, "loss": 0.528, "step": 4703 }, { "epoch": 0.91, "learning_rate": 4.6828596835730487e-07, "loss": 0.5681, "step": 4704 }, { "epoch": 0.91, "learning_rate": 4.664030470280467e-07, "loss": 0.5845, "step": 4705 }, { "epoch": 0.91, "learning_rate": 4.645238284379883e-07, "loss": 0.4585, "step": 4706 }, { "epoch": 0.91, "learning_rate": 4.626483133169968e-07, "loss": 0.489, "step": 4707 }, { "epoch": 0.91, "learning_rate": 4.60776502393504e-07, "loss": 0.5786, "step": 4708 }, { "epoch": 0.91, "learning_rate": 4.5890839639449514e-07, "loss": 0.5658, "step": 4709 }, { "epoch": 0.91, "learning_rate": 4.5704399604552417e-07, "loss": 0.5683, "step": 4710 }, { "epoch": 0.91, "learning_rate": 4.551833020707008e-07, "loss": 0.5611, "step": 4711 }, { "epoch": 0.91, "learning_rate": 4.533263151926981e-07, "loss": 0.5986, "step": 4712 }, { "epoch": 0.91, "learning_rate": 4.514730361327502e-07, "loss": 0.517, "step": 4713 }, { "epoch": 0.91, "learning_rate": 4.4962346561064574e-07, "loss": 0.6004, "step": 4714 }, { "epoch": 0.91, "learning_rate": 4.4777760434473796e-07, "loss": 0.6156, "step": 4715 }, { "epoch": 0.91, "learning_rate": 4.4593545305193774e-07, "loss": 0.5486, "step": 4716 }, { "epoch": 0.91, "learning_rate": 4.440970124477173e-07, "loss": 0.6051, "step": 4717 }, { "epoch": 0.91, "learning_rate": 4.4226228324610544e-07, "loss": 0.5739, "step": 4718 }, { "epoch": 0.91, "learning_rate": 4.404312661596877e-07, "loss": 0.5426, "step": 4719 }, { "epoch": 0.91, "learning_rate": 4.386039618996119e-07, "loss": 0.5945, "step": 4720 }, { "epoch": 0.91, "learning_rate": 4.3678037117558135e-07, "loss": 0.5843, "step": 4721 }, { "epoch": 0.91, "learning_rate": 4.349604946958563e-07, "loss": 0.5915, "step": 4722 }, { "epoch": 0.91, "learning_rate": 4.331443331672591e-07, "loss": 0.444, "step": 4723 }, { "epoch": 0.91, "learning_rate": 4.313318872951633e-07, "loss": 0.5716, "step": 4724 }, { "epoch": 0.91, "learning_rate": 4.295231577835024e-07, "loss": 0.5885, "step": 4725 }, { "epoch": 0.91, "learning_rate": 4.2771814533476787e-07, "loss": 0.5733, "step": 4726 }, { "epoch": 0.91, "learning_rate": 4.2591685065000223e-07, "loss": 0.5375, "step": 4727 }, { "epoch": 0.91, "learning_rate": 4.241192744288092e-07, "loss": 0.5667, "step": 4728 }, { "epoch": 0.91, "learning_rate": 4.223254173693492e-07, "loss": 0.5601, "step": 4729 }, { "epoch": 0.91, "learning_rate": 4.2053528016833267e-07, "loss": 0.5267, "step": 4730 }, { "epoch": 0.91, "learning_rate": 4.1874886352103015e-07, "loss": 0.5872, "step": 4731 }, { "epoch": 0.91, "learning_rate": 4.1696616812126333e-07, "loss": 0.53, "step": 4732 }, { "epoch": 0.91, "learning_rate": 4.1518719466141165e-07, "loss": 0.5602, "step": 4733 }, { "epoch": 0.91, "learning_rate": 4.13411943832408e-07, "loss": 0.5825, "step": 4734 }, { "epoch": 0.91, "learning_rate": 4.116404163237386e-07, "loss": 0.5572, "step": 4735 }, { "epoch": 0.91, "learning_rate": 4.0987261282344425e-07, "loss": 0.5579, "step": 4736 }, { "epoch": 0.91, "learning_rate": 4.081085340181223e-07, "loss": 0.5465, "step": 4737 }, { "epoch": 0.91, "learning_rate": 4.06348180592917e-07, "loss": 0.5276, "step": 4738 }, { "epoch": 0.91, "learning_rate": 4.0459155323153034e-07, "loss": 0.5773, "step": 4739 }, { "epoch": 0.91, "learning_rate": 4.028386526162176e-07, "loss": 0.5956, "step": 4740 }, { "epoch": 0.91, "learning_rate": 4.010894794277831e-07, "loss": 0.5645, "step": 4741 }, { "epoch": 0.91, "learning_rate": 3.993440343455879e-07, "loss": 0.5805, "step": 4742 }, { "epoch": 0.91, "learning_rate": 3.976023180475397e-07, "loss": 0.5714, "step": 4743 }, { "epoch": 0.91, "learning_rate": 3.95864331210104e-07, "loss": 0.6002, "step": 4744 }, { "epoch": 0.91, "learning_rate": 3.941300745082932e-07, "loss": 0.6158, "step": 4745 }, { "epoch": 0.91, "learning_rate": 3.9239954861567177e-07, "loss": 0.5947, "step": 4746 }, { "epoch": 0.91, "learning_rate": 3.906727542043598e-07, "loss": 0.5532, "step": 4747 }, { "epoch": 0.91, "learning_rate": 3.8894969194502083e-07, "loss": 0.475, "step": 4748 }, { "epoch": 0.91, "learning_rate": 3.87230362506873e-07, "loss": 0.5424, "step": 4749 }, { "epoch": 0.91, "learning_rate": 3.8551476655768527e-07, "loss": 0.6011, "step": 4750 }, { "epoch": 0.91, "learning_rate": 3.8380290476377255e-07, "loss": 0.5379, "step": 4751 }, { "epoch": 0.91, "learning_rate": 3.8209477779000637e-07, "loss": 0.552, "step": 4752 }, { "epoch": 0.91, "learning_rate": 3.803903862998004e-07, "loss": 0.4762, "step": 4753 }, { "epoch": 0.91, "learning_rate": 3.7868973095512185e-07, "loss": 0.5851, "step": 4754 }, { "epoch": 0.91, "learning_rate": 3.7699281241648565e-07, "loss": 0.4765, "step": 4755 }, { "epoch": 0.92, "learning_rate": 3.7529963134295466e-07, "loss": 0.564, "step": 4756 }, { "epoch": 0.92, "learning_rate": 3.736101883921406e-07, "loss": 0.5721, "step": 4757 }, { "epoch": 0.92, "learning_rate": 3.719244842202074e-07, "loss": 0.3321, "step": 4758 }, { "epoch": 0.92, "learning_rate": 3.702425194818582e-07, "loss": 0.5694, "step": 4759 }, { "epoch": 0.92, "learning_rate": 3.685642948303503e-07, "loss": 0.5678, "step": 4760 }, { "epoch": 0.92, "learning_rate": 3.66889810917489e-07, "loss": 0.5697, "step": 4761 }, { "epoch": 0.92, "learning_rate": 3.6521906839362187e-07, "loss": 0.4932, "step": 4762 }, { "epoch": 0.92, "learning_rate": 3.635520679076465e-07, "loss": 0.4787, "step": 4763 }, { "epoch": 0.92, "learning_rate": 3.6188881010700725e-07, "loss": 0.6201, "step": 4764 }, { "epoch": 0.92, "learning_rate": 3.6022929563769513e-07, "loss": 0.5376, "step": 4765 }, { "epoch": 0.92, "learning_rate": 3.5857352514424573e-07, "loss": 0.4546, "step": 4766 }, { "epoch": 0.92, "learning_rate": 3.5692149926974006e-07, "loss": 0.5553, "step": 4767 }, { "epoch": 0.92, "learning_rate": 3.552732186558072e-07, "loss": 0.5215, "step": 4768 }, { "epoch": 0.92, "learning_rate": 3.536286839426195e-07, "loss": 0.5903, "step": 4769 }, { "epoch": 0.92, "learning_rate": 3.51987895768896e-07, "loss": 0.5093, "step": 4770 }, { "epoch": 0.92, "learning_rate": 3.5035085477190143e-07, "loss": 0.5989, "step": 4771 }, { "epoch": 0.92, "learning_rate": 3.4871756158744054e-07, "loss": 0.5446, "step": 4772 }, { "epoch": 0.92, "learning_rate": 3.4708801684986693e-07, "loss": 0.6071, "step": 4773 }, { "epoch": 0.92, "learning_rate": 3.454622211920766e-07, "loss": 0.5521, "step": 4774 }, { "epoch": 0.92, "learning_rate": 3.4384017524551116e-07, "loss": 0.586, "step": 4775 }, { "epoch": 0.92, "learning_rate": 3.422218796401544e-07, "loss": 0.542, "step": 4776 }, { "epoch": 0.92, "learning_rate": 3.4060733500453247e-07, "loss": 0.5927, "step": 4777 }, { "epoch": 0.92, "learning_rate": 3.38996541965716e-07, "loss": 0.5713, "step": 4778 }, { "epoch": 0.92, "learning_rate": 3.3738950114932e-07, "loss": 0.5308, "step": 4779 }, { "epoch": 0.92, "learning_rate": 3.3578621317949755e-07, "loss": 0.5308, "step": 4780 }, { "epoch": 0.92, "learning_rate": 3.341866786789505e-07, "loss": 0.546, "step": 4781 }, { "epoch": 0.92, "learning_rate": 3.325908982689185e-07, "loss": 0.5885, "step": 4782 }, { "epoch": 0.92, "learning_rate": 3.309988725691837e-07, "loss": 0.5705, "step": 4783 }, { "epoch": 0.92, "learning_rate": 3.294106021980714e-07, "loss": 0.565, "step": 4784 }, { "epoch": 0.92, "learning_rate": 3.278260877724471e-07, "loss": 0.5879, "step": 4785 }, { "epoch": 0.92, "learning_rate": 3.2624532990771507e-07, "loss": 0.4474, "step": 4786 }, { "epoch": 0.92, "learning_rate": 3.2466832921782986e-07, "loss": 0.5498, "step": 4787 }, { "epoch": 0.92, "learning_rate": 3.2309508631527486e-07, "loss": 0.5475, "step": 4788 }, { "epoch": 0.92, "learning_rate": 3.215256018110824e-07, "loss": 0.4619, "step": 4789 }, { "epoch": 0.92, "learning_rate": 3.199598763148215e-07, "loss": 0.4699, "step": 4790 }, { "epoch": 0.92, "learning_rate": 3.183979104346002e-07, "loss": 0.5596, "step": 4791 }, { "epoch": 0.92, "learning_rate": 3.1683970477706994e-07, "loss": 0.4533, "step": 4792 }, { "epoch": 0.92, "learning_rate": 3.1528525994741876e-07, "loss": 0.574, "step": 4793 }, { "epoch": 0.92, "learning_rate": 3.13734576549376e-07, "loss": 0.5822, "step": 4794 }, { "epoch": 0.92, "learning_rate": 3.121876551852099e-07, "loss": 0.4738, "step": 4795 }, { "epoch": 0.92, "learning_rate": 3.1064449645572536e-07, "loss": 0.4356, "step": 4796 }, { "epoch": 0.92, "learning_rate": 3.091051009602675e-07, "loss": 0.5664, "step": 4797 }, { "epoch": 0.92, "learning_rate": 3.0756946929672017e-07, "loss": 0.5574, "step": 4798 }, { "epoch": 0.92, "learning_rate": 3.060376020615052e-07, "loss": 0.5392, "step": 4799 }, { "epoch": 0.92, "learning_rate": 3.0450949984958347e-07, "loss": 0.4353, "step": 4800 }, { "epoch": 0.92, "learning_rate": 3.0298516325444893e-07, "loss": 0.5703, "step": 4801 }, { "epoch": 0.92, "learning_rate": 3.0146459286813924e-07, "loss": 0.5758, "step": 4802 }, { "epoch": 0.92, "learning_rate": 2.999477892812264e-07, "loss": 0.6025, "step": 4803 }, { "epoch": 0.92, "learning_rate": 2.984347530828158e-07, "loss": 0.6089, "step": 4804 }, { "epoch": 0.92, "learning_rate": 2.969254848605585e-07, "loss": 0.5351, "step": 4805 }, { "epoch": 0.92, "learning_rate": 2.9541998520063344e-07, "loss": 0.5625, "step": 4806 }, { "epoch": 0.92, "learning_rate": 2.9391825468775946e-07, "loss": 0.5432, "step": 4807 }, { "epoch": 0.93, "learning_rate": 2.9242029390519454e-07, "loss": 0.5857, "step": 4808 }, { "epoch": 0.93, "learning_rate": 2.909261034347255e-07, "loss": 0.3846, "step": 4809 }, { "epoch": 0.93, "learning_rate": 2.894356838566792e-07, "loss": 0.5493, "step": 4810 }, { "epoch": 0.93, "learning_rate": 2.879490357499204e-07, "loss": 0.5368, "step": 4811 }, { "epoch": 0.93, "learning_rate": 2.864661596918428e-07, "loss": 0.4546, "step": 4812 }, { "epoch": 0.93, "learning_rate": 2.849870562583812e-07, "loss": 0.4614, "step": 4813 }, { "epoch": 0.93, "learning_rate": 2.8351172602399945e-07, "loss": 0.579, "step": 4814 }, { "epoch": 0.93, "learning_rate": 2.8204016956169924e-07, "loss": 0.6141, "step": 4815 }, { "epoch": 0.93, "learning_rate": 2.8057238744301994e-07, "loss": 0.5631, "step": 4816 }, { "epoch": 0.93, "learning_rate": 2.7910838023802676e-07, "loss": 0.5303, "step": 4817 }, { "epoch": 0.93, "learning_rate": 2.7764814851532485e-07, "loss": 0.494, "step": 4818 }, { "epoch": 0.93, "learning_rate": 2.761916928420527e-07, "loss": 0.5594, "step": 4819 }, { "epoch": 0.93, "learning_rate": 2.74739013783879e-07, "loss": 0.5662, "step": 4820 }, { "epoch": 0.93, "learning_rate": 2.7329011190500797e-07, "loss": 0.55, "step": 4821 }, { "epoch": 0.93, "learning_rate": 2.7184498776817615e-07, "loss": 0.5168, "step": 4822 }, { "epoch": 0.93, "learning_rate": 2.704036419346534e-07, "loss": 0.5462, "step": 4823 }, { "epoch": 0.93, "learning_rate": 2.689660749642442e-07, "loss": 0.5583, "step": 4824 }, { "epoch": 0.93, "learning_rate": 2.675322874152786e-07, "loss": 0.5332, "step": 4825 }, { "epoch": 0.93, "learning_rate": 2.6610227984462556e-07, "loss": 0.5442, "step": 4826 }, { "epoch": 0.93, "learning_rate": 2.646760528076842e-07, "loss": 0.4558, "step": 4827 }, { "epoch": 0.93, "learning_rate": 2.6325360685838243e-07, "loss": 0.5576, "step": 4828 }, { "epoch": 0.93, "learning_rate": 2.618349425491851e-07, "loss": 0.57, "step": 4829 }, { "epoch": 0.93, "learning_rate": 2.604200604310825e-07, "loss": 0.5753, "step": 4830 }, { "epoch": 0.93, "learning_rate": 2.590089610535984e-07, "loss": 0.4617, "step": 4831 }, { "epoch": 0.93, "learning_rate": 2.5760164496479e-07, "loss": 0.4595, "step": 4832 }, { "epoch": 0.93, "learning_rate": 2.5619811271123897e-07, "loss": 0.5843, "step": 4833 }, { "epoch": 0.93, "learning_rate": 2.5479836483806586e-07, "loss": 0.5708, "step": 4834 }, { "epoch": 0.93, "learning_rate": 2.5340240188891143e-07, "loss": 0.5738, "step": 4835 }, { "epoch": 0.93, "learning_rate": 2.520102244059552e-07, "loss": 0.4815, "step": 4836 }, { "epoch": 0.93, "learning_rate": 2.506218329299026e-07, "loss": 0.4311, "step": 4837 }, { "epoch": 0.93, "learning_rate": 2.4923722799998664e-07, "loss": 0.5842, "step": 4838 }, { "epoch": 0.93, "learning_rate": 2.4785641015397375e-07, "loss": 0.5744, "step": 4839 }, { "epoch": 0.93, "learning_rate": 2.464793799281573e-07, "loss": 0.5209, "step": 4840 }, { "epoch": 0.93, "learning_rate": 2.4510613785735936e-07, "loss": 0.5919, "step": 4841 }, { "epoch": 0.93, "learning_rate": 2.4373668447493225e-07, "loss": 0.5591, "step": 4842 }, { "epoch": 0.93, "learning_rate": 2.423710203127561e-07, "loss": 0.4587, "step": 4843 }, { "epoch": 0.93, "learning_rate": 2.410091459012376e-07, "loss": 0.5784, "step": 4844 }, { "epoch": 0.93, "learning_rate": 2.3965106176931375e-07, "loss": 0.5366, "step": 4845 }, { "epoch": 0.93, "learning_rate": 2.3829676844444926e-07, "loss": 0.611, "step": 4846 }, { "epoch": 0.93, "learning_rate": 2.3694626645263675e-07, "loss": 0.4746, "step": 4847 }, { "epoch": 0.93, "learning_rate": 2.3559955631839436e-07, "loss": 0.554, "step": 4848 }, { "epoch": 0.93, "learning_rate": 2.3425663856476932e-07, "loss": 0.5647, "step": 4849 }, { "epoch": 0.93, "learning_rate": 2.3291751371333438e-07, "loss": 0.5537, "step": 4850 }, { "epoch": 0.93, "learning_rate": 2.3158218228419127e-07, "loss": 0.5582, "step": 4851 }, { "epoch": 0.93, "learning_rate": 2.3025064479596625e-07, "loss": 0.6211, "step": 4852 }, { "epoch": 0.93, "learning_rate": 2.2892290176581678e-07, "loss": 0.4486, "step": 4853 }, { "epoch": 0.93, "learning_rate": 2.2759895370941809e-07, "loss": 0.5408, "step": 4854 }, { "epoch": 0.93, "learning_rate": 2.2627880114097779e-07, "loss": 0.4469, "step": 4855 }, { "epoch": 0.93, "learning_rate": 2.2496244457323013e-07, "loss": 0.5654, "step": 4856 }, { "epoch": 0.93, "learning_rate": 2.2364988451742953e-07, "loss": 0.5451, "step": 4857 }, { "epoch": 0.93, "learning_rate": 2.2234112148336373e-07, "loss": 0.3666, "step": 4858 }, { "epoch": 0.93, "learning_rate": 2.2103615597933613e-07, "loss": 0.528, "step": 4859 }, { "epoch": 0.94, "learning_rate": 2.1973498851218244e-07, "loss": 0.5606, "step": 4860 }, { "epoch": 0.94, "learning_rate": 2.1843761958726283e-07, "loss": 0.4854, "step": 4861 }, { "epoch": 0.94, "learning_rate": 2.1714404970845647e-07, "loss": 0.5519, "step": 4862 }, { "epoch": 0.94, "learning_rate": 2.1585427937817594e-07, "loss": 0.4923, "step": 4863 }, { "epoch": 0.94, "learning_rate": 2.145683090973494e-07, "loss": 0.4777, "step": 4864 }, { "epoch": 0.94, "learning_rate": 2.1328613936543396e-07, "loss": 0.5192, "step": 4865 }, { "epoch": 0.94, "learning_rate": 2.1200777068041134e-07, "loss": 0.5413, "step": 4866 }, { "epoch": 0.94, "learning_rate": 2.1073320353878102e-07, "loss": 0.5555, "step": 4867 }, { "epoch": 0.94, "learning_rate": 2.0946243843557367e-07, "loss": 0.4817, "step": 4868 }, { "epoch": 0.94, "learning_rate": 2.0819547586434008e-07, "loss": 0.52, "step": 4869 }, { "epoch": 0.94, "learning_rate": 2.0693231631715105e-07, "loss": 0.464, "step": 4870 }, { "epoch": 0.94, "learning_rate": 2.0567296028460638e-07, "loss": 0.5735, "step": 4871 }, { "epoch": 0.94, "learning_rate": 2.0441740825582258e-07, "loss": 0.5939, "step": 4872 }, { "epoch": 0.94, "learning_rate": 2.0316566071844402e-07, "loss": 0.4409, "step": 4873 }, { "epoch": 0.94, "learning_rate": 2.0191771815863292e-07, "loss": 0.5276, "step": 4874 }, { "epoch": 0.94, "learning_rate": 2.0067358106107714e-07, "loss": 0.5762, "step": 4875 }, { "epoch": 0.94, "learning_rate": 1.994332499089846e-07, "loss": 0.4607, "step": 4876 }, { "epoch": 0.94, "learning_rate": 1.9819672518408662e-07, "loss": 0.553, "step": 4877 }, { "epoch": 0.94, "learning_rate": 1.9696400736663457e-07, "loss": 0.4725, "step": 4878 }, { "epoch": 0.94, "learning_rate": 1.9573509693540104e-07, "loss": 0.5846, "step": 4879 }, { "epoch": 0.94, "learning_rate": 1.9450999436768093e-07, "loss": 0.5808, "step": 4880 }, { "epoch": 0.94, "learning_rate": 1.9328870013929134e-07, "loss": 0.4759, "step": 4881 }, { "epoch": 0.94, "learning_rate": 1.9207121472456846e-07, "loss": 0.6031, "step": 4882 }, { "epoch": 0.94, "learning_rate": 1.9085753859636736e-07, "loss": 0.4354, "step": 4883 }, { "epoch": 0.94, "learning_rate": 1.8964767222606873e-07, "loss": 0.542, "step": 4884 }, { "epoch": 0.94, "learning_rate": 1.8844161608356782e-07, "loss": 0.5579, "step": 4885 }, { "epoch": 0.94, "learning_rate": 1.872393706372866e-07, "loss": 0.541, "step": 4886 }, { "epoch": 0.94, "learning_rate": 1.8604093635416155e-07, "loss": 0.5519, "step": 4887 }, { "epoch": 0.94, "learning_rate": 1.8484631369964922e-07, "loss": 0.4824, "step": 4888 }, { "epoch": 0.94, "learning_rate": 1.8365550313772852e-07, "loss": 0.5713, "step": 4889 }, { "epoch": 0.94, "learning_rate": 1.8246850513089832e-07, "loss": 0.5714, "step": 4890 }, { "epoch": 0.94, "learning_rate": 1.8128532014017098e-07, "loss": 0.52, "step": 4891 }, { "epoch": 0.94, "learning_rate": 1.8010594862508669e-07, "loss": 0.6153, "step": 4892 }, { "epoch": 0.94, "learning_rate": 1.789303910436968e-07, "loss": 0.5786, "step": 4893 }, { "epoch": 0.94, "learning_rate": 1.777586478525739e-07, "loss": 0.5966, "step": 4894 }, { "epoch": 0.94, "learning_rate": 1.7659071950681172e-07, "loss": 0.5586, "step": 4895 }, { "epoch": 0.94, "learning_rate": 1.754266064600174e-07, "loss": 0.5697, "step": 4896 }, { "epoch": 0.94, "learning_rate": 1.742663091643204e-07, "loss": 0.5604, "step": 4897 }, { "epoch": 0.94, "learning_rate": 1.7310982807036915e-07, "loss": 0.5866, "step": 4898 }, { "epoch": 0.94, "learning_rate": 1.719571636273243e-07, "loss": 0.5307, "step": 4899 }, { "epoch": 0.94, "learning_rate": 1.7080831628286886e-07, "loss": 0.5611, "step": 4900 }, { "epoch": 0.94, "learning_rate": 1.6966328648320152e-07, "loss": 0.5297, "step": 4901 }, { "epoch": 0.94, "learning_rate": 1.685220746730387e-07, "loss": 0.5305, "step": 4902 }, { "epoch": 0.94, "learning_rate": 1.673846812956137e-07, "loss": 0.5327, "step": 4903 }, { "epoch": 0.94, "learning_rate": 1.6625110679267642e-07, "loss": 0.4798, "step": 4904 }, { "epoch": 0.94, "learning_rate": 1.6512135160449583e-07, "loss": 0.4697, "step": 4905 }, { "epoch": 0.94, "learning_rate": 1.6399541616985648e-07, "loss": 0.568, "step": 4906 }, { "epoch": 0.94, "learning_rate": 1.6287330092605525e-07, "loss": 0.562, "step": 4907 }, { "epoch": 0.94, "learning_rate": 1.6175500630891128e-07, "loss": 0.5825, "step": 4908 }, { "epoch": 0.94, "learning_rate": 1.6064053275275716e-07, "loss": 0.5617, "step": 4909 }, { "epoch": 0.94, "learning_rate": 1.5952988069044105e-07, "loss": 0.618, "step": 4910 }, { "epoch": 0.94, "learning_rate": 1.5842305055332796e-07, "loss": 0.5338, "step": 4911 }, { "epoch": 0.95, "learning_rate": 1.573200427712973e-07, "loss": 0.5922, "step": 4912 }, { "epoch": 0.95, "learning_rate": 1.562208577727442e-07, "loss": 0.486, "step": 4913 }, { "epoch": 0.95, "learning_rate": 1.551254959845805e-07, "loss": 0.5082, "step": 4914 }, { "epoch": 0.95, "learning_rate": 1.540339578322314e-07, "loss": 0.4243, "step": 4915 }, { "epoch": 0.95, "learning_rate": 1.5294624373963894e-07, "loss": 0.5585, "step": 4916 }, { "epoch": 0.95, "learning_rate": 1.5186235412925744e-07, "loss": 0.5798, "step": 4917 }, { "epoch": 0.95, "learning_rate": 1.5078228942205674e-07, "loss": 0.5716, "step": 4918 }, { "epoch": 0.95, "learning_rate": 1.4970605003752359e-07, "loss": 0.4559, "step": 4919 }, { "epoch": 0.95, "learning_rate": 1.4863363639365357e-07, "loss": 0.5603, "step": 4920 }, { "epoch": 0.95, "learning_rate": 1.4756504890696466e-07, "loss": 0.5531, "step": 4921 }, { "epoch": 0.95, "learning_rate": 1.4650028799247928e-07, "loss": 0.5928, "step": 4922 }, { "epoch": 0.95, "learning_rate": 1.454393540637411e-07, "loss": 0.4877, "step": 4923 }, { "epoch": 0.95, "learning_rate": 1.4438224753280384e-07, "loss": 0.5288, "step": 4924 }, { "epoch": 0.95, "learning_rate": 1.4332896881023462e-07, "loss": 0.555, "step": 4925 }, { "epoch": 0.95, "learning_rate": 1.422795183051151e-07, "loss": 0.5731, "step": 4926 }, { "epoch": 0.95, "learning_rate": 1.4123389642504148e-07, "loss": 0.5601, "step": 4927 }, { "epoch": 0.95, "learning_rate": 1.401921035761189e-07, "loss": 0.5702, "step": 4928 }, { "epoch": 0.95, "learning_rate": 1.3915414016296925e-07, "loss": 0.5722, "step": 4929 }, { "epoch": 0.95, "learning_rate": 1.381200065887256e-07, "loss": 0.5475, "step": 4930 }, { "epoch": 0.95, "learning_rate": 1.3708970325503222e-07, "loss": 0.5791, "step": 4931 }, { "epoch": 0.95, "learning_rate": 1.3606323056204795e-07, "loss": 0.5443, "step": 4932 }, { "epoch": 0.95, "learning_rate": 1.3504058890844274e-07, "loss": 0.4389, "step": 4933 }, { "epoch": 0.95, "learning_rate": 1.3402177869139887e-07, "loss": 0.5908, "step": 4934 }, { "epoch": 0.95, "learning_rate": 1.3300680030661096e-07, "loss": 0.4538, "step": 4935 }, { "epoch": 0.95, "learning_rate": 1.3199565414828363e-07, "loss": 0.5727, "step": 4936 }, { "epoch": 0.95, "learning_rate": 1.3098834060913612e-07, "loss": 0.5582, "step": 4937 }, { "epoch": 0.95, "learning_rate": 1.2998486008039545e-07, "loss": 0.5562, "step": 4938 }, { "epoch": 0.95, "learning_rate": 1.289852129518032e-07, "loss": 0.5837, "step": 4939 }, { "epoch": 0.95, "learning_rate": 1.2798939961161217e-07, "loss": 0.4581, "step": 4940 }, { "epoch": 0.95, "learning_rate": 1.269974204465818e-07, "loss": 0.551, "step": 4941 }, { "epoch": 0.95, "learning_rate": 1.2600927584198618e-07, "loss": 0.5485, "step": 4942 }, { "epoch": 0.95, "learning_rate": 1.2502496618161165e-07, "loss": 0.6018, "step": 4943 }, { "epoch": 0.95, "learning_rate": 1.2404449184774904e-07, "loss": 0.6105, "step": 4944 }, { "epoch": 0.95, "learning_rate": 1.2306785322120596e-07, "loss": 0.5976, "step": 4945 }, { "epoch": 0.95, "learning_rate": 1.220950506812968e-07, "loss": 0.5848, "step": 4946 }, { "epoch": 0.95, "learning_rate": 1.2112608460584707e-07, "loss": 0.5384, "step": 4947 }, { "epoch": 0.95, "learning_rate": 1.2016095537119242e-07, "loss": 0.4476, "step": 4948 }, { "epoch": 0.95, "learning_rate": 1.1919966335217636e-07, "loss": 0.5635, "step": 4949 }, { "epoch": 0.95, "learning_rate": 1.1824220892215465e-07, "loss": 0.5542, "step": 4950 }, { "epoch": 0.95, "learning_rate": 1.1728859245299207e-07, "loss": 0.4341, "step": 4951 }, { "epoch": 0.95, "learning_rate": 1.1633881431506122e-07, "loss": 0.5795, "step": 4952 }, { "epoch": 0.95, "learning_rate": 1.1539287487724594e-07, "loss": 0.5797, "step": 4953 }, { "epoch": 0.95, "learning_rate": 1.1445077450693786e-07, "loss": 0.4544, "step": 4954 }, { "epoch": 0.95, "learning_rate": 1.1351251357003656e-07, "loss": 0.5571, "step": 4955 }, { "epoch": 0.95, "learning_rate": 1.1257809243095385e-07, "loss": 0.545, "step": 4956 }, { "epoch": 0.95, "learning_rate": 1.1164751145260722e-07, "loss": 0.568, "step": 4957 }, { "epoch": 0.95, "learning_rate": 1.107207709964242e-07, "loss": 0.6013, "step": 4958 }, { "epoch": 0.95, "learning_rate": 1.0979787142233911e-07, "loss": 0.4612, "step": 4959 }, { "epoch": 0.95, "learning_rate": 1.0887881308879633e-07, "loss": 0.5945, "step": 4960 }, { "epoch": 0.95, "learning_rate": 1.0796359635274701e-07, "loss": 0.4728, "step": 4961 }, { "epoch": 0.95, "learning_rate": 1.0705222156965011e-07, "loss": 0.5896, "step": 4962 }, { "epoch": 0.95, "learning_rate": 1.0614468909347476e-07, "loss": 0.5987, "step": 4963 }, { "epoch": 0.96, "learning_rate": 1.0524099927669563e-07, "loss": 0.4491, "step": 4964 }, { "epoch": 0.96, "learning_rate": 1.0434115247029419e-07, "loss": 0.5698, "step": 4965 }, { "epoch": 0.96, "learning_rate": 1.0344514902376201e-07, "loss": 0.4962, "step": 4966 }, { "epoch": 0.96, "learning_rate": 1.0255298928509627e-07, "loss": 0.57, "step": 4967 }, { "epoch": 0.96, "learning_rate": 1.0166467360079979e-07, "loss": 0.5, "step": 4968 }, { "epoch": 0.96, "learning_rate": 1.007802023158877e-07, "loss": 0.3969, "step": 4969 }, { "epoch": 0.96, "learning_rate": 9.989957577387521e-08, "loss": 0.5727, "step": 4970 }, { "epoch": 0.96, "learning_rate": 9.902279431678874e-08, "loss": 0.5812, "step": 4971 }, { "epoch": 0.96, "learning_rate": 9.814985828516033e-08, "loss": 0.3473, "step": 4972 }, { "epoch": 0.96, "learning_rate": 9.728076801802656e-08, "loss": 0.5714, "step": 4973 }, { "epoch": 0.96, "learning_rate": 9.641552385293518e-08, "loss": 0.567, "step": 4974 }, { "epoch": 0.96, "learning_rate": 9.555412612593518e-08, "loss": 0.6231, "step": 4975 }, { "epoch": 0.96, "learning_rate": 9.469657517158226e-08, "loss": 0.4423, "step": 4976 }, { "epoch": 0.96, "learning_rate": 9.384287132294223e-08, "loss": 0.5602, "step": 4977 }, { "epoch": 0.96, "learning_rate": 9.299301491158207e-08, "loss": 0.5484, "step": 4978 }, { "epoch": 0.96, "learning_rate": 9.214700626757667e-08, "loss": 0.581, "step": 4979 }, { "epoch": 0.96, "learning_rate": 9.130484571950538e-08, "loss": 0.5775, "step": 4980 }, { "epoch": 0.96, "learning_rate": 9.046653359445323e-08, "loss": 0.4666, "step": 4981 }, { "epoch": 0.96, "learning_rate": 8.963207021801423e-08, "loss": 0.5714, "step": 4982 }, { "epoch": 0.96, "learning_rate": 8.880145591428024e-08, "loss": 0.5864, "step": 4983 }, { "epoch": 0.96, "learning_rate": 8.797469100585432e-08, "loss": 0.5961, "step": 4984 }, { "epoch": 0.96, "learning_rate": 8.715177581384182e-08, "loss": 0.5807, "step": 4985 }, { "epoch": 0.96, "learning_rate": 8.633271065785486e-08, "loss": 0.5437, "step": 4986 }, { "epoch": 0.96, "learning_rate": 8.551749585600678e-08, "loss": 0.5417, "step": 4987 }, { "epoch": 0.96, "learning_rate": 8.470613172491981e-08, "loss": 0.5635, "step": 4988 }, { "epoch": 0.96, "learning_rate": 8.389861857971748e-08, "loss": 0.6033, "step": 4989 }, { "epoch": 0.96, "learning_rate": 8.309495673402778e-08, "loss": 0.598, "step": 4990 }, { "epoch": 0.96, "learning_rate": 8.229514649998438e-08, "loss": 0.5319, "step": 4991 }, { "epoch": 0.96, "learning_rate": 8.149918818822433e-08, "loss": 0.5861, "step": 4992 }, { "epoch": 0.96, "learning_rate": 8.070708210788925e-08, "loss": 0.5113, "step": 4993 }, { "epoch": 0.96, "learning_rate": 7.991882856662303e-08, "loss": 0.5392, "step": 4994 }, { "epoch": 0.96, "learning_rate": 7.913442787057523e-08, "loss": 0.5573, "step": 4995 }, { "epoch": 0.96, "learning_rate": 7.835388032439661e-08, "loss": 0.4956, "step": 4996 }, { "epoch": 0.96, "learning_rate": 7.757718623124466e-08, "loss": 0.5578, "step": 4997 }, { "epoch": 0.96, "learning_rate": 7.680434589277696e-08, "loss": 0.6162, "step": 4998 }, { "epoch": 0.96, "learning_rate": 7.603535960915675e-08, "loss": 0.5544, "step": 4999 }, { "epoch": 0.96, "learning_rate": 7.527022767904957e-08, "loss": 0.5714, "step": 5000 }, { "epoch": 0.96, "learning_rate": 7.450895039962214e-08, "loss": 0.5179, "step": 5001 }, { "epoch": 0.96, "learning_rate": 7.375152806654685e-08, "loss": 0.595, "step": 5002 }, { "epoch": 0.96, "learning_rate": 7.299796097399947e-08, "loss": 0.4951, "step": 5003 }, { "epoch": 0.96, "learning_rate": 7.224824941465369e-08, "loss": 0.5844, "step": 5004 }, { "epoch": 0.96, "learning_rate": 7.150239367969102e-08, "loss": 0.5104, "step": 5005 }, { "epoch": 0.96, "learning_rate": 7.076039405879309e-08, "loss": 0.5253, "step": 5006 }, { "epoch": 0.96, "learning_rate": 7.002225084014269e-08, "loss": 0.5123, "step": 5007 }, { "epoch": 0.96, "learning_rate": 6.928796431042717e-08, "loss": 0.5594, "step": 5008 }, { "epoch": 0.96, "learning_rate": 6.855753475483507e-08, "loss": 0.3575, "step": 5009 }, { "epoch": 0.96, "learning_rate": 6.783096245705612e-08, "loss": 0.5502, "step": 5010 }, { "epoch": 0.96, "learning_rate": 6.710824769928349e-08, "loss": 0.5526, "step": 5011 }, { "epoch": 0.96, "learning_rate": 6.63893907622104e-08, "loss": 0.4594, "step": 5012 }, { "epoch": 0.96, "learning_rate": 6.567439192503244e-08, "loss": 0.5669, "step": 5013 }, { "epoch": 0.96, "learning_rate": 6.496325146544746e-08, "loss": 0.5926, "step": 5014 }, { "epoch": 0.96, "learning_rate": 6.425596965965453e-08, "loss": 0.5542, "step": 5015 }, { "epoch": 0.97, "learning_rate": 6.35525467823539e-08, "loss": 0.5333, "step": 5016 }, { "epoch": 0.97, "learning_rate": 6.285298310674703e-08, "loss": 0.5433, "step": 5017 }, { "epoch": 0.97, "learning_rate": 6.215727890453438e-08, "loss": 0.5513, "step": 5018 }, { "epoch": 0.97, "learning_rate": 6.14654344459209e-08, "loss": 0.5498, "step": 5019 }, { "epoch": 0.97, "learning_rate": 6.077744999961166e-08, "loss": 0.5673, "step": 5020 }, { "epoch": 0.97, "learning_rate": 6.009332583281069e-08, "loss": 0.4717, "step": 5021 }, { "epoch": 0.97, "learning_rate": 5.941306221122545e-08, "loss": 0.5643, "step": 5022 }, { "epoch": 0.97, "learning_rate": 5.873665939906015e-08, "loss": 0.5329, "step": 5023 }, { "epoch": 0.97, "learning_rate": 5.806411765902353e-08, "loss": 0.5643, "step": 5024 }, { "epoch": 0.97, "learning_rate": 5.739543725232333e-08, "loss": 0.5567, "step": 5025 }, { "epoch": 0.97, "learning_rate": 5.673061843866623e-08, "loss": 0.496, "step": 5026 }, { "epoch": 0.97, "learning_rate": 5.606966147626125e-08, "loss": 0.5313, "step": 5027 }, { "epoch": 0.97, "learning_rate": 5.5412566621815266e-08, "loss": 0.5512, "step": 5028 }, { "epoch": 0.97, "learning_rate": 5.4759334130536355e-08, "loss": 0.5694, "step": 5029 }, { "epoch": 0.97, "learning_rate": 5.41099642561338e-08, "loss": 0.5143, "step": 5030 }, { "epoch": 0.97, "learning_rate": 5.3464457250814726e-08, "loss": 0.5405, "step": 5031 }, { "epoch": 0.97, "learning_rate": 5.2822813365286386e-08, "loss": 0.57, "step": 5032 }, { "epoch": 0.97, "learning_rate": 5.218503284875609e-08, "loss": 0.4591, "step": 5033 }, { "epoch": 0.97, "learning_rate": 5.155111594893014e-08, "loss": 0.5926, "step": 5034 }, { "epoch": 0.97, "learning_rate": 5.092106291201604e-08, "loss": 0.4633, "step": 5035 }, { "epoch": 0.97, "learning_rate": 5.029487398271693e-08, "loss": 0.3965, "step": 5036 }, { "epoch": 0.97, "learning_rate": 4.967254940423716e-08, "loss": 0.551, "step": 5037 }, { "epoch": 0.97, "learning_rate": 4.905408941828338e-08, "loss": 0.6023, "step": 5038 }, { "epoch": 0.97, "learning_rate": 4.8439494265055674e-08, "loss": 0.5154, "step": 5039 }, { "epoch": 0.97, "learning_rate": 4.7828764183257545e-08, "loss": 0.5432, "step": 5040 }, { "epoch": 0.97, "learning_rate": 4.722189941008703e-08, "loss": 0.5446, "step": 5041 }, { "epoch": 0.97, "learning_rate": 4.66189001812456e-08, "loss": 0.5495, "step": 5042 }, { "epoch": 0.97, "learning_rate": 4.6019766730930336e-08, "loss": 0.5865, "step": 5043 }, { "epoch": 0.97, "learning_rate": 4.542449929183845e-08, "loss": 0.5763, "step": 5044 }, { "epoch": 0.97, "learning_rate": 4.4833098095165005e-08, "loss": 0.5659, "step": 5045 }, { "epoch": 0.97, "learning_rate": 4.424556337060182e-08, "loss": 0.5412, "step": 5046 }, { "epoch": 0.97, "learning_rate": 4.366189534634191e-08, "loss": 0.6151, "step": 5047 }, { "epoch": 0.97, "learning_rate": 4.308209424907506e-08, "loss": 0.5756, "step": 5048 }, { "epoch": 0.97, "learning_rate": 4.250616030399002e-08, "loss": 0.5638, "step": 5049 }, { "epoch": 0.97, "learning_rate": 4.1934093734771194e-08, "loss": 0.5495, "step": 5050 }, { "epoch": 0.97, "learning_rate": 4.13658947636042e-08, "loss": 0.4007, "step": 5051 }, { "epoch": 0.97, "learning_rate": 4.080156361117027e-08, "loss": 0.547, "step": 5052 }, { "epoch": 0.97, "learning_rate": 4.024110049664853e-08, "loss": 0.6088, "step": 5053 }, { "epoch": 0.97, "learning_rate": 3.9684505637718194e-08, "loss": 0.5399, "step": 5054 }, { "epoch": 0.97, "learning_rate": 3.913177925055189e-08, "loss": 0.5346, "step": 5055 }, { "epoch": 0.97, "learning_rate": 3.858292154982457e-08, "loss": 0.5438, "step": 5056 }, { "epoch": 0.97, "learning_rate": 3.8037932748704596e-08, "loss": 0.4357, "step": 5057 }, { "epoch": 0.97, "learning_rate": 3.7496813058859325e-08, "loss": 0.5387, "step": 5058 }, { "epoch": 0.97, "learning_rate": 3.695956269045509e-08, "loss": 0.6171, "step": 5059 }, { "epoch": 0.97, "learning_rate": 3.642618185215163e-08, "loss": 0.5781, "step": 5060 }, { "epoch": 0.97, "learning_rate": 3.589667075110992e-08, "loss": 0.6121, "step": 5061 }, { "epoch": 0.97, "learning_rate": 3.537102959298322e-08, "loss": 0.5571, "step": 5062 }, { "epoch": 0.97, "learning_rate": 3.4849258581928224e-08, "loss": 0.532, "step": 5063 }, { "epoch": 0.97, "learning_rate": 3.4331357920591726e-08, "loss": 0.4028, "step": 5064 }, { "epoch": 0.97, "learning_rate": 3.3817327810121706e-08, "loss": 0.5519, "step": 5065 }, { "epoch": 0.97, "learning_rate": 3.3307168450160685e-08, "loss": 0.5605, "step": 5066 }, { "epoch": 0.97, "learning_rate": 3.280088003885018e-08, "loss": 0.5657, "step": 5067 }, { "epoch": 0.98, "learning_rate": 3.229846277282511e-08, "loss": 0.4876, "step": 5068 }, { "epoch": 0.98, "learning_rate": 3.1799916847220505e-08, "loss": 0.5986, "step": 5069 }, { "epoch": 0.98, "learning_rate": 3.130524245566369e-08, "loss": 0.5477, "step": 5070 }, { "epoch": 0.98, "learning_rate": 3.0814439790280984e-08, "loss": 0.5999, "step": 5071 }, { "epoch": 0.98, "learning_rate": 3.032750904169546e-08, "loss": 0.5249, "step": 5072 }, { "epoch": 0.98, "learning_rate": 2.9844450399024725e-08, "loss": 0.4899, "step": 5073 }, { "epoch": 0.98, "learning_rate": 2.936526404988427e-08, "loss": 0.3783, "step": 5074 }, { "epoch": 0.98, "learning_rate": 2.8889950180382985e-08, "loss": 0.4771, "step": 5075 }, { "epoch": 0.98, "learning_rate": 2.8418508975127655e-08, "loss": 0.6038, "step": 5076 }, { "epoch": 0.98, "learning_rate": 2.7950940617221812e-08, "loss": 0.5755, "step": 5077 }, { "epoch": 0.98, "learning_rate": 2.7487245288261298e-08, "loss": 0.4384, "step": 5078 }, { "epoch": 0.98, "learning_rate": 2.702742316834206e-08, "loss": 0.5892, "step": 5079 }, { "epoch": 0.98, "learning_rate": 2.657147443605457e-08, "loss": 0.5842, "step": 5080 }, { "epoch": 0.98, "learning_rate": 2.6119399268480505e-08, "loss": 0.5456, "step": 5081 }, { "epoch": 0.98, "learning_rate": 2.5671197841203867e-08, "loss": 0.585, "step": 5082 }, { "epoch": 0.98, "learning_rate": 2.5226870328299845e-08, "loss": 0.5975, "step": 5083 }, { "epoch": 0.98, "learning_rate": 2.4786416902339293e-08, "loss": 0.5675, "step": 5084 }, { "epoch": 0.98, "learning_rate": 2.4349837734390923e-08, "loss": 0.5511, "step": 5085 }, { "epoch": 0.98, "learning_rate": 2.3917132994016878e-08, "loss": 0.4564, "step": 5086 }, { "epoch": 0.98, "learning_rate": 2.3488302849272726e-08, "loss": 0.5359, "step": 5087 }, { "epoch": 0.98, "learning_rate": 2.3063347466713014e-08, "loss": 0.5697, "step": 5088 }, { "epoch": 0.98, "learning_rate": 2.264226701138461e-08, "loss": 0.631, "step": 5089 }, { "epoch": 0.98, "learning_rate": 2.2225061646830025e-08, "loss": 0.5564, "step": 5090 }, { "epoch": 0.98, "learning_rate": 2.181173153508853e-08, "loss": 0.4514, "step": 5091 }, { "epoch": 0.98, "learning_rate": 2.1402276836691715e-08, "loss": 0.5865, "step": 5092 }, { "epoch": 0.98, "learning_rate": 2.0996697710666812e-08, "loss": 0.5255, "step": 5093 }, { "epoch": 0.98, "learning_rate": 2.059499431453671e-08, "loss": 0.5915, "step": 5094 }, { "epoch": 0.98, "learning_rate": 2.0197166804317713e-08, "loss": 0.5614, "step": 5095 }, { "epoch": 0.98, "learning_rate": 1.9803215334522895e-08, "loss": 0.4755, "step": 5096 }, { "epoch": 0.98, "learning_rate": 1.9413140058156533e-08, "loss": 0.4594, "step": 5097 }, { "epoch": 0.98, "learning_rate": 1.9026941126721876e-08, "loss": 0.5333, "step": 5098 }, { "epoch": 0.98, "learning_rate": 1.864461869021117e-08, "loss": 0.5074, "step": 5099 }, { "epoch": 0.98, "learning_rate": 1.826617289711563e-08, "loss": 0.5691, "step": 5100 }, { "epoch": 0.98, "learning_rate": 1.7891603894418797e-08, "loss": 0.5748, "step": 5101 }, { "epoch": 0.98, "learning_rate": 1.7520911827598742e-08, "loss": 0.5665, "step": 5102 }, { "epoch": 0.98, "learning_rate": 1.7154096840629186e-08, "loss": 0.5856, "step": 5103 }, { "epoch": 0.98, "learning_rate": 1.679115907597617e-08, "loss": 0.4299, "step": 5104 }, { "epoch": 0.98, "learning_rate": 1.6432098674600272e-08, "loss": 0.5204, "step": 5105 }, { "epoch": 0.98, "learning_rate": 1.6076915775956604e-08, "loss": 0.4549, "step": 5106 }, { "epoch": 0.98, "learning_rate": 1.5725610517994816e-08, "loss": 0.5423, "step": 5107 }, { "epoch": 0.98, "learning_rate": 1.537818303715688e-08, "loss": 0.387, "step": 5108 }, { "epoch": 0.98, "learning_rate": 1.50346334683793e-08, "loss": 0.5244, "step": 5109 }, { "epoch": 0.98, "learning_rate": 1.4694961945093122e-08, "loss": 0.534, "step": 5110 }, { "epoch": 0.98, "learning_rate": 1.4359168599223926e-08, "loss": 0.5398, "step": 5111 }, { "epoch": 0.98, "learning_rate": 1.4027253561188502e-08, "loss": 0.5647, "step": 5112 }, { "epoch": 0.98, "learning_rate": 1.3699216959899287e-08, "loss": 0.5601, "step": 5113 }, { "epoch": 0.98, "learning_rate": 1.337505892276325e-08, "loss": 0.5934, "step": 5114 }, { "epoch": 0.98, "learning_rate": 1.3054779575677468e-08, "loss": 0.4897, "step": 5115 }, { "epoch": 0.98, "learning_rate": 1.2738379043035765e-08, "loss": 0.576, "step": 5116 }, { "epoch": 0.98, "learning_rate": 1.2425857447725397e-08, "loss": 0.4784, "step": 5117 }, { "epoch": 0.98, "learning_rate": 1.211721491112372e-08, "loss": 0.5653, "step": 5118 }, { "epoch": 0.98, "learning_rate": 1.1812451553107063e-08, "loss": 0.5974, "step": 5119 }, { "epoch": 0.99, "learning_rate": 1.1511567492038522e-08, "loss": 0.4177, "step": 5120 }, { "epoch": 0.99, "learning_rate": 1.1214562844781285e-08, "loss": 0.5944, "step": 5121 }, { "epoch": 0.99, "learning_rate": 1.0921437726686413e-08, "loss": 0.5518, "step": 5122 }, { "epoch": 0.99, "learning_rate": 1.0632192251601726e-08, "loss": 0.5421, "step": 5123 }, { "epoch": 0.99, "learning_rate": 1.0346826531865139e-08, "loss": 0.4526, "step": 5124 }, { "epoch": 0.99, "learning_rate": 1.006534067831022e-08, "loss": 0.5217, "step": 5125 }, { "epoch": 0.99, "learning_rate": 9.78773480026396e-09, "loss": 0.3514, "step": 5126 }, { "epoch": 0.99, "learning_rate": 9.514009005543445e-09, "loss": 0.5349, "step": 5127 }, { "epoch": 0.99, "learning_rate": 9.244163400462525e-09, "loss": 0.3716, "step": 5128 }, { "epoch": 0.99, "learning_rate": 8.97819808982403e-09, "loss": 0.5377, "step": 5129 }, { "epoch": 0.99, "learning_rate": 8.716113176927554e-09, "loss": 0.5433, "step": 5130 }, { "epoch": 0.99, "learning_rate": 8.457908763562783e-09, "loss": 0.4467, "step": 5131 }, { "epoch": 0.99, "learning_rate": 8.203584950013942e-09, "loss": 0.5461, "step": 5132 }, { "epoch": 0.99, "learning_rate": 7.953141835057576e-09, "loss": 0.5543, "step": 5133 }, { "epoch": 0.99, "learning_rate": 7.706579515962542e-09, "loss": 0.5774, "step": 5134 }, { "epoch": 0.99, "learning_rate": 7.463898088490019e-09, "loss": 0.4214, "step": 5135 }, { "epoch": 0.99, "learning_rate": 7.225097646895718e-09, "loss": 0.5717, "step": 5136 }, { "epoch": 0.99, "learning_rate": 6.990178283927673e-09, "loss": 0.4919, "step": 5137 }, { "epoch": 0.99, "learning_rate": 6.759140090824012e-09, "loss": 0.4772, "step": 5138 }, { "epoch": 0.99, "learning_rate": 6.531983157318511e-09, "loss": 0.5396, "step": 5139 }, { "epoch": 0.99, "learning_rate": 6.308707571636152e-09, "loss": 0.4773, "step": 5140 }, { "epoch": 0.99, "learning_rate": 6.089313420494236e-09, "loss": 0.5469, "step": 5141 }, { "epoch": 0.99, "learning_rate": 5.87380078910349e-09, "loss": 0.5554, "step": 5142 }, { "epoch": 0.99, "learning_rate": 5.662169761165848e-09, "loss": 0.5509, "step": 5143 }, { "epoch": 0.99, "learning_rate": 5.4544204188777815e-09, "loss": 0.6112, "step": 5144 }, { "epoch": 0.99, "learning_rate": 5.25055284292475e-09, "loss": 0.5704, "step": 5145 }, { "epoch": 0.99, "learning_rate": 5.050567112487858e-09, "loss": 0.5874, "step": 5146 }, { "epoch": 0.99, "learning_rate": 4.85446330523831e-09, "loss": 0.5903, "step": 5147 }, { "epoch": 0.99, "learning_rate": 4.662241497341846e-09, "loss": 0.5643, "step": 5148 }, { "epoch": 0.99, "learning_rate": 4.473901763454302e-09, "loss": 0.4439, "step": 5149 }, { "epoch": 0.99, "learning_rate": 4.289444176724944e-09, "loss": 0.5832, "step": 5150 }, { "epoch": 0.99, "learning_rate": 4.108868808794242e-09, "loss": 0.5883, "step": 5151 }, { "epoch": 0.99, "learning_rate": 3.932175729797205e-09, "loss": 0.5725, "step": 5152 }, { "epoch": 0.99, "learning_rate": 3.759365008357829e-09, "loss": 0.5806, "step": 5153 }, { "epoch": 0.99, "learning_rate": 3.590436711594647e-09, "loss": 0.5392, "step": 5154 }, { "epoch": 0.99, "learning_rate": 3.4253909051173985e-09, "loss": 0.5682, "step": 5155 }, { "epoch": 0.99, "learning_rate": 3.2642276530281404e-09, "loss": 0.5972, "step": 5156 }, { "epoch": 0.99, "learning_rate": 3.1069470179201365e-09, "loss": 0.5411, "step": 5157 }, { "epoch": 0.99, "learning_rate": 2.9535490608789687e-09, "loss": 0.4614, "step": 5158 }, { "epoch": 0.99, "learning_rate": 2.8040338414847545e-09, "loss": 0.5527, "step": 5159 }, { "epoch": 0.99, "learning_rate": 2.6584014178054894e-09, "loss": 0.5503, "step": 5160 }, { "epoch": 0.99, "learning_rate": 2.516651846403706e-09, "loss": 0.5719, "step": 5161 }, { "epoch": 0.99, "learning_rate": 2.378785182333143e-09, "loss": 0.4909, "step": 5162 }, { "epoch": 0.99, "learning_rate": 2.2448014791398574e-09, "loss": 0.492, "step": 5163 }, { "epoch": 0.99, "learning_rate": 2.1147007888622227e-09, "loss": 0.5629, "step": 5164 }, { "epoch": 0.99, "learning_rate": 1.9884831620287094e-09, "loss": 0.5403, "step": 5165 }, { "epoch": 0.99, "learning_rate": 1.8661486476612144e-09, "loss": 0.5986, "step": 5166 }, { "epoch": 0.99, "learning_rate": 1.747697293272843e-09, "loss": 0.4587, "step": 5167 }, { "epoch": 0.99, "learning_rate": 1.633129144870127e-09, "loss": 0.5259, "step": 5168 }, { "epoch": 0.99, "learning_rate": 1.5224442469474743e-09, "loss": 0.5399, "step": 5169 }, { "epoch": 0.99, "learning_rate": 1.4156426424960513e-09, "loss": 0.5652, "step": 5170 }, { "epoch": 0.99, "learning_rate": 1.3127243729949003e-09, "loss": 0.5677, "step": 5171 }, { "epoch": 1.0, "learning_rate": 1.2136894784176011e-09, "loss": 0.5788, "step": 5172 }, { "epoch": 1.0, "learning_rate": 1.1185379972256105e-09, "loss": 0.5186, "step": 5173 }, { "epoch": 1.0, "learning_rate": 1.0272699663782525e-09, "loss": 0.5597, "step": 5174 }, { "epoch": 1.0, "learning_rate": 9.398854213193976e-10, "loss": 0.5908, "step": 5175 }, { "epoch": 1.0, "learning_rate": 8.563843959907836e-10, "loss": 0.5416, "step": 5176 }, { "epoch": 1.0, "learning_rate": 7.767669228231356e-10, "loss": 0.4838, "step": 5177 }, { "epoch": 1.0, "learning_rate": 7.010330327372749e-10, "loss": 0.4995, "step": 5178 }, { "epoch": 1.0, "learning_rate": 6.291827551474505e-10, "loss": 0.5284, "step": 5179 }, { "epoch": 1.0, "learning_rate": 5.612161179613385e-10, "loss": 0.4694, "step": 5180 }, { "epoch": 1.0, "learning_rate": 4.971331475756013e-10, "loss": 0.5267, "step": 5181 }, { "epoch": 1.0, "learning_rate": 4.369338688781088e-10, "loss": 0.5283, "step": 5182 }, { "epoch": 1.0, "learning_rate": 3.8061830525126797e-10, "loss": 0.5639, "step": 5183 }, { "epoch": 1.0, "learning_rate": 3.2818647856647234e-10, "loss": 0.5808, "step": 5184 }, { "epoch": 1.0, "learning_rate": 2.796384091885429e-10, "loss": 0.584, "step": 5185 }, { "epoch": 1.0, "learning_rate": 2.3497411597128705e-10, "loss": 0.5782, "step": 5186 }, { "epoch": 1.0, "learning_rate": 1.9419361626416e-10, "loss": 0.5625, "step": 5187 }, { "epoch": 1.0, "learning_rate": 1.5729692590338296e-10, "loss": 0.5057, "step": 5188 }, { "epoch": 1.0, "learning_rate": 1.2428405921971476e-10, "loss": 0.5492, "step": 5189 }, { "epoch": 1.0, "learning_rate": 9.515502903734153e-11, "loss": 0.5432, "step": 5190 }, { "epoch": 1.0, "learning_rate": 6.990984666610523e-11, "loss": 0.4651, "step": 5191 }, { "epoch": 1.0, "learning_rate": 4.854852191371606e-11, "loss": 0.5835, "step": 5192 }, { "epoch": 1.0, "learning_rate": 3.1071063075760466e-11, "loss": 0.5448, "step": 5193 }, { "epoch": 1.0, "learning_rate": 1.7477476940142013e-11, "loss": 0.5775, "step": 5194 }, { "epoch": 1.0, "learning_rate": 7.767768785971185e-12, "loss": 0.5749, "step": 5195 }, { "epoch": 1.0, "learning_rate": 1.9419423846755993e-12, "loss": 0.5898, "step": 5196 }, { "epoch": 1.0, "learning_rate": 0.0, "loss": 0.5901, "step": 5197 }, { "epoch": 1.0, "step": 5197, "total_flos": 4.698809544841626e+16, "train_loss": 0.5654000673519778, "train_runtime": 62319.4225, "train_samples_per_second": 10.676, "train_steps_per_second": 0.083 } ], "max_steps": 5197, "num_train_epochs": 1, "total_flos": 4.698809544841626e+16, "trial_name": null, "trial_params": null }