{ "best_metric": null, "best_model_checkpoint": null, "epoch": 5.0, "global_step": 2800, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 2.3809523809523814e-07, "loss": 4.2266, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.761904761904763e-07, "loss": 4.5547, "step": 2 }, { "epoch": 0.01, "learning_rate": 7.142857142857143e-07, "loss": 4.457, "step": 3 }, { "epoch": 0.01, "learning_rate": 9.523809523809526e-07, "loss": 4.6562, "step": 4 }, { "epoch": 0.01, "learning_rate": 1.1904761904761904e-06, "loss": 5.4297, "step": 5 }, { "epoch": 0.01, "learning_rate": 1.4285714285714286e-06, "loss": 4.3594, "step": 6 }, { "epoch": 0.01, "learning_rate": 1.6666666666666667e-06, "loss": 4.582, "step": 7 }, { "epoch": 0.01, "learning_rate": 1.9047619047619051e-06, "loss": 4.7734, "step": 8 }, { "epoch": 0.02, "learning_rate": 2.142857142857143e-06, "loss": 4.5117, "step": 9 }, { "epoch": 0.02, "learning_rate": 2.3809523809523808e-06, "loss": 5.0312, "step": 10 }, { "epoch": 0.02, "learning_rate": 2.6190476190476192e-06, "loss": 4.5664, "step": 11 }, { "epoch": 0.02, "learning_rate": 2.8571428571428573e-06, "loss": 4.6328, "step": 12 }, { "epoch": 0.02, "learning_rate": 3.0952380952380953e-06, "loss": 4.4766, "step": 13 }, { "epoch": 0.03, "learning_rate": 3.3333333333333333e-06, "loss": 4.1719, "step": 14 }, { "epoch": 0.03, "learning_rate": 3.5714285714285714e-06, "loss": 4.8594, "step": 15 }, { "epoch": 0.03, "learning_rate": 3.8095238095238102e-06, "loss": 4.418, "step": 16 }, { "epoch": 0.03, "learning_rate": 4.047619047619048e-06, "loss": 4.2383, "step": 17 }, { "epoch": 0.03, "learning_rate": 4.285714285714286e-06, "loss": 4.5977, "step": 18 }, { "epoch": 0.03, "learning_rate": 4.5238095238095235e-06, "loss": 4.3398, "step": 19 }, { "epoch": 0.04, "learning_rate": 4.7619047619047615e-06, "loss": 4.0938, "step": 20 }, { "epoch": 0.04, "learning_rate": 5e-06, "loss": 4.6406, "step": 21 }, { "epoch": 0.04, "learning_rate": 5.2380952380952384e-06, "loss": 4.5273, "step": 22 }, { "epoch": 0.04, "learning_rate": 5.4761904761904765e-06, "loss": 4.6484, "step": 23 }, { "epoch": 0.04, "learning_rate": 5.7142857142857145e-06, "loss": 3.8438, "step": 24 }, { "epoch": 0.04, "learning_rate": 5.9523809523809525e-06, "loss": 3.9648, "step": 25 }, { "epoch": 0.05, "learning_rate": 6.190476190476191e-06, "loss": 4.2109, "step": 26 }, { "epoch": 0.05, "learning_rate": 6.428571428571429e-06, "loss": 3.8242, "step": 27 }, { "epoch": 0.05, "learning_rate": 6.666666666666667e-06, "loss": 4.9609, "step": 28 }, { "epoch": 0.05, "learning_rate": 6.9047619047619055e-06, "loss": 3.8535, "step": 29 }, { "epoch": 0.05, "learning_rate": 7.142857142857143e-06, "loss": 3.7344, "step": 30 }, { "epoch": 0.06, "learning_rate": 7.380952380952382e-06, "loss": 4.3438, "step": 31 }, { "epoch": 0.06, "learning_rate": 7.6190476190476205e-06, "loss": 3.8047, "step": 32 }, { "epoch": 0.06, "learning_rate": 7.857142857142858e-06, "loss": 3.8672, "step": 33 }, { "epoch": 0.06, "learning_rate": 8.095238095238097e-06, "loss": 3.668, "step": 34 }, { "epoch": 0.06, "learning_rate": 8.333333333333334e-06, "loss": 4.1602, "step": 35 }, { "epoch": 0.06, "learning_rate": 8.571428571428573e-06, "loss": 4.1055, "step": 36 }, { "epoch": 0.07, "learning_rate": 8.80952380952381e-06, "loss": 3.7344, "step": 37 }, { "epoch": 0.07, "learning_rate": 9.047619047619047e-06, "loss": 3.7285, "step": 38 }, { "epoch": 0.07, "learning_rate": 9.285714285714286e-06, "loss": 3.8438, "step": 39 }, { "epoch": 0.07, "learning_rate": 9.523809523809523e-06, "loss": 3.6738, "step": 40 }, { "epoch": 0.07, "learning_rate": 9.761904761904762e-06, "loss": 4.0742, "step": 41 }, { "epoch": 0.07, "learning_rate": 1e-05, "loss": 3.6914, "step": 42 }, { "epoch": 0.08, "learning_rate": 1.0238095238095238e-05, "loss": 3.8633, "step": 43 }, { "epoch": 0.08, "learning_rate": 1.0476190476190477e-05, "loss": 4.1445, "step": 44 }, { "epoch": 0.08, "learning_rate": 1.0714285714285714e-05, "loss": 4.4922, "step": 45 }, { "epoch": 0.08, "learning_rate": 1.0952380952380953e-05, "loss": 4.0234, "step": 46 }, { "epoch": 0.08, "learning_rate": 1.1190476190476192e-05, "loss": 3.7637, "step": 47 }, { "epoch": 0.09, "learning_rate": 1.1428571428571429e-05, "loss": 4.1953, "step": 48 }, { "epoch": 0.09, "learning_rate": 1.1666666666666668e-05, "loss": 3.7578, "step": 49 }, { "epoch": 0.09, "learning_rate": 1.1904761904761905e-05, "loss": 4.418, "step": 50 }, { "epoch": 0.09, "learning_rate": 1.2142857142857144e-05, "loss": 4.0078, "step": 51 }, { "epoch": 0.09, "learning_rate": 1.2380952380952381e-05, "loss": 4.5078, "step": 52 }, { "epoch": 0.09, "learning_rate": 1.261904761904762e-05, "loss": 3.875, "step": 53 }, { "epoch": 0.1, "learning_rate": 1.2857142857142857e-05, "loss": 4.0547, "step": 54 }, { "epoch": 0.1, "learning_rate": 1.3095238095238096e-05, "loss": 3.6836, "step": 55 }, { "epoch": 0.1, "learning_rate": 1.3333333333333333e-05, "loss": 3.9336, "step": 56 }, { "epoch": 0.1, "learning_rate": 1.357142857142857e-05, "loss": 3.1445, "step": 57 }, { "epoch": 0.1, "learning_rate": 1.3809523809523811e-05, "loss": 3.7363, "step": 58 }, { "epoch": 0.11, "learning_rate": 1.4047619047619048e-05, "loss": 3.5312, "step": 59 }, { "epoch": 0.11, "learning_rate": 1.4285714285714285e-05, "loss": 3.8066, "step": 60 }, { "epoch": 0.11, "learning_rate": 1.4523809523809526e-05, "loss": 3.2168, "step": 61 }, { "epoch": 0.11, "learning_rate": 1.4761904761904763e-05, "loss": 3.4902, "step": 62 }, { "epoch": 0.11, "learning_rate": 1.5e-05, "loss": 3.7617, "step": 63 }, { "epoch": 0.11, "learning_rate": 1.5238095238095241e-05, "loss": 2.9531, "step": 64 }, { "epoch": 0.12, "learning_rate": 1.5476190476190476e-05, "loss": 3.5039, "step": 65 }, { "epoch": 0.12, "learning_rate": 1.5714285714285715e-05, "loss": 4.5273, "step": 66 }, { "epoch": 0.12, "learning_rate": 1.595238095238095e-05, "loss": 3.5332, "step": 67 }, { "epoch": 0.12, "learning_rate": 1.6190476190476193e-05, "loss": 3.4062, "step": 68 }, { "epoch": 0.12, "learning_rate": 1.642857142857143e-05, "loss": 3.2715, "step": 69 }, { "epoch": 0.12, "learning_rate": 1.6666666666666667e-05, "loss": 3.8633, "step": 70 }, { "epoch": 0.13, "learning_rate": 1.6904761904761906e-05, "loss": 3.2734, "step": 71 }, { "epoch": 0.13, "learning_rate": 1.7142857142857145e-05, "loss": 4.2422, "step": 72 }, { "epoch": 0.13, "learning_rate": 1.738095238095238e-05, "loss": 3.9492, "step": 73 }, { "epoch": 0.13, "learning_rate": 1.761904761904762e-05, "loss": 3.2715, "step": 74 }, { "epoch": 0.13, "learning_rate": 1.785714285714286e-05, "loss": 3.1914, "step": 75 }, { "epoch": 0.14, "learning_rate": 1.8095238095238094e-05, "loss": 3.6602, "step": 76 }, { "epoch": 0.14, "learning_rate": 1.8333333333333333e-05, "loss": 3.084, "step": 77 }, { "epoch": 0.14, "learning_rate": 1.8571428571428572e-05, "loss": 3.2988, "step": 78 }, { "epoch": 0.14, "learning_rate": 1.880952380952381e-05, "loss": 4.1289, "step": 79 }, { "epoch": 0.14, "learning_rate": 1.9047619047619046e-05, "loss": 3.707, "step": 80 }, { "epoch": 0.14, "learning_rate": 1.928571428571429e-05, "loss": 3.707, "step": 81 }, { "epoch": 0.15, "learning_rate": 1.9523809523809524e-05, "loss": 3.5508, "step": 82 }, { "epoch": 0.15, "learning_rate": 1.9761904761904763e-05, "loss": 3.0645, "step": 83 }, { "epoch": 0.15, "learning_rate": 2e-05, "loss": 3.2344, "step": 84 }, { "epoch": 0.15, "learning_rate": 2.023809523809524e-05, "loss": 3.1641, "step": 85 }, { "epoch": 0.15, "learning_rate": 2.0476190476190476e-05, "loss": 3.2188, "step": 86 }, { "epoch": 0.16, "learning_rate": 2.0714285714285718e-05, "loss": 2.5312, "step": 87 }, { "epoch": 0.16, "learning_rate": 2.0952380952380954e-05, "loss": 3.8164, "step": 88 }, { "epoch": 0.16, "learning_rate": 2.1190476190476193e-05, "loss": 3.3203, "step": 89 }, { "epoch": 0.16, "learning_rate": 2.1428571428571428e-05, "loss": 3.9023, "step": 90 }, { "epoch": 0.16, "learning_rate": 2.1666666666666667e-05, "loss": 3.5547, "step": 91 }, { "epoch": 0.16, "learning_rate": 2.1904761904761906e-05, "loss": 3.7012, "step": 92 }, { "epoch": 0.17, "learning_rate": 2.214285714285714e-05, "loss": 2.6992, "step": 93 }, { "epoch": 0.17, "learning_rate": 2.2380952380952384e-05, "loss": 3.9375, "step": 94 }, { "epoch": 0.17, "learning_rate": 2.261904761904762e-05, "loss": 3.1992, "step": 95 }, { "epoch": 0.17, "learning_rate": 2.2857142857142858e-05, "loss": 2.8398, "step": 96 }, { "epoch": 0.17, "learning_rate": 2.3095238095238097e-05, "loss": 3.4688, "step": 97 }, { "epoch": 0.17, "learning_rate": 2.3333333333333336e-05, "loss": 2.8203, "step": 98 }, { "epoch": 0.18, "learning_rate": 2.357142857142857e-05, "loss": 2.8926, "step": 99 }, { "epoch": 0.18, "learning_rate": 2.380952380952381e-05, "loss": 3.3047, "step": 100 }, { "epoch": 0.18, "learning_rate": 2.404761904761905e-05, "loss": 3.6641, "step": 101 }, { "epoch": 0.18, "learning_rate": 2.4285714285714288e-05, "loss": 2.4883, "step": 102 }, { "epoch": 0.18, "learning_rate": 2.4523809523809523e-05, "loss": 2.9023, "step": 103 }, { "epoch": 0.19, "learning_rate": 2.4761904761904762e-05, "loss": 3.6895, "step": 104 }, { "epoch": 0.19, "learning_rate": 2.5e-05, "loss": 3.4922, "step": 105 }, { "epoch": 0.19, "learning_rate": 2.523809523809524e-05, "loss": 3.1211, "step": 106 }, { "epoch": 0.19, "learning_rate": 2.5476190476190476e-05, "loss": 3.0254, "step": 107 }, { "epoch": 0.19, "learning_rate": 2.5714285714285714e-05, "loss": 2.4922, "step": 108 }, { "epoch": 0.19, "learning_rate": 2.5952380952380957e-05, "loss": 2.9844, "step": 109 }, { "epoch": 0.2, "learning_rate": 2.6190476190476192e-05, "loss": 3.0117, "step": 110 }, { "epoch": 0.2, "learning_rate": 2.642857142857143e-05, "loss": 3.6152, "step": 111 }, { "epoch": 0.2, "learning_rate": 2.6666666666666667e-05, "loss": 2.6953, "step": 112 }, { "epoch": 0.2, "learning_rate": 2.6904761904761905e-05, "loss": 3.3242, "step": 113 }, { "epoch": 0.2, "learning_rate": 2.714285714285714e-05, "loss": 3.7266, "step": 114 }, { "epoch": 0.21, "learning_rate": 2.7380952380952383e-05, "loss": 3.4512, "step": 115 }, { "epoch": 0.21, "learning_rate": 2.7619047619047622e-05, "loss": 3.4062, "step": 116 }, { "epoch": 0.21, "learning_rate": 2.785714285714286e-05, "loss": 2.9102, "step": 117 }, { "epoch": 0.21, "learning_rate": 2.8095238095238096e-05, "loss": 2.8281, "step": 118 }, { "epoch": 0.21, "learning_rate": 2.8333333333333335e-05, "loss": 2.6367, "step": 119 }, { "epoch": 0.21, "learning_rate": 2.857142857142857e-05, "loss": 3.0625, "step": 120 }, { "epoch": 0.22, "learning_rate": 2.880952380952381e-05, "loss": 3.2812, "step": 121 }, { "epoch": 0.22, "learning_rate": 2.9047619047619052e-05, "loss": 2.9062, "step": 122 }, { "epoch": 0.22, "learning_rate": 2.9285714285714288e-05, "loss": 3.0137, "step": 123 }, { "epoch": 0.22, "learning_rate": 2.9523809523809526e-05, "loss": 2.543, "step": 124 }, { "epoch": 0.22, "learning_rate": 2.9761904761904762e-05, "loss": 2.8242, "step": 125 }, { "epoch": 0.23, "learning_rate": 3e-05, "loss": 2.9062, "step": 126 }, { "epoch": 0.23, "learning_rate": 3.0238095238095236e-05, "loss": 3.1309, "step": 127 }, { "epoch": 0.23, "learning_rate": 3.0476190476190482e-05, "loss": 3.1621, "step": 128 }, { "epoch": 0.23, "learning_rate": 3.071428571428572e-05, "loss": 2.9258, "step": 129 }, { "epoch": 0.23, "learning_rate": 3.095238095238095e-05, "loss": 3.0879, "step": 130 }, { "epoch": 0.23, "learning_rate": 3.1190476190476195e-05, "loss": 3.5703, "step": 131 }, { "epoch": 0.24, "learning_rate": 3.142857142857143e-05, "loss": 2.8945, "step": 132 }, { "epoch": 0.24, "learning_rate": 3.1666666666666666e-05, "loss": 3.5566, "step": 133 }, { "epoch": 0.24, "learning_rate": 3.19047619047619e-05, "loss": 2.5312, "step": 134 }, { "epoch": 0.24, "learning_rate": 3.2142857142857144e-05, "loss": 2.8398, "step": 135 }, { "epoch": 0.24, "learning_rate": 3.2380952380952386e-05, "loss": 3.293, "step": 136 }, { "epoch": 0.24, "learning_rate": 3.261904761904762e-05, "loss": 2.8125, "step": 137 }, { "epoch": 0.25, "learning_rate": 3.285714285714286e-05, "loss": 2.5273, "step": 138 }, { "epoch": 0.25, "learning_rate": 3.309523809523809e-05, "loss": 2.9219, "step": 139 }, { "epoch": 0.25, "learning_rate": 3.3333333333333335e-05, "loss": 3.5039, "step": 140 }, { "epoch": 0.25, "learning_rate": 3.357142857142857e-05, "loss": 3.2656, "step": 141 }, { "epoch": 0.25, "learning_rate": 3.380952380952381e-05, "loss": 3.0625, "step": 142 }, { "epoch": 0.26, "learning_rate": 3.404761904761905e-05, "loss": 3.1309, "step": 143 }, { "epoch": 0.26, "learning_rate": 3.428571428571429e-05, "loss": 2.1836, "step": 144 }, { "epoch": 0.26, "learning_rate": 3.4523809523809526e-05, "loss": 2.5664, "step": 145 }, { "epoch": 0.26, "learning_rate": 3.476190476190476e-05, "loss": 2.9043, "step": 146 }, { "epoch": 0.26, "learning_rate": 3.5e-05, "loss": 2.5156, "step": 147 }, { "epoch": 0.26, "learning_rate": 3.523809523809524e-05, "loss": 2.9531, "step": 148 }, { "epoch": 0.27, "learning_rate": 3.547619047619048e-05, "loss": 2.5059, "step": 149 }, { "epoch": 0.27, "learning_rate": 3.571428571428572e-05, "loss": 3.625, "step": 150 }, { "epoch": 0.27, "learning_rate": 3.595238095238095e-05, "loss": 2.75, "step": 151 }, { "epoch": 0.27, "learning_rate": 3.619047619047619e-05, "loss": 2.7734, "step": 152 }, { "epoch": 0.27, "learning_rate": 3.642857142857143e-05, "loss": 2.3516, "step": 153 }, { "epoch": 0.28, "learning_rate": 3.6666666666666666e-05, "loss": 3.3086, "step": 154 }, { "epoch": 0.28, "learning_rate": 3.690476190476191e-05, "loss": 3.2188, "step": 155 }, { "epoch": 0.28, "learning_rate": 3.7142857142857143e-05, "loss": 2.9023, "step": 156 }, { "epoch": 0.28, "learning_rate": 3.7380952380952386e-05, "loss": 2.625, "step": 157 }, { "epoch": 0.28, "learning_rate": 3.761904761904762e-05, "loss": 2.5703, "step": 158 }, { "epoch": 0.28, "learning_rate": 3.785714285714286e-05, "loss": 3.2188, "step": 159 }, { "epoch": 0.29, "learning_rate": 3.809523809523809e-05, "loss": 3.1445, "step": 160 }, { "epoch": 0.29, "learning_rate": 3.8333333333333334e-05, "loss": 2.7852, "step": 161 }, { "epoch": 0.29, "learning_rate": 3.857142857142858e-05, "loss": 3.0586, "step": 162 }, { "epoch": 0.29, "learning_rate": 3.880952380952381e-05, "loss": 2.9883, "step": 163 }, { "epoch": 0.29, "learning_rate": 3.904761904761905e-05, "loss": 2.4238, "step": 164 }, { "epoch": 0.29, "learning_rate": 3.928571428571429e-05, "loss": 3.0703, "step": 165 }, { "epoch": 0.3, "learning_rate": 3.9523809523809526e-05, "loss": 3.041, "step": 166 }, { "epoch": 0.3, "learning_rate": 3.976190476190476e-05, "loss": 3.0898, "step": 167 }, { "epoch": 0.3, "learning_rate": 4e-05, "loss": 2.3516, "step": 168 }, { "epoch": 0.3, "learning_rate": 4.023809523809524e-05, "loss": 2.6504, "step": 169 }, { "epoch": 0.3, "learning_rate": 4.047619047619048e-05, "loss": 3.0996, "step": 170 }, { "epoch": 0.31, "learning_rate": 4.0714285714285717e-05, "loss": 2.2852, "step": 171 }, { "epoch": 0.31, "learning_rate": 4.095238095238095e-05, "loss": 2.7695, "step": 172 }, { "epoch": 0.31, "learning_rate": 4.119047619047619e-05, "loss": 2.7773, "step": 173 }, { "epoch": 0.31, "learning_rate": 4.1428571428571437e-05, "loss": 2.7461, "step": 174 }, { "epoch": 0.31, "learning_rate": 4.166666666666667e-05, "loss": 2.8535, "step": 175 }, { "epoch": 0.31, "learning_rate": 4.190476190476191e-05, "loss": 2.9609, "step": 176 }, { "epoch": 0.32, "learning_rate": 4.214285714285714e-05, "loss": 2.8047, "step": 177 }, { "epoch": 0.32, "learning_rate": 4.2380952380952385e-05, "loss": 2.7812, "step": 178 }, { "epoch": 0.32, "learning_rate": 4.261904761904762e-05, "loss": 2.4688, "step": 179 }, { "epoch": 0.32, "learning_rate": 4.2857142857142856e-05, "loss": 2.8027, "step": 180 }, { "epoch": 0.32, "learning_rate": 4.30952380952381e-05, "loss": 2.2012, "step": 181 }, { "epoch": 0.33, "learning_rate": 4.3333333333333334e-05, "loss": 2.709, "step": 182 }, { "epoch": 0.33, "learning_rate": 4.3571428571428576e-05, "loss": 2.6953, "step": 183 }, { "epoch": 0.33, "learning_rate": 4.380952380952381e-05, "loss": 2.207, "step": 184 }, { "epoch": 0.33, "learning_rate": 4.404761904761905e-05, "loss": 2.2051, "step": 185 }, { "epoch": 0.33, "learning_rate": 4.428571428571428e-05, "loss": 2.5742, "step": 186 }, { "epoch": 0.33, "learning_rate": 4.4523809523809525e-05, "loss": 2.5742, "step": 187 }, { "epoch": 0.34, "learning_rate": 4.476190476190477e-05, "loss": 2.5, "step": 188 }, { "epoch": 0.34, "learning_rate": 4.5e-05, "loss": 2.5508, "step": 189 }, { "epoch": 0.34, "learning_rate": 4.523809523809524e-05, "loss": 2.5059, "step": 190 }, { "epoch": 0.34, "learning_rate": 4.547619047619048e-05, "loss": 2.6836, "step": 191 }, { "epoch": 0.34, "learning_rate": 4.5714285714285716e-05, "loss": 2.9629, "step": 192 }, { "epoch": 0.34, "learning_rate": 4.595238095238095e-05, "loss": 2.7422, "step": 193 }, { "epoch": 0.35, "learning_rate": 4.6190476190476194e-05, "loss": 2.832, "step": 194 }, { "epoch": 0.35, "learning_rate": 4.642857142857143e-05, "loss": 3.1348, "step": 195 }, { "epoch": 0.35, "learning_rate": 4.666666666666667e-05, "loss": 2.4492, "step": 196 }, { "epoch": 0.35, "learning_rate": 4.690476190476191e-05, "loss": 2.3125, "step": 197 }, { "epoch": 0.35, "learning_rate": 4.714285714285714e-05, "loss": 1.9961, "step": 198 }, { "epoch": 0.36, "learning_rate": 4.738095238095238e-05, "loss": 2.9336, "step": 199 }, { "epoch": 0.36, "learning_rate": 4.761904761904762e-05, "loss": 2.3438, "step": 200 }, { "epoch": 0.36, "learning_rate": 4.785714285714286e-05, "loss": 3.0469, "step": 201 }, { "epoch": 0.36, "learning_rate": 4.80952380952381e-05, "loss": 2.8945, "step": 202 }, { "epoch": 0.36, "learning_rate": 4.8333333333333334e-05, "loss": 2.6289, "step": 203 }, { "epoch": 0.36, "learning_rate": 4.8571428571428576e-05, "loss": 2.4414, "step": 204 }, { "epoch": 0.37, "learning_rate": 4.880952380952381e-05, "loss": 2.498, "step": 205 }, { "epoch": 0.37, "learning_rate": 4.904761904761905e-05, "loss": 3.0, "step": 206 }, { "epoch": 0.37, "learning_rate": 4.928571428571429e-05, "loss": 2.6387, "step": 207 }, { "epoch": 0.37, "learning_rate": 4.9523809523809525e-05, "loss": 2.082, "step": 208 }, { "epoch": 0.37, "learning_rate": 4.976190476190477e-05, "loss": 2.4785, "step": 209 }, { "epoch": 0.38, "learning_rate": 5e-05, "loss": 2.8066, "step": 210 }, { "epoch": 0.38, "learning_rate": 5.023809523809524e-05, "loss": 3.1484, "step": 211 }, { "epoch": 0.38, "learning_rate": 5.047619047619048e-05, "loss": 2.582, "step": 212 }, { "epoch": 0.38, "learning_rate": 5.0714285714285716e-05, "loss": 2.7012, "step": 213 }, { "epoch": 0.38, "learning_rate": 5.095238095238095e-05, "loss": 2.7832, "step": 214 }, { "epoch": 0.38, "learning_rate": 5.119047619047619e-05, "loss": 2.3984, "step": 215 }, { "epoch": 0.39, "learning_rate": 5.142857142857143e-05, "loss": 2.5938, "step": 216 }, { "epoch": 0.39, "learning_rate": 5.166666666666667e-05, "loss": 2.7676, "step": 217 }, { "epoch": 0.39, "learning_rate": 5.1904761904761913e-05, "loss": 2.291, "step": 218 }, { "epoch": 0.39, "learning_rate": 5.214285714285715e-05, "loss": 2.2441, "step": 219 }, { "epoch": 0.39, "learning_rate": 5.2380952380952384e-05, "loss": 2.4492, "step": 220 }, { "epoch": 0.39, "learning_rate": 5.261904761904763e-05, "loss": 2.4062, "step": 221 }, { "epoch": 0.4, "learning_rate": 5.285714285714286e-05, "loss": 2.6621, "step": 222 }, { "epoch": 0.4, "learning_rate": 5.30952380952381e-05, "loss": 2.2891, "step": 223 }, { "epoch": 0.4, "learning_rate": 5.333333333333333e-05, "loss": 2.7812, "step": 224 }, { "epoch": 0.4, "learning_rate": 5.3571428571428575e-05, "loss": 2.2441, "step": 225 }, { "epoch": 0.4, "learning_rate": 5.380952380952381e-05, "loss": 2.7676, "step": 226 }, { "epoch": 0.41, "learning_rate": 5.4047619047619046e-05, "loss": 2.6348, "step": 227 }, { "epoch": 0.41, "learning_rate": 5.428571428571428e-05, "loss": 3.4023, "step": 228 }, { "epoch": 0.41, "learning_rate": 5.4523809523809524e-05, "loss": 2.6426, "step": 229 }, { "epoch": 0.41, "learning_rate": 5.4761904761904766e-05, "loss": 2.8887, "step": 230 }, { "epoch": 0.41, "learning_rate": 5.500000000000001e-05, "loss": 2.707, "step": 231 }, { "epoch": 0.41, "learning_rate": 5.5238095238095244e-05, "loss": 2.6172, "step": 232 }, { "epoch": 0.42, "learning_rate": 5.547619047619048e-05, "loss": 2.6484, "step": 233 }, { "epoch": 0.42, "learning_rate": 5.571428571428572e-05, "loss": 2.4512, "step": 234 }, { "epoch": 0.42, "learning_rate": 5.595238095238096e-05, "loss": 2.3203, "step": 235 }, { "epoch": 0.42, "learning_rate": 5.619047619047619e-05, "loss": 2.6074, "step": 236 }, { "epoch": 0.42, "learning_rate": 5.642857142857143e-05, "loss": 2.7539, "step": 237 }, { "epoch": 0.42, "learning_rate": 5.666666666666667e-05, "loss": 2.6035, "step": 238 }, { "epoch": 0.43, "learning_rate": 5.6904761904761906e-05, "loss": 2.791, "step": 239 }, { "epoch": 0.43, "learning_rate": 5.714285714285714e-05, "loss": 2.9609, "step": 240 }, { "epoch": 0.43, "learning_rate": 5.738095238095238e-05, "loss": 2.3281, "step": 241 }, { "epoch": 0.43, "learning_rate": 5.761904761904762e-05, "loss": 2.4414, "step": 242 }, { "epoch": 0.43, "learning_rate": 5.785714285714287e-05, "loss": 2.2969, "step": 243 }, { "epoch": 0.44, "learning_rate": 5.8095238095238104e-05, "loss": 3.2949, "step": 244 }, { "epoch": 0.44, "learning_rate": 5.833333333333334e-05, "loss": 3.082, "step": 245 }, { "epoch": 0.44, "learning_rate": 5.8571428571428575e-05, "loss": 2.75, "step": 246 }, { "epoch": 0.44, "learning_rate": 5.880952380952382e-05, "loss": 2.7812, "step": 247 }, { "epoch": 0.44, "learning_rate": 5.904761904761905e-05, "loss": 3.0723, "step": 248 }, { "epoch": 0.44, "learning_rate": 5.928571428571429e-05, "loss": 3.25, "step": 249 }, { "epoch": 0.45, "learning_rate": 5.9523809523809524e-05, "loss": 2.2383, "step": 250 }, { "epoch": 0.45, "learning_rate": 5.9761904761904766e-05, "loss": 2.1504, "step": 251 }, { "epoch": 0.45, "learning_rate": 6e-05, "loss": 2.3711, "step": 252 }, { "epoch": 0.45, "learning_rate": 6.023809523809524e-05, "loss": 2.793, "step": 253 }, { "epoch": 0.45, "learning_rate": 6.047619047619047e-05, "loss": 2.9688, "step": 254 }, { "epoch": 0.46, "learning_rate": 6.0714285714285715e-05, "loss": 2.0117, "step": 255 }, { "epoch": 0.46, "learning_rate": 6.0952380952380964e-05, "loss": 2.4961, "step": 256 }, { "epoch": 0.46, "learning_rate": 6.11904761904762e-05, "loss": 2.0195, "step": 257 }, { "epoch": 0.46, "learning_rate": 6.142857142857143e-05, "loss": 2.1152, "step": 258 }, { "epoch": 0.46, "learning_rate": 6.166666666666667e-05, "loss": 2.4648, "step": 259 }, { "epoch": 0.46, "learning_rate": 6.19047619047619e-05, "loss": 2.7305, "step": 260 }, { "epoch": 0.47, "learning_rate": 6.214285714285714e-05, "loss": 2.293, "step": 261 }, { "epoch": 0.47, "learning_rate": 6.238095238095239e-05, "loss": 2.4102, "step": 262 }, { "epoch": 0.47, "learning_rate": 6.261904761904763e-05, "loss": 2.3848, "step": 263 }, { "epoch": 0.47, "learning_rate": 6.285714285714286e-05, "loss": 2.2188, "step": 264 }, { "epoch": 0.47, "learning_rate": 6.30952380952381e-05, "loss": 2.873, "step": 265 }, { "epoch": 0.47, "learning_rate": 6.333333333333333e-05, "loss": 3.6621, "step": 266 }, { "epoch": 0.48, "learning_rate": 6.357142857142857e-05, "loss": 2.8535, "step": 267 }, { "epoch": 0.48, "learning_rate": 6.38095238095238e-05, "loss": 3.1758, "step": 268 }, { "epoch": 0.48, "learning_rate": 6.404761904761904e-05, "loss": 2.6699, "step": 269 }, { "epoch": 0.48, "learning_rate": 6.428571428571429e-05, "loss": 2.6172, "step": 270 }, { "epoch": 0.48, "learning_rate": 6.452380952380954e-05, "loss": 2.4551, "step": 271 }, { "epoch": 0.49, "learning_rate": 6.476190476190477e-05, "loss": 2.6719, "step": 272 }, { "epoch": 0.49, "learning_rate": 6.500000000000001e-05, "loss": 2.6191, "step": 273 }, { "epoch": 0.49, "learning_rate": 6.523809523809524e-05, "loss": 2.873, "step": 274 }, { "epoch": 0.49, "learning_rate": 6.547619047619048e-05, "loss": 2.4551, "step": 275 }, { "epoch": 0.49, "learning_rate": 6.571428571428571e-05, "loss": 2.4375, "step": 276 }, { "epoch": 0.49, "learning_rate": 6.595238095238095e-05, "loss": 2.6895, "step": 277 }, { "epoch": 0.5, "learning_rate": 6.619047619047619e-05, "loss": 2.4473, "step": 278 }, { "epoch": 0.5, "learning_rate": 6.642857142857143e-05, "loss": 2.0273, "step": 279 }, { "epoch": 0.5, "learning_rate": 6.666666666666667e-05, "loss": 2.5938, "step": 280 }, { "epoch": 0.5, "learning_rate": 6.69047619047619e-05, "loss": 2.6699, "step": 281 }, { "epoch": 0.5, "learning_rate": 6.714285714285714e-05, "loss": 2.373, "step": 282 }, { "epoch": 0.51, "learning_rate": 6.738095238095239e-05, "loss": 2.6523, "step": 283 }, { "epoch": 0.51, "learning_rate": 6.761904761904763e-05, "loss": 2.3184, "step": 284 }, { "epoch": 0.51, "learning_rate": 6.785714285714286e-05, "loss": 2.9141, "step": 285 }, { "epoch": 0.51, "learning_rate": 6.80952380952381e-05, "loss": 2.1211, "step": 286 }, { "epoch": 0.51, "learning_rate": 6.833333333333333e-05, "loss": 2.3926, "step": 287 }, { "epoch": 0.51, "learning_rate": 6.857142857142858e-05, "loss": 2.7266, "step": 288 }, { "epoch": 0.52, "learning_rate": 6.880952380952382e-05, "loss": 2.8535, "step": 289 }, { "epoch": 0.52, "learning_rate": 6.904761904761905e-05, "loss": 2.3633, "step": 290 }, { "epoch": 0.52, "learning_rate": 6.928571428571429e-05, "loss": 2.5762, "step": 291 }, { "epoch": 0.52, "learning_rate": 6.952380952380952e-05, "loss": 2.4766, "step": 292 }, { "epoch": 0.52, "learning_rate": 6.976190476190476e-05, "loss": 1.7285, "step": 293 }, { "epoch": 0.53, "learning_rate": 7e-05, "loss": 2.1484, "step": 294 }, { "epoch": 0.53, "learning_rate": 7.023809523809524e-05, "loss": 2.5527, "step": 295 }, { "epoch": 0.53, "learning_rate": 7.047619047619048e-05, "loss": 2.8613, "step": 296 }, { "epoch": 0.53, "learning_rate": 7.071428571428573e-05, "loss": 2.4805, "step": 297 }, { "epoch": 0.53, "learning_rate": 7.095238095238096e-05, "loss": 2.3164, "step": 298 }, { "epoch": 0.53, "learning_rate": 7.11904761904762e-05, "loss": 2.1797, "step": 299 }, { "epoch": 0.54, "learning_rate": 7.142857142857143e-05, "loss": 1.8604, "step": 300 }, { "epoch": 0.54, "learning_rate": 7.166666666666667e-05, "loss": 2.8926, "step": 301 }, { "epoch": 0.54, "learning_rate": 7.19047619047619e-05, "loss": 1.8184, "step": 302 }, { "epoch": 0.54, "learning_rate": 7.214285714285714e-05, "loss": 2.3828, "step": 303 }, { "epoch": 0.54, "learning_rate": 7.238095238095238e-05, "loss": 2.5352, "step": 304 }, { "epoch": 0.54, "learning_rate": 7.261904761904762e-05, "loss": 2.7305, "step": 305 }, { "epoch": 0.55, "learning_rate": 7.285714285714286e-05, "loss": 2.8281, "step": 306 }, { "epoch": 0.55, "learning_rate": 7.30952380952381e-05, "loss": 2.5527, "step": 307 }, { "epoch": 0.55, "learning_rate": 7.333333333333333e-05, "loss": 2.5508, "step": 308 }, { "epoch": 0.55, "learning_rate": 7.357142857142858e-05, "loss": 2.4062, "step": 309 }, { "epoch": 0.55, "learning_rate": 7.380952380952382e-05, "loss": 3.1602, "step": 310 }, { "epoch": 0.56, "learning_rate": 7.404761904761905e-05, "loss": 2.6055, "step": 311 }, { "epoch": 0.56, "learning_rate": 7.428571428571429e-05, "loss": 2.4668, "step": 312 }, { "epoch": 0.56, "learning_rate": 7.452380952380952e-05, "loss": 2.4062, "step": 313 }, { "epoch": 0.56, "learning_rate": 7.476190476190477e-05, "loss": 2.1699, "step": 314 }, { "epoch": 0.56, "learning_rate": 7.500000000000001e-05, "loss": 2.6445, "step": 315 }, { "epoch": 0.56, "learning_rate": 7.523809523809524e-05, "loss": 2.6152, "step": 316 }, { "epoch": 0.57, "learning_rate": 7.547619047619048e-05, "loss": 2.5039, "step": 317 }, { "epoch": 0.57, "learning_rate": 7.571428571428571e-05, "loss": 2.5078, "step": 318 }, { "epoch": 0.57, "learning_rate": 7.595238095238095e-05, "loss": 2.3984, "step": 319 }, { "epoch": 0.57, "learning_rate": 7.619047619047618e-05, "loss": 2.9785, "step": 320 }, { "epoch": 0.57, "learning_rate": 7.642857142857143e-05, "loss": 2.9688, "step": 321 }, { "epoch": 0.57, "learning_rate": 7.666666666666667e-05, "loss": 2.5781, "step": 322 }, { "epoch": 0.58, "learning_rate": 7.690476190476192e-05, "loss": 2.3359, "step": 323 }, { "epoch": 0.58, "learning_rate": 7.714285714285715e-05, "loss": 2.9336, "step": 324 }, { "epoch": 0.58, "learning_rate": 7.738095238095239e-05, "loss": 2.5234, "step": 325 }, { "epoch": 0.58, "learning_rate": 7.761904761904762e-05, "loss": 3.1406, "step": 326 }, { "epoch": 0.58, "learning_rate": 7.785714285714286e-05, "loss": 2.4473, "step": 327 }, { "epoch": 0.59, "learning_rate": 7.80952380952381e-05, "loss": 2.4648, "step": 328 }, { "epoch": 0.59, "learning_rate": 7.833333333333333e-05, "loss": 2.709, "step": 329 }, { "epoch": 0.59, "learning_rate": 7.857142857142858e-05, "loss": 2.2852, "step": 330 }, { "epoch": 0.59, "learning_rate": 7.880952380952382e-05, "loss": 2.5469, "step": 331 }, { "epoch": 0.59, "learning_rate": 7.904761904761905e-05, "loss": 2.2734, "step": 332 }, { "epoch": 0.59, "learning_rate": 7.928571428571429e-05, "loss": 2.4023, "step": 333 }, { "epoch": 0.6, "learning_rate": 7.952380952380952e-05, "loss": 2.0039, "step": 334 }, { "epoch": 0.6, "learning_rate": 7.976190476190477e-05, "loss": 2.6133, "step": 335 }, { "epoch": 0.6, "learning_rate": 8e-05, "loss": 2.8086, "step": 336 }, { "epoch": 0.6, "learning_rate": 8.023809523809524e-05, "loss": 2.1211, "step": 337 }, { "epoch": 0.6, "learning_rate": 8.047619047619048e-05, "loss": 1.8906, "step": 338 }, { "epoch": 0.61, "learning_rate": 8.071428571428573e-05, "loss": 2.3105, "step": 339 }, { "epoch": 0.61, "learning_rate": 8.095238095238096e-05, "loss": 2.875, "step": 340 }, { "epoch": 0.61, "learning_rate": 8.11904761904762e-05, "loss": 1.7012, "step": 341 }, { "epoch": 0.61, "learning_rate": 8.142857142857143e-05, "loss": 1.9688, "step": 342 }, { "epoch": 0.61, "learning_rate": 8.166666666666667e-05, "loss": 2.3125, "step": 343 }, { "epoch": 0.61, "learning_rate": 8.19047619047619e-05, "loss": 2.6523, "step": 344 }, { "epoch": 0.62, "learning_rate": 8.214285714285714e-05, "loss": 2.5781, "step": 345 }, { "epoch": 0.62, "learning_rate": 8.238095238095238e-05, "loss": 2.7578, "step": 346 }, { "epoch": 0.62, "learning_rate": 8.261904761904762e-05, "loss": 1.8359, "step": 347 }, { "epoch": 0.62, "learning_rate": 8.285714285714287e-05, "loss": 2.1875, "step": 348 }, { "epoch": 0.62, "learning_rate": 8.309523809523811e-05, "loss": 2.4805, "step": 349 }, { "epoch": 0.62, "learning_rate": 8.333333333333334e-05, "loss": 2.9551, "step": 350 }, { "epoch": 0.63, "learning_rate": 8.357142857142858e-05, "loss": 2.5547, "step": 351 }, { "epoch": 0.63, "learning_rate": 8.380952380952382e-05, "loss": 2.4688, "step": 352 }, { "epoch": 0.63, "learning_rate": 8.404761904761905e-05, "loss": 1.6758, "step": 353 }, { "epoch": 0.63, "learning_rate": 8.428571428571429e-05, "loss": 2.6211, "step": 354 }, { "epoch": 0.63, "learning_rate": 8.452380952380952e-05, "loss": 3.1328, "step": 355 }, { "epoch": 0.64, "learning_rate": 8.476190476190477e-05, "loss": 2.2949, "step": 356 }, { "epoch": 0.64, "learning_rate": 8.5e-05, "loss": 3.0859, "step": 357 }, { "epoch": 0.64, "learning_rate": 8.523809523809524e-05, "loss": 2.6914, "step": 358 }, { "epoch": 0.64, "learning_rate": 8.547619047619048e-05, "loss": 1.9902, "step": 359 }, { "epoch": 0.64, "learning_rate": 8.571428571428571e-05, "loss": 3.0527, "step": 360 }, { "epoch": 0.64, "learning_rate": 8.595238095238096e-05, "loss": 2.7031, "step": 361 }, { "epoch": 0.65, "learning_rate": 8.61904761904762e-05, "loss": 2.002, "step": 362 }, { "epoch": 0.65, "learning_rate": 8.642857142857143e-05, "loss": 2.3242, "step": 363 }, { "epoch": 0.65, "learning_rate": 8.666666666666667e-05, "loss": 2.0547, "step": 364 }, { "epoch": 0.65, "learning_rate": 8.690476190476192e-05, "loss": 2.3086, "step": 365 }, { "epoch": 0.65, "learning_rate": 8.714285714285715e-05, "loss": 2.6758, "step": 366 }, { "epoch": 0.66, "learning_rate": 8.738095238095239e-05, "loss": 2.457, "step": 367 }, { "epoch": 0.66, "learning_rate": 8.761904761904762e-05, "loss": 2.5312, "step": 368 }, { "epoch": 0.66, "learning_rate": 8.785714285714286e-05, "loss": 2.1445, "step": 369 }, { "epoch": 0.66, "learning_rate": 8.80952380952381e-05, "loss": 2.5664, "step": 370 }, { "epoch": 0.66, "learning_rate": 8.833333333333333e-05, "loss": 2.5156, "step": 371 }, { "epoch": 0.66, "learning_rate": 8.857142857142857e-05, "loss": 2.5156, "step": 372 }, { "epoch": 0.67, "learning_rate": 8.880952380952381e-05, "loss": 2.5391, "step": 373 }, { "epoch": 0.67, "learning_rate": 8.904761904761905e-05, "loss": 2.4609, "step": 374 }, { "epoch": 0.67, "learning_rate": 8.92857142857143e-05, "loss": 2.5352, "step": 375 }, { "epoch": 0.67, "learning_rate": 8.952380952380953e-05, "loss": 2.084, "step": 376 }, { "epoch": 0.67, "learning_rate": 8.976190476190477e-05, "loss": 2.1133, "step": 377 }, { "epoch": 0.68, "learning_rate": 9e-05, "loss": 2.5449, "step": 378 }, { "epoch": 0.68, "learning_rate": 9.023809523809524e-05, "loss": 3.1875, "step": 379 }, { "epoch": 0.68, "learning_rate": 9.047619047619048e-05, "loss": 2.1211, "step": 380 }, { "epoch": 0.68, "learning_rate": 9.071428571428571e-05, "loss": 3.0391, "step": 381 }, { "epoch": 0.68, "learning_rate": 9.095238095238096e-05, "loss": 2.1602, "step": 382 }, { "epoch": 0.68, "learning_rate": 9.11904761904762e-05, "loss": 2.2969, "step": 383 }, { "epoch": 0.69, "learning_rate": 9.142857142857143e-05, "loss": 2.6016, "step": 384 }, { "epoch": 0.69, "learning_rate": 9.166666666666667e-05, "loss": 3.043, "step": 385 }, { "epoch": 0.69, "learning_rate": 9.19047619047619e-05, "loss": 2.5469, "step": 386 }, { "epoch": 0.69, "learning_rate": 9.214285714285714e-05, "loss": 1.9902, "step": 387 }, { "epoch": 0.69, "learning_rate": 9.238095238095239e-05, "loss": 2.3047, "step": 388 }, { "epoch": 0.69, "learning_rate": 9.261904761904762e-05, "loss": 2.041, "step": 389 }, { "epoch": 0.7, "learning_rate": 9.285714285714286e-05, "loss": 2.9297, "step": 390 }, { "epoch": 0.7, "learning_rate": 9.309523809523811e-05, "loss": 2.2695, "step": 391 }, { "epoch": 0.7, "learning_rate": 9.333333333333334e-05, "loss": 2.1016, "step": 392 }, { "epoch": 0.7, "learning_rate": 9.357142857142858e-05, "loss": 2.8242, "step": 393 }, { "epoch": 0.7, "learning_rate": 9.380952380952381e-05, "loss": 2.0312, "step": 394 }, { "epoch": 0.71, "learning_rate": 9.404761904761905e-05, "loss": 1.9766, "step": 395 }, { "epoch": 0.71, "learning_rate": 9.428571428571429e-05, "loss": 2.4277, "step": 396 }, { "epoch": 0.71, "learning_rate": 9.452380952380952e-05, "loss": 2.1953, "step": 397 }, { "epoch": 0.71, "learning_rate": 9.476190476190476e-05, "loss": 2.6426, "step": 398 }, { "epoch": 0.71, "learning_rate": 9.5e-05, "loss": 2.4453, "step": 399 }, { "epoch": 0.71, "learning_rate": 9.523809523809524e-05, "loss": 2.0703, "step": 400 }, { "epoch": 0.72, "learning_rate": 9.547619047619049e-05, "loss": 2.3242, "step": 401 }, { "epoch": 0.72, "learning_rate": 9.571428571428573e-05, "loss": 2.8555, "step": 402 }, { "epoch": 0.72, "learning_rate": 9.595238095238096e-05, "loss": 2.2969, "step": 403 }, { "epoch": 0.72, "learning_rate": 9.61904761904762e-05, "loss": 2.6035, "step": 404 }, { "epoch": 0.72, "learning_rate": 9.642857142857143e-05, "loss": 2.0273, "step": 405 }, { "epoch": 0.72, "learning_rate": 9.666666666666667e-05, "loss": 2.4102, "step": 406 }, { "epoch": 0.73, "learning_rate": 9.69047619047619e-05, "loss": 2.7031, "step": 407 }, { "epoch": 0.73, "learning_rate": 9.714285714285715e-05, "loss": 2.3047, "step": 408 }, { "epoch": 0.73, "learning_rate": 9.738095238095239e-05, "loss": 2.8086, "step": 409 }, { "epoch": 0.73, "learning_rate": 9.761904761904762e-05, "loss": 2.1328, "step": 410 }, { "epoch": 0.73, "learning_rate": 9.785714285714286e-05, "loss": 2.8379, "step": 411 }, { "epoch": 0.74, "learning_rate": 9.80952380952381e-05, "loss": 1.6934, "step": 412 }, { "epoch": 0.74, "learning_rate": 9.833333333333333e-05, "loss": 2.4219, "step": 413 }, { "epoch": 0.74, "learning_rate": 9.857142857142858e-05, "loss": 1.8232, "step": 414 }, { "epoch": 0.74, "learning_rate": 9.880952380952381e-05, "loss": 2.4805, "step": 415 }, { "epoch": 0.74, "learning_rate": 9.904761904761905e-05, "loss": 2.1953, "step": 416 }, { "epoch": 0.74, "learning_rate": 9.92857142857143e-05, "loss": 2.4297, "step": 417 }, { "epoch": 0.75, "learning_rate": 9.952380952380953e-05, "loss": 2.1406, "step": 418 }, { "epoch": 0.75, "learning_rate": 9.976190476190477e-05, "loss": 2.2227, "step": 419 }, { "epoch": 0.75, "learning_rate": 0.0001, "loss": 2.3965, "step": 420 }, { "epoch": 0.75, "learning_rate": 9.999995644021005e-05, "loss": 2.6992, "step": 421 }, { "epoch": 0.75, "learning_rate": 9.99998257609161e-05, "loss": 2.4473, "step": 422 }, { "epoch": 0.76, "learning_rate": 9.999960796234583e-05, "loss": 2.6934, "step": 423 }, { "epoch": 0.76, "learning_rate": 9.999930304487874e-05, "loss": 2.7441, "step": 424 }, { "epoch": 0.76, "learning_rate": 9.999891100904613e-05, "loss": 2.623, "step": 425 }, { "epoch": 0.76, "learning_rate": 9.999843185553105e-05, "loss": 2.1465, "step": 426 }, { "epoch": 0.76, "learning_rate": 9.99978655851684e-05, "loss": 2.4121, "step": 427 }, { "epoch": 0.76, "learning_rate": 9.999721219894483e-05, "loss": 2.2637, "step": 428 }, { "epoch": 0.77, "learning_rate": 9.99964716979988e-05, "loss": 2.6172, "step": 429 }, { "epoch": 0.77, "learning_rate": 9.999564408362054e-05, "loss": 2.168, "step": 430 }, { "epoch": 0.77, "learning_rate": 9.999472935725209e-05, "loss": 1.8506, "step": 431 }, { "epoch": 0.77, "learning_rate": 9.999372752048728e-05, "loss": 2.4141, "step": 432 }, { "epoch": 0.77, "learning_rate": 9.999263857507166e-05, "loss": 2.7363, "step": 433 }, { "epoch": 0.78, "learning_rate": 9.999146252290264e-05, "loss": 2.5312, "step": 434 }, { "epoch": 0.78, "learning_rate": 9.999019936602934e-05, "loss": 2.4512, "step": 435 }, { "epoch": 0.78, "learning_rate": 9.998884910665267e-05, "loss": 2.3672, "step": 436 }, { "epoch": 0.78, "learning_rate": 9.998741174712533e-05, "loss": 2.0156, "step": 437 }, { "epoch": 0.78, "learning_rate": 9.998588728995176e-05, "loss": 2.3574, "step": 438 }, { "epoch": 0.78, "learning_rate": 9.998427573778815e-05, "loss": 1.9258, "step": 439 }, { "epoch": 0.79, "learning_rate": 9.998257709344246e-05, "loss": 3.1328, "step": 440 }, { "epoch": 0.79, "learning_rate": 9.998079135987437e-05, "loss": 2.168, "step": 441 }, { "epoch": 0.79, "learning_rate": 9.997891854019536e-05, "loss": 1.8887, "step": 442 }, { "epoch": 0.79, "learning_rate": 9.997695863766863e-05, "loss": 2.0273, "step": 443 }, { "epoch": 0.79, "learning_rate": 9.997491165570906e-05, "loss": 2.9375, "step": 444 }, { "epoch": 0.79, "learning_rate": 9.997277759788331e-05, "loss": 1.998, "step": 445 }, { "epoch": 0.8, "learning_rate": 9.997055646790972e-05, "loss": 2.1953, "step": 446 }, { "epoch": 0.8, "learning_rate": 9.996824826965841e-05, "loss": 2.9961, "step": 447 }, { "epoch": 0.8, "learning_rate": 9.996585300715116e-05, "loss": 2.4688, "step": 448 }, { "epoch": 0.8, "learning_rate": 9.99633706845614e-05, "loss": 1.752, "step": 449 }, { "epoch": 0.8, "learning_rate": 9.996080130621439e-05, "loss": 2.6562, "step": 450 }, { "epoch": 0.81, "learning_rate": 9.995814487658692e-05, "loss": 2.7695, "step": 451 }, { "epoch": 0.81, "learning_rate": 9.995540140030759e-05, "loss": 2.1562, "step": 452 }, { "epoch": 0.81, "learning_rate": 9.995257088215656e-05, "loss": 3.0527, "step": 453 }, { "epoch": 0.81, "learning_rate": 9.994965332706573e-05, "loss": 2.3184, "step": 454 }, { "epoch": 0.81, "learning_rate": 9.994664874011863e-05, "loss": 2.6445, "step": 455 }, { "epoch": 0.81, "learning_rate": 9.994355712655041e-05, "loss": 1.6992, "step": 456 }, { "epoch": 0.82, "learning_rate": 9.994037849174787e-05, "loss": 2.1758, "step": 457 }, { "epoch": 0.82, "learning_rate": 9.993711284124942e-05, "loss": 1.5977, "step": 458 }, { "epoch": 0.82, "learning_rate": 9.993376018074517e-05, "loss": 2.0508, "step": 459 }, { "epoch": 0.82, "learning_rate": 9.993032051607669e-05, "loss": 1.9609, "step": 460 }, { "epoch": 0.82, "learning_rate": 9.992679385323726e-05, "loss": 2.373, "step": 461 }, { "epoch": 0.82, "learning_rate": 9.99231801983717e-05, "loss": 2.377, "step": 462 }, { "epoch": 0.83, "learning_rate": 9.991947955777643e-05, "loss": 2.3301, "step": 463 }, { "epoch": 0.83, "learning_rate": 9.991569193789937e-05, "loss": 1.6982, "step": 464 }, { "epoch": 0.83, "learning_rate": 9.99118173453401e-05, "loss": 2.6387, "step": 465 }, { "epoch": 0.83, "learning_rate": 9.990785578684962e-05, "loss": 1.9414, "step": 466 }, { "epoch": 0.83, "learning_rate": 9.990380726933055e-05, "loss": 1.6963, "step": 467 }, { "epoch": 0.84, "learning_rate": 9.989967179983698e-05, "loss": 1.7988, "step": 468 }, { "epoch": 0.84, "learning_rate": 9.989544938557452e-05, "loss": 1.9971, "step": 469 }, { "epoch": 0.84, "learning_rate": 9.989114003390027e-05, "loss": 2.0, "step": 470 }, { "epoch": 0.84, "learning_rate": 9.98867437523228e-05, "loss": 2.5332, "step": 471 }, { "epoch": 0.84, "learning_rate": 9.988226054850218e-05, "loss": 2.4355, "step": 472 }, { "epoch": 0.84, "learning_rate": 9.987769043024988e-05, "loss": 1.6562, "step": 473 }, { "epoch": 0.85, "learning_rate": 9.987303340552884e-05, "loss": 2.125, "step": 474 }, { "epoch": 0.85, "learning_rate": 9.986828948245343e-05, "loss": 1.9004, "step": 475 }, { "epoch": 0.85, "learning_rate": 9.986345866928941e-05, "loss": 2.0039, "step": 476 }, { "epoch": 0.85, "learning_rate": 9.985854097445395e-05, "loss": 2.0664, "step": 477 }, { "epoch": 0.85, "learning_rate": 9.985353640651562e-05, "loss": 2.1211, "step": 478 }, { "epoch": 0.86, "learning_rate": 9.984844497419432e-05, "loss": 2.3281, "step": 479 }, { "epoch": 0.86, "learning_rate": 9.984326668636131e-05, "loss": 1.9531, "step": 480 }, { "epoch": 0.86, "learning_rate": 9.98380015520392e-05, "loss": 2.0156, "step": 481 }, { "epoch": 0.86, "learning_rate": 9.983264958040194e-05, "loss": 2.4102, "step": 482 }, { "epoch": 0.86, "learning_rate": 9.982721078077473e-05, "loss": 2.6875, "step": 483 }, { "epoch": 0.86, "learning_rate": 9.98216851626341e-05, "loss": 2.5156, "step": 484 }, { "epoch": 0.87, "learning_rate": 9.981607273560785e-05, "loss": 2.2441, "step": 485 }, { "epoch": 0.87, "learning_rate": 9.981037350947503e-05, "loss": 2.709, "step": 486 }, { "epoch": 0.87, "learning_rate": 9.980458749416587e-05, "loss": 1.9854, "step": 487 }, { "epoch": 0.87, "learning_rate": 9.979871469976196e-05, "loss": 2.207, "step": 488 }, { "epoch": 0.87, "learning_rate": 9.979275513649595e-05, "loss": 2.3945, "step": 489 }, { "epoch": 0.88, "learning_rate": 9.978670881475172e-05, "loss": 2.0918, "step": 490 }, { "epoch": 0.88, "learning_rate": 9.978057574506439e-05, "loss": 2.6992, "step": 491 }, { "epoch": 0.88, "learning_rate": 9.977435593812012e-05, "loss": 2.918, "step": 492 }, { "epoch": 0.88, "learning_rate": 9.976804940475626e-05, "loss": 1.875, "step": 493 }, { "epoch": 0.88, "learning_rate": 9.976165615596128e-05, "loss": 2.0156, "step": 494 }, { "epoch": 0.88, "learning_rate": 9.97551762028747e-05, "loss": 2.5742, "step": 495 }, { "epoch": 0.89, "learning_rate": 9.974860955678715e-05, "loss": 2.3945, "step": 496 }, { "epoch": 0.89, "learning_rate": 9.974195622914028e-05, "loss": 2.2344, "step": 497 }, { "epoch": 0.89, "learning_rate": 9.973521623152682e-05, "loss": 2.293, "step": 498 }, { "epoch": 0.89, "learning_rate": 9.972838957569046e-05, "loss": 2.3594, "step": 499 }, { "epoch": 0.89, "learning_rate": 9.972147627352592e-05, "loss": 2.5703, "step": 500 }, { "epoch": 0.89, "learning_rate": 9.971447633707889e-05, "loss": 2.207, "step": 501 }, { "epoch": 0.9, "learning_rate": 9.970738977854597e-05, "loss": 1.9883, "step": 502 }, { "epoch": 0.9, "learning_rate": 9.970021661027476e-05, "loss": 2.5117, "step": 503 }, { "epoch": 0.9, "learning_rate": 9.96929568447637e-05, "loss": 2.7793, "step": 504 }, { "epoch": 0.9, "learning_rate": 9.968561049466214e-05, "loss": 2.3828, "step": 505 }, { "epoch": 0.9, "learning_rate": 9.967817757277032e-05, "loss": 2.3516, "step": 506 }, { "epoch": 0.91, "learning_rate": 9.967065809203928e-05, "loss": 2.3789, "step": 507 }, { "epoch": 0.91, "learning_rate": 9.966305206557091e-05, "loss": 2.4727, "step": 508 }, { "epoch": 0.91, "learning_rate": 9.96553595066179e-05, "loss": 1.9062, "step": 509 }, { "epoch": 0.91, "learning_rate": 9.964758042858367e-05, "loss": 2.3125, "step": 510 }, { "epoch": 0.91, "learning_rate": 9.963971484502246e-05, "loss": 1.6787, "step": 511 }, { "epoch": 0.91, "learning_rate": 9.963176276963915e-05, "loss": 2.0039, "step": 512 }, { "epoch": 0.92, "learning_rate": 9.962372421628941e-05, "loss": 2.5566, "step": 513 }, { "epoch": 0.92, "learning_rate": 9.961559919897954e-05, "loss": 1.835, "step": 514 }, { "epoch": 0.92, "learning_rate": 9.960738773186648e-05, "loss": 2.6094, "step": 515 }, { "epoch": 0.92, "learning_rate": 9.959908982925784e-05, "loss": 2.7109, "step": 516 }, { "epoch": 0.92, "learning_rate": 9.959070550561179e-05, "loss": 2.8281, "step": 517 }, { "epoch": 0.93, "learning_rate": 9.958223477553714e-05, "loss": 2.4844, "step": 518 }, { "epoch": 0.93, "learning_rate": 9.95736776537932e-05, "loss": 1.9873, "step": 519 }, { "epoch": 0.93, "learning_rate": 9.956503415528984e-05, "loss": 2.2305, "step": 520 }, { "epoch": 0.93, "learning_rate": 9.955630429508741e-05, "loss": 2.5215, "step": 521 }, { "epoch": 0.93, "learning_rate": 9.954748808839674e-05, "loss": 1.5635, "step": 522 }, { "epoch": 0.93, "learning_rate": 9.953858555057913e-05, "loss": 1.9844, "step": 523 }, { "epoch": 0.94, "learning_rate": 9.952959669714627e-05, "loss": 2.207, "step": 524 }, { "epoch": 0.94, "learning_rate": 9.952052154376026e-05, "loss": 2.5723, "step": 525 }, { "epoch": 0.94, "learning_rate": 9.951136010623359e-05, "loss": 1.6797, "step": 526 }, { "epoch": 0.94, "learning_rate": 9.950211240052907e-05, "loss": 1.6309, "step": 527 }, { "epoch": 0.94, "learning_rate": 9.94927784427598e-05, "loss": 2.3086, "step": 528 }, { "epoch": 0.94, "learning_rate": 9.948335824918922e-05, "loss": 1.7793, "step": 529 }, { "epoch": 0.95, "learning_rate": 9.947385183623098e-05, "loss": 2.5312, "step": 530 }, { "epoch": 0.95, "learning_rate": 9.946425922044897e-05, "loss": 2.252, "step": 531 }, { "epoch": 0.95, "learning_rate": 9.94545804185573e-05, "loss": 2.0156, "step": 532 }, { "epoch": 0.95, "learning_rate": 9.944481544742022e-05, "loss": 2.2246, "step": 533 }, { "epoch": 0.95, "learning_rate": 9.943496432405214e-05, "loss": 2.1289, "step": 534 }, { "epoch": 0.96, "learning_rate": 9.942502706561757e-05, "loss": 2.1992, "step": 535 }, { "epoch": 0.96, "learning_rate": 9.941500368943111e-05, "loss": 2.0918, "step": 536 }, { "epoch": 0.96, "learning_rate": 9.94048942129574e-05, "loss": 2.6543, "step": 537 }, { "epoch": 0.96, "learning_rate": 9.93946986538111e-05, "loss": 2.2539, "step": 538 }, { "epoch": 0.96, "learning_rate": 9.938441702975689e-05, "loss": 2.4199, "step": 539 }, { "epoch": 0.96, "learning_rate": 9.937404935870938e-05, "loss": 1.9355, "step": 540 }, { "epoch": 0.97, "learning_rate": 9.936359565873308e-05, "loss": 2.4004, "step": 541 }, { "epoch": 0.97, "learning_rate": 9.935305594804247e-05, "loss": 1.9609, "step": 542 }, { "epoch": 0.97, "learning_rate": 9.934243024500186e-05, "loss": 2.0566, "step": 543 }, { "epoch": 0.97, "learning_rate": 9.933171856812532e-05, "loss": 2.9453, "step": 544 }, { "epoch": 0.97, "learning_rate": 9.932092093607686e-05, "loss": 2.1641, "step": 545 }, { "epoch": 0.97, "learning_rate": 9.931003736767013e-05, "loss": 2.0488, "step": 546 }, { "epoch": 0.98, "learning_rate": 9.92990678818686e-05, "loss": 2.2266, "step": 547 }, { "epoch": 0.98, "learning_rate": 9.928801249778542e-05, "loss": 2.8203, "step": 548 }, { "epoch": 0.98, "learning_rate": 9.927687123468334e-05, "loss": 2.3398, "step": 549 }, { "epoch": 0.98, "learning_rate": 9.926564411197488e-05, "loss": 1.7266, "step": 550 }, { "epoch": 0.98, "learning_rate": 9.925433114922203e-05, "loss": 2.3457, "step": 551 }, { "epoch": 0.99, "learning_rate": 9.924293236613641e-05, "loss": 1.9941, "step": 552 }, { "epoch": 0.99, "learning_rate": 9.923144778257918e-05, "loss": 2.8945, "step": 553 }, { "epoch": 0.99, "learning_rate": 9.921987741856098e-05, "loss": 1.5039, "step": 554 }, { "epoch": 0.99, "learning_rate": 9.92082212942419e-05, "loss": 2.1973, "step": 555 }, { "epoch": 0.99, "learning_rate": 9.919647942993148e-05, "loss": 1.834, "step": 556 }, { "epoch": 0.99, "learning_rate": 9.918465184608867e-05, "loss": 2.0625, "step": 557 }, { "epoch": 1.0, "learning_rate": 9.91727385633217e-05, "loss": 2.6973, "step": 558 }, { "epoch": 1.0, "learning_rate": 9.916073960238821e-05, "loss": 1.8242, "step": 559 }, { "epoch": 1.0, "learning_rate": 9.91486549841951e-05, "loss": 1.9531, "step": 560 }, { "epoch": 1.0, "learning_rate": 9.913648472979848e-05, "loss": 2.2031, "step": 561 }, { "epoch": 1.0, "learning_rate": 9.912422886040369e-05, "loss": 2.1426, "step": 562 }, { "epoch": 1.01, "learning_rate": 9.91118873973653e-05, "loss": 2.457, "step": 563 }, { "epoch": 1.01, "learning_rate": 9.909946036218694e-05, "loss": 1.9434, "step": 564 }, { "epoch": 1.01, "learning_rate": 9.908694777652136e-05, "loss": 2.5625, "step": 565 }, { "epoch": 1.01, "learning_rate": 9.90743496621704e-05, "loss": 2.375, "step": 566 }, { "epoch": 1.01, "learning_rate": 9.906166604108493e-05, "loss": 2.5195, "step": 567 }, { "epoch": 1.01, "learning_rate": 9.904889693536474e-05, "loss": 1.8711, "step": 568 }, { "epoch": 1.02, "learning_rate": 9.903604236725865e-05, "loss": 2.1387, "step": 569 }, { "epoch": 1.02, "learning_rate": 9.902310235916435e-05, "loss": 2.1445, "step": 570 }, { "epoch": 1.02, "learning_rate": 9.901007693362836e-05, "loss": 2.4902, "step": 571 }, { "epoch": 1.02, "learning_rate": 9.899696611334611e-05, "loss": 2.3164, "step": 572 }, { "epoch": 1.02, "learning_rate": 9.898376992116179e-05, "loss": 1.9424, "step": 573 }, { "epoch": 1.02, "learning_rate": 9.89704883800683e-05, "loss": 1.7041, "step": 574 }, { "epoch": 1.03, "learning_rate": 9.89571215132073e-05, "loss": 1.7139, "step": 575 }, { "epoch": 1.03, "learning_rate": 9.894366934386913e-05, "loss": 1.8809, "step": 576 }, { "epoch": 1.03, "learning_rate": 9.89301318954927e-05, "loss": 2.3242, "step": 577 }, { "epoch": 1.03, "learning_rate": 9.891650919166557e-05, "loss": 2.2539, "step": 578 }, { "epoch": 1.03, "learning_rate": 9.890280125612382e-05, "loss": 1.9219, "step": 579 }, { "epoch": 1.04, "learning_rate": 9.888900811275204e-05, "loss": 2.748, "step": 580 }, { "epoch": 1.04, "learning_rate": 9.887512978558329e-05, "loss": 1.6855, "step": 581 }, { "epoch": 1.04, "learning_rate": 9.886116629879905e-05, "loss": 2.6055, "step": 582 }, { "epoch": 1.04, "learning_rate": 9.884711767672916e-05, "loss": 2.1016, "step": 583 }, { "epoch": 1.04, "learning_rate": 9.883298394385186e-05, "loss": 1.9258, "step": 584 }, { "epoch": 1.04, "learning_rate": 9.881876512479362e-05, "loss": 2.4492, "step": 585 }, { "epoch": 1.05, "learning_rate": 9.88044612443292e-05, "loss": 1.5957, "step": 586 }, { "epoch": 1.05, "learning_rate": 9.879007232738157e-05, "loss": 2.2324, "step": 587 }, { "epoch": 1.05, "learning_rate": 9.877559839902184e-05, "loss": 2.5059, "step": 588 }, { "epoch": 1.05, "learning_rate": 9.876103948446927e-05, "loss": 1.999, "step": 589 }, { "epoch": 1.05, "learning_rate": 9.874639560909117e-05, "loss": 1.7969, "step": 590 }, { "epoch": 1.06, "learning_rate": 9.873166679840295e-05, "loss": 2.1777, "step": 591 }, { "epoch": 1.06, "learning_rate": 9.871685307806795e-05, "loss": 1.9609, "step": 592 }, { "epoch": 1.06, "learning_rate": 9.870195447389745e-05, "loss": 1.8926, "step": 593 }, { "epoch": 1.06, "learning_rate": 9.868697101185066e-05, "loss": 1.8965, "step": 594 }, { "epoch": 1.06, "learning_rate": 9.867190271803465e-05, "loss": 2.5293, "step": 595 }, { "epoch": 1.06, "learning_rate": 9.865674961870428e-05, "loss": 2.6953, "step": 596 }, { "epoch": 1.07, "learning_rate": 9.864151174026218e-05, "loss": 2.3359, "step": 597 }, { "epoch": 1.07, "learning_rate": 9.862618910925872e-05, "loss": 2.1445, "step": 598 }, { "epoch": 1.07, "learning_rate": 9.86107817523919e-05, "loss": 2.4414, "step": 599 }, { "epoch": 1.07, "learning_rate": 9.859528969650738e-05, "loss": 2.1445, "step": 600 }, { "epoch": 1.07, "learning_rate": 9.85797129685984e-05, "loss": 1.3672, "step": 601 }, { "epoch": 1.07, "learning_rate": 9.85640515958057e-05, "loss": 2.1758, "step": 602 }, { "epoch": 1.08, "learning_rate": 9.854830560541752e-05, "loss": 2.6992, "step": 603 }, { "epoch": 1.08, "learning_rate": 9.853247502486958e-05, "loss": 1.9355, "step": 604 }, { "epoch": 1.08, "learning_rate": 9.85165598817449e-05, "loss": 2.0059, "step": 605 }, { "epoch": 1.08, "learning_rate": 9.850056020377392e-05, "loss": 2.2266, "step": 606 }, { "epoch": 1.08, "learning_rate": 9.848447601883435e-05, "loss": 1.9473, "step": 607 }, { "epoch": 1.09, "learning_rate": 9.846830735495112e-05, "loss": 2.375, "step": 608 }, { "epoch": 1.09, "learning_rate": 9.845205424029639e-05, "loss": 1.8516, "step": 609 }, { "epoch": 1.09, "learning_rate": 9.843571670318943e-05, "loss": 1.8906, "step": 610 }, { "epoch": 1.09, "learning_rate": 9.841929477209665e-05, "loss": 2.25, "step": 611 }, { "epoch": 1.09, "learning_rate": 9.840278847563146e-05, "loss": 1.8574, "step": 612 }, { "epoch": 1.09, "learning_rate": 9.838619784255434e-05, "loss": 2.3516, "step": 613 }, { "epoch": 1.1, "learning_rate": 9.836952290177261e-05, "loss": 1.9941, "step": 614 }, { "epoch": 1.1, "learning_rate": 9.835276368234059e-05, "loss": 1.7578, "step": 615 }, { "epoch": 1.1, "learning_rate": 9.833592021345937e-05, "loss": 1.7246, "step": 616 }, { "epoch": 1.1, "learning_rate": 9.83189925244769e-05, "loss": 2.3145, "step": 617 }, { "epoch": 1.1, "learning_rate": 9.830198064488782e-05, "loss": 2.1309, "step": 618 }, { "epoch": 1.11, "learning_rate": 9.82848846043335e-05, "loss": 1.7129, "step": 619 }, { "epoch": 1.11, "learning_rate": 9.826770443260193e-05, "loss": 1.9219, "step": 620 }, { "epoch": 1.11, "learning_rate": 9.825044015962772e-05, "loss": 2.0391, "step": 621 }, { "epoch": 1.11, "learning_rate": 9.823309181549194e-05, "loss": 2.2676, "step": 622 }, { "epoch": 1.11, "learning_rate": 9.821565943042225e-05, "loss": 1.998, "step": 623 }, { "epoch": 1.11, "learning_rate": 9.819814303479267e-05, "loss": 2.832, "step": 624 }, { "epoch": 1.12, "learning_rate": 9.818054265912362e-05, "loss": 2.2031, "step": 625 }, { "epoch": 1.12, "learning_rate": 9.816285833408186e-05, "loss": 2.4453, "step": 626 }, { "epoch": 1.12, "learning_rate": 9.81450900904804e-05, "loss": 2.4102, "step": 627 }, { "epoch": 1.12, "learning_rate": 9.812723795927848e-05, "loss": 2.5586, "step": 628 }, { "epoch": 1.12, "learning_rate": 9.810930197158148e-05, "loss": 1.9746, "step": 629 }, { "epoch": 1.12, "learning_rate": 9.809128215864097e-05, "loss": 2.5371, "step": 630 }, { "epoch": 1.13, "learning_rate": 9.807317855185445e-05, "loss": 1.7852, "step": 631 }, { "epoch": 1.13, "learning_rate": 9.805499118276555e-05, "loss": 2.0508, "step": 632 }, { "epoch": 1.13, "learning_rate": 9.803672008306377e-05, "loss": 2.375, "step": 633 }, { "epoch": 1.13, "learning_rate": 9.801836528458452e-05, "loss": 2.1758, "step": 634 }, { "epoch": 1.13, "learning_rate": 9.799992681930905e-05, "loss": 2.4453, "step": 635 }, { "epoch": 1.14, "learning_rate": 9.798140471936437e-05, "loss": 2.0117, "step": 636 }, { "epoch": 1.14, "learning_rate": 9.796279901702325e-05, "loss": 1.9219, "step": 637 }, { "epoch": 1.14, "learning_rate": 9.794410974470411e-05, "loss": 2.1465, "step": 638 }, { "epoch": 1.14, "learning_rate": 9.792533693497095e-05, "loss": 2.5078, "step": 639 }, { "epoch": 1.14, "learning_rate": 9.79064806205334e-05, "loss": 2.1211, "step": 640 }, { "epoch": 1.14, "learning_rate": 9.788754083424652e-05, "loss": 2.2188, "step": 641 }, { "epoch": 1.15, "learning_rate": 9.786851760911084e-05, "loss": 1.9131, "step": 642 }, { "epoch": 1.15, "learning_rate": 9.784941097827227e-05, "loss": 1.9492, "step": 643 }, { "epoch": 1.15, "learning_rate": 9.783022097502204e-05, "loss": 1.9531, "step": 644 }, { "epoch": 1.15, "learning_rate": 9.781094763279666e-05, "loss": 2.9766, "step": 645 }, { "epoch": 1.15, "learning_rate": 9.77915909851778e-05, "loss": 2.2031, "step": 646 }, { "epoch": 1.16, "learning_rate": 9.777215106589236e-05, "loss": 2.1445, "step": 647 }, { "epoch": 1.16, "learning_rate": 9.77526279088123e-05, "loss": 2.5137, "step": 648 }, { "epoch": 1.16, "learning_rate": 9.773302154795459e-05, "loss": 2.4023, "step": 649 }, { "epoch": 1.16, "learning_rate": 9.771333201748117e-05, "loss": 2.2988, "step": 650 }, { "epoch": 1.16, "learning_rate": 9.76935593516989e-05, "loss": 1.7578, "step": 651 }, { "epoch": 1.16, "learning_rate": 9.767370358505957e-05, "loss": 2.4121, "step": 652 }, { "epoch": 1.17, "learning_rate": 9.765376475215966e-05, "loss": 2.3594, "step": 653 }, { "epoch": 1.17, "learning_rate": 9.763374288774044e-05, "loss": 1.7969, "step": 654 }, { "epoch": 1.17, "learning_rate": 9.76136380266878e-05, "loss": 1.5801, "step": 655 }, { "epoch": 1.17, "learning_rate": 9.759345020403232e-05, "loss": 2.1758, "step": 656 }, { "epoch": 1.17, "learning_rate": 9.757317945494909e-05, "loss": 2.0898, "step": 657 }, { "epoch": 1.18, "learning_rate": 9.755282581475769e-05, "loss": 2.1523, "step": 658 }, { "epoch": 1.18, "learning_rate": 9.753238931892211e-05, "loss": 2.3516, "step": 659 }, { "epoch": 1.18, "learning_rate": 9.751187000305076e-05, "loss": 1.9414, "step": 660 }, { "epoch": 1.18, "learning_rate": 9.749126790289628e-05, "loss": 1.998, "step": 661 }, { "epoch": 1.18, "learning_rate": 9.747058305435565e-05, "loss": 2.3672, "step": 662 }, { "epoch": 1.18, "learning_rate": 9.744981549346996e-05, "loss": 1.8398, "step": 663 }, { "epoch": 1.19, "learning_rate": 9.742896525642442e-05, "loss": 1.9814, "step": 664 }, { "epoch": 1.19, "learning_rate": 9.74080323795483e-05, "loss": 2.6348, "step": 665 }, { "epoch": 1.19, "learning_rate": 9.738701689931488e-05, "loss": 1.9102, "step": 666 }, { "epoch": 1.19, "learning_rate": 9.736591885234137e-05, "loss": 2.4883, "step": 667 }, { "epoch": 1.19, "learning_rate": 9.734473827538882e-05, "loss": 2.0234, "step": 668 }, { "epoch": 1.19, "learning_rate": 9.732347520536207e-05, "loss": 2.3906, "step": 669 }, { "epoch": 1.2, "learning_rate": 9.730212967930974e-05, "loss": 2.418, "step": 670 }, { "epoch": 1.2, "learning_rate": 9.728070173442409e-05, "loss": 2.3945, "step": 671 }, { "epoch": 1.2, "learning_rate": 9.725919140804099e-05, "loss": 2.4727, "step": 672 }, { "epoch": 1.2, "learning_rate": 9.723759873763983e-05, "loss": 2.1016, "step": 673 }, { "epoch": 1.2, "learning_rate": 9.721592376084355e-05, "loss": 2.4258, "step": 674 }, { "epoch": 1.21, "learning_rate": 9.719416651541839e-05, "loss": 1.8145, "step": 675 }, { "epoch": 1.21, "learning_rate": 9.717232703927401e-05, "loss": 2.2305, "step": 676 }, { "epoch": 1.21, "learning_rate": 9.715040537046336e-05, "loss": 2.2598, "step": 677 }, { "epoch": 1.21, "learning_rate": 9.712840154718252e-05, "loss": 2.0742, "step": 678 }, { "epoch": 1.21, "learning_rate": 9.710631560777081e-05, "loss": 2.125, "step": 679 }, { "epoch": 1.21, "learning_rate": 9.708414759071059e-05, "loss": 2.7266, "step": 680 }, { "epoch": 1.22, "learning_rate": 9.706189753462718e-05, "loss": 1.748, "step": 681 }, { "epoch": 1.22, "learning_rate": 9.703956547828891e-05, "loss": 1.8887, "step": 682 }, { "epoch": 1.22, "learning_rate": 9.701715146060699e-05, "loss": 1.9746, "step": 683 }, { "epoch": 1.22, "learning_rate": 9.69946555206354e-05, "loss": 2.1465, "step": 684 }, { "epoch": 1.22, "learning_rate": 9.697207769757089e-05, "loss": 1.8252, "step": 685 }, { "epoch": 1.23, "learning_rate": 9.694941803075283e-05, "loss": 2.4004, "step": 686 }, { "epoch": 1.23, "learning_rate": 9.692667655966327e-05, "loss": 1.7305, "step": 687 }, { "epoch": 1.23, "learning_rate": 9.690385332392675e-05, "loss": 2.4727, "step": 688 }, { "epoch": 1.23, "learning_rate": 9.688094836331028e-05, "loss": 1.9717, "step": 689 }, { "epoch": 1.23, "learning_rate": 9.685796171772328e-05, "loss": 1.4219, "step": 690 }, { "epoch": 1.23, "learning_rate": 9.683489342721748e-05, "loss": 1.4443, "step": 691 }, { "epoch": 1.24, "learning_rate": 9.681174353198687e-05, "loss": 1.8955, "step": 692 }, { "epoch": 1.24, "learning_rate": 9.678851207236763e-05, "loss": 2.4102, "step": 693 }, { "epoch": 1.24, "learning_rate": 9.676519908883809e-05, "loss": 1.9824, "step": 694 }, { "epoch": 1.24, "learning_rate": 9.674180462201856e-05, "loss": 2.2773, "step": 695 }, { "epoch": 1.24, "learning_rate": 9.671832871267139e-05, "loss": 1.9561, "step": 696 }, { "epoch": 1.24, "learning_rate": 9.66947714017008e-05, "loss": 2.0234, "step": 697 }, { "epoch": 1.25, "learning_rate": 9.667113273015282e-05, "loss": 2.291, "step": 698 }, { "epoch": 1.25, "learning_rate": 9.664741273921532e-05, "loss": 2.457, "step": 699 }, { "epoch": 1.25, "learning_rate": 9.662361147021779e-05, "loss": 1.6797, "step": 700 }, { "epoch": 1.25, "learning_rate": 9.659972896463136e-05, "loss": 2.2852, "step": 701 }, { "epoch": 1.25, "learning_rate": 9.657576526406872e-05, "loss": 2.2012, "step": 702 }, { "epoch": 1.26, "learning_rate": 9.655172041028399e-05, "loss": 2.1523, "step": 703 }, { "epoch": 1.26, "learning_rate": 9.652759444517277e-05, "loss": 2.0469, "step": 704 }, { "epoch": 1.26, "learning_rate": 9.65033874107719e-05, "loss": 2.5195, "step": 705 }, { "epoch": 1.26, "learning_rate": 9.647909934925951e-05, "loss": 1.8828, "step": 706 }, { "epoch": 1.26, "learning_rate": 9.645473030295496e-05, "loss": 2.041, "step": 707 }, { "epoch": 1.26, "learning_rate": 9.64302803143186e-05, "loss": 1.8125, "step": 708 }, { "epoch": 1.27, "learning_rate": 9.640574942595196e-05, "loss": 1.6826, "step": 709 }, { "epoch": 1.27, "learning_rate": 9.63811376805974e-05, "loss": 1.7695, "step": 710 }, { "epoch": 1.27, "learning_rate": 9.635644512113824e-05, "loss": 1.624, "step": 711 }, { "epoch": 1.27, "learning_rate": 9.633167179059858e-05, "loss": 1.9443, "step": 712 }, { "epoch": 1.27, "learning_rate": 9.630681773214328e-05, "loss": 2.1211, "step": 713 }, { "epoch": 1.27, "learning_rate": 9.628188298907782e-05, "loss": 1.957, "step": 714 }, { "epoch": 1.28, "learning_rate": 9.62568676048483e-05, "loss": 2.2598, "step": 715 }, { "epoch": 1.28, "learning_rate": 9.623177162304131e-05, "loss": 2.5195, "step": 716 }, { "epoch": 1.28, "learning_rate": 9.62065950873839e-05, "loss": 2.6992, "step": 717 }, { "epoch": 1.28, "learning_rate": 9.618133804174341e-05, "loss": 2.2188, "step": 718 }, { "epoch": 1.28, "learning_rate": 9.615600053012754e-05, "loss": 2.1953, "step": 719 }, { "epoch": 1.29, "learning_rate": 9.613058259668416e-05, "loss": 2.5625, "step": 720 }, { "epoch": 1.29, "learning_rate": 9.610508428570122e-05, "loss": 2.0254, "step": 721 }, { "epoch": 1.29, "learning_rate": 9.607950564160682e-05, "loss": 2.4805, "step": 722 }, { "epoch": 1.29, "learning_rate": 9.605384670896893e-05, "loss": 1.8203, "step": 723 }, { "epoch": 1.29, "learning_rate": 9.602810753249549e-05, "loss": 1.7246, "step": 724 }, { "epoch": 1.29, "learning_rate": 9.60022881570342e-05, "loss": 2.5742, "step": 725 }, { "epoch": 1.3, "learning_rate": 9.597638862757255e-05, "loss": 1.8926, "step": 726 }, { "epoch": 1.3, "learning_rate": 9.595040898923762e-05, "loss": 1.9727, "step": 727 }, { "epoch": 1.3, "learning_rate": 9.592434928729616e-05, "loss": 1.8301, "step": 728 }, { "epoch": 1.3, "learning_rate": 9.589820956715434e-05, "loss": 2.4883, "step": 729 }, { "epoch": 1.3, "learning_rate": 9.587198987435782e-05, "loss": 1.623, "step": 730 }, { "epoch": 1.31, "learning_rate": 9.584569025459155e-05, "loss": 2.8066, "step": 731 }, { "epoch": 1.31, "learning_rate": 9.581931075367979e-05, "loss": 1.457, "step": 732 }, { "epoch": 1.31, "learning_rate": 9.579285141758591e-05, "loss": 1.7754, "step": 733 }, { "epoch": 1.31, "learning_rate": 9.576631229241249e-05, "loss": 2.1523, "step": 734 }, { "epoch": 1.31, "learning_rate": 9.573969342440106e-05, "loss": 2.1719, "step": 735 }, { "epoch": 1.31, "learning_rate": 9.57129948599321e-05, "loss": 2.2539, "step": 736 }, { "epoch": 1.32, "learning_rate": 9.568621664552497e-05, "loss": 2.2715, "step": 737 }, { "epoch": 1.32, "learning_rate": 9.565935882783783e-05, "loss": 1.666, "step": 738 }, { "epoch": 1.32, "learning_rate": 9.563242145366749e-05, "loss": 1.7363, "step": 739 }, { "epoch": 1.32, "learning_rate": 9.56054045699494e-05, "loss": 1.9824, "step": 740 }, { "epoch": 1.32, "learning_rate": 9.557830822375755e-05, "loss": 1.6738, "step": 741 }, { "epoch": 1.32, "learning_rate": 9.555113246230442e-05, "loss": 2.0117, "step": 742 }, { "epoch": 1.33, "learning_rate": 9.552387733294081e-05, "loss": 2.6641, "step": 743 }, { "epoch": 1.33, "learning_rate": 9.549654288315581e-05, "loss": 2.291, "step": 744 }, { "epoch": 1.33, "learning_rate": 9.546912916057676e-05, "loss": 2.2969, "step": 745 }, { "epoch": 1.33, "learning_rate": 9.544163621296906e-05, "loss": 1.8242, "step": 746 }, { "epoch": 1.33, "learning_rate": 9.541406408823626e-05, "loss": 2.3496, "step": 747 }, { "epoch": 1.34, "learning_rate": 9.538641283441974e-05, "loss": 1.8965, "step": 748 }, { "epoch": 1.34, "learning_rate": 9.535868249969882e-05, "loss": 2.3008, "step": 749 }, { "epoch": 1.34, "learning_rate": 9.533087313239064e-05, "loss": 2.6836, "step": 750 }, { "epoch": 1.34, "learning_rate": 9.530298478094997e-05, "loss": 2.127, "step": 751 }, { "epoch": 1.34, "learning_rate": 9.527501749396924e-05, "loss": 2.1465, "step": 752 }, { "epoch": 1.34, "learning_rate": 9.524697132017844e-05, "loss": 2.1582, "step": 753 }, { "epoch": 1.35, "learning_rate": 9.521884630844498e-05, "loss": 1.9414, "step": 754 }, { "epoch": 1.35, "learning_rate": 9.51906425077736e-05, "loss": 1.8008, "step": 755 }, { "epoch": 1.35, "learning_rate": 9.516235996730645e-05, "loss": 1.5156, "step": 756 }, { "epoch": 1.35, "learning_rate": 9.513399873632273e-05, "loss": 2.7656, "step": 757 }, { "epoch": 1.35, "learning_rate": 9.510555886423881e-05, "loss": 1.5811, "step": 758 }, { "epoch": 1.36, "learning_rate": 9.507704040060812e-05, "loss": 1.7871, "step": 759 }, { "epoch": 1.36, "learning_rate": 9.504844339512095e-05, "loss": 1.9961, "step": 760 }, { "epoch": 1.36, "learning_rate": 9.501976789760453e-05, "loss": 1.8086, "step": 761 }, { "epoch": 1.36, "learning_rate": 9.499101395802277e-05, "loss": 2.2383, "step": 762 }, { "epoch": 1.36, "learning_rate": 9.496218162647629e-05, "loss": 2.2227, "step": 763 }, { "epoch": 1.36, "learning_rate": 9.493327095320232e-05, "loss": 1.7344, "step": 764 }, { "epoch": 1.37, "learning_rate": 9.490428198857457e-05, "loss": 2.7363, "step": 765 }, { "epoch": 1.37, "learning_rate": 9.487521478310316e-05, "loss": 2.6738, "step": 766 }, { "epoch": 1.37, "learning_rate": 9.484606938743454e-05, "loss": 2.002, "step": 767 }, { "epoch": 1.37, "learning_rate": 9.481684585235144e-05, "loss": 2.3555, "step": 768 }, { "epoch": 1.37, "learning_rate": 9.478754422877266e-05, "loss": 2.4453, "step": 769 }, { "epoch": 1.38, "learning_rate": 9.475816456775313e-05, "loss": 2.418, "step": 770 }, { "epoch": 1.38, "learning_rate": 9.47287069204837e-05, "loss": 2.1758, "step": 771 }, { "epoch": 1.38, "learning_rate": 9.469917133829114e-05, "loss": 1.3818, "step": 772 }, { "epoch": 1.38, "learning_rate": 9.4669557872638e-05, "loss": 1.5996, "step": 773 }, { "epoch": 1.38, "learning_rate": 9.463986657512253e-05, "loss": 1.8975, "step": 774 }, { "epoch": 1.38, "learning_rate": 9.461009749747863e-05, "loss": 2.1523, "step": 775 }, { "epoch": 1.39, "learning_rate": 9.458025069157563e-05, "loss": 2.1172, "step": 776 }, { "epoch": 1.39, "learning_rate": 9.45503262094184e-05, "loss": 2.0586, "step": 777 }, { "epoch": 1.39, "learning_rate": 9.452032410314708e-05, "loss": 2.2969, "step": 778 }, { "epoch": 1.39, "learning_rate": 9.449024442503711e-05, "loss": 2.1738, "step": 779 }, { "epoch": 1.39, "learning_rate": 9.446008722749905e-05, "loss": 2.0684, "step": 780 }, { "epoch": 1.39, "learning_rate": 9.442985256307857e-05, "loss": 1.6699, "step": 781 }, { "epoch": 1.4, "learning_rate": 9.439954048445628e-05, "loss": 2.2852, "step": 782 }, { "epoch": 1.4, "learning_rate": 9.436915104444768e-05, "loss": 2.1055, "step": 783 }, { "epoch": 1.4, "learning_rate": 9.43386842960031e-05, "loss": 2.0449, "step": 784 }, { "epoch": 1.4, "learning_rate": 9.430814029220753e-05, "loss": 2.4414, "step": 785 }, { "epoch": 1.4, "learning_rate": 9.427751908628058e-05, "loss": 1.623, "step": 786 }, { "epoch": 1.41, "learning_rate": 9.42468207315764e-05, "loss": 1.8555, "step": 787 }, { "epoch": 1.41, "learning_rate": 9.421604528158354e-05, "loss": 2.2109, "step": 788 }, { "epoch": 1.41, "learning_rate": 9.418519278992489e-05, "loss": 2.5898, "step": 789 }, { "epoch": 1.41, "learning_rate": 9.415426331035753e-05, "loss": 1.873, "step": 790 }, { "epoch": 1.41, "learning_rate": 9.41232568967728e-05, "loss": 2.2773, "step": 791 }, { "epoch": 1.41, "learning_rate": 9.409217360319594e-05, "loss": 2.0996, "step": 792 }, { "epoch": 1.42, "learning_rate": 9.406101348378626e-05, "loss": 2.2266, "step": 793 }, { "epoch": 1.42, "learning_rate": 9.40297765928369e-05, "loss": 2.2539, "step": 794 }, { "epoch": 1.42, "learning_rate": 9.399846298477474e-05, "loss": 2.7266, "step": 795 }, { "epoch": 1.42, "learning_rate": 9.396707271416034e-05, "loss": 1.6816, "step": 796 }, { "epoch": 1.42, "learning_rate": 9.393560583568785e-05, "loss": 2.4453, "step": 797 }, { "epoch": 1.43, "learning_rate": 9.39040624041849e-05, "loss": 1.9248, "step": 798 }, { "epoch": 1.43, "learning_rate": 9.38724424746125e-05, "loss": 2.4922, "step": 799 }, { "epoch": 1.43, "learning_rate": 9.384074610206495e-05, "loss": 2.5742, "step": 800 }, { "epoch": 1.43, "learning_rate": 9.380897334176973e-05, "loss": 2.4102, "step": 801 }, { "epoch": 1.43, "learning_rate": 9.377712424908743e-05, "loss": 2.2207, "step": 802 }, { "epoch": 1.43, "learning_rate": 9.374519887951166e-05, "loss": 2.7324, "step": 803 }, { "epoch": 1.44, "learning_rate": 9.371319728866892e-05, "loss": 2.2695, "step": 804 }, { "epoch": 1.44, "learning_rate": 9.368111953231848e-05, "loss": 1.9932, "step": 805 }, { "epoch": 1.44, "learning_rate": 9.364896566635238e-05, "loss": 2.0332, "step": 806 }, { "epoch": 1.44, "learning_rate": 9.361673574679524e-05, "loss": 1.7256, "step": 807 }, { "epoch": 1.44, "learning_rate": 9.35844298298042e-05, "loss": 2.6719, "step": 808 }, { "epoch": 1.44, "learning_rate": 9.355204797166883e-05, "loss": 1.8105, "step": 809 }, { "epoch": 1.45, "learning_rate": 9.351959022881099e-05, "loss": 2.1484, "step": 810 }, { "epoch": 1.45, "learning_rate": 9.348705665778478e-05, "loss": 2.0742, "step": 811 }, { "epoch": 1.45, "learning_rate": 9.345444731527642e-05, "loss": 1.9219, "step": 812 }, { "epoch": 1.45, "learning_rate": 9.342176225810417e-05, "loss": 2.5742, "step": 813 }, { "epoch": 1.45, "learning_rate": 9.338900154321819e-05, "loss": 1.7051, "step": 814 }, { "epoch": 1.46, "learning_rate": 9.335616522770045e-05, "loss": 1.6074, "step": 815 }, { "epoch": 1.46, "learning_rate": 9.332325336876472e-05, "loss": 2.4883, "step": 816 }, { "epoch": 1.46, "learning_rate": 9.329026602375631e-05, "loss": 1.623, "step": 817 }, { "epoch": 1.46, "learning_rate": 9.32572032501521e-05, "loss": 2.4648, "step": 818 }, { "epoch": 1.46, "learning_rate": 9.32240651055604e-05, "loss": 2.2539, "step": 819 }, { "epoch": 1.46, "learning_rate": 9.319085164772082e-05, "loss": 2.1152, "step": 820 }, { "epoch": 1.47, "learning_rate": 9.315756293450422e-05, "loss": 2.4844, "step": 821 }, { "epoch": 1.47, "learning_rate": 9.312419902391256e-05, "loss": 2.2383, "step": 822 }, { "epoch": 1.47, "learning_rate": 9.309075997407886e-05, "loss": 2.5684, "step": 823 }, { "epoch": 1.47, "learning_rate": 9.305724584326702e-05, "loss": 2.3359, "step": 824 }, { "epoch": 1.47, "learning_rate": 9.302365668987179e-05, "loss": 2.0508, "step": 825 }, { "epoch": 1.48, "learning_rate": 9.298999257241863e-05, "loss": 2.0938, "step": 826 }, { "epoch": 1.48, "learning_rate": 9.295625354956361e-05, "loss": 2.5898, "step": 827 }, { "epoch": 1.48, "learning_rate": 9.292243968009331e-05, "loss": 1.8477, "step": 828 }, { "epoch": 1.48, "learning_rate": 9.288855102292475e-05, "loss": 2.3457, "step": 829 }, { "epoch": 1.48, "learning_rate": 9.285458763710524e-05, "loss": 2.2129, "step": 830 }, { "epoch": 1.48, "learning_rate": 9.282054958181229e-05, "loss": 2.1387, "step": 831 }, { "epoch": 1.49, "learning_rate": 9.278643691635351e-05, "loss": 2.252, "step": 832 }, { "epoch": 1.49, "learning_rate": 9.275224970016656e-05, "loss": 2.5742, "step": 833 }, { "epoch": 1.49, "learning_rate": 9.271798799281892e-05, "loss": 2.4805, "step": 834 }, { "epoch": 1.49, "learning_rate": 9.268365185400793e-05, "loss": 1.7949, "step": 835 }, { "epoch": 1.49, "learning_rate": 9.264924134356056e-05, "loss": 2.6953, "step": 836 }, { "epoch": 1.49, "learning_rate": 9.261475652143341e-05, "loss": 3.4531, "step": 837 }, { "epoch": 1.5, "learning_rate": 9.258019744771256e-05, "loss": 2.4199, "step": 838 }, { "epoch": 1.5, "learning_rate": 9.254556418261342e-05, "loss": 2.3926, "step": 839 }, { "epoch": 1.5, "learning_rate": 9.251085678648072e-05, "loss": 2.3066, "step": 840 }, { "epoch": 1.5, "learning_rate": 9.247607531978832e-05, "loss": 1.9395, "step": 841 }, { "epoch": 1.5, "learning_rate": 9.244121984313915e-05, "loss": 2.0078, "step": 842 }, { "epoch": 1.51, "learning_rate": 9.240629041726516e-05, "loss": 2.2656, "step": 843 }, { "epoch": 1.51, "learning_rate": 9.237128710302701e-05, "loss": 1.8613, "step": 844 }, { "epoch": 1.51, "learning_rate": 9.233620996141421e-05, "loss": 2.123, "step": 845 }, { "epoch": 1.51, "learning_rate": 9.230105905354489e-05, "loss": 2.543, "step": 846 }, { "epoch": 1.51, "learning_rate": 9.226583444066569e-05, "loss": 2.3516, "step": 847 }, { "epoch": 1.51, "learning_rate": 9.223053618415167e-05, "loss": 2.1797, "step": 848 }, { "epoch": 1.52, "learning_rate": 9.219516434550622e-05, "loss": 1.7725, "step": 849 }, { "epoch": 1.52, "learning_rate": 9.215971898636094e-05, "loss": 1.9541, "step": 850 }, { "epoch": 1.52, "learning_rate": 9.212420016847552e-05, "loss": 1.8438, "step": 851 }, { "epoch": 1.52, "learning_rate": 9.208860795373764e-05, "loss": 2.0801, "step": 852 }, { "epoch": 1.52, "learning_rate": 9.20529424041629e-05, "loss": 1.8564, "step": 853 }, { "epoch": 1.52, "learning_rate": 9.201720358189464e-05, "loss": 1.54, "step": 854 }, { "epoch": 1.53, "learning_rate": 9.198139154920389e-05, "loss": 2.1094, "step": 855 }, { "epoch": 1.53, "learning_rate": 9.194550636848921e-05, "loss": 2.0078, "step": 856 }, { "epoch": 1.53, "learning_rate": 9.190954810227668e-05, "loss": 1.7637, "step": 857 }, { "epoch": 1.53, "learning_rate": 9.187351681321965e-05, "loss": 1.7852, "step": 858 }, { "epoch": 1.53, "learning_rate": 9.183741256409875e-05, "loss": 2.0371, "step": 859 }, { "epoch": 1.54, "learning_rate": 9.180123541782171e-05, "loss": 1.9502, "step": 860 }, { "epoch": 1.54, "learning_rate": 9.176498543742328e-05, "loss": 1.2959, "step": 861 }, { "epoch": 1.54, "learning_rate": 9.172866268606513e-05, "loss": 1.9395, "step": 862 }, { "epoch": 1.54, "learning_rate": 9.169226722703574e-05, "loss": 1.9141, "step": 863 }, { "epoch": 1.54, "learning_rate": 9.165579912375021e-05, "loss": 1.7461, "step": 864 }, { "epoch": 1.54, "learning_rate": 9.161925843975028e-05, "loss": 2.4414, "step": 865 }, { "epoch": 1.55, "learning_rate": 9.158264523870412e-05, "loss": 1.6055, "step": 866 }, { "epoch": 1.55, "learning_rate": 9.154595958440628e-05, "loss": 1.9219, "step": 867 }, { "epoch": 1.55, "learning_rate": 9.150920154077754e-05, "loss": 2.0195, "step": 868 }, { "epoch": 1.55, "learning_rate": 9.147237117186478e-05, "loss": 1.8984, "step": 869 }, { "epoch": 1.55, "learning_rate": 9.143546854184094e-05, "loss": 1.8262, "step": 870 }, { "epoch": 1.56, "learning_rate": 9.139849371500485e-05, "loss": 2.5547, "step": 871 }, { "epoch": 1.56, "learning_rate": 9.136144675578113e-05, "loss": 2.3281, "step": 872 }, { "epoch": 1.56, "learning_rate": 9.132432772872012e-05, "loss": 1.8516, "step": 873 }, { "epoch": 1.56, "learning_rate": 9.128713669849766e-05, "loss": 2.6758, "step": 874 }, { "epoch": 1.56, "learning_rate": 9.124987372991511e-05, "loss": 1.334, "step": 875 }, { "epoch": 1.56, "learning_rate": 9.121253888789917e-05, "loss": 1.5469, "step": 876 }, { "epoch": 1.57, "learning_rate": 9.117513223750173e-05, "loss": 1.4629, "step": 877 }, { "epoch": 1.57, "learning_rate": 9.113765384389983e-05, "loss": 1.9824, "step": 878 }, { "epoch": 1.57, "learning_rate": 9.110010377239551e-05, "loss": 1.5352, "step": 879 }, { "epoch": 1.57, "learning_rate": 9.106248208841569e-05, "loss": 1.0273, "step": 880 }, { "epoch": 1.57, "learning_rate": 9.102478885751208e-05, "loss": 1.6221, "step": 881 }, { "epoch": 1.57, "learning_rate": 9.098702414536107e-05, "loss": 2.375, "step": 882 }, { "epoch": 1.58, "learning_rate": 9.094918801776353e-05, "loss": 2.293, "step": 883 }, { "epoch": 1.58, "learning_rate": 9.091128054064487e-05, "loss": 1.4902, "step": 884 }, { "epoch": 1.58, "learning_rate": 9.087330178005472e-05, "loss": 2.4863, "step": 885 }, { "epoch": 1.58, "learning_rate": 9.083525180216696e-05, "loss": 1.9473, "step": 886 }, { "epoch": 1.58, "learning_rate": 9.079713067327957e-05, "loss": 1.8398, "step": 887 }, { "epoch": 1.59, "learning_rate": 9.075893845981446e-05, "loss": 1.9453, "step": 888 }, { "epoch": 1.59, "learning_rate": 9.072067522831742e-05, "loss": 2.2578, "step": 889 }, { "epoch": 1.59, "learning_rate": 9.068234104545801e-05, "loss": 1.8086, "step": 890 }, { "epoch": 1.59, "learning_rate": 9.064393597802938e-05, "loss": 2.3086, "step": 891 }, { "epoch": 1.59, "learning_rate": 9.060546009294817e-05, "loss": 2.25, "step": 892 }, { "epoch": 1.59, "learning_rate": 9.056691345725447e-05, "loss": 1.9199, "step": 893 }, { "epoch": 1.6, "learning_rate": 9.05282961381116e-05, "loss": 1.8477, "step": 894 }, { "epoch": 1.6, "learning_rate": 9.048960820280605e-05, "loss": 1.9102, "step": 895 }, { "epoch": 1.6, "learning_rate": 9.045084971874738e-05, "loss": 1.877, "step": 896 }, { "epoch": 1.6, "learning_rate": 9.041202075346801e-05, "loss": 2.1914, "step": 897 }, { "epoch": 1.6, "learning_rate": 9.037312137462321e-05, "loss": 2.3828, "step": 898 }, { "epoch": 1.61, "learning_rate": 9.033415164999095e-05, "loss": 2.5781, "step": 899 }, { "epoch": 1.61, "learning_rate": 9.029511164747175e-05, "loss": 2.7656, "step": 900 }, { "epoch": 1.61, "learning_rate": 9.025600143508856e-05, "loss": 1.5137, "step": 901 }, { "epoch": 1.61, "learning_rate": 9.021682108098671e-05, "loss": 2.3145, "step": 902 }, { "epoch": 1.61, "learning_rate": 9.017757065343369e-05, "loss": 2.3711, "step": 903 }, { "epoch": 1.61, "learning_rate": 9.013825022081915e-05, "loss": 2.084, "step": 904 }, { "epoch": 1.62, "learning_rate": 9.009885985165464e-05, "loss": 1.6582, "step": 905 }, { "epoch": 1.62, "learning_rate": 9.005939961457366e-05, "loss": 2.1973, "step": 906 }, { "epoch": 1.62, "learning_rate": 9.001986957833135e-05, "loss": 2.1758, "step": 907 }, { "epoch": 1.62, "learning_rate": 8.998026981180455e-05, "loss": 2.3984, "step": 908 }, { "epoch": 1.62, "learning_rate": 8.994060038399153e-05, "loss": 2.1543, "step": 909 }, { "epoch": 1.62, "learning_rate": 8.9900861364012e-05, "loss": 1.625, "step": 910 }, { "epoch": 1.63, "learning_rate": 8.986105282110684e-05, "loss": 1.8887, "step": 911 }, { "epoch": 1.63, "learning_rate": 8.982117482463817e-05, "loss": 2.2773, "step": 912 }, { "epoch": 1.63, "learning_rate": 8.978122744408906e-05, "loss": 2.0625, "step": 913 }, { "epoch": 1.63, "learning_rate": 8.974121074906348e-05, "loss": 1.5645, "step": 914 }, { "epoch": 1.63, "learning_rate": 8.970112480928621e-05, "loss": 1.8662, "step": 915 }, { "epoch": 1.64, "learning_rate": 8.966096969460264e-05, "loss": 2.3711, "step": 916 }, { "epoch": 1.64, "learning_rate": 8.962074547497869e-05, "loss": 1.6602, "step": 917 }, { "epoch": 1.64, "learning_rate": 8.958045222050073e-05, "loss": 1.8945, "step": 918 }, { "epoch": 1.64, "learning_rate": 8.954009000137537e-05, "loss": 2.0078, "step": 919 }, { "epoch": 1.64, "learning_rate": 8.949965888792941e-05, "loss": 2.3867, "step": 920 }, { "epoch": 1.64, "learning_rate": 8.945915895060966e-05, "loss": 2.584, "step": 921 }, { "epoch": 1.65, "learning_rate": 8.941859025998289e-05, "loss": 1.4805, "step": 922 }, { "epoch": 1.65, "learning_rate": 8.937795288673566e-05, "loss": 1.7695, "step": 923 }, { "epoch": 1.65, "learning_rate": 8.933724690167417e-05, "loss": 2.6719, "step": 924 }, { "epoch": 1.65, "learning_rate": 8.929647237572418e-05, "loss": 1.5635, "step": 925 }, { "epoch": 1.65, "learning_rate": 8.925562937993089e-05, "loss": 2.4043, "step": 926 }, { "epoch": 1.66, "learning_rate": 8.92147179854588e-05, "loss": 2.0273, "step": 927 }, { "epoch": 1.66, "learning_rate": 8.917373826359157e-05, "loss": 1.9434, "step": 928 }, { "epoch": 1.66, "learning_rate": 8.91326902857319e-05, "loss": 1.5195, "step": 929 }, { "epoch": 1.66, "learning_rate": 8.90915741234015e-05, "loss": 1.9688, "step": 930 }, { "epoch": 1.66, "learning_rate": 8.905038984824078e-05, "loss": 1.9258, "step": 931 }, { "epoch": 1.66, "learning_rate": 8.900913753200888e-05, "loss": 1.6426, "step": 932 }, { "epoch": 1.67, "learning_rate": 8.896781724658351e-05, "loss": 1.833, "step": 933 }, { "epoch": 1.67, "learning_rate": 8.892642906396076e-05, "loss": 1.2812, "step": 934 }, { "epoch": 1.67, "learning_rate": 8.888497305625508e-05, "loss": 1.6465, "step": 935 }, { "epoch": 1.67, "learning_rate": 8.884344929569905e-05, "loss": 2.5586, "step": 936 }, { "epoch": 1.67, "learning_rate": 8.880185785464334e-05, "loss": 1.8643, "step": 937 }, { "epoch": 1.68, "learning_rate": 8.876019880555649e-05, "loss": 1.1807, "step": 938 }, { "epoch": 1.68, "learning_rate": 8.871847222102492e-05, "loss": 2.5098, "step": 939 }, { "epoch": 1.68, "learning_rate": 8.867667817375266e-05, "loss": 2.0781, "step": 940 }, { "epoch": 1.68, "learning_rate": 8.863481673656131e-05, "loss": 2.2383, "step": 941 }, { "epoch": 1.68, "learning_rate": 8.859288798238988e-05, "loss": 1.6094, "step": 942 }, { "epoch": 1.68, "learning_rate": 8.855089198429469e-05, "loss": 2.0664, "step": 943 }, { "epoch": 1.69, "learning_rate": 8.850882881544921e-05, "loss": 1.4951, "step": 944 }, { "epoch": 1.69, "learning_rate": 8.846669854914396e-05, "loss": 1.8125, "step": 945 }, { "epoch": 1.69, "learning_rate": 8.842450125878633e-05, "loss": 2.6211, "step": 946 }, { "epoch": 1.69, "learning_rate": 8.838223701790055e-05, "loss": 2.0879, "step": 947 }, { "epoch": 1.69, "learning_rate": 8.833990590012749e-05, "loss": 2.7246, "step": 948 }, { "epoch": 1.69, "learning_rate": 8.829750797922452e-05, "loss": 1.8848, "step": 949 }, { "epoch": 1.7, "learning_rate": 8.825504332906541e-05, "loss": 2.2891, "step": 950 }, { "epoch": 1.7, "learning_rate": 8.821251202364023e-05, "loss": 2.3574, "step": 951 }, { "epoch": 1.7, "learning_rate": 8.816991413705516e-05, "loss": 2.4863, "step": 952 }, { "epoch": 1.7, "learning_rate": 8.812724974353239e-05, "loss": 2.5703, "step": 953 }, { "epoch": 1.7, "learning_rate": 8.808451891741001e-05, "loss": 1.6836, "step": 954 }, { "epoch": 1.71, "learning_rate": 8.804172173314184e-05, "loss": 1.418, "step": 955 }, { "epoch": 1.71, "learning_rate": 8.799885826529737e-05, "loss": 1.7979, "step": 956 }, { "epoch": 1.71, "learning_rate": 8.795592858856152e-05, "loss": 2.0117, "step": 957 }, { "epoch": 1.71, "learning_rate": 8.79129327777346e-05, "loss": 1.543, "step": 958 }, { "epoch": 1.71, "learning_rate": 8.786987090773214e-05, "loss": 1.998, "step": 959 }, { "epoch": 1.71, "learning_rate": 8.78267430535848e-05, "loss": 2.1777, "step": 960 }, { "epoch": 1.72, "learning_rate": 8.778354929043818e-05, "loss": 2.3223, "step": 961 }, { "epoch": 1.72, "learning_rate": 8.774028969355273e-05, "loss": 2.1055, "step": 962 }, { "epoch": 1.72, "learning_rate": 8.769696433830361e-05, "loss": 1.7256, "step": 963 }, { "epoch": 1.72, "learning_rate": 8.765357330018056e-05, "loss": 2.2344, "step": 964 }, { "epoch": 1.72, "learning_rate": 8.761011665478774e-05, "loss": 1.7188, "step": 965 }, { "epoch": 1.73, "learning_rate": 8.756659447784368e-05, "loss": 1.4033, "step": 966 }, { "epoch": 1.73, "learning_rate": 8.752300684518102e-05, "loss": 2.0605, "step": 967 }, { "epoch": 1.73, "learning_rate": 8.74793538327465e-05, "loss": 1.9863, "step": 968 }, { "epoch": 1.73, "learning_rate": 8.743563551660076e-05, "loss": 2.9648, "step": 969 }, { "epoch": 1.73, "learning_rate": 8.739185197291823e-05, "loss": 1.4092, "step": 970 }, { "epoch": 1.73, "learning_rate": 8.734800327798698e-05, "loss": 1.3516, "step": 971 }, { "epoch": 1.74, "learning_rate": 8.730408950820863e-05, "loss": 1.8066, "step": 972 }, { "epoch": 1.74, "learning_rate": 8.726011074009814e-05, "loss": 1.8369, "step": 973 }, { "epoch": 1.74, "learning_rate": 8.721606705028375e-05, "loss": 1.9219, "step": 974 }, { "epoch": 1.74, "learning_rate": 8.717195851550682e-05, "loss": 2.3203, "step": 975 }, { "epoch": 1.74, "learning_rate": 8.71277852126217e-05, "loss": 1.5723, "step": 976 }, { "epoch": 1.74, "learning_rate": 8.708354721859556e-05, "loss": 2.4102, "step": 977 }, { "epoch": 1.75, "learning_rate": 8.703924461050831e-05, "loss": 1.6064, "step": 978 }, { "epoch": 1.75, "learning_rate": 8.699487746555248e-05, "loss": 2.0645, "step": 979 }, { "epoch": 1.75, "learning_rate": 8.695044586103296e-05, "loss": 2.5039, "step": 980 }, { "epoch": 1.75, "learning_rate": 8.690594987436704e-05, "loss": 1.5234, "step": 981 }, { "epoch": 1.75, "learning_rate": 8.686138958308414e-05, "loss": 1.9873, "step": 982 }, { "epoch": 1.76, "learning_rate": 8.681676506482572e-05, "loss": 1.8672, "step": 983 }, { "epoch": 1.76, "learning_rate": 8.67720763973452e-05, "loss": 1.5205, "step": 984 }, { "epoch": 1.76, "learning_rate": 8.672732365850774e-05, "loss": 2.4648, "step": 985 }, { "epoch": 1.76, "learning_rate": 8.668250692629007e-05, "loss": 2.1367, "step": 986 }, { "epoch": 1.76, "learning_rate": 8.663762627878058e-05, "loss": 1.3809, "step": 987 }, { "epoch": 1.76, "learning_rate": 8.659268179417887e-05, "loss": 1.7871, "step": 988 }, { "epoch": 1.77, "learning_rate": 8.654767355079585e-05, "loss": 1.7363, "step": 989 }, { "epoch": 1.77, "learning_rate": 8.650260162705351e-05, "loss": 2.3047, "step": 990 }, { "epoch": 1.77, "learning_rate": 8.645746610148479e-05, "loss": 1.8711, "step": 991 }, { "epoch": 1.77, "learning_rate": 8.641226705273344e-05, "loss": 1.9629, "step": 992 }, { "epoch": 1.77, "learning_rate": 8.636700455955393e-05, "loss": 1.8281, "step": 993 }, { "epoch": 1.77, "learning_rate": 8.632167870081121e-05, "loss": 2.0078, "step": 994 }, { "epoch": 1.78, "learning_rate": 8.62762895554807e-05, "loss": 1.5879, "step": 995 }, { "epoch": 1.78, "learning_rate": 8.623083720264806e-05, "loss": 1.877, "step": 996 }, { "epoch": 1.78, "learning_rate": 8.61853217215091e-05, "loss": 2.332, "step": 997 }, { "epoch": 1.78, "learning_rate": 8.613974319136958e-05, "loss": 2.4355, "step": 998 }, { "epoch": 1.78, "learning_rate": 8.609410169164518e-05, "loss": 1.9121, "step": 999 }, { "epoch": 1.79, "learning_rate": 8.604839730186125e-05, "loss": 1.7754, "step": 1000 }, { "epoch": 1.79, "learning_rate": 8.600263010165273e-05, "loss": 2.2344, "step": 1001 }, { "epoch": 1.79, "learning_rate": 8.595680017076401e-05, "loss": 1.9639, "step": 1002 }, { "epoch": 1.79, "learning_rate": 8.59109075890488e-05, "loss": 2.5117, "step": 1003 }, { "epoch": 1.79, "learning_rate": 8.586495243646992e-05, "loss": 1.7334, "step": 1004 }, { "epoch": 1.79, "learning_rate": 8.581893479309925e-05, "loss": 2.1016, "step": 1005 }, { "epoch": 1.8, "learning_rate": 8.577285473911754e-05, "loss": 2.3867, "step": 1006 }, { "epoch": 1.8, "learning_rate": 8.572671235481428e-05, "loss": 1.6953, "step": 1007 }, { "epoch": 1.8, "learning_rate": 8.568050772058762e-05, "loss": 1.7363, "step": 1008 }, { "epoch": 1.8, "learning_rate": 8.563424091694407e-05, "loss": 2.5156, "step": 1009 }, { "epoch": 1.8, "learning_rate": 8.558791202449856e-05, "loss": 2.3457, "step": 1010 }, { "epoch": 1.81, "learning_rate": 8.554152112397413e-05, "loss": 2.2461, "step": 1011 }, { "epoch": 1.81, "learning_rate": 8.549506829620191e-05, "loss": 2.2188, "step": 1012 }, { "epoch": 1.81, "learning_rate": 8.544855362212094e-05, "loss": 2.0039, "step": 1013 }, { "epoch": 1.81, "learning_rate": 8.540197718277796e-05, "loss": 1.5625, "step": 1014 }, { "epoch": 1.81, "learning_rate": 8.535533905932738e-05, "loss": 2.4707, "step": 1015 }, { "epoch": 1.81, "learning_rate": 8.530863933303109e-05, "loss": 2.2676, "step": 1016 }, { "epoch": 1.82, "learning_rate": 8.526187808525827e-05, "loss": 2.6543, "step": 1017 }, { "epoch": 1.82, "learning_rate": 8.521505539748535e-05, "loss": 2.2227, "step": 1018 }, { "epoch": 1.82, "learning_rate": 8.516817135129578e-05, "loss": 1.2197, "step": 1019 }, { "epoch": 1.82, "learning_rate": 8.512122602837993e-05, "loss": 2.7891, "step": 1020 }, { "epoch": 1.82, "learning_rate": 8.507421951053493e-05, "loss": 1.9141, "step": 1021 }, { "epoch": 1.82, "learning_rate": 8.502715187966455e-05, "loss": 1.1709, "step": 1022 }, { "epoch": 1.83, "learning_rate": 8.498002321777903e-05, "loss": 1.3906, "step": 1023 }, { "epoch": 1.83, "learning_rate": 8.493283360699495e-05, "loss": 1.5488, "step": 1024 }, { "epoch": 1.83, "learning_rate": 8.488558312953512e-05, "loss": 1.7695, "step": 1025 }, { "epoch": 1.83, "learning_rate": 8.483827186772832e-05, "loss": 1.167, "step": 1026 }, { "epoch": 1.83, "learning_rate": 8.479089990400935e-05, "loss": 1.5879, "step": 1027 }, { "epoch": 1.84, "learning_rate": 8.47434673209187e-05, "loss": 1.8711, "step": 1028 }, { "epoch": 1.84, "learning_rate": 8.46959742011025e-05, "loss": 2.082, "step": 1029 }, { "epoch": 1.84, "learning_rate": 8.464842062731235e-05, "loss": 2.0625, "step": 1030 }, { "epoch": 1.84, "learning_rate": 8.460080668240522e-05, "loss": 1.9141, "step": 1031 }, { "epoch": 1.84, "learning_rate": 8.455313244934324e-05, "loss": 1.8262, "step": 1032 }, { "epoch": 1.84, "learning_rate": 8.450539801119359e-05, "loss": 2.0352, "step": 1033 }, { "epoch": 1.85, "learning_rate": 8.445760345112836e-05, "loss": 2.1133, "step": 1034 }, { "epoch": 1.85, "learning_rate": 8.44097488524244e-05, "loss": 2.3125, "step": 1035 }, { "epoch": 1.85, "learning_rate": 8.436183429846313e-05, "loss": 1.9219, "step": 1036 }, { "epoch": 1.85, "learning_rate": 8.43138598727305e-05, "loss": 1.9668, "step": 1037 }, { "epoch": 1.85, "learning_rate": 8.426582565881673e-05, "loss": 2.2344, "step": 1038 }, { "epoch": 1.86, "learning_rate": 8.421773174041622e-05, "loss": 2.1211, "step": 1039 }, { "epoch": 1.86, "learning_rate": 8.416957820132742e-05, "loss": 2.2676, "step": 1040 }, { "epoch": 1.86, "learning_rate": 8.412136512545268e-05, "loss": 2.1641, "step": 1041 }, { "epoch": 1.86, "learning_rate": 8.4073092596798e-05, "loss": 1.3535, "step": 1042 }, { "epoch": 1.86, "learning_rate": 8.402476069947308e-05, "loss": 2.0312, "step": 1043 }, { "epoch": 1.86, "learning_rate": 8.397636951769099e-05, "loss": 1.8672, "step": 1044 }, { "epoch": 1.87, "learning_rate": 8.392791913576809e-05, "loss": 2.1465, "step": 1045 }, { "epoch": 1.87, "learning_rate": 8.387940963812399e-05, "loss": 1.7051, "step": 1046 }, { "epoch": 1.87, "learning_rate": 8.383084110928117e-05, "loss": 2.2637, "step": 1047 }, { "epoch": 1.87, "learning_rate": 8.378221363386505e-05, "loss": 2.084, "step": 1048 }, { "epoch": 1.87, "learning_rate": 8.373352729660373e-05, "loss": 1.915, "step": 1049 }, { "epoch": 1.88, "learning_rate": 8.368478218232787e-05, "loss": 1.7949, "step": 1050 }, { "epoch": 1.88, "learning_rate": 8.363597837597055e-05, "loss": 1.6328, "step": 1051 }, { "epoch": 1.88, "learning_rate": 8.358711596256712e-05, "loss": 2.0156, "step": 1052 }, { "epoch": 1.88, "learning_rate": 8.353819502725505e-05, "loss": 2.5527, "step": 1053 }, { "epoch": 1.88, "learning_rate": 8.348921565527372e-05, "loss": 1.9355, "step": 1054 }, { "epoch": 1.88, "learning_rate": 8.344017793196442e-05, "loss": 2.2148, "step": 1055 }, { "epoch": 1.89, "learning_rate": 8.339108194277006e-05, "loss": 1.9941, "step": 1056 }, { "epoch": 1.89, "learning_rate": 8.334192777323507e-05, "loss": 2.3379, "step": 1057 }, { "epoch": 1.89, "learning_rate": 8.329271550900528e-05, "loss": 2.5762, "step": 1058 }, { "epoch": 1.89, "learning_rate": 8.324344523582767e-05, "loss": 2.1895, "step": 1059 }, { "epoch": 1.89, "learning_rate": 8.319411703955042e-05, "loss": 1.834, "step": 1060 }, { "epoch": 1.89, "learning_rate": 8.314473100612254e-05, "loss": 2.6133, "step": 1061 }, { "epoch": 1.9, "learning_rate": 8.309528722159383e-05, "loss": 1.7812, "step": 1062 }, { "epoch": 1.9, "learning_rate": 8.304578577211472e-05, "loss": 2.0156, "step": 1063 }, { "epoch": 1.9, "learning_rate": 8.299622674393614e-05, "loss": 2.4512, "step": 1064 }, { "epoch": 1.9, "learning_rate": 8.294661022340931e-05, "loss": 1.71, "step": 1065 }, { "epoch": 1.9, "learning_rate": 8.289693629698564e-05, "loss": 1.6602, "step": 1066 }, { "epoch": 1.91, "learning_rate": 8.284720505121657e-05, "loss": 1.9102, "step": 1067 }, { "epoch": 1.91, "learning_rate": 8.27974165727534e-05, "loss": 1.9648, "step": 1068 }, { "epoch": 1.91, "learning_rate": 8.274757094834714e-05, "loss": 1.9648, "step": 1069 }, { "epoch": 1.91, "learning_rate": 8.269766826484842e-05, "loss": 2.0234, "step": 1070 }, { "epoch": 1.91, "learning_rate": 8.264770860920722e-05, "loss": 1.9561, "step": 1071 }, { "epoch": 1.91, "learning_rate": 8.259769206847285e-05, "loss": 2.0996, "step": 1072 }, { "epoch": 1.92, "learning_rate": 8.254761872979371e-05, "loss": 2.3555, "step": 1073 }, { "epoch": 1.92, "learning_rate": 8.249748868041717e-05, "loss": 1.623, "step": 1074 }, { "epoch": 1.92, "learning_rate": 8.244730200768937e-05, "loss": 2.0762, "step": 1075 }, { "epoch": 1.92, "learning_rate": 8.239705879905519e-05, "loss": 1.5088, "step": 1076 }, { "epoch": 1.92, "learning_rate": 8.234675914205796e-05, "loss": 1.4355, "step": 1077 }, { "epoch": 1.93, "learning_rate": 8.229640312433937e-05, "loss": 1.7754, "step": 1078 }, { "epoch": 1.93, "learning_rate": 8.224599083363933e-05, "loss": 3.2695, "step": 1079 }, { "epoch": 1.93, "learning_rate": 8.219552235779578e-05, "loss": 2.0371, "step": 1080 }, { "epoch": 1.93, "learning_rate": 8.214499778474459e-05, "loss": 2.7363, "step": 1081 }, { "epoch": 1.93, "learning_rate": 8.209441720251933e-05, "loss": 1.6123, "step": 1082 }, { "epoch": 1.93, "learning_rate": 8.20437806992512e-05, "loss": 1.7754, "step": 1083 }, { "epoch": 1.94, "learning_rate": 8.199308836316882e-05, "loss": 2.5977, "step": 1084 }, { "epoch": 1.94, "learning_rate": 8.194234028259806e-05, "loss": 2.293, "step": 1085 }, { "epoch": 1.94, "learning_rate": 8.189153654596199e-05, "loss": 2.4727, "step": 1086 }, { "epoch": 1.94, "learning_rate": 8.184067724178058e-05, "loss": 2.3379, "step": 1087 }, { "epoch": 1.94, "learning_rate": 8.178976245867068e-05, "loss": 2.0938, "step": 1088 }, { "epoch": 1.94, "learning_rate": 8.173879228534575e-05, "loss": 2.1328, "step": 1089 }, { "epoch": 1.95, "learning_rate": 8.168776681061582e-05, "loss": 2.5547, "step": 1090 }, { "epoch": 1.95, "learning_rate": 8.163668612338723e-05, "loss": 2.4219, "step": 1091 }, { "epoch": 1.95, "learning_rate": 8.158555031266254e-05, "loss": 2.2695, "step": 1092 }, { "epoch": 1.95, "learning_rate": 8.153435946754039e-05, "loss": 2.5449, "step": 1093 }, { "epoch": 1.95, "learning_rate": 8.148311367721524e-05, "loss": 2.6328, "step": 1094 }, { "epoch": 1.96, "learning_rate": 8.143181303097734e-05, "loss": 2.2539, "step": 1095 }, { "epoch": 1.96, "learning_rate": 8.138045761821251e-05, "loss": 1.8301, "step": 1096 }, { "epoch": 1.96, "learning_rate": 8.132904752840198e-05, "loss": 2.1172, "step": 1097 }, { "epoch": 1.96, "learning_rate": 8.127758285112225e-05, "loss": 1.9668, "step": 1098 }, { "epoch": 1.96, "learning_rate": 8.122606367604497e-05, "loss": 2.2539, "step": 1099 }, { "epoch": 1.96, "learning_rate": 8.117449009293668e-05, "loss": 1.8848, "step": 1100 }, { "epoch": 1.97, "learning_rate": 8.11228621916588e-05, "loss": 1.7949, "step": 1101 }, { "epoch": 1.97, "learning_rate": 8.107118006216732e-05, "loss": 2.043, "step": 1102 }, { "epoch": 1.97, "learning_rate": 8.101944379451275e-05, "loss": 1.7793, "step": 1103 }, { "epoch": 1.97, "learning_rate": 8.096765347883994e-05, "loss": 2.0625, "step": 1104 }, { "epoch": 1.97, "learning_rate": 8.09158092053879e-05, "loss": 2.3945, "step": 1105 }, { "epoch": 1.98, "learning_rate": 8.086391106448965e-05, "loss": 2.0234, "step": 1106 }, { "epoch": 1.98, "learning_rate": 8.081195914657208e-05, "loss": 2.3789, "step": 1107 }, { "epoch": 1.98, "learning_rate": 8.075995354215577e-05, "loss": 2.0039, "step": 1108 }, { "epoch": 1.98, "learning_rate": 8.070789434185486e-05, "loss": 1.6719, "step": 1109 }, { "epoch": 1.98, "learning_rate": 8.065578163637686e-05, "loss": 2.1113, "step": 1110 }, { "epoch": 1.98, "learning_rate": 8.060361551652249e-05, "loss": 2.125, "step": 1111 }, { "epoch": 1.99, "learning_rate": 8.055139607318558e-05, "loss": 2.4336, "step": 1112 }, { "epoch": 1.99, "learning_rate": 8.049912339735283e-05, "loss": 1.8525, "step": 1113 }, { "epoch": 1.99, "learning_rate": 8.044679758010375e-05, "loss": 2.332, "step": 1114 }, { "epoch": 1.99, "learning_rate": 8.039441871261038e-05, "loss": 2.8555, "step": 1115 }, { "epoch": 1.99, "learning_rate": 8.034198688613719e-05, "loss": 2.0039, "step": 1116 }, { "epoch": 1.99, "learning_rate": 8.0289502192041e-05, "loss": 1.709, "step": 1117 }, { "epoch": 2.0, "learning_rate": 8.023696472177067e-05, "loss": 2.3047, "step": 1118 }, { "epoch": 2.0, "learning_rate": 8.018437456686707e-05, "loss": 1.6094, "step": 1119 }, { "epoch": 2.0, "learning_rate": 8.013173181896283e-05, "loss": 2.2383, "step": 1120 }, { "epoch": 2.0, "learning_rate": 8.007903656978221e-05, "loss": 1.6816, "step": 1121 }, { "epoch": 2.0, "learning_rate": 8.002628891114103e-05, "loss": 1.7871, "step": 1122 }, { "epoch": 2.01, "learning_rate": 7.997348893494631e-05, "loss": 2.1113, "step": 1123 }, { "epoch": 2.01, "learning_rate": 7.992063673319631e-05, "loss": 1.8359, "step": 1124 }, { "epoch": 2.01, "learning_rate": 7.986773239798026e-05, "loss": 1.8105, "step": 1125 }, { "epoch": 2.01, "learning_rate": 7.981477602147822e-05, "loss": 2.3164, "step": 1126 }, { "epoch": 2.01, "learning_rate": 7.976176769596094e-05, "loss": 1.9131, "step": 1127 }, { "epoch": 2.01, "learning_rate": 7.970870751378969e-05, "loss": 1.6318, "step": 1128 }, { "epoch": 2.02, "learning_rate": 7.965559556741609e-05, "loss": 2.1426, "step": 1129 }, { "epoch": 2.02, "learning_rate": 7.960243194938192e-05, "loss": 1.7246, "step": 1130 }, { "epoch": 2.02, "learning_rate": 7.954921675231905e-05, "loss": 1.873, "step": 1131 }, { "epoch": 2.02, "learning_rate": 7.949595006894917e-05, "loss": 1.7852, "step": 1132 }, { "epoch": 2.02, "learning_rate": 7.944263199208372e-05, "loss": 2.127, "step": 1133 }, { "epoch": 2.02, "learning_rate": 7.938926261462366e-05, "loss": 1.4512, "step": 1134 }, { "epoch": 2.03, "learning_rate": 7.933584202955934e-05, "loss": 1.75, "step": 1135 }, { "epoch": 2.03, "learning_rate": 7.928237032997036e-05, "loss": 2.3984, "step": 1136 }, { "epoch": 2.03, "learning_rate": 7.922884760902535e-05, "loss": 2.1055, "step": 1137 }, { "epoch": 2.03, "learning_rate": 7.917527395998183e-05, "loss": 1.707, "step": 1138 }, { "epoch": 2.03, "learning_rate": 7.91216494761861e-05, "loss": 2.2969, "step": 1139 }, { "epoch": 2.04, "learning_rate": 7.9067974251073e-05, "loss": 1.4111, "step": 1140 }, { "epoch": 2.04, "learning_rate": 7.901424837816579e-05, "loss": 1.373, "step": 1141 }, { "epoch": 2.04, "learning_rate": 7.896047195107599e-05, "loss": 1.6182, "step": 1142 }, { "epoch": 2.04, "learning_rate": 7.890664506350318e-05, "loss": 2.0352, "step": 1143 }, { "epoch": 2.04, "learning_rate": 7.885276780923488e-05, "loss": 2.1406, "step": 1144 }, { "epoch": 2.04, "learning_rate": 7.879884028214637e-05, "loss": 2.1602, "step": 1145 }, { "epoch": 2.05, "learning_rate": 7.87448625762005e-05, "loss": 1.8379, "step": 1146 }, { "epoch": 2.05, "learning_rate": 7.869083478544763e-05, "loss": 1.75, "step": 1147 }, { "epoch": 2.05, "learning_rate": 7.863675700402526e-05, "loss": 2.2461, "step": 1148 }, { "epoch": 2.05, "learning_rate": 7.85826293261581e-05, "loss": 2.2773, "step": 1149 }, { "epoch": 2.05, "learning_rate": 7.852845184615776e-05, "loss": 1.7979, "step": 1150 }, { "epoch": 2.06, "learning_rate": 7.84742246584226e-05, "loss": 2.3203, "step": 1151 }, { "epoch": 2.06, "learning_rate": 7.841994785743765e-05, "loss": 1.7725, "step": 1152 }, { "epoch": 2.06, "learning_rate": 7.836562153777433e-05, "loss": 3.3398, "step": 1153 }, { "epoch": 2.06, "learning_rate": 7.831124579409037e-05, "loss": 1.7178, "step": 1154 }, { "epoch": 2.06, "learning_rate": 7.82568207211296e-05, "loss": 1.9297, "step": 1155 }, { "epoch": 2.06, "learning_rate": 7.820234641372181e-05, "loss": 1.7217, "step": 1156 }, { "epoch": 2.07, "learning_rate": 7.814782296678261e-05, "loss": 2.1914, "step": 1157 }, { "epoch": 2.07, "learning_rate": 7.809325047531315e-05, "loss": 2.2812, "step": 1158 }, { "epoch": 2.07, "learning_rate": 7.803862903440012e-05, "loss": 2.4395, "step": 1159 }, { "epoch": 2.07, "learning_rate": 7.79839587392154e-05, "loss": 1.9805, "step": 1160 }, { "epoch": 2.07, "learning_rate": 7.792923968501613e-05, "loss": 1.6582, "step": 1161 }, { "epoch": 2.08, "learning_rate": 7.787447196714427e-05, "loss": 2.123, "step": 1162 }, { "epoch": 2.08, "learning_rate": 7.781965568102667e-05, "loss": 1.0859, "step": 1163 }, { "epoch": 2.08, "learning_rate": 7.776479092217474e-05, "loss": 1.8652, "step": 1164 }, { "epoch": 2.08, "learning_rate": 7.770987778618441e-05, "loss": 1.6152, "step": 1165 }, { "epoch": 2.08, "learning_rate": 7.765491636873582e-05, "loss": 2.4453, "step": 1166 }, { "epoch": 2.08, "learning_rate": 7.75999067655933e-05, "loss": 1.6523, "step": 1167 }, { "epoch": 2.09, "learning_rate": 7.754484907260513e-05, "loss": 2.8125, "step": 1168 }, { "epoch": 2.09, "learning_rate": 7.748974338570337e-05, "loss": 2.1523, "step": 1169 }, { "epoch": 2.09, "learning_rate": 7.74345898009037e-05, "loss": 1.8574, "step": 1170 }, { "epoch": 2.09, "learning_rate": 7.737938841430527e-05, "loss": 1.9414, "step": 1171 }, { "epoch": 2.09, "learning_rate": 7.732413932209049e-05, "loss": 2.0, "step": 1172 }, { "epoch": 2.09, "learning_rate": 7.726884262052496e-05, "loss": 1.3691, "step": 1173 }, { "epoch": 2.1, "learning_rate": 7.721349840595712e-05, "loss": 1.5195, "step": 1174 }, { "epoch": 2.1, "learning_rate": 7.715810677481834e-05, "loss": 1.7041, "step": 1175 }, { "epoch": 2.1, "learning_rate": 7.710266782362247e-05, "loss": 3.0039, "step": 1176 }, { "epoch": 2.1, "learning_rate": 7.704718164896591e-05, "loss": 2.3594, "step": 1177 }, { "epoch": 2.1, "learning_rate": 7.69916483475273e-05, "loss": 2.5, "step": 1178 }, { "epoch": 2.11, "learning_rate": 7.693606801606738e-05, "loss": 1.7285, "step": 1179 }, { "epoch": 2.11, "learning_rate": 7.688044075142887e-05, "loss": 2.2031, "step": 1180 }, { "epoch": 2.11, "learning_rate": 7.682476665053625e-05, "loss": 2.1816, "step": 1181 }, { "epoch": 2.11, "learning_rate": 7.67690458103956e-05, "loss": 1.8633, "step": 1182 }, { "epoch": 2.11, "learning_rate": 7.671327832809441e-05, "loss": 2.9707, "step": 1183 }, { "epoch": 2.11, "learning_rate": 7.665746430080154e-05, "loss": 2.7734, "step": 1184 }, { "epoch": 2.12, "learning_rate": 7.660160382576683e-05, "loss": 2.1445, "step": 1185 }, { "epoch": 2.12, "learning_rate": 7.654569700032113e-05, "loss": 2.0938, "step": 1186 }, { "epoch": 2.12, "learning_rate": 7.6489743921876e-05, "loss": 1.2236, "step": 1187 }, { "epoch": 2.12, "learning_rate": 7.643374468792364e-05, "loss": 1.5918, "step": 1188 }, { "epoch": 2.12, "learning_rate": 7.63776993960366e-05, "loss": 1.2998, "step": 1189 }, { "epoch": 2.12, "learning_rate": 7.63216081438678e-05, "loss": 2.0078, "step": 1190 }, { "epoch": 2.13, "learning_rate": 7.62654710291501e-05, "loss": 1.6055, "step": 1191 }, { "epoch": 2.13, "learning_rate": 7.620928814969635e-05, "loss": 2.0059, "step": 1192 }, { "epoch": 2.13, "learning_rate": 7.615305960339916e-05, "loss": 1.4287, "step": 1193 }, { "epoch": 2.13, "learning_rate": 7.609678548823065e-05, "loss": 1.6143, "step": 1194 }, { "epoch": 2.13, "learning_rate": 7.604046590224235e-05, "loss": 1.8252, "step": 1195 }, { "epoch": 2.14, "learning_rate": 7.598410094356505e-05, "loss": 2.4805, "step": 1196 }, { "epoch": 2.14, "learning_rate": 7.592769071040861e-05, "loss": 2.0938, "step": 1197 }, { "epoch": 2.14, "learning_rate": 7.587123530106171e-05, "loss": 1.1855, "step": 1198 }, { "epoch": 2.14, "learning_rate": 7.581473481389178e-05, "loss": 1.5781, "step": 1199 }, { "epoch": 2.14, "learning_rate": 7.57581893473448e-05, "loss": 1.7188, "step": 1200 }, { "epoch": 2.14, "learning_rate": 7.570159899994513e-05, "loss": 2.5605, "step": 1201 }, { "epoch": 2.15, "learning_rate": 7.564496387029532e-05, "loss": 2.2832, "step": 1202 }, { "epoch": 2.15, "learning_rate": 7.55882840570759e-05, "loss": 2.1328, "step": 1203 }, { "epoch": 2.15, "learning_rate": 7.553155965904535e-05, "loss": 2.0117, "step": 1204 }, { "epoch": 2.15, "learning_rate": 7.547479077503976e-05, "loss": 1.9062, "step": 1205 }, { "epoch": 2.15, "learning_rate": 7.541797750397277e-05, "loss": 1.0674, "step": 1206 }, { "epoch": 2.16, "learning_rate": 7.536111994483533e-05, "loss": 2.5625, "step": 1207 }, { "epoch": 2.16, "learning_rate": 7.530421819669557e-05, "loss": 2.1621, "step": 1208 }, { "epoch": 2.16, "learning_rate": 7.524727235869864e-05, "loss": 2.2754, "step": 1209 }, { "epoch": 2.16, "learning_rate": 7.519028253006648e-05, "loss": 2.2148, "step": 1210 }, { "epoch": 2.16, "learning_rate": 7.513324881009769e-05, "loss": 1.8359, "step": 1211 }, { "epoch": 2.16, "learning_rate": 7.507617129816733e-05, "loss": 1.7461, "step": 1212 }, { "epoch": 2.17, "learning_rate": 7.501905009372681e-05, "loss": 1.7305, "step": 1213 }, { "epoch": 2.17, "learning_rate": 7.49618852963036e-05, "loss": 1.2275, "step": 1214 }, { "epoch": 2.17, "learning_rate": 7.490467700550117e-05, "loss": 2.4297, "step": 1215 }, { "epoch": 2.17, "learning_rate": 7.484742532099878e-05, "loss": 2.3926, "step": 1216 }, { "epoch": 2.17, "learning_rate": 7.479013034255128e-05, "loss": 1.7734, "step": 1217 }, { "epoch": 2.17, "learning_rate": 7.473279216998895e-05, "loss": 2.5977, "step": 1218 }, { "epoch": 2.18, "learning_rate": 7.467541090321735e-05, "loss": 2.252, "step": 1219 }, { "epoch": 2.18, "learning_rate": 7.461798664221711e-05, "loss": 1.9688, "step": 1220 }, { "epoch": 2.18, "learning_rate": 7.456051948704379e-05, "loss": 1.6934, "step": 1221 }, { "epoch": 2.18, "learning_rate": 7.450300953782767e-05, "loss": 1.7148, "step": 1222 }, { "epoch": 2.18, "learning_rate": 7.44454568947736e-05, "loss": 2.4844, "step": 1223 }, { "epoch": 2.19, "learning_rate": 7.438786165816084e-05, "loss": 2.0957, "step": 1224 }, { "epoch": 2.19, "learning_rate": 7.433022392834282e-05, "loss": 1.8496, "step": 1225 }, { "epoch": 2.19, "learning_rate": 7.427254380574705e-05, "loss": 2.1426, "step": 1226 }, { "epoch": 2.19, "learning_rate": 7.421482139087489e-05, "loss": 1.5137, "step": 1227 }, { "epoch": 2.19, "learning_rate": 7.415705678430138e-05, "loss": 1.7783, "step": 1228 }, { "epoch": 2.19, "learning_rate": 7.409925008667512e-05, "loss": 2.3379, "step": 1229 }, { "epoch": 2.2, "learning_rate": 7.404140139871797e-05, "loss": 1.9727, "step": 1230 }, { "epoch": 2.2, "learning_rate": 7.398351082122503e-05, "loss": 2.2832, "step": 1231 }, { "epoch": 2.2, "learning_rate": 7.392557845506432e-05, "loss": 1.4062, "step": 1232 }, { "epoch": 2.2, "learning_rate": 7.386760440117676e-05, "loss": 1.627, "step": 1233 }, { "epoch": 2.2, "learning_rate": 7.380958876057581e-05, "loss": 2.0078, "step": 1234 }, { "epoch": 2.21, "learning_rate": 7.375153163434746e-05, "loss": 2.5977, "step": 1235 }, { "epoch": 2.21, "learning_rate": 7.369343312364993e-05, "loss": 2.0078, "step": 1236 }, { "epoch": 2.21, "learning_rate": 7.363529332971362e-05, "loss": 2.2305, "step": 1237 }, { "epoch": 2.21, "learning_rate": 7.357711235384078e-05, "loss": 1.5996, "step": 1238 }, { "epoch": 2.21, "learning_rate": 7.351889029740547e-05, "loss": 2.5391, "step": 1239 }, { "epoch": 2.21, "learning_rate": 7.346062726185332e-05, "loss": 1.8066, "step": 1240 }, { "epoch": 2.22, "learning_rate": 7.340232334870133e-05, "loss": 1.9043, "step": 1241 }, { "epoch": 2.22, "learning_rate": 7.334397865953777e-05, "loss": 1.6328, "step": 1242 }, { "epoch": 2.22, "learning_rate": 7.328559329602193e-05, "loss": 3.1523, "step": 1243 }, { "epoch": 2.22, "learning_rate": 7.322716735988396e-05, "loss": 2.0723, "step": 1244 }, { "epoch": 2.22, "learning_rate": 7.316870095292476e-05, "loss": 2.0039, "step": 1245 }, { "epoch": 2.23, "learning_rate": 7.311019417701566e-05, "loss": 1.791, "step": 1246 }, { "epoch": 2.23, "learning_rate": 7.305164713409841e-05, "loss": 1.5596, "step": 1247 }, { "epoch": 2.23, "learning_rate": 7.299305992618487e-05, "loss": 2.3457, "step": 1248 }, { "epoch": 2.23, "learning_rate": 7.29344326553569e-05, "loss": 1.623, "step": 1249 }, { "epoch": 2.23, "learning_rate": 7.287576542376617e-05, "loss": 2.1738, "step": 1250 }, { "epoch": 2.23, "learning_rate": 7.281705833363395e-05, "loss": 2.4473, "step": 1251 }, { "epoch": 2.24, "learning_rate": 7.2758311487251e-05, "loss": 1.8408, "step": 1252 }, { "epoch": 2.24, "learning_rate": 7.269952498697734e-05, "loss": 2.1133, "step": 1253 }, { "epoch": 2.24, "learning_rate": 7.264069893524206e-05, "loss": 2.1621, "step": 1254 }, { "epoch": 2.24, "learning_rate": 7.258183343454318e-05, "loss": 1.6875, "step": 1255 }, { "epoch": 2.24, "learning_rate": 7.252292858744748e-05, "loss": 1.6523, "step": 1256 }, { "epoch": 2.24, "learning_rate": 7.24639844965902e-05, "loss": 1.6777, "step": 1257 }, { "epoch": 2.25, "learning_rate": 7.24050012646751e-05, "loss": 1.6416, "step": 1258 }, { "epoch": 2.25, "learning_rate": 7.234597899447403e-05, "loss": 2.0117, "step": 1259 }, { "epoch": 2.25, "learning_rate": 7.228691778882693e-05, "loss": 1.8203, "step": 1260 }, { "epoch": 2.25, "learning_rate": 7.22278177506415e-05, "loss": 2.0332, "step": 1261 }, { "epoch": 2.25, "learning_rate": 7.216867898289319e-05, "loss": 2.1445, "step": 1262 }, { "epoch": 2.26, "learning_rate": 7.210950158862488e-05, "loss": 1.3203, "step": 1263 }, { "epoch": 2.26, "learning_rate": 7.205028567094677e-05, "loss": 2.5273, "step": 1264 }, { "epoch": 2.26, "learning_rate": 7.199103133303618e-05, "loss": 1.9238, "step": 1265 }, { "epoch": 2.26, "learning_rate": 7.193173867813735e-05, "loss": 1.582, "step": 1266 }, { "epoch": 2.26, "learning_rate": 7.187240780956133e-05, "loss": 1.918, "step": 1267 }, { "epoch": 2.26, "learning_rate": 7.181303883068568e-05, "loss": 1.9092, "step": 1268 }, { "epoch": 2.27, "learning_rate": 7.175363184495447e-05, "loss": 1.8828, "step": 1269 }, { "epoch": 2.27, "learning_rate": 7.169418695587791e-05, "loss": 1.915, "step": 1270 }, { "epoch": 2.27, "learning_rate": 7.163470426703226e-05, "loss": 1.7744, "step": 1271 }, { "epoch": 2.27, "learning_rate": 7.157518388205968e-05, "loss": 1.8184, "step": 1272 }, { "epoch": 2.27, "learning_rate": 7.151562590466796e-05, "loss": 2.1875, "step": 1273 }, { "epoch": 2.27, "learning_rate": 7.145603043863045e-05, "loss": 2.1641, "step": 1274 }, { "epoch": 2.28, "learning_rate": 7.139639758778578e-05, "loss": 2.3828, "step": 1275 }, { "epoch": 2.28, "learning_rate": 7.13367274560377e-05, "loss": 2.5, "step": 1276 }, { "epoch": 2.28, "learning_rate": 7.1277020147355e-05, "loss": 1.9863, "step": 1277 }, { "epoch": 2.28, "learning_rate": 7.121727576577115e-05, "loss": 1.9434, "step": 1278 }, { "epoch": 2.28, "learning_rate": 7.115749441538428e-05, "loss": 2.3906, "step": 1279 }, { "epoch": 2.29, "learning_rate": 7.109767620035689e-05, "loss": 2.2852, "step": 1280 }, { "epoch": 2.29, "learning_rate": 7.103782122491577e-05, "loss": 1.6855, "step": 1281 }, { "epoch": 2.29, "learning_rate": 7.097792959335168e-05, "loss": 2.5, "step": 1282 }, { "epoch": 2.29, "learning_rate": 7.091800141001935e-05, "loss": 1.9932, "step": 1283 }, { "epoch": 2.29, "learning_rate": 7.08580367793371e-05, "loss": 1.752, "step": 1284 }, { "epoch": 2.29, "learning_rate": 7.079803580578681e-05, "loss": 1.875, "step": 1285 }, { "epoch": 2.3, "learning_rate": 7.073799859391368e-05, "loss": 1.6152, "step": 1286 }, { "epoch": 2.3, "learning_rate": 7.067792524832604e-05, "loss": 1.5762, "step": 1287 }, { "epoch": 2.3, "learning_rate": 7.061781587369519e-05, "loss": 2.1621, "step": 1288 }, { "epoch": 2.3, "learning_rate": 7.055767057475517e-05, "loss": 1.8809, "step": 1289 }, { "epoch": 2.3, "learning_rate": 7.049748945630269e-05, "loss": 2.1211, "step": 1290 }, { "epoch": 2.31, "learning_rate": 7.043727262319678e-05, "loss": 1.8008, "step": 1291 }, { "epoch": 2.31, "learning_rate": 7.037702018035877e-05, "loss": 2.4922, "step": 1292 }, { "epoch": 2.31, "learning_rate": 7.031673223277202e-05, "loss": 2.0137, "step": 1293 }, { "epoch": 2.31, "learning_rate": 7.025640888548169e-05, "loss": 1.6621, "step": 1294 }, { "epoch": 2.31, "learning_rate": 7.019605024359474e-05, "loss": 1.6758, "step": 1295 }, { "epoch": 2.31, "learning_rate": 7.013565641227954e-05, "loss": 2.5, "step": 1296 }, { "epoch": 2.32, "learning_rate": 7.007522749676577e-05, "loss": 1.5664, "step": 1297 }, { "epoch": 2.32, "learning_rate": 7.001476360234428e-05, "loss": 1.7588, "step": 1298 }, { "epoch": 2.32, "learning_rate": 6.995426483436687e-05, "loss": 1.6738, "step": 1299 }, { "epoch": 2.32, "learning_rate": 6.989373129824604e-05, "loss": 1.502, "step": 1300 }, { "epoch": 2.32, "learning_rate": 6.983316309945497e-05, "loss": 1.625, "step": 1301 }, { "epoch": 2.33, "learning_rate": 6.977256034352712e-05, "loss": 1.7402, "step": 1302 }, { "epoch": 2.33, "learning_rate": 6.971192313605628e-05, "loss": 1.6133, "step": 1303 }, { "epoch": 2.33, "learning_rate": 6.965125158269619e-05, "loss": 1.7939, "step": 1304 }, { "epoch": 2.33, "learning_rate": 6.959054578916044e-05, "loss": 1.7383, "step": 1305 }, { "epoch": 2.33, "learning_rate": 6.952980586122231e-05, "loss": 2.1445, "step": 1306 }, { "epoch": 2.33, "learning_rate": 6.946903190471452e-05, "loss": 2.0547, "step": 1307 }, { "epoch": 2.34, "learning_rate": 6.940822402552914e-05, "loss": 1.3906, "step": 1308 }, { "epoch": 2.34, "learning_rate": 6.934738232961728e-05, "loss": 1.3594, "step": 1309 }, { "epoch": 2.34, "learning_rate": 6.928650692298898e-05, "loss": 1.6055, "step": 1310 }, { "epoch": 2.34, "learning_rate": 6.922559791171309e-05, "loss": 2.2383, "step": 1311 }, { "epoch": 2.34, "learning_rate": 6.91646554019169e-05, "loss": 1.6406, "step": 1312 }, { "epoch": 2.34, "learning_rate": 6.91036794997862e-05, "loss": 2.0996, "step": 1313 }, { "epoch": 2.35, "learning_rate": 6.904267031156482e-05, "loss": 1.9072, "step": 1314 }, { "epoch": 2.35, "learning_rate": 6.89816279435547e-05, "loss": 2.0879, "step": 1315 }, { "epoch": 2.35, "learning_rate": 6.892055250211552e-05, "loss": 1.7734, "step": 1316 }, { "epoch": 2.35, "learning_rate": 6.885944409366463e-05, "loss": 2.6172, "step": 1317 }, { "epoch": 2.35, "learning_rate": 6.879830282467682e-05, "loss": 1.7246, "step": 1318 }, { "epoch": 2.36, "learning_rate": 6.87371288016841e-05, "loss": 1.8193, "step": 1319 }, { "epoch": 2.36, "learning_rate": 6.867592213127558e-05, "loss": 1.9277, "step": 1320 }, { "epoch": 2.36, "learning_rate": 6.861468292009727e-05, "loss": 1.9863, "step": 1321 }, { "epoch": 2.36, "learning_rate": 6.855341127485183e-05, "loss": 2.3633, "step": 1322 }, { "epoch": 2.36, "learning_rate": 6.849210730229846e-05, "loss": 1.7324, "step": 1323 }, { "epoch": 2.36, "learning_rate": 6.84307711092527e-05, "loss": 1.4473, "step": 1324 }, { "epoch": 2.37, "learning_rate": 6.836940280258623e-05, "loss": 2.4531, "step": 1325 }, { "epoch": 2.37, "learning_rate": 6.830800248922664e-05, "loss": 1.9141, "step": 1326 }, { "epoch": 2.37, "learning_rate": 6.824657027615733e-05, "loss": 2.7969, "step": 1327 }, { "epoch": 2.37, "learning_rate": 6.81851062704173e-05, "loss": 1.8574, "step": 1328 }, { "epoch": 2.37, "learning_rate": 6.812361057910088e-05, "loss": 1.9883, "step": 1329 }, { "epoch": 2.38, "learning_rate": 6.806208330935766e-05, "loss": 1.9873, "step": 1330 }, { "epoch": 2.38, "learning_rate": 6.800052456839225e-05, "loss": 1.3809, "step": 1331 }, { "epoch": 2.38, "learning_rate": 6.793893446346405e-05, "loss": 2.1309, "step": 1332 }, { "epoch": 2.38, "learning_rate": 6.787731310188716e-05, "loss": 2.1406, "step": 1333 }, { "epoch": 2.38, "learning_rate": 6.781566059103015e-05, "loss": 1.8447, "step": 1334 }, { "epoch": 2.38, "learning_rate": 6.775397703831579e-05, "loss": 1.8281, "step": 1335 }, { "epoch": 2.39, "learning_rate": 6.769226255122103e-05, "loss": 2.6133, "step": 1336 }, { "epoch": 2.39, "learning_rate": 6.763051723727662e-05, "loss": 2.0137, "step": 1337 }, { "epoch": 2.39, "learning_rate": 6.756874120406714e-05, "loss": 1.5469, "step": 1338 }, { "epoch": 2.39, "learning_rate": 6.750693455923058e-05, "loss": 2.1758, "step": 1339 }, { "epoch": 2.39, "learning_rate": 6.744509741045835e-05, "loss": 2.1738, "step": 1340 }, { "epoch": 2.39, "learning_rate": 6.738322986549494e-05, "loss": 1.7559, "step": 1341 }, { "epoch": 2.4, "learning_rate": 6.732133203213787e-05, "loss": 2.4414, "step": 1342 }, { "epoch": 2.4, "learning_rate": 6.725940401823743e-05, "loss": 1.8398, "step": 1343 }, { "epoch": 2.4, "learning_rate": 6.719744593169641e-05, "loss": 1.8838, "step": 1344 }, { "epoch": 2.4, "learning_rate": 6.713545788047011e-05, "loss": 1.7656, "step": 1345 }, { "epoch": 2.4, "learning_rate": 6.707343997256595e-05, "loss": 2.1211, "step": 1346 }, { "epoch": 2.41, "learning_rate": 6.701139231604345e-05, "loss": 2.2441, "step": 1347 }, { "epoch": 2.41, "learning_rate": 6.69493150190139e-05, "loss": 1.8418, "step": 1348 }, { "epoch": 2.41, "learning_rate": 6.688720818964028e-05, "loss": 1.8418, "step": 1349 }, { "epoch": 2.41, "learning_rate": 6.682507193613696e-05, "loss": 1.7676, "step": 1350 }, { "epoch": 2.41, "learning_rate": 6.676290636676969e-05, "loss": 2.5, "step": 1351 }, { "epoch": 2.41, "learning_rate": 6.67007115898552e-05, "loss": 1.0322, "step": 1352 }, { "epoch": 2.42, "learning_rate": 6.663848771376115e-05, "loss": 1.7559, "step": 1353 }, { "epoch": 2.42, "learning_rate": 6.657623484690593e-05, "loss": 2.084, "step": 1354 }, { "epoch": 2.42, "learning_rate": 6.651395309775837e-05, "loss": 1.2217, "step": 1355 }, { "epoch": 2.42, "learning_rate": 6.645164257483766e-05, "loss": 2.0625, "step": 1356 }, { "epoch": 2.42, "learning_rate": 6.638930338671319e-05, "loss": 2.3555, "step": 1357 }, { "epoch": 2.42, "learning_rate": 6.632693564200416e-05, "loss": 1.791, "step": 1358 }, { "epoch": 2.43, "learning_rate": 6.626453944937969e-05, "loss": 1.0059, "step": 1359 }, { "epoch": 2.43, "learning_rate": 6.62021149175583e-05, "loss": 1.2998, "step": 1360 }, { "epoch": 2.43, "learning_rate": 6.613966215530803e-05, "loss": 1.7812, "step": 1361 }, { "epoch": 2.43, "learning_rate": 6.607718127144601e-05, "loss": 1.793, "step": 1362 }, { "epoch": 2.43, "learning_rate": 6.601467237483843e-05, "loss": 2.0762, "step": 1363 }, { "epoch": 2.44, "learning_rate": 6.595213557440025e-05, "loss": 2.0703, "step": 1364 }, { "epoch": 2.44, "learning_rate": 6.588957097909508e-05, "loss": 2.2148, "step": 1365 }, { "epoch": 2.44, "learning_rate": 6.582697869793495e-05, "loss": 1.1816, "step": 1366 }, { "epoch": 2.44, "learning_rate": 6.57643588399801e-05, "loss": 1.4502, "step": 1367 }, { "epoch": 2.44, "learning_rate": 6.570171151433887e-05, "loss": 2.0859, "step": 1368 }, { "epoch": 2.44, "learning_rate": 6.56390368301674e-05, "loss": 1.7812, "step": 1369 }, { "epoch": 2.45, "learning_rate": 6.557633489666958e-05, "loss": 1.876, "step": 1370 }, { "epoch": 2.45, "learning_rate": 6.55136058230967e-05, "loss": 2.3125, "step": 1371 }, { "epoch": 2.45, "learning_rate": 6.545084971874738e-05, "loss": 1.6592, "step": 1372 }, { "epoch": 2.45, "learning_rate": 6.538806669296732e-05, "loss": 1.5195, "step": 1373 }, { "epoch": 2.45, "learning_rate": 6.532525685514916e-05, "loss": 1.9004, "step": 1374 }, { "epoch": 2.46, "learning_rate": 6.52624203147322e-05, "loss": 2.2852, "step": 1375 }, { "epoch": 2.46, "learning_rate": 6.519955718120231e-05, "loss": 1.3398, "step": 1376 }, { "epoch": 2.46, "learning_rate": 6.513666756409171e-05, "loss": 1.4727, "step": 1377 }, { "epoch": 2.46, "learning_rate": 6.507375157297872e-05, "loss": 2.1113, "step": 1378 }, { "epoch": 2.46, "learning_rate": 6.501080931748764e-05, "loss": 2.0566, "step": 1379 }, { "epoch": 2.46, "learning_rate": 6.494784090728852e-05, "loss": 1.7373, "step": 1380 }, { "epoch": 2.47, "learning_rate": 6.4884846452097e-05, "loss": 2.3008, "step": 1381 }, { "epoch": 2.47, "learning_rate": 6.482182606167408e-05, "loss": 1.7715, "step": 1382 }, { "epoch": 2.47, "learning_rate": 6.475877984582596e-05, "loss": 2.1191, "step": 1383 }, { "epoch": 2.47, "learning_rate": 6.469570791440384e-05, "loss": 2.4805, "step": 1384 }, { "epoch": 2.47, "learning_rate": 6.463261037730373e-05, "loss": 1.7119, "step": 1385 }, { "epoch": 2.48, "learning_rate": 6.456948734446624e-05, "loss": 2.582, "step": 1386 }, { "epoch": 2.48, "learning_rate": 6.450633892587641e-05, "loss": 1.5869, "step": 1387 }, { "epoch": 2.48, "learning_rate": 6.444316523156352e-05, "loss": 2.6113, "step": 1388 }, { "epoch": 2.48, "learning_rate": 6.437996637160087e-05, "loss": 1.332, "step": 1389 }, { "epoch": 2.48, "learning_rate": 6.431674245610564e-05, "loss": 1.8691, "step": 1390 }, { "epoch": 2.48, "learning_rate": 6.425349359523864e-05, "loss": 2.1621, "step": 1391 }, { "epoch": 2.49, "learning_rate": 6.419021989920416e-05, "loss": 1.625, "step": 1392 }, { "epoch": 2.49, "learning_rate": 6.412692147824976e-05, "loss": 2.3984, "step": 1393 }, { "epoch": 2.49, "learning_rate": 6.406359844266607e-05, "loss": 1.8848, "step": 1394 }, { "epoch": 2.49, "learning_rate": 6.400025090278661e-05, "loss": 1.9414, "step": 1395 }, { "epoch": 2.49, "learning_rate": 6.393687896898759e-05, "loss": 1.5527, "step": 1396 }, { "epoch": 2.49, "learning_rate": 6.387348275168778e-05, "loss": 1.4844, "step": 1397 }, { "epoch": 2.5, "learning_rate": 6.381006236134817e-05, "loss": 2.4727, "step": 1398 }, { "epoch": 2.5, "learning_rate": 6.374661790847193e-05, "loss": 1.9922, "step": 1399 }, { "epoch": 2.5, "learning_rate": 6.368314950360415e-05, "loss": 2.6094, "step": 1400 }, { "epoch": 2.5, "learning_rate": 6.361965725733164e-05, "loss": 1.9932, "step": 1401 }, { "epoch": 2.5, "learning_rate": 6.355614128028276e-05, "loss": 1.9277, "step": 1402 }, { "epoch": 2.51, "learning_rate": 6.349260168312721e-05, "loss": 1.6348, "step": 1403 }, { "epoch": 2.51, "learning_rate": 6.342903857657584e-05, "loss": 1.9434, "step": 1404 }, { "epoch": 2.51, "learning_rate": 6.336545207138051e-05, "loss": 1.5645, "step": 1405 }, { "epoch": 2.51, "learning_rate": 6.330184227833376e-05, "loss": 1.7031, "step": 1406 }, { "epoch": 2.51, "learning_rate": 6.323820930826879e-05, "loss": 2.5508, "step": 1407 }, { "epoch": 2.51, "learning_rate": 6.317455327205915e-05, "loss": 1.6738, "step": 1408 }, { "epoch": 2.52, "learning_rate": 6.311087428061859e-05, "loss": 1.4678, "step": 1409 }, { "epoch": 2.52, "learning_rate": 6.304717244490084e-05, "loss": 2.2656, "step": 1410 }, { "epoch": 2.52, "learning_rate": 6.298344787589944e-05, "loss": 1.8643, "step": 1411 }, { "epoch": 2.52, "learning_rate": 6.291970068464754e-05, "loss": 2.3984, "step": 1412 }, { "epoch": 2.52, "learning_rate": 6.285593098221774e-05, "loss": 2.2578, "step": 1413 }, { "epoch": 2.52, "learning_rate": 6.279213887972179e-05, "loss": 2.2266, "step": 1414 }, { "epoch": 2.53, "learning_rate": 6.272832448831054e-05, "loss": 2.6016, "step": 1415 }, { "epoch": 2.53, "learning_rate": 6.266448791917365e-05, "loss": 1.377, "step": 1416 }, { "epoch": 2.53, "learning_rate": 6.26006292835394e-05, "loss": 2.3906, "step": 1417 }, { "epoch": 2.53, "learning_rate": 6.253674869267456e-05, "loss": 1.8711, "step": 1418 }, { "epoch": 2.53, "learning_rate": 6.247284625788414e-05, "loss": 1.2715, "step": 1419 }, { "epoch": 2.54, "learning_rate": 6.240892209051121e-05, "loss": 1.6279, "step": 1420 }, { "epoch": 2.54, "learning_rate": 6.234497630193666e-05, "loss": 2.2305, "step": 1421 }, { "epoch": 2.54, "learning_rate": 6.228100900357914e-05, "loss": 2.2891, "step": 1422 }, { "epoch": 2.54, "learning_rate": 6.22170203068947e-05, "loss": 2.2617, "step": 1423 }, { "epoch": 2.54, "learning_rate": 6.215301032337674e-05, "loss": 1.9775, "step": 1424 }, { "epoch": 2.54, "learning_rate": 6.20889791645557e-05, "loss": 2.1289, "step": 1425 }, { "epoch": 2.55, "learning_rate": 6.202492694199893e-05, "loss": 2.0664, "step": 1426 }, { "epoch": 2.55, "learning_rate": 6.19608537673105e-05, "loss": 2.1016, "step": 1427 }, { "epoch": 2.55, "learning_rate": 6.189675975213094e-05, "loss": 1.8789, "step": 1428 }, { "epoch": 2.55, "learning_rate": 6.183264500813717e-05, "loss": 2.1934, "step": 1429 }, { "epoch": 2.55, "learning_rate": 6.176850964704213e-05, "loss": 2.0664, "step": 1430 }, { "epoch": 2.56, "learning_rate": 6.170435378059478e-05, "loss": 1.2012, "step": 1431 }, { "epoch": 2.56, "learning_rate": 6.164017752057972e-05, "loss": 1.6055, "step": 1432 }, { "epoch": 2.56, "learning_rate": 6.157598097881716e-05, "loss": 1.207, "step": 1433 }, { "epoch": 2.56, "learning_rate": 6.15117642671626e-05, "loss": 2.0156, "step": 1434 }, { "epoch": 2.56, "learning_rate": 6.14475274975067e-05, "loss": 1.9824, "step": 1435 }, { "epoch": 2.56, "learning_rate": 6.138327078177509e-05, "loss": 1.5898, "step": 1436 }, { "epoch": 2.57, "learning_rate": 6.13189942319281e-05, "loss": 2.3594, "step": 1437 }, { "epoch": 2.57, "learning_rate": 6.125469795996065e-05, "loss": 1.709, "step": 1438 }, { "epoch": 2.57, "learning_rate": 6.119038207790205e-05, "loss": 1.9121, "step": 1439 }, { "epoch": 2.57, "learning_rate": 6.112604669781572e-05, "loss": 2.2383, "step": 1440 }, { "epoch": 2.57, "learning_rate": 6.106169193179912e-05, "loss": 2.3125, "step": 1441 }, { "epoch": 2.58, "learning_rate": 6.099731789198344e-05, "loss": 1.5977, "step": 1442 }, { "epoch": 2.58, "learning_rate": 6.093292469053345e-05, "loss": 1.6406, "step": 1443 }, { "epoch": 2.58, "learning_rate": 6.086851243964734e-05, "loss": 2.2656, "step": 1444 }, { "epoch": 2.58, "learning_rate": 6.080408125155649e-05, "loss": 1.2256, "step": 1445 }, { "epoch": 2.58, "learning_rate": 6.073963123852522e-05, "loss": 2.4512, "step": 1446 }, { "epoch": 2.58, "learning_rate": 6.0675162512850735e-05, "loss": 2.4082, "step": 1447 }, { "epoch": 2.59, "learning_rate": 6.061067518686276e-05, "loss": 1.7246, "step": 1448 }, { "epoch": 2.59, "learning_rate": 6.05461693729235e-05, "loss": 2.1289, "step": 1449 }, { "epoch": 2.59, "learning_rate": 6.048164518342734e-05, "loss": 1.7051, "step": 1450 }, { "epoch": 2.59, "learning_rate": 6.0417102730800654e-05, "loss": 1.7031, "step": 1451 }, { "epoch": 2.59, "learning_rate": 6.035254212750172e-05, "loss": 2.8477, "step": 1452 }, { "epoch": 2.59, "learning_rate": 6.028796348602035e-05, "loss": 2.3691, "step": 1453 }, { "epoch": 2.6, "learning_rate": 6.0223366918877845e-05, "loss": 1.3721, "step": 1454 }, { "epoch": 2.6, "learning_rate": 6.015875253862672e-05, "loss": 1.4492, "step": 1455 }, { "epoch": 2.6, "learning_rate": 6.009412045785051e-05, "loss": 2.6953, "step": 1456 }, { "epoch": 2.6, "learning_rate": 6.0029470789163646e-05, "loss": 2.4648, "step": 1457 }, { "epoch": 2.6, "learning_rate": 5.996480364521113e-05, "loss": 1.71, "step": 1458 }, { "epoch": 2.61, "learning_rate": 5.990011913866847e-05, "loss": 2.2773, "step": 1459 }, { "epoch": 2.61, "learning_rate": 5.983541738224141e-05, "loss": 1.5068, "step": 1460 }, { "epoch": 2.61, "learning_rate": 5.977069848866574e-05, "loss": 1.5664, "step": 1461 }, { "epoch": 2.61, "learning_rate": 5.970596257070711e-05, "loss": 1.9805, "step": 1462 }, { "epoch": 2.61, "learning_rate": 5.9641209741160855e-05, "loss": 1.3945, "step": 1463 }, { "epoch": 2.61, "learning_rate": 5.957644011285173e-05, "loss": 2.459, "step": 1464 }, { "epoch": 2.62, "learning_rate": 5.951165379863382e-05, "loss": 1.7363, "step": 1465 }, { "epoch": 2.62, "learning_rate": 5.9446850911390265e-05, "loss": 2.3164, "step": 1466 }, { "epoch": 2.62, "learning_rate": 5.938203156403304e-05, "loss": 1.9785, "step": 1467 }, { "epoch": 2.62, "learning_rate": 5.931719586950286e-05, "loss": 1.9248, "step": 1468 }, { "epoch": 2.62, "learning_rate": 5.925234394076886e-05, "loss": 1.7246, "step": 1469 }, { "epoch": 2.62, "learning_rate": 5.918747589082853e-05, "loss": 2.082, "step": 1470 }, { "epoch": 2.63, "learning_rate": 5.912259183270739e-05, "loss": 1.6719, "step": 1471 }, { "epoch": 2.63, "learning_rate": 5.905769187945889e-05, "loss": 1.8994, "step": 1472 }, { "epoch": 2.63, "learning_rate": 5.8992776144164164e-05, "loss": 1.4961, "step": 1473 }, { "epoch": 2.63, "learning_rate": 5.8927844739931834e-05, "loss": 1.8203, "step": 1474 }, { "epoch": 2.63, "learning_rate": 5.886289777989786e-05, "loss": 1.5566, "step": 1475 }, { "epoch": 2.64, "learning_rate": 5.8797935377225246e-05, "loss": 1.4629, "step": 1476 }, { "epoch": 2.64, "learning_rate": 5.873295764510395e-05, "loss": 1.9248, "step": 1477 }, { "epoch": 2.64, "learning_rate": 5.866796469675062e-05, "loss": 1.998, "step": 1478 }, { "epoch": 2.64, "learning_rate": 5.860295664540844e-05, "loss": 2.1699, "step": 1479 }, { "epoch": 2.64, "learning_rate": 5.853793360434687e-05, "loss": 1.665, "step": 1480 }, { "epoch": 2.64, "learning_rate": 5.847289568686155e-05, "loss": 1.793, "step": 1481 }, { "epoch": 2.65, "learning_rate": 5.840784300627396e-05, "loss": 2.2812, "step": 1482 }, { "epoch": 2.65, "learning_rate": 5.834277567593134e-05, "loss": 2.2227, "step": 1483 }, { "epoch": 2.65, "learning_rate": 5.82776938092065e-05, "loss": 1.9775, "step": 1484 }, { "epoch": 2.65, "learning_rate": 5.82125975194975e-05, "loss": 1.918, "step": 1485 }, { "epoch": 2.65, "learning_rate": 5.814748692022761e-05, "loss": 1.8574, "step": 1486 }, { "epoch": 2.66, "learning_rate": 5.8082362124844946e-05, "loss": 2.0508, "step": 1487 }, { "epoch": 2.66, "learning_rate": 5.8017223246822426e-05, "loss": 2.498, "step": 1488 }, { "epoch": 2.66, "learning_rate": 5.7952070399657475e-05, "loss": 2.4824, "step": 1489 }, { "epoch": 2.66, "learning_rate": 5.788690369687187e-05, "loss": 1.5186, "step": 1490 }, { "epoch": 2.66, "learning_rate": 5.782172325201155e-05, "loss": 2.3516, "step": 1491 }, { "epoch": 2.66, "learning_rate": 5.775652917864634e-05, "loss": 2.2227, "step": 1492 }, { "epoch": 2.67, "learning_rate": 5.769132159036986e-05, "loss": 2.0859, "step": 1493 }, { "epoch": 2.67, "learning_rate": 5.762610060079926e-05, "loss": 1.4746, "step": 1494 }, { "epoch": 2.67, "learning_rate": 5.7560866323575056e-05, "loss": 1.502, "step": 1495 }, { "epoch": 2.67, "learning_rate": 5.749561887236088e-05, "loss": 1.6855, "step": 1496 }, { "epoch": 2.67, "learning_rate": 5.7430358360843364e-05, "loss": 2.0781, "step": 1497 }, { "epoch": 2.67, "learning_rate": 5.736508490273188e-05, "loss": 1.7441, "step": 1498 }, { "epoch": 2.68, "learning_rate": 5.7299798611758336e-05, "loss": 2.3945, "step": 1499 }, { "epoch": 2.68, "learning_rate": 5.7234499601677026e-05, "loss": 1.7168, "step": 1500 }, { "epoch": 2.68, "learning_rate": 5.716918798626438e-05, "loss": 2.0254, "step": 1501 }, { "epoch": 2.68, "learning_rate": 5.7103863879318854e-05, "loss": 2.418, "step": 1502 }, { "epoch": 2.68, "learning_rate": 5.7038527394660555e-05, "loss": 2.0801, "step": 1503 }, { "epoch": 2.69, "learning_rate": 5.697317864613127e-05, "loss": 2.291, "step": 1504 }, { "epoch": 2.69, "learning_rate": 5.6907817747594116e-05, "loss": 1.5479, "step": 1505 }, { "epoch": 2.69, "learning_rate": 5.684244481293335e-05, "loss": 1.4326, "step": 1506 }, { "epoch": 2.69, "learning_rate": 5.6777059956054236e-05, "loss": 1.9258, "step": 1507 }, { "epoch": 2.69, "learning_rate": 5.6711663290882776e-05, "loss": 1.6543, "step": 1508 }, { "epoch": 2.69, "learning_rate": 5.6646254931365615e-05, "loss": 2.5312, "step": 1509 }, { "epoch": 2.7, "learning_rate": 5.6580834991469675e-05, "loss": 2.1621, "step": 1510 }, { "epoch": 2.7, "learning_rate": 5.651540358518217e-05, "loss": 1.6914, "step": 1511 }, { "epoch": 2.7, "learning_rate": 5.644996082651017e-05, "loss": 1.5664, "step": 1512 }, { "epoch": 2.7, "learning_rate": 5.6384506829480634e-05, "loss": 1.9971, "step": 1513 }, { "epoch": 2.7, "learning_rate": 5.631904170814004e-05, "loss": 1.7676, "step": 1514 }, { "epoch": 2.71, "learning_rate": 5.625356557655426e-05, "loss": 1.2041, "step": 1515 }, { "epoch": 2.71, "learning_rate": 5.6188078548808366e-05, "loss": 1.3105, "step": 1516 }, { "epoch": 2.71, "learning_rate": 5.6122580739006406e-05, "loss": 1.8486, "step": 1517 }, { "epoch": 2.71, "learning_rate": 5.6057072261271194e-05, "loss": 2.7812, "step": 1518 }, { "epoch": 2.71, "learning_rate": 5.599155322974417e-05, "loss": 2.2734, "step": 1519 }, { "epoch": 2.71, "learning_rate": 5.5926023758585146e-05, "loss": 1.6699, "step": 1520 }, { "epoch": 2.72, "learning_rate": 5.58604839619721e-05, "loss": 1.8184, "step": 1521 }, { "epoch": 2.72, "learning_rate": 5.579493395410105e-05, "loss": 2.2734, "step": 1522 }, { "epoch": 2.72, "learning_rate": 5.572937384918576e-05, "loss": 2.1016, "step": 1523 }, { "epoch": 2.72, "learning_rate": 5.5663803761457624e-05, "loss": 2.1289, "step": 1524 }, { "epoch": 2.72, "learning_rate": 5.559822380516539e-05, "loss": 1.6592, "step": 1525 }, { "epoch": 2.73, "learning_rate": 5.553263409457504e-05, "loss": 2.0625, "step": 1526 }, { "epoch": 2.73, "learning_rate": 5.546703474396954e-05, "loss": 1.9131, "step": 1527 }, { "epoch": 2.73, "learning_rate": 5.5401425867648614e-05, "loss": 1.8906, "step": 1528 }, { "epoch": 2.73, "learning_rate": 5.5335807579928655e-05, "loss": 1.748, "step": 1529 }, { "epoch": 2.73, "learning_rate": 5.527017999514239e-05, "loss": 2.2969, "step": 1530 }, { "epoch": 2.73, "learning_rate": 5.5204543227638784e-05, "loss": 1.998, "step": 1531 }, { "epoch": 2.74, "learning_rate": 5.51388973917828e-05, "loss": 2.5059, "step": 1532 }, { "epoch": 2.74, "learning_rate": 5.507324260195515e-05, "loss": 2.6406, "step": 1533 }, { "epoch": 2.74, "learning_rate": 5.5007578972552244e-05, "loss": 1.8848, "step": 1534 }, { "epoch": 2.74, "learning_rate": 5.4941906617985785e-05, "loss": 2.3398, "step": 1535 }, { "epoch": 2.74, "learning_rate": 5.487622565268278e-05, "loss": 2.0039, "step": 1536 }, { "epoch": 2.74, "learning_rate": 5.481053619108515e-05, "loss": 1.8799, "step": 1537 }, { "epoch": 2.75, "learning_rate": 5.474483834764967e-05, "loss": 2.2461, "step": 1538 }, { "epoch": 2.75, "learning_rate": 5.467913223684774e-05, "loss": 1.5791, "step": 1539 }, { "epoch": 2.75, "learning_rate": 5.4613417973165106e-05, "loss": 1.5371, "step": 1540 }, { "epoch": 2.75, "learning_rate": 5.4547695671101764e-05, "loss": 1.7871, "step": 1541 }, { "epoch": 2.75, "learning_rate": 5.448196544517168e-05, "loss": 2.918, "step": 1542 }, { "epoch": 2.76, "learning_rate": 5.441622740990268e-05, "loss": 1.625, "step": 1543 }, { "epoch": 2.76, "learning_rate": 5.435048167983613e-05, "loss": 1.9375, "step": 1544 }, { "epoch": 2.76, "learning_rate": 5.4284728369526846e-05, "loss": 1.9199, "step": 1545 }, { "epoch": 2.76, "learning_rate": 5.421896759354288e-05, "loss": 2.0293, "step": 1546 }, { "epoch": 2.76, "learning_rate": 5.41531994664652e-05, "loss": 1.5156, "step": 1547 }, { "epoch": 2.76, "learning_rate": 5.408742410288769e-05, "loss": 1.6963, "step": 1548 }, { "epoch": 2.77, "learning_rate": 5.402164161741674e-05, "loss": 1.1953, "step": 1549 }, { "epoch": 2.77, "learning_rate": 5.3955852124671234e-05, "loss": 2.0449, "step": 1550 }, { "epoch": 2.77, "learning_rate": 5.389005573928223e-05, "loss": 2.3066, "step": 1551 }, { "epoch": 2.77, "learning_rate": 5.3824252575892776e-05, "loss": 1.8633, "step": 1552 }, { "epoch": 2.77, "learning_rate": 5.375844274915778e-05, "loss": 1.8809, "step": 1553 }, { "epoch": 2.77, "learning_rate": 5.3692626373743706e-05, "loss": 1.8027, "step": 1554 }, { "epoch": 2.78, "learning_rate": 5.3626803564328465e-05, "loss": 2.2852, "step": 1555 }, { "epoch": 2.78, "learning_rate": 5.356097443560116e-05, "loss": 2.1836, "step": 1556 }, { "epoch": 2.78, "learning_rate": 5.3495139102261924e-05, "loss": 1.4766, "step": 1557 }, { "epoch": 2.78, "learning_rate": 5.3429297679021684e-05, "loss": 2.0996, "step": 1558 }, { "epoch": 2.78, "learning_rate": 5.336345028060199e-05, "loss": 1.7402, "step": 1559 }, { "epoch": 2.79, "learning_rate": 5.329759702173477e-05, "loss": 1.7031, "step": 1560 }, { "epoch": 2.79, "learning_rate": 5.3231738017162214e-05, "loss": 2.2578, "step": 1561 }, { "epoch": 2.79, "learning_rate": 5.3165873381636486e-05, "loss": 1.7734, "step": 1562 }, { "epoch": 2.79, "learning_rate": 5.3100003229919583e-05, "loss": 1.8213, "step": 1563 }, { "epoch": 2.79, "learning_rate": 5.3034127676783094e-05, "loss": 1.4424, "step": 1564 }, { "epoch": 2.79, "learning_rate": 5.2968246837008026e-05, "loss": 2.2285, "step": 1565 }, { "epoch": 2.8, "learning_rate": 5.2902360825384636e-05, "loss": 2.0664, "step": 1566 }, { "epoch": 2.8, "learning_rate": 5.2836469756712106e-05, "loss": 2.0664, "step": 1567 }, { "epoch": 2.8, "learning_rate": 5.27705737457985e-05, "loss": 1.9355, "step": 1568 }, { "epoch": 2.8, "learning_rate": 5.270467290746048e-05, "loss": 1.9375, "step": 1569 }, { "epoch": 2.8, "learning_rate": 5.2638767356523125e-05, "loss": 1.6943, "step": 1570 }, { "epoch": 2.81, "learning_rate": 5.257285720781969e-05, "loss": 1.9346, "step": 1571 }, { "epoch": 2.81, "learning_rate": 5.2506942576191464e-05, "loss": 2.0273, "step": 1572 }, { "epoch": 2.81, "learning_rate": 5.244102357648757e-05, "loss": 2.3535, "step": 1573 }, { "epoch": 2.81, "learning_rate": 5.2375100323564695e-05, "loss": 1.5195, "step": 1574 }, { "epoch": 2.81, "learning_rate": 5.230917293228699e-05, "loss": 1.5684, "step": 1575 }, { "epoch": 2.81, "learning_rate": 5.2243241517525754e-05, "loss": 2.1562, "step": 1576 }, { "epoch": 2.82, "learning_rate": 5.217730619415935e-05, "loss": 2.0996, "step": 1577 }, { "epoch": 2.82, "learning_rate": 5.2111367077072924e-05, "loss": 2.0664, "step": 1578 }, { "epoch": 2.82, "learning_rate": 5.204542428115824e-05, "loss": 2.6992, "step": 1579 }, { "epoch": 2.82, "learning_rate": 5.197947792131348e-05, "loss": 1.7266, "step": 1580 }, { "epoch": 2.82, "learning_rate": 5.191352811244301e-05, "loss": 1.5205, "step": 1581 }, { "epoch": 2.83, "learning_rate": 5.184757496945726e-05, "loss": 2.1465, "step": 1582 }, { "epoch": 2.83, "learning_rate": 5.1781618607272386e-05, "loss": 1.8486, "step": 1583 }, { "epoch": 2.83, "learning_rate": 5.171565914081023e-05, "loss": 2.1699, "step": 1584 }, { "epoch": 2.83, "learning_rate": 5.164969668499798e-05, "loss": 2.0273, "step": 1585 }, { "epoch": 2.83, "learning_rate": 5.1583731354768105e-05, "loss": 2.1582, "step": 1586 }, { "epoch": 2.83, "learning_rate": 5.151776326505803e-05, "loss": 1.6475, "step": 1587 }, { "epoch": 2.84, "learning_rate": 5.1451792530809973e-05, "loss": 1.9912, "step": 1588 }, { "epoch": 2.84, "learning_rate": 5.138581926697082e-05, "loss": 2.1523, "step": 1589 }, { "epoch": 2.84, "learning_rate": 5.131984358849182e-05, "loss": 2.707, "step": 1590 }, { "epoch": 2.84, "learning_rate": 5.125386561032846e-05, "loss": 2.3047, "step": 1591 }, { "epoch": 2.84, "learning_rate": 5.118788544744015e-05, "loss": 2.4453, "step": 1592 }, { "epoch": 2.84, "learning_rate": 5.112190321479026e-05, "loss": 1.8438, "step": 1593 }, { "epoch": 2.85, "learning_rate": 5.105591902734561e-05, "loss": 1.6553, "step": 1594 }, { "epoch": 2.85, "learning_rate": 5.0989933000076526e-05, "loss": 2.293, "step": 1595 }, { "epoch": 2.85, "learning_rate": 5.092394524795649e-05, "loss": 1.9707, "step": 1596 }, { "epoch": 2.85, "learning_rate": 5.085795588596203e-05, "loss": 1.959, "step": 1597 }, { "epoch": 2.85, "learning_rate": 5.079196502907245e-05, "loss": 1.6074, "step": 1598 }, { "epoch": 2.86, "learning_rate": 5.0725972792269637e-05, "loss": 1.3848, "step": 1599 }, { "epoch": 2.86, "learning_rate": 5.0659979290537954e-05, "loss": 1.8691, "step": 1600 }, { "epoch": 2.86, "learning_rate": 5.059398463886388e-05, "loss": 1.7773, "step": 1601 }, { "epoch": 2.86, "learning_rate": 5.052798895223597e-05, "loss": 2.6758, "step": 1602 }, { "epoch": 2.86, "learning_rate": 5.046199234564455e-05, "loss": 1.8135, "step": 1603 }, { "epoch": 2.86, "learning_rate": 5.0395994934081534e-05, "loss": 2.3242, "step": 1604 }, { "epoch": 2.87, "learning_rate": 5.0329996832540285e-05, "loss": 1.4238, "step": 1605 }, { "epoch": 2.87, "learning_rate": 5.0263998156015324e-05, "loss": 2.1816, "step": 1606 }, { "epoch": 2.87, "learning_rate": 5.0197999019502195e-05, "loss": 1.3535, "step": 1607 }, { "epoch": 2.87, "learning_rate": 5.013199953799723e-05, "loss": 2.2422, "step": 1608 }, { "epoch": 2.87, "learning_rate": 5.00659998264974e-05, "loss": 1.7598, "step": 1609 }, { "epoch": 2.88, "learning_rate": 5e-05, "loss": 1.6631, "step": 1610 }, { "epoch": 2.88, "learning_rate": 4.993400017350262e-05, "loss": 2.0625, "step": 1611 }, { "epoch": 2.88, "learning_rate": 4.9868000462002775e-05, "loss": 1.8623, "step": 1612 }, { "epoch": 2.88, "learning_rate": 4.9802000980497816e-05, "loss": 1.8203, "step": 1613 }, { "epoch": 2.88, "learning_rate": 4.9736001843984695e-05, "loss": 2.332, "step": 1614 }, { "epoch": 2.88, "learning_rate": 4.967000316745972e-05, "loss": 1.877, "step": 1615 }, { "epoch": 2.89, "learning_rate": 4.960400506591848e-05, "loss": 1.957, "step": 1616 }, { "epoch": 2.89, "learning_rate": 4.953800765435547e-05, "loss": 1.7207, "step": 1617 }, { "epoch": 2.89, "learning_rate": 4.947201104776404e-05, "loss": 2.3164, "step": 1618 }, { "epoch": 2.89, "learning_rate": 4.940601536113613e-05, "loss": 2.3516, "step": 1619 }, { "epoch": 2.89, "learning_rate": 4.934002070946206e-05, "loss": 1.3467, "step": 1620 }, { "epoch": 2.89, "learning_rate": 4.9274027207730375e-05, "loss": 2.002, "step": 1621 }, { "epoch": 2.9, "learning_rate": 4.920803497092756e-05, "loss": 1.7383, "step": 1622 }, { "epoch": 2.9, "learning_rate": 4.9142044114037975e-05, "loss": 2.0352, "step": 1623 }, { "epoch": 2.9, "learning_rate": 4.907605475204352e-05, "loss": 1.9609, "step": 1624 }, { "epoch": 2.9, "learning_rate": 4.901006699992348e-05, "loss": 1.8447, "step": 1625 }, { "epoch": 2.9, "learning_rate": 4.894408097265441e-05, "loss": 1.9629, "step": 1626 }, { "epoch": 2.91, "learning_rate": 4.887809678520976e-05, "loss": 1.6455, "step": 1627 }, { "epoch": 2.91, "learning_rate": 4.8812114552559854e-05, "loss": 1.4023, "step": 1628 }, { "epoch": 2.91, "learning_rate": 4.874613438967157e-05, "loss": 1.4023, "step": 1629 }, { "epoch": 2.91, "learning_rate": 4.86801564115082e-05, "loss": 1.4199, "step": 1630 }, { "epoch": 2.91, "learning_rate": 4.861418073302919e-05, "loss": 1.7285, "step": 1631 }, { "epoch": 2.91, "learning_rate": 4.8548207469190045e-05, "loss": 2.2305, "step": 1632 }, { "epoch": 2.92, "learning_rate": 4.8482236734942e-05, "loss": 1.2568, "step": 1633 }, { "epoch": 2.92, "learning_rate": 4.841626864523191e-05, "loss": 2.166, "step": 1634 }, { "epoch": 2.92, "learning_rate": 4.835030331500204e-05, "loss": 2.0176, "step": 1635 }, { "epoch": 2.92, "learning_rate": 4.82843408591898e-05, "loss": 1.3535, "step": 1636 }, { "epoch": 2.92, "learning_rate": 4.821838139272764e-05, "loss": 2.4258, "step": 1637 }, { "epoch": 2.92, "learning_rate": 4.8152425030542766e-05, "loss": 2.1719, "step": 1638 }, { "epoch": 2.93, "learning_rate": 4.8086471887557004e-05, "loss": 1.876, "step": 1639 }, { "epoch": 2.93, "learning_rate": 4.802052207868654e-05, "loss": 2.1582, "step": 1640 }, { "epoch": 2.93, "learning_rate": 4.795457571884177e-05, "loss": 1.0742, "step": 1641 }, { "epoch": 2.93, "learning_rate": 4.78886329229271e-05, "loss": 1.6104, "step": 1642 }, { "epoch": 2.93, "learning_rate": 4.782269380584067e-05, "loss": 2.1895, "step": 1643 }, { "epoch": 2.94, "learning_rate": 4.775675848247427e-05, "loss": 1.9844, "step": 1644 }, { "epoch": 2.94, "learning_rate": 4.7690827067713035e-05, "loss": 1.9414, "step": 1645 }, { "epoch": 2.94, "learning_rate": 4.762489967643532e-05, "loss": 2.5879, "step": 1646 }, { "epoch": 2.94, "learning_rate": 4.7558976423512446e-05, "loss": 1.7969, "step": 1647 }, { "epoch": 2.94, "learning_rate": 4.7493057423808534e-05, "loss": 1.335, "step": 1648 }, { "epoch": 2.94, "learning_rate": 4.7427142792180314e-05, "loss": 2.7461, "step": 1649 }, { "epoch": 2.95, "learning_rate": 4.736123264347688e-05, "loss": 1.959, "step": 1650 }, { "epoch": 2.95, "learning_rate": 4.7295327092539517e-05, "loss": 2.4316, "step": 1651 }, { "epoch": 2.95, "learning_rate": 4.72294262542015e-05, "loss": 1.6738, "step": 1652 }, { "epoch": 2.95, "learning_rate": 4.71635302432879e-05, "loss": 1.8027, "step": 1653 }, { "epoch": 2.95, "learning_rate": 4.709763917461537e-05, "loss": 1.623, "step": 1654 }, { "epoch": 2.96, "learning_rate": 4.7031753162991966e-05, "loss": 1.666, "step": 1655 }, { "epoch": 2.96, "learning_rate": 4.6965872323216904e-05, "loss": 2.0195, "step": 1656 }, { "epoch": 2.96, "learning_rate": 4.689999677008042e-05, "loss": 2.6445, "step": 1657 }, { "epoch": 2.96, "learning_rate": 4.683412661836351e-05, "loss": 1.334, "step": 1658 }, { "epoch": 2.96, "learning_rate": 4.676826198283779e-05, "loss": 1.9502, "step": 1659 }, { "epoch": 2.96, "learning_rate": 4.670240297826523e-05, "loss": 1.6562, "step": 1660 }, { "epoch": 2.97, "learning_rate": 4.663654971939802e-05, "loss": 1.7695, "step": 1661 }, { "epoch": 2.97, "learning_rate": 4.6570702320978314e-05, "loss": 1.8555, "step": 1662 }, { "epoch": 2.97, "learning_rate": 4.650486089773807e-05, "loss": 1.7793, "step": 1663 }, { "epoch": 2.97, "learning_rate": 4.6439025564398844e-05, "loss": 1.5205, "step": 1664 }, { "epoch": 2.97, "learning_rate": 4.637319643567154e-05, "loss": 1.915, "step": 1665 }, { "epoch": 2.98, "learning_rate": 4.6307373626256306e-05, "loss": 1.6631, "step": 1666 }, { "epoch": 2.98, "learning_rate": 4.624155725084223e-05, "loss": 2.5391, "step": 1667 }, { "epoch": 2.98, "learning_rate": 4.6175747424107236e-05, "loss": 1.8184, "step": 1668 }, { "epoch": 2.98, "learning_rate": 4.610994426071779e-05, "loss": 2.0195, "step": 1669 }, { "epoch": 2.98, "learning_rate": 4.604414787532877e-05, "loss": 1.2041, "step": 1670 }, { "epoch": 2.98, "learning_rate": 4.597835838258327e-05, "loss": 2.9355, "step": 1671 }, { "epoch": 2.99, "learning_rate": 4.591257589711233e-05, "loss": 2.3438, "step": 1672 }, { "epoch": 2.99, "learning_rate": 4.5846800533534815e-05, "loss": 1.6914, "step": 1673 }, { "epoch": 2.99, "learning_rate": 4.578103240645714e-05, "loss": 1.7832, "step": 1674 }, { "epoch": 2.99, "learning_rate": 4.571527163047316e-05, "loss": 1.457, "step": 1675 }, { "epoch": 2.99, "learning_rate": 4.5649518320163885e-05, "loss": 1.6201, "step": 1676 }, { "epoch": 2.99, "learning_rate": 4.558377259009733e-05, "loss": 1.1855, "step": 1677 }, { "epoch": 3.0, "learning_rate": 4.551803455482833e-05, "loss": 2.0938, "step": 1678 }, { "epoch": 3.0, "learning_rate": 4.545230432889825e-05, "loss": 2.0586, "step": 1679 }, { "epoch": 3.0, "learning_rate": 4.5386582026834906e-05, "loss": 1.4951, "step": 1680 }, { "epoch": 3.0, "learning_rate": 4.532086776315227e-05, "loss": 1.3633, "step": 1681 }, { "epoch": 3.0, "learning_rate": 4.525516165235034e-05, "loss": 2.5117, "step": 1682 }, { "epoch": 3.01, "learning_rate": 4.518946380891487e-05, "loss": 2.4805, "step": 1683 }, { "epoch": 3.01, "learning_rate": 4.512377434731724e-05, "loss": 2.3535, "step": 1684 }, { "epoch": 3.01, "learning_rate": 4.5058093382014226e-05, "loss": 1.8975, "step": 1685 }, { "epoch": 3.01, "learning_rate": 4.4992421027447775e-05, "loss": 2.1484, "step": 1686 }, { "epoch": 3.01, "learning_rate": 4.492675739804486e-05, "loss": 1.8291, "step": 1687 }, { "epoch": 3.01, "learning_rate": 4.486110260821722e-05, "loss": 2.1172, "step": 1688 }, { "epoch": 3.02, "learning_rate": 4.479545677236123e-05, "loss": 1.8984, "step": 1689 }, { "epoch": 3.02, "learning_rate": 4.472982000485762e-05, "loss": 1.8975, "step": 1690 }, { "epoch": 3.02, "learning_rate": 4.466419242007135e-05, "loss": 1.9395, "step": 1691 }, { "epoch": 3.02, "learning_rate": 4.45985741323514e-05, "loss": 2.1719, "step": 1692 }, { "epoch": 3.02, "learning_rate": 4.4532965256030473e-05, "loss": 1.71, "step": 1693 }, { "epoch": 3.02, "learning_rate": 4.446736590542497e-05, "loss": 2.418, "step": 1694 }, { "epoch": 3.03, "learning_rate": 4.4401776194834613e-05, "loss": 2.0059, "step": 1695 }, { "epoch": 3.03, "learning_rate": 4.4336196238542394e-05, "loss": 1.6699, "step": 1696 }, { "epoch": 3.03, "learning_rate": 4.4270626150814244e-05, "loss": 1.9355, "step": 1697 }, { "epoch": 3.03, "learning_rate": 4.420506604589897e-05, "loss": 2.2441, "step": 1698 }, { "epoch": 3.03, "learning_rate": 4.4139516038027914e-05, "loss": 1.7041, "step": 1699 }, { "epoch": 3.04, "learning_rate": 4.407397624141487e-05, "loss": 1.4453, "step": 1700 }, { "epoch": 3.04, "learning_rate": 4.400844677025585e-05, "loss": 2.5547, "step": 1701 }, { "epoch": 3.04, "learning_rate": 4.3942927738728824e-05, "loss": 2.0059, "step": 1702 }, { "epoch": 3.04, "learning_rate": 4.3877419260993626e-05, "loss": 1.4629, "step": 1703 }, { "epoch": 3.04, "learning_rate": 4.381192145119165e-05, "loss": 2.1562, "step": 1704 }, { "epoch": 3.04, "learning_rate": 4.374643442344576e-05, "loss": 1.6738, "step": 1705 }, { "epoch": 3.05, "learning_rate": 4.3680958291859984e-05, "loss": 1.791, "step": 1706 }, { "epoch": 3.05, "learning_rate": 4.3615493170519384e-05, "loss": 1.5762, "step": 1707 }, { "epoch": 3.05, "learning_rate": 4.3550039173489845e-05, "loss": 1.5, "step": 1708 }, { "epoch": 3.05, "learning_rate": 4.3484596414817855e-05, "loss": 2.793, "step": 1709 }, { "epoch": 3.05, "learning_rate": 4.341916500853034e-05, "loss": 1.7754, "step": 1710 }, { "epoch": 3.06, "learning_rate": 4.335374506863441e-05, "loss": 1.5762, "step": 1711 }, { "epoch": 3.06, "learning_rate": 4.328833670911724e-05, "loss": 2.6328, "step": 1712 }, { "epoch": 3.06, "learning_rate": 4.322294004394579e-05, "loss": 2.1016, "step": 1713 }, { "epoch": 3.06, "learning_rate": 4.315755518706667e-05, "loss": 2.1953, "step": 1714 }, { "epoch": 3.06, "learning_rate": 4.30921822524059e-05, "loss": 1.3184, "step": 1715 }, { "epoch": 3.06, "learning_rate": 4.302682135386873e-05, "loss": 1.8984, "step": 1716 }, { "epoch": 3.07, "learning_rate": 4.2961472605339464e-05, "loss": 1.9893, "step": 1717 }, { "epoch": 3.07, "learning_rate": 4.289613612068117e-05, "loss": 2.3242, "step": 1718 }, { "epoch": 3.07, "learning_rate": 4.283081201373563e-05, "loss": 2.207, "step": 1719 }, { "epoch": 3.07, "learning_rate": 4.276550039832299e-05, "loss": 1.7451, "step": 1720 }, { "epoch": 3.07, "learning_rate": 4.2700201388241676e-05, "loss": 2.0527, "step": 1721 }, { "epoch": 3.08, "learning_rate": 4.2634915097268115e-05, "loss": 2.6934, "step": 1722 }, { "epoch": 3.08, "learning_rate": 4.2569641639156634e-05, "loss": 1.3799, "step": 1723 }, { "epoch": 3.08, "learning_rate": 4.250438112763912e-05, "loss": 2.2832, "step": 1724 }, { "epoch": 3.08, "learning_rate": 4.2439133676424955e-05, "loss": 2.1641, "step": 1725 }, { "epoch": 3.08, "learning_rate": 4.237389939920075e-05, "loss": 2.0918, "step": 1726 }, { "epoch": 3.08, "learning_rate": 4.230867840963015e-05, "loss": 2.2246, "step": 1727 }, { "epoch": 3.09, "learning_rate": 4.224347082135367e-05, "loss": 1.5244, "step": 1728 }, { "epoch": 3.09, "learning_rate": 4.2178276747988446e-05, "loss": 1.3633, "step": 1729 }, { "epoch": 3.09, "learning_rate": 4.2113096303128125e-05, "loss": 1.6006, "step": 1730 }, { "epoch": 3.09, "learning_rate": 4.204792960034252e-05, "loss": 2.2695, "step": 1731 }, { "epoch": 3.09, "learning_rate": 4.1982776753177585e-05, "loss": 1.8076, "step": 1732 }, { "epoch": 3.09, "learning_rate": 4.191763787515506e-05, "loss": 1.3047, "step": 1733 }, { "epoch": 3.1, "learning_rate": 4.1852513079772406e-05, "loss": 1.5898, "step": 1734 }, { "epoch": 3.1, "learning_rate": 4.178740248050249e-05, "loss": 1.2715, "step": 1735 }, { "epoch": 3.1, "learning_rate": 4.1722306190793495e-05, "loss": 1.9355, "step": 1736 }, { "epoch": 3.1, "learning_rate": 4.165722432406866e-05, "loss": 2.2969, "step": 1737 }, { "epoch": 3.1, "learning_rate": 4.159215699372605e-05, "loss": 1.4062, "step": 1738 }, { "epoch": 3.11, "learning_rate": 4.152710431313847e-05, "loss": 1.5488, "step": 1739 }, { "epoch": 3.11, "learning_rate": 4.146206639565312e-05, "loss": 1.3721, "step": 1740 }, { "epoch": 3.11, "learning_rate": 4.139704335459157e-05, "loss": 1.4287, "step": 1741 }, { "epoch": 3.11, "learning_rate": 4.133203530324938e-05, "loss": 1.4062, "step": 1742 }, { "epoch": 3.11, "learning_rate": 4.1267042354896056e-05, "loss": 2.207, "step": 1743 }, { "epoch": 3.11, "learning_rate": 4.120206462277477e-05, "loss": 2.0703, "step": 1744 }, { "epoch": 3.12, "learning_rate": 4.113710222010215e-05, "loss": 2.0723, "step": 1745 }, { "epoch": 3.12, "learning_rate": 4.107215526006817e-05, "loss": 2.0586, "step": 1746 }, { "epoch": 3.12, "learning_rate": 4.100722385583585e-05, "loss": 2.0215, "step": 1747 }, { "epoch": 3.12, "learning_rate": 4.094230812054112e-05, "loss": 1.2803, "step": 1748 }, { "epoch": 3.12, "learning_rate": 4.087740816729262e-05, "loss": 1.3379, "step": 1749 }, { "epoch": 3.12, "learning_rate": 4.0812524109171476e-05, "loss": 1.7559, "step": 1750 }, { "epoch": 3.13, "learning_rate": 4.074765605923115e-05, "loss": 2.0488, "step": 1751 }, { "epoch": 3.13, "learning_rate": 4.0682804130497156e-05, "loss": 2.1328, "step": 1752 }, { "epoch": 3.13, "learning_rate": 4.0617968435966966e-05, "loss": 1.9297, "step": 1753 }, { "epoch": 3.13, "learning_rate": 4.055314908860974e-05, "loss": 1.3154, "step": 1754 }, { "epoch": 3.13, "learning_rate": 4.048834620136618e-05, "loss": 1.583, "step": 1755 }, { "epoch": 3.14, "learning_rate": 4.042355988714828e-05, "loss": 2.0098, "step": 1756 }, { "epoch": 3.14, "learning_rate": 4.0358790258839164e-05, "loss": 1.9531, "step": 1757 }, { "epoch": 3.14, "learning_rate": 4.0294037429292905e-05, "loss": 2.2129, "step": 1758 }, { "epoch": 3.14, "learning_rate": 4.0229301511334275e-05, "loss": 1.8838, "step": 1759 }, { "epoch": 3.14, "learning_rate": 4.01645826177586e-05, "loss": 2.1094, "step": 1760 }, { "epoch": 3.14, "learning_rate": 4.009988086133153e-05, "loss": 1.8691, "step": 1761 }, { "epoch": 3.15, "learning_rate": 4.0035196354788886e-05, "loss": 1.8438, "step": 1762 }, { "epoch": 3.15, "learning_rate": 3.9970529210836366e-05, "loss": 2.1992, "step": 1763 }, { "epoch": 3.15, "learning_rate": 3.99058795421495e-05, "loss": 1.9844, "step": 1764 }, { "epoch": 3.15, "learning_rate": 3.98412474613733e-05, "loss": 1.8018, "step": 1765 }, { "epoch": 3.15, "learning_rate": 3.977663308112216e-05, "loss": 1.7822, "step": 1766 }, { "epoch": 3.16, "learning_rate": 3.971203651397966e-05, "loss": 1.3418, "step": 1767 }, { "epoch": 3.16, "learning_rate": 3.964745787249829e-05, "loss": 2.2578, "step": 1768 }, { "epoch": 3.16, "learning_rate": 3.958289726919935e-05, "loss": 1.9961, "step": 1769 }, { "epoch": 3.16, "learning_rate": 3.951835481657268e-05, "loss": 2.3164, "step": 1770 }, { "epoch": 3.16, "learning_rate": 3.9453830627076516e-05, "loss": 1.7637, "step": 1771 }, { "epoch": 3.16, "learning_rate": 3.9389324813137255e-05, "loss": 1.8555, "step": 1772 }, { "epoch": 3.17, "learning_rate": 3.932483748714928e-05, "loss": 1.2871, "step": 1773 }, { "epoch": 3.17, "learning_rate": 3.9260368761474786e-05, "loss": 1.415, "step": 1774 }, { "epoch": 3.17, "learning_rate": 3.919591874844352e-05, "loss": 2.0723, "step": 1775 }, { "epoch": 3.17, "learning_rate": 3.913148756035266e-05, "loss": 1.5703, "step": 1776 }, { "epoch": 3.17, "learning_rate": 3.906707530946656e-05, "loss": 1.8184, "step": 1777 }, { "epoch": 3.17, "learning_rate": 3.9002682108016585e-05, "loss": 1.873, "step": 1778 }, { "epoch": 3.18, "learning_rate": 3.893830806820089e-05, "loss": 1.291, "step": 1779 }, { "epoch": 3.18, "learning_rate": 3.887395330218429e-05, "loss": 2.1191, "step": 1780 }, { "epoch": 3.18, "learning_rate": 3.880961792209797e-05, "loss": 1.6143, "step": 1781 }, { "epoch": 3.18, "learning_rate": 3.8745302040039365e-05, "loss": 1.3691, "step": 1782 }, { "epoch": 3.18, "learning_rate": 3.868100576807193e-05, "loss": 2.25, "step": 1783 }, { "epoch": 3.19, "learning_rate": 3.8616729218224925e-05, "loss": 1.292, "step": 1784 }, { "epoch": 3.19, "learning_rate": 3.855247250249331e-05, "loss": 2.2246, "step": 1785 }, { "epoch": 3.19, "learning_rate": 3.848823573283741e-05, "loss": 1.8027, "step": 1786 }, { "epoch": 3.19, "learning_rate": 3.842401902118285e-05, "loss": 1.9053, "step": 1787 }, { "epoch": 3.19, "learning_rate": 3.83598224794203e-05, "loss": 2.5352, "step": 1788 }, { "epoch": 3.19, "learning_rate": 3.829564621940524e-05, "loss": 1.9092, "step": 1789 }, { "epoch": 3.2, "learning_rate": 3.823149035295789e-05, "loss": 1.957, "step": 1790 }, { "epoch": 3.2, "learning_rate": 3.816735499186286e-05, "loss": 1.1738, "step": 1791 }, { "epoch": 3.2, "learning_rate": 3.8103240247869075e-05, "loss": 1.709, "step": 1792 }, { "epoch": 3.2, "learning_rate": 3.8039146232689524e-05, "loss": 2.1211, "step": 1793 }, { "epoch": 3.2, "learning_rate": 3.797507305800109e-05, "loss": 1.8457, "step": 1794 }, { "epoch": 3.21, "learning_rate": 3.791102083544432e-05, "loss": 2.4297, "step": 1795 }, { "epoch": 3.21, "learning_rate": 3.784698967662326e-05, "loss": 1.7383, "step": 1796 }, { "epoch": 3.21, "learning_rate": 3.778297969310529e-05, "loss": 1.9902, "step": 1797 }, { "epoch": 3.21, "learning_rate": 3.771899099642087e-05, "loss": 1.7119, "step": 1798 }, { "epoch": 3.21, "learning_rate": 3.765502369806334e-05, "loss": 2.4023, "step": 1799 }, { "epoch": 3.21, "learning_rate": 3.759107790948882e-05, "loss": 1.5059, "step": 1800 }, { "epoch": 3.22, "learning_rate": 3.752715374211586e-05, "loss": 1.3945, "step": 1801 }, { "epoch": 3.22, "learning_rate": 3.746325130732543e-05, "loss": 1.3613, "step": 1802 }, { "epoch": 3.22, "learning_rate": 3.73993707164606e-05, "loss": 1.6211, "step": 1803 }, { "epoch": 3.22, "learning_rate": 3.7335512080826365e-05, "loss": 1.4658, "step": 1804 }, { "epoch": 3.22, "learning_rate": 3.7271675511689475e-05, "loss": 1.8643, "step": 1805 }, { "epoch": 3.23, "learning_rate": 3.720786112027822e-05, "loss": 1.5117, "step": 1806 }, { "epoch": 3.23, "learning_rate": 3.714406901778228e-05, "loss": 1.6777, "step": 1807 }, { "epoch": 3.23, "learning_rate": 3.7080299315352455e-05, "loss": 2.1855, "step": 1808 }, { "epoch": 3.23, "learning_rate": 3.7016552124100555e-05, "loss": 1.8574, "step": 1809 }, { "epoch": 3.23, "learning_rate": 3.695282755509917e-05, "loss": 1.4678, "step": 1810 }, { "epoch": 3.23, "learning_rate": 3.68891257193814e-05, "loss": 1.4902, "step": 1811 }, { "epoch": 3.24, "learning_rate": 3.682544672794085e-05, "loss": 1.8604, "step": 1812 }, { "epoch": 3.24, "learning_rate": 3.676179069173121e-05, "loss": 2.3125, "step": 1813 }, { "epoch": 3.24, "learning_rate": 3.6698157721666246e-05, "loss": 1.9941, "step": 1814 }, { "epoch": 3.24, "learning_rate": 3.6634547928619505e-05, "loss": 1.3203, "step": 1815 }, { "epoch": 3.24, "learning_rate": 3.657096142342415e-05, "loss": 2.1484, "step": 1816 }, { "epoch": 3.24, "learning_rate": 3.65073983168728e-05, "loss": 1.8066, "step": 1817 }, { "epoch": 3.25, "learning_rate": 3.644385871971725e-05, "loss": 1.5049, "step": 1818 }, { "epoch": 3.25, "learning_rate": 3.6380342742668374e-05, "loss": 1.5449, "step": 1819 }, { "epoch": 3.25, "learning_rate": 3.631685049639586e-05, "loss": 2.2031, "step": 1820 }, { "epoch": 3.25, "learning_rate": 3.625338209152809e-05, "loss": 1.9648, "step": 1821 }, { "epoch": 3.25, "learning_rate": 3.618993763865185e-05, "loss": 1.3594, "step": 1822 }, { "epoch": 3.26, "learning_rate": 3.6126517248312235e-05, "loss": 2.1094, "step": 1823 }, { "epoch": 3.26, "learning_rate": 3.606312103101241e-05, "loss": 2.0586, "step": 1824 }, { "epoch": 3.26, "learning_rate": 3.599974909721341e-05, "loss": 1.6934, "step": 1825 }, { "epoch": 3.26, "learning_rate": 3.593640155733396e-05, "loss": 1.5723, "step": 1826 }, { "epoch": 3.26, "learning_rate": 3.5873078521750246e-05, "loss": 1.6074, "step": 1827 }, { "epoch": 3.26, "learning_rate": 3.580978010079585e-05, "loss": 2.4766, "step": 1828 }, { "epoch": 3.27, "learning_rate": 3.5746506404761366e-05, "loss": 1.9902, "step": 1829 }, { "epoch": 3.27, "learning_rate": 3.568325754389438e-05, "loss": 2.1016, "step": 1830 }, { "epoch": 3.27, "learning_rate": 3.562003362839914e-05, "loss": 2.2832, "step": 1831 }, { "epoch": 3.27, "learning_rate": 3.55568347684365e-05, "loss": 1.3105, "step": 1832 }, { "epoch": 3.27, "learning_rate": 3.5493661074123605e-05, "loss": 2.1523, "step": 1833 }, { "epoch": 3.27, "learning_rate": 3.543051265553377e-05, "loss": 2.3984, "step": 1834 }, { "epoch": 3.28, "learning_rate": 3.536738962269628e-05, "loss": 2.0762, "step": 1835 }, { "epoch": 3.28, "learning_rate": 3.530429208559616e-05, "loss": 1.5156, "step": 1836 }, { "epoch": 3.28, "learning_rate": 3.524122015417405e-05, "loss": 1.5195, "step": 1837 }, { "epoch": 3.28, "learning_rate": 3.517817393832593e-05, "loss": 2.0352, "step": 1838 }, { "epoch": 3.28, "learning_rate": 3.511515354790301e-05, "loss": 1.7842, "step": 1839 }, { "epoch": 3.29, "learning_rate": 3.505215909271149e-05, "loss": 2.0176, "step": 1840 }, { "epoch": 3.29, "learning_rate": 3.498919068251237e-05, "loss": 2.0996, "step": 1841 }, { "epoch": 3.29, "learning_rate": 3.492624842702129e-05, "loss": 2.3828, "step": 1842 }, { "epoch": 3.29, "learning_rate": 3.4863332435908304e-05, "loss": 2.2031, "step": 1843 }, { "epoch": 3.29, "learning_rate": 3.48004428187977e-05, "loss": 1.6211, "step": 1844 }, { "epoch": 3.29, "learning_rate": 3.4737579685267815e-05, "loss": 1.5918, "step": 1845 }, { "epoch": 3.3, "learning_rate": 3.467474314485086e-05, "loss": 1.9883, "step": 1846 }, { "epoch": 3.3, "learning_rate": 3.4611933307032696e-05, "loss": 2.0625, "step": 1847 }, { "epoch": 3.3, "learning_rate": 3.4549150281252636e-05, "loss": 1.6396, "step": 1848 }, { "epoch": 3.3, "learning_rate": 3.448639417690331e-05, "loss": 2.377, "step": 1849 }, { "epoch": 3.3, "learning_rate": 3.442366510333043e-05, "loss": 1.9941, "step": 1850 }, { "epoch": 3.31, "learning_rate": 3.4360963169832604e-05, "loss": 2.2305, "step": 1851 }, { "epoch": 3.31, "learning_rate": 3.4298288485661143e-05, "loss": 2.0859, "step": 1852 }, { "epoch": 3.31, "learning_rate": 3.4235641160019906e-05, "loss": 1.5488, "step": 1853 }, { "epoch": 3.31, "learning_rate": 3.4173021302065065e-05, "loss": 1.8721, "step": 1854 }, { "epoch": 3.31, "learning_rate": 3.411042902090492e-05, "loss": 1.9453, "step": 1855 }, { "epoch": 3.31, "learning_rate": 3.404786442559976e-05, "loss": 1.9727, "step": 1856 }, { "epoch": 3.32, "learning_rate": 3.3985327625161586e-05, "loss": 1.707, "step": 1857 }, { "epoch": 3.32, "learning_rate": 3.3922818728554006e-05, "loss": 2.2578, "step": 1858 }, { "epoch": 3.32, "learning_rate": 3.386033784469199e-05, "loss": 1.2939, "step": 1859 }, { "epoch": 3.32, "learning_rate": 3.379788508244171e-05, "loss": 1.5137, "step": 1860 }, { "epoch": 3.32, "learning_rate": 3.373546055062034e-05, "loss": 1.4551, "step": 1861 }, { "epoch": 3.33, "learning_rate": 3.367306435799584e-05, "loss": 1.875, "step": 1862 }, { "epoch": 3.33, "learning_rate": 3.361069661328684e-05, "loss": 1.5938, "step": 1863 }, { "epoch": 3.33, "learning_rate": 3.354835742516235e-05, "loss": 2.207, "step": 1864 }, { "epoch": 3.33, "learning_rate": 3.3486046902241664e-05, "loss": 2.0703, "step": 1865 }, { "epoch": 3.33, "learning_rate": 3.342376515309409e-05, "loss": 1.9727, "step": 1866 }, { "epoch": 3.33, "learning_rate": 3.336151228623886e-05, "loss": 1.582, "step": 1867 }, { "epoch": 3.34, "learning_rate": 3.3299288410144814e-05, "loss": 2.5059, "step": 1868 }, { "epoch": 3.34, "learning_rate": 3.323709363323032e-05, "loss": 1.5576, "step": 1869 }, { "epoch": 3.34, "learning_rate": 3.3174928063863055e-05, "loss": 1.7139, "step": 1870 }, { "epoch": 3.34, "learning_rate": 3.311279181035974e-05, "loss": 2.0, "step": 1871 }, { "epoch": 3.34, "learning_rate": 3.30506849809861e-05, "loss": 1.7295, "step": 1872 }, { "epoch": 3.34, "learning_rate": 3.2988607683956564e-05, "loss": 1.8047, "step": 1873 }, { "epoch": 3.35, "learning_rate": 3.292656002743405e-05, "loss": 2.1426, "step": 1874 }, { "epoch": 3.35, "learning_rate": 3.286454211952989e-05, "loss": 1.3867, "step": 1875 }, { "epoch": 3.35, "learning_rate": 3.2802554068303596e-05, "loss": 1.3906, "step": 1876 }, { "epoch": 3.35, "learning_rate": 3.274059598176258e-05, "loss": 1.7871, "step": 1877 }, { "epoch": 3.35, "learning_rate": 3.267866796786212e-05, "loss": 1.4941, "step": 1878 }, { "epoch": 3.36, "learning_rate": 3.261677013450506e-05, "loss": 1.5039, "step": 1879 }, { "epoch": 3.36, "learning_rate": 3.255490258954167e-05, "loss": 1.5947, "step": 1880 }, { "epoch": 3.36, "learning_rate": 3.249306544076942e-05, "loss": 2.0469, "step": 1881 }, { "epoch": 3.36, "learning_rate": 3.243125879593286e-05, "loss": 1.5703, "step": 1882 }, { "epoch": 3.36, "learning_rate": 3.236948276272337e-05, "loss": 2.1348, "step": 1883 }, { "epoch": 3.36, "learning_rate": 3.230773744877898e-05, "loss": 1.8535, "step": 1884 }, { "epoch": 3.37, "learning_rate": 3.2246022961684206e-05, "loss": 2.1855, "step": 1885 }, { "epoch": 3.37, "learning_rate": 3.218433940896986e-05, "loss": 1.7197, "step": 1886 }, { "epoch": 3.37, "learning_rate": 3.212268689811284e-05, "loss": 1.918, "step": 1887 }, { "epoch": 3.37, "learning_rate": 3.206106553653596e-05, "loss": 2.1914, "step": 1888 }, { "epoch": 3.37, "learning_rate": 3.1999475431607775e-05, "loss": 2.6523, "step": 1889 }, { "epoch": 3.38, "learning_rate": 3.1937916690642356e-05, "loss": 2.0195, "step": 1890 }, { "epoch": 3.38, "learning_rate": 3.187638942089913e-05, "loss": 1.4482, "step": 1891 }, { "epoch": 3.38, "learning_rate": 3.1814893729582715e-05, "loss": 2.1211, "step": 1892 }, { "epoch": 3.38, "learning_rate": 3.175342972384267e-05, "loss": 1.9941, "step": 1893 }, { "epoch": 3.38, "learning_rate": 3.1691997510773375e-05, "loss": 1.6152, "step": 1894 }, { "epoch": 3.38, "learning_rate": 3.163059719741378e-05, "loss": 2.4707, "step": 1895 }, { "epoch": 3.39, "learning_rate": 3.156922889074731e-05, "loss": 2.2656, "step": 1896 }, { "epoch": 3.39, "learning_rate": 3.150789269770155e-05, "loss": 1.8008, "step": 1897 }, { "epoch": 3.39, "learning_rate": 3.1446588725148184e-05, "loss": 1.376, "step": 1898 }, { "epoch": 3.39, "learning_rate": 3.1385317079902745e-05, "loss": 2.0234, "step": 1899 }, { "epoch": 3.39, "learning_rate": 3.132407786872442e-05, "loss": 2.0195, "step": 1900 }, { "epoch": 3.39, "learning_rate": 3.1262871198315915e-05, "loss": 1.7969, "step": 1901 }, { "epoch": 3.4, "learning_rate": 3.120169717532319e-05, "loss": 2.1094, "step": 1902 }, { "epoch": 3.4, "learning_rate": 3.114055590633538e-05, "loss": 1.792, "step": 1903 }, { "epoch": 3.4, "learning_rate": 3.107944749788449e-05, "loss": 1.959, "step": 1904 }, { "epoch": 3.4, "learning_rate": 3.101837205644531e-05, "loss": 2.168, "step": 1905 }, { "epoch": 3.4, "learning_rate": 3.095732968843519e-05, "loss": 1.8652, "step": 1906 }, { "epoch": 3.41, "learning_rate": 3.089632050021381e-05, "loss": 1.4883, "step": 1907 }, { "epoch": 3.41, "learning_rate": 3.083534459808309e-05, "loss": 2.7285, "step": 1908 }, { "epoch": 3.41, "learning_rate": 3.077440208828692e-05, "loss": 2.4453, "step": 1909 }, { "epoch": 3.41, "learning_rate": 3.071349307701103e-05, "loss": 1.3379, "step": 1910 }, { "epoch": 3.41, "learning_rate": 3.065261767038275e-05, "loss": 2.4219, "step": 1911 }, { "epoch": 3.41, "learning_rate": 3.059177597447087e-05, "loss": 1.9727, "step": 1912 }, { "epoch": 3.42, "learning_rate": 3.053096809528548e-05, "loss": 1.5586, "step": 1913 }, { "epoch": 3.42, "learning_rate": 3.0470194138777707e-05, "loss": 1.7871, "step": 1914 }, { "epoch": 3.42, "learning_rate": 3.0409454210839577e-05, "loss": 2.6641, "step": 1915 }, { "epoch": 3.42, "learning_rate": 3.0348748417303823e-05, "loss": 1.5869, "step": 1916 }, { "epoch": 3.42, "learning_rate": 3.0288076863943726e-05, "loss": 2.2148, "step": 1917 }, { "epoch": 3.42, "learning_rate": 3.0227439656472877e-05, "loss": 1.5117, "step": 1918 }, { "epoch": 3.43, "learning_rate": 3.0166836900545042e-05, "loss": 1.627, "step": 1919 }, { "epoch": 3.43, "learning_rate": 3.0106268701753965e-05, "loss": 2.4668, "step": 1920 }, { "epoch": 3.43, "learning_rate": 3.0045735165633144e-05, "loss": 1.9268, "step": 1921 }, { "epoch": 3.43, "learning_rate": 2.9985236397655725e-05, "loss": 1.6152, "step": 1922 }, { "epoch": 3.43, "learning_rate": 2.9924772503234233e-05, "loss": 2.2383, "step": 1923 }, { "epoch": 3.44, "learning_rate": 2.9864343587720477e-05, "loss": 2.2031, "step": 1924 }, { "epoch": 3.44, "learning_rate": 2.980394975640526e-05, "loss": 2.1855, "step": 1925 }, { "epoch": 3.44, "learning_rate": 2.974359111451831e-05, "loss": 1.9434, "step": 1926 }, { "epoch": 3.44, "learning_rate": 2.9683267767228007e-05, "loss": 1.6934, "step": 1927 }, { "epoch": 3.44, "learning_rate": 2.962297981964124e-05, "loss": 1.7383, "step": 1928 }, { "epoch": 3.44, "learning_rate": 2.9562727376803235e-05, "loss": 2.959, "step": 1929 }, { "epoch": 3.45, "learning_rate": 2.9502510543697325e-05, "loss": 2.2852, "step": 1930 }, { "epoch": 3.45, "learning_rate": 2.944232942524484e-05, "loss": 1.9336, "step": 1931 }, { "epoch": 3.45, "learning_rate": 2.9382184126304834e-05, "loss": 2.7148, "step": 1932 }, { "epoch": 3.45, "learning_rate": 2.932207475167398e-05, "loss": 1.7676, "step": 1933 }, { "epoch": 3.45, "learning_rate": 2.9262001406086336e-05, "loss": 2.0586, "step": 1934 }, { "epoch": 3.46, "learning_rate": 2.9201964194213204e-05, "loss": 2.0566, "step": 1935 }, { "epoch": 3.46, "learning_rate": 2.9141963220662917e-05, "loss": 2.8945, "step": 1936 }, { "epoch": 3.46, "learning_rate": 2.9081998589980676e-05, "loss": 1.7539, "step": 1937 }, { "epoch": 3.46, "learning_rate": 2.9022070406648337e-05, "loss": 1.9336, "step": 1938 }, { "epoch": 3.46, "learning_rate": 2.8962178775084263e-05, "loss": 1.79, "step": 1939 }, { "epoch": 3.46, "learning_rate": 2.8902323799643116e-05, "loss": 1.6787, "step": 1940 }, { "epoch": 3.47, "learning_rate": 2.884250558461574e-05, "loss": 1.5195, "step": 1941 }, { "epoch": 3.47, "learning_rate": 2.8782724234228876e-05, "loss": 1.5977, "step": 1942 }, { "epoch": 3.47, "learning_rate": 2.8722979852645003e-05, "loss": 1.5918, "step": 1943 }, { "epoch": 3.47, "learning_rate": 2.8663272543962306e-05, "loss": 1.7441, "step": 1944 }, { "epoch": 3.47, "learning_rate": 2.8603602412214226e-05, "loss": 1.7344, "step": 1945 }, { "epoch": 3.48, "learning_rate": 2.8543969561369556e-05, "loss": 1.4707, "step": 1946 }, { "epoch": 3.48, "learning_rate": 2.8484374095332044e-05, "loss": 1.6816, "step": 1947 }, { "epoch": 3.48, "learning_rate": 2.8424816117940322e-05, "loss": 1.8271, "step": 1948 }, { "epoch": 3.48, "learning_rate": 2.836529573296774e-05, "loss": 2.0723, "step": 1949 }, { "epoch": 3.48, "learning_rate": 2.8305813044122097e-05, "loss": 2.6484, "step": 1950 }, { "epoch": 3.48, "learning_rate": 2.8246368155045515e-05, "loss": 2.2461, "step": 1951 }, { "epoch": 3.49, "learning_rate": 2.8186961169314307e-05, "loss": 1.666, "step": 1952 }, { "epoch": 3.49, "learning_rate": 2.8127592190438688e-05, "loss": 1.5859, "step": 1953 }, { "epoch": 3.49, "learning_rate": 2.8068261321862664e-05, "loss": 2.1523, "step": 1954 }, { "epoch": 3.49, "learning_rate": 2.800896866696382e-05, "loss": 2.0664, "step": 1955 }, { "epoch": 3.49, "learning_rate": 2.7949714329053228e-05, "loss": 1.4473, "step": 1956 }, { "epoch": 3.49, "learning_rate": 2.789049841137512e-05, "loss": 1.9883, "step": 1957 }, { "epoch": 3.5, "learning_rate": 2.7831321017106803e-05, "loss": 2.8281, "step": 1958 }, { "epoch": 3.5, "learning_rate": 2.7772182249358503e-05, "loss": 1.1426, "step": 1959 }, { "epoch": 3.5, "learning_rate": 2.771308221117309e-05, "loss": 2.0332, "step": 1960 }, { "epoch": 3.5, "learning_rate": 2.7654021005525983e-05, "loss": 2.168, "step": 1961 }, { "epoch": 3.5, "learning_rate": 2.7594998735324908e-05, "loss": 1.9395, "step": 1962 }, { "epoch": 3.51, "learning_rate": 2.7536015503409807e-05, "loss": 1.8193, "step": 1963 }, { "epoch": 3.51, "learning_rate": 2.7477071412552553e-05, "loss": 1.3887, "step": 1964 }, { "epoch": 3.51, "learning_rate": 2.7418166565456815e-05, "loss": 1.4492, "step": 1965 }, { "epoch": 3.51, "learning_rate": 2.7359301064757942e-05, "loss": 1.6309, "step": 1966 }, { "epoch": 3.51, "learning_rate": 2.7300475013022663e-05, "loss": 1.9219, "step": 1967 }, { "epoch": 3.51, "learning_rate": 2.7241688512749007e-05, "loss": 1.8633, "step": 1968 }, { "epoch": 3.52, "learning_rate": 2.7182941666366053e-05, "loss": 1.4678, "step": 1969 }, { "epoch": 3.52, "learning_rate": 2.7124234576233847e-05, "loss": 2.1719, "step": 1970 }, { "epoch": 3.52, "learning_rate": 2.7065567344643117e-05, "loss": 1.998, "step": 1971 }, { "epoch": 3.52, "learning_rate": 2.7006940073815133e-05, "loss": 1.4414, "step": 1972 }, { "epoch": 3.52, "learning_rate": 2.6948352865901595e-05, "loss": 1.5977, "step": 1973 }, { "epoch": 3.52, "learning_rate": 2.688980582298435e-05, "loss": 2.543, "step": 1974 }, { "epoch": 3.53, "learning_rate": 2.6831299047075265e-05, "loss": 1.832, "step": 1975 }, { "epoch": 3.53, "learning_rate": 2.6772832640116035e-05, "loss": 1.791, "step": 1976 }, { "epoch": 3.53, "learning_rate": 2.6714406703978077e-05, "loss": 2.082, "step": 1977 }, { "epoch": 3.53, "learning_rate": 2.6656021340462245e-05, "loss": 1.8623, "step": 1978 }, { "epoch": 3.53, "learning_rate": 2.659767665129867e-05, "loss": 1.7148, "step": 1979 }, { "epoch": 3.54, "learning_rate": 2.6539372738146695e-05, "loss": 2.6523, "step": 1980 }, { "epoch": 3.54, "learning_rate": 2.6481109702594542e-05, "loss": 1.791, "step": 1981 }, { "epoch": 3.54, "learning_rate": 2.6422887646159235e-05, "loss": 1.8008, "step": 1982 }, { "epoch": 3.54, "learning_rate": 2.636470667028639e-05, "loss": 1.3516, "step": 1983 }, { "epoch": 3.54, "learning_rate": 2.630656687635007e-05, "loss": 2.166, "step": 1984 }, { "epoch": 3.54, "learning_rate": 2.6248468365652557e-05, "loss": 1.5996, "step": 1985 }, { "epoch": 3.55, "learning_rate": 2.619041123942419e-05, "loss": 2.3691, "step": 1986 }, { "epoch": 3.55, "learning_rate": 2.6132395598823244e-05, "loss": 1.2471, "step": 1987 }, { "epoch": 3.55, "learning_rate": 2.607442154493568e-05, "loss": 1.3975, "step": 1988 }, { "epoch": 3.55, "learning_rate": 2.6016489178774995e-05, "loss": 2.1133, "step": 1989 }, { "epoch": 3.55, "learning_rate": 2.5958598601282035e-05, "loss": 1.9473, "step": 1990 }, { "epoch": 3.56, "learning_rate": 2.5900749913324895e-05, "loss": 2.1484, "step": 1991 }, { "epoch": 3.56, "learning_rate": 2.5842943215698624e-05, "loss": 2.0332, "step": 1992 }, { "epoch": 3.56, "learning_rate": 2.5785178609125114e-05, "loss": 2.4492, "step": 1993 }, { "epoch": 3.56, "learning_rate": 2.572745619425296e-05, "loss": 1.5605, "step": 1994 }, { "epoch": 3.56, "learning_rate": 2.5669776071657192e-05, "loss": 1.1045, "step": 1995 }, { "epoch": 3.56, "learning_rate": 2.5612138341839186e-05, "loss": 2.6094, "step": 1996 }, { "epoch": 3.57, "learning_rate": 2.5554543105226403e-05, "loss": 2.1367, "step": 1997 }, { "epoch": 3.57, "learning_rate": 2.5496990462172343e-05, "loss": 2.084, "step": 1998 }, { "epoch": 3.57, "learning_rate": 2.5439480512956225e-05, "loss": 1.7803, "step": 1999 }, { "epoch": 3.57, "learning_rate": 2.5382013357782893e-05, "loss": 1.8223, "step": 2000 }, { "epoch": 3.57, "learning_rate": 2.5324589096782657e-05, "loss": 2.0488, "step": 2001 }, { "epoch": 3.58, "learning_rate": 2.5267207830011068e-05, "loss": 1.6084, "step": 2002 }, { "epoch": 3.58, "learning_rate": 2.5209869657448748e-05, "loss": 1.5273, "step": 2003 }, { "epoch": 3.58, "learning_rate": 2.5152574679001234e-05, "loss": 1.7754, "step": 2004 }, { "epoch": 3.58, "learning_rate": 2.509532299449885e-05, "loss": 1.6094, "step": 2005 }, { "epoch": 3.58, "learning_rate": 2.5038114703696434e-05, "loss": 1.6826, "step": 2006 }, { "epoch": 3.58, "learning_rate": 2.4980949906273226e-05, "loss": 1.6094, "step": 2007 }, { "epoch": 3.59, "learning_rate": 2.492382870183268e-05, "loss": 2.2773, "step": 2008 }, { "epoch": 3.59, "learning_rate": 2.4866751189902334e-05, "loss": 2.0273, "step": 2009 }, { "epoch": 3.59, "learning_rate": 2.480971746993354e-05, "loss": 1.9023, "step": 2010 }, { "epoch": 3.59, "learning_rate": 2.4752727641301365e-05, "loss": 1.5938, "step": 2011 }, { "epoch": 3.59, "learning_rate": 2.469578180330444e-05, "loss": 1.5596, "step": 2012 }, { "epoch": 3.59, "learning_rate": 2.4638880055164692e-05, "loss": 2.5332, "step": 2013 }, { "epoch": 3.6, "learning_rate": 2.4582022496027257e-05, "loss": 1.7656, "step": 2014 }, { "epoch": 3.6, "learning_rate": 2.452520922496025e-05, "loss": 2.2012, "step": 2015 }, { "epoch": 3.6, "learning_rate": 2.446844034095466e-05, "loss": 2.1738, "step": 2016 }, { "epoch": 3.6, "learning_rate": 2.4411715942924117e-05, "loss": 2.2949, "step": 2017 }, { "epoch": 3.6, "learning_rate": 2.4355036129704693e-05, "loss": 2.1211, "step": 2018 }, { "epoch": 3.61, "learning_rate": 2.4298401000054878e-05, "loss": 2.0312, "step": 2019 }, { "epoch": 3.61, "learning_rate": 2.4241810652655196e-05, "loss": 2.3086, "step": 2020 }, { "epoch": 3.61, "learning_rate": 2.418526518610823e-05, "loss": 1.8848, "step": 2021 }, { "epoch": 3.61, "learning_rate": 2.4128764698938293e-05, "loss": 2.2188, "step": 2022 }, { "epoch": 3.61, "learning_rate": 2.4072309289591394e-05, "loss": 2.2031, "step": 2023 }, { "epoch": 3.61, "learning_rate": 2.4015899056434945e-05, "loss": 2.0371, "step": 2024 }, { "epoch": 3.62, "learning_rate": 2.3959534097757664e-05, "loss": 1.7979, "step": 2025 }, { "epoch": 3.62, "learning_rate": 2.390321451176936e-05, "loss": 2.123, "step": 2026 }, { "epoch": 3.62, "learning_rate": 2.3846940396600846e-05, "loss": 1.377, "step": 2027 }, { "epoch": 3.62, "learning_rate": 2.379071185030365e-05, "loss": 2.5547, "step": 2028 }, { "epoch": 3.62, "learning_rate": 2.37345289708499e-05, "loss": 2.5, "step": 2029 }, { "epoch": 3.62, "learning_rate": 2.3678391856132204e-05, "loss": 2.0977, "step": 2030 }, { "epoch": 3.63, "learning_rate": 2.3622300603963388e-05, "loss": 1.6777, "step": 2031 }, { "epoch": 3.63, "learning_rate": 2.356625531207638e-05, "loss": 1.957, "step": 2032 }, { "epoch": 3.63, "learning_rate": 2.3510256078123994e-05, "loss": 1.75, "step": 2033 }, { "epoch": 3.63, "learning_rate": 2.3454302999678878e-05, "loss": 1.7393, "step": 2034 }, { "epoch": 3.63, "learning_rate": 2.3398396174233178e-05, "loss": 1.5791, "step": 2035 }, { "epoch": 3.64, "learning_rate": 2.334253569919846e-05, "loss": 1.5137, "step": 2036 }, { "epoch": 3.64, "learning_rate": 2.328672167190558e-05, "loss": 1.6484, "step": 2037 }, { "epoch": 3.64, "learning_rate": 2.323095418960442e-05, "loss": 2.1855, "step": 2038 }, { "epoch": 3.64, "learning_rate": 2.3175233349463766e-05, "loss": 2.2422, "step": 2039 }, { "epoch": 3.64, "learning_rate": 2.3119559248571128e-05, "loss": 2.4473, "step": 2040 }, { "epoch": 3.64, "learning_rate": 2.3063931983932624e-05, "loss": 1.918, "step": 2041 }, { "epoch": 3.65, "learning_rate": 2.3008351652472716e-05, "loss": 1.8145, "step": 2042 }, { "epoch": 3.65, "learning_rate": 2.2952818351034107e-05, "loss": 1.8086, "step": 2043 }, { "epoch": 3.65, "learning_rate": 2.2897332176377528e-05, "loss": 2.293, "step": 2044 }, { "epoch": 3.65, "learning_rate": 2.2841893225181676e-05, "loss": 2.6367, "step": 2045 }, { "epoch": 3.65, "learning_rate": 2.2786501594042887e-05, "loss": 1.624, "step": 2046 }, { "epoch": 3.66, "learning_rate": 2.2731157379475056e-05, "loss": 2.0566, "step": 2047 }, { "epoch": 3.66, "learning_rate": 2.2675860677909516e-05, "loss": 2.2188, "step": 2048 }, { "epoch": 3.66, "learning_rate": 2.262061158569475e-05, "loss": 1.3672, "step": 2049 }, { "epoch": 3.66, "learning_rate": 2.256541019909632e-05, "loss": 1.834, "step": 2050 }, { "epoch": 3.66, "learning_rate": 2.251025661429664e-05, "loss": 2.3125, "step": 2051 }, { "epoch": 3.66, "learning_rate": 2.245515092739488e-05, "loss": 1.4883, "step": 2052 }, { "epoch": 3.67, "learning_rate": 2.2400093234406715e-05, "loss": 1.625, "step": 2053 }, { "epoch": 3.67, "learning_rate": 2.234508363126419e-05, "loss": 1.6836, "step": 2054 }, { "epoch": 3.67, "learning_rate": 2.2290122213815607e-05, "loss": 1.9688, "step": 2055 }, { "epoch": 3.67, "learning_rate": 2.2235209077825263e-05, "loss": 2.5449, "step": 2056 }, { "epoch": 3.67, "learning_rate": 2.218034431897334e-05, "loss": 2.4258, "step": 2057 }, { "epoch": 3.67, "learning_rate": 2.2125528032855724e-05, "loss": 1.3535, "step": 2058 }, { "epoch": 3.68, "learning_rate": 2.2070760314983878e-05, "loss": 2.3711, "step": 2059 }, { "epoch": 3.68, "learning_rate": 2.2016041260784605e-05, "loss": 1.75, "step": 2060 }, { "epoch": 3.68, "learning_rate": 2.196137096559989e-05, "loss": 1.9688, "step": 2061 }, { "epoch": 3.68, "learning_rate": 2.1906749524686853e-05, "loss": 1.5352, "step": 2062 }, { "epoch": 3.68, "learning_rate": 2.18521770332174e-05, "loss": 1.541, "step": 2063 }, { "epoch": 3.69, "learning_rate": 2.1797653586278192e-05, "loss": 1.6846, "step": 2064 }, { "epoch": 3.69, "learning_rate": 2.1743179278870407e-05, "loss": 2.1777, "step": 2065 }, { "epoch": 3.69, "learning_rate": 2.168875420590965e-05, "loss": 1.8916, "step": 2066 }, { "epoch": 3.69, "learning_rate": 2.163437846222569e-05, "loss": 1.6895, "step": 2067 }, { "epoch": 3.69, "learning_rate": 2.1580052142562356e-05, "loss": 1.7793, "step": 2068 }, { "epoch": 3.69, "learning_rate": 2.1525775341577405e-05, "loss": 1.8164, "step": 2069 }, { "epoch": 3.7, "learning_rate": 2.1471548153842258e-05, "loss": 1.5488, "step": 2070 }, { "epoch": 3.7, "learning_rate": 2.1417370673841913e-05, "loss": 2.9492, "step": 2071 }, { "epoch": 3.7, "learning_rate": 2.136324299597474e-05, "loss": 1.709, "step": 2072 }, { "epoch": 3.7, "learning_rate": 2.1309165214552385e-05, "loss": 2.207, "step": 2073 }, { "epoch": 3.7, "learning_rate": 2.1255137423799505e-05, "loss": 2.0312, "step": 2074 }, { "epoch": 3.71, "learning_rate": 2.1201159717853637e-05, "loss": 1.6045, "step": 2075 }, { "epoch": 3.71, "learning_rate": 2.1147232190765136e-05, "loss": 1.3125, "step": 2076 }, { "epoch": 3.71, "learning_rate": 2.1093354936496846e-05, "loss": 1.4316, "step": 2077 }, { "epoch": 3.71, "learning_rate": 2.103952804892404e-05, "loss": 1.9375, "step": 2078 }, { "epoch": 3.71, "learning_rate": 2.098575162183422e-05, "loss": 2.1992, "step": 2079 }, { "epoch": 3.71, "learning_rate": 2.0932025748927013e-05, "loss": 1.6104, "step": 2080 }, { "epoch": 3.72, "learning_rate": 2.0878350523813916e-05, "loss": 2.2871, "step": 2081 }, { "epoch": 3.72, "learning_rate": 2.082472604001817e-05, "loss": 1.5928, "step": 2082 }, { "epoch": 3.72, "learning_rate": 2.0771152390974664e-05, "loss": 1.6084, "step": 2083 }, { "epoch": 3.72, "learning_rate": 2.0717629670029654e-05, "loss": 2.0391, "step": 2084 }, { "epoch": 3.72, "learning_rate": 2.0664157970440674e-05, "loss": 2.875, "step": 2085 }, { "epoch": 3.73, "learning_rate": 2.061073738537635e-05, "loss": 2.2266, "step": 2086 }, { "epoch": 3.73, "learning_rate": 2.0557368007916295e-05, "loss": 2.334, "step": 2087 }, { "epoch": 3.73, "learning_rate": 2.050404993105085e-05, "loss": 1.7207, "step": 2088 }, { "epoch": 3.73, "learning_rate": 2.0450783247680966e-05, "loss": 1.4512, "step": 2089 }, { "epoch": 3.73, "learning_rate": 2.0397568050618094e-05, "loss": 2.2344, "step": 2090 }, { "epoch": 3.73, "learning_rate": 2.0344404432583936e-05, "loss": 1.7197, "step": 2091 }, { "epoch": 3.74, "learning_rate": 2.0291292486210327e-05, "loss": 1.459, "step": 2092 }, { "epoch": 3.74, "learning_rate": 2.0238232304039068e-05, "loss": 2.4551, "step": 2093 }, { "epoch": 3.74, "learning_rate": 2.018522397852178e-05, "loss": 2.2031, "step": 2094 }, { "epoch": 3.74, "learning_rate": 2.0132267602019743e-05, "loss": 1.9502, "step": 2095 }, { "epoch": 3.74, "learning_rate": 2.0079363266803698e-05, "loss": 1.4414, "step": 2096 }, { "epoch": 3.74, "learning_rate": 2.002651106505368e-05, "loss": 1.7051, "step": 2097 }, { "epoch": 3.75, "learning_rate": 1.9973711088858972e-05, "loss": 2.123, "step": 2098 }, { "epoch": 3.75, "learning_rate": 1.992096343021778e-05, "loss": 2.0977, "step": 2099 }, { "epoch": 3.75, "learning_rate": 1.9868268181037185e-05, "loss": 1.7656, "step": 2100 }, { "epoch": 3.75, "learning_rate": 1.981562543313293e-05, "loss": 1.9902, "step": 2101 }, { "epoch": 3.75, "learning_rate": 1.976303527822933e-05, "loss": 1.8877, "step": 2102 }, { "epoch": 3.76, "learning_rate": 1.971049780795901e-05, "loss": 2.1523, "step": 2103 }, { "epoch": 3.76, "learning_rate": 1.9658013113862807e-05, "loss": 1.9336, "step": 2104 }, { "epoch": 3.76, "learning_rate": 1.9605581287389635e-05, "loss": 1.7471, "step": 2105 }, { "epoch": 3.76, "learning_rate": 1.9553202419896255e-05, "loss": 2.3672, "step": 2106 }, { "epoch": 3.76, "learning_rate": 1.9500876602647167e-05, "loss": 2.3965, "step": 2107 }, { "epoch": 3.76, "learning_rate": 1.944860392681442e-05, "loss": 1.79, "step": 2108 }, { "epoch": 3.77, "learning_rate": 1.9396384483477515e-05, "loss": 2.0781, "step": 2109 }, { "epoch": 3.77, "learning_rate": 1.934421836362315e-05, "loss": 2.3477, "step": 2110 }, { "epoch": 3.77, "learning_rate": 1.9292105658145133e-05, "loss": 2.6113, "step": 2111 }, { "epoch": 3.77, "learning_rate": 1.9240046457844223e-05, "loss": 1.9307, "step": 2112 }, { "epoch": 3.77, "learning_rate": 1.9188040853427923e-05, "loss": 2.043, "step": 2113 }, { "epoch": 3.77, "learning_rate": 1.9136088935510362e-05, "loss": 1.7812, "step": 2114 }, { "epoch": 3.78, "learning_rate": 1.90841907946121e-05, "loss": 2.125, "step": 2115 }, { "epoch": 3.78, "learning_rate": 1.9032346521160067e-05, "loss": 2.1133, "step": 2116 }, { "epoch": 3.78, "learning_rate": 1.8980556205487264e-05, "loss": 1.4727, "step": 2117 }, { "epoch": 3.78, "learning_rate": 1.8928819937832686e-05, "loss": 1.748, "step": 2118 }, { "epoch": 3.78, "learning_rate": 1.887713780834121e-05, "loss": 2.4355, "step": 2119 }, { "epoch": 3.79, "learning_rate": 1.8825509907063327e-05, "loss": 1.5801, "step": 2120 }, { "epoch": 3.79, "learning_rate": 1.8773936323955054e-05, "loss": 1.9707, "step": 2121 }, { "epoch": 3.79, "learning_rate": 1.8722417148877752e-05, "loss": 1.5215, "step": 2122 }, { "epoch": 3.79, "learning_rate": 1.8670952471598035e-05, "loss": 2.2441, "step": 2123 }, { "epoch": 3.79, "learning_rate": 1.861954238178751e-05, "loss": 1.5898, "step": 2124 }, { "epoch": 3.79, "learning_rate": 1.8568186969022654e-05, "loss": 1.498, "step": 2125 }, { "epoch": 3.8, "learning_rate": 1.851688632278476e-05, "loss": 2.5801, "step": 2126 }, { "epoch": 3.8, "learning_rate": 1.8465640532459623e-05, "loss": 2.4434, "step": 2127 }, { "epoch": 3.8, "learning_rate": 1.8414449687337464e-05, "loss": 1.8867, "step": 2128 }, { "epoch": 3.8, "learning_rate": 1.8363313876612774e-05, "loss": 2.1777, "step": 2129 }, { "epoch": 3.8, "learning_rate": 1.831223318938419e-05, "loss": 1.6992, "step": 2130 }, { "epoch": 3.81, "learning_rate": 1.8261207714654267e-05, "loss": 1.999, "step": 2131 }, { "epoch": 3.81, "learning_rate": 1.821023754132933e-05, "loss": 1.583, "step": 2132 }, { "epoch": 3.81, "learning_rate": 1.8159322758219426e-05, "loss": 1.5039, "step": 2133 }, { "epoch": 3.81, "learning_rate": 1.8108463454038022e-05, "loss": 2.1953, "step": 2134 }, { "epoch": 3.81, "learning_rate": 1.8057659717401947e-05, "loss": 2.3945, "step": 2135 }, { "epoch": 3.81, "learning_rate": 1.800691163683118e-05, "loss": 1.7754, "step": 2136 }, { "epoch": 3.82, "learning_rate": 1.7956219300748793e-05, "loss": 1.9844, "step": 2137 }, { "epoch": 3.82, "learning_rate": 1.790558279748067e-05, "loss": 1.6279, "step": 2138 }, { "epoch": 3.82, "learning_rate": 1.7855002215255422e-05, "loss": 2.4688, "step": 2139 }, { "epoch": 3.82, "learning_rate": 1.7804477642204222e-05, "loss": 1.4082, "step": 2140 }, { "epoch": 3.82, "learning_rate": 1.7754009166360687e-05, "loss": 2.0234, "step": 2141 }, { "epoch": 3.83, "learning_rate": 1.7703596875660645e-05, "loss": 1.8906, "step": 2142 }, { "epoch": 3.83, "learning_rate": 1.7653240857942042e-05, "loss": 1.8086, "step": 2143 }, { "epoch": 3.83, "learning_rate": 1.760294120094481e-05, "loss": 2.3711, "step": 2144 }, { "epoch": 3.83, "learning_rate": 1.7552697992310635e-05, "loss": 1.5098, "step": 2145 }, { "epoch": 3.83, "learning_rate": 1.7502511319582855e-05, "loss": 2.1719, "step": 2146 }, { "epoch": 3.83, "learning_rate": 1.7452381270206286e-05, "loss": 1.9805, "step": 2147 }, { "epoch": 3.84, "learning_rate": 1.7402307931527157e-05, "loss": 1.9561, "step": 2148 }, { "epoch": 3.84, "learning_rate": 1.7352291390792798e-05, "loss": 1.9121, "step": 2149 }, { "epoch": 3.84, "learning_rate": 1.7302331735151594e-05, "loss": 1.8477, "step": 2150 }, { "epoch": 3.84, "learning_rate": 1.7252429051652872e-05, "loss": 2.3105, "step": 2151 }, { "epoch": 3.84, "learning_rate": 1.720258342724663e-05, "loss": 1.8516, "step": 2152 }, { "epoch": 3.84, "learning_rate": 1.7152794948783457e-05, "loss": 2.0645, "step": 2153 }, { "epoch": 3.85, "learning_rate": 1.7103063703014372e-05, "loss": 1.5957, "step": 2154 }, { "epoch": 3.85, "learning_rate": 1.705338977659071e-05, "loss": 2.6836, "step": 2155 }, { "epoch": 3.85, "learning_rate": 1.700377325606388e-05, "loss": 2.1621, "step": 2156 }, { "epoch": 3.85, "learning_rate": 1.695421422788529e-05, "loss": 1.7119, "step": 2157 }, { "epoch": 3.85, "learning_rate": 1.690471277840619e-05, "loss": 2.0293, "step": 2158 }, { "epoch": 3.86, "learning_rate": 1.6855268993877476e-05, "loss": 2.0078, "step": 2159 }, { "epoch": 3.86, "learning_rate": 1.6805882960449594e-05, "loss": 1.7109, "step": 2160 }, { "epoch": 3.86, "learning_rate": 1.6756554764172328e-05, "loss": 1.8359, "step": 2161 }, { "epoch": 3.86, "learning_rate": 1.6707284490994745e-05, "loss": 2.1895, "step": 2162 }, { "epoch": 3.86, "learning_rate": 1.665807222676495e-05, "loss": 1.8828, "step": 2163 }, { "epoch": 3.86, "learning_rate": 1.6608918057229942e-05, "loss": 1.5977, "step": 2164 }, { "epoch": 3.87, "learning_rate": 1.6559822068035584e-05, "loss": 1.875, "step": 2165 }, { "epoch": 3.87, "learning_rate": 1.6510784344726293e-05, "loss": 2.084, "step": 2166 }, { "epoch": 3.87, "learning_rate": 1.6461804972744983e-05, "loss": 2.0977, "step": 2167 }, { "epoch": 3.87, "learning_rate": 1.6412884037432872e-05, "loss": 1.7529, "step": 2168 }, { "epoch": 3.87, "learning_rate": 1.6364021624029445e-05, "loss": 1.8027, "step": 2169 }, { "epoch": 3.88, "learning_rate": 1.631521781767214e-05, "loss": 1.5576, "step": 2170 }, { "epoch": 3.88, "learning_rate": 1.6266472703396286e-05, "loss": 1.6152, "step": 2171 }, { "epoch": 3.88, "learning_rate": 1.6217786366134953e-05, "loss": 2.3594, "step": 2172 }, { "epoch": 3.88, "learning_rate": 1.6169158890718834e-05, "loss": 1.4238, "step": 2173 }, { "epoch": 3.88, "learning_rate": 1.612059036187602e-05, "loss": 2.0977, "step": 2174 }, { "epoch": 3.88, "learning_rate": 1.6072080864231904e-05, "loss": 1.334, "step": 2175 }, { "epoch": 3.89, "learning_rate": 1.6023630482309016e-05, "loss": 2.0391, "step": 2176 }, { "epoch": 3.89, "learning_rate": 1.5975239300526923e-05, "loss": 1.3984, "step": 2177 }, { "epoch": 3.89, "learning_rate": 1.5926907403202002e-05, "loss": 1.416, "step": 2178 }, { "epoch": 3.89, "learning_rate": 1.5878634874547322e-05, "loss": 2.625, "step": 2179 }, { "epoch": 3.89, "learning_rate": 1.5830421798672568e-05, "loss": 1.916, "step": 2180 }, { "epoch": 3.89, "learning_rate": 1.5782268259583783e-05, "loss": 1.3281, "step": 2181 }, { "epoch": 3.9, "learning_rate": 1.5734174341183284e-05, "loss": 1.9639, "step": 2182 }, { "epoch": 3.9, "learning_rate": 1.5686140127269494e-05, "loss": 1.2227, "step": 2183 }, { "epoch": 3.9, "learning_rate": 1.5638165701536868e-05, "loss": 3.0391, "step": 2184 }, { "epoch": 3.9, "learning_rate": 1.5590251147575608e-05, "loss": 1.7871, "step": 2185 }, { "epoch": 3.9, "learning_rate": 1.554239654887163e-05, "loss": 1.2441, "step": 2186 }, { "epoch": 3.91, "learning_rate": 1.5494601988806407e-05, "loss": 1.9307, "step": 2187 }, { "epoch": 3.91, "learning_rate": 1.544686755065677e-05, "loss": 2.0449, "step": 2188 }, { "epoch": 3.91, "learning_rate": 1.5399193317594794e-05, "loss": 1.8516, "step": 2189 }, { "epoch": 3.91, "learning_rate": 1.535157937268766e-05, "loss": 1.1719, "step": 2190 }, { "epoch": 3.91, "learning_rate": 1.530402579889752e-05, "loss": 1.791, "step": 2191 }, { "epoch": 3.91, "learning_rate": 1.525653267908132e-05, "loss": 2.4141, "step": 2192 }, { "epoch": 3.92, "learning_rate": 1.5209100095990647e-05, "loss": 1.6504, "step": 2193 }, { "epoch": 3.92, "learning_rate": 1.5161728132271675e-05, "loss": 1.9766, "step": 2194 }, { "epoch": 3.92, "learning_rate": 1.5114416870464898e-05, "loss": 2.0039, "step": 2195 }, { "epoch": 3.92, "learning_rate": 1.5067166393005055e-05, "loss": 2.3066, "step": 2196 }, { "epoch": 3.92, "learning_rate": 1.5019976782220967e-05, "loss": 2.0586, "step": 2197 }, { "epoch": 3.92, "learning_rate": 1.4972848120335453e-05, "loss": 2.3516, "step": 2198 }, { "epoch": 3.93, "learning_rate": 1.492578048946508e-05, "loss": 1.9297, "step": 2199 }, { "epoch": 3.93, "learning_rate": 1.4878773971620074e-05, "loss": 2.1465, "step": 2200 }, { "epoch": 3.93, "learning_rate": 1.4831828648704227e-05, "loss": 2.0449, "step": 2201 }, { "epoch": 3.93, "learning_rate": 1.4784944602514661e-05, "loss": 1.6572, "step": 2202 }, { "epoch": 3.93, "learning_rate": 1.4738121914741749e-05, "loss": 1.9902, "step": 2203 }, { "epoch": 3.94, "learning_rate": 1.4691360666968923e-05, "loss": 2.5625, "step": 2204 }, { "epoch": 3.94, "learning_rate": 1.4644660940672627e-05, "loss": 2.791, "step": 2205 }, { "epoch": 3.94, "learning_rate": 1.4598022817222056e-05, "loss": 1.6299, "step": 2206 }, { "epoch": 3.94, "learning_rate": 1.4551446377879069e-05, "loss": 2.2285, "step": 2207 }, { "epoch": 3.94, "learning_rate": 1.4504931703798086e-05, "loss": 2.6094, "step": 2208 }, { "epoch": 3.94, "learning_rate": 1.4458478876025877e-05, "loss": 1.3926, "step": 2209 }, { "epoch": 3.95, "learning_rate": 1.4412087975501458e-05, "loss": 2.3027, "step": 2210 }, { "epoch": 3.95, "learning_rate": 1.436575908305593e-05, "loss": 1.9463, "step": 2211 }, { "epoch": 3.95, "learning_rate": 1.4319492279412388e-05, "loss": 1.9688, "step": 2212 }, { "epoch": 3.95, "learning_rate": 1.4273287645185718e-05, "loss": 1.5781, "step": 2213 }, { "epoch": 3.95, "learning_rate": 1.4227145260882463e-05, "loss": 1.4961, "step": 2214 }, { "epoch": 3.96, "learning_rate": 1.4181065206900762e-05, "loss": 1.7812, "step": 2215 }, { "epoch": 3.96, "learning_rate": 1.4135047563530091e-05, "loss": 1.3994, "step": 2216 }, { "epoch": 3.96, "learning_rate": 1.4089092410951216e-05, "loss": 2.1875, "step": 2217 }, { "epoch": 3.96, "learning_rate": 1.404319982923598e-05, "loss": 1.5205, "step": 2218 }, { "epoch": 3.96, "learning_rate": 1.3997369898347278e-05, "loss": 1.8467, "step": 2219 }, { "epoch": 3.96, "learning_rate": 1.3951602698138771e-05, "loss": 1.459, "step": 2220 }, { "epoch": 3.97, "learning_rate": 1.3905898308354826e-05, "loss": 1.7734, "step": 2221 }, { "epoch": 3.97, "learning_rate": 1.3860256808630428e-05, "loss": 1.7715, "step": 2222 }, { "epoch": 3.97, "learning_rate": 1.3814678278490923e-05, "loss": 1.3457, "step": 2223 }, { "epoch": 3.97, "learning_rate": 1.3769162797351954e-05, "loss": 2.1113, "step": 2224 }, { "epoch": 3.97, "learning_rate": 1.3723710444519306e-05, "loss": 1.9033, "step": 2225 }, { "epoch": 3.98, "learning_rate": 1.3678321299188801e-05, "loss": 1.4844, "step": 2226 }, { "epoch": 3.98, "learning_rate": 1.3632995440446088e-05, "loss": 1.9854, "step": 2227 }, { "epoch": 3.98, "learning_rate": 1.3587732947266557e-05, "loss": 2.2441, "step": 2228 }, { "epoch": 3.98, "learning_rate": 1.354253389851522e-05, "loss": 1.4463, "step": 2229 }, { "epoch": 3.98, "learning_rate": 1.3497398372946501e-05, "loss": 1.3037, "step": 2230 }, { "epoch": 3.98, "learning_rate": 1.3452326449204167e-05, "loss": 2.3867, "step": 2231 }, { "epoch": 3.99, "learning_rate": 1.340731820582114e-05, "loss": 1.877, "step": 2232 }, { "epoch": 3.99, "learning_rate": 1.336237372121944e-05, "loss": 1.7988, "step": 2233 }, { "epoch": 3.99, "learning_rate": 1.3317493073709935e-05, "loss": 1.7305, "step": 2234 }, { "epoch": 3.99, "learning_rate": 1.3272676341492285e-05, "loss": 1.499, "step": 2235 }, { "epoch": 3.99, "learning_rate": 1.3227923602654806e-05, "loss": 1.5176, "step": 2236 }, { "epoch": 3.99, "learning_rate": 1.3183234935174288e-05, "loss": 2.084, "step": 2237 }, { "epoch": 4.0, "learning_rate": 1.3138610416915887e-05, "loss": 2.3574, "step": 2238 }, { "epoch": 4.0, "learning_rate": 1.3094050125632972e-05, "loss": 1.0996, "step": 2239 }, { "epoch": 4.0, "learning_rate": 1.3049554138967051e-05, "loss": 2.3965, "step": 2240 }, { "epoch": 4.0, "learning_rate": 1.3005122534447544e-05, "loss": 1.8926, "step": 2241 }, { "epoch": 4.0, "learning_rate": 1.2960755389491702e-05, "loss": 1.5234, "step": 2242 }, { "epoch": 4.01, "learning_rate": 1.2916452781404443e-05, "loss": 2.1074, "step": 2243 }, { "epoch": 4.01, "learning_rate": 1.2872214787378306e-05, "loss": 1.291, "step": 2244 }, { "epoch": 4.01, "learning_rate": 1.2828041484493187e-05, "loss": 1.6738, "step": 2245 }, { "epoch": 4.01, "learning_rate": 1.278393294971626e-05, "loss": 2.1875, "step": 2246 }, { "epoch": 4.01, "learning_rate": 1.2739889259901866e-05, "loss": 1.8047, "step": 2247 }, { "epoch": 4.01, "learning_rate": 1.269591049179138e-05, "loss": 2.2148, "step": 2248 }, { "epoch": 4.02, "learning_rate": 1.2651996722013021e-05, "loss": 2.002, "step": 2249 }, { "epoch": 4.02, "learning_rate": 1.2608148027081773e-05, "loss": 1.7988, "step": 2250 }, { "epoch": 4.02, "learning_rate": 1.2564364483399244e-05, "loss": 1.7715, "step": 2251 }, { "epoch": 4.02, "learning_rate": 1.2520646167253514e-05, "loss": 2.5117, "step": 2252 }, { "epoch": 4.02, "learning_rate": 1.2476993154819e-05, "loss": 2.1914, "step": 2253 }, { "epoch": 4.03, "learning_rate": 1.2433405522156332e-05, "loss": 1.6807, "step": 2254 }, { "epoch": 4.03, "learning_rate": 1.238988334521226e-05, "loss": 1.123, "step": 2255 }, { "epoch": 4.03, "learning_rate": 1.2346426699819458e-05, "loss": 1.4219, "step": 2256 }, { "epoch": 4.03, "learning_rate": 1.2303035661696393e-05, "loss": 1.4629, "step": 2257 }, { "epoch": 4.03, "learning_rate": 1.2259710306447275e-05, "loss": 2.0039, "step": 2258 }, { "epoch": 4.03, "learning_rate": 1.2216450709561828e-05, "loss": 1.4375, "step": 2259 }, { "epoch": 4.04, "learning_rate": 1.2173256946415212e-05, "loss": 1.9453, "step": 2260 }, { "epoch": 4.04, "learning_rate": 1.2130129092267862e-05, "loss": 1.791, "step": 2261 }, { "epoch": 4.04, "learning_rate": 1.2087067222265407e-05, "loss": 1.917, "step": 2262 }, { "epoch": 4.04, "learning_rate": 1.204407141143849e-05, "loss": 1.8613, "step": 2263 }, { "epoch": 4.04, "learning_rate": 1.2001141734702626e-05, "loss": 1.5068, "step": 2264 }, { "epoch": 4.04, "learning_rate": 1.1958278266858147e-05, "loss": 1.7305, "step": 2265 }, { "epoch": 4.05, "learning_rate": 1.1915481082589996e-05, "loss": 1.4727, "step": 2266 }, { "epoch": 4.05, "learning_rate": 1.1872750256467625e-05, "loss": 1.1152, "step": 2267 }, { "epoch": 4.05, "learning_rate": 1.183008586294485e-05, "loss": 1.8027, "step": 2268 }, { "epoch": 4.05, "learning_rate": 1.1787487976359779e-05, "loss": 2.6523, "step": 2269 }, { "epoch": 4.05, "learning_rate": 1.1744956670934597e-05, "loss": 1.9707, "step": 2270 }, { "epoch": 4.06, "learning_rate": 1.170249202077548e-05, "loss": 1.7383, "step": 2271 }, { "epoch": 4.06, "learning_rate": 1.166009409987251e-05, "loss": 1.5557, "step": 2272 }, { "epoch": 4.06, "learning_rate": 1.1617762982099446e-05, "loss": 1.6729, "step": 2273 }, { "epoch": 4.06, "learning_rate": 1.1575498741213681e-05, "loss": 1.7979, "step": 2274 }, { "epoch": 4.06, "learning_rate": 1.1533301450856054e-05, "loss": 1.3594, "step": 2275 }, { "epoch": 4.06, "learning_rate": 1.1491171184550798e-05, "loss": 2.0469, "step": 2276 }, { "epoch": 4.07, "learning_rate": 1.1449108015705318e-05, "loss": 1.9199, "step": 2277 }, { "epoch": 4.07, "learning_rate": 1.1407112017610133e-05, "loss": 1.3418, "step": 2278 }, { "epoch": 4.07, "learning_rate": 1.1365183263438694e-05, "loss": 1.9141, "step": 2279 }, { "epoch": 4.07, "learning_rate": 1.1323321826247346e-05, "loss": 1.9492, "step": 2280 }, { "epoch": 4.07, "learning_rate": 1.1281527778975093e-05, "loss": 1.7363, "step": 2281 }, { "epoch": 4.08, "learning_rate": 1.1239801194443506e-05, "loss": 2.1816, "step": 2282 }, { "epoch": 4.08, "learning_rate": 1.1198142145356677e-05, "loss": 1.1797, "step": 2283 }, { "epoch": 4.08, "learning_rate": 1.1156550704300961e-05, "loss": 1.7539, "step": 2284 }, { "epoch": 4.08, "learning_rate": 1.1115026943744934e-05, "loss": 1.7666, "step": 2285 }, { "epoch": 4.08, "learning_rate": 1.107357093603924e-05, "loss": 2.625, "step": 2286 }, { "epoch": 4.08, "learning_rate": 1.1032182753416503e-05, "loss": 1.8281, "step": 2287 }, { "epoch": 4.09, "learning_rate": 1.099086246799113e-05, "loss": 2.0, "step": 2288 }, { "epoch": 4.09, "learning_rate": 1.0949610151759232e-05, "loss": 1.8203, "step": 2289 }, { "epoch": 4.09, "learning_rate": 1.090842587659851e-05, "loss": 1.9043, "step": 2290 }, { "epoch": 4.09, "learning_rate": 1.0867309714268103e-05, "loss": 1.791, "step": 2291 }, { "epoch": 4.09, "learning_rate": 1.082626173640846e-05, "loss": 1.8438, "step": 2292 }, { "epoch": 4.09, "learning_rate": 1.0785282014541215e-05, "loss": 1.5547, "step": 2293 }, { "epoch": 4.1, "learning_rate": 1.0744370620069122e-05, "loss": 2.2031, "step": 2294 }, { "epoch": 4.1, "learning_rate": 1.0703527624275832e-05, "loss": 2.168, "step": 2295 }, { "epoch": 4.1, "learning_rate": 1.066275309832584e-05, "loss": 1.1875, "step": 2296 }, { "epoch": 4.1, "learning_rate": 1.0622047113264344e-05, "loss": 1.5742, "step": 2297 }, { "epoch": 4.1, "learning_rate": 1.0581409740017112e-05, "loss": 1.7725, "step": 2298 }, { "epoch": 4.11, "learning_rate": 1.0540841049390354e-05, "loss": 2.2891, "step": 2299 }, { "epoch": 4.11, "learning_rate": 1.0500341112070605e-05, "loss": 2.1543, "step": 2300 }, { "epoch": 4.11, "learning_rate": 1.0459909998624646e-05, "loss": 3.3008, "step": 2301 }, { "epoch": 4.11, "learning_rate": 1.0419547779499284e-05, "loss": 1.7637, "step": 2302 }, { "epoch": 4.11, "learning_rate": 1.0379254525021309e-05, "loss": 2.1133, "step": 2303 }, { "epoch": 4.11, "learning_rate": 1.0339030305397374e-05, "loss": 1.9238, "step": 2304 }, { "epoch": 4.12, "learning_rate": 1.0298875190713802e-05, "loss": 2.1309, "step": 2305 }, { "epoch": 4.12, "learning_rate": 1.0258789250936528e-05, "loss": 1.9707, "step": 2306 }, { "epoch": 4.12, "learning_rate": 1.0218772555910955e-05, "loss": 1.9209, "step": 2307 }, { "epoch": 4.12, "learning_rate": 1.0178825175361844e-05, "loss": 1.6973, "step": 2308 }, { "epoch": 4.12, "learning_rate": 1.0138947178893183e-05, "loss": 2.0508, "step": 2309 }, { "epoch": 4.12, "learning_rate": 1.0099138635988026e-05, "loss": 1.5098, "step": 2310 }, { "epoch": 4.13, "learning_rate": 1.005939961600848e-05, "loss": 2.2344, "step": 2311 }, { "epoch": 4.13, "learning_rate": 1.0019730188195464e-05, "loss": 2.2266, "step": 2312 }, { "epoch": 4.13, "learning_rate": 9.98013042166866e-06, "loss": 2.3477, "step": 2313 }, { "epoch": 4.13, "learning_rate": 9.940600385426347e-06, "loss": 1.5137, "step": 2314 }, { "epoch": 4.13, "learning_rate": 9.901140148345361e-06, "loss": 2.3184, "step": 2315 }, { "epoch": 4.14, "learning_rate": 9.861749779180873e-06, "loss": 1.9961, "step": 2316 }, { "epoch": 4.14, "learning_rate": 9.822429346566314e-06, "loss": 1.418, "step": 2317 }, { "epoch": 4.14, "learning_rate": 9.783178919013297e-06, "loss": 1.9883, "step": 2318 }, { "epoch": 4.14, "learning_rate": 9.74399856491144e-06, "loss": 2.582, "step": 2319 }, { "epoch": 4.14, "learning_rate": 9.704888352528258e-06, "loss": 1.7715, "step": 2320 }, { "epoch": 4.14, "learning_rate": 9.66584835000905e-06, "loss": 2.2969, "step": 2321 }, { "epoch": 4.15, "learning_rate": 9.626878625376784e-06, "loss": 2.0684, "step": 2322 }, { "epoch": 4.15, "learning_rate": 9.587979246531997e-06, "loss": 2.0566, "step": 2323 }, { "epoch": 4.15, "learning_rate": 9.549150281252633e-06, "loss": 1.584, "step": 2324 }, { "epoch": 4.15, "learning_rate": 9.51039179719394e-06, "loss": 2.1699, "step": 2325 }, { "epoch": 4.15, "learning_rate": 9.471703861888398e-06, "loss": 2.1602, "step": 2326 }, { "epoch": 4.16, "learning_rate": 9.433086542745534e-06, "loss": 1.6445, "step": 2327 }, { "epoch": 4.16, "learning_rate": 9.394539907051836e-06, "loss": 1.6943, "step": 2328 }, { "epoch": 4.16, "learning_rate": 9.356064021970629e-06, "loss": 1.9229, "step": 2329 }, { "epoch": 4.16, "learning_rate": 9.317658954541992e-06, "loss": 1.459, "step": 2330 }, { "epoch": 4.16, "learning_rate": 9.279324771682585e-06, "loss": 2.2324, "step": 2331 }, { "epoch": 4.16, "learning_rate": 9.241061540185547e-06, "loss": 2.1836, "step": 2332 }, { "epoch": 4.17, "learning_rate": 9.202869326720437e-06, "loss": 1.7578, "step": 2333 }, { "epoch": 4.17, "learning_rate": 9.164748197833039e-06, "loss": 1.4375, "step": 2334 }, { "epoch": 4.17, "learning_rate": 9.12669821994529e-06, "loss": 1.7822, "step": 2335 }, { "epoch": 4.17, "learning_rate": 9.088719459355132e-06, "loss": 2.584, "step": 2336 }, { "epoch": 4.17, "learning_rate": 9.050811982236468e-06, "loss": 1.7422, "step": 2337 }, { "epoch": 4.17, "learning_rate": 9.012975854638949e-06, "loss": 2.1328, "step": 2338 }, { "epoch": 4.18, "learning_rate": 8.975211142487922e-06, "loss": 1.9453, "step": 2339 }, { "epoch": 4.18, "learning_rate": 8.93751791158432e-06, "loss": 1.8281, "step": 2340 }, { "epoch": 4.18, "learning_rate": 8.899896227604509e-06, "loss": 1.4961, "step": 2341 }, { "epoch": 4.18, "learning_rate": 8.862346156100187e-06, "loss": 2.1094, "step": 2342 }, { "epoch": 4.18, "learning_rate": 8.824867762498279e-06, "loss": 1.9766, "step": 2343 }, { "epoch": 4.19, "learning_rate": 8.787461112100836e-06, "loss": 1.5664, "step": 2344 }, { "epoch": 4.19, "learning_rate": 8.75012627008489e-06, "loss": 1.6748, "step": 2345 }, { "epoch": 4.19, "learning_rate": 8.71286330150234e-06, "loss": 1.9766, "step": 2346 }, { "epoch": 4.19, "learning_rate": 8.675672271279888e-06, "loss": 1.4033, "step": 2347 }, { "epoch": 4.19, "learning_rate": 8.638553244218872e-06, "loss": 1.5527, "step": 2348 }, { "epoch": 4.19, "learning_rate": 8.601506284995164e-06, "loss": 2.1875, "step": 2349 }, { "epoch": 4.2, "learning_rate": 8.56453145815907e-06, "loss": 2.418, "step": 2350 }, { "epoch": 4.2, "learning_rate": 8.527628828135232e-06, "loss": 1.5781, "step": 2351 }, { "epoch": 4.2, "learning_rate": 8.490798459222476e-06, "loss": 2.0391, "step": 2352 }, { "epoch": 4.2, "learning_rate": 8.454040415593712e-06, "loss": 2.002, "step": 2353 }, { "epoch": 4.2, "learning_rate": 8.417354761295875e-06, "loss": 1.2969, "step": 2354 }, { "epoch": 4.21, "learning_rate": 8.380741560249727e-06, "loss": 1.7617, "step": 2355 }, { "epoch": 4.21, "learning_rate": 8.344200876249803e-06, "loss": 1.8105, "step": 2356 }, { "epoch": 4.21, "learning_rate": 8.307732772964272e-06, "loss": 2.3047, "step": 2357 }, { "epoch": 4.21, "learning_rate": 8.271337313934869e-06, "loss": 2.2285, "step": 2358 }, { "epoch": 4.21, "learning_rate": 8.235014562576732e-06, "loss": 1.917, "step": 2359 }, { "epoch": 4.21, "learning_rate": 8.198764582178303e-06, "loss": 2.4512, "step": 2360 }, { "epoch": 4.22, "learning_rate": 8.162587435901264e-06, "loss": 1.6836, "step": 2361 }, { "epoch": 4.22, "learning_rate": 8.12648318678036e-06, "loss": 1.5381, "step": 2362 }, { "epoch": 4.22, "learning_rate": 8.090451897723333e-06, "loss": 2.6172, "step": 2363 }, { "epoch": 4.22, "learning_rate": 8.054493631510784e-06, "loss": 2.6328, "step": 2364 }, { "epoch": 4.22, "learning_rate": 8.018608450796122e-06, "loss": 1.8271, "step": 2365 }, { "epoch": 4.22, "learning_rate": 7.982796418105371e-06, "loss": 2.0586, "step": 2366 }, { "epoch": 4.23, "learning_rate": 7.947057595837103e-06, "loss": 1.5332, "step": 2367 }, { "epoch": 4.23, "learning_rate": 7.911392046262366e-06, "loss": 1.4609, "step": 2368 }, { "epoch": 4.23, "learning_rate": 7.8757998315245e-06, "loss": 2.2891, "step": 2369 }, { "epoch": 4.23, "learning_rate": 7.840281013639078e-06, "loss": 1.9336, "step": 2370 }, { "epoch": 4.23, "learning_rate": 7.804835654493792e-06, "loss": 1.7666, "step": 2371 }, { "epoch": 4.24, "learning_rate": 7.769463815848343e-06, "loss": 2.0117, "step": 2372 }, { "epoch": 4.24, "learning_rate": 7.734165559334328e-06, "loss": 2.4062, "step": 2373 }, { "epoch": 4.24, "learning_rate": 7.698940946455125e-06, "loss": 2.2734, "step": 2374 }, { "epoch": 4.24, "learning_rate": 7.663790038585793e-06, "loss": 1.8047, "step": 2375 }, { "epoch": 4.24, "learning_rate": 7.628712896973006e-06, "loss": 1.2354, "step": 2376 }, { "epoch": 4.24, "learning_rate": 7.593709582734865e-06, "loss": 2.1465, "step": 2377 }, { "epoch": 4.25, "learning_rate": 7.5587801568608395e-06, "loss": 1.6934, "step": 2378 }, { "epoch": 4.25, "learning_rate": 7.523924680211691e-06, "loss": 1.5488, "step": 2379 }, { "epoch": 4.25, "learning_rate": 7.489143213519301e-06, "loss": 2.291, "step": 2380 }, { "epoch": 4.25, "learning_rate": 7.454435817386601e-06, "loss": 1.5781, "step": 2381 }, { "epoch": 4.25, "learning_rate": 7.419802552287453e-06, "loss": 2.1523, "step": 2382 }, { "epoch": 4.26, "learning_rate": 7.385243478566594e-06, "loss": 1.4297, "step": 2383 }, { "epoch": 4.26, "learning_rate": 7.350758656439455e-06, "loss": 1.9414, "step": 2384 }, { "epoch": 4.26, "learning_rate": 7.31634814599208e-06, "loss": 2.4824, "step": 2385 }, { "epoch": 4.26, "learning_rate": 7.282012007181083e-06, "loss": 1.2109, "step": 2386 }, { "epoch": 4.26, "learning_rate": 7.2477502998334505e-06, "loss": 2.1094, "step": 2387 }, { "epoch": 4.26, "learning_rate": 7.213563083646496e-06, "loss": 1.8086, "step": 2388 }, { "epoch": 4.27, "learning_rate": 7.179450418187722e-06, "loss": 2.0352, "step": 2389 }, { "epoch": 4.27, "learning_rate": 7.145412362894771e-06, "loss": 1.8301, "step": 2390 }, { "epoch": 4.27, "learning_rate": 7.111448977075252e-06, "loss": 1.168, "step": 2391 }, { "epoch": 4.27, "learning_rate": 7.077560319906695e-06, "loss": 2.1133, "step": 2392 }, { "epoch": 4.27, "learning_rate": 7.0437464504363945e-06, "loss": 1.8164, "step": 2393 }, { "epoch": 4.28, "learning_rate": 7.010007427581378e-06, "loss": 2.0938, "step": 2394 }, { "epoch": 4.28, "learning_rate": 6.976343310128214e-06, "loss": 2.4453, "step": 2395 }, { "epoch": 4.28, "learning_rate": 6.942754156732978e-06, "loss": 2.1387, "step": 2396 }, { "epoch": 4.28, "learning_rate": 6.909240025921143e-06, "loss": 2.0469, "step": 2397 }, { "epoch": 4.28, "learning_rate": 6.875800976087443e-06, "loss": 1.9277, "step": 2398 }, { "epoch": 4.28, "learning_rate": 6.842437065495794e-06, "loss": 1.8945, "step": 2399 }, { "epoch": 4.29, "learning_rate": 6.809148352279182e-06, "loss": 2.3945, "step": 2400 }, { "epoch": 4.29, "learning_rate": 6.775934894439606e-06, "loss": 2.0273, "step": 2401 }, { "epoch": 4.29, "learning_rate": 6.742796749847907e-06, "loss": 1.9668, "step": 2402 }, { "epoch": 4.29, "learning_rate": 6.709733976243687e-06, "loss": 1.2246, "step": 2403 }, { "epoch": 4.29, "learning_rate": 6.676746631235281e-06, "loss": 1.9609, "step": 2404 }, { "epoch": 4.29, "learning_rate": 6.643834772299545e-06, "loss": 1.6836, "step": 2405 }, { "epoch": 4.3, "learning_rate": 6.61099845678183e-06, "loss": 2.0742, "step": 2406 }, { "epoch": 4.3, "learning_rate": 6.578237741895838e-06, "loss": 1.7549, "step": 2407 }, { "epoch": 4.3, "learning_rate": 6.5455526847235825e-06, "loss": 1.418, "step": 2408 }, { "epoch": 4.3, "learning_rate": 6.5129433422152334e-06, "loss": 2.0391, "step": 2409 }, { "epoch": 4.3, "learning_rate": 6.480409771189028e-06, "loss": 1.5625, "step": 2410 }, { "epoch": 4.31, "learning_rate": 6.447952028331178e-06, "loss": 1.7012, "step": 2411 }, { "epoch": 4.31, "learning_rate": 6.4155701701958e-06, "loss": 1.9785, "step": 2412 }, { "epoch": 4.31, "learning_rate": 6.383264253204763e-06, "loss": 1.666, "step": 2413 }, { "epoch": 4.31, "learning_rate": 6.351034333647615e-06, "loss": 1.6934, "step": 2414 }, { "epoch": 4.31, "learning_rate": 6.318880467681526e-06, "loss": 1.7832, "step": 2415 }, { "epoch": 4.31, "learning_rate": 6.286802711331097e-06, "loss": 1.8262, "step": 2416 }, { "epoch": 4.32, "learning_rate": 6.254801120488346e-06, "loss": 2.5312, "step": 2417 }, { "epoch": 4.32, "learning_rate": 6.222875750912571e-06, "loss": 1.5127, "step": 2418 }, { "epoch": 4.32, "learning_rate": 6.191026658230287e-06, "loss": 1.9707, "step": 2419 }, { "epoch": 4.32, "learning_rate": 6.159253897935069e-06, "loss": 1.7793, "step": 2420 }, { "epoch": 4.32, "learning_rate": 6.127557525387506e-06, "loss": 1.8379, "step": 2421 }, { "epoch": 4.33, "learning_rate": 6.0959375958151045e-06, "loss": 1.1533, "step": 2422 }, { "epoch": 4.33, "learning_rate": 6.064394164312159e-06, "loss": 1.2559, "step": 2423 }, { "epoch": 4.33, "learning_rate": 6.032927285839673e-06, "loss": 1.4443, "step": 2424 }, { "epoch": 4.33, "learning_rate": 6.001537015225267e-06, "loss": 1.6621, "step": 2425 }, { "epoch": 4.33, "learning_rate": 5.9702234071631e-06, "loss": 1.7627, "step": 2426 }, { "epoch": 4.33, "learning_rate": 5.93898651621374e-06, "loss": 2.0391, "step": 2427 }, { "epoch": 4.34, "learning_rate": 5.907826396804061e-06, "loss": 2.0625, "step": 2428 }, { "epoch": 4.34, "learning_rate": 5.876743103227217e-06, "loss": 1.9043, "step": 2429 }, { "epoch": 4.34, "learning_rate": 5.845736689642472e-06, "loss": 1.4385, "step": 2430 }, { "epoch": 4.34, "learning_rate": 5.814807210075135e-06, "loss": 1.7129, "step": 2431 }, { "epoch": 4.34, "learning_rate": 5.783954718416468e-06, "loss": 2.4844, "step": 2432 }, { "epoch": 4.34, "learning_rate": 5.753179268423603e-06, "loss": 2.0195, "step": 2433 }, { "epoch": 4.35, "learning_rate": 5.7224809137194244e-06, "loss": 1.5781, "step": 2434 }, { "epoch": 4.35, "learning_rate": 5.691859707792474e-06, "loss": 1.6133, "step": 2435 }, { "epoch": 4.35, "learning_rate": 5.6613157039969055e-06, "loss": 2.7812, "step": 2436 }, { "epoch": 4.35, "learning_rate": 5.630848955552326e-06, "loss": 2.9062, "step": 2437 }, { "epoch": 4.35, "learning_rate": 5.600459515543732e-06, "loss": 1.8789, "step": 2438 }, { "epoch": 4.36, "learning_rate": 5.57014743692143e-06, "loss": 1.7178, "step": 2439 }, { "epoch": 4.36, "learning_rate": 5.539912772500944e-06, "loss": 1.5664, "step": 2440 }, { "epoch": 4.36, "learning_rate": 5.509755574962894e-06, "loss": 2.2031, "step": 2441 }, { "epoch": 4.36, "learning_rate": 5.479675896852921e-06, "loss": 1.9824, "step": 2442 }, { "epoch": 4.36, "learning_rate": 5.449673790581611e-06, "loss": 2.543, "step": 2443 }, { "epoch": 4.36, "learning_rate": 5.419749308424377e-06, "loss": 1.8477, "step": 2444 }, { "epoch": 4.37, "learning_rate": 5.3899025025213935e-06, "loss": 1.6387, "step": 2445 }, { "epoch": 4.37, "learning_rate": 5.360133424877467e-06, "loss": 1.6992, "step": 2446 }, { "epoch": 4.37, "learning_rate": 5.3304421273620065e-06, "loss": 1.3496, "step": 2447 }, { "epoch": 4.37, "learning_rate": 5.300828661708873e-06, "loss": 1.7969, "step": 2448 }, { "epoch": 4.37, "learning_rate": 5.27129307951631e-06, "loss": 1.4717, "step": 2449 }, { "epoch": 4.38, "learning_rate": 5.241835432246889e-06, "loss": 2.4336, "step": 2450 }, { "epoch": 4.38, "learning_rate": 5.212455771227353e-06, "loss": 2.6016, "step": 2451 }, { "epoch": 4.38, "learning_rate": 5.183154147648578e-06, "loss": 2.0957, "step": 2452 }, { "epoch": 4.38, "learning_rate": 5.153930612565461e-06, "loss": 1.5918, "step": 2453 }, { "epoch": 4.38, "learning_rate": 5.124785216896855e-06, "loss": 1.5488, "step": 2454 }, { "epoch": 4.38, "learning_rate": 5.095718011425454e-06, "loss": 1.5381, "step": 2455 }, { "epoch": 4.39, "learning_rate": 5.066729046797691e-06, "loss": 2.4141, "step": 2456 }, { "epoch": 4.39, "learning_rate": 5.037818373523723e-06, "loss": 2.2949, "step": 2457 }, { "epoch": 4.39, "learning_rate": 5.0089860419772536e-06, "loss": 1.9004, "step": 2458 }, { "epoch": 4.39, "learning_rate": 4.980232102395488e-06, "loss": 1.3809, "step": 2459 }, { "epoch": 4.39, "learning_rate": 4.951556604879048e-06, "loss": 2.0234, "step": 2460 }, { "epoch": 4.39, "learning_rate": 4.922959599391891e-06, "loss": 2.2246, "step": 2461 }, { "epoch": 4.4, "learning_rate": 4.894441135761196e-06, "loss": 1.2744, "step": 2462 }, { "epoch": 4.4, "learning_rate": 4.866001263677283e-06, "loss": 1.5098, "step": 2463 }, { "epoch": 4.4, "learning_rate": 4.837640032693558e-06, "loss": 2.0371, "step": 2464 }, { "epoch": 4.4, "learning_rate": 4.809357492226396e-06, "loss": 1.8887, "step": 2465 }, { "epoch": 4.4, "learning_rate": 4.781153691555035e-06, "loss": 2.1445, "step": 2466 }, { "epoch": 4.41, "learning_rate": 4.75302867982157e-06, "loss": 2.125, "step": 2467 }, { "epoch": 4.41, "learning_rate": 4.724982506030762e-06, "loss": 2.0781, "step": 2468 }, { "epoch": 4.41, "learning_rate": 4.697015219050044e-06, "loss": 2.0, "step": 2469 }, { "epoch": 4.41, "learning_rate": 4.669126867609375e-06, "loss": 1.1074, "step": 2470 }, { "epoch": 4.41, "learning_rate": 4.641317500301173e-06, "loss": 1.6318, "step": 2471 }, { "epoch": 4.41, "learning_rate": 4.6135871655802685e-06, "loss": 2.1094, "step": 2472 }, { "epoch": 4.42, "learning_rate": 4.585935911763756e-06, "loss": 2.0312, "step": 2473 }, { "epoch": 4.42, "learning_rate": 4.558363787030939e-06, "loss": 1.9883, "step": 2474 }, { "epoch": 4.42, "learning_rate": 4.530870839423257e-06, "loss": 1.5332, "step": 2475 }, { "epoch": 4.42, "learning_rate": 4.5034571168442005e-06, "loss": 1.958, "step": 2476 }, { "epoch": 4.42, "learning_rate": 4.476122667059207e-06, "loss": 2.2285, "step": 2477 }, { "epoch": 4.42, "learning_rate": 4.448867537695578e-06, "loss": 2.2383, "step": 2478 }, { "epoch": 4.43, "learning_rate": 4.421691776242442e-06, "loss": 2.2773, "step": 2479 }, { "epoch": 4.43, "learning_rate": 4.394595430050613e-06, "loss": 1.7637, "step": 2480 }, { "epoch": 4.43, "learning_rate": 4.367578546332529e-06, "loss": 1.6465, "step": 2481 }, { "epoch": 4.43, "learning_rate": 4.34064117216218e-06, "loss": 2.0547, "step": 2482 }, { "epoch": 4.43, "learning_rate": 4.313783354475026e-06, "loss": 1.7432, "step": 2483 }, { "epoch": 4.44, "learning_rate": 4.287005140067912e-06, "loss": 1.5234, "step": 2484 }, { "epoch": 4.44, "learning_rate": 4.260306575598949e-06, "loss": 2.0762, "step": 2485 }, { "epoch": 4.44, "learning_rate": 4.233687707587513e-06, "loss": 1.4199, "step": 2486 }, { "epoch": 4.44, "learning_rate": 4.207148582414094e-06, "loss": 1.6504, "step": 2487 }, { "epoch": 4.44, "learning_rate": 4.180689246320235e-06, "loss": 1.6309, "step": 2488 }, { "epoch": 4.44, "learning_rate": 4.154309745408452e-06, "loss": 2.2539, "step": 2489 }, { "epoch": 4.45, "learning_rate": 4.12801012564219e-06, "loss": 2.2344, "step": 2490 }, { "epoch": 4.45, "learning_rate": 4.1017904328456644e-06, "loss": 1.7666, "step": 2491 }, { "epoch": 4.45, "learning_rate": 4.075650712703849e-06, "loss": 2.0254, "step": 2492 }, { "epoch": 4.45, "learning_rate": 4.0495910107623825e-06, "loss": 1.7773, "step": 2493 }, { "epoch": 4.45, "learning_rate": 4.023611372427471e-06, "loss": 1.6289, "step": 2494 }, { "epoch": 4.46, "learning_rate": 3.997711842965813e-06, "loss": 2.6934, "step": 2495 }, { "epoch": 4.46, "learning_rate": 3.971892467504517e-06, "loss": 1.6738, "step": 2496 }, { "epoch": 4.46, "learning_rate": 3.946153291031074e-06, "loss": 1.9463, "step": 2497 }, { "epoch": 4.46, "learning_rate": 3.920494358393195e-06, "loss": 1.0938, "step": 2498 }, { "epoch": 4.46, "learning_rate": 3.8949157142987746e-06, "loss": 1.9482, "step": 2499 }, { "epoch": 4.46, "learning_rate": 3.8694174033158555e-06, "loss": 1.8242, "step": 2500 }, { "epoch": 4.47, "learning_rate": 3.8439994698724664e-06, "loss": 1.0, "step": 2501 }, { "epoch": 4.47, "learning_rate": 3.8186619582565974e-06, "loss": 1.5469, "step": 2502 }, { "epoch": 4.47, "learning_rate": 3.793404912616111e-06, "loss": 1.5801, "step": 2503 }, { "epoch": 4.47, "learning_rate": 3.7682283769586878e-06, "loss": 2.2266, "step": 2504 }, { "epoch": 4.47, "learning_rate": 3.743132395151705e-06, "loss": 1.2139, "step": 2505 }, { "epoch": 4.47, "learning_rate": 3.71811701092219e-06, "loss": 2.6582, "step": 2506 }, { "epoch": 4.48, "learning_rate": 3.6931822678567317e-06, "loss": 1.2139, "step": 2507 }, { "epoch": 4.48, "learning_rate": 3.6683282094014282e-06, "loss": 1.8633, "step": 2508 }, { "epoch": 4.48, "learning_rate": 3.643554878861777e-06, "loss": 1.7832, "step": 2509 }, { "epoch": 4.48, "learning_rate": 3.6188623194026104e-06, "loss": 2.5781, "step": 2510 }, { "epoch": 4.48, "learning_rate": 3.5942505740480582e-06, "loss": 1.5977, "step": 2511 }, { "epoch": 4.49, "learning_rate": 3.5697196856814054e-06, "loss": 2.0234, "step": 2512 }, { "epoch": 4.49, "learning_rate": 3.545269697045067e-06, "loss": 2.1172, "step": 2513 }, { "epoch": 4.49, "learning_rate": 3.5209006507404885e-06, "loss": 1.8691, "step": 2514 }, { "epoch": 4.49, "learning_rate": 3.4966125892281097e-06, "loss": 1.7988, "step": 2515 }, { "epoch": 4.49, "learning_rate": 3.4724055548272403e-06, "loss": 1.9209, "step": 2516 }, { "epoch": 4.49, "learning_rate": 3.4482795897160025e-06, "loss": 1.7695, "step": 2517 }, { "epoch": 4.5, "learning_rate": 3.4242347359312865e-06, "loss": 2.3086, "step": 2518 }, { "epoch": 4.5, "learning_rate": 3.4002710353686405e-06, "loss": 1.7979, "step": 2519 }, { "epoch": 4.5, "learning_rate": 3.376388529782215e-06, "loss": 2.3223, "step": 2520 }, { "epoch": 4.5, "learning_rate": 3.352587260784679e-06, "loss": 1.6924, "step": 2521 }, { "epoch": 4.5, "learning_rate": 3.32886726984718e-06, "loss": 1.7881, "step": 2522 }, { "epoch": 4.51, "learning_rate": 3.3052285982992202e-06, "loss": 1.6494, "step": 2523 }, { "epoch": 4.51, "learning_rate": 3.2816712873286136e-06, "loss": 2.2832, "step": 2524 }, { "epoch": 4.51, "learning_rate": 3.258195377981443e-06, "loss": 2.332, "step": 2525 }, { "epoch": 4.51, "learning_rate": 3.2348009111619226e-06, "loss": 1.4531, "step": 2526 }, { "epoch": 4.51, "learning_rate": 3.2114879276323783e-06, "loss": 1.749, "step": 2527 }, { "epoch": 4.51, "learning_rate": 3.18825646801314e-06, "loss": 1.8594, "step": 2528 }, { "epoch": 4.52, "learning_rate": 3.1651065727825336e-06, "loss": 1.7783, "step": 2529 }, { "epoch": 4.52, "learning_rate": 3.1420382822767323e-06, "loss": 0.9819, "step": 2530 }, { "epoch": 4.52, "learning_rate": 3.119051636689724e-06, "loss": 1.6123, "step": 2531 }, { "epoch": 4.52, "learning_rate": 3.096146676073253e-06, "loss": 2.2227, "step": 2532 }, { "epoch": 4.52, "learning_rate": 3.073323440336734e-06, "loss": 1.4248, "step": 2533 }, { "epoch": 4.53, "learning_rate": 3.0505819692471792e-06, "loss": 1.4111, "step": 2534 }, { "epoch": 4.53, "learning_rate": 3.027922302429126e-06, "loss": 2.3457, "step": 2535 }, { "epoch": 4.53, "learning_rate": 3.0053444793646023e-06, "loss": 2.7812, "step": 2536 }, { "epoch": 4.53, "learning_rate": 2.9828485393930128e-06, "loss": 1.4395, "step": 2537 }, { "epoch": 4.53, "learning_rate": 2.960434521711086e-06, "loss": 1.4043, "step": 2538 }, { "epoch": 4.53, "learning_rate": 2.938102465372833e-06, "loss": 1.6406, "step": 2539 }, { "epoch": 4.54, "learning_rate": 2.9158524092894214e-06, "loss": 1.5254, "step": 2540 }, { "epoch": 4.54, "learning_rate": 2.8936843922291847e-06, "loss": 1.7969, "step": 2541 }, { "epoch": 4.54, "learning_rate": 2.8715984528174754e-06, "loss": 2.3359, "step": 2542 }, { "epoch": 4.54, "learning_rate": 2.8495946295366505e-06, "loss": 1.7285, "step": 2543 }, { "epoch": 4.54, "learning_rate": 2.827672960725991e-06, "loss": 2.2148, "step": 2544 }, { "epoch": 4.54, "learning_rate": 2.8058334845816213e-06, "loss": 1.1973, "step": 2545 }, { "epoch": 4.55, "learning_rate": 2.7840762391564636e-06, "loss": 2.2539, "step": 2546 }, { "epoch": 4.55, "learning_rate": 2.7624012623601704e-06, "loss": 1.9492, "step": 2547 }, { "epoch": 4.55, "learning_rate": 2.7408085919590264e-06, "loss": 1.6807, "step": 2548 }, { "epoch": 4.55, "learning_rate": 2.7192982655759205e-06, "loss": 1.8311, "step": 2549 }, { "epoch": 4.55, "learning_rate": 2.6978703206902657e-06, "loss": 1.8057, "step": 2550 }, { "epoch": 4.56, "learning_rate": 2.6765247946379357e-06, "loss": 1.5381, "step": 2551 }, { "epoch": 4.56, "learning_rate": 2.6552617246111964e-06, "loss": 1.7227, "step": 2552 }, { "epoch": 4.56, "learning_rate": 2.634081147658635e-06, "loss": 1.6621, "step": 2553 }, { "epoch": 4.56, "learning_rate": 2.612983100685118e-06, "loss": 1.8574, "step": 2554 }, { "epoch": 4.56, "learning_rate": 2.591967620451707e-06, "loss": 2.4766, "step": 2555 }, { "epoch": 4.56, "learning_rate": 2.5710347435755957e-06, "loss": 1.2715, "step": 2556 }, { "epoch": 4.57, "learning_rate": 2.5501845065300458e-06, "loss": 1.8135, "step": 2557 }, { "epoch": 4.57, "learning_rate": 2.5294169456443416e-06, "loss": 2.0195, "step": 2558 }, { "epoch": 4.57, "learning_rate": 2.508732097103711e-06, "loss": 1.4297, "step": 2559 }, { "epoch": 4.57, "learning_rate": 2.488129996949251e-06, "loss": 1.9316, "step": 2560 }, { "epoch": 4.57, "learning_rate": 2.467610681077892e-06, "loss": 1.584, "step": 2561 }, { "epoch": 4.58, "learning_rate": 2.4471741852423237e-06, "loss": 1.8271, "step": 2562 }, { "epoch": 4.58, "learning_rate": 2.4268205450509174e-06, "loss": 1.9746, "step": 2563 }, { "epoch": 4.58, "learning_rate": 2.4065497959676776e-06, "loss": 1.6211, "step": 2564 }, { "epoch": 4.58, "learning_rate": 2.3863619733122068e-06, "loss": 2.334, "step": 2565 }, { "epoch": 4.58, "learning_rate": 2.3662571122595787e-06, "loss": 1.5371, "step": 2566 }, { "epoch": 4.58, "learning_rate": 2.346235247840345e-06, "loss": 2.2715, "step": 2567 }, { "epoch": 4.59, "learning_rate": 2.326296414940432e-06, "loss": 2.7539, "step": 2568 }, { "epoch": 4.59, "learning_rate": 2.306440648301095e-06, "loss": 1.8984, "step": 2569 }, { "epoch": 4.59, "learning_rate": 2.286667982518853e-06, "loss": 2.2578, "step": 2570 }, { "epoch": 4.59, "learning_rate": 2.266978452045426e-06, "loss": 1.8555, "step": 2571 }, { "epoch": 4.59, "learning_rate": 2.247372091187705e-06, "loss": 2.2148, "step": 2572 }, { "epoch": 4.59, "learning_rate": 2.227848934107635e-06, "loss": 1.4531, "step": 2573 }, { "epoch": 4.6, "learning_rate": 2.2084090148221935e-06, "loss": 2.3145, "step": 2574 }, { "epoch": 4.6, "learning_rate": 2.1890523672033536e-06, "loss": 1.9248, "step": 2575 }, { "epoch": 4.6, "learning_rate": 2.1697790249779636e-06, "loss": 2.8672, "step": 2576 }, { "epoch": 4.6, "learning_rate": 2.1505890217277335e-06, "loss": 2.0, "step": 2577 }, { "epoch": 4.6, "learning_rate": 2.1314823908891556e-06, "loss": 2.1211, "step": 2578 }, { "epoch": 4.61, "learning_rate": 2.1124591657534774e-06, "loss": 1.71, "step": 2579 }, { "epoch": 4.61, "learning_rate": 2.093519379466602e-06, "loss": 2.2305, "step": 2580 }, { "epoch": 4.61, "learning_rate": 2.074663065029042e-06, "loss": 2.6738, "step": 2581 }, { "epoch": 4.61, "learning_rate": 2.055890255295906e-06, "loss": 1.9756, "step": 2582 }, { "epoch": 4.61, "learning_rate": 2.0372009829767557e-06, "loss": 1.8809, "step": 2583 }, { "epoch": 4.61, "learning_rate": 2.018595280635638e-06, "loss": 2.4258, "step": 2584 }, { "epoch": 4.62, "learning_rate": 2.000073180690959e-06, "loss": 2.4492, "step": 2585 }, { "epoch": 4.62, "learning_rate": 1.9816347154154866e-06, "loss": 1.2861, "step": 2586 }, { "epoch": 4.62, "learning_rate": 1.963279916936234e-06, "loss": 1.9912, "step": 2587 }, { "epoch": 4.62, "learning_rate": 1.945008817234445e-06, "loss": 1.5957, "step": 2588 }, { "epoch": 4.62, "learning_rate": 1.9268214481455493e-06, "loss": 1.4951, "step": 2589 }, { "epoch": 4.62, "learning_rate": 1.908717841359048e-06, "loss": 2.7852, "step": 2590 }, { "epoch": 4.63, "learning_rate": 1.890698028418525e-06, "loss": 1.4521, "step": 2591 }, { "epoch": 4.63, "learning_rate": 1.8727620407215374e-06, "loss": 1.0488, "step": 2592 }, { "epoch": 4.63, "learning_rate": 1.8549099095196143e-06, "loss": 1.8809, "step": 2593 }, { "epoch": 4.63, "learning_rate": 1.8371416659181517e-06, "loss": 2.3477, "step": 2594 }, { "epoch": 4.63, "learning_rate": 1.8194573408763794e-06, "loss": 2.0078, "step": 2595 }, { "epoch": 4.64, "learning_rate": 1.8018569652073381e-06, "loss": 2.2031, "step": 2596 }, { "epoch": 4.64, "learning_rate": 1.784340569577758e-06, "loss": 1.8174, "step": 2597 }, { "epoch": 4.64, "learning_rate": 1.7669081845080648e-06, "loss": 1.5273, "step": 2598 }, { "epoch": 4.64, "learning_rate": 1.7495598403722947e-06, "loss": 1.8301, "step": 2599 }, { "epoch": 4.64, "learning_rate": 1.7322955673980678e-06, "loss": 1.6211, "step": 2600 }, { "epoch": 4.64, "learning_rate": 1.7151153956665e-06, "loss": 1.6348, "step": 2601 }, { "epoch": 4.65, "learning_rate": 1.6980193551121847e-06, "loss": 1.3672, "step": 2602 }, { "epoch": 4.65, "learning_rate": 1.6810074755231053e-06, "loss": 1.9316, "step": 2603 }, { "epoch": 4.65, "learning_rate": 1.6640797865406288e-06, "loss": 2.2637, "step": 2604 }, { "epoch": 4.65, "learning_rate": 1.647236317659423e-06, "loss": 1.2324, "step": 2605 }, { "epoch": 4.65, "learning_rate": 1.6304770982273898e-06, "loss": 1.916, "step": 2606 }, { "epoch": 4.66, "learning_rate": 1.6138021574456708e-06, "loss": 2.4258, "step": 2607 }, { "epoch": 4.66, "learning_rate": 1.597211524368536e-06, "loss": 1.8789, "step": 2608 }, { "epoch": 4.66, "learning_rate": 1.5807052279033619e-06, "loss": 1.6309, "step": 2609 }, { "epoch": 4.66, "learning_rate": 1.5642832968105758e-06, "loss": 2.4473, "step": 2610 }, { "epoch": 4.66, "learning_rate": 1.547945759703623e-06, "loss": 1.4082, "step": 2611 }, { "epoch": 4.66, "learning_rate": 1.5316926450488878e-06, "loss": 2.4805, "step": 2612 }, { "epoch": 4.67, "learning_rate": 1.5155239811656563e-06, "loss": 2.2988, "step": 2613 }, { "epoch": 4.67, "learning_rate": 1.499439796226082e-06, "loss": 2.2734, "step": 2614 }, { "epoch": 4.67, "learning_rate": 1.4834401182551027e-06, "loss": 2.3574, "step": 2615 }, { "epoch": 4.67, "learning_rate": 1.4675249751304354e-06, "loss": 1.8896, "step": 2616 }, { "epoch": 4.67, "learning_rate": 1.4516943945824758e-06, "loss": 2.1094, "step": 2617 }, { "epoch": 4.67, "learning_rate": 1.4359484041943038e-06, "loss": 2.5527, "step": 2618 }, { "epoch": 4.68, "learning_rate": 1.420287031401607e-06, "loss": 2.0273, "step": 2619 }, { "epoch": 4.68, "learning_rate": 1.404710303492618e-06, "loss": 1.6865, "step": 2620 }, { "epoch": 4.68, "learning_rate": 1.3892182476080983e-06, "loss": 1.7324, "step": 2621 }, { "epoch": 4.68, "learning_rate": 1.3738108907412838e-06, "loss": 1.6562, "step": 2622 }, { "epoch": 4.68, "learning_rate": 1.3584882597378223e-06, "loss": 1.4883, "step": 2623 }, { "epoch": 4.69, "learning_rate": 1.3432503812957242e-06, "loss": 2.0859, "step": 2624 }, { "epoch": 4.69, "learning_rate": 1.328097281965357e-06, "loss": 1.9512, "step": 2625 }, { "epoch": 4.69, "learning_rate": 1.3130289881493451e-06, "loss": 1.9316, "step": 2626 }, { "epoch": 4.69, "learning_rate": 1.2980455261025648e-06, "loss": 2.0977, "step": 2627 }, { "epoch": 4.69, "learning_rate": 1.2831469219320601e-06, "loss": 1.7324, "step": 2628 }, { "epoch": 4.69, "learning_rate": 1.268333201597044e-06, "loss": 1.5332, "step": 2629 }, { "epoch": 4.7, "learning_rate": 1.2536043909088191e-06, "loss": 1.3379, "step": 2630 }, { "epoch": 4.7, "learning_rate": 1.238960515530735e-06, "loss": 1.9443, "step": 2631 }, { "epoch": 4.7, "learning_rate": 1.2244016009781701e-06, "loss": 1.7051, "step": 2632 }, { "epoch": 4.7, "learning_rate": 1.2099276726184383e-06, "loss": 1.3701, "step": 2633 }, { "epoch": 4.7, "learning_rate": 1.1955387556707998e-06, "loss": 2.1465, "step": 2634 }, { "epoch": 4.71, "learning_rate": 1.1812348752063772e-06, "loss": 1.6455, "step": 2635 }, { "epoch": 4.71, "learning_rate": 1.1670160561481459e-06, "loss": 1.8672, "step": 2636 }, { "epoch": 4.71, "learning_rate": 1.1528823232708375e-06, "loss": 2.6934, "step": 2637 }, { "epoch": 4.71, "learning_rate": 1.1388337012009643e-06, "loss": 1.5605, "step": 2638 }, { "epoch": 4.71, "learning_rate": 1.1248702144167122e-06, "loss": 1.6846, "step": 2639 }, { "epoch": 4.71, "learning_rate": 1.1109918872479642e-06, "loss": 1.6816, "step": 2640 }, { "epoch": 4.72, "learning_rate": 1.0971987438761822e-06, "loss": 2.3086, "step": 2641 }, { "epoch": 4.72, "learning_rate": 1.0834908083344253e-06, "loss": 1.5107, "step": 2642 }, { "epoch": 4.72, "learning_rate": 1.0698681045072989e-06, "loss": 2.6367, "step": 2643 }, { "epoch": 4.72, "learning_rate": 1.0563306561308772e-06, "loss": 1.4746, "step": 2644 }, { "epoch": 4.72, "learning_rate": 1.0428784867926978e-06, "loss": 2.1289, "step": 2645 }, { "epoch": 4.72, "learning_rate": 1.0295116199317057e-06, "loss": 1.8672, "step": 2646 }, { "epoch": 4.73, "learning_rate": 1.016230078838226e-06, "loss": 1.8242, "step": 2647 }, { "epoch": 4.73, "learning_rate": 1.0030338866538924e-06, "loss": 2.6914, "step": 2648 }, { "epoch": 4.73, "learning_rate": 9.899230663716453e-07, "loss": 2.0234, "step": 2649 }, { "epoch": 4.73, "learning_rate": 9.768976408356666e-07, "loss": 2.1094, "step": 2650 }, { "epoch": 4.73, "learning_rate": 9.639576327413525e-07, "loss": 2.0742, "step": 2651 }, { "epoch": 4.74, "learning_rate": 9.511030646352615e-07, "loss": 1.4316, "step": 2652 }, { "epoch": 4.74, "learning_rate": 9.383339589150775e-07, "loss": 1.5967, "step": 2653 }, { "epoch": 4.74, "learning_rate": 9.256503378295978e-07, "loss": 1.9785, "step": 2654 }, { "epoch": 4.74, "learning_rate": 9.130522234786499e-07, "loss": 2.1113, "step": 2655 }, { "epoch": 4.74, "learning_rate": 9.005396378130749e-07, "loss": 1.3027, "step": 2656 }, { "epoch": 4.74, "learning_rate": 8.881126026347109e-07, "loss": 2.0742, "step": 2657 }, { "epoch": 4.75, "learning_rate": 8.757711395963098e-07, "loss": 1.6689, "step": 2658 }, { "epoch": 4.75, "learning_rate": 8.63515270201537e-07, "loss": 1.9199, "step": 2659 }, { "epoch": 4.75, "learning_rate": 8.513450158049108e-07, "loss": 1.8516, "step": 2660 }, { "epoch": 4.75, "learning_rate": 8.392603976117908e-07, "loss": 1.4902, "step": 2661 }, { "epoch": 4.75, "learning_rate": 8.27261436678306e-07, "loss": 1.8496, "step": 2662 }, { "epoch": 4.76, "learning_rate": 8.153481539113438e-07, "loss": 2.4316, "step": 2663 }, { "epoch": 4.76, "learning_rate": 8.035205700685167e-07, "loss": 1.8652, "step": 2664 }, { "epoch": 4.76, "learning_rate": 7.917787057581061e-07, "loss": 1.5332, "step": 2665 }, { "epoch": 4.76, "learning_rate": 7.801225814390245e-07, "loss": 2.3262, "step": 2666 }, { "epoch": 4.76, "learning_rate": 7.685522174208204e-07, "loss": 1.7012, "step": 2667 }, { "epoch": 4.76, "learning_rate": 7.570676338635896e-07, "loss": 1.5059, "step": 2668 }, { "epoch": 4.77, "learning_rate": 7.456688507779808e-07, "loss": 1.7148, "step": 2669 }, { "epoch": 4.77, "learning_rate": 7.343558880251289e-07, "loss": 1.5352, "step": 2670 }, { "epoch": 4.77, "learning_rate": 7.231287653166552e-07, "loss": 1.3838, "step": 2671 }, { "epoch": 4.77, "learning_rate": 7.11987502214595e-07, "loss": 1.8877, "step": 2672 }, { "epoch": 4.77, "learning_rate": 7.009321181314033e-07, "loss": 2.0723, "step": 2673 }, { "epoch": 4.78, "learning_rate": 6.899626323298713e-07, "loss": 2.5352, "step": 2674 }, { "epoch": 4.78, "learning_rate": 6.790790639231548e-07, "loss": 1.5234, "step": 2675 }, { "epoch": 4.78, "learning_rate": 6.682814318746844e-07, "loss": 1.877, "step": 2676 }, { "epoch": 4.78, "learning_rate": 6.575697549981663e-07, "loss": 1.7549, "step": 2677 }, { "epoch": 4.78, "learning_rate": 6.46944051957532e-07, "loss": 1.7324, "step": 2678 }, { "epoch": 4.78, "learning_rate": 6.364043412669218e-07, "loss": 1.8672, "step": 2679 }, { "epoch": 4.79, "learning_rate": 6.259506412906402e-07, "loss": 1.7871, "step": 2680 }, { "epoch": 4.79, "learning_rate": 6.15582970243117e-07, "loss": 2.1348, "step": 2681 }, { "epoch": 4.79, "learning_rate": 6.053013461889023e-07, "loss": 1.9082, "step": 2682 }, { "epoch": 4.79, "learning_rate": 5.951057870426158e-07, "loss": 1.7539, "step": 2683 }, { "epoch": 4.79, "learning_rate": 5.849963105689027e-07, "loss": 1.752, "step": 2684 }, { "epoch": 4.79, "learning_rate": 5.749729343824395e-07, "loss": 1.8359, "step": 2685 }, { "epoch": 4.8, "learning_rate": 5.650356759478669e-07, "loss": 1.1562, "step": 2686 }, { "epoch": 4.8, "learning_rate": 5.551845525797849e-07, "loss": 1.3906, "step": 2687 }, { "epoch": 4.8, "learning_rate": 5.454195814427021e-07, "loss": 2.418, "step": 2688 }, { "epoch": 4.8, "learning_rate": 5.35740779551025e-07, "loss": 2.0293, "step": 2689 }, { "epoch": 4.8, "learning_rate": 5.261481637690246e-07, "loss": 2.9785, "step": 2690 }, { "epoch": 4.81, "learning_rate": 5.166417508107868e-07, "loss": 1.5547, "step": 2691 }, { "epoch": 4.81, "learning_rate": 5.072215572402006e-07, "loss": 1.7578, "step": 2692 }, { "epoch": 4.81, "learning_rate": 4.97887599470942e-07, "loss": 1.6025, "step": 2693 }, { "epoch": 4.81, "learning_rate": 4.886398937664127e-07, "loss": 1.4492, "step": 2694 }, { "epoch": 4.81, "learning_rate": 4.794784562397458e-07, "loss": 1.335, "step": 2695 }, { "epoch": 4.81, "learning_rate": 4.704033028537391e-07, "loss": 1.7422, "step": 2696 }, { "epoch": 4.82, "learning_rate": 4.614144494208827e-07, "loss": 2.0527, "step": 2697 }, { "epoch": 4.82, "learning_rate": 4.52511911603265e-07, "loss": 1.8281, "step": 2698 }, { "epoch": 4.82, "learning_rate": 4.436957049125945e-07, "loss": 1.7871, "step": 2699 }, { "epoch": 4.82, "learning_rate": 4.349658447101612e-07, "loss": 1.9297, "step": 2700 }, { "epoch": 4.82, "learning_rate": 4.263223462067978e-07, "loss": 1.8447, "step": 2701 }, { "epoch": 4.83, "learning_rate": 4.177652244628627e-07, "loss": 1.4258, "step": 2702 }, { "epoch": 4.83, "learning_rate": 4.0929449438821265e-07, "loss": 1.9199, "step": 2703 }, { "epoch": 4.83, "learning_rate": 4.009101707421803e-07, "loss": 1.5078, "step": 2704 }, { "epoch": 4.83, "learning_rate": 3.9261226813353537e-07, "loss": 1.9512, "step": 2705 }, { "epoch": 4.83, "learning_rate": 3.844008010204736e-07, "loss": 1.293, "step": 2706 }, { "epoch": 4.83, "learning_rate": 3.7627578371058905e-07, "loss": 1.7109, "step": 2707 }, { "epoch": 4.84, "learning_rate": 3.682372303608461e-07, "loss": 1.4766, "step": 2708 }, { "epoch": 4.84, "learning_rate": 3.602851549775521e-07, "loss": 1.251, "step": 2709 }, { "epoch": 4.84, "learning_rate": 3.524195714163292e-07, "loss": 1.834, "step": 2710 }, { "epoch": 4.84, "learning_rate": 3.4464049338210904e-07, "loss": 1.4121, "step": 2711 }, { "epoch": 4.84, "learning_rate": 3.369479344290938e-07, "loss": 2.1133, "step": 2712 }, { "epoch": 4.84, "learning_rate": 3.293419079607285e-07, "loss": 1.7266, "step": 2713 }, { "epoch": 4.85, "learning_rate": 3.2182242722969545e-07, "loss": 2.4102, "step": 2714 }, { "epoch": 4.85, "learning_rate": 3.143895053378698e-07, "loss": 1.7324, "step": 2715 }, { "epoch": 4.85, "learning_rate": 3.0704315523631953e-07, "loss": 1.9834, "step": 2716 }, { "epoch": 4.85, "learning_rate": 2.9978338972525e-07, "loss": 2.6699, "step": 2717 }, { "epoch": 4.85, "learning_rate": 2.926102214540316e-07, "loss": 1.8311, "step": 2718 }, { "epoch": 4.86, "learning_rate": 2.855236629211222e-07, "loss": 1.7998, "step": 2719 }, { "epoch": 4.86, "learning_rate": 2.785237264740781e-07, "loss": 2.0742, "step": 2720 }, { "epoch": 4.86, "learning_rate": 2.71610424309543e-07, "loss": 1.6211, "step": 2721 }, { "epoch": 4.86, "learning_rate": 2.647837684731869e-07, "loss": 1.3906, "step": 2722 }, { "epoch": 4.86, "learning_rate": 2.5804377085972274e-07, "loss": 1.3516, "step": 2723 }, { "epoch": 4.86, "learning_rate": 2.513904432128622e-07, "loss": 1.041, "step": 2724 }, { "epoch": 4.87, "learning_rate": 2.4482379712530423e-07, "loss": 2.1309, "step": 2725 }, { "epoch": 4.87, "learning_rate": 2.3834384403872978e-07, "loss": 2.0293, "step": 2726 }, { "epoch": 4.87, "learning_rate": 2.3195059524374064e-07, "loss": 1.707, "step": 2727 }, { "epoch": 4.87, "learning_rate": 2.2564406187988717e-07, "loss": 1.8945, "step": 2728 }, { "epoch": 4.87, "learning_rate": 2.1942425493561847e-07, "loss": 1.9219, "step": 2729 }, { "epoch": 4.88, "learning_rate": 2.1329118524827662e-07, "loss": 1.8486, "step": 2730 }, { "epoch": 4.88, "learning_rate": 2.0724486350406357e-07, "loss": 2.1641, "step": 2731 }, { "epoch": 4.88, "learning_rate": 2.012853002380466e-07, "loss": 1.6094, "step": 2732 }, { "epoch": 4.88, "learning_rate": 1.954125058341194e-07, "loss": 1.7754, "step": 2733 }, { "epoch": 4.88, "learning_rate": 1.896264905249856e-07, "loss": 2.1152, "step": 2734 }, { "epoch": 4.88, "learning_rate": 1.8392726439214746e-07, "loss": 1.6104, "step": 2735 }, { "epoch": 4.89, "learning_rate": 1.7831483736590048e-07, "loss": 1.2393, "step": 2736 }, { "epoch": 4.89, "learning_rate": 1.7278921922527225e-07, "loss": 2.2734, "step": 2737 }, { "epoch": 4.89, "learning_rate": 1.6735041959806687e-07, "loss": 2.0742, "step": 2738 }, { "epoch": 4.89, "learning_rate": 1.6199844796079834e-07, "loss": 2.5312, "step": 2739 }, { "epoch": 4.89, "learning_rate": 1.567333136387017e-07, "loss": 2.4141, "step": 2740 }, { "epoch": 4.89, "learning_rate": 1.5155502580569413e-07, "loss": 1.7266, "step": 2741 }, { "epoch": 4.9, "learning_rate": 1.4646359348439166e-07, "loss": 1.4414, "step": 2742 }, { "epoch": 4.9, "learning_rate": 1.414590255460535e-07, "loss": 1.5586, "step": 2743 }, { "epoch": 4.9, "learning_rate": 1.3654133071059893e-07, "loss": 2.0195, "step": 2744 }, { "epoch": 4.9, "learning_rate": 1.3171051754657938e-07, "loss": 2.1875, "step": 2745 }, { "epoch": 4.9, "learning_rate": 1.2696659447116733e-07, "loss": 2.1328, "step": 2746 }, { "epoch": 4.91, "learning_rate": 1.2230956975012863e-07, "loss": 2.082, "step": 2747 }, { "epoch": 4.91, "learning_rate": 1.1773945149782805e-07, "loss": 2.0234, "step": 2748 }, { "epoch": 4.91, "learning_rate": 1.132562476771959e-07, "loss": 1.7266, "step": 2749 }, { "epoch": 4.91, "learning_rate": 1.088599660997336e-07, "loss": 1.3896, "step": 2750 }, { "epoch": 4.91, "learning_rate": 1.0455061442548597e-07, "loss": 1.3408, "step": 2751 }, { "epoch": 4.91, "learning_rate": 1.0032820016302457e-07, "loss": 1.0957, "step": 2752 }, { "epoch": 4.92, "learning_rate": 9.619273066945322e-08, "loss": 1.2168, "step": 2753 }, { "epoch": 4.92, "learning_rate": 9.214421315038579e-08, "loss": 1.9951, "step": 2754 }, { "epoch": 4.92, "learning_rate": 8.818265465991293e-08, "loss": 1.4355, "step": 2755 }, { "epoch": 4.92, "learning_rate": 8.430806210062425e-08, "loss": 2.0723, "step": 2756 }, { "epoch": 4.92, "learning_rate": 8.052044222358612e-08, "loss": 2.1094, "step": 2757 }, { "epoch": 4.92, "learning_rate": 7.681980162830282e-08, "loss": 1.248, "step": 2758 }, { "epoch": 4.93, "learning_rate": 7.320614676274429e-08, "loss": 1.8672, "step": 2759 }, { "epoch": 4.93, "learning_rate": 6.967948392331835e-08, "loss": 2.0527, "step": 2760 }, { "epoch": 4.93, "learning_rate": 6.623981925484857e-08, "loss": 2.0293, "step": 2761 }, { "epoch": 4.93, "learning_rate": 6.288715875057416e-08, "loss": 1.5898, "step": 2762 }, { "epoch": 4.93, "learning_rate": 5.962150825214452e-08, "loss": 1.3828, "step": 2763 }, { "epoch": 4.94, "learning_rate": 5.644287344960253e-08, "loss": 2.4883, "step": 2764 }, { "epoch": 4.94, "learning_rate": 5.3351259881379014e-08, "loss": 2.2324, "step": 2765 }, { "epoch": 4.94, "learning_rate": 5.0346672934270534e-08, "loss": 1.333, "step": 2766 }, { "epoch": 4.94, "learning_rate": 4.74291178434394e-08, "loss": 2.0703, "step": 2767 }, { "epoch": 4.94, "learning_rate": 4.4598599692419195e-08, "loss": 2.1367, "step": 2768 }, { "epoch": 4.94, "learning_rate": 4.185512341307596e-08, "loss": 2.2402, "step": 2769 }, { "epoch": 4.95, "learning_rate": 3.9198693785619244e-08, "loss": 1.1123, "step": 2770 }, { "epoch": 4.95, "learning_rate": 3.662931543859105e-08, "loss": 2.2324, "step": 2771 }, { "epoch": 4.95, "learning_rate": 3.4146992848854695e-08, "loss": 1.1172, "step": 2772 }, { "epoch": 4.95, "learning_rate": 3.1751730341589295e-08, "loss": 1.4756, "step": 2773 }, { "epoch": 4.95, "learning_rate": 2.9443532090273064e-08, "loss": 1.2559, "step": 2774 }, { "epoch": 4.96, "learning_rate": 2.722240211670557e-08, "loss": 1.8398, "step": 2775 }, { "epoch": 4.96, "learning_rate": 2.5088344290946642e-08, "loss": 2.0859, "step": 2776 }, { "epoch": 4.96, "learning_rate": 2.3041362331377438e-08, "loss": 1.6895, "step": 2777 }, { "epoch": 4.96, "learning_rate": 2.1081459804628277e-08, "loss": 2.1699, "step": 2778 }, { "epoch": 4.96, "learning_rate": 1.920864012562862e-08, "loss": 1.6465, "step": 2779 }, { "epoch": 4.96, "learning_rate": 1.742290655755707e-08, "loss": 2.0391, "step": 2780 }, { "epoch": 4.97, "learning_rate": 1.5724262211863626e-08, "loss": 1.9453, "step": 2781 }, { "epoch": 4.97, "learning_rate": 1.4112710048247436e-08, "loss": 1.2617, "step": 2782 }, { "epoch": 4.97, "learning_rate": 1.2588252874673468e-08, "loss": 2.4609, "step": 2783 }, { "epoch": 4.97, "learning_rate": 1.1150893347328106e-08, "loss": 2.4375, "step": 2784 }, { "epoch": 4.97, "learning_rate": 9.800633970669104e-09, "loss": 1.7725, "step": 2785 }, { "epoch": 4.97, "learning_rate": 8.537477097364522e-09, "loss": 2.1797, "step": 2786 }, { "epoch": 4.98, "learning_rate": 7.361424928337135e-09, "loss": 2.0273, "step": 2787 }, { "epoch": 4.98, "learning_rate": 6.272479512731133e-09, "loss": 1.8643, "step": 2788 }, { "epoch": 4.98, "learning_rate": 5.270642747906562e-09, "loss": 1.208, "step": 2789 }, { "epoch": 4.98, "learning_rate": 4.355916379467084e-09, "loss": 1.7656, "step": 2790 }, { "epoch": 4.98, "learning_rate": 3.5283020012100152e-09, "loss": 2.1602, "step": 2791 }, { "epoch": 4.99, "learning_rate": 2.787801055181838e-09, "loss": 2.1543, "step": 2792 }, { "epoch": 4.99, "learning_rate": 2.1344148316060354e-09, "loss": 1.9648, "step": 2793 }, { "epoch": 4.99, "learning_rate": 1.5681444689552572e-09, "loss": 1.666, "step": 2794 }, { "epoch": 4.99, "learning_rate": 1.0889909538791543e-09, "loss": 1.8096, "step": 2795 }, { "epoch": 4.99, "learning_rate": 6.969551212598902e-10, "loss": 1.3584, "step": 2796 }, { "epoch": 4.99, "learning_rate": 3.920376541732829e-10, "loss": 1.7852, "step": 2797 }, { "epoch": 5.0, "learning_rate": 1.7423908390545863e-10, "loss": 1.0449, "step": 2798 }, { "epoch": 5.0, "learning_rate": 4.3559789952851704e-11, "loss": 2.3008, "step": 2799 }, { "epoch": 5.0, "learning_rate": 0.0, "loss": 1.8457, "step": 2800 }, { "epoch": 5.0, "step": 2800, "total_flos": 8.939540806725796e+20, "train_loss": 2.1092245047433034, "train_runtime": 7227.2285, "train_samples_per_second": 198.633, "train_steps_per_second": 0.387 } ], "max_steps": 2800, "num_train_epochs": 5, "total_flos": 8.939540806725796e+20, "trial_name": null, "trial_params": null }