{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.49449215214463893, "eval_steps": 500, "global_step": 5600, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 0.0, "loss": 2.9636, "step": 1 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 3.255, "step": 2 }, { "epoch": 0.0, "learning_rate": 5.882352941176471e-08, "loss": 3.6124, "step": 3 }, { "epoch": 0.0, "learning_rate": 5.882352941176471e-08, "loss": 3.4175, "step": 4 }, { "epoch": 0.0, "learning_rate": 1.1764705882352942e-07, "loss": 3.5507, "step": 5 }, { "epoch": 0.0, "learning_rate": 1.7647058823529414e-07, "loss": 3.1186, "step": 6 }, { "epoch": 0.0, "learning_rate": 1.7647058823529414e-07, "loss": 3.4127, "step": 7 }, { "epoch": 0.0, "learning_rate": 2.3529411764705883e-07, "loss": 3.0368, "step": 8 }, { "epoch": 0.0, "learning_rate": 2.9411764705882356e-07, "loss": 3.5186, "step": 9 }, { "epoch": 0.0, "learning_rate": 3.529411764705883e-07, "loss": 3.6207, "step": 10 }, { "epoch": 0.0, "learning_rate": 4.1176470588235295e-07, "loss": 3.2086, "step": 11 }, { "epoch": 0.0, "learning_rate": 4.7058823529411767e-07, "loss": 2.8715, "step": 12 }, { "epoch": 0.0, "learning_rate": 5.294117647058824e-07, "loss": 2.9927, "step": 13 }, { "epoch": 0.0, "learning_rate": 5.882352941176471e-07, "loss": 3.4283, "step": 14 }, { "epoch": 0.0, "learning_rate": 6.470588235294118e-07, "loss": 3.1952, "step": 15 }, { "epoch": 0.0, "learning_rate": 7.058823529411766e-07, "loss": 3.6315, "step": 16 }, { "epoch": 0.0, "learning_rate": 7.647058823529413e-07, "loss": 3.5689, "step": 17 }, { "epoch": 0.0, "learning_rate": 8.235294117647059e-07, "loss": 3.5208, "step": 18 }, { "epoch": 0.0, "learning_rate": 8.823529411764707e-07, "loss": 3.1917, "step": 19 }, { "epoch": 0.0, "learning_rate": 9.411764705882353e-07, "loss": 3.3514, "step": 20 }, { "epoch": 0.0, "learning_rate": 1.0000000000000002e-06, "loss": 3.1205, "step": 21 }, { "epoch": 0.0, "learning_rate": 1.0588235294117648e-06, "loss": 3.2811, "step": 22 }, { "epoch": 0.0, "learning_rate": 1.1176470588235296e-06, "loss": 3.0976, "step": 23 }, { "epoch": 0.0, "learning_rate": 1.1764705882352942e-06, "loss": 3.2949, "step": 24 }, { "epoch": 0.0, "learning_rate": 1.235294117647059e-06, "loss": 3.3681, "step": 25 }, { "epoch": 0.0, "learning_rate": 1.2941176470588237e-06, "loss": 3.403, "step": 26 }, { "epoch": 0.0, "learning_rate": 1.3529411764705883e-06, "loss": 3.0557, "step": 27 }, { "epoch": 0.0, "learning_rate": 1.4117647058823531e-06, "loss": 3.5182, "step": 28 }, { "epoch": 0.0, "learning_rate": 1.4705882352941177e-06, "loss": 2.8665, "step": 29 }, { "epoch": 0.0, "learning_rate": 1.5294117647058826e-06, "loss": 3.2878, "step": 30 }, { "epoch": 0.0, "learning_rate": 1.5882352941176472e-06, "loss": 3.2177, "step": 31 }, { "epoch": 0.0, "learning_rate": 1.6470588235294118e-06, "loss": 3.2192, "step": 32 }, { "epoch": 0.0, "learning_rate": 1.7058823529411766e-06, "loss": 3.1215, "step": 33 }, { "epoch": 0.0, "learning_rate": 1.7058823529411766e-06, "loss": 3.3609, "step": 34 }, { "epoch": 0.0, "learning_rate": 1.7647058823529414e-06, "loss": 3.1926, "step": 35 }, { "epoch": 0.0, "learning_rate": 1.8235294117647058e-06, "loss": 2.9637, "step": 36 }, { "epoch": 0.0, "learning_rate": 1.8823529411764707e-06, "loss": 3.4635, "step": 37 }, { "epoch": 0.0, "learning_rate": 1.9411764705882353e-06, "loss": 2.8308, "step": 38 }, { "epoch": 0.0, "learning_rate": 2.0000000000000003e-06, "loss": 3.2403, "step": 39 }, { "epoch": 0.0, "learning_rate": 2.058823529411765e-06, "loss": 3.0046, "step": 40 }, { "epoch": 0.0, "learning_rate": 2.1176470588235296e-06, "loss": 3.1984, "step": 41 }, { "epoch": 0.0, "learning_rate": 2.176470588235294e-06, "loss": 3.2416, "step": 42 }, { "epoch": 0.0, "learning_rate": 2.2352941176470592e-06, "loss": 3.1161, "step": 43 }, { "epoch": 0.0, "learning_rate": 2.2941176470588234e-06, "loss": 3.051, "step": 44 }, { "epoch": 0.0, "learning_rate": 2.3529411764705885e-06, "loss": 2.8774, "step": 45 }, { "epoch": 0.0, "learning_rate": 2.411764705882353e-06, "loss": 3.1534, "step": 46 }, { "epoch": 0.0, "learning_rate": 2.470588235294118e-06, "loss": 3.0378, "step": 47 }, { "epoch": 0.0, "learning_rate": 2.5294117647058823e-06, "loss": 2.9408, "step": 48 }, { "epoch": 0.0, "learning_rate": 2.5882352941176473e-06, "loss": 3.0045, "step": 49 }, { "epoch": 0.0, "learning_rate": 2.647058823529412e-06, "loss": 3.0065, "step": 50 }, { "epoch": 0.0, "learning_rate": 2.7058823529411766e-06, "loss": 2.6637, "step": 51 }, { "epoch": 0.0, "learning_rate": 2.7647058823529416e-06, "loss": 2.3709, "step": 52 }, { "epoch": 0.0, "learning_rate": 2.8235294117647062e-06, "loss": 2.3, "step": 53 }, { "epoch": 0.0, "learning_rate": 2.8823529411764704e-06, "loss": 2.2974, "step": 54 }, { "epoch": 0.0, "learning_rate": 2.9411764705882355e-06, "loss": 2.2971, "step": 55 }, { "epoch": 0.0, "learning_rate": 3e-06, "loss": 2.2308, "step": 56 }, { "epoch": 0.01, "learning_rate": 3.058823529411765e-06, "loss": 2.1108, "step": 57 }, { "epoch": 0.01, "learning_rate": 3.1176470588235297e-06, "loss": 2.2843, "step": 58 }, { "epoch": 0.01, "learning_rate": 3.1764705882352943e-06, "loss": 2.0902, "step": 59 }, { "epoch": 0.01, "learning_rate": 3.2352941176470594e-06, "loss": 2.0705, "step": 60 }, { "epoch": 0.01, "learning_rate": 3.2941176470588236e-06, "loss": 2.1023, "step": 61 }, { "epoch": 0.01, "learning_rate": 3.352941176470588e-06, "loss": 2.1283, "step": 62 }, { "epoch": 0.01, "learning_rate": 3.4117647058823532e-06, "loss": 2.0138, "step": 63 }, { "epoch": 0.01, "learning_rate": 3.470588235294118e-06, "loss": 2.2479, "step": 64 }, { "epoch": 0.01, "learning_rate": 3.529411764705883e-06, "loss": 2.078, "step": 65 }, { "epoch": 0.01, "learning_rate": 3.5882352941176475e-06, "loss": 2.0143, "step": 66 }, { "epoch": 0.01, "learning_rate": 3.6470588235294117e-06, "loss": 2.0438, "step": 67 }, { "epoch": 0.01, "learning_rate": 3.7058823529411767e-06, "loss": 1.8931, "step": 68 }, { "epoch": 0.01, "learning_rate": 3.7647058823529414e-06, "loss": 2.0753, "step": 69 }, { "epoch": 0.01, "learning_rate": 3.8235294117647055e-06, "loss": 2.1203, "step": 70 }, { "epoch": 0.01, "learning_rate": 3.882352941176471e-06, "loss": 1.8253, "step": 71 }, { "epoch": 0.01, "learning_rate": 3.941176470588236e-06, "loss": 2.1516, "step": 72 }, { "epoch": 0.01, "learning_rate": 4.000000000000001e-06, "loss": 2.0502, "step": 73 }, { "epoch": 0.01, "learning_rate": 4.058823529411765e-06, "loss": 2.0454, "step": 74 }, { "epoch": 0.01, "learning_rate": 4.11764705882353e-06, "loss": 1.9959, "step": 75 }, { "epoch": 0.01, "learning_rate": 4.176470588235295e-06, "loss": 2.0194, "step": 76 }, { "epoch": 0.01, "learning_rate": 4.235294117647059e-06, "loss": 2.039, "step": 77 }, { "epoch": 0.01, "learning_rate": 4.294117647058823e-06, "loss": 2.0229, "step": 78 }, { "epoch": 0.01, "learning_rate": 4.352941176470588e-06, "loss": 1.9874, "step": 79 }, { "epoch": 0.01, "learning_rate": 4.411764705882353e-06, "loss": 1.7915, "step": 80 }, { "epoch": 0.01, "learning_rate": 4.4705882352941184e-06, "loss": 1.9745, "step": 81 }, { "epoch": 0.01, "learning_rate": 4.529411764705883e-06, "loss": 2.0146, "step": 82 }, { "epoch": 0.01, "learning_rate": 4.588235294117647e-06, "loss": 1.9273, "step": 83 }, { "epoch": 0.01, "learning_rate": 4.647058823529412e-06, "loss": 1.8348, "step": 84 }, { "epoch": 0.01, "learning_rate": 4.705882352941177e-06, "loss": 1.9028, "step": 85 }, { "epoch": 0.01, "learning_rate": 4.764705882352941e-06, "loss": 1.9158, "step": 86 }, { "epoch": 0.01, "learning_rate": 4.823529411764706e-06, "loss": 1.9379, "step": 87 }, { "epoch": 0.01, "learning_rate": 4.882352941176471e-06, "loss": 1.9091, "step": 88 }, { "epoch": 0.01, "learning_rate": 4.941176470588236e-06, "loss": 2.0235, "step": 89 }, { "epoch": 0.01, "learning_rate": 5e-06, "loss": 1.9395, "step": 90 }, { "epoch": 0.01, "learning_rate": 5.058823529411765e-06, "loss": 1.8915, "step": 91 }, { "epoch": 0.01, "learning_rate": 5.11764705882353e-06, "loss": 1.9144, "step": 92 }, { "epoch": 0.01, "learning_rate": 5.176470588235295e-06, "loss": 1.9143, "step": 93 }, { "epoch": 0.01, "learning_rate": 5.23529411764706e-06, "loss": 1.9596, "step": 94 }, { "epoch": 0.01, "learning_rate": 5.294117647058824e-06, "loss": 2.0647, "step": 95 }, { "epoch": 0.01, "learning_rate": 5.352941176470589e-06, "loss": 1.8178, "step": 96 }, { "epoch": 0.01, "learning_rate": 5.411764705882353e-06, "loss": 1.9181, "step": 97 }, { "epoch": 0.01, "learning_rate": 5.470588235294119e-06, "loss": 2.061, "step": 98 }, { "epoch": 0.01, "learning_rate": 5.529411764705883e-06, "loss": 2.0527, "step": 99 }, { "epoch": 0.01, "learning_rate": 5.588235294117647e-06, "loss": 1.9509, "step": 100 }, { "epoch": 0.01, "learning_rate": 5.6470588235294125e-06, "loss": 2.0561, "step": 101 }, { "epoch": 0.01, "learning_rate": 5.705882352941177e-06, "loss": 1.9207, "step": 102 }, { "epoch": 0.01, "learning_rate": 5.764705882352941e-06, "loss": 1.9326, "step": 103 }, { "epoch": 0.01, "learning_rate": 5.823529411764707e-06, "loss": 2.0566, "step": 104 }, { "epoch": 0.01, "learning_rate": 5.882352941176471e-06, "loss": 1.6838, "step": 105 }, { "epoch": 0.01, "learning_rate": 5.941176470588236e-06, "loss": 1.8667, "step": 106 }, { "epoch": 0.01, "learning_rate": 6e-06, "loss": 1.915, "step": 107 }, { "epoch": 0.01, "learning_rate": 6.058823529411765e-06, "loss": 1.8784, "step": 108 }, { "epoch": 0.01, "learning_rate": 6.11764705882353e-06, "loss": 1.9572, "step": 109 }, { "epoch": 0.01, "learning_rate": 6.176470588235295e-06, "loss": 1.7827, "step": 110 }, { "epoch": 0.01, "learning_rate": 6.2352941176470595e-06, "loss": 1.7817, "step": 111 }, { "epoch": 0.01, "learning_rate": 6.294117647058824e-06, "loss": 1.8444, "step": 112 }, { "epoch": 0.01, "learning_rate": 6.352941176470589e-06, "loss": 1.7721, "step": 113 }, { "epoch": 0.01, "learning_rate": 6.411764705882354e-06, "loss": 1.7561, "step": 114 }, { "epoch": 0.01, "learning_rate": 6.470588235294119e-06, "loss": 1.8318, "step": 115 }, { "epoch": 0.01, "learning_rate": 6.529411764705883e-06, "loss": 1.9731, "step": 116 }, { "epoch": 0.01, "learning_rate": 6.588235294117647e-06, "loss": 1.9562, "step": 117 }, { "epoch": 0.01, "learning_rate": 6.647058823529412e-06, "loss": 1.9408, "step": 118 }, { "epoch": 0.01, "learning_rate": 6.705882352941176e-06, "loss": 1.9437, "step": 119 }, { "epoch": 0.01, "learning_rate": 6.764705882352942e-06, "loss": 1.8397, "step": 120 }, { "epoch": 0.01, "learning_rate": 6.8235294117647065e-06, "loss": 1.9309, "step": 121 }, { "epoch": 0.01, "learning_rate": 6.8823529411764715e-06, "loss": 1.8366, "step": 122 }, { "epoch": 0.01, "learning_rate": 6.941176470588236e-06, "loss": 2.003, "step": 123 }, { "epoch": 0.01, "learning_rate": 7e-06, "loss": 1.8588, "step": 124 }, { "epoch": 0.01, "learning_rate": 7.058823529411766e-06, "loss": 1.8238, "step": 125 }, { "epoch": 0.01, "learning_rate": 7.11764705882353e-06, "loss": 1.792, "step": 126 }, { "epoch": 0.01, "learning_rate": 7.176470588235295e-06, "loss": 1.991, "step": 127 }, { "epoch": 0.01, "learning_rate": 7.235294117647059e-06, "loss": 1.9142, "step": 128 }, { "epoch": 0.01, "learning_rate": 7.294117647058823e-06, "loss": 1.8821, "step": 129 }, { "epoch": 0.01, "learning_rate": 7.352941176470589e-06, "loss": 1.9564, "step": 130 }, { "epoch": 0.01, "learning_rate": 7.4117647058823535e-06, "loss": 1.8554, "step": 131 }, { "epoch": 0.01, "learning_rate": 7.4705882352941185e-06, "loss": 1.8582, "step": 132 }, { "epoch": 0.01, "learning_rate": 7.529411764705883e-06, "loss": 1.9303, "step": 133 }, { "epoch": 0.01, "learning_rate": 7.588235294117648e-06, "loss": 1.9452, "step": 134 }, { "epoch": 0.01, "learning_rate": 7.647058823529411e-06, "loss": 1.9576, "step": 135 }, { "epoch": 0.01, "learning_rate": 7.705882352941178e-06, "loss": 1.8328, "step": 136 }, { "epoch": 0.01, "learning_rate": 7.764705882352941e-06, "loss": 1.8514, "step": 137 }, { "epoch": 0.01, "learning_rate": 7.823529411764706e-06, "loss": 1.7921, "step": 138 }, { "epoch": 0.01, "learning_rate": 7.882352941176471e-06, "loss": 1.7766, "step": 139 }, { "epoch": 0.01, "learning_rate": 7.941176470588236e-06, "loss": 1.7484, "step": 140 }, { "epoch": 0.01, "learning_rate": 8.000000000000001e-06, "loss": 1.8292, "step": 141 }, { "epoch": 0.01, "learning_rate": 8.058823529411766e-06, "loss": 1.8914, "step": 142 }, { "epoch": 0.01, "learning_rate": 8.11764705882353e-06, "loss": 1.9321, "step": 143 }, { "epoch": 0.01, "learning_rate": 8.176470588235295e-06, "loss": 1.9507, "step": 144 }, { "epoch": 0.01, "learning_rate": 8.23529411764706e-06, "loss": 1.8346, "step": 145 }, { "epoch": 0.01, "learning_rate": 8.294117647058825e-06, "loss": 1.9211, "step": 146 }, { "epoch": 0.01, "learning_rate": 8.35294117647059e-06, "loss": 2.0087, "step": 147 }, { "epoch": 0.01, "learning_rate": 8.411764705882353e-06, "loss": 1.6486, "step": 148 }, { "epoch": 0.01, "learning_rate": 8.470588235294118e-06, "loss": 1.717, "step": 149 }, { "epoch": 0.01, "learning_rate": 8.529411764705883e-06, "loss": 1.568, "step": 150 }, { "epoch": 0.01, "learning_rate": 8.588235294117647e-06, "loss": 1.8613, "step": 151 }, { "epoch": 0.01, "learning_rate": 8.647058823529413e-06, "loss": 1.9449, "step": 152 }, { "epoch": 0.01, "learning_rate": 8.705882352941177e-06, "loss": 1.8637, "step": 153 }, { "epoch": 0.01, "learning_rate": 8.764705882352942e-06, "loss": 2.0128, "step": 154 }, { "epoch": 0.01, "learning_rate": 8.823529411764707e-06, "loss": 1.7952, "step": 155 }, { "epoch": 0.01, "learning_rate": 8.88235294117647e-06, "loss": 1.9479, "step": 156 }, { "epoch": 0.01, "learning_rate": 8.941176470588237e-06, "loss": 1.8059, "step": 157 }, { "epoch": 0.01, "learning_rate": 9e-06, "loss": 1.8203, "step": 158 }, { "epoch": 0.01, "learning_rate": 9.058823529411765e-06, "loss": 1.7718, "step": 159 }, { "epoch": 0.01, "learning_rate": 9.11764705882353e-06, "loss": 1.9796, "step": 160 }, { "epoch": 0.01, "learning_rate": 9.176470588235294e-06, "loss": 1.7742, "step": 161 }, { "epoch": 0.01, "learning_rate": 9.23529411764706e-06, "loss": 1.9743, "step": 162 }, { "epoch": 0.01, "learning_rate": 9.294117647058824e-06, "loss": 1.8927, "step": 163 }, { "epoch": 0.01, "learning_rate": 9.352941176470589e-06, "loss": 1.8728, "step": 164 }, { "epoch": 0.01, "learning_rate": 9.411764705882354e-06, "loss": 1.8889, "step": 165 }, { "epoch": 0.01, "learning_rate": 9.470588235294119e-06, "loss": 1.8821, "step": 166 }, { "epoch": 0.01, "learning_rate": 9.529411764705882e-06, "loss": 1.8214, "step": 167 }, { "epoch": 0.01, "learning_rate": 9.588235294117649e-06, "loss": 2.0099, "step": 168 }, { "epoch": 0.01, "learning_rate": 9.647058823529412e-06, "loss": 1.8011, "step": 169 }, { "epoch": 0.02, "learning_rate": 9.705882352941177e-06, "loss": 1.8846, "step": 170 }, { "epoch": 0.02, "learning_rate": 9.764705882352942e-06, "loss": 1.7883, "step": 171 }, { "epoch": 0.02, "learning_rate": 9.823529411764706e-06, "loss": 1.8289, "step": 172 }, { "epoch": 0.02, "learning_rate": 9.882352941176472e-06, "loss": 1.893, "step": 173 }, { "epoch": 0.02, "learning_rate": 9.941176470588236e-06, "loss": 1.7083, "step": 174 }, { "epoch": 0.02, "learning_rate": 1e-05, "loss": 1.7471, "step": 175 }, { "epoch": 0.02, "learning_rate": 1.0058823529411766e-05, "loss": 1.8829, "step": 176 }, { "epoch": 0.02, "learning_rate": 1.011764705882353e-05, "loss": 1.7948, "step": 177 }, { "epoch": 0.02, "learning_rate": 1.0176470588235294e-05, "loss": 1.7624, "step": 178 }, { "epoch": 0.02, "learning_rate": 1.023529411764706e-05, "loss": 1.777, "step": 179 }, { "epoch": 0.02, "learning_rate": 1.0294117647058823e-05, "loss": 1.7181, "step": 180 }, { "epoch": 0.02, "learning_rate": 1.035294117647059e-05, "loss": 1.8327, "step": 181 }, { "epoch": 0.02, "learning_rate": 1.0411764705882354e-05, "loss": 1.7789, "step": 182 }, { "epoch": 0.02, "learning_rate": 1.047058823529412e-05, "loss": 1.8552, "step": 183 }, { "epoch": 0.02, "learning_rate": 1.0529411764705883e-05, "loss": 1.5922, "step": 184 }, { "epoch": 0.02, "learning_rate": 1.0588235294117648e-05, "loss": 1.7897, "step": 185 }, { "epoch": 0.02, "learning_rate": 1.0647058823529413e-05, "loss": 1.8304, "step": 186 }, { "epoch": 0.02, "learning_rate": 1.0705882352941178e-05, "loss": 1.8134, "step": 187 }, { "epoch": 0.02, "learning_rate": 1.0764705882352941e-05, "loss": 1.9701, "step": 188 }, { "epoch": 0.02, "learning_rate": 1.0823529411764706e-05, "loss": 1.8111, "step": 189 }, { "epoch": 0.02, "learning_rate": 1.0882352941176471e-05, "loss": 1.7513, "step": 190 }, { "epoch": 0.02, "learning_rate": 1.0941176470588238e-05, "loss": 1.8043, "step": 191 }, { "epoch": 0.02, "learning_rate": 1.1000000000000001e-05, "loss": 1.8381, "step": 192 }, { "epoch": 0.02, "learning_rate": 1.1058823529411766e-05, "loss": 1.8956, "step": 193 }, { "epoch": 0.02, "learning_rate": 1.1117647058823531e-05, "loss": 1.7934, "step": 194 }, { "epoch": 0.02, "learning_rate": 1.1176470588235295e-05, "loss": 1.7373, "step": 195 }, { "epoch": 0.02, "learning_rate": 1.123529411764706e-05, "loss": 1.7318, "step": 196 }, { "epoch": 0.02, "learning_rate": 1.1294117647058825e-05, "loss": 1.8098, "step": 197 }, { "epoch": 0.02, "learning_rate": 1.1352941176470588e-05, "loss": 1.9791, "step": 198 }, { "epoch": 0.02, "learning_rate": 1.1411764705882353e-05, "loss": 1.5328, "step": 199 }, { "epoch": 0.02, "learning_rate": 1.1470588235294118e-05, "loss": 1.9219, "step": 200 }, { "epoch": 0.02, "learning_rate": 1.1529411764705882e-05, "loss": 1.7725, "step": 201 }, { "epoch": 0.02, "learning_rate": 1.1588235294117648e-05, "loss": 1.7413, "step": 202 }, { "epoch": 0.02, "learning_rate": 1.1647058823529413e-05, "loss": 1.8547, "step": 203 }, { "epoch": 0.02, "learning_rate": 1.1705882352941178e-05, "loss": 1.7321, "step": 204 }, { "epoch": 0.02, "learning_rate": 1.1764705882352942e-05, "loss": 1.8301, "step": 205 }, { "epoch": 0.02, "learning_rate": 1.1823529411764707e-05, "loss": 1.7957, "step": 206 }, { "epoch": 0.02, "learning_rate": 1.1882352941176472e-05, "loss": 1.8399, "step": 207 }, { "epoch": 0.02, "learning_rate": 1.1941176470588235e-05, "loss": 1.777, "step": 208 }, { "epoch": 0.02, "learning_rate": 1.2e-05, "loss": 1.7621, "step": 209 }, { "epoch": 0.02, "learning_rate": 1.2058823529411765e-05, "loss": 1.7479, "step": 210 }, { "epoch": 0.02, "learning_rate": 1.211764705882353e-05, "loss": 1.7845, "step": 211 }, { "epoch": 0.02, "learning_rate": 1.2176470588235294e-05, "loss": 1.7771, "step": 212 }, { "epoch": 0.02, "learning_rate": 1.223529411764706e-05, "loss": 1.8029, "step": 213 }, { "epoch": 0.02, "learning_rate": 1.2294117647058826e-05, "loss": 1.8568, "step": 214 }, { "epoch": 0.02, "learning_rate": 1.235294117647059e-05, "loss": 1.7339, "step": 215 }, { "epoch": 0.02, "learning_rate": 1.2411764705882354e-05, "loss": 1.7136, "step": 216 }, { "epoch": 0.02, "learning_rate": 1.2470588235294119e-05, "loss": 1.8174, "step": 217 }, { "epoch": 0.02, "learning_rate": 1.2529411764705884e-05, "loss": 1.7329, "step": 218 }, { "epoch": 0.02, "learning_rate": 1.2588235294117647e-05, "loss": 1.8153, "step": 219 }, { "epoch": 0.02, "learning_rate": 1.2647058823529412e-05, "loss": 1.7235, "step": 220 }, { "epoch": 0.02, "learning_rate": 1.2705882352941177e-05, "loss": 1.7458, "step": 221 }, { "epoch": 0.02, "learning_rate": 1.276470588235294e-05, "loss": 1.8214, "step": 222 }, { "epoch": 0.02, "learning_rate": 1.2823529411764707e-05, "loss": 1.8265, "step": 223 }, { "epoch": 0.02, "learning_rate": 1.2882352941176473e-05, "loss": 1.752, "step": 224 }, { "epoch": 0.02, "learning_rate": 1.2941176470588238e-05, "loss": 1.6561, "step": 225 }, { "epoch": 0.02, "learning_rate": 1.3000000000000001e-05, "loss": 1.7669, "step": 226 }, { "epoch": 0.02, "learning_rate": 1.3058823529411766e-05, "loss": 1.6806, "step": 227 }, { "epoch": 0.02, "learning_rate": 1.3117647058823531e-05, "loss": 1.6759, "step": 228 }, { "epoch": 0.02, "learning_rate": 1.3176470588235294e-05, "loss": 1.8423, "step": 229 }, { "epoch": 0.02, "learning_rate": 1.323529411764706e-05, "loss": 1.7432, "step": 230 }, { "epoch": 0.02, "learning_rate": 1.3294117647058824e-05, "loss": 1.7614, "step": 231 }, { "epoch": 0.02, "learning_rate": 1.3352941176470588e-05, "loss": 1.8269, "step": 232 }, { "epoch": 0.02, "learning_rate": 1.3411764705882353e-05, "loss": 1.9406, "step": 233 }, { "epoch": 0.02, "learning_rate": 1.347058823529412e-05, "loss": 1.7704, "step": 234 }, { "epoch": 0.02, "learning_rate": 1.3529411764705885e-05, "loss": 1.7758, "step": 235 }, { "epoch": 0.02, "learning_rate": 1.3588235294117648e-05, "loss": 1.9139, "step": 236 }, { "epoch": 0.02, "learning_rate": 1.3647058823529413e-05, "loss": 1.7656, "step": 237 }, { "epoch": 0.02, "learning_rate": 1.3705882352941178e-05, "loss": 1.7995, "step": 238 }, { "epoch": 0.02, "learning_rate": 1.3764705882352943e-05, "loss": 1.5958, "step": 239 }, { "epoch": 0.02, "learning_rate": 1.3823529411764706e-05, "loss": 1.8755, "step": 240 }, { "epoch": 0.02, "learning_rate": 1.3882352941176471e-05, "loss": 1.8004, "step": 241 }, { "epoch": 0.02, "learning_rate": 1.3941176470588236e-05, "loss": 1.774, "step": 242 }, { "epoch": 0.02, "learning_rate": 1.4e-05, "loss": 1.7721, "step": 243 }, { "epoch": 0.02, "learning_rate": 1.4058823529411765e-05, "loss": 1.8833, "step": 244 }, { "epoch": 0.02, "learning_rate": 1.4117647058823532e-05, "loss": 1.8525, "step": 245 }, { "epoch": 0.02, "learning_rate": 1.4176470588235297e-05, "loss": 1.7835, "step": 246 }, { "epoch": 0.02, "learning_rate": 1.423529411764706e-05, "loss": 1.753, "step": 247 }, { "epoch": 0.02, "learning_rate": 1.4294117647058825e-05, "loss": 1.8284, "step": 248 }, { "epoch": 0.02, "learning_rate": 1.435294117647059e-05, "loss": 1.8015, "step": 249 }, { "epoch": 0.02, "learning_rate": 1.4411764705882353e-05, "loss": 1.871, "step": 250 }, { "epoch": 0.02, "learning_rate": 1.4470588235294118e-05, "loss": 1.6578, "step": 251 }, { "epoch": 0.02, "learning_rate": 1.4529411764705883e-05, "loss": 1.7746, "step": 252 }, { "epoch": 0.02, "learning_rate": 1.4588235294117647e-05, "loss": 1.7694, "step": 253 }, { "epoch": 0.02, "learning_rate": 1.4647058823529412e-05, "loss": 1.8512, "step": 254 }, { "epoch": 0.02, "learning_rate": 1.4705882352941179e-05, "loss": 1.7924, "step": 255 }, { "epoch": 0.02, "learning_rate": 1.4764705882352944e-05, "loss": 1.8199, "step": 256 }, { "epoch": 0.02, "learning_rate": 1.4823529411764707e-05, "loss": 1.7804, "step": 257 }, { "epoch": 0.02, "learning_rate": 1.4882352941176472e-05, "loss": 1.6964, "step": 258 }, { "epoch": 0.02, "learning_rate": 1.4941176470588237e-05, "loss": 1.7473, "step": 259 }, { "epoch": 0.02, "learning_rate": 1.5000000000000002e-05, "loss": 1.6956, "step": 260 }, { "epoch": 0.02, "learning_rate": 1.5058823529411765e-05, "loss": 1.7133, "step": 261 }, { "epoch": 0.02, "learning_rate": 1.511764705882353e-05, "loss": 1.8428, "step": 262 }, { "epoch": 0.02, "learning_rate": 1.5176470588235295e-05, "loss": 1.7368, "step": 263 }, { "epoch": 0.02, "learning_rate": 1.5235294117647059e-05, "loss": 1.7058, "step": 264 }, { "epoch": 0.02, "learning_rate": 1.5294117647058822e-05, "loss": 1.6386, "step": 265 }, { "epoch": 0.02, "learning_rate": 1.535294117647059e-05, "loss": 1.6243, "step": 266 }, { "epoch": 0.02, "learning_rate": 1.5411764705882356e-05, "loss": 1.7292, "step": 267 }, { "epoch": 0.02, "learning_rate": 1.547058823529412e-05, "loss": 1.7552, "step": 268 }, { "epoch": 0.02, "learning_rate": 1.5529411764705882e-05, "loss": 1.7844, "step": 269 }, { "epoch": 0.02, "learning_rate": 1.558823529411765e-05, "loss": 1.8425, "step": 270 }, { "epoch": 0.02, "learning_rate": 1.5647058823529412e-05, "loss": 1.8513, "step": 271 }, { "epoch": 0.02, "learning_rate": 1.570588235294118e-05, "loss": 1.5936, "step": 272 }, { "epoch": 0.02, "learning_rate": 1.5764705882352943e-05, "loss": 1.5872, "step": 273 }, { "epoch": 0.02, "learning_rate": 1.5823529411764706e-05, "loss": 1.7768, "step": 274 }, { "epoch": 0.02, "learning_rate": 1.5882352941176473e-05, "loss": 1.8093, "step": 275 }, { "epoch": 0.02, "learning_rate": 1.594117647058824e-05, "loss": 1.7098, "step": 276 }, { "epoch": 0.02, "learning_rate": 1.6000000000000003e-05, "loss": 1.6385, "step": 277 }, { "epoch": 0.02, "learning_rate": 1.6058823529411766e-05, "loss": 1.8189, "step": 278 }, { "epoch": 0.02, "learning_rate": 1.6117647058823533e-05, "loss": 1.6522, "step": 279 }, { "epoch": 0.02, "learning_rate": 1.6176470588235296e-05, "loss": 1.8119, "step": 280 }, { "epoch": 0.02, "learning_rate": 1.623529411764706e-05, "loss": 1.691, "step": 281 }, { "epoch": 0.02, "learning_rate": 1.6294117647058826e-05, "loss": 1.8178, "step": 282 }, { "epoch": 0.02, "learning_rate": 1.635294117647059e-05, "loss": 1.7884, "step": 283 }, { "epoch": 0.03, "learning_rate": 1.6411764705882353e-05, "loss": 1.7638, "step": 284 }, { "epoch": 0.03, "learning_rate": 1.647058823529412e-05, "loss": 1.8189, "step": 285 }, { "epoch": 0.03, "learning_rate": 1.6529411764705883e-05, "loss": 1.7446, "step": 286 }, { "epoch": 0.03, "learning_rate": 1.658823529411765e-05, "loss": 1.8517, "step": 287 }, { "epoch": 0.03, "learning_rate": 1.6647058823529413e-05, "loss": 1.7914, "step": 288 }, { "epoch": 0.03, "learning_rate": 1.670588235294118e-05, "loss": 1.8452, "step": 289 }, { "epoch": 0.03, "learning_rate": 1.670588235294118e-05, "loss": 1.8341, "step": 290 }, { "epoch": 0.03, "learning_rate": 1.6764705882352943e-05, "loss": 1.8618, "step": 291 }, { "epoch": 0.03, "learning_rate": 1.6823529411764706e-05, "loss": 1.7503, "step": 292 }, { "epoch": 0.03, "learning_rate": 1.6882352941176473e-05, "loss": 1.6964, "step": 293 }, { "epoch": 0.03, "learning_rate": 1.6941176470588237e-05, "loss": 1.7998, "step": 294 }, { "epoch": 0.03, "learning_rate": 1.7e-05, "loss": 1.8688, "step": 295 }, { "epoch": 0.03, "learning_rate": 1.7058823529411767e-05, "loss": 1.8799, "step": 296 }, { "epoch": 0.03, "learning_rate": 1.711764705882353e-05, "loss": 1.6653, "step": 297 }, { "epoch": 0.03, "learning_rate": 1.7176470588235293e-05, "loss": 1.6181, "step": 298 }, { "epoch": 0.03, "learning_rate": 1.723529411764706e-05, "loss": 1.8295, "step": 299 }, { "epoch": 0.03, "learning_rate": 1.7294117647058827e-05, "loss": 1.6648, "step": 300 }, { "epoch": 0.03, "learning_rate": 1.735294117647059e-05, "loss": 1.7477, "step": 301 }, { "epoch": 0.03, "learning_rate": 1.7411764705882353e-05, "loss": 1.6922, "step": 302 }, { "epoch": 0.03, "learning_rate": 1.747058823529412e-05, "loss": 1.652, "step": 303 }, { "epoch": 0.03, "learning_rate": 1.7529411764705884e-05, "loss": 1.7404, "step": 304 }, { "epoch": 0.03, "learning_rate": 1.7588235294117647e-05, "loss": 1.8752, "step": 305 }, { "epoch": 0.03, "learning_rate": 1.7647058823529414e-05, "loss": 1.6192, "step": 306 }, { "epoch": 0.03, "learning_rate": 1.7705882352941177e-05, "loss": 1.6793, "step": 307 }, { "epoch": 0.03, "learning_rate": 1.776470588235294e-05, "loss": 1.7085, "step": 308 }, { "epoch": 0.03, "learning_rate": 1.7823529411764707e-05, "loss": 1.796, "step": 309 }, { "epoch": 0.03, "learning_rate": 1.7882352941176474e-05, "loss": 1.7767, "step": 310 }, { "epoch": 0.03, "learning_rate": 1.7941176470588237e-05, "loss": 1.6429, "step": 311 }, { "epoch": 0.03, "learning_rate": 1.8e-05, "loss": 1.8337, "step": 312 }, { "epoch": 0.03, "learning_rate": 1.8058823529411767e-05, "loss": 1.8093, "step": 313 }, { "epoch": 0.03, "learning_rate": 1.811764705882353e-05, "loss": 1.6414, "step": 314 }, { "epoch": 0.03, "learning_rate": 1.8176470588235294e-05, "loss": 1.7883, "step": 315 }, { "epoch": 0.03, "learning_rate": 1.823529411764706e-05, "loss": 1.7759, "step": 316 }, { "epoch": 0.03, "learning_rate": 1.8294117647058824e-05, "loss": 1.6168, "step": 317 }, { "epoch": 0.03, "learning_rate": 1.8352941176470587e-05, "loss": 1.7067, "step": 318 }, { "epoch": 0.03, "learning_rate": 1.8411764705882354e-05, "loss": 1.7285, "step": 319 }, { "epoch": 0.03, "learning_rate": 1.847058823529412e-05, "loss": 1.8618, "step": 320 }, { "epoch": 0.03, "learning_rate": 1.8529411764705884e-05, "loss": 1.8168, "step": 321 }, { "epoch": 0.03, "learning_rate": 1.8588235294117647e-05, "loss": 1.585, "step": 322 }, { "epoch": 0.03, "learning_rate": 1.8647058823529414e-05, "loss": 1.7558, "step": 323 }, { "epoch": 0.03, "learning_rate": 1.8705882352941178e-05, "loss": 1.8046, "step": 324 }, { "epoch": 0.03, "learning_rate": 1.8764705882352944e-05, "loss": 1.6669, "step": 325 }, { "epoch": 0.03, "learning_rate": 1.8823529411764708e-05, "loss": 1.7708, "step": 326 }, { "epoch": 0.03, "learning_rate": 1.888235294117647e-05, "loss": 1.7561, "step": 327 }, { "epoch": 0.03, "learning_rate": 1.8941176470588238e-05, "loss": 1.8269, "step": 328 }, { "epoch": 0.03, "learning_rate": 1.9e-05, "loss": 1.7875, "step": 329 }, { "epoch": 0.03, "learning_rate": 1.9058823529411764e-05, "loss": 1.6255, "step": 330 }, { "epoch": 0.03, "learning_rate": 1.911764705882353e-05, "loss": 1.8127, "step": 331 }, { "epoch": 0.03, "learning_rate": 1.9176470588235298e-05, "loss": 1.7961, "step": 332 }, { "epoch": 0.03, "learning_rate": 1.923529411764706e-05, "loss": 1.5232, "step": 333 }, { "epoch": 0.03, "learning_rate": 1.9294117647058825e-05, "loss": 1.7592, "step": 334 }, { "epoch": 0.03, "learning_rate": 1.935294117647059e-05, "loss": 1.605, "step": 335 }, { "epoch": 0.03, "learning_rate": 1.9411764705882355e-05, "loss": 1.7067, "step": 336 }, { "epoch": 0.03, "learning_rate": 1.9470588235294118e-05, "loss": 1.7928, "step": 337 }, { "epoch": 0.03, "learning_rate": 1.9529411764705885e-05, "loss": 1.8148, "step": 338 }, { "epoch": 0.03, "learning_rate": 1.9588235294117648e-05, "loss": 1.6558, "step": 339 }, { "epoch": 0.03, "learning_rate": 1.964705882352941e-05, "loss": 1.6352, "step": 340 }, { "epoch": 0.03, "learning_rate": 1.9705882352941178e-05, "loss": 1.6702, "step": 341 }, { "epoch": 0.03, "learning_rate": 1.9764705882352945e-05, "loss": 1.6362, "step": 342 }, { "epoch": 0.03, "learning_rate": 1.9823529411764708e-05, "loss": 1.6018, "step": 343 }, { "epoch": 0.03, "learning_rate": 1.988235294117647e-05, "loss": 1.663, "step": 344 }, { "epoch": 0.03, "learning_rate": 1.9941176470588238e-05, "loss": 1.731, "step": 345 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 1.6835, "step": 346 }, { "epoch": 0.03, "learning_rate": 1.999999959097609e-05, "loss": 1.6298, "step": 347 }, { "epoch": 0.03, "learning_rate": 1.9999998363904388e-05, "loss": 1.709, "step": 348 }, { "epoch": 0.03, "learning_rate": 1.9999996318784994e-05, "loss": 1.7654, "step": 349 }, { "epoch": 0.03, "learning_rate": 1.9999993455618076e-05, "loss": 1.794, "step": 350 }, { "epoch": 0.03, "learning_rate": 1.9999989774403868e-05, "loss": 1.7347, "step": 351 }, { "epoch": 0.03, "learning_rate": 1.999998527514268e-05, "loss": 1.8049, "step": 352 }, { "epoch": 0.03, "learning_rate": 1.999997995783486e-05, "loss": 1.6482, "step": 353 }, { "epoch": 0.03, "learning_rate": 1.999997382248086e-05, "loss": 1.7947, "step": 354 }, { "epoch": 0.03, "learning_rate": 1.9999966869081183e-05, "loss": 1.8877, "step": 355 }, { "epoch": 0.03, "learning_rate": 1.9999959097636386e-05, "loss": 1.6704, "step": 356 }, { "epoch": 0.03, "learning_rate": 1.999995050814711e-05, "loss": 1.5764, "step": 357 }, { "epoch": 0.03, "learning_rate": 1.9999941100614063e-05, "loss": 1.809, "step": 358 }, { "epoch": 0.03, "learning_rate": 1.9999930875038008e-05, "loss": 1.7173, "step": 359 }, { "epoch": 0.03, "learning_rate": 1.9999919831419782e-05, "loss": 1.6244, "step": 360 }, { "epoch": 0.03, "learning_rate": 1.9999907969760287e-05, "loss": 1.7726, "step": 361 }, { "epoch": 0.03, "learning_rate": 1.99998952900605e-05, "loss": 1.7427, "step": 362 }, { "epoch": 0.03, "learning_rate": 1.9999881792321456e-05, "loss": 1.8286, "step": 363 }, { "epoch": 0.03, "learning_rate": 1.9999867476544254e-05, "loss": 1.7537, "step": 364 }, { "epoch": 0.03, "learning_rate": 1.999985234273007e-05, "loss": 1.6806, "step": 365 }, { "epoch": 0.03, "learning_rate": 1.999983639088014e-05, "loss": 1.856, "step": 366 }, { "epoch": 0.03, "learning_rate": 1.999981962099577e-05, "loss": 1.601, "step": 367 }, { "epoch": 0.03, "learning_rate": 1.9999802033078334e-05, "loss": 1.7786, "step": 368 }, { "epoch": 0.03, "learning_rate": 1.9999783627129267e-05, "loss": 1.7506, "step": 369 }, { "epoch": 0.03, "learning_rate": 1.9999764403150074e-05, "loss": 1.8417, "step": 370 }, { "epoch": 0.03, "learning_rate": 1.9999744361142327e-05, "loss": 1.7112, "step": 371 }, { "epoch": 0.03, "learning_rate": 1.9999723501107674e-05, "loss": 1.7228, "step": 372 }, { "epoch": 0.03, "learning_rate": 1.9999701823047814e-05, "loss": 1.7763, "step": 373 }, { "epoch": 0.03, "learning_rate": 1.999967932696452e-05, "loss": 1.8038, "step": 374 }, { "epoch": 0.03, "learning_rate": 1.9999656012859634e-05, "loss": 1.8009, "step": 375 }, { "epoch": 0.03, "learning_rate": 1.9999631880735066e-05, "loss": 1.59, "step": 376 }, { "epoch": 0.03, "learning_rate": 1.9999606930592785e-05, "loss": 1.8018, "step": 377 }, { "epoch": 0.03, "learning_rate": 1.9999581162434836e-05, "loss": 1.7858, "step": 378 }, { "epoch": 0.03, "learning_rate": 1.9999554576263326e-05, "loss": 1.7133, "step": 379 }, { "epoch": 0.03, "learning_rate": 1.999952717208043e-05, "loss": 1.7246, "step": 380 }, { "epoch": 0.03, "learning_rate": 1.9999498949888387e-05, "loss": 1.9036, "step": 381 }, { "epoch": 0.03, "learning_rate": 1.9999469909689505e-05, "loss": 1.724, "step": 382 }, { "epoch": 0.03, "learning_rate": 1.999944005148617e-05, "loss": 1.7641, "step": 383 }, { "epoch": 0.03, "learning_rate": 1.9999409375280812e-05, "loss": 1.6301, "step": 384 }, { "epoch": 0.03, "learning_rate": 1.9999377881075944e-05, "loss": 1.7611, "step": 385 }, { "epoch": 0.03, "learning_rate": 1.9999345568874148e-05, "loss": 1.7637, "step": 386 }, { "epoch": 0.03, "learning_rate": 1.999931243867806e-05, "loss": 1.8763, "step": 387 }, { "epoch": 0.03, "learning_rate": 1.9999278490490396e-05, "loss": 1.6765, "step": 388 }, { "epoch": 0.03, "learning_rate": 1.999924372431393e-05, "loss": 1.6147, "step": 389 }, { "epoch": 0.03, "learning_rate": 1.999920814015151e-05, "loss": 1.7381, "step": 390 }, { "epoch": 0.03, "learning_rate": 1.9999171738006038e-05, "loss": 1.7676, "step": 391 }, { "epoch": 0.03, "learning_rate": 1.99991345178805e-05, "loss": 1.6784, "step": 392 }, { "epoch": 0.03, "learning_rate": 1.9999096479777936e-05, "loss": 1.799, "step": 393 }, { "epoch": 0.03, "learning_rate": 1.9999057623701464e-05, "loss": 1.9104, "step": 394 }, { "epoch": 0.03, "learning_rate": 1.9999017949654254e-05, "loss": 1.7197, "step": 395 }, { "epoch": 0.03, "learning_rate": 1.9998977457639558e-05, "loss": 1.6644, "step": 396 }, { "epoch": 0.04, "learning_rate": 1.9998936147660687e-05, "loss": 1.8483, "step": 397 }, { "epoch": 0.04, "learning_rate": 1.9998894019721017e-05, "loss": 1.905, "step": 398 }, { "epoch": 0.04, "learning_rate": 1.9998851073824e-05, "loss": 1.7903, "step": 399 }, { "epoch": 0.04, "learning_rate": 1.9998807309973145e-05, "loss": 1.7973, "step": 400 }, { "epoch": 0.04, "learning_rate": 1.9998762728172034e-05, "loss": 1.7439, "step": 401 }, { "epoch": 0.04, "learning_rate": 1.9998717328424315e-05, "loss": 1.7145, "step": 402 }, { "epoch": 0.04, "learning_rate": 1.9998671110733697e-05, "loss": 1.6827, "step": 403 }, { "epoch": 0.04, "learning_rate": 1.9998624075103966e-05, "loss": 1.7397, "step": 404 }, { "epoch": 0.04, "learning_rate": 1.9998576221538967e-05, "loss": 1.6826, "step": 405 }, { "epoch": 0.04, "learning_rate": 1.9998527550042615e-05, "loss": 1.8806, "step": 406 }, { "epoch": 0.04, "learning_rate": 1.9998478060618894e-05, "loss": 1.7194, "step": 407 }, { "epoch": 0.04, "learning_rate": 1.999842775327185e-05, "loss": 1.7549, "step": 408 }, { "epoch": 0.04, "learning_rate": 1.9998376628005598e-05, "loss": 1.6952, "step": 409 }, { "epoch": 0.04, "learning_rate": 1.9998324684824322e-05, "loss": 1.8066, "step": 410 }, { "epoch": 0.04, "learning_rate": 1.999827192373227e-05, "loss": 1.5253, "step": 411 }, { "epoch": 0.04, "learning_rate": 1.9998218344733757e-05, "loss": 1.6773, "step": 412 }, { "epoch": 0.04, "learning_rate": 1.9998163947833168e-05, "loss": 1.7386, "step": 413 }, { "epoch": 0.04, "learning_rate": 1.9998108733034956e-05, "loss": 1.6878, "step": 414 }, { "epoch": 0.04, "learning_rate": 1.9998052700343632e-05, "loss": 1.7962, "step": 415 }, { "epoch": 0.04, "learning_rate": 1.9997995849763782e-05, "loss": 1.6081, "step": 416 }, { "epoch": 0.04, "learning_rate": 1.9997938181300055e-05, "loss": 1.6685, "step": 417 }, { "epoch": 0.04, "learning_rate": 1.9997879694957173e-05, "loss": 1.7464, "step": 418 }, { "epoch": 0.04, "learning_rate": 1.9997820390739915e-05, "loss": 1.7406, "step": 419 }, { "epoch": 0.04, "learning_rate": 1.9997760268653136e-05, "loss": 1.6809, "step": 420 }, { "epoch": 0.04, "learning_rate": 1.9997699328701754e-05, "loss": 1.7831, "step": 421 }, { "epoch": 0.04, "learning_rate": 1.9997637570890755e-05, "loss": 1.7175, "step": 422 }, { "epoch": 0.04, "learning_rate": 1.9997574995225185e-05, "loss": 1.8123, "step": 423 }, { "epoch": 0.04, "learning_rate": 1.9997511601710172e-05, "loss": 1.796, "step": 424 }, { "epoch": 0.04, "learning_rate": 1.9997447390350896e-05, "loss": 1.8005, "step": 425 }, { "epoch": 0.04, "learning_rate": 1.999738236115261e-05, "loss": 1.7869, "step": 426 }, { "epoch": 0.04, "learning_rate": 1.9997316514120636e-05, "loss": 1.8293, "step": 427 }, { "epoch": 0.04, "learning_rate": 1.9997249849260362e-05, "loss": 1.6663, "step": 428 }, { "epoch": 0.04, "learning_rate": 1.9997182366577235e-05, "loss": 1.7549, "step": 429 }, { "epoch": 0.04, "learning_rate": 1.9997114066076783e-05, "loss": 1.6802, "step": 430 }, { "epoch": 0.04, "learning_rate": 1.9997044947764584e-05, "loss": 1.7847, "step": 431 }, { "epoch": 0.04, "learning_rate": 1.9996975011646307e-05, "loss": 1.7504, "step": 432 }, { "epoch": 0.04, "learning_rate": 1.999690425772766e-05, "loss": 1.7082, "step": 433 }, { "epoch": 0.04, "learning_rate": 1.9996832686014432e-05, "loss": 1.5792, "step": 434 }, { "epoch": 0.04, "learning_rate": 1.9996760296512484e-05, "loss": 1.6767, "step": 435 }, { "epoch": 0.04, "learning_rate": 1.9996687089227734e-05, "loss": 1.5983, "step": 436 }, { "epoch": 0.04, "learning_rate": 1.999661306416617e-05, "loss": 1.7717, "step": 437 }, { "epoch": 0.04, "learning_rate": 1.999653822133385e-05, "loss": 1.788, "step": 438 }, { "epoch": 0.04, "learning_rate": 1.99964625607369e-05, "loss": 1.6891, "step": 439 }, { "epoch": 0.04, "learning_rate": 1.9996386082381502e-05, "loss": 1.6956, "step": 440 }, { "epoch": 0.04, "learning_rate": 1.9996308786273914e-05, "loss": 1.7415, "step": 441 }, { "epoch": 0.04, "learning_rate": 1.999623067242046e-05, "loss": 1.6741, "step": 442 }, { "epoch": 0.04, "learning_rate": 1.9996151740827534e-05, "loss": 1.7461, "step": 443 }, { "epoch": 0.04, "learning_rate": 1.999607199150159e-05, "loss": 1.7167, "step": 444 }, { "epoch": 0.04, "learning_rate": 1.9995991424449147e-05, "loss": 1.6863, "step": 445 }, { "epoch": 0.04, "learning_rate": 1.9995910039676802e-05, "loss": 1.856, "step": 446 }, { "epoch": 0.04, "learning_rate": 1.999582783719121e-05, "loss": 1.5794, "step": 447 }, { "epoch": 0.04, "learning_rate": 1.99957448169991e-05, "loss": 1.6738, "step": 448 }, { "epoch": 0.04, "learning_rate": 1.9995660979107257e-05, "loss": 1.7672, "step": 449 }, { "epoch": 0.04, "learning_rate": 1.9995576323522547e-05, "loss": 1.5931, "step": 450 }, { "epoch": 0.04, "learning_rate": 1.9995490850251884e-05, "loss": 1.6992, "step": 451 }, { "epoch": 0.04, "learning_rate": 1.9995404559302268e-05, "loss": 1.7247, "step": 452 }, { "epoch": 0.04, "learning_rate": 1.999531745068076e-05, "loss": 1.69, "step": 453 }, { "epoch": 0.04, "learning_rate": 1.999522952439448e-05, "loss": 1.7269, "step": 454 }, { "epoch": 0.04, "learning_rate": 1.9995140780450622e-05, "loss": 1.8018, "step": 455 }, { "epoch": 0.04, "learning_rate": 1.9995051218856448e-05, "loss": 1.7617, "step": 456 }, { "epoch": 0.04, "learning_rate": 1.9994960839619283e-05, "loss": 1.7335, "step": 457 }, { "epoch": 0.04, "learning_rate": 1.9994869642746523e-05, "loss": 1.7295, "step": 458 }, { "epoch": 0.04, "learning_rate": 1.999477762824563e-05, "loss": 1.7897, "step": 459 }, { "epoch": 0.04, "learning_rate": 1.999468479612412e-05, "loss": 1.7575, "step": 460 }, { "epoch": 0.04, "learning_rate": 1.99945911463896e-05, "loss": 1.749, "step": 461 }, { "epoch": 0.04, "learning_rate": 1.999449667904972e-05, "loss": 1.6572, "step": 462 }, { "epoch": 0.04, "learning_rate": 1.999440139411222e-05, "loss": 1.7904, "step": 463 }, { "epoch": 0.04, "learning_rate": 1.9994305291584885e-05, "loss": 1.6993, "step": 464 }, { "epoch": 0.04, "learning_rate": 1.999420837147558e-05, "loss": 1.5911, "step": 465 }, { "epoch": 0.04, "learning_rate": 1.9994110633792236e-05, "loss": 1.7075, "step": 466 }, { "epoch": 0.04, "learning_rate": 1.9994012078542843e-05, "loss": 1.7536, "step": 467 }, { "epoch": 0.04, "learning_rate": 1.9993912705735466e-05, "loss": 1.5574, "step": 468 }, { "epoch": 0.04, "learning_rate": 1.9993812515378238e-05, "loss": 1.7151, "step": 469 }, { "epoch": 0.04, "learning_rate": 1.9993711507479348e-05, "loss": 1.8526, "step": 470 }, { "epoch": 0.04, "learning_rate": 1.9993609682047062e-05, "loss": 1.6143, "step": 471 }, { "epoch": 0.04, "learning_rate": 1.9993507039089713e-05, "loss": 1.6404, "step": 472 }, { "epoch": 0.04, "learning_rate": 1.9993403578615693e-05, "loss": 1.5782, "step": 473 }, { "epoch": 0.04, "learning_rate": 1.9993299300633468e-05, "loss": 1.8153, "step": 474 }, { "epoch": 0.04, "learning_rate": 1.999319420515157e-05, "loss": 1.6097, "step": 475 }, { "epoch": 0.04, "learning_rate": 1.999308829217859e-05, "loss": 1.7078, "step": 476 }, { "epoch": 0.04, "learning_rate": 1.99929815617232e-05, "loss": 1.8618, "step": 477 }, { "epoch": 0.04, "learning_rate": 1.9992874013794123e-05, "loss": 1.7151, "step": 478 }, { "epoch": 0.04, "learning_rate": 1.9992765648400164e-05, "loss": 1.6612, "step": 479 }, { "epoch": 0.04, "learning_rate": 1.9992656465550184e-05, "loss": 1.8678, "step": 480 }, { "epoch": 0.04, "learning_rate": 1.999254646525312e-05, "loss": 1.7804, "step": 481 }, { "epoch": 0.04, "learning_rate": 1.9992435647517963e-05, "loss": 1.6065, "step": 482 }, { "epoch": 0.04, "learning_rate": 1.999232401235378e-05, "loss": 1.6964, "step": 483 }, { "epoch": 0.04, "learning_rate": 1.999221155976971e-05, "loss": 1.5443, "step": 484 }, { "epoch": 0.04, "learning_rate": 1.9992098289774944e-05, "loss": 1.626, "step": 485 }, { "epoch": 0.04, "learning_rate": 1.999198420237875e-05, "loss": 1.6804, "step": 486 }, { "epoch": 0.04, "learning_rate": 1.9991869297590467e-05, "loss": 1.705, "step": 487 }, { "epoch": 0.04, "learning_rate": 1.9991753575419485e-05, "loss": 1.7655, "step": 488 }, { "epoch": 0.04, "learning_rate": 1.999163703587528e-05, "loss": 1.7009, "step": 489 }, { "epoch": 0.04, "learning_rate": 1.9991519678967376e-05, "loss": 1.7248, "step": 490 }, { "epoch": 0.04, "learning_rate": 1.999140150470538e-05, "loss": 1.8169, "step": 491 }, { "epoch": 0.04, "learning_rate": 1.999128251309896e-05, "loss": 1.8236, "step": 492 }, { "epoch": 0.04, "learning_rate": 1.9991162704157847e-05, "loss": 1.6331, "step": 493 }, { "epoch": 0.04, "learning_rate": 1.9991042077891842e-05, "loss": 1.7346, "step": 494 }, { "epoch": 0.04, "learning_rate": 1.9990920634310814e-05, "loss": 1.7176, "step": 495 }, { "epoch": 0.04, "learning_rate": 1.9990798373424696e-05, "loss": 1.7984, "step": 496 }, { "epoch": 0.04, "learning_rate": 1.999067529524349e-05, "loss": 1.8326, "step": 497 }, { "epoch": 0.04, "learning_rate": 1.9990551399777268e-05, "loss": 1.6732, "step": 498 }, { "epoch": 0.04, "learning_rate": 1.999042668703616e-05, "loss": 1.7063, "step": 499 }, { "epoch": 0.04, "learning_rate": 1.999030115703037e-05, "loss": 1.7056, "step": 500 }, { "epoch": 0.04, "learning_rate": 1.9990174809770168e-05, "loss": 1.7776, "step": 501 }, { "epoch": 0.04, "learning_rate": 1.999004764526589e-05, "loss": 1.785, "step": 502 }, { "epoch": 0.04, "learning_rate": 1.998991966352794e-05, "loss": 1.6077, "step": 503 }, { "epoch": 0.04, "learning_rate": 1.9989790864566782e-05, "loss": 1.7705, "step": 504 }, { "epoch": 0.04, "learning_rate": 1.9989661248392957e-05, "loss": 1.5893, "step": 505 }, { "epoch": 0.04, "learning_rate": 1.9989530815017068e-05, "loss": 1.6323, "step": 506 }, { "epoch": 0.04, "learning_rate": 1.9989399564449782e-05, "loss": 1.8712, "step": 507 }, { "epoch": 0.04, "learning_rate": 1.9989267496701836e-05, "loss": 1.7619, "step": 508 }, { "epoch": 0.04, "learning_rate": 1.998913461178404e-05, "loss": 1.7071, "step": 509 }, { "epoch": 0.05, "learning_rate": 1.9989000909707257e-05, "loss": 1.7012, "step": 510 }, { "epoch": 0.05, "learning_rate": 1.9988866390482427e-05, "loss": 1.688, "step": 511 }, { "epoch": 0.05, "learning_rate": 1.998873105412056e-05, "loss": 1.8532, "step": 512 }, { "epoch": 0.05, "learning_rate": 1.998859490063272e-05, "loss": 1.6968, "step": 513 }, { "epoch": 0.05, "learning_rate": 1.9988457930030046e-05, "loss": 1.7395, "step": 514 }, { "epoch": 0.05, "learning_rate": 1.9988320142323742e-05, "loss": 1.6721, "step": 515 }, { "epoch": 0.05, "learning_rate": 1.9988181537525086e-05, "loss": 1.7051, "step": 516 }, { "epoch": 0.05, "learning_rate": 1.9988042115645412e-05, "loss": 1.6107, "step": 517 }, { "epoch": 0.05, "learning_rate": 1.9987901876696124e-05, "loss": 1.5935, "step": 518 }, { "epoch": 0.05, "learning_rate": 1.9987760820688697e-05, "loss": 1.7454, "step": 519 }, { "epoch": 0.05, "learning_rate": 1.9987618947634672e-05, "loss": 1.7007, "step": 520 }, { "epoch": 0.05, "learning_rate": 1.9987476257545645e-05, "loss": 1.6582, "step": 521 }, { "epoch": 0.05, "learning_rate": 1.99873327504333e-05, "loss": 1.7397, "step": 522 }, { "epoch": 0.05, "learning_rate": 1.9987188426309373e-05, "loss": 1.6816, "step": 523 }, { "epoch": 0.05, "learning_rate": 1.9987043285185668e-05, "loss": 1.806, "step": 524 }, { "epoch": 0.05, "learning_rate": 1.9986897327074056e-05, "loss": 1.7432, "step": 525 }, { "epoch": 0.05, "learning_rate": 1.9986750551986486e-05, "loss": 1.7233, "step": 526 }, { "epoch": 0.05, "learning_rate": 1.998660295993496e-05, "loss": 1.7281, "step": 527 }, { "epoch": 0.05, "learning_rate": 1.998645455093155e-05, "loss": 1.6617, "step": 528 }, { "epoch": 0.05, "learning_rate": 1.9986305324988397e-05, "loss": 1.8334, "step": 529 }, { "epoch": 0.05, "learning_rate": 1.9986155282117712e-05, "loss": 1.6351, "step": 530 }, { "epoch": 0.05, "learning_rate": 1.9986004422331765e-05, "loss": 1.598, "step": 531 }, { "epoch": 0.05, "learning_rate": 1.99858527456429e-05, "loss": 1.7319, "step": 532 }, { "epoch": 0.05, "learning_rate": 1.998570025206352e-05, "loss": 1.6817, "step": 533 }, { "epoch": 0.05, "learning_rate": 1.998554694160611e-05, "loss": 1.6216, "step": 534 }, { "epoch": 0.05, "learning_rate": 1.99853928142832e-05, "loss": 1.6777, "step": 535 }, { "epoch": 0.05, "learning_rate": 1.9985237870107406e-05, "loss": 1.5826, "step": 536 }, { "epoch": 0.05, "learning_rate": 1.99850821090914e-05, "loss": 1.782, "step": 537 }, { "epoch": 0.05, "learning_rate": 1.9984925531247922e-05, "loss": 1.729, "step": 538 }, { "epoch": 0.05, "learning_rate": 1.9984768136589783e-05, "loss": 1.8016, "step": 539 }, { "epoch": 0.05, "learning_rate": 1.9984609925129862e-05, "loss": 1.6273, "step": 540 }, { "epoch": 0.05, "learning_rate": 1.99844508968811e-05, "loss": 1.7393, "step": 541 }, { "epoch": 0.05, "learning_rate": 1.99842910518565e-05, "loss": 1.6582, "step": 542 }, { "epoch": 0.05, "learning_rate": 1.998413039006914e-05, "loss": 1.7326, "step": 543 }, { "epoch": 0.05, "learning_rate": 1.998396891153217e-05, "loss": 1.6471, "step": 544 }, { "epoch": 0.05, "learning_rate": 1.9983806616258797e-05, "loss": 1.7829, "step": 545 }, { "epoch": 0.05, "learning_rate": 1.9983643504262292e-05, "loss": 1.6677, "step": 546 }, { "epoch": 0.05, "learning_rate": 1.9983479575556006e-05, "loss": 1.5984, "step": 547 }, { "epoch": 0.05, "learning_rate": 1.9983314830153343e-05, "loss": 1.6699, "step": 548 }, { "epoch": 0.05, "learning_rate": 1.9983149268067784e-05, "loss": 1.6179, "step": 549 }, { "epoch": 0.05, "learning_rate": 1.9982982889312867e-05, "loss": 1.8063, "step": 550 }, { "epoch": 0.05, "learning_rate": 1.998281569390221e-05, "loss": 1.8534, "step": 551 }, { "epoch": 0.05, "learning_rate": 1.9982647681849487e-05, "loss": 1.606, "step": 552 }, { "epoch": 0.05, "learning_rate": 1.998247885316844e-05, "loss": 1.7163, "step": 553 }, { "epoch": 0.05, "learning_rate": 1.9982309207872885e-05, "loss": 1.6835, "step": 554 }, { "epoch": 0.05, "learning_rate": 1.99821387459767e-05, "loss": 1.6474, "step": 555 }, { "epoch": 0.05, "learning_rate": 1.998196746749382e-05, "loss": 1.5789, "step": 556 }, { "epoch": 0.05, "learning_rate": 1.9981795372438265e-05, "loss": 1.726, "step": 557 }, { "epoch": 0.05, "learning_rate": 1.998162246082411e-05, "loss": 1.7356, "step": 558 }, { "epoch": 0.05, "learning_rate": 1.99814487326655e-05, "loss": 1.6233, "step": 559 }, { "epoch": 0.05, "learning_rate": 1.9981274187976652e-05, "loss": 1.7018, "step": 560 }, { "epoch": 0.05, "learning_rate": 1.9981098826771843e-05, "loss": 1.6933, "step": 561 }, { "epoch": 0.05, "learning_rate": 1.998092264906541e-05, "loss": 1.5709, "step": 562 }, { "epoch": 0.05, "learning_rate": 1.9980745654871776e-05, "loss": 1.638, "step": 563 }, { "epoch": 0.05, "learning_rate": 1.9980567844205412e-05, "loss": 1.665, "step": 564 }, { "epoch": 0.05, "learning_rate": 1.9980389217080866e-05, "loss": 1.5793, "step": 565 }, { "epoch": 0.05, "learning_rate": 1.9980209773512754e-05, "loss": 1.7, "step": 566 }, { "epoch": 0.05, "learning_rate": 1.998002951351575e-05, "loss": 1.6557, "step": 567 }, { "epoch": 0.05, "learning_rate": 1.9979848437104606e-05, "loss": 1.72, "step": 568 }, { "epoch": 0.05, "learning_rate": 1.997966654429413e-05, "loss": 1.6481, "step": 569 }, { "epoch": 0.05, "learning_rate": 1.9979483835099206e-05, "loss": 1.5901, "step": 570 }, { "epoch": 0.05, "learning_rate": 1.9979300309534774e-05, "loss": 1.7717, "step": 571 }, { "epoch": 0.05, "learning_rate": 1.9979115967615854e-05, "loss": 1.8542, "step": 572 }, { "epoch": 0.05, "learning_rate": 1.9978930809357523e-05, "loss": 1.6269, "step": 573 }, { "epoch": 0.05, "learning_rate": 1.9978744834774926e-05, "loss": 1.7523, "step": 574 }, { "epoch": 0.05, "learning_rate": 1.997855804388328e-05, "loss": 1.5917, "step": 575 }, { "epoch": 0.05, "learning_rate": 1.9978370436697868e-05, "loss": 1.5745, "step": 576 }, { "epoch": 0.05, "learning_rate": 1.9978182013234028e-05, "loss": 1.5683, "step": 577 }, { "epoch": 0.05, "learning_rate": 1.9977992773507183e-05, "loss": 1.7107, "step": 578 }, { "epoch": 0.05, "learning_rate": 1.9977802717532807e-05, "loss": 1.7573, "step": 579 }, { "epoch": 0.05, "learning_rate": 1.997761184532645e-05, "loss": 1.7078, "step": 580 }, { "epoch": 0.05, "learning_rate": 1.997742015690373e-05, "loss": 1.6691, "step": 581 }, { "epoch": 0.05, "learning_rate": 1.9977227652280323e-05, "loss": 1.6576, "step": 582 }, { "epoch": 0.05, "learning_rate": 1.997703433147198e-05, "loss": 1.6566, "step": 583 }, { "epoch": 0.05, "learning_rate": 1.997684019449451e-05, "loss": 1.7468, "step": 584 }, { "epoch": 0.05, "learning_rate": 1.9976645241363804e-05, "loss": 1.6503, "step": 585 }, { "epoch": 0.05, "learning_rate": 1.99764494720958e-05, "loss": 1.7132, "step": 586 }, { "epoch": 0.05, "learning_rate": 1.997625288670652e-05, "loss": 1.6236, "step": 587 }, { "epoch": 0.05, "learning_rate": 1.9976055485212043e-05, "loss": 1.6542, "step": 588 }, { "epoch": 0.05, "learning_rate": 1.9975857267628515e-05, "loss": 1.7549, "step": 589 }, { "epoch": 0.05, "learning_rate": 1.9975658233972158e-05, "loss": 1.6929, "step": 590 }, { "epoch": 0.05, "learning_rate": 1.9975458384259245e-05, "loss": 1.7554, "step": 591 }, { "epoch": 0.05, "learning_rate": 1.997525771850613e-05, "loss": 1.7939, "step": 592 }, { "epoch": 0.05, "learning_rate": 1.997505623672923e-05, "loss": 1.7701, "step": 593 }, { "epoch": 0.05, "learning_rate": 1.9974853938945023e-05, "loss": 1.7059, "step": 594 }, { "epoch": 0.05, "learning_rate": 1.9974650825170058e-05, "loss": 1.5971, "step": 595 }, { "epoch": 0.05, "learning_rate": 1.9974446895420957e-05, "loss": 1.5631, "step": 596 }, { "epoch": 0.05, "learning_rate": 1.9974242149714392e-05, "loss": 1.71, "step": 597 }, { "epoch": 0.05, "learning_rate": 1.9974036588067122e-05, "loss": 1.6442, "step": 598 }, { "epoch": 0.05, "learning_rate": 1.997383021049596e-05, "loss": 1.6441, "step": 599 }, { "epoch": 0.05, "learning_rate": 1.9973623017017784e-05, "loss": 1.6484, "step": 600 }, { "epoch": 0.05, "learning_rate": 1.997341500764955e-05, "loss": 1.8247, "step": 601 }, { "epoch": 0.05, "learning_rate": 1.9973206182408268e-05, "loss": 1.6509, "step": 602 }, { "epoch": 0.05, "learning_rate": 1.9972996541311024e-05, "loss": 1.6845, "step": 603 }, { "epoch": 0.05, "learning_rate": 1.9972786084374973e-05, "loss": 1.6624, "step": 604 }, { "epoch": 0.05, "learning_rate": 1.997257481161732e-05, "loss": 1.7201, "step": 605 }, { "epoch": 0.05, "learning_rate": 1.9972362723055355e-05, "loss": 1.6387, "step": 606 }, { "epoch": 0.05, "learning_rate": 1.9972149818706426e-05, "loss": 1.8114, "step": 607 }, { "epoch": 0.05, "learning_rate": 1.9971936098587957e-05, "loss": 1.7834, "step": 608 }, { "epoch": 0.05, "learning_rate": 1.997172156271742e-05, "loss": 1.6528, "step": 609 }, { "epoch": 0.05, "learning_rate": 1.997150621111237e-05, "loss": 1.5522, "step": 610 }, { "epoch": 0.05, "learning_rate": 1.9971290043790427e-05, "loss": 1.6249, "step": 611 }, { "epoch": 0.05, "learning_rate": 1.997107306076927e-05, "loss": 1.6368, "step": 612 }, { "epoch": 0.05, "learning_rate": 1.997085526206665e-05, "loss": 1.588, "step": 613 }, { "epoch": 0.05, "learning_rate": 1.9970636647700382e-05, "loss": 1.7188, "step": 614 }, { "epoch": 0.05, "learning_rate": 1.9970417217688353e-05, "loss": 1.6705, "step": 615 }, { "epoch": 0.05, "learning_rate": 1.9970196972048515e-05, "loss": 1.7336, "step": 616 }, { "epoch": 0.05, "learning_rate": 1.9969975910798884e-05, "loss": 1.7185, "step": 617 }, { "epoch": 0.05, "learning_rate": 1.9969754033957538e-05, "loss": 1.6426, "step": 618 }, { "epoch": 0.05, "learning_rate": 1.9969531341542635e-05, "loss": 1.6413, "step": 619 }, { "epoch": 0.05, "learning_rate": 1.9969307833572392e-05, "loss": 1.7157, "step": 620 }, { "epoch": 0.05, "learning_rate": 1.9969083510065086e-05, "loss": 1.5556, "step": 621 }, { "epoch": 0.05, "learning_rate": 1.9968858371039076e-05, "loss": 1.7554, "step": 622 }, { "epoch": 0.06, "learning_rate": 1.9968632416512773e-05, "loss": 1.8446, "step": 623 }, { "epoch": 0.06, "learning_rate": 1.9968405646504664e-05, "loss": 1.6297, "step": 624 }, { "epoch": 0.06, "learning_rate": 1.99681780610333e-05, "loss": 1.6703, "step": 625 }, { "epoch": 0.06, "learning_rate": 1.9967949660117302e-05, "loss": 1.6758, "step": 626 }, { "epoch": 0.06, "learning_rate": 1.9967720443775347e-05, "loss": 1.6613, "step": 627 }, { "epoch": 0.06, "learning_rate": 1.996749041202619e-05, "loss": 1.694, "step": 628 }, { "epoch": 0.06, "learning_rate": 1.9967259564888646e-05, "loss": 1.6763, "step": 629 }, { "epoch": 0.06, "learning_rate": 1.9967027902381608e-05, "loss": 1.5985, "step": 630 }, { "epoch": 0.06, "learning_rate": 1.9966795424524016e-05, "loss": 1.7343, "step": 631 }, { "epoch": 0.06, "learning_rate": 1.9966562131334897e-05, "loss": 1.7285, "step": 632 }, { "epoch": 0.06, "learning_rate": 1.9966328022833327e-05, "loss": 1.6728, "step": 633 }, { "epoch": 0.06, "learning_rate": 1.9966093099038463e-05, "loss": 1.5664, "step": 634 }, { "epoch": 0.06, "learning_rate": 1.996585735996952e-05, "loss": 1.6224, "step": 635 }, { "epoch": 0.06, "learning_rate": 1.9965620805645788e-05, "loss": 1.6544, "step": 636 }, { "epoch": 0.06, "learning_rate": 1.996538343608661e-05, "loss": 1.6424, "step": 637 }, { "epoch": 0.06, "learning_rate": 1.996514525131141e-05, "loss": 1.7161, "step": 638 }, { "epoch": 0.06, "learning_rate": 1.9964906251339672e-05, "loss": 1.7598, "step": 639 }, { "epoch": 0.06, "learning_rate": 1.9964666436190942e-05, "loss": 1.623, "step": 640 }, { "epoch": 0.06, "learning_rate": 1.9964425805884845e-05, "loss": 1.6503, "step": 641 }, { "epoch": 0.06, "learning_rate": 1.996418436044106e-05, "loss": 1.6574, "step": 642 }, { "epoch": 0.06, "learning_rate": 1.9963942099879342e-05, "loss": 1.7469, "step": 643 }, { "epoch": 0.06, "learning_rate": 1.9963699024219512e-05, "loss": 1.6658, "step": 644 }, { "epoch": 0.06, "learning_rate": 1.996345513348145e-05, "loss": 1.5724, "step": 645 }, { "epoch": 0.06, "learning_rate": 1.9963210427685105e-05, "loss": 1.7084, "step": 646 }, { "epoch": 0.06, "learning_rate": 1.9962964906850502e-05, "loss": 1.7312, "step": 647 }, { "epoch": 0.06, "learning_rate": 1.9962718570997717e-05, "loss": 1.5772, "step": 648 }, { "epoch": 0.06, "learning_rate": 1.996247142014691e-05, "loss": 1.7042, "step": 649 }, { "epoch": 0.06, "learning_rate": 1.9962223454318298e-05, "loss": 1.6898, "step": 650 }, { "epoch": 0.06, "learning_rate": 1.996197467353216e-05, "loss": 1.8022, "step": 651 }, { "epoch": 0.06, "learning_rate": 1.9961725077808857e-05, "loss": 1.678, "step": 652 }, { "epoch": 0.06, "learning_rate": 1.99614746671688e-05, "loss": 1.6902, "step": 653 }, { "epoch": 0.06, "learning_rate": 1.9961223441632473e-05, "loss": 1.7637, "step": 654 }, { "epoch": 0.06, "learning_rate": 1.9960971401220427e-05, "loss": 1.5835, "step": 655 }, { "epoch": 0.06, "learning_rate": 1.9960718545953286e-05, "loss": 1.6549, "step": 656 }, { "epoch": 0.06, "learning_rate": 1.996046487585173e-05, "loss": 1.859, "step": 657 }, { "epoch": 0.06, "learning_rate": 1.9960210390936513e-05, "loss": 1.8011, "step": 658 }, { "epoch": 0.06, "learning_rate": 1.9959955091228453e-05, "loss": 1.6664, "step": 659 }, { "epoch": 0.06, "learning_rate": 1.9959698976748426e-05, "loss": 1.7014, "step": 660 }, { "epoch": 0.06, "learning_rate": 1.99594420475174e-05, "loss": 1.7006, "step": 661 }, { "epoch": 0.06, "learning_rate": 1.995918430355638e-05, "loss": 1.6732, "step": 662 }, { "epoch": 0.06, "learning_rate": 1.9958925744886457e-05, "loss": 1.6764, "step": 663 }, { "epoch": 0.06, "learning_rate": 1.9958666371528776e-05, "loss": 1.6921, "step": 664 }, { "epoch": 0.06, "learning_rate": 1.9958406183504563e-05, "loss": 1.7047, "step": 665 }, { "epoch": 0.06, "learning_rate": 1.9958145180835098e-05, "loss": 1.5729, "step": 666 }, { "epoch": 0.06, "learning_rate": 1.995788336354173e-05, "loss": 1.5842, "step": 667 }, { "epoch": 0.06, "learning_rate": 1.9957620731645882e-05, "loss": 1.6206, "step": 668 }, { "epoch": 0.06, "learning_rate": 1.995735728516904e-05, "loss": 1.6335, "step": 669 }, { "epoch": 0.06, "learning_rate": 1.9957093024132748e-05, "loss": 1.6971, "step": 670 }, { "epoch": 0.06, "learning_rate": 1.9956827948558627e-05, "loss": 1.6988, "step": 671 }, { "epoch": 0.06, "learning_rate": 1.9956562058468364e-05, "loss": 1.5884, "step": 672 }, { "epoch": 0.06, "learning_rate": 1.9956295353883708e-05, "loss": 1.7587, "step": 673 }, { "epoch": 0.06, "learning_rate": 1.9956027834826475e-05, "loss": 1.7491, "step": 674 }, { "epoch": 0.06, "learning_rate": 1.995575950131855e-05, "loss": 1.6145, "step": 675 }, { "epoch": 0.06, "learning_rate": 1.9955490353381886e-05, "loss": 1.6843, "step": 676 }, { "epoch": 0.06, "learning_rate": 1.9955220391038503e-05, "loss": 1.6026, "step": 677 }, { "epoch": 0.06, "learning_rate": 1.9954949614310477e-05, "loss": 1.6138, "step": 678 }, { "epoch": 0.06, "learning_rate": 1.9954678023219965e-05, "loss": 1.7048, "step": 679 }, { "epoch": 0.06, "learning_rate": 1.9954405617789183e-05, "loss": 1.6691, "step": 680 }, { "epoch": 0.06, "learning_rate": 1.9954132398040416e-05, "loss": 1.7233, "step": 681 }, { "epoch": 0.06, "learning_rate": 1.9953858363996015e-05, "loss": 1.7887, "step": 682 }, { "epoch": 0.06, "learning_rate": 1.9953583515678393e-05, "loss": 1.6657, "step": 683 }, { "epoch": 0.06, "learning_rate": 1.995330785311004e-05, "loss": 1.7538, "step": 684 }, { "epoch": 0.06, "learning_rate": 1.9953031376313498e-05, "loss": 1.6549, "step": 685 }, { "epoch": 0.06, "learning_rate": 1.9952754085311397e-05, "loss": 1.7633, "step": 686 }, { "epoch": 0.06, "learning_rate": 1.995247598012641e-05, "loss": 1.5842, "step": 687 }, { "epoch": 0.06, "learning_rate": 1.995219706078129e-05, "loss": 1.6167, "step": 688 }, { "epoch": 0.06, "learning_rate": 1.9951917327298855e-05, "loss": 1.6931, "step": 689 }, { "epoch": 0.06, "learning_rate": 1.995163677970199e-05, "loss": 1.663, "step": 690 }, { "epoch": 0.06, "learning_rate": 1.995135541801364e-05, "loss": 1.5378, "step": 691 }, { "epoch": 0.06, "learning_rate": 1.9951073242256826e-05, "loss": 1.6987, "step": 692 }, { "epoch": 0.06, "learning_rate": 1.9950790252454634e-05, "loss": 1.7273, "step": 693 }, { "epoch": 0.06, "learning_rate": 1.9950506448630206e-05, "loss": 1.764, "step": 694 }, { "epoch": 0.06, "learning_rate": 1.9950221830806766e-05, "loss": 1.6179, "step": 695 }, { "epoch": 0.06, "learning_rate": 1.994993639900759e-05, "loss": 1.7453, "step": 696 }, { "epoch": 0.06, "learning_rate": 1.9949650153256032e-05, "loss": 1.6725, "step": 697 }, { "epoch": 0.06, "learning_rate": 1.9949363093575512e-05, "loss": 1.7908, "step": 698 }, { "epoch": 0.06, "learning_rate": 1.9949075219989503e-05, "loss": 1.6694, "step": 699 }, { "epoch": 0.06, "learning_rate": 1.9948786532521567e-05, "loss": 1.6546, "step": 700 }, { "epoch": 0.06, "learning_rate": 1.9948497031195308e-05, "loss": 1.597, "step": 701 }, { "epoch": 0.06, "learning_rate": 1.9948206716034416e-05, "loss": 1.6485, "step": 702 }, { "epoch": 0.06, "learning_rate": 1.994791558706264e-05, "loss": 1.7308, "step": 703 }, { "epoch": 0.06, "learning_rate": 1.9947623644303792e-05, "loss": 1.571, "step": 704 }, { "epoch": 0.06, "learning_rate": 1.9947330887781754e-05, "loss": 1.6692, "step": 705 }, { "epoch": 0.06, "learning_rate": 1.9947037317520473e-05, "loss": 1.665, "step": 706 }, { "epoch": 0.06, "learning_rate": 1.9946742933543978e-05, "loss": 1.702, "step": 707 }, { "epoch": 0.06, "learning_rate": 1.9946447735876334e-05, "loss": 1.5466, "step": 708 }, { "epoch": 0.06, "learning_rate": 1.99461517245417e-05, "loss": 1.6145, "step": 709 }, { "epoch": 0.06, "learning_rate": 1.9945854899564286e-05, "loss": 1.5716, "step": 710 }, { "epoch": 0.06, "learning_rate": 1.994555726096838e-05, "loss": 1.6951, "step": 711 }, { "epoch": 0.06, "learning_rate": 1.9945258808778322e-05, "loss": 1.6438, "step": 712 }, { "epoch": 0.06, "learning_rate": 1.9944959543018536e-05, "loss": 1.6881, "step": 713 }, { "epoch": 0.06, "learning_rate": 1.9944659463713493e-05, "loss": 1.6211, "step": 714 }, { "epoch": 0.06, "learning_rate": 1.9944358570887745e-05, "loss": 1.756, "step": 715 }, { "epoch": 0.06, "learning_rate": 1.9944056864565913e-05, "loss": 1.7082, "step": 716 }, { "epoch": 0.06, "learning_rate": 1.9943754344772667e-05, "loss": 1.7092, "step": 717 }, { "epoch": 0.06, "learning_rate": 1.9943451011532763e-05, "loss": 1.5085, "step": 718 }, { "epoch": 0.06, "learning_rate": 1.9943146864871012e-05, "loss": 1.7132, "step": 719 }, { "epoch": 0.06, "learning_rate": 1.9942841904812297e-05, "loss": 1.6353, "step": 720 }, { "epoch": 0.06, "learning_rate": 1.994253613138156e-05, "loss": 1.7058, "step": 721 }, { "epoch": 0.06, "learning_rate": 1.994222954460382e-05, "loss": 1.73, "step": 722 }, { "epoch": 0.06, "learning_rate": 1.9941922144504153e-05, "loss": 1.6068, "step": 723 }, { "epoch": 0.06, "learning_rate": 1.994161393110771e-05, "loss": 1.7487, "step": 724 }, { "epoch": 0.06, "learning_rate": 1.9941304904439703e-05, "loss": 1.7303, "step": 725 }, { "epoch": 0.06, "learning_rate": 1.994099506452541e-05, "loss": 1.6055, "step": 726 }, { "epoch": 0.06, "learning_rate": 1.9940684411390177e-05, "loss": 1.5497, "step": 727 }, { "epoch": 0.06, "learning_rate": 1.994037294505942e-05, "loss": 1.6642, "step": 728 }, { "epoch": 0.06, "learning_rate": 1.9940060665558615e-05, "loss": 1.7244, "step": 729 }, { "epoch": 0.06, "learning_rate": 1.9939747572913312e-05, "loss": 1.6295, "step": 730 }, { "epoch": 0.06, "learning_rate": 1.993943366714912e-05, "loss": 1.7389, "step": 731 }, { "epoch": 0.06, "learning_rate": 1.993911894829172e-05, "loss": 1.7947, "step": 732 }, { "epoch": 0.06, "learning_rate": 1.9938803416366856e-05, "loss": 1.7441, "step": 733 }, { "epoch": 0.06, "learning_rate": 1.9938487071400344e-05, "loss": 1.6723, "step": 734 }, { "epoch": 0.06, "learning_rate": 1.9938169913418054e-05, "loss": 1.7572, "step": 735 }, { "epoch": 0.06, "learning_rate": 1.9937851942445944e-05, "loss": 1.6934, "step": 736 }, { "epoch": 0.07, "learning_rate": 1.993753315851001e-05, "loss": 1.7943, "step": 737 }, { "epoch": 0.07, "learning_rate": 1.993721356163634e-05, "loss": 1.6555, "step": 738 }, { "epoch": 0.07, "learning_rate": 1.993689315185108e-05, "loss": 1.7137, "step": 739 }, { "epoch": 0.07, "learning_rate": 1.9936571929180434e-05, "loss": 1.6418, "step": 740 }, { "epoch": 0.07, "learning_rate": 1.9936249893650684e-05, "loss": 1.7406, "step": 741 }, { "epoch": 0.07, "learning_rate": 1.9935927045288176e-05, "loss": 1.5906, "step": 742 }, { "epoch": 0.07, "learning_rate": 1.9935603384119315e-05, "loss": 1.6078, "step": 743 }, { "epoch": 0.07, "learning_rate": 1.993527891017058e-05, "loss": 1.7172, "step": 744 }, { "epoch": 0.07, "learning_rate": 1.993527891017058e-05, "loss": 1.6652, "step": 745 }, { "epoch": 0.07, "learning_rate": 1.9934953623468517e-05, "loss": 1.5267, "step": 746 }, { "epoch": 0.07, "learning_rate": 1.9934627524039733e-05, "loss": 1.6347, "step": 747 }, { "epoch": 0.07, "learning_rate": 1.9934300611910906e-05, "loss": 1.799, "step": 748 }, { "epoch": 0.07, "learning_rate": 1.993397288710878e-05, "loss": 1.6606, "step": 749 }, { "epoch": 0.07, "learning_rate": 1.9933644349660163e-05, "loss": 1.6828, "step": 750 }, { "epoch": 0.07, "learning_rate": 1.993331499959193e-05, "loss": 1.7092, "step": 751 }, { "epoch": 0.07, "learning_rate": 1.9932984836931026e-05, "loss": 1.7309, "step": 752 }, { "epoch": 0.07, "learning_rate": 1.9932653861704457e-05, "loss": 1.5772, "step": 753 }, { "epoch": 0.07, "learning_rate": 1.99323220739393e-05, "loss": 1.7301, "step": 754 }, { "epoch": 0.07, "learning_rate": 1.9931989473662696e-05, "loss": 1.6615, "step": 755 }, { "epoch": 0.07, "learning_rate": 1.9931656060901856e-05, "loss": 1.7386, "step": 756 }, { "epoch": 0.07, "learning_rate": 1.9931321835684052e-05, "loss": 1.7203, "step": 757 }, { "epoch": 0.07, "learning_rate": 1.9930986798036626e-05, "loss": 1.7669, "step": 758 }, { "epoch": 0.07, "learning_rate": 1.9930650947986984e-05, "loss": 1.7348, "step": 759 }, { "epoch": 0.07, "learning_rate": 1.9930314285562602e-05, "loss": 1.6473, "step": 760 }, { "epoch": 0.07, "learning_rate": 1.992997681079102e-05, "loss": 1.7376, "step": 761 }, { "epoch": 0.07, "learning_rate": 1.9929638523699845e-05, "loss": 1.6791, "step": 762 }, { "epoch": 0.07, "learning_rate": 1.9929299424316753e-05, "loss": 1.6316, "step": 763 }, { "epoch": 0.07, "learning_rate": 1.9928959512669483e-05, "loss": 1.5387, "step": 764 }, { "epoch": 0.07, "learning_rate": 1.992861878878584e-05, "loss": 1.6585, "step": 765 }, { "epoch": 0.07, "learning_rate": 1.9928277252693693e-05, "loss": 1.6228, "step": 766 }, { "epoch": 0.07, "learning_rate": 1.992793490442099e-05, "loss": 1.6081, "step": 767 }, { "epoch": 0.07, "learning_rate": 1.9927591743995727e-05, "loss": 1.6613, "step": 768 }, { "epoch": 0.07, "learning_rate": 1.9927247771445987e-05, "loss": 1.6821, "step": 769 }, { "epoch": 0.07, "learning_rate": 1.9926902986799898e-05, "loss": 1.6437, "step": 770 }, { "epoch": 0.07, "learning_rate": 1.992655739008567e-05, "loss": 1.6632, "step": 771 }, { "epoch": 0.07, "learning_rate": 1.9926210981331576e-05, "loss": 1.5146, "step": 772 }, { "epoch": 0.07, "learning_rate": 1.9925863760565954e-05, "loss": 1.6515, "step": 773 }, { "epoch": 0.07, "learning_rate": 1.9925515727817206e-05, "loss": 1.5897, "step": 774 }, { "epoch": 0.07, "learning_rate": 1.99251668831138e-05, "loss": 1.7371, "step": 775 }, { "epoch": 0.07, "learning_rate": 1.992481722648428e-05, "loss": 1.6048, "step": 776 }, { "epoch": 0.07, "learning_rate": 1.992446675795724e-05, "loss": 1.5352, "step": 777 }, { "epoch": 0.07, "learning_rate": 1.9924115477561362e-05, "loss": 1.638, "step": 778 }, { "epoch": 0.07, "learning_rate": 1.9923763385325377e-05, "loss": 1.5346, "step": 779 }, { "epoch": 0.07, "learning_rate": 1.9923410481278083e-05, "loss": 1.6795, "step": 780 }, { "epoch": 0.07, "learning_rate": 1.992305676544836e-05, "loss": 1.723, "step": 781 }, { "epoch": 0.07, "learning_rate": 1.992270223786513e-05, "loss": 1.6283, "step": 782 }, { "epoch": 0.07, "learning_rate": 1.9922346898557406e-05, "loss": 1.6929, "step": 783 }, { "epoch": 0.07, "learning_rate": 1.9921990747554253e-05, "loss": 1.577, "step": 784 }, { "epoch": 0.07, "learning_rate": 1.9921633784884804e-05, "loss": 1.5632, "step": 785 }, { "epoch": 0.07, "learning_rate": 1.9921276010578262e-05, "loss": 1.7768, "step": 786 }, { "epoch": 0.07, "learning_rate": 1.9920917424663892e-05, "loss": 1.6631, "step": 787 }, { "epoch": 0.07, "learning_rate": 1.9920558027171037e-05, "loss": 1.7551, "step": 788 }, { "epoch": 0.07, "learning_rate": 1.9920197818129084e-05, "loss": 1.6821, "step": 789 }, { "epoch": 0.07, "learning_rate": 1.991983679756751e-05, "loss": 1.5972, "step": 790 }, { "epoch": 0.07, "learning_rate": 1.9919474965515845e-05, "loss": 1.5883, "step": 791 }, { "epoch": 0.07, "learning_rate": 1.9919112322003688e-05, "loss": 1.6178, "step": 792 }, { "epoch": 0.07, "learning_rate": 1.9918748867060707e-05, "loss": 1.8619, "step": 793 }, { "epoch": 0.07, "learning_rate": 1.9918384600716632e-05, "loss": 1.6912, "step": 794 }, { "epoch": 0.07, "learning_rate": 1.9918019523001262e-05, "loss": 1.7664, "step": 795 }, { "epoch": 0.07, "learning_rate": 1.991765363394446e-05, "loss": 1.603, "step": 796 }, { "epoch": 0.07, "learning_rate": 1.9917286933576165e-05, "loss": 1.6901, "step": 797 }, { "epoch": 0.07, "learning_rate": 1.9916919421926367e-05, "loss": 1.8147, "step": 798 }, { "epoch": 0.07, "learning_rate": 1.9916551099025135e-05, "loss": 1.6404, "step": 799 }, { "epoch": 0.07, "learning_rate": 1.9916181964902595e-05, "loss": 1.6706, "step": 800 }, { "epoch": 0.07, "learning_rate": 1.991581201958895e-05, "loss": 1.5947, "step": 801 }, { "epoch": 0.07, "learning_rate": 1.991544126311446e-05, "loss": 1.7423, "step": 802 }, { "epoch": 0.07, "learning_rate": 1.9915069695509453e-05, "loss": 1.6972, "step": 803 }, { "epoch": 0.07, "learning_rate": 1.9914697316804328e-05, "loss": 1.5948, "step": 804 }, { "epoch": 0.07, "learning_rate": 1.9914324127029544e-05, "loss": 1.5757, "step": 805 }, { "epoch": 0.07, "learning_rate": 1.9913950126215633e-05, "loss": 1.5908, "step": 806 }, { "epoch": 0.07, "learning_rate": 1.991357531439319e-05, "loss": 1.6402, "step": 807 }, { "epoch": 0.07, "learning_rate": 1.9913199691592872e-05, "loss": 1.6246, "step": 808 }, { "epoch": 0.07, "learning_rate": 1.9912823257845414e-05, "loss": 1.7421, "step": 809 }, { "epoch": 0.07, "learning_rate": 1.9912446013181603e-05, "loss": 1.687, "step": 810 }, { "epoch": 0.07, "learning_rate": 1.99120679576323e-05, "loss": 1.8023, "step": 811 }, { "epoch": 0.07, "learning_rate": 1.9911689091228438e-05, "loss": 1.5419, "step": 812 }, { "epoch": 0.07, "learning_rate": 1.991130941400101e-05, "loss": 1.626, "step": 813 }, { "epoch": 0.07, "learning_rate": 1.9910928925981066e-05, "loss": 1.5943, "step": 814 }, { "epoch": 0.07, "learning_rate": 1.9910547627199736e-05, "loss": 1.7124, "step": 815 }, { "epoch": 0.07, "learning_rate": 1.991016551768822e-05, "loss": 1.7341, "step": 816 }, { "epoch": 0.07, "learning_rate": 1.9909782597477764e-05, "loss": 1.64, "step": 817 }, { "epoch": 0.07, "learning_rate": 1.99093988665997e-05, "loss": 1.6214, "step": 818 }, { "epoch": 0.07, "learning_rate": 1.990901432508542e-05, "loss": 1.6714, "step": 819 }, { "epoch": 0.07, "learning_rate": 1.9908628972966378e-05, "loss": 1.7048, "step": 820 }, { "epoch": 0.07, "learning_rate": 1.9908242810274098e-05, "loss": 1.6217, "step": 821 }, { "epoch": 0.07, "learning_rate": 1.990785583704017e-05, "loss": 1.6567, "step": 822 }, { "epoch": 0.07, "learning_rate": 1.990746805329625e-05, "loss": 1.6915, "step": 823 }, { "epoch": 0.07, "learning_rate": 1.9907079459074063e-05, "loss": 1.6161, "step": 824 }, { "epoch": 0.07, "learning_rate": 1.9906690054405397e-05, "loss": 1.6102, "step": 825 }, { "epoch": 0.07, "learning_rate": 1.9906299839322108e-05, "loss": 1.7997, "step": 826 }, { "epoch": 0.07, "learning_rate": 1.990590881385611e-05, "loss": 1.673, "step": 827 }, { "epoch": 0.07, "learning_rate": 1.99055169780394e-05, "loss": 1.6889, "step": 828 }, { "epoch": 0.07, "learning_rate": 1.9905124331904032e-05, "loss": 1.5728, "step": 829 }, { "epoch": 0.07, "learning_rate": 1.9904730875482116e-05, "loss": 1.5085, "step": 830 }, { "epoch": 0.07, "learning_rate": 1.990433660880585e-05, "loss": 1.6022, "step": 831 }, { "epoch": 0.07, "learning_rate": 1.990394153190748e-05, "loss": 1.7006, "step": 832 }, { "epoch": 0.07, "learning_rate": 1.990354564481933e-05, "loss": 1.5267, "step": 833 }, { "epoch": 0.07, "learning_rate": 1.9903148947573783e-05, "loss": 1.6396, "step": 834 }, { "epoch": 0.07, "learning_rate": 1.990275144020329e-05, "loss": 1.6901, "step": 835 }, { "epoch": 0.07, "learning_rate": 1.9902353122740372e-05, "loss": 1.6048, "step": 836 }, { "epoch": 0.07, "learning_rate": 1.990195399521761e-05, "loss": 1.5872, "step": 837 }, { "epoch": 0.07, "learning_rate": 1.9901554057667653e-05, "loss": 1.7888, "step": 838 }, { "epoch": 0.07, "learning_rate": 1.9901153310123224e-05, "loss": 1.7598, "step": 839 }, { "epoch": 0.07, "learning_rate": 1.9900751752617102e-05, "loss": 1.7741, "step": 840 }, { "epoch": 0.07, "learning_rate": 1.9900349385182135e-05, "loss": 1.711, "step": 841 }, { "epoch": 0.07, "learning_rate": 1.989994620785124e-05, "loss": 1.5311, "step": 842 }, { "epoch": 0.07, "learning_rate": 1.9899542220657402e-05, "loss": 1.674, "step": 843 }, { "epoch": 0.07, "learning_rate": 1.9899137423633664e-05, "loss": 1.6461, "step": 844 }, { "epoch": 0.07, "learning_rate": 1.9898731816813147e-05, "loss": 1.6953, "step": 845 }, { "epoch": 0.07, "learning_rate": 1.9898325400229025e-05, "loss": 1.6357, "step": 846 }, { "epoch": 0.07, "learning_rate": 1.9897918173914547e-05, "loss": 1.6274, "step": 847 }, { "epoch": 0.07, "learning_rate": 1.9897510137903023e-05, "loss": 1.5571, "step": 848 }, { "epoch": 0.07, "learning_rate": 1.989710129222784e-05, "loss": 1.589, "step": 849 }, { "epoch": 0.08, "learning_rate": 1.9896691636922438e-05, "loss": 1.6933, "step": 850 }, { "epoch": 0.08, "learning_rate": 1.9896281172020326e-05, "loss": 1.5634, "step": 851 }, { "epoch": 0.08, "learning_rate": 1.989586989755509e-05, "loss": 1.6062, "step": 852 }, { "epoch": 0.08, "learning_rate": 1.9895457813560365e-05, "loss": 1.6033, "step": 853 }, { "epoch": 0.08, "learning_rate": 1.989504492006987e-05, "loss": 1.513, "step": 854 }, { "epoch": 0.08, "learning_rate": 1.989463121711738e-05, "loss": 1.6009, "step": 855 }, { "epoch": 0.08, "learning_rate": 1.9894216704736732e-05, "loss": 1.5883, "step": 856 }, { "epoch": 0.08, "learning_rate": 1.9893801382961842e-05, "loss": 1.6505, "step": 857 }, { "epoch": 0.08, "learning_rate": 1.9893385251826684e-05, "loss": 1.6334, "step": 858 }, { "epoch": 0.08, "learning_rate": 1.9892968311365293e-05, "loss": 1.5869, "step": 859 }, { "epoch": 0.08, "learning_rate": 1.9892550561611785e-05, "loss": 1.6669, "step": 860 }, { "epoch": 0.08, "learning_rate": 1.989213200260033e-05, "loss": 1.5419, "step": 861 }, { "epoch": 0.08, "learning_rate": 1.9891712634365167e-05, "loss": 1.6113, "step": 862 }, { "epoch": 0.08, "learning_rate": 1.9891292456940606e-05, "loss": 1.6194, "step": 863 }, { "epoch": 0.08, "learning_rate": 1.9890871470361014e-05, "loss": 1.6173, "step": 864 }, { "epoch": 0.08, "learning_rate": 1.9890449674660837e-05, "loss": 1.631, "step": 865 }, { "epoch": 0.08, "learning_rate": 1.9890027069874575e-05, "loss": 1.7144, "step": 866 }, { "epoch": 0.08, "learning_rate": 1.98896036560368e-05, "loss": 1.5611, "step": 867 }, { "epoch": 0.08, "learning_rate": 1.988917943318215e-05, "loss": 1.6289, "step": 868 }, { "epoch": 0.08, "learning_rate": 1.9888754401345328e-05, "loss": 1.6634, "step": 869 }, { "epoch": 0.08, "learning_rate": 1.9888328560561103e-05, "loss": 1.6169, "step": 870 }, { "epoch": 0.08, "learning_rate": 1.9887901910864313e-05, "loss": 1.5745, "step": 871 }, { "epoch": 0.08, "learning_rate": 1.9887474452289855e-05, "loss": 1.6602, "step": 872 }, { "epoch": 0.08, "learning_rate": 1.9887046184872703e-05, "loss": 1.6829, "step": 873 }, { "epoch": 0.08, "learning_rate": 1.988661710864789e-05, "loss": 1.6125, "step": 874 }, { "epoch": 0.08, "learning_rate": 1.9886187223650512e-05, "loss": 1.5865, "step": 875 }, { "epoch": 0.08, "learning_rate": 1.988575652991574e-05, "loss": 1.6333, "step": 876 }, { "epoch": 0.08, "learning_rate": 1.988532502747881e-05, "loss": 1.7185, "step": 877 }, { "epoch": 0.08, "learning_rate": 1.988489271637501e-05, "loss": 1.5636, "step": 878 }, { "epoch": 0.08, "learning_rate": 1.9884459596639715e-05, "loss": 1.7278, "step": 879 }, { "epoch": 0.08, "learning_rate": 1.9884025668308355e-05, "loss": 1.8341, "step": 880 }, { "epoch": 0.08, "learning_rate": 1.9883590931416426e-05, "loss": 1.6602, "step": 881 }, { "epoch": 0.08, "learning_rate": 1.9883155385999487e-05, "loss": 1.7589, "step": 882 }, { "epoch": 0.08, "learning_rate": 1.9882719032093174e-05, "loss": 1.7762, "step": 883 }, { "epoch": 0.08, "learning_rate": 1.9882281869733182e-05, "loss": 1.5523, "step": 884 }, { "epoch": 0.08, "learning_rate": 1.988184389895527e-05, "loss": 1.6122, "step": 885 }, { "epoch": 0.08, "learning_rate": 1.9881405119795266e-05, "loss": 1.684, "step": 886 }, { "epoch": 0.08, "learning_rate": 1.988096553228907e-05, "loss": 1.7497, "step": 887 }, { "epoch": 0.08, "learning_rate": 1.9880525136472636e-05, "loss": 1.4579, "step": 888 }, { "epoch": 0.08, "learning_rate": 1.9880083932381993e-05, "loss": 1.5604, "step": 889 }, { "epoch": 0.08, "learning_rate": 1.9879641920053236e-05, "loss": 1.7547, "step": 890 }, { "epoch": 0.08, "learning_rate": 1.9879199099522517e-05, "loss": 1.6771, "step": 891 }, { "epoch": 0.08, "learning_rate": 1.9878755470826066e-05, "loss": 1.692, "step": 892 }, { "epoch": 0.08, "learning_rate": 1.9878311034000176e-05, "loss": 1.7411, "step": 893 }, { "epoch": 0.08, "learning_rate": 1.9877865789081195e-05, "loss": 1.6252, "step": 894 }, { "epoch": 0.08, "learning_rate": 1.9877419736105556e-05, "loss": 1.6535, "step": 895 }, { "epoch": 0.08, "learning_rate": 1.9876972875109745e-05, "loss": 1.6212, "step": 896 }, { "epoch": 0.08, "learning_rate": 1.9876525206130316e-05, "loss": 1.549, "step": 897 }, { "epoch": 0.08, "learning_rate": 1.9876076729203893e-05, "loss": 1.6571, "step": 898 }, { "epoch": 0.08, "learning_rate": 1.987562744436716e-05, "loss": 1.5872, "step": 899 }, { "epoch": 0.08, "learning_rate": 1.987517735165687e-05, "loss": 1.6255, "step": 900 }, { "epoch": 0.08, "learning_rate": 1.987472645110985e-05, "loss": 1.6776, "step": 901 }, { "epoch": 0.08, "learning_rate": 1.9874274742762977e-05, "loss": 1.6626, "step": 902 }, { "epoch": 0.08, "learning_rate": 1.9873822226653212e-05, "loss": 1.6104, "step": 903 }, { "epoch": 0.08, "learning_rate": 1.9873368902817566e-05, "loss": 1.6045, "step": 904 }, { "epoch": 0.08, "learning_rate": 1.987291477129312e-05, "loss": 1.5726, "step": 905 }, { "epoch": 0.08, "learning_rate": 1.9872459832117034e-05, "loss": 1.6684, "step": 906 }, { "epoch": 0.08, "learning_rate": 1.9872004085326518e-05, "loss": 1.61, "step": 907 }, { "epoch": 0.08, "learning_rate": 1.9871547530958855e-05, "loss": 1.4888, "step": 908 }, { "epoch": 0.08, "learning_rate": 1.9871090169051394e-05, "loss": 1.5923, "step": 909 }, { "epoch": 0.08, "learning_rate": 1.987063199964155e-05, "loss": 1.5901, "step": 910 }, { "epoch": 0.08, "learning_rate": 1.98701730227668e-05, "loss": 1.616, "step": 911 }, { "epoch": 0.08, "learning_rate": 1.9869713238464697e-05, "loss": 1.6858, "step": 912 }, { "epoch": 0.08, "learning_rate": 1.986925264677285e-05, "loss": 1.6378, "step": 913 }, { "epoch": 0.08, "learning_rate": 1.9868791247728935e-05, "loss": 1.6745, "step": 914 }, { "epoch": 0.08, "learning_rate": 1.98683290413707e-05, "loss": 1.6742, "step": 915 }, { "epoch": 0.08, "learning_rate": 1.9867866027735955e-05, "loss": 1.7648, "step": 916 }, { "epoch": 0.08, "learning_rate": 1.9867402206862577e-05, "loss": 1.6667, "step": 917 }, { "epoch": 0.08, "learning_rate": 1.9866937578788508e-05, "loss": 1.7044, "step": 918 }, { "epoch": 0.08, "learning_rate": 1.9866472143551756e-05, "loss": 1.6967, "step": 919 }, { "epoch": 0.08, "learning_rate": 1.9866005901190397e-05, "loss": 1.6487, "step": 920 }, { "epoch": 0.08, "learning_rate": 1.986553885174257e-05, "loss": 1.5816, "step": 921 }, { "epoch": 0.08, "learning_rate": 1.9865070995246486e-05, "loss": 1.635, "step": 922 }, { "epoch": 0.08, "learning_rate": 1.9864602331740415e-05, "loss": 1.5593, "step": 923 }, { "epoch": 0.08, "learning_rate": 1.9864132861262697e-05, "loss": 1.5953, "step": 924 }, { "epoch": 0.08, "learning_rate": 1.9863662583851734e-05, "loss": 1.591, "step": 925 }, { "epoch": 0.08, "learning_rate": 1.9863191499546e-05, "loss": 1.7676, "step": 926 }, { "epoch": 0.08, "learning_rate": 1.9862719608384034e-05, "loss": 1.5013, "step": 927 }, { "epoch": 0.08, "learning_rate": 1.9862246910404436e-05, "loss": 1.7493, "step": 928 }, { "epoch": 0.08, "learning_rate": 1.986177340564587e-05, "loss": 1.6609, "step": 929 }, { "epoch": 0.08, "learning_rate": 1.986129909414708e-05, "loss": 1.5415, "step": 930 }, { "epoch": 0.08, "learning_rate": 1.9860823975946863e-05, "loss": 1.6117, "step": 931 }, { "epoch": 0.08, "learning_rate": 1.9860348051084088e-05, "loss": 1.7041, "step": 932 }, { "epoch": 0.08, "learning_rate": 1.985987131959768e-05, "loss": 1.5929, "step": 933 }, { "epoch": 0.08, "learning_rate": 1.985939378152665e-05, "loss": 1.6681, "step": 934 }, { "epoch": 0.08, "learning_rate": 1.9858915436910055e-05, "loss": 1.8414, "step": 935 }, { "epoch": 0.08, "learning_rate": 1.9858436285787024e-05, "loss": 1.6628, "step": 936 }, { "epoch": 0.08, "learning_rate": 1.9857956328196758e-05, "loss": 1.5373, "step": 937 }, { "epoch": 0.08, "learning_rate": 1.9857475564178525e-05, "loss": 1.6197, "step": 938 }, { "epoch": 0.08, "learning_rate": 1.985699399377164e-05, "loss": 1.6992, "step": 939 }, { "epoch": 0.08, "learning_rate": 1.985651161701551e-05, "loss": 1.6806, "step": 940 }, { "epoch": 0.08, "learning_rate": 1.9856028433949596e-05, "loss": 1.7678, "step": 941 }, { "epoch": 0.08, "learning_rate": 1.9855544444613415e-05, "loss": 1.6664, "step": 942 }, { "epoch": 0.08, "learning_rate": 1.9855059649046565e-05, "loss": 1.5658, "step": 943 }, { "epoch": 0.08, "learning_rate": 1.9854574047288704e-05, "loss": 1.6764, "step": 944 }, { "epoch": 0.08, "learning_rate": 1.985408763937956e-05, "loss": 1.5557, "step": 945 }, { "epoch": 0.08, "learning_rate": 1.985360042535892e-05, "loss": 1.6356, "step": 946 }, { "epoch": 0.08, "learning_rate": 1.9853112405266638e-05, "loss": 1.6108, "step": 947 }, { "epoch": 0.08, "learning_rate": 1.9852623579142643e-05, "loss": 1.4991, "step": 948 }, { "epoch": 0.08, "learning_rate": 1.9852133947026917e-05, "loss": 1.7585, "step": 949 }, { "epoch": 0.08, "learning_rate": 1.985164350895952e-05, "loss": 1.6139, "step": 950 }, { "epoch": 0.08, "learning_rate": 1.9851152264980565e-05, "loss": 1.5618, "step": 951 }, { "epoch": 0.08, "learning_rate": 1.985066021513024e-05, "loss": 1.7542, "step": 952 }, { "epoch": 0.08, "learning_rate": 1.9850167359448807e-05, "loss": 1.6628, "step": 953 }, { "epoch": 0.08, "learning_rate": 1.9849673697976572e-05, "loss": 1.5465, "step": 954 }, { "epoch": 0.08, "learning_rate": 1.9849179230753922e-05, "loss": 1.6443, "step": 955 }, { "epoch": 0.08, "learning_rate": 1.9848683957821308e-05, "loss": 1.4571, "step": 956 }, { "epoch": 0.08, "learning_rate": 1.984818787921925e-05, "loss": 1.5432, "step": 957 }, { "epoch": 0.08, "learning_rate": 1.984769099498832e-05, "loss": 1.686, "step": 958 }, { "epoch": 0.08, "learning_rate": 1.9847193305169172e-05, "loss": 1.7305, "step": 959 }, { "epoch": 0.08, "learning_rate": 1.984669480980252e-05, "loss": 1.7307, "step": 960 }, { "epoch": 0.08, "learning_rate": 1.984619550892914e-05, "loss": 1.6398, "step": 961 }, { "epoch": 0.08, "learning_rate": 1.9845695402589876e-05, "loss": 1.5541, "step": 962 }, { "epoch": 0.09, "learning_rate": 1.9845194490825647e-05, "loss": 1.5642, "step": 963 }, { "epoch": 0.09, "learning_rate": 1.9844692773677422e-05, "loss": 1.7834, "step": 964 }, { "epoch": 0.09, "learning_rate": 1.9844190251186243e-05, "loss": 1.6157, "step": 965 }, { "epoch": 0.09, "learning_rate": 1.9843686923393228e-05, "loss": 1.5462, "step": 966 }, { "epoch": 0.09, "learning_rate": 1.9843182790339542e-05, "loss": 1.5926, "step": 967 }, { "epoch": 0.09, "learning_rate": 1.984267785206643e-05, "loss": 1.7417, "step": 968 }, { "epoch": 0.09, "learning_rate": 1.9842172108615195e-05, "loss": 1.7378, "step": 969 }, { "epoch": 0.09, "learning_rate": 1.9841665560027218e-05, "loss": 1.5515, "step": 970 }, { "epoch": 0.09, "learning_rate": 1.9841158206343925e-05, "loss": 1.6918, "step": 971 }, { "epoch": 0.09, "learning_rate": 1.9840650047606834e-05, "loss": 1.7037, "step": 972 }, { "epoch": 0.09, "learning_rate": 1.98401410838575e-05, "loss": 1.6507, "step": 973 }, { "epoch": 0.09, "learning_rate": 1.9839631315137565e-05, "loss": 1.5911, "step": 974 }, { "epoch": 0.09, "learning_rate": 1.9839120741488735e-05, "loss": 1.6445, "step": 975 }, { "epoch": 0.09, "learning_rate": 1.983860936295277e-05, "loss": 1.7135, "step": 976 }, { "epoch": 0.09, "learning_rate": 1.983809717957151e-05, "loss": 1.6937, "step": 977 }, { "epoch": 0.09, "learning_rate": 1.9837584191386848e-05, "loss": 1.6609, "step": 978 }, { "epoch": 0.09, "learning_rate": 1.9837070398440753e-05, "loss": 1.6807, "step": 979 }, { "epoch": 0.09, "learning_rate": 1.983655580077525e-05, "loss": 1.7884, "step": 980 }, { "epoch": 0.09, "learning_rate": 1.9836040398432447e-05, "loss": 1.7242, "step": 981 }, { "epoch": 0.09, "learning_rate": 1.9835524191454494e-05, "loss": 1.5511, "step": 982 }, { "epoch": 0.09, "learning_rate": 1.9835007179883626e-05, "loss": 1.6706, "step": 983 }, { "epoch": 0.09, "learning_rate": 1.9834489363762136e-05, "loss": 1.6913, "step": 984 }, { "epoch": 0.09, "learning_rate": 1.983397074313238e-05, "loss": 1.5958, "step": 985 }, { "epoch": 0.09, "learning_rate": 1.983345131803679e-05, "loss": 1.587, "step": 986 }, { "epoch": 0.09, "learning_rate": 1.9832931088517853e-05, "loss": 1.494, "step": 987 }, { "epoch": 0.09, "learning_rate": 1.983241005461813e-05, "loss": 1.6663, "step": 988 }, { "epoch": 0.09, "learning_rate": 1.9831888216380238e-05, "loss": 1.6417, "step": 989 }, { "epoch": 0.09, "learning_rate": 1.983136557384687e-05, "loss": 1.5633, "step": 990 }, { "epoch": 0.09, "learning_rate": 1.983084212706078e-05, "loss": 1.6077, "step": 991 }, { "epoch": 0.09, "learning_rate": 1.9830317876064794e-05, "loss": 1.612, "step": 992 }, { "epoch": 0.09, "learning_rate": 1.9829792820901787e-05, "loss": 1.5774, "step": 993 }, { "epoch": 0.09, "learning_rate": 1.9829266961614716e-05, "loss": 1.5228, "step": 994 }, { "epoch": 0.09, "learning_rate": 1.9828740298246604e-05, "loss": 1.7516, "step": 995 }, { "epoch": 0.09, "learning_rate": 1.9828212830840527e-05, "loss": 1.5928, "step": 996 }, { "epoch": 0.09, "learning_rate": 1.982768455943964e-05, "loss": 1.6592, "step": 997 }, { "epoch": 0.09, "learning_rate": 1.9827155484087155e-05, "loss": 1.6148, "step": 998 }, { "epoch": 0.09, "learning_rate": 1.9826625604826352e-05, "loss": 1.693, "step": 999 }, { "epoch": 0.09, "learning_rate": 1.9826094921700582e-05, "loss": 1.6408, "step": 1000 }, { "epoch": 0.09, "learning_rate": 1.9825563434753254e-05, "loss": 1.6378, "step": 1001 }, { "epoch": 0.09, "learning_rate": 1.9825031144027848e-05, "loss": 1.6169, "step": 1002 }, { "epoch": 0.09, "learning_rate": 1.98244980495679e-05, "loss": 1.571, "step": 1003 }, { "epoch": 0.09, "learning_rate": 1.9823964151417033e-05, "loss": 1.6256, "step": 1004 }, { "epoch": 0.09, "learning_rate": 1.9823429449618915e-05, "loss": 1.5694, "step": 1005 }, { "epoch": 0.09, "learning_rate": 1.9822893944217286e-05, "loss": 1.6189, "step": 1006 }, { "epoch": 0.09, "learning_rate": 1.9822357635255958e-05, "loss": 1.6376, "step": 1007 }, { "epoch": 0.09, "learning_rate": 1.98218205227788e-05, "loss": 1.6217, "step": 1008 }, { "epoch": 0.09, "learning_rate": 1.982128260682975e-05, "loss": 1.4837, "step": 1009 }, { "epoch": 0.09, "learning_rate": 1.982074388745281e-05, "loss": 1.7877, "step": 1010 }, { "epoch": 0.09, "learning_rate": 1.9820204364692056e-05, "loss": 1.6399, "step": 1011 }, { "epoch": 0.09, "learning_rate": 1.981966403859162e-05, "loss": 1.7391, "step": 1012 }, { "epoch": 0.09, "learning_rate": 1.9819122909195707e-05, "loss": 1.6767, "step": 1013 }, { "epoch": 0.09, "learning_rate": 1.9818580976548577e-05, "loss": 1.5784, "step": 1014 }, { "epoch": 0.09, "learning_rate": 1.9818038240694566e-05, "loss": 1.6921, "step": 1015 }, { "epoch": 0.09, "learning_rate": 1.9817494701678073e-05, "loss": 1.6864, "step": 1016 }, { "epoch": 0.09, "learning_rate": 1.981695035954356e-05, "loss": 1.5893, "step": 1017 }, { "epoch": 0.09, "learning_rate": 1.9816405214335564e-05, "loss": 1.6609, "step": 1018 }, { "epoch": 0.09, "learning_rate": 1.9815859266098672e-05, "loss": 1.6577, "step": 1019 }, { "epoch": 0.09, "learning_rate": 1.9815312514877548e-05, "loss": 1.6969, "step": 1020 }, { "epoch": 0.09, "learning_rate": 1.9814764960716922e-05, "loss": 1.5792, "step": 1021 }, { "epoch": 0.09, "learning_rate": 1.981421660366158e-05, "loss": 1.5292, "step": 1022 }, { "epoch": 0.09, "learning_rate": 1.9813667443756386e-05, "loss": 1.7487, "step": 1023 }, { "epoch": 0.09, "learning_rate": 1.9813117481046262e-05, "loss": 1.6651, "step": 1024 }, { "epoch": 0.09, "learning_rate": 1.9812566715576196e-05, "loss": 1.7806, "step": 1025 }, { "epoch": 0.09, "learning_rate": 1.9812015147391245e-05, "loss": 1.7477, "step": 1026 }, { "epoch": 0.09, "learning_rate": 1.981146277653653e-05, "loss": 1.636, "step": 1027 }, { "epoch": 0.09, "learning_rate": 1.9810909603057235e-05, "loss": 1.6807, "step": 1028 }, { "epoch": 0.09, "learning_rate": 1.981035562699862e-05, "loss": 1.7886, "step": 1029 }, { "epoch": 0.09, "learning_rate": 1.9809800848405992e-05, "loss": 1.5196, "step": 1030 }, { "epoch": 0.09, "learning_rate": 1.9809245267324743e-05, "loss": 1.553, "step": 1031 }, { "epoch": 0.09, "learning_rate": 1.9808688883800316e-05, "loss": 1.6517, "step": 1032 }, { "epoch": 0.09, "learning_rate": 1.980813169787823e-05, "loss": 1.6775, "step": 1033 }, { "epoch": 0.09, "learning_rate": 1.9807573709604063e-05, "loss": 1.6759, "step": 1034 }, { "epoch": 0.09, "learning_rate": 1.9807014919023467e-05, "loss": 1.6128, "step": 1035 }, { "epoch": 0.09, "learning_rate": 1.9806455326182147e-05, "loss": 1.7069, "step": 1036 }, { "epoch": 0.09, "learning_rate": 1.980589493112588e-05, "loss": 1.5374, "step": 1037 }, { "epoch": 0.09, "learning_rate": 1.9805333733900515e-05, "loss": 1.6084, "step": 1038 }, { "epoch": 0.09, "learning_rate": 1.9804771734551954e-05, "loss": 1.7372, "step": 1039 }, { "epoch": 0.09, "learning_rate": 1.9804208933126173e-05, "loss": 1.7344, "step": 1040 }, { "epoch": 0.09, "learning_rate": 1.980364532966922e-05, "loss": 1.6221, "step": 1041 }, { "epoch": 0.09, "learning_rate": 1.980308092422719e-05, "loss": 1.7638, "step": 1042 }, { "epoch": 0.09, "learning_rate": 1.9802515716846257e-05, "loss": 1.6415, "step": 1043 }, { "epoch": 0.09, "learning_rate": 1.9801949707572657e-05, "loss": 1.6368, "step": 1044 }, { "epoch": 0.09, "learning_rate": 1.9801382896452696e-05, "loss": 1.5338, "step": 1045 }, { "epoch": 0.09, "learning_rate": 1.980081528353274e-05, "loss": 1.6231, "step": 1046 }, { "epoch": 0.09, "learning_rate": 1.980024686885922e-05, "loss": 1.5618, "step": 1047 }, { "epoch": 0.09, "learning_rate": 1.9799677652478637e-05, "loss": 1.5787, "step": 1048 }, { "epoch": 0.09, "learning_rate": 1.9799107634437558e-05, "loss": 1.5293, "step": 1049 }, { "epoch": 0.09, "learning_rate": 1.979853681478261e-05, "loss": 1.599, "step": 1050 }, { "epoch": 0.09, "learning_rate": 1.979796519356049e-05, "loss": 1.6215, "step": 1051 }, { "epoch": 0.09, "learning_rate": 1.979739277081796e-05, "loss": 1.5888, "step": 1052 }, { "epoch": 0.09, "learning_rate": 1.9796819546601842e-05, "loss": 1.768, "step": 1053 }, { "epoch": 0.09, "learning_rate": 1.9796245520959034e-05, "loss": 1.4042, "step": 1054 }, { "epoch": 0.09, "learning_rate": 1.9795670693936494e-05, "loss": 1.6577, "step": 1055 }, { "epoch": 0.09, "learning_rate": 1.9795095065581244e-05, "loss": 1.6679, "step": 1056 }, { "epoch": 0.09, "learning_rate": 1.979451863594037e-05, "loss": 1.6206, "step": 1057 }, { "epoch": 0.09, "learning_rate": 1.9793941405061034e-05, "loss": 1.6392, "step": 1058 }, { "epoch": 0.09, "learning_rate": 1.9793363372990448e-05, "loss": 1.7884, "step": 1059 }, { "epoch": 0.09, "learning_rate": 1.9792784539775907e-05, "loss": 1.6853, "step": 1060 }, { "epoch": 0.09, "learning_rate": 1.9792204905464754e-05, "loss": 1.6302, "step": 1061 }, { "epoch": 0.09, "learning_rate": 1.9791624470104408e-05, "loss": 1.7269, "step": 1062 }, { "epoch": 0.09, "learning_rate": 1.9791043233742355e-05, "loss": 1.6025, "step": 1063 }, { "epoch": 0.09, "learning_rate": 1.979046119642614e-05, "loss": 1.6854, "step": 1064 }, { "epoch": 0.09, "learning_rate": 1.9789878358203377e-05, "loss": 1.692, "step": 1065 }, { "epoch": 0.09, "learning_rate": 1.9789294719121744e-05, "loss": 1.6203, "step": 1066 }, { "epoch": 0.09, "learning_rate": 1.9788710279228985e-05, "loss": 1.6838, "step": 1067 }, { "epoch": 0.09, "learning_rate": 1.978812503857291e-05, "loss": 1.6385, "step": 1068 }, { "epoch": 0.09, "learning_rate": 1.97875389972014e-05, "loss": 1.406, "step": 1069 }, { "epoch": 0.09, "learning_rate": 1.978695215516239e-05, "loss": 1.6678, "step": 1070 }, { "epoch": 0.09, "learning_rate": 1.978636451250389e-05, "loss": 1.7045, "step": 1071 }, { "epoch": 0.09, "learning_rate": 1.9785776069273967e-05, "loss": 1.441, "step": 1072 }, { "epoch": 0.09, "learning_rate": 1.9785186825520763e-05, "loss": 1.6537, "step": 1073 }, { "epoch": 0.09, "learning_rate": 1.978459678129248e-05, "loss": 1.7099, "step": 1074 }, { "epoch": 0.09, "learning_rate": 1.9784005936637384e-05, "loss": 1.6583, "step": 1075 }, { "epoch": 0.1, "learning_rate": 1.9783414291603814e-05, "loss": 1.6839, "step": 1076 }, { "epoch": 0.1, "learning_rate": 1.9782821846240166e-05, "loss": 1.521, "step": 1077 }, { "epoch": 0.1, "learning_rate": 1.9782228600594903e-05, "loss": 1.68, "step": 1078 }, { "epoch": 0.1, "learning_rate": 1.978163455471656e-05, "loss": 1.6557, "step": 1079 }, { "epoch": 0.1, "learning_rate": 1.978103970865373e-05, "loss": 1.621, "step": 1080 }, { "epoch": 0.1, "learning_rate": 1.9780444062455075e-05, "loss": 1.737, "step": 1081 }, { "epoch": 0.1, "learning_rate": 1.9779847616169315e-05, "loss": 1.4995, "step": 1082 }, { "epoch": 0.1, "learning_rate": 1.9779250369845256e-05, "loss": 1.6589, "step": 1083 }, { "epoch": 0.1, "learning_rate": 1.9778652323531745e-05, "loss": 1.6934, "step": 1084 }, { "epoch": 0.1, "learning_rate": 1.977805347727771e-05, "loss": 1.5934, "step": 1085 }, { "epoch": 0.1, "learning_rate": 1.9777453831132136e-05, "loss": 1.4651, "step": 1086 }, { "epoch": 0.1, "learning_rate": 1.977685338514408e-05, "loss": 1.5891, "step": 1087 }, { "epoch": 0.1, "learning_rate": 1.9776252139362657e-05, "loss": 1.5241, "step": 1088 }, { "epoch": 0.1, "learning_rate": 1.977565009383706e-05, "loss": 1.6309, "step": 1089 }, { "epoch": 0.1, "learning_rate": 1.9775047248616528e-05, "loss": 1.6733, "step": 1090 }, { "epoch": 0.1, "learning_rate": 1.9774443603750383e-05, "loss": 1.579, "step": 1091 }, { "epoch": 0.1, "learning_rate": 1.977383915928801e-05, "loss": 1.6995, "step": 1092 }, { "epoch": 0.1, "learning_rate": 1.9773233915278845e-05, "loss": 1.7218, "step": 1093 }, { "epoch": 0.1, "learning_rate": 1.977262787177241e-05, "loss": 1.7101, "step": 1094 }, { "epoch": 0.1, "learning_rate": 1.9772021028818277e-05, "loss": 1.727, "step": 1095 }, { "epoch": 0.1, "learning_rate": 1.977141338646609e-05, "loss": 1.5332, "step": 1096 }, { "epoch": 0.1, "learning_rate": 1.977080494476556e-05, "loss": 1.6018, "step": 1097 }, { "epoch": 0.1, "learning_rate": 1.977019570376645e-05, "loss": 1.5564, "step": 1098 }, { "epoch": 0.1, "learning_rate": 1.976958566351861e-05, "loss": 1.6785, "step": 1099 }, { "epoch": 0.1, "learning_rate": 1.976897482407194e-05, "loss": 1.5741, "step": 1100 }, { "epoch": 0.1, "learning_rate": 1.9768363185476415e-05, "loss": 1.6211, "step": 1101 }, { "epoch": 0.1, "learning_rate": 1.976775074778206e-05, "loss": 1.623, "step": 1102 }, { "epoch": 0.1, "learning_rate": 1.9767137511038976e-05, "loss": 1.6434, "step": 1103 }, { "epoch": 0.1, "learning_rate": 1.9766523475297337e-05, "loss": 1.7056, "step": 1104 }, { "epoch": 0.1, "learning_rate": 1.976590864060737e-05, "loss": 1.4433, "step": 1105 }, { "epoch": 0.1, "learning_rate": 1.976529300701937e-05, "loss": 1.5913, "step": 1106 }, { "epoch": 0.1, "learning_rate": 1.9764676574583704e-05, "loss": 1.6765, "step": 1107 }, { "epoch": 0.1, "learning_rate": 1.9764059343350793e-05, "loss": 1.5504, "step": 1108 }, { "epoch": 0.1, "learning_rate": 1.976344131337113e-05, "loss": 1.5891, "step": 1109 }, { "epoch": 0.1, "learning_rate": 1.9762822484695275e-05, "loss": 1.7854, "step": 1110 }, { "epoch": 0.1, "learning_rate": 1.976220285737385e-05, "loss": 1.6304, "step": 1111 }, { "epoch": 0.1, "learning_rate": 1.9761582431457548e-05, "loss": 1.6015, "step": 1112 }, { "epoch": 0.1, "learning_rate": 1.9760961206997115e-05, "loss": 1.5706, "step": 1113 }, { "epoch": 0.1, "learning_rate": 1.9760339184043378e-05, "loss": 1.631, "step": 1114 }, { "epoch": 0.1, "learning_rate": 1.9759716362647216e-05, "loss": 1.6683, "step": 1115 }, { "epoch": 0.1, "learning_rate": 1.9759092742859577e-05, "loss": 1.5919, "step": 1116 }, { "epoch": 0.1, "learning_rate": 1.9758468324731483e-05, "loss": 1.7165, "step": 1117 }, { "epoch": 0.1, "learning_rate": 1.9757843108314006e-05, "loss": 1.5555, "step": 1118 }, { "epoch": 0.1, "learning_rate": 1.9757217093658304e-05, "loss": 1.6912, "step": 1119 }, { "epoch": 0.1, "learning_rate": 1.9756590280815575e-05, "loss": 1.5994, "step": 1120 }, { "epoch": 0.1, "learning_rate": 1.9755962669837102e-05, "loss": 1.6801, "step": 1121 }, { "epoch": 0.1, "learning_rate": 1.9755334260774224e-05, "loss": 1.4708, "step": 1122 }, { "epoch": 0.1, "learning_rate": 1.9754705053678348e-05, "loss": 1.5116, "step": 1123 }, { "epoch": 0.1, "learning_rate": 1.9754075048600954e-05, "loss": 1.562, "step": 1124 }, { "epoch": 0.1, "learning_rate": 1.9753444245593568e-05, "loss": 1.7133, "step": 1125 }, { "epoch": 0.1, "learning_rate": 1.9752812644707796e-05, "loss": 1.6919, "step": 1126 }, { "epoch": 0.1, "learning_rate": 1.975218024599531e-05, "loss": 1.6412, "step": 1127 }, { "epoch": 0.1, "learning_rate": 1.975154704950784e-05, "loss": 1.6308, "step": 1128 }, { "epoch": 0.1, "learning_rate": 1.9750913055297184e-05, "loss": 1.542, "step": 1129 }, { "epoch": 0.1, "learning_rate": 1.975027826341521e-05, "loss": 1.7459, "step": 1130 }, { "epoch": 0.1, "learning_rate": 1.9749642673913843e-05, "loss": 1.6575, "step": 1131 }, { "epoch": 0.1, "learning_rate": 1.9749006286845075e-05, "loss": 1.65, "step": 1132 }, { "epoch": 0.1, "learning_rate": 1.974836910226097e-05, "loss": 1.7304, "step": 1133 }, { "epoch": 0.1, "learning_rate": 1.9747731120213656e-05, "loss": 1.6906, "step": 1134 }, { "epoch": 0.1, "learning_rate": 1.9747092340755314e-05, "loss": 1.5453, "step": 1135 }, { "epoch": 0.1, "learning_rate": 1.9746452763938205e-05, "loss": 1.5819, "step": 1136 }, { "epoch": 0.1, "learning_rate": 1.9745812389814648e-05, "loss": 1.5482, "step": 1137 }, { "epoch": 0.1, "learning_rate": 1.9745171218437026e-05, "loss": 1.6405, "step": 1138 }, { "epoch": 0.1, "learning_rate": 1.9744529249857795e-05, "loss": 1.5706, "step": 1139 }, { "epoch": 0.1, "learning_rate": 1.9743886484129468e-05, "loss": 1.7271, "step": 1140 }, { "epoch": 0.1, "learning_rate": 1.9743242921304627e-05, "loss": 1.4901, "step": 1141 }, { "epoch": 0.1, "learning_rate": 1.9742598561435922e-05, "loss": 1.6312, "step": 1142 }, { "epoch": 0.1, "learning_rate": 1.9741953404576058e-05, "loss": 1.5037, "step": 1143 }, { "epoch": 0.1, "learning_rate": 1.9741307450777815e-05, "loss": 1.6799, "step": 1144 }, { "epoch": 0.1, "learning_rate": 1.974066070009404e-05, "loss": 1.5684, "step": 1145 }, { "epoch": 0.1, "learning_rate": 1.974001315257763e-05, "loss": 1.5064, "step": 1146 }, { "epoch": 0.1, "learning_rate": 1.9739364808281564e-05, "loss": 1.6642, "step": 1147 }, { "epoch": 0.1, "learning_rate": 1.973871566725888e-05, "loss": 1.5682, "step": 1148 }, { "epoch": 0.1, "learning_rate": 1.973806572956268e-05, "loss": 1.5775, "step": 1149 }, { "epoch": 0.1, "learning_rate": 1.973741499524613e-05, "loss": 1.5888, "step": 1150 }, { "epoch": 0.1, "learning_rate": 1.9736763464362465e-05, "loss": 1.6259, "step": 1151 }, { "epoch": 0.1, "learning_rate": 1.9736111136964983e-05, "loss": 1.7059, "step": 1152 }, { "epoch": 0.1, "learning_rate": 1.9735458013107048e-05, "loss": 1.5936, "step": 1153 }, { "epoch": 0.1, "learning_rate": 1.973480409284209e-05, "loss": 1.633, "step": 1154 }, { "epoch": 0.1, "learning_rate": 1.97341493762236e-05, "loss": 1.5201, "step": 1155 }, { "epoch": 0.1, "learning_rate": 1.973349386330514e-05, "loss": 1.566, "step": 1156 }, { "epoch": 0.1, "learning_rate": 1.9732837554140328e-05, "loss": 1.6099, "step": 1157 }, { "epoch": 0.1, "learning_rate": 1.973218044878286e-05, "loss": 1.6711, "step": 1158 }, { "epoch": 0.1, "learning_rate": 1.9731522547286482e-05, "loss": 1.6458, "step": 1159 }, { "epoch": 0.1, "learning_rate": 1.9730863849705028e-05, "loss": 1.5851, "step": 1160 }, { "epoch": 0.1, "learning_rate": 1.9730204356092367e-05, "loss": 1.6074, "step": 1161 }, { "epoch": 0.1, "learning_rate": 1.9729544066502463e-05, "loss": 1.6818, "step": 1162 }, { "epoch": 0.1, "learning_rate": 1.9728882980989318e-05, "loss": 1.6816, "step": 1163 }, { "epoch": 0.1, "learning_rate": 1.972822109960702e-05, "loss": 1.5713, "step": 1164 }, { "epoch": 0.1, "learning_rate": 1.9727558422409706e-05, "loss": 1.7104, "step": 1165 }, { "epoch": 0.1, "learning_rate": 1.9726894949451597e-05, "loss": 1.6013, "step": 1166 }, { "epoch": 0.1, "learning_rate": 1.9726230680786963e-05, "loss": 1.5691, "step": 1167 }, { "epoch": 0.1, "learning_rate": 1.972556561647014e-05, "loss": 1.5564, "step": 1168 }, { "epoch": 0.1, "learning_rate": 1.972489975655554e-05, "loss": 1.5963, "step": 1169 }, { "epoch": 0.1, "learning_rate": 1.972423310109763e-05, "loss": 1.6947, "step": 1170 }, { "epoch": 0.1, "learning_rate": 1.972356565015095e-05, "loss": 1.5476, "step": 1171 }, { "epoch": 0.1, "learning_rate": 1.9722897403770095e-05, "loss": 1.7148, "step": 1172 }, { "epoch": 0.1, "learning_rate": 1.9722228362009734e-05, "loss": 1.6869, "step": 1173 }, { "epoch": 0.1, "learning_rate": 1.9721558524924596e-05, "loss": 1.6289, "step": 1174 }, { "epoch": 0.1, "learning_rate": 1.9720887892569476e-05, "loss": 1.6259, "step": 1175 }, { "epoch": 0.1, "learning_rate": 1.9720216464999243e-05, "loss": 1.7051, "step": 1176 }, { "epoch": 0.1, "learning_rate": 1.971954424226881e-05, "loss": 1.6395, "step": 1177 }, { "epoch": 0.1, "learning_rate": 1.971887122443318e-05, "loss": 1.6443, "step": 1178 }, { "epoch": 0.1, "learning_rate": 1.9718197411547402e-05, "loss": 1.7312, "step": 1179 }, { "epoch": 0.1, "learning_rate": 1.9717522803666597e-05, "loss": 1.5216, "step": 1180 }, { "epoch": 0.1, "learning_rate": 1.9716847400845958e-05, "loss": 1.7734, "step": 1181 }, { "epoch": 0.1, "learning_rate": 1.9716171203140727e-05, "loss": 1.7814, "step": 1182 }, { "epoch": 0.1, "learning_rate": 1.9715494210606226e-05, "loss": 1.6015, "step": 1183 }, { "epoch": 0.1, "learning_rate": 1.9714816423297834e-05, "loss": 1.621, "step": 1184 }, { "epoch": 0.1, "learning_rate": 1.9714137841270998e-05, "loss": 1.697, "step": 1185 }, { "epoch": 0.1, "learning_rate": 1.9713458464581233e-05, "loss": 1.7055, "step": 1186 }, { "epoch": 0.1, "learning_rate": 1.971277829328411e-05, "loss": 1.5747, "step": 1187 }, { "epoch": 0.1, "learning_rate": 1.971209732743527e-05, "loss": 1.7082, "step": 1188 }, { "epoch": 0.1, "learning_rate": 1.9711415567090422e-05, "loss": 1.5992, "step": 1189 }, { "epoch": 0.11, "learning_rate": 1.9710733012305337e-05, "loss": 1.6545, "step": 1190 }, { "epoch": 0.11, "learning_rate": 1.971004966313585e-05, "loss": 1.657, "step": 1191 }, { "epoch": 0.11, "learning_rate": 1.970936551963786e-05, "loss": 1.626, "step": 1192 }, { "epoch": 0.11, "learning_rate": 1.970868058186734e-05, "loss": 1.605, "step": 1193 }, { "epoch": 0.11, "learning_rate": 1.9707994849880315e-05, "loss": 1.5911, "step": 1194 }, { "epoch": 0.11, "learning_rate": 1.9707308323732884e-05, "loss": 1.6311, "step": 1195 }, { "epoch": 0.11, "learning_rate": 1.970662100348121e-05, "loss": 1.6286, "step": 1196 }, { "epoch": 0.11, "learning_rate": 1.9705932889181515e-05, "loss": 1.5296, "step": 1197 }, { "epoch": 0.11, "learning_rate": 1.970524398089009e-05, "loss": 1.6159, "step": 1198 }, { "epoch": 0.11, "learning_rate": 1.9704554278663294e-05, "loss": 1.5604, "step": 1199 }, { "epoch": 0.11, "learning_rate": 1.9703863782557543e-05, "loss": 1.6678, "step": 1200 }, { "epoch": 0.11, "learning_rate": 1.9703172492629332e-05, "loss": 1.6733, "step": 1201 }, { "epoch": 0.11, "learning_rate": 1.9702480408935203e-05, "loss": 1.6546, "step": 1202 }, { "epoch": 0.11, "learning_rate": 1.9701787531531775e-05, "loss": 1.5201, "step": 1203 }, { "epoch": 0.11, "learning_rate": 1.970109386047573e-05, "loss": 1.5107, "step": 1204 }, { "epoch": 0.11, "learning_rate": 1.970039939582381e-05, "loss": 1.5378, "step": 1205 }, { "epoch": 0.11, "learning_rate": 1.9699704137632833e-05, "loss": 1.5297, "step": 1206 }, { "epoch": 0.11, "learning_rate": 1.9699008085959666e-05, "loss": 1.5234, "step": 1207 }, { "epoch": 0.11, "learning_rate": 1.9698311240861253e-05, "loss": 1.574, "step": 1208 }, { "epoch": 0.11, "learning_rate": 1.9697613602394597e-05, "loss": 1.6562, "step": 1209 }, { "epoch": 0.11, "learning_rate": 1.969691517061677e-05, "loss": 1.4743, "step": 1210 }, { "epoch": 0.11, "learning_rate": 1.9696215945584913e-05, "loss": 1.64, "step": 1211 }, { "epoch": 0.11, "learning_rate": 1.9695515927356213e-05, "loss": 1.6761, "step": 1212 }, { "epoch": 0.11, "learning_rate": 1.9694815115987946e-05, "loss": 1.5287, "step": 1213 }, { "epoch": 0.11, "learning_rate": 1.9694113511537434e-05, "loss": 1.6602, "step": 1214 }, { "epoch": 0.11, "learning_rate": 1.969341111406208e-05, "loss": 1.6052, "step": 1215 }, { "epoch": 0.11, "learning_rate": 1.9692707923619336e-05, "loss": 1.6582, "step": 1216 }, { "epoch": 0.11, "learning_rate": 1.969200394026673e-05, "loss": 1.6268, "step": 1217 }, { "epoch": 0.11, "learning_rate": 1.969129916406185e-05, "loss": 1.6792, "step": 1218 }, { "epoch": 0.11, "learning_rate": 1.9690593595062348e-05, "loss": 1.6388, "step": 1219 }, { "epoch": 0.11, "learning_rate": 1.968988723332595e-05, "loss": 1.5887, "step": 1220 }, { "epoch": 0.11, "learning_rate": 1.968918007891043e-05, "loss": 1.5661, "step": 1221 }, { "epoch": 0.11, "learning_rate": 1.9688472131873644e-05, "loss": 1.6887, "step": 1222 }, { "epoch": 0.11, "learning_rate": 1.9687763392273504e-05, "loss": 1.6484, "step": 1223 }, { "epoch": 0.11, "learning_rate": 1.9687053860167987e-05, "loss": 1.6105, "step": 1224 }, { "epoch": 0.11, "learning_rate": 1.9686343535615134e-05, "loss": 1.534, "step": 1225 }, { "epoch": 0.11, "learning_rate": 1.968563241867306e-05, "loss": 1.7624, "step": 1226 }, { "epoch": 0.11, "learning_rate": 1.9684920509399926e-05, "loss": 1.6402, "step": 1227 }, { "epoch": 0.11, "learning_rate": 1.9684207807853983e-05, "loss": 1.6031, "step": 1228 }, { "epoch": 0.11, "learning_rate": 1.9683494314093528e-05, "loss": 1.6716, "step": 1229 }, { "epoch": 0.11, "learning_rate": 1.9682780028176923e-05, "loss": 1.6222, "step": 1230 }, { "epoch": 0.11, "learning_rate": 1.9682064950162604e-05, "loss": 1.7035, "step": 1231 }, { "epoch": 0.11, "learning_rate": 1.968134908010907e-05, "loss": 1.5769, "step": 1232 }, { "epoch": 0.11, "learning_rate": 1.9680632418074883e-05, "loss": 1.6004, "step": 1233 }, { "epoch": 0.11, "learning_rate": 1.9679914964118663e-05, "loss": 1.6635, "step": 1234 }, { "epoch": 0.11, "learning_rate": 1.9679196718299113e-05, "loss": 1.5338, "step": 1235 }, { "epoch": 0.11, "learning_rate": 1.9678477680674974e-05, "loss": 1.6392, "step": 1236 }, { "epoch": 0.11, "learning_rate": 1.9677757851305075e-05, "loss": 1.5446, "step": 1237 }, { "epoch": 0.11, "learning_rate": 1.96770372302483e-05, "loss": 1.6267, "step": 1238 }, { "epoch": 0.11, "learning_rate": 1.96763158175636e-05, "loss": 1.561, "step": 1239 }, { "epoch": 0.11, "learning_rate": 1.9675593613309997e-05, "loss": 1.6946, "step": 1240 }, { "epoch": 0.11, "learning_rate": 1.9674870617546556e-05, "loss": 1.7802, "step": 1241 }, { "epoch": 0.11, "learning_rate": 1.9674146830332427e-05, "loss": 1.6318, "step": 1242 }, { "epoch": 0.11, "learning_rate": 1.9673422251726825e-05, "loss": 1.7667, "step": 1243 }, { "epoch": 0.11, "learning_rate": 1.967269688178902e-05, "loss": 1.6844, "step": 1244 }, { "epoch": 0.11, "learning_rate": 1.967197072057835e-05, "loss": 1.6447, "step": 1245 }, { "epoch": 0.11, "learning_rate": 1.9671243768154224e-05, "loss": 1.67, "step": 1246 }, { "epoch": 0.11, "learning_rate": 1.96705160245761e-05, "loss": 1.6326, "step": 1247 }, { "epoch": 0.11, "learning_rate": 1.9669787489903518e-05, "loss": 1.7358, "step": 1248 }, { "epoch": 0.11, "learning_rate": 1.9669058164196076e-05, "loss": 1.5144, "step": 1249 }, { "epoch": 0.11, "learning_rate": 1.966832804751343e-05, "loss": 1.4912, "step": 1250 }, { "epoch": 0.11, "learning_rate": 1.9667597139915314e-05, "loss": 1.5992, "step": 1251 }, { "epoch": 0.11, "learning_rate": 1.966686544146152e-05, "loss": 1.6614, "step": 1252 }, { "epoch": 0.11, "learning_rate": 1.96661329522119e-05, "loss": 1.5237, "step": 1253 }, { "epoch": 0.11, "learning_rate": 1.9665399672226377e-05, "loss": 1.5648, "step": 1254 }, { "epoch": 0.11, "learning_rate": 1.9664665601564933e-05, "loss": 1.6361, "step": 1255 }, { "epoch": 0.11, "learning_rate": 1.9663930740287628e-05, "loss": 1.7735, "step": 1256 }, { "epoch": 0.11, "learning_rate": 1.9663195088454568e-05, "loss": 1.5776, "step": 1257 }, { "epoch": 0.11, "learning_rate": 1.9662458646125936e-05, "loss": 1.7055, "step": 1258 }, { "epoch": 0.11, "learning_rate": 1.9661721413361978e-05, "loss": 1.7449, "step": 1259 }, { "epoch": 0.11, "learning_rate": 1.9660983390222997e-05, "loss": 1.5118, "step": 1260 }, { "epoch": 0.11, "learning_rate": 1.9660244576769378e-05, "loss": 1.6796, "step": 1261 }, { "epoch": 0.11, "learning_rate": 1.965950497306155e-05, "loss": 1.5328, "step": 1262 }, { "epoch": 0.11, "learning_rate": 1.9658764579160018e-05, "loss": 1.7145, "step": 1263 }, { "epoch": 0.11, "learning_rate": 1.9658023395125356e-05, "loss": 1.5063, "step": 1264 }, { "epoch": 0.11, "learning_rate": 1.9657281421018188e-05, "loss": 1.6444, "step": 1265 }, { "epoch": 0.11, "learning_rate": 1.9656538656899212e-05, "loss": 1.6489, "step": 1266 }, { "epoch": 0.11, "learning_rate": 1.9655795102829195e-05, "loss": 1.4381, "step": 1267 }, { "epoch": 0.11, "learning_rate": 1.965505075886896e-05, "loss": 1.4672, "step": 1268 }, { "epoch": 0.11, "learning_rate": 1.9654305625079398e-05, "loss": 1.5615, "step": 1269 }, { "epoch": 0.11, "learning_rate": 1.9653559701521464e-05, "loss": 1.5894, "step": 1270 }, { "epoch": 0.11, "learning_rate": 1.965281298825618e-05, "loss": 1.6075, "step": 1271 }, { "epoch": 0.11, "learning_rate": 1.9652065485344633e-05, "loss": 1.5293, "step": 1272 }, { "epoch": 0.11, "learning_rate": 1.9651317192847965e-05, "loss": 1.6572, "step": 1273 }, { "epoch": 0.11, "learning_rate": 1.965056811082739e-05, "loss": 1.555, "step": 1274 }, { "epoch": 0.11, "learning_rate": 1.9649818239344194e-05, "loss": 1.5362, "step": 1275 }, { "epoch": 0.11, "learning_rate": 1.964906757845972e-05, "loss": 1.6086, "step": 1276 }, { "epoch": 0.11, "learning_rate": 1.9648316128235367e-05, "loss": 1.4726, "step": 1277 }, { "epoch": 0.11, "learning_rate": 1.9647563888732613e-05, "loss": 1.6676, "step": 1278 }, { "epoch": 0.11, "learning_rate": 1.9646810860012996e-05, "loss": 1.621, "step": 1279 }, { "epoch": 0.11, "learning_rate": 1.9646057042138114e-05, "loss": 1.6129, "step": 1280 }, { "epoch": 0.11, "learning_rate": 1.9645302435169633e-05, "loss": 1.5465, "step": 1281 }, { "epoch": 0.11, "learning_rate": 1.9644547039169287e-05, "loss": 1.6038, "step": 1282 }, { "epoch": 0.11, "learning_rate": 1.964379085419887e-05, "loss": 1.6243, "step": 1283 }, { "epoch": 0.11, "learning_rate": 1.9643033880320236e-05, "loss": 1.6303, "step": 1284 }, { "epoch": 0.11, "learning_rate": 1.9642276117595313e-05, "loss": 1.5928, "step": 1285 }, { "epoch": 0.11, "learning_rate": 1.9641517566086096e-05, "loss": 1.5561, "step": 1286 }, { "epoch": 0.11, "learning_rate": 1.964075822585463e-05, "loss": 1.8442, "step": 1287 }, { "epoch": 0.11, "learning_rate": 1.963999809696303e-05, "loss": 1.6148, "step": 1288 }, { "epoch": 0.11, "learning_rate": 1.9639237179473486e-05, "loss": 1.6383, "step": 1289 }, { "epoch": 0.11, "learning_rate": 1.9638475473448243e-05, "loss": 1.5501, "step": 1290 }, { "epoch": 0.11, "learning_rate": 1.9637712978949607e-05, "loss": 1.6263, "step": 1291 }, { "epoch": 0.11, "learning_rate": 1.9636949696039962e-05, "loss": 1.6875, "step": 1292 }, { "epoch": 0.11, "learning_rate": 1.9636185624781743e-05, "loss": 1.6246, "step": 1293 }, { "epoch": 0.11, "learning_rate": 1.9635420765237455e-05, "loss": 1.6488, "step": 1294 }, { "epoch": 0.11, "learning_rate": 1.9634655117469667e-05, "loss": 1.6746, "step": 1295 }, { "epoch": 0.11, "learning_rate": 1.9633888681541014e-05, "loss": 1.6432, "step": 1296 }, { "epoch": 0.11, "learning_rate": 1.963312145751419e-05, "loss": 1.7458, "step": 1297 }, { "epoch": 0.11, "learning_rate": 1.9632353445451967e-05, "loss": 1.6676, "step": 1298 }, { "epoch": 0.11, "learning_rate": 1.9631584645417157e-05, "loss": 1.687, "step": 1299 }, { "epoch": 0.11, "learning_rate": 1.963081505747267e-05, "loss": 1.6234, "step": 1300 }, { "epoch": 0.11, "learning_rate": 1.9630044681681447e-05, "loss": 1.5131, "step": 1301 }, { "epoch": 0.11, "learning_rate": 1.9629273518106515e-05, "loss": 1.6635, "step": 1302 }, { "epoch": 0.12, "learning_rate": 1.962850156681096e-05, "loss": 1.5995, "step": 1303 }, { "epoch": 0.12, "learning_rate": 1.962772882785793e-05, "loss": 1.5012, "step": 1304 }, { "epoch": 0.12, "learning_rate": 1.9626955301310633e-05, "loss": 1.553, "step": 1305 }, { "epoch": 0.12, "learning_rate": 1.9626180987232354e-05, "loss": 1.6524, "step": 1306 }, { "epoch": 0.12, "learning_rate": 1.9625405885686436e-05, "loss": 1.6761, "step": 1307 }, { "epoch": 0.12, "learning_rate": 1.962462999673628e-05, "loss": 1.6083, "step": 1308 }, { "epoch": 0.12, "learning_rate": 1.9623853320445363e-05, "loss": 1.7003, "step": 1309 }, { "epoch": 0.12, "learning_rate": 1.962307585687722e-05, "loss": 1.5813, "step": 1310 }, { "epoch": 0.12, "learning_rate": 1.9622297606095447e-05, "loss": 1.5567, "step": 1311 }, { "epoch": 0.12, "learning_rate": 1.962151856816371e-05, "loss": 1.7252, "step": 1312 }, { "epoch": 0.12, "learning_rate": 1.9620738743145746e-05, "loss": 1.6575, "step": 1313 }, { "epoch": 0.12, "learning_rate": 1.9619958131105342e-05, "loss": 1.6943, "step": 1314 }, { "epoch": 0.12, "learning_rate": 1.961917673210635e-05, "loss": 1.6309, "step": 1315 }, { "epoch": 0.12, "learning_rate": 1.96183945462127e-05, "loss": 1.6612, "step": 1316 }, { "epoch": 0.12, "learning_rate": 1.9617611573488383e-05, "loss": 1.6765, "step": 1317 }, { "epoch": 0.12, "learning_rate": 1.9616827813997438e-05, "loss": 1.6721, "step": 1318 }, { "epoch": 0.12, "learning_rate": 1.9616043267803985e-05, "loss": 1.7237, "step": 1319 }, { "epoch": 0.12, "learning_rate": 1.9615257934972207e-05, "loss": 1.6387, "step": 1320 }, { "epoch": 0.12, "learning_rate": 1.9614471815566345e-05, "loss": 1.7202, "step": 1321 }, { "epoch": 0.12, "learning_rate": 1.9613684909650706e-05, "loss": 1.543, "step": 1322 }, { "epoch": 0.12, "learning_rate": 1.961289721728967e-05, "loss": 1.7042, "step": 1323 }, { "epoch": 0.12, "learning_rate": 1.9612108738547664e-05, "loss": 1.7296, "step": 1324 }, { "epoch": 0.12, "learning_rate": 1.9611319473489193e-05, "loss": 1.5687, "step": 1325 }, { "epoch": 0.12, "learning_rate": 1.9610529422178826e-05, "loss": 1.5798, "step": 1326 }, { "epoch": 0.12, "learning_rate": 1.9609738584681192e-05, "loss": 1.7329, "step": 1327 }, { "epoch": 0.12, "learning_rate": 1.960894696106098e-05, "loss": 1.5073, "step": 1328 }, { "epoch": 0.12, "learning_rate": 1.9608154551382955e-05, "loss": 1.5892, "step": 1329 }, { "epoch": 0.12, "learning_rate": 1.9607361355711935e-05, "loss": 1.6183, "step": 1330 }, { "epoch": 0.12, "learning_rate": 1.960656737411281e-05, "loss": 1.5746, "step": 1331 }, { "epoch": 0.12, "learning_rate": 1.960577260665053e-05, "loss": 1.6804, "step": 1332 }, { "epoch": 0.12, "learning_rate": 1.9604977053390116e-05, "loss": 1.6208, "step": 1333 }, { "epoch": 0.12, "learning_rate": 1.960418071439664e-05, "loss": 1.6267, "step": 1334 }, { "epoch": 0.12, "learning_rate": 1.960338358973525e-05, "loss": 1.5411, "step": 1335 }, { "epoch": 0.12, "learning_rate": 1.9602585679471158e-05, "loss": 1.611, "step": 1336 }, { "epoch": 0.12, "learning_rate": 1.9601786983669632e-05, "loss": 1.5265, "step": 1337 }, { "epoch": 0.12, "learning_rate": 1.960098750239601e-05, "loss": 1.6018, "step": 1338 }, { "epoch": 0.12, "learning_rate": 1.9600187235715693e-05, "loss": 1.56, "step": 1339 }, { "epoch": 0.12, "learning_rate": 1.959938618369415e-05, "loss": 1.5786, "step": 1340 }, { "epoch": 0.12, "learning_rate": 1.959858434639691e-05, "loss": 1.6581, "step": 1341 }, { "epoch": 0.12, "learning_rate": 1.959778172388956e-05, "loss": 1.7354, "step": 1342 }, { "epoch": 0.12, "learning_rate": 1.959697831623777e-05, "loss": 1.5792, "step": 1343 }, { "epoch": 0.12, "learning_rate": 1.9596174123507256e-05, "loss": 1.5256, "step": 1344 }, { "epoch": 0.12, "learning_rate": 1.9595369145763807e-05, "loss": 1.5255, "step": 1345 }, { "epoch": 0.12, "learning_rate": 1.9594563383073265e-05, "loss": 1.5762, "step": 1346 }, { "epoch": 0.12, "learning_rate": 1.959375683550156e-05, "loss": 1.5578, "step": 1347 }, { "epoch": 0.12, "learning_rate": 1.9592949503114664e-05, "loss": 1.593, "step": 1348 }, { "epoch": 0.12, "learning_rate": 1.9592141385978617e-05, "loss": 1.76, "step": 1349 }, { "epoch": 0.12, "learning_rate": 1.9591332484159534e-05, "loss": 1.6275, "step": 1350 }, { "epoch": 0.12, "learning_rate": 1.959052279772358e-05, "loss": 1.5318, "step": 1351 }, { "epoch": 0.12, "learning_rate": 1.9589712326737e-05, "loss": 1.7499, "step": 1352 }, { "epoch": 0.12, "learning_rate": 1.958890107126609e-05, "loss": 1.5743, "step": 1353 }, { "epoch": 0.12, "learning_rate": 1.958808903137721e-05, "loss": 1.5454, "step": 1354 }, { "epoch": 0.12, "learning_rate": 1.9587276207136794e-05, "loss": 1.5831, "step": 1355 }, { "epoch": 0.12, "learning_rate": 1.9586462598611332e-05, "loss": 1.6962, "step": 1356 }, { "epoch": 0.12, "learning_rate": 1.9585648205867386e-05, "loss": 1.4973, "step": 1357 }, { "epoch": 0.12, "learning_rate": 1.958483302897157e-05, "loss": 1.6989, "step": 1358 }, { "epoch": 0.12, "learning_rate": 1.9584017067990576e-05, "loss": 1.6487, "step": 1359 }, { "epoch": 0.12, "learning_rate": 1.958320032299115e-05, "loss": 1.559, "step": 1360 }, { "epoch": 0.12, "learning_rate": 1.9582382794040108e-05, "loss": 1.6426, "step": 1361 }, { "epoch": 0.12, "learning_rate": 1.9581564481204327e-05, "loss": 1.5439, "step": 1362 }, { "epoch": 0.12, "learning_rate": 1.958074538455075e-05, "loss": 1.5896, "step": 1363 }, { "epoch": 0.12, "learning_rate": 1.9579925504146374e-05, "loss": 1.6054, "step": 1364 }, { "epoch": 0.12, "learning_rate": 1.957910484005828e-05, "loss": 1.5327, "step": 1365 }, { "epoch": 0.12, "learning_rate": 1.9578283392353598e-05, "loss": 1.6272, "step": 1366 }, { "epoch": 0.12, "learning_rate": 1.9577461161099527e-05, "loss": 1.5732, "step": 1367 }, { "epoch": 0.12, "learning_rate": 1.957663814636333e-05, "loss": 1.6355, "step": 1368 }, { "epoch": 0.12, "learning_rate": 1.9575814348212336e-05, "loss": 1.5627, "step": 1369 }, { "epoch": 0.12, "learning_rate": 1.957498976671393e-05, "loss": 1.7038, "step": 1370 }, { "epoch": 0.12, "learning_rate": 1.9574164401935566e-05, "loss": 1.5249, "step": 1371 }, { "epoch": 0.12, "learning_rate": 1.9573338253944772e-05, "loss": 1.5153, "step": 1372 }, { "epoch": 0.12, "learning_rate": 1.957251132280912e-05, "loss": 1.5624, "step": 1373 }, { "epoch": 0.12, "learning_rate": 1.9571683608596264e-05, "loss": 1.6265, "step": 1374 }, { "epoch": 0.12, "learning_rate": 1.9570855111373912e-05, "loss": 1.6839, "step": 1375 }, { "epoch": 0.12, "learning_rate": 1.9570025831209842e-05, "loss": 1.5554, "step": 1376 }, { "epoch": 0.12, "learning_rate": 1.9569195768171888e-05, "loss": 1.572, "step": 1377 }, { "epoch": 0.12, "learning_rate": 1.9568364922327954e-05, "loss": 1.5755, "step": 1378 }, { "epoch": 0.12, "learning_rate": 1.9567533293746014e-05, "loss": 1.7434, "step": 1379 }, { "epoch": 0.12, "learning_rate": 1.9566700882494097e-05, "loss": 1.574, "step": 1380 }, { "epoch": 0.12, "learning_rate": 1.956586768864029e-05, "loss": 1.5409, "step": 1381 }, { "epoch": 0.12, "learning_rate": 1.956503371225276e-05, "loss": 1.5493, "step": 1382 }, { "epoch": 0.12, "learning_rate": 1.9564198953399726e-05, "loss": 1.4953, "step": 1383 }, { "epoch": 0.12, "learning_rate": 1.956336341214948e-05, "loss": 1.511, "step": 1384 }, { "epoch": 0.12, "learning_rate": 1.9562527088570368e-05, "loss": 1.5458, "step": 1385 }, { "epoch": 0.12, "learning_rate": 1.9561689982730813e-05, "loss": 1.6115, "step": 1386 }, { "epoch": 0.12, "learning_rate": 1.9560852094699286e-05, "loss": 1.6004, "step": 1387 }, { "epoch": 0.12, "learning_rate": 1.956001342454433e-05, "loss": 1.6109, "step": 1388 }, { "epoch": 0.12, "learning_rate": 1.955917397233456e-05, "loss": 1.6567, "step": 1389 }, { "epoch": 0.12, "learning_rate": 1.955833373813864e-05, "loss": 1.7462, "step": 1390 }, { "epoch": 0.12, "learning_rate": 1.9557492722025313e-05, "loss": 1.4687, "step": 1391 }, { "epoch": 0.12, "learning_rate": 1.955665092406337e-05, "loss": 1.6123, "step": 1392 }, { "epoch": 0.12, "learning_rate": 1.955580834432168e-05, "loss": 1.6644, "step": 1393 }, { "epoch": 0.12, "learning_rate": 1.955496498286916e-05, "loss": 1.5694, "step": 1394 }, { "epoch": 0.12, "learning_rate": 1.9554120839774814e-05, "loss": 1.5655, "step": 1395 }, { "epoch": 0.12, "learning_rate": 1.9553275915107694e-05, "loss": 1.6146, "step": 1396 }, { "epoch": 0.12, "learning_rate": 1.955243020893691e-05, "loss": 1.6075, "step": 1397 }, { "epoch": 0.12, "learning_rate": 1.9551583721331653e-05, "loss": 1.5163, "step": 1398 }, { "epoch": 0.12, "learning_rate": 1.955073645236117e-05, "loss": 1.5552, "step": 1399 }, { "epoch": 0.12, "learning_rate": 1.9549888402094768e-05, "loss": 1.4653, "step": 1400 }, { "epoch": 0.12, "learning_rate": 1.9549039570601823e-05, "loss": 1.5546, "step": 1401 }, { "epoch": 0.12, "learning_rate": 1.954818995795177e-05, "loss": 1.6605, "step": 1402 }, { "epoch": 0.12, "learning_rate": 1.954733956421412e-05, "loss": 1.5816, "step": 1403 }, { "epoch": 0.12, "learning_rate": 1.9546488389458435e-05, "loss": 1.5537, "step": 1404 }, { "epoch": 0.12, "learning_rate": 1.9545636433754337e-05, "loss": 1.6833, "step": 1405 }, { "epoch": 0.12, "learning_rate": 1.9544783697171534e-05, "loss": 1.7031, "step": 1406 }, { "epoch": 0.12, "learning_rate": 1.9543930179779775e-05, "loss": 1.561, "step": 1407 }, { "epoch": 0.12, "learning_rate": 1.9543075881648884e-05, "loss": 1.6228, "step": 1408 }, { "epoch": 0.12, "learning_rate": 1.9542220802848746e-05, "loss": 1.6013, "step": 1409 }, { "epoch": 0.12, "learning_rate": 1.954136494344931e-05, "loss": 1.554, "step": 1410 }, { "epoch": 0.12, "learning_rate": 1.9540508303520595e-05, "loss": 1.6969, "step": 1411 }, { "epoch": 0.12, "learning_rate": 1.9539650883132667e-05, "loss": 1.4449, "step": 1412 }, { "epoch": 0.12, "learning_rate": 1.953879268235568e-05, "loss": 1.623, "step": 1413 }, { "epoch": 0.12, "learning_rate": 1.953793370125983e-05, "loss": 1.5798, "step": 1414 }, { "epoch": 0.12, "learning_rate": 1.953707393991539e-05, "loss": 1.6472, "step": 1415 }, { "epoch": 0.13, "learning_rate": 1.9536213398392694e-05, "loss": 1.6102, "step": 1416 }, { "epoch": 0.13, "learning_rate": 1.9535352076762132e-05, "loss": 1.6129, "step": 1417 }, { "epoch": 0.13, "learning_rate": 1.953448997509417e-05, "loss": 1.5535, "step": 1418 }, { "epoch": 0.13, "learning_rate": 1.9533627093459328e-05, "loss": 1.7267, "step": 1419 }, { "epoch": 0.13, "learning_rate": 1.9532763431928194e-05, "loss": 1.5807, "step": 1420 }, { "epoch": 0.13, "learning_rate": 1.9531898990571425e-05, "loss": 1.6431, "step": 1421 }, { "epoch": 0.13, "learning_rate": 1.953103376945973e-05, "loss": 1.6596, "step": 1422 }, { "epoch": 0.13, "learning_rate": 1.953016776866389e-05, "loss": 1.6134, "step": 1423 }, { "epoch": 0.13, "learning_rate": 1.9529300988254752e-05, "loss": 1.6518, "step": 1424 }, { "epoch": 0.13, "learning_rate": 1.952843342830322e-05, "loss": 1.4922, "step": 1425 }, { "epoch": 0.13, "learning_rate": 1.9527565088880265e-05, "loss": 1.6277, "step": 1426 }, { "epoch": 0.13, "learning_rate": 1.952669597005692e-05, "loss": 1.5829, "step": 1427 }, { "epoch": 0.13, "learning_rate": 1.9525826071904284e-05, "loss": 1.6207, "step": 1428 }, { "epoch": 0.13, "learning_rate": 1.9524955394493517e-05, "loss": 1.6251, "step": 1429 }, { "epoch": 0.13, "learning_rate": 1.9524083937895847e-05, "loss": 1.5736, "step": 1430 }, { "epoch": 0.13, "learning_rate": 1.952321170218256e-05, "loss": 1.5633, "step": 1431 }, { "epoch": 0.13, "learning_rate": 1.9522338687425016e-05, "loss": 1.6627, "step": 1432 }, { "epoch": 0.13, "learning_rate": 1.9521464893694624e-05, "loss": 1.741, "step": 1433 }, { "epoch": 0.13, "learning_rate": 1.9520590321062872e-05, "loss": 1.5942, "step": 1434 }, { "epoch": 0.13, "learning_rate": 1.9519714969601295e-05, "loss": 1.6541, "step": 1435 }, { "epoch": 0.13, "learning_rate": 1.951883883938151e-05, "loss": 1.582, "step": 1436 }, { "epoch": 0.13, "learning_rate": 1.9517961930475185e-05, "loss": 1.6691, "step": 1437 }, { "epoch": 0.13, "learning_rate": 1.9517084242954053e-05, "loss": 1.6452, "step": 1438 }, { "epoch": 0.13, "learning_rate": 1.9516205776889913e-05, "loss": 1.559, "step": 1439 }, { "epoch": 0.13, "learning_rate": 1.951532653235463e-05, "loss": 1.6224, "step": 1440 }, { "epoch": 0.13, "learning_rate": 1.9514446509420136e-05, "loss": 1.5992, "step": 1441 }, { "epoch": 0.13, "learning_rate": 1.951356570815841e-05, "loss": 1.528, "step": 1442 }, { "epoch": 0.13, "learning_rate": 1.9512684128641513e-05, "loss": 1.634, "step": 1443 }, { "epoch": 0.13, "learning_rate": 1.951180177094156e-05, "loss": 1.5785, "step": 1444 }, { "epoch": 0.13, "learning_rate": 1.951091863513073e-05, "loss": 1.5504, "step": 1445 }, { "epoch": 0.13, "learning_rate": 1.9510034721281275e-05, "loss": 1.6448, "step": 1446 }, { "epoch": 0.13, "learning_rate": 1.9509150029465494e-05, "loss": 1.5411, "step": 1447 }, { "epoch": 0.13, "learning_rate": 1.9508264559755766e-05, "loss": 1.7738, "step": 1448 }, { "epoch": 0.13, "learning_rate": 1.9507378312224525e-05, "loss": 1.721, "step": 1449 }, { "epoch": 0.13, "learning_rate": 1.9506491286944272e-05, "loss": 1.6391, "step": 1450 }, { "epoch": 0.13, "learning_rate": 1.9505603483987562e-05, "loss": 1.5366, "step": 1451 }, { "epoch": 0.13, "learning_rate": 1.950471490342703e-05, "loss": 1.7163, "step": 1452 }, { "epoch": 0.13, "learning_rate": 1.9503825545335363e-05, "loss": 1.6603, "step": 1453 }, { "epoch": 0.13, "learning_rate": 1.9502935409785316e-05, "loss": 1.6792, "step": 1454 }, { "epoch": 0.13, "learning_rate": 1.95020444968497e-05, "loss": 1.6857, "step": 1455 }, { "epoch": 0.13, "learning_rate": 1.950115280660141e-05, "loss": 1.7001, "step": 1456 }, { "epoch": 0.13, "learning_rate": 1.950026033911337e-05, "loss": 1.5026, "step": 1457 }, { "epoch": 0.13, "learning_rate": 1.949936709445861e-05, "loss": 1.5694, "step": 1458 }, { "epoch": 0.13, "learning_rate": 1.949847307271019e-05, "loss": 1.6282, "step": 1459 }, { "epoch": 0.13, "learning_rate": 1.9497578273941244e-05, "loss": 1.6174, "step": 1460 }, { "epoch": 0.13, "learning_rate": 1.9496682698224973e-05, "loss": 1.5331, "step": 1461 }, { "epoch": 0.13, "learning_rate": 1.9495786345634642e-05, "loss": 1.6376, "step": 1462 }, { "epoch": 0.13, "learning_rate": 1.9494889216243576e-05, "loss": 1.619, "step": 1463 }, { "epoch": 0.13, "learning_rate": 1.949399131012516e-05, "loss": 1.6879, "step": 1464 }, { "epoch": 0.13, "learning_rate": 1.949309262735285e-05, "loss": 1.6438, "step": 1465 }, { "epoch": 0.13, "learning_rate": 1.949219316800017e-05, "loss": 1.5987, "step": 1466 }, { "epoch": 0.13, "learning_rate": 1.9491292932140686e-05, "loss": 1.5884, "step": 1467 }, { "epoch": 0.13, "learning_rate": 1.949039191984805e-05, "loss": 1.5244, "step": 1468 }, { "epoch": 0.13, "learning_rate": 1.948949013119597e-05, "loss": 1.5334, "step": 1469 }, { "epoch": 0.13, "learning_rate": 1.948858756625821e-05, "loss": 1.5342, "step": 1470 }, { "epoch": 0.13, "learning_rate": 1.9487684225108614e-05, "loss": 1.5305, "step": 1471 }, { "epoch": 0.13, "learning_rate": 1.948678010782107e-05, "loss": 1.5567, "step": 1472 }, { "epoch": 0.13, "learning_rate": 1.9485875214469546e-05, "loss": 1.5481, "step": 1473 }, { "epoch": 0.13, "learning_rate": 1.9484969545128063e-05, "loss": 1.5449, "step": 1474 }, { "epoch": 0.13, "learning_rate": 1.9484063099870708e-05, "loss": 1.5847, "step": 1475 }, { "epoch": 0.13, "learning_rate": 1.9483155878771635e-05, "loss": 1.7067, "step": 1476 }, { "epoch": 0.13, "learning_rate": 1.948224788190506e-05, "loss": 1.5956, "step": 1477 }, { "epoch": 0.13, "learning_rate": 1.9481339109345258e-05, "loss": 1.5434, "step": 1478 }, { "epoch": 0.13, "learning_rate": 1.9480429561166575e-05, "loss": 1.6454, "step": 1479 }, { "epoch": 0.13, "learning_rate": 1.9479519237443414e-05, "loss": 1.5707, "step": 1480 }, { "epoch": 0.13, "learning_rate": 1.9478608138250243e-05, "loss": 1.5828, "step": 1481 }, { "epoch": 0.13, "learning_rate": 1.9477696263661598e-05, "loss": 1.5659, "step": 1482 }, { "epoch": 0.13, "learning_rate": 1.9476783613752067e-05, "loss": 1.5468, "step": 1483 }, { "epoch": 0.13, "learning_rate": 1.9475870188596316e-05, "loss": 1.6534, "step": 1484 }, { "epoch": 0.13, "learning_rate": 1.947495598826907e-05, "loss": 1.6771, "step": 1485 }, { "epoch": 0.13, "learning_rate": 1.9474041012845108e-05, "loss": 1.6196, "step": 1486 }, { "epoch": 0.13, "learning_rate": 1.9473125262399284e-05, "loss": 1.5772, "step": 1487 }, { "epoch": 0.13, "learning_rate": 1.9472208737006503e-05, "loss": 1.5974, "step": 1488 }, { "epoch": 0.13, "learning_rate": 1.947129143674175e-05, "loss": 1.599, "step": 1489 }, { "epoch": 0.13, "learning_rate": 1.9470373361680065e-05, "loss": 1.7317, "step": 1490 }, { "epoch": 0.13, "learning_rate": 1.9469454511896547e-05, "loss": 1.6358, "step": 1491 }, { "epoch": 0.13, "learning_rate": 1.946853488746636e-05, "loss": 1.6228, "step": 1492 }, { "epoch": 0.13, "learning_rate": 1.946761448846474e-05, "loss": 1.6808, "step": 1493 }, { "epoch": 0.13, "learning_rate": 1.946669331496697e-05, "loss": 1.5772, "step": 1494 }, { "epoch": 0.13, "learning_rate": 1.946577136704842e-05, "loss": 1.6701, "step": 1495 }, { "epoch": 0.13, "learning_rate": 1.9464848644784497e-05, "loss": 1.5404, "step": 1496 }, { "epoch": 0.13, "learning_rate": 1.946392514825069e-05, "loss": 1.745, "step": 1497 }, { "epoch": 0.13, "learning_rate": 1.9463000877522552e-05, "loss": 1.527, "step": 1498 }, { "epoch": 0.13, "learning_rate": 1.946207583267568e-05, "loss": 1.4983, "step": 1499 }, { "epoch": 0.13, "learning_rate": 1.9461150013785755e-05, "loss": 1.465, "step": 1500 }, { "epoch": 0.13, "learning_rate": 1.9460223420928514e-05, "loss": 1.6579, "step": 1501 }, { "epoch": 0.13, "learning_rate": 1.9459296054179752e-05, "loss": 1.534, "step": 1502 }, { "epoch": 0.13, "learning_rate": 1.9458367913615334e-05, "loss": 1.6509, "step": 1503 }, { "epoch": 0.13, "learning_rate": 1.9457438999311184e-05, "loss": 1.7674, "step": 1504 }, { "epoch": 0.13, "learning_rate": 1.9456509311343298e-05, "loss": 1.6664, "step": 1505 }, { "epoch": 0.13, "learning_rate": 1.9455578849787722e-05, "loss": 1.5486, "step": 1506 }, { "epoch": 0.13, "learning_rate": 1.945464761472058e-05, "loss": 1.5088, "step": 1507 }, { "epoch": 0.13, "learning_rate": 1.945371560621804e-05, "loss": 1.5756, "step": 1508 }, { "epoch": 0.13, "learning_rate": 1.9452782824356356e-05, "loss": 1.6755, "step": 1509 }, { "epoch": 0.13, "learning_rate": 1.9451849269211827e-05, "loss": 1.6826, "step": 1510 }, { "epoch": 0.13, "learning_rate": 1.945091494086083e-05, "loss": 1.5641, "step": 1511 }, { "epoch": 0.13, "learning_rate": 1.944997983937979e-05, "loss": 1.6191, "step": 1512 }, { "epoch": 0.13, "learning_rate": 1.9449043964845202e-05, "loss": 1.63, "step": 1513 }, { "epoch": 0.13, "learning_rate": 1.9448107317333633e-05, "loss": 1.5812, "step": 1514 }, { "epoch": 0.13, "learning_rate": 1.9447169896921698e-05, "loss": 1.6738, "step": 1515 }, { "epoch": 0.13, "learning_rate": 1.9446231703686084e-05, "loss": 1.6037, "step": 1516 }, { "epoch": 0.13, "learning_rate": 1.944529273770354e-05, "loss": 1.6241, "step": 1517 }, { "epoch": 0.13, "learning_rate": 1.9444352999050884e-05, "loss": 1.5359, "step": 1518 }, { "epoch": 0.13, "learning_rate": 1.944341248780498e-05, "loss": 1.531, "step": 1519 }, { "epoch": 0.13, "learning_rate": 1.9442471204042776e-05, "loss": 1.5367, "step": 1520 }, { "epoch": 0.13, "learning_rate": 1.9441529147841266e-05, "loss": 1.602, "step": 1521 }, { "epoch": 0.13, "learning_rate": 1.944058631927752e-05, "loss": 1.5998, "step": 1522 }, { "epoch": 0.13, "learning_rate": 1.943964271842866e-05, "loss": 1.6407, "step": 1523 }, { "epoch": 0.13, "learning_rate": 1.9438698345371886e-05, "loss": 1.5845, "step": 1524 }, { "epoch": 0.13, "learning_rate": 1.9437753200184447e-05, "loss": 1.7166, "step": 1525 }, { "epoch": 0.13, "learning_rate": 1.943680728294366e-05, "loss": 1.5949, "step": 1526 }, { "epoch": 0.13, "learning_rate": 1.9435860593726906e-05, "loss": 1.6648, "step": 1527 }, { "epoch": 0.13, "learning_rate": 1.943491313261163e-05, "loss": 1.6498, "step": 1528 }, { "epoch": 0.14, "learning_rate": 1.9433964899675336e-05, "loss": 1.604, "step": 1529 }, { "epoch": 0.14, "learning_rate": 1.9433015894995594e-05, "loss": 1.5774, "step": 1530 }, { "epoch": 0.14, "learning_rate": 1.943206611865004e-05, "loss": 1.7916, "step": 1531 }, { "epoch": 0.14, "learning_rate": 1.9431115570716375e-05, "loss": 1.6965, "step": 1532 }, { "epoch": 0.14, "learning_rate": 1.9430164251272345e-05, "loss": 1.5743, "step": 1533 }, { "epoch": 0.14, "learning_rate": 1.9429212160395785e-05, "loss": 1.5203, "step": 1534 }, { "epoch": 0.14, "learning_rate": 1.9428259298164572e-05, "loss": 1.4887, "step": 1535 }, { "epoch": 0.14, "learning_rate": 1.942730566465666e-05, "loss": 1.5349, "step": 1536 }, { "epoch": 0.14, "learning_rate": 1.942635125995006e-05, "loss": 1.5717, "step": 1537 }, { "epoch": 0.14, "learning_rate": 1.9425396084122848e-05, "loss": 1.6702, "step": 1538 }, { "epoch": 0.14, "learning_rate": 1.9424440137253157e-05, "loss": 1.6518, "step": 1539 }, { "epoch": 0.14, "learning_rate": 1.942348341941919e-05, "loss": 1.5977, "step": 1540 }, { "epoch": 0.14, "learning_rate": 1.9422525930699217e-05, "loss": 1.6131, "step": 1541 }, { "epoch": 0.14, "learning_rate": 1.9421567671171554e-05, "loss": 1.6763, "step": 1542 }, { "epoch": 0.14, "learning_rate": 1.94206086409146e-05, "loss": 1.5655, "step": 1543 }, { "epoch": 0.14, "learning_rate": 1.9419648840006806e-05, "loss": 1.5147, "step": 1544 }, { "epoch": 0.14, "learning_rate": 1.9418688268526688e-05, "loss": 1.5501, "step": 1545 }, { "epoch": 0.14, "learning_rate": 1.9417726926552824e-05, "loss": 1.5666, "step": 1546 }, { "epoch": 0.14, "learning_rate": 1.941676481416386e-05, "loss": 1.5445, "step": 1547 }, { "epoch": 0.14, "learning_rate": 1.9415801931438497e-05, "loss": 1.5486, "step": 1548 }, { "epoch": 0.14, "learning_rate": 1.9414838278455508e-05, "loss": 1.4432, "step": 1549 }, { "epoch": 0.14, "learning_rate": 1.9413873855293717e-05, "loss": 1.6783, "step": 1550 }, { "epoch": 0.14, "learning_rate": 1.941290866203203e-05, "loss": 1.7261, "step": 1551 }, { "epoch": 0.14, "learning_rate": 1.9411942698749392e-05, "loss": 1.6805, "step": 1552 }, { "epoch": 0.14, "learning_rate": 1.9410975965524833e-05, "loss": 1.5439, "step": 1553 }, { "epoch": 0.14, "learning_rate": 1.941000846243743e-05, "loss": 1.6244, "step": 1554 }, { "epoch": 0.14, "learning_rate": 1.9409040189566336e-05, "loss": 1.5769, "step": 1555 }, { "epoch": 0.14, "learning_rate": 1.9408071146990752e-05, "loss": 1.6646, "step": 1556 }, { "epoch": 0.14, "learning_rate": 1.9407101334789957e-05, "loss": 1.5178, "step": 1557 }, { "epoch": 0.14, "learning_rate": 1.940613075304328e-05, "loss": 1.629, "step": 1558 }, { "epoch": 0.14, "learning_rate": 1.940515940183013e-05, "loss": 1.4244, "step": 1559 }, { "epoch": 0.14, "learning_rate": 1.9404187281229955e-05, "loss": 1.6637, "step": 1560 }, { "epoch": 0.14, "learning_rate": 1.940321439132229e-05, "loss": 1.6418, "step": 1561 }, { "epoch": 0.14, "learning_rate": 1.9402240732186716e-05, "loss": 1.6956, "step": 1562 }, { "epoch": 0.14, "learning_rate": 1.9401266303902882e-05, "loss": 1.5445, "step": 1563 }, { "epoch": 0.14, "learning_rate": 1.9400291106550507e-05, "loss": 1.5376, "step": 1564 }, { "epoch": 0.14, "learning_rate": 1.9399315140209357e-05, "loss": 1.573, "step": 1565 }, { "epoch": 0.14, "learning_rate": 1.9398338404959284e-05, "loss": 1.4994, "step": 1566 }, { "epoch": 0.14, "learning_rate": 1.939736090088018e-05, "loss": 1.6132, "step": 1567 }, { "epoch": 0.14, "learning_rate": 1.9396382628052013e-05, "loss": 1.605, "step": 1568 }, { "epoch": 0.14, "learning_rate": 1.9395403586554807e-05, "loss": 1.6055, "step": 1569 }, { "epoch": 0.14, "learning_rate": 1.9394423776468657e-05, "loss": 1.4297, "step": 1570 }, { "epoch": 0.14, "learning_rate": 1.939344319787371e-05, "loss": 1.4857, "step": 1571 }, { "epoch": 0.14, "learning_rate": 1.9392461850850187e-05, "loss": 1.6166, "step": 1572 }, { "epoch": 0.14, "learning_rate": 1.9391479735478367e-05, "loss": 1.5848, "step": 1573 }, { "epoch": 0.14, "learning_rate": 1.9390496851838595e-05, "loss": 1.6394, "step": 1574 }, { "epoch": 0.14, "learning_rate": 1.9389513200011266e-05, "loss": 1.7311, "step": 1575 }, { "epoch": 0.14, "learning_rate": 1.9388528780076852e-05, "loss": 1.6819, "step": 1576 }, { "epoch": 0.14, "learning_rate": 1.9387543592115886e-05, "loss": 1.5908, "step": 1577 }, { "epoch": 0.14, "learning_rate": 1.9386557636208958e-05, "loss": 1.6129, "step": 1578 }, { "epoch": 0.14, "learning_rate": 1.9385570912436722e-05, "loss": 1.5341, "step": 1579 }, { "epoch": 0.14, "learning_rate": 1.9384583420879905e-05, "loss": 1.5416, "step": 1580 }, { "epoch": 0.14, "learning_rate": 1.938359516161928e-05, "loss": 1.6991, "step": 1581 }, { "epoch": 0.14, "learning_rate": 1.938260613473569e-05, "loss": 1.616, "step": 1582 }, { "epoch": 0.14, "learning_rate": 1.9381616340310053e-05, "loss": 1.4888, "step": 1583 }, { "epoch": 0.14, "learning_rate": 1.9380625778423326e-05, "loss": 1.5464, "step": 1584 }, { "epoch": 0.14, "learning_rate": 1.9379634449156554e-05, "loss": 1.5959, "step": 1585 }, { "epoch": 0.14, "learning_rate": 1.9378642352590823e-05, "loss": 1.6557, "step": 1586 }, { "epoch": 0.14, "learning_rate": 1.9377649488807297e-05, "loss": 1.5753, "step": 1587 }, { "epoch": 0.14, "learning_rate": 1.9376655857887194e-05, "loss": 1.7596, "step": 1588 }, { "epoch": 0.14, "learning_rate": 1.93756614599118e-05, "loss": 1.605, "step": 1589 }, { "epoch": 0.14, "learning_rate": 1.937466629496246e-05, "loss": 1.5374, "step": 1590 }, { "epoch": 0.14, "learning_rate": 1.937367036312058e-05, "loss": 1.5101, "step": 1591 }, { "epoch": 0.14, "learning_rate": 1.9372673664467634e-05, "loss": 1.5401, "step": 1592 }, { "epoch": 0.14, "learning_rate": 1.9371676199085164e-05, "loss": 1.6209, "step": 1593 }, { "epoch": 0.14, "learning_rate": 1.937067796705476e-05, "loss": 1.6598, "step": 1594 }, { "epoch": 0.14, "learning_rate": 1.936967896845808e-05, "loss": 1.6007, "step": 1595 }, { "epoch": 0.14, "learning_rate": 1.936867920337685e-05, "loss": 1.5697, "step": 1596 }, { "epoch": 0.14, "learning_rate": 1.9367678671892864e-05, "loss": 1.5993, "step": 1597 }, { "epoch": 0.14, "learning_rate": 1.9366677374087953e-05, "loss": 1.6094, "step": 1598 }, { "epoch": 0.14, "learning_rate": 1.9365675310044042e-05, "loss": 1.5996, "step": 1599 }, { "epoch": 0.14, "learning_rate": 1.9364672479843102e-05, "loss": 1.6196, "step": 1600 }, { "epoch": 0.14, "learning_rate": 1.9363668883567164e-05, "loss": 1.6556, "step": 1601 }, { "epoch": 0.14, "learning_rate": 1.9362664521298328e-05, "loss": 1.6409, "step": 1602 }, { "epoch": 0.14, "learning_rate": 1.936165939311876e-05, "loss": 1.5963, "step": 1603 }, { "epoch": 0.14, "learning_rate": 1.9360653499110685e-05, "loss": 1.5851, "step": 1604 }, { "epoch": 0.14, "learning_rate": 1.9359646839356382e-05, "loss": 1.6981, "step": 1605 }, { "epoch": 0.14, "learning_rate": 1.9358639413938206e-05, "loss": 1.6201, "step": 1606 }, { "epoch": 0.14, "learning_rate": 1.9357631222938574e-05, "loss": 1.5384, "step": 1607 }, { "epoch": 0.14, "learning_rate": 1.935662226643995e-05, "loss": 1.4922, "step": 1608 }, { "epoch": 0.14, "learning_rate": 1.935561254452488e-05, "loss": 1.5299, "step": 1609 }, { "epoch": 0.14, "learning_rate": 1.935460205727596e-05, "loss": 1.6069, "step": 1610 }, { "epoch": 0.14, "learning_rate": 1.9353590804775856e-05, "loss": 1.6097, "step": 1611 }, { "epoch": 0.14, "learning_rate": 1.935257878710729e-05, "loss": 1.6147, "step": 1612 }, { "epoch": 0.14, "learning_rate": 1.9351566004353048e-05, "loss": 1.6204, "step": 1613 }, { "epoch": 0.14, "learning_rate": 1.935055245659599e-05, "loss": 1.5601, "step": 1614 }, { "epoch": 0.14, "learning_rate": 1.9349538143919018e-05, "loss": 1.5834, "step": 1615 }, { "epoch": 0.14, "learning_rate": 1.9348523066405112e-05, "loss": 1.6036, "step": 1616 }, { "epoch": 0.14, "learning_rate": 1.934750722413731e-05, "loss": 1.6512, "step": 1617 }, { "epoch": 0.14, "learning_rate": 1.9346490617198717e-05, "loss": 1.5999, "step": 1618 }, { "epoch": 0.14, "learning_rate": 1.934547324567249e-05, "loss": 1.6548, "step": 1619 }, { "epoch": 0.14, "learning_rate": 1.9344455109641857e-05, "loss": 1.5969, "step": 1620 }, { "epoch": 0.14, "learning_rate": 1.9343436209190108e-05, "loss": 1.5736, "step": 1621 }, { "epoch": 0.14, "learning_rate": 1.934241654440059e-05, "loss": 1.6129, "step": 1622 }, { "epoch": 0.14, "learning_rate": 1.9341396115356722e-05, "loss": 1.5568, "step": 1623 }, { "epoch": 0.14, "learning_rate": 1.9340374922141975e-05, "loss": 1.5547, "step": 1624 }, { "epoch": 0.14, "learning_rate": 1.933935296483989e-05, "loss": 1.7322, "step": 1625 }, { "epoch": 0.14, "learning_rate": 1.9338330243534068e-05, "loss": 1.7368, "step": 1626 }, { "epoch": 0.14, "learning_rate": 1.9337306758308176e-05, "loss": 1.5583, "step": 1627 }, { "epoch": 0.14, "learning_rate": 1.933628250924593e-05, "loss": 1.5701, "step": 1628 }, { "epoch": 0.14, "learning_rate": 1.9335257496431126e-05, "loss": 1.5647, "step": 1629 }, { "epoch": 0.14, "learning_rate": 1.9334231719947617e-05, "loss": 1.6121, "step": 1630 }, { "epoch": 0.14, "learning_rate": 1.9333205179879312e-05, "loss": 1.5698, "step": 1631 }, { "epoch": 0.14, "learning_rate": 1.9332177876310182e-05, "loss": 1.6847, "step": 1632 }, { "epoch": 0.14, "learning_rate": 1.9331149809324276e-05, "loss": 1.6359, "step": 1633 }, { "epoch": 0.14, "learning_rate": 1.933012097900569e-05, "loss": 1.5863, "step": 1634 }, { "epoch": 0.14, "learning_rate": 1.9329091385438588e-05, "loss": 1.6452, "step": 1635 }, { "epoch": 0.14, "learning_rate": 1.9328061028707192e-05, "loss": 1.568, "step": 1636 }, { "epoch": 0.14, "learning_rate": 1.9327029908895794e-05, "loss": 1.6428, "step": 1637 }, { "epoch": 0.14, "learning_rate": 1.9325998026088742e-05, "loss": 1.6861, "step": 1638 }, { "epoch": 0.14, "learning_rate": 1.932496538037045e-05, "loss": 1.768, "step": 1639 }, { "epoch": 0.14, "learning_rate": 1.93239319718254e-05, "loss": 1.6197, "step": 1640 }, { "epoch": 0.14, "learning_rate": 1.932289780053812e-05, "loss": 1.5573, "step": 1641 }, { "epoch": 0.14, "learning_rate": 1.9321862866593213e-05, "loss": 1.5407, "step": 1642 }, { "epoch": 0.15, "learning_rate": 1.9320827170075343e-05, "loss": 1.5554, "step": 1643 }, { "epoch": 0.15, "learning_rate": 1.931979071106923e-05, "loss": 1.6259, "step": 1644 }, { "epoch": 0.15, "learning_rate": 1.931875348965967e-05, "loss": 1.6721, "step": 1645 }, { "epoch": 0.15, "learning_rate": 1.9317715505931513e-05, "loss": 1.5662, "step": 1646 }, { "epoch": 0.15, "learning_rate": 1.9316676759969658e-05, "loss": 1.6141, "step": 1647 }, { "epoch": 0.15, "learning_rate": 1.931563725185909e-05, "loss": 1.4809, "step": 1648 }, { "epoch": 0.15, "learning_rate": 1.9314596981684842e-05, "loss": 1.5438, "step": 1649 }, { "epoch": 0.15, "learning_rate": 1.9313555949532016e-05, "loss": 1.5225, "step": 1650 }, { "epoch": 0.15, "learning_rate": 1.9312514155485773e-05, "loss": 1.6166, "step": 1651 }, { "epoch": 0.15, "learning_rate": 1.9311471599631334e-05, "loss": 1.538, "step": 1652 }, { "epoch": 0.15, "learning_rate": 1.931042828205399e-05, "loss": 1.6643, "step": 1653 }, { "epoch": 0.15, "learning_rate": 1.930938420283908e-05, "loss": 1.4604, "step": 1654 }, { "epoch": 0.15, "learning_rate": 1.9308339362072025e-05, "loss": 1.4745, "step": 1655 }, { "epoch": 0.15, "learning_rate": 1.930729375983829e-05, "loss": 1.5927, "step": 1656 }, { "epoch": 0.15, "learning_rate": 1.9306247396223415e-05, "loss": 1.4499, "step": 1657 }, { "epoch": 0.15, "learning_rate": 1.9305200271312997e-05, "loss": 1.542, "step": 1658 }, { "epoch": 0.15, "learning_rate": 1.9304152385192698e-05, "loss": 1.649, "step": 1659 }, { "epoch": 0.15, "learning_rate": 1.9303103737948236e-05, "loss": 1.6344, "step": 1660 }, { "epoch": 0.15, "learning_rate": 1.9302054329665393e-05, "loss": 1.5847, "step": 1661 }, { "epoch": 0.15, "learning_rate": 1.9301004160430024e-05, "loss": 1.6257, "step": 1662 }, { "epoch": 0.15, "learning_rate": 1.929995323032803e-05, "loss": 1.5288, "step": 1663 }, { "epoch": 0.15, "learning_rate": 1.9298901539445386e-05, "loss": 1.6359, "step": 1664 }, { "epoch": 0.15, "learning_rate": 1.9297849087868126e-05, "loss": 1.7774, "step": 1665 }, { "epoch": 0.15, "learning_rate": 1.9296795875682344e-05, "loss": 1.638, "step": 1666 }, { "epoch": 0.15, "learning_rate": 1.9295741902974196e-05, "loss": 1.5129, "step": 1667 }, { "epoch": 0.15, "learning_rate": 1.9294687169829904e-05, "loss": 1.6472, "step": 1668 }, { "epoch": 0.15, "learning_rate": 1.9293631676335753e-05, "loss": 1.5237, "step": 1669 }, { "epoch": 0.15, "learning_rate": 1.9292575422578084e-05, "loss": 1.4292, "step": 1670 }, { "epoch": 0.15, "learning_rate": 1.9291518408643303e-05, "loss": 1.671, "step": 1671 }, { "epoch": 0.15, "learning_rate": 1.929046063461788e-05, "loss": 1.5421, "step": 1672 }, { "epoch": 0.15, "learning_rate": 1.9289402100588348e-05, "loss": 1.4578, "step": 1673 }, { "epoch": 0.15, "learning_rate": 1.9288342806641295e-05, "loss": 1.6203, "step": 1674 }, { "epoch": 0.15, "learning_rate": 1.9287282752863383e-05, "loss": 1.5786, "step": 1675 }, { "epoch": 0.15, "learning_rate": 1.9286221939341323e-05, "loss": 1.6356, "step": 1676 }, { "epoch": 0.15, "learning_rate": 1.9285160366161898e-05, "loss": 1.4809, "step": 1677 }, { "epoch": 0.15, "learning_rate": 1.9284098033411948e-05, "loss": 1.512, "step": 1678 }, { "epoch": 0.15, "learning_rate": 1.9283034941178384e-05, "loss": 1.6348, "step": 1679 }, { "epoch": 0.15, "learning_rate": 1.928197108954816e-05, "loss": 1.5221, "step": 1680 }, { "epoch": 0.15, "learning_rate": 1.9280906478608313e-05, "loss": 1.6164, "step": 1681 }, { "epoch": 0.15, "learning_rate": 1.927984110844593e-05, "loss": 1.5999, "step": 1682 }, { "epoch": 0.15, "learning_rate": 1.9278774979148167e-05, "loss": 1.5768, "step": 1683 }, { "epoch": 0.15, "learning_rate": 1.9277708090802232e-05, "loss": 1.519, "step": 1684 }, { "epoch": 0.15, "learning_rate": 1.9276640443495406e-05, "loss": 1.5514, "step": 1685 }, { "epoch": 0.15, "learning_rate": 1.9275572037315026e-05, "loss": 1.5599, "step": 1686 }, { "epoch": 0.15, "learning_rate": 1.9274502872348497e-05, "loss": 1.5867, "step": 1687 }, { "epoch": 0.15, "learning_rate": 1.9273432948683276e-05, "loss": 1.5519, "step": 1688 }, { "epoch": 0.15, "learning_rate": 1.9272362266406886e-05, "loss": 1.6293, "step": 1689 }, { "epoch": 0.15, "learning_rate": 1.9271290825606926e-05, "loss": 1.7367, "step": 1690 }, { "epoch": 0.15, "learning_rate": 1.9270218626371032e-05, "loss": 1.4929, "step": 1691 }, { "epoch": 0.15, "learning_rate": 1.9269145668786923e-05, "loss": 1.6148, "step": 1692 }, { "epoch": 0.15, "learning_rate": 1.9268071952942366e-05, "loss": 1.5923, "step": 1693 }, { "epoch": 0.15, "learning_rate": 1.92669974789252e-05, "loss": 1.7188, "step": 1694 }, { "epoch": 0.15, "learning_rate": 1.9265922246823322e-05, "loss": 1.5659, "step": 1695 }, { "epoch": 0.15, "learning_rate": 1.926484625672469e-05, "loss": 1.6223, "step": 1696 }, { "epoch": 0.15, "learning_rate": 1.9263769508717328e-05, "loss": 1.4676, "step": 1697 }, { "epoch": 0.15, "learning_rate": 1.9262692002889313e-05, "loss": 1.4969, "step": 1698 }, { "epoch": 0.15, "learning_rate": 1.9261613739328796e-05, "loss": 1.5171, "step": 1699 }, { "epoch": 0.15, "learning_rate": 1.926053471812398e-05, "loss": 1.5542, "step": 1700 }, { "epoch": 0.15, "learning_rate": 1.9259454939363142e-05, "loss": 1.6, "step": 1701 }, { "epoch": 0.15, "learning_rate": 1.9258374403134602e-05, "loss": 1.5696, "step": 1702 }, { "epoch": 0.15, "learning_rate": 1.9257293109526758e-05, "loss": 1.5561, "step": 1703 }, { "epoch": 0.15, "learning_rate": 1.9256211058628067e-05, "loss": 1.6386, "step": 1704 }, { "epoch": 0.15, "learning_rate": 1.9255128250527045e-05, "loss": 1.7072, "step": 1705 }, { "epoch": 0.15, "learning_rate": 1.9254044685312266e-05, "loss": 1.6008, "step": 1706 }, { "epoch": 0.15, "learning_rate": 1.9252960363072376e-05, "loss": 1.6084, "step": 1707 }, { "epoch": 0.15, "learning_rate": 1.925187528389608e-05, "loss": 1.6605, "step": 1708 }, { "epoch": 0.15, "learning_rate": 1.925078944787214e-05, "loss": 1.4974, "step": 1709 }, { "epoch": 0.15, "learning_rate": 1.9249702855089378e-05, "loss": 1.5885, "step": 1710 }, { "epoch": 0.15, "learning_rate": 1.9248615505636685e-05, "loss": 1.6249, "step": 1711 }, { "epoch": 0.15, "learning_rate": 1.9247527399603018e-05, "loss": 1.6955, "step": 1712 }, { "epoch": 0.15, "learning_rate": 1.924643853707738e-05, "loss": 1.5889, "step": 1713 }, { "epoch": 0.15, "learning_rate": 1.924534891814885e-05, "loss": 1.6631, "step": 1714 }, { "epoch": 0.15, "learning_rate": 1.9244258542906564e-05, "loss": 1.5693, "step": 1715 }, { "epoch": 0.15, "learning_rate": 1.9243167411439717e-05, "loss": 1.6, "step": 1716 }, { "epoch": 0.15, "learning_rate": 1.924207552383757e-05, "loss": 1.6715, "step": 1717 }, { "epoch": 0.15, "learning_rate": 1.924098288018945e-05, "loss": 1.68, "step": 1718 }, { "epoch": 0.15, "learning_rate": 1.9239889480584734e-05, "loss": 1.6524, "step": 1719 }, { "epoch": 0.15, "learning_rate": 1.9238795325112867e-05, "loss": 1.5388, "step": 1720 }, { "epoch": 0.15, "learning_rate": 1.923770041386336e-05, "loss": 1.5777, "step": 1721 }, { "epoch": 0.15, "learning_rate": 1.923660474692578e-05, "loss": 1.6396, "step": 1722 }, { "epoch": 0.15, "learning_rate": 1.9235508324389764e-05, "loss": 1.5682, "step": 1723 }, { "epoch": 0.15, "learning_rate": 1.9234411146344993e-05, "loss": 1.6237, "step": 1724 }, { "epoch": 0.15, "learning_rate": 1.9233313212881226e-05, "loss": 1.65, "step": 1725 }, { "epoch": 0.15, "learning_rate": 1.9232214524088284e-05, "loss": 1.6995, "step": 1726 }, { "epoch": 0.15, "learning_rate": 1.923111508005604e-05, "loss": 1.5357, "step": 1727 }, { "epoch": 0.15, "learning_rate": 1.9230014880874433e-05, "loss": 1.4753, "step": 1728 }, { "epoch": 0.15, "learning_rate": 1.922891392663347e-05, "loss": 1.4512, "step": 1729 }, { "epoch": 0.15, "learning_rate": 1.9227812217423212e-05, "loss": 1.6236, "step": 1730 }, { "epoch": 0.15, "learning_rate": 1.9226709753333778e-05, "loss": 1.5244, "step": 1731 }, { "epoch": 0.15, "learning_rate": 1.9225606534455364e-05, "loss": 1.6454, "step": 1732 }, { "epoch": 0.15, "learning_rate": 1.922450256087821e-05, "loss": 1.637, "step": 1733 }, { "epoch": 0.15, "learning_rate": 1.9223397832692637e-05, "loss": 1.5302, "step": 1734 }, { "epoch": 0.15, "learning_rate": 1.922229234998901e-05, "loss": 1.6538, "step": 1735 }, { "epoch": 0.15, "learning_rate": 1.9221186112857762e-05, "loss": 1.5904, "step": 1736 }, { "epoch": 0.15, "learning_rate": 1.922007912138939e-05, "loss": 1.5184, "step": 1737 }, { "epoch": 0.15, "learning_rate": 1.9218971375674454e-05, "loss": 1.5753, "step": 1738 }, { "epoch": 0.15, "learning_rate": 1.9217862875803567e-05, "loss": 1.4915, "step": 1739 }, { "epoch": 0.15, "learning_rate": 1.9216753621867412e-05, "loss": 1.6938, "step": 1740 }, { "epoch": 0.15, "learning_rate": 1.9215643613956735e-05, "loss": 1.6696, "step": 1741 }, { "epoch": 0.15, "learning_rate": 1.9214532852162336e-05, "loss": 1.5342, "step": 1742 }, { "epoch": 0.15, "learning_rate": 1.9213421336575083e-05, "loss": 1.5268, "step": 1743 }, { "epoch": 0.15, "learning_rate": 1.9212309067285902e-05, "loss": 1.564, "step": 1744 }, { "epoch": 0.15, "learning_rate": 1.921119604438578e-05, "loss": 1.6088, "step": 1745 }, { "epoch": 0.15, "learning_rate": 1.9210082267965773e-05, "loss": 1.4998, "step": 1746 }, { "epoch": 0.15, "learning_rate": 1.9208967738116987e-05, "loss": 1.498, "step": 1747 }, { "epoch": 0.15, "learning_rate": 1.9207852454930596e-05, "loss": 1.5553, "step": 1748 }, { "epoch": 0.15, "learning_rate": 1.9206736418497843e-05, "loss": 1.5728, "step": 1749 }, { "epoch": 0.15, "learning_rate": 1.920561962891002e-05, "loss": 1.6686, "step": 1750 }, { "epoch": 0.15, "learning_rate": 1.9204502086258482e-05, "loss": 1.5062, "step": 1751 }, { "epoch": 0.15, "learning_rate": 1.9203383790634658e-05, "loss": 1.5393, "step": 1752 }, { "epoch": 0.15, "learning_rate": 1.9202264742130023e-05, "loss": 1.5952, "step": 1753 }, { "epoch": 0.15, "learning_rate": 1.9201144940836122e-05, "loss": 1.5734, "step": 1754 }, { "epoch": 0.15, "learning_rate": 1.9200024386844563e-05, "loss": 1.7076, "step": 1755 }, { "epoch": 0.16, "learning_rate": 1.919890308024701e-05, "loss": 1.4823, "step": 1756 }, { "epoch": 0.16, "learning_rate": 1.9197781021135192e-05, "loss": 1.6082, "step": 1757 }, { "epoch": 0.16, "learning_rate": 1.9196658209600895e-05, "loss": 1.6179, "step": 1758 }, { "epoch": 0.16, "learning_rate": 1.919553464573598e-05, "loss": 1.6374, "step": 1759 }, { "epoch": 0.16, "learning_rate": 1.919441032963235e-05, "loss": 1.5099, "step": 1760 }, { "epoch": 0.16, "learning_rate": 1.9193285261381984e-05, "loss": 1.5526, "step": 1761 }, { "epoch": 0.16, "learning_rate": 1.9192159441076917e-05, "loss": 1.4766, "step": 1762 }, { "epoch": 0.16, "learning_rate": 1.9191032868809247e-05, "loss": 1.6244, "step": 1763 }, { "epoch": 0.16, "learning_rate": 1.9189905544671134e-05, "loss": 1.5749, "step": 1764 }, { "epoch": 0.16, "learning_rate": 1.9188777468754794e-05, "loss": 1.6731, "step": 1765 }, { "epoch": 0.16, "learning_rate": 1.918764864115252e-05, "loss": 1.6082, "step": 1766 }, { "epoch": 0.16, "learning_rate": 1.918651906195664e-05, "loss": 1.5296, "step": 1767 }, { "epoch": 0.16, "learning_rate": 1.9185388731259567e-05, "loss": 1.6143, "step": 1768 }, { "epoch": 0.16, "learning_rate": 1.918425764915377e-05, "loss": 1.4714, "step": 1769 }, { "epoch": 0.16, "learning_rate": 1.9183125815731774e-05, "loss": 1.6069, "step": 1770 }, { "epoch": 0.16, "learning_rate": 1.9181993231086168e-05, "loss": 1.548, "step": 1771 }, { "epoch": 0.16, "learning_rate": 1.9180859895309606e-05, "loss": 1.5857, "step": 1772 }, { "epoch": 0.16, "learning_rate": 1.9179725808494795e-05, "loss": 1.6604, "step": 1773 }, { "epoch": 0.16, "learning_rate": 1.9178590970734512e-05, "loss": 1.4999, "step": 1774 }, { "epoch": 0.16, "learning_rate": 1.917745538212159e-05, "loss": 1.5599, "step": 1775 }, { "epoch": 0.16, "learning_rate": 1.9176319042748926e-05, "loss": 1.4947, "step": 1776 }, { "epoch": 0.16, "learning_rate": 1.9175181952709483e-05, "loss": 1.4807, "step": 1777 }, { "epoch": 0.16, "learning_rate": 1.9174044112096276e-05, "loss": 1.4685, "step": 1778 }, { "epoch": 0.16, "learning_rate": 1.9172905521002384e-05, "loss": 1.4716, "step": 1779 }, { "epoch": 0.16, "learning_rate": 1.9171766179520954e-05, "loss": 1.4685, "step": 1780 }, { "epoch": 0.16, "learning_rate": 1.9170626087745187e-05, "loss": 1.6115, "step": 1781 }, { "epoch": 0.16, "learning_rate": 1.9169485245768343e-05, "loss": 1.6393, "step": 1782 }, { "epoch": 0.16, "learning_rate": 1.9168343653683757e-05, "loss": 1.5035, "step": 1783 }, { "epoch": 0.16, "learning_rate": 1.9167201311584815e-05, "loss": 1.6723, "step": 1784 }, { "epoch": 0.16, "learning_rate": 1.916605821956496e-05, "loss": 1.6277, "step": 1785 }, { "epoch": 0.16, "learning_rate": 1.9164914377717712e-05, "loss": 1.5659, "step": 1786 }, { "epoch": 0.16, "learning_rate": 1.916376978613663e-05, "loss": 1.5065, "step": 1787 }, { "epoch": 0.16, "learning_rate": 1.9162624444915363e-05, "loss": 1.5176, "step": 1788 }, { "epoch": 0.16, "learning_rate": 1.916147835414759e-05, "loss": 1.5814, "step": 1789 }, { "epoch": 0.16, "learning_rate": 1.9160331513927075e-05, "loss": 1.7708, "step": 1790 }, { "epoch": 0.16, "learning_rate": 1.9159183924347636e-05, "loss": 1.5215, "step": 1791 }, { "epoch": 0.16, "learning_rate": 1.9158035585503147e-05, "loss": 1.5229, "step": 1792 }, { "epoch": 0.16, "learning_rate": 1.9156886497487547e-05, "loss": 1.5479, "step": 1793 }, { "epoch": 0.16, "learning_rate": 1.9155736660394843e-05, "loss": 1.6413, "step": 1794 }, { "epoch": 0.16, "learning_rate": 1.915458607431909e-05, "loss": 1.613, "step": 1795 }, { "epoch": 0.16, "learning_rate": 1.9153434739354417e-05, "loss": 1.5729, "step": 1796 }, { "epoch": 0.16, "learning_rate": 1.9152282655595007e-05, "loss": 1.6142, "step": 1797 }, { "epoch": 0.16, "learning_rate": 1.9151129823135105e-05, "loss": 1.6173, "step": 1798 }, { "epoch": 0.16, "learning_rate": 1.914997624206902e-05, "loss": 1.4426, "step": 1799 }, { "epoch": 0.16, "learning_rate": 1.9148821912491118e-05, "loss": 1.4195, "step": 1800 }, { "epoch": 0.16, "learning_rate": 1.914766683449583e-05, "loss": 1.5204, "step": 1801 }, { "epoch": 0.16, "learning_rate": 1.9146511008177643e-05, "loss": 1.6266, "step": 1802 }, { "epoch": 0.16, "learning_rate": 1.9145354433631117e-05, "loss": 1.5791, "step": 1803 }, { "epoch": 0.16, "learning_rate": 1.914419711095086e-05, "loss": 1.5683, "step": 1804 }, { "epoch": 0.16, "learning_rate": 1.914303904023155e-05, "loss": 1.5721, "step": 1805 }, { "epoch": 0.16, "learning_rate": 1.9141880221567918e-05, "loss": 1.6314, "step": 1806 }, { "epoch": 0.16, "learning_rate": 1.9140720655054762e-05, "loss": 1.4461, "step": 1807 }, { "epoch": 0.16, "learning_rate": 1.9139560340786945e-05, "loss": 1.5478, "step": 1808 }, { "epoch": 0.16, "learning_rate": 1.9138399278859384e-05, "loss": 1.7204, "step": 1809 }, { "epoch": 0.16, "learning_rate": 1.9137237469367053e-05, "loss": 1.5245, "step": 1810 }, { "epoch": 0.16, "learning_rate": 1.9136074912405002e-05, "loss": 1.5516, "step": 1811 }, { "epoch": 0.16, "learning_rate": 1.913491160806833e-05, "loss": 1.6611, "step": 1812 }, { "epoch": 0.16, "learning_rate": 1.91337475564522e-05, "loss": 1.5652, "step": 1813 }, { "epoch": 0.16, "learning_rate": 1.913258275765184e-05, "loss": 1.5774, "step": 1814 }, { "epoch": 0.16, "learning_rate": 1.9131417211762534e-05, "loss": 1.6966, "step": 1815 }, { "epoch": 0.16, "learning_rate": 1.913025091887963e-05, "loss": 1.5835, "step": 1816 }, { "epoch": 0.16, "learning_rate": 1.912908387909853e-05, "loss": 1.5575, "step": 1817 }, { "epoch": 0.16, "learning_rate": 1.9127916092514718e-05, "loss": 1.5145, "step": 1818 }, { "epoch": 0.16, "learning_rate": 1.912674755922371e-05, "loss": 1.5933, "step": 1819 }, { "epoch": 0.16, "learning_rate": 1.9125578279321107e-05, "loss": 1.5621, "step": 1820 }, { "epoch": 0.16, "learning_rate": 1.912440825290256e-05, "loss": 1.5597, "step": 1821 }, { "epoch": 0.16, "learning_rate": 1.9123237480063778e-05, "loss": 1.6718, "step": 1822 }, { "epoch": 0.16, "learning_rate": 1.9122065960900537e-05, "loss": 1.5779, "step": 1823 }, { "epoch": 0.16, "learning_rate": 1.9120893695508676e-05, "loss": 1.5891, "step": 1824 }, { "epoch": 0.16, "learning_rate": 1.9119720683984093e-05, "loss": 1.3651, "step": 1825 }, { "epoch": 0.16, "learning_rate": 1.9118546926422742e-05, "loss": 1.6689, "step": 1826 }, { "epoch": 0.16, "learning_rate": 1.9117372422920645e-05, "loss": 1.5227, "step": 1827 }, { "epoch": 0.16, "learning_rate": 1.9116197173573877e-05, "loss": 1.7265, "step": 1828 }, { "epoch": 0.16, "learning_rate": 1.9115021178478583e-05, "loss": 1.6488, "step": 1829 }, { "epoch": 0.16, "learning_rate": 1.9113844437730967e-05, "loss": 1.6286, "step": 1830 }, { "epoch": 0.16, "learning_rate": 1.911266695142729e-05, "loss": 1.7063, "step": 1831 }, { "epoch": 0.16, "learning_rate": 1.911148871966387e-05, "loss": 1.7115, "step": 1832 }, { "epoch": 0.16, "learning_rate": 1.91103097425371e-05, "loss": 1.664, "step": 1833 }, { "epoch": 0.16, "learning_rate": 1.9109130020143428e-05, "loss": 1.5884, "step": 1834 }, { "epoch": 0.16, "learning_rate": 1.9107949552579356e-05, "loss": 1.663, "step": 1835 }, { "epoch": 0.16, "learning_rate": 1.9106768339941448e-05, "loss": 1.4719, "step": 1836 }, { "epoch": 0.16, "learning_rate": 1.910558638232634e-05, "loss": 1.6097, "step": 1837 }, { "epoch": 0.16, "learning_rate": 1.9104403679830715e-05, "loss": 1.4989, "step": 1838 }, { "epoch": 0.16, "learning_rate": 1.9103220232551335e-05, "loss": 1.4986, "step": 1839 }, { "epoch": 0.16, "learning_rate": 1.9102036040585e-05, "loss": 1.7596, "step": 1840 }, { "epoch": 0.16, "learning_rate": 1.9102036040585e-05, "loss": 1.6209, "step": 1841 }, { "epoch": 0.16, "learning_rate": 1.9100851104028587e-05, "loss": 1.6189, "step": 1842 }, { "epoch": 0.16, "learning_rate": 1.909966542297903e-05, "loss": 1.5872, "step": 1843 }, { "epoch": 0.16, "learning_rate": 1.9098478997533324e-05, "loss": 1.5874, "step": 1844 }, { "epoch": 0.16, "learning_rate": 1.909729182778852e-05, "loss": 1.6577, "step": 1845 }, { "epoch": 0.16, "learning_rate": 1.909610391384174e-05, "loss": 1.5929, "step": 1846 }, { "epoch": 0.16, "learning_rate": 1.9094915255790157e-05, "loss": 1.515, "step": 1847 }, { "epoch": 0.16, "learning_rate": 1.9093725853731008e-05, "loss": 1.5503, "step": 1848 }, { "epoch": 0.16, "learning_rate": 1.9092535707761598e-05, "loss": 1.4657, "step": 1849 }, { "epoch": 0.16, "learning_rate": 1.9091344817979282e-05, "loss": 1.4811, "step": 1850 }, { "epoch": 0.16, "learning_rate": 1.9090153184481476e-05, "loss": 1.6181, "step": 1851 }, { "epoch": 0.16, "learning_rate": 1.9088960807365673e-05, "loss": 1.7895, "step": 1852 }, { "epoch": 0.16, "learning_rate": 1.9087767686729408e-05, "loss": 1.509, "step": 1853 }, { "epoch": 0.16, "learning_rate": 1.908657382267028e-05, "loss": 1.5458, "step": 1854 }, { "epoch": 0.16, "learning_rate": 1.908537921528596e-05, "loss": 1.5678, "step": 1855 }, { "epoch": 0.16, "learning_rate": 1.908418386467417e-05, "loss": 1.5652, "step": 1856 }, { "epoch": 0.16, "learning_rate": 1.9082987770932696e-05, "loss": 1.5008, "step": 1857 }, { "epoch": 0.16, "learning_rate": 1.9081790934159383e-05, "loss": 1.5659, "step": 1858 }, { "epoch": 0.16, "learning_rate": 1.9080593354452137e-05, "loss": 1.4408, "step": 1859 }, { "epoch": 0.16, "learning_rate": 1.907939503190893e-05, "loss": 1.5964, "step": 1860 }, { "epoch": 0.16, "learning_rate": 1.9078195966627785e-05, "loss": 1.5009, "step": 1861 }, { "epoch": 0.16, "learning_rate": 1.9076996158706794e-05, "loss": 1.6273, "step": 1862 }, { "epoch": 0.16, "learning_rate": 1.9075795608244108e-05, "loss": 1.5946, "step": 1863 }, { "epoch": 0.16, "learning_rate": 1.9074594315337934e-05, "loss": 1.5522, "step": 1864 }, { "epoch": 0.16, "learning_rate": 1.9073392280086553e-05, "loss": 1.5671, "step": 1865 }, { "epoch": 0.16, "learning_rate": 1.9072189502588287e-05, "loss": 1.6898, "step": 1866 }, { "epoch": 0.16, "learning_rate": 1.907098598294153e-05, "loss": 1.534, "step": 1867 }, { "epoch": 0.16, "learning_rate": 1.906978172124474e-05, "loss": 1.5875, "step": 1868 }, { "epoch": 0.17, "learning_rate": 1.9068576717596433e-05, "loss": 1.6264, "step": 1869 }, { "epoch": 0.17, "learning_rate": 1.9067370972095178e-05, "loss": 1.5927, "step": 1870 }, { "epoch": 0.17, "learning_rate": 1.9066164484839608e-05, "loss": 1.7418, "step": 1871 }, { "epoch": 0.17, "learning_rate": 1.906495725592843e-05, "loss": 1.4923, "step": 1872 }, { "epoch": 0.17, "learning_rate": 1.90637492854604e-05, "loss": 1.507, "step": 1873 }, { "epoch": 0.17, "learning_rate": 1.906254057353432e-05, "loss": 1.4484, "step": 1874 }, { "epoch": 0.17, "learning_rate": 1.906133112024909e-05, "loss": 1.5541, "step": 1875 }, { "epoch": 0.17, "learning_rate": 1.9060120925703635e-05, "loss": 1.5022, "step": 1876 }, { "epoch": 0.17, "learning_rate": 1.905890998999696e-05, "loss": 1.5413, "step": 1877 }, { "epoch": 0.17, "learning_rate": 1.9057698313228122e-05, "loss": 1.5185, "step": 1878 }, { "epoch": 0.17, "learning_rate": 1.905648589549625e-05, "loss": 1.5998, "step": 1879 }, { "epoch": 0.17, "learning_rate": 1.9055272736900514e-05, "loss": 1.6173, "step": 1880 }, { "epoch": 0.17, "learning_rate": 1.9054058837540164e-05, "loss": 1.612, "step": 1881 }, { "epoch": 0.17, "learning_rate": 1.9052844197514502e-05, "loss": 1.6222, "step": 1882 }, { "epoch": 0.17, "learning_rate": 1.9051628816922887e-05, "loss": 1.72, "step": 1883 }, { "epoch": 0.17, "learning_rate": 1.9050412695864745e-05, "loss": 1.462, "step": 1884 }, { "epoch": 0.17, "learning_rate": 1.9049195834439563e-05, "loss": 1.5526, "step": 1885 }, { "epoch": 0.17, "learning_rate": 1.9047978232746888e-05, "loss": 1.5618, "step": 1886 }, { "epoch": 0.17, "learning_rate": 1.9046759890886316e-05, "loss": 1.7095, "step": 1887 }, { "epoch": 0.17, "learning_rate": 1.9045540808957524e-05, "loss": 1.6348, "step": 1888 }, { "epoch": 0.17, "learning_rate": 1.9044320987060234e-05, "loss": 1.6776, "step": 1889 }, { "epoch": 0.17, "learning_rate": 1.904310042529423e-05, "loss": 1.5962, "step": 1890 }, { "epoch": 0.17, "learning_rate": 1.9041879123759365e-05, "loss": 1.5557, "step": 1891 }, { "epoch": 0.17, "learning_rate": 1.9040657082555546e-05, "loss": 1.5456, "step": 1892 }, { "epoch": 0.17, "learning_rate": 1.903943430178274e-05, "loss": 1.5676, "step": 1893 }, { "epoch": 0.17, "learning_rate": 1.9038210781540977e-05, "loss": 1.5089, "step": 1894 }, { "epoch": 0.17, "learning_rate": 1.903698652193035e-05, "loss": 1.5656, "step": 1895 }, { "epoch": 0.17, "learning_rate": 1.9035761523051005e-05, "loss": 1.5498, "step": 1896 }, { "epoch": 0.17, "learning_rate": 1.9034535785003157e-05, "loss": 1.5291, "step": 1897 }, { "epoch": 0.17, "learning_rate": 1.9033309307887072e-05, "loss": 1.682, "step": 1898 }, { "epoch": 0.17, "learning_rate": 1.9032082091803082e-05, "loss": 1.5326, "step": 1899 }, { "epoch": 0.17, "learning_rate": 1.9030854136851588e-05, "loss": 1.6016, "step": 1900 }, { "epoch": 0.17, "learning_rate": 1.902962544313303e-05, "loss": 1.7979, "step": 1901 }, { "epoch": 0.17, "learning_rate": 1.9028396010747928e-05, "loss": 1.4982, "step": 1902 }, { "epoch": 0.17, "learning_rate": 1.9027165839796857e-05, "loss": 1.5948, "step": 1903 }, { "epoch": 0.17, "learning_rate": 1.9025934930380447e-05, "loss": 1.65, "step": 1904 }, { "epoch": 0.17, "learning_rate": 1.9024703282599396e-05, "loss": 1.5636, "step": 1905 }, { "epoch": 0.17, "learning_rate": 1.9023470896554454e-05, "loss": 1.6819, "step": 1906 }, { "epoch": 0.17, "learning_rate": 1.9023470896554454e-05, "loss": 1.5608, "step": 1907 }, { "epoch": 0.17, "learning_rate": 1.9022237772346437e-05, "loss": 1.5037, "step": 1908 }, { "epoch": 0.17, "learning_rate": 1.9021003910076225e-05, "loss": 1.6658, "step": 1909 }, { "epoch": 0.17, "learning_rate": 1.901976930984475e-05, "loss": 1.5476, "step": 1910 }, { "epoch": 0.17, "learning_rate": 1.9018533971753007e-05, "loss": 1.4183, "step": 1911 }, { "epoch": 0.17, "learning_rate": 1.901729789590206e-05, "loss": 1.5592, "step": 1912 }, { "epoch": 0.17, "learning_rate": 1.9016061082393015e-05, "loss": 1.5785, "step": 1913 }, { "epoch": 0.17, "learning_rate": 1.9014823531327057e-05, "loss": 1.5906, "step": 1914 }, { "epoch": 0.17, "learning_rate": 1.901358524280542e-05, "loss": 1.563, "step": 1915 }, { "epoch": 0.17, "learning_rate": 1.90123462169294e-05, "loss": 1.6125, "step": 1916 }, { "epoch": 0.17, "learning_rate": 1.9011106453800363e-05, "loss": 1.5042, "step": 1917 }, { "epoch": 0.17, "learning_rate": 1.9009865953519724e-05, "loss": 1.6662, "step": 1918 }, { "epoch": 0.17, "learning_rate": 1.9008624716188958e-05, "loss": 1.5789, "step": 1919 }, { "epoch": 0.17, "learning_rate": 1.9007382741909607e-05, "loss": 1.5085, "step": 1920 }, { "epoch": 0.17, "learning_rate": 1.9006140030783275e-05, "loss": 1.5725, "step": 1921 }, { "epoch": 0.17, "learning_rate": 1.9004896582911612e-05, "loss": 1.5787, "step": 1922 }, { "epoch": 0.17, "learning_rate": 1.9003652398396344e-05, "loss": 1.5519, "step": 1923 }, { "epoch": 0.17, "learning_rate": 1.9002407477339253e-05, "loss": 1.6727, "step": 1924 }, { "epoch": 0.17, "learning_rate": 1.9001161819842173e-05, "loss": 1.4996, "step": 1925 }, { "epoch": 0.17, "learning_rate": 1.899991542600701e-05, "loss": 1.5985, "step": 1926 }, { "epoch": 0.17, "learning_rate": 1.8998668295935725e-05, "loss": 1.5462, "step": 1927 }, { "epoch": 0.17, "learning_rate": 1.8997420429730338e-05, "loss": 1.6241, "step": 1928 }, { "epoch": 0.17, "learning_rate": 1.8996171827492927e-05, "loss": 1.718, "step": 1929 }, { "epoch": 0.17, "learning_rate": 1.899492248932564e-05, "loss": 1.655, "step": 1930 }, { "epoch": 0.17, "learning_rate": 1.8993672415330677e-05, "loss": 1.6517, "step": 1931 }, { "epoch": 0.17, "learning_rate": 1.8992421605610297e-05, "loss": 1.6335, "step": 1932 }, { "epoch": 0.17, "learning_rate": 1.899117006026682e-05, "loss": 1.5731, "step": 1933 }, { "epoch": 0.17, "learning_rate": 1.8989917779402637e-05, "loss": 1.6281, "step": 1934 }, { "epoch": 0.17, "learning_rate": 1.8988664763120187e-05, "loss": 1.4883, "step": 1935 }, { "epoch": 0.17, "learning_rate": 1.8987411011521966e-05, "loss": 1.6082, "step": 1936 }, { "epoch": 0.17, "learning_rate": 1.898615652471055e-05, "loss": 1.6086, "step": 1937 }, { "epoch": 0.17, "learning_rate": 1.8984901302788547e-05, "loss": 1.5841, "step": 1938 }, { "epoch": 0.17, "learning_rate": 1.8983645345858652e-05, "loss": 1.5197, "step": 1939 }, { "epoch": 0.17, "learning_rate": 1.8982388654023602e-05, "loss": 1.5856, "step": 1940 }, { "epoch": 0.17, "learning_rate": 1.8981131227386203e-05, "loss": 1.6107, "step": 1941 }, { "epoch": 0.17, "learning_rate": 1.897987306604932e-05, "loss": 1.5007, "step": 1942 }, { "epoch": 0.17, "learning_rate": 1.8978614170115874e-05, "loss": 1.548, "step": 1943 }, { "epoch": 0.17, "learning_rate": 1.8977354539688848e-05, "loss": 1.6008, "step": 1944 }, { "epoch": 0.17, "learning_rate": 1.8976094174871287e-05, "loss": 1.6364, "step": 1945 }, { "epoch": 0.17, "learning_rate": 1.8974833075766294e-05, "loss": 1.5075, "step": 1946 }, { "epoch": 0.17, "learning_rate": 1.8973571242477033e-05, "loss": 1.6856, "step": 1947 }, { "epoch": 0.17, "learning_rate": 1.8972308675106733e-05, "loss": 1.6577, "step": 1948 }, { "epoch": 0.17, "learning_rate": 1.8971045373758673e-05, "loss": 1.5623, "step": 1949 }, { "epoch": 0.17, "learning_rate": 1.8969781338536193e-05, "loss": 1.5063, "step": 1950 }, { "epoch": 0.17, "learning_rate": 1.8968516569542704e-05, "loss": 1.6502, "step": 1951 }, { "epoch": 0.17, "learning_rate": 1.8967251066881674e-05, "loss": 1.625, "step": 1952 }, { "epoch": 0.17, "learning_rate": 1.8965984830656613e-05, "loss": 1.5599, "step": 1953 }, { "epoch": 0.17, "learning_rate": 1.896471786097112e-05, "loss": 1.6487, "step": 1954 }, { "epoch": 0.17, "learning_rate": 1.8963450157928828e-05, "loss": 1.6547, "step": 1955 }, { "epoch": 0.17, "learning_rate": 1.896218172163345e-05, "loss": 1.4428, "step": 1956 }, { "epoch": 0.17, "learning_rate": 1.8960912552188743e-05, "loss": 1.5718, "step": 1957 }, { "epoch": 0.17, "learning_rate": 1.895964264969854e-05, "loss": 1.6323, "step": 1958 }, { "epoch": 0.17, "learning_rate": 1.8958372014266713e-05, "loss": 1.5146, "step": 1959 }, { "epoch": 0.17, "learning_rate": 1.8957100645997214e-05, "loss": 1.617, "step": 1960 }, { "epoch": 0.17, "learning_rate": 1.8955828544994045e-05, "loss": 1.5888, "step": 1961 }, { "epoch": 0.17, "learning_rate": 1.895455571136127e-05, "loss": 1.5781, "step": 1962 }, { "epoch": 0.17, "learning_rate": 1.8953282145203014e-05, "loss": 1.5838, "step": 1963 }, { "epoch": 0.17, "learning_rate": 1.8952007846623458e-05, "loss": 1.6594, "step": 1964 }, { "epoch": 0.17, "learning_rate": 1.8950732815726852e-05, "loss": 1.5265, "step": 1965 }, { "epoch": 0.17, "learning_rate": 1.8949457052617495e-05, "loss": 1.6046, "step": 1966 }, { "epoch": 0.17, "learning_rate": 1.894818055739975e-05, "loss": 1.6528, "step": 1967 }, { "epoch": 0.17, "learning_rate": 1.894690333017804e-05, "loss": 1.5571, "step": 1968 }, { "epoch": 0.17, "learning_rate": 1.8945625371056845e-05, "loss": 1.4058, "step": 1969 }, { "epoch": 0.17, "learning_rate": 1.894434668014072e-05, "loss": 1.5972, "step": 1970 }, { "epoch": 0.17, "learning_rate": 1.8943067257534257e-05, "loss": 1.5468, "step": 1971 }, { "epoch": 0.17, "learning_rate": 1.8941787103342126e-05, "loss": 1.5785, "step": 1972 }, { "epoch": 0.17, "learning_rate": 1.8940506217669042e-05, "loss": 1.5631, "step": 1973 }, { "epoch": 0.17, "learning_rate": 1.8939224600619796e-05, "loss": 1.5432, "step": 1974 }, { "epoch": 0.17, "learning_rate": 1.8937942252299224e-05, "loss": 1.6388, "step": 1975 }, { "epoch": 0.17, "learning_rate": 1.893665917281223e-05, "loss": 1.6213, "step": 1976 }, { "epoch": 0.17, "learning_rate": 1.893537536226378e-05, "loss": 1.6501, "step": 1977 }, { "epoch": 0.17, "learning_rate": 1.893409082075889e-05, "loss": 1.5435, "step": 1978 }, { "epoch": 0.17, "learning_rate": 1.8932805548402645e-05, "loss": 1.613, "step": 1979 }, { "epoch": 0.17, "learning_rate": 1.8931519545300183e-05, "loss": 1.6366, "step": 1980 }, { "epoch": 0.17, "learning_rate": 1.8930232811556712e-05, "loss": 1.5644, "step": 1981 }, { "epoch": 0.18, "learning_rate": 1.8928945347277485e-05, "loss": 1.5536, "step": 1982 }, { "epoch": 0.18, "learning_rate": 1.8927657152567826e-05, "loss": 1.4418, "step": 1983 }, { "epoch": 0.18, "learning_rate": 1.892636822753312e-05, "loss": 1.5102, "step": 1984 }, { "epoch": 0.18, "learning_rate": 1.8925078572278803e-05, "loss": 1.6047, "step": 1985 }, { "epoch": 0.18, "learning_rate": 1.8923788186910373e-05, "loss": 1.4727, "step": 1986 }, { "epoch": 0.18, "learning_rate": 1.892249707153339e-05, "loss": 1.6215, "step": 1987 }, { "epoch": 0.18, "learning_rate": 1.892120522625348e-05, "loss": 1.5658, "step": 1988 }, { "epoch": 0.18, "learning_rate": 1.8919912651176312e-05, "loss": 1.5499, "step": 1989 }, { "epoch": 0.18, "learning_rate": 1.8918619346407633e-05, "loss": 1.5447, "step": 1990 }, { "epoch": 0.18, "learning_rate": 1.8917325312053237e-05, "loss": 1.6001, "step": 1991 }, { "epoch": 0.18, "learning_rate": 1.8916030548218987e-05, "loss": 1.5216, "step": 1992 }, { "epoch": 0.18, "learning_rate": 1.8914735055010795e-05, "loss": 1.5052, "step": 1993 }, { "epoch": 0.18, "learning_rate": 1.891343883253464e-05, "loss": 1.4433, "step": 1994 }, { "epoch": 0.18, "learning_rate": 1.8912141880896566e-05, "loss": 1.6121, "step": 1995 }, { "epoch": 0.18, "learning_rate": 1.891084420020266e-05, "loss": 1.7313, "step": 1996 }, { "epoch": 0.18, "learning_rate": 1.8909545790559084e-05, "loss": 1.5735, "step": 1997 }, { "epoch": 0.18, "learning_rate": 1.8908246652072057e-05, "loss": 1.6139, "step": 1998 }, { "epoch": 0.18, "learning_rate": 1.8906946784847843e-05, "loss": 1.6095, "step": 1999 }, { "epoch": 0.18, "learning_rate": 1.890564618899279e-05, "loss": 1.6387, "step": 2000 }, { "epoch": 0.18, "learning_rate": 1.8904344864613287e-05, "loss": 1.6177, "step": 2001 }, { "epoch": 0.18, "learning_rate": 1.890304281181579e-05, "loss": 1.5065, "step": 2002 }, { "epoch": 0.18, "learning_rate": 1.8901740030706813e-05, "loss": 1.5016, "step": 2003 }, { "epoch": 0.18, "learning_rate": 1.8900436521392928e-05, "loss": 1.5659, "step": 2004 }, { "epoch": 0.18, "learning_rate": 1.889913228398077e-05, "loss": 1.6097, "step": 2005 }, { "epoch": 0.18, "learning_rate": 1.8897827318577034e-05, "loss": 1.508, "step": 2006 }, { "epoch": 0.18, "learning_rate": 1.889652162528847e-05, "loss": 1.5513, "step": 2007 }, { "epoch": 0.18, "learning_rate": 1.8895215204221882e-05, "loss": 1.6617, "step": 2008 }, { "epoch": 0.18, "learning_rate": 1.889390805548416e-05, "loss": 1.6844, "step": 2009 }, { "epoch": 0.18, "learning_rate": 1.889260017918222e-05, "loss": 1.5214, "step": 2010 }, { "epoch": 0.18, "learning_rate": 1.8891291575423056e-05, "loss": 1.6018, "step": 2011 }, { "epoch": 0.18, "learning_rate": 1.888998224431372e-05, "loss": 1.5952, "step": 2012 }, { "epoch": 0.18, "learning_rate": 1.888867218596132e-05, "loss": 1.5784, "step": 2013 }, { "epoch": 0.18, "learning_rate": 1.8887361400473025e-05, "loss": 1.6536, "step": 2014 }, { "epoch": 0.18, "learning_rate": 1.888604988795607e-05, "loss": 1.4903, "step": 2015 }, { "epoch": 0.18, "learning_rate": 1.8884737648517732e-05, "loss": 1.5424, "step": 2016 }, { "epoch": 0.18, "learning_rate": 1.8883424682265363e-05, "loss": 1.5129, "step": 2017 }, { "epoch": 0.18, "learning_rate": 1.8882110989306372e-05, "loss": 1.5622, "step": 2018 }, { "epoch": 0.18, "learning_rate": 1.8880796569748224e-05, "loss": 1.5718, "step": 2019 }, { "epoch": 0.18, "learning_rate": 1.8879481423698443e-05, "loss": 1.5107, "step": 2020 }, { "epoch": 0.18, "learning_rate": 1.8878165551264617e-05, "loss": 1.5386, "step": 2021 }, { "epoch": 0.18, "learning_rate": 1.887684895255439e-05, "loss": 1.4444, "step": 2022 }, { "epoch": 0.18, "learning_rate": 1.8875531627675467e-05, "loss": 1.5307, "step": 2023 }, { "epoch": 0.18, "learning_rate": 1.8874213576735608e-05, "loss": 1.5382, "step": 2024 }, { "epoch": 0.18, "learning_rate": 1.887289479984264e-05, "loss": 1.5135, "step": 2025 }, { "epoch": 0.18, "learning_rate": 1.8871575297104442e-05, "loss": 1.6246, "step": 2026 }, { "epoch": 0.18, "learning_rate": 1.887025506862896e-05, "loss": 1.5513, "step": 2027 }, { "epoch": 0.18, "learning_rate": 1.8868934114524185e-05, "loss": 1.5285, "step": 2028 }, { "epoch": 0.18, "learning_rate": 1.886761243489819e-05, "loss": 1.604, "step": 2029 }, { "epoch": 0.18, "learning_rate": 1.8866290029859086e-05, "loss": 1.5303, "step": 2030 }, { "epoch": 0.18, "learning_rate": 1.8864966899515058e-05, "loss": 1.5636, "step": 2031 }, { "epoch": 0.18, "learning_rate": 1.886364304397434e-05, "loss": 1.6461, "step": 2032 }, { "epoch": 0.18, "learning_rate": 1.886231846334523e-05, "loss": 1.5252, "step": 2033 }, { "epoch": 0.18, "learning_rate": 1.886099315773609e-05, "loss": 1.4859, "step": 2034 }, { "epoch": 0.18, "learning_rate": 1.8859667127255328e-05, "loss": 1.5716, "step": 2035 }, { "epoch": 0.18, "learning_rate": 1.8858340372011427e-05, "loss": 1.5989, "step": 2036 }, { "epoch": 0.18, "learning_rate": 1.8857012892112917e-05, "loss": 1.533, "step": 2037 }, { "epoch": 0.18, "learning_rate": 1.8855684687668393e-05, "loss": 1.6462, "step": 2038 }, { "epoch": 0.18, "learning_rate": 1.8854355758786514e-05, "loss": 1.518, "step": 2039 }, { "epoch": 0.18, "learning_rate": 1.8853026105575987e-05, "loss": 1.5341, "step": 2040 }, { "epoch": 0.18, "learning_rate": 1.8851695728145586e-05, "loss": 1.5794, "step": 2041 }, { "epoch": 0.18, "learning_rate": 1.8850364626604142e-05, "loss": 1.6625, "step": 2042 }, { "epoch": 0.18, "learning_rate": 1.8849032801060542e-05, "loss": 1.4632, "step": 2043 }, { "epoch": 0.18, "learning_rate": 1.8847700251623743e-05, "loss": 1.5173, "step": 2044 }, { "epoch": 0.18, "learning_rate": 1.8846366978402746e-05, "loss": 1.5298, "step": 2045 }, { "epoch": 0.18, "learning_rate": 1.8845032981506625e-05, "loss": 1.5661, "step": 2046 }, { "epoch": 0.18, "learning_rate": 1.8843698261044504e-05, "loss": 1.5065, "step": 2047 }, { "epoch": 0.18, "learning_rate": 1.8842362817125573e-05, "loss": 1.5321, "step": 2048 }, { "epoch": 0.18, "learning_rate": 1.8841026649859072e-05, "loss": 1.5448, "step": 2049 }, { "epoch": 0.18, "learning_rate": 1.8839689759354312e-05, "loss": 1.6714, "step": 2050 }, { "epoch": 0.18, "learning_rate": 1.8838352145720656e-05, "loss": 1.4534, "step": 2051 }, { "epoch": 0.18, "learning_rate": 1.8837013809067523e-05, "loss": 1.5901, "step": 2052 }, { "epoch": 0.18, "learning_rate": 1.8837013809067523e-05, "loss": 1.5366, "step": 2053 }, { "epoch": 0.18, "learning_rate": 1.8835674749504397e-05, "loss": 1.5079, "step": 2054 }, { "epoch": 0.18, "learning_rate": 1.8834334967140823e-05, "loss": 1.637, "step": 2055 }, { "epoch": 0.18, "learning_rate": 1.88329944620864e-05, "loss": 1.5209, "step": 2056 }, { "epoch": 0.18, "learning_rate": 1.8831653234450784e-05, "loss": 1.4857, "step": 2057 }, { "epoch": 0.18, "learning_rate": 1.8830311284343697e-05, "loss": 1.5328, "step": 2058 }, { "epoch": 0.18, "learning_rate": 1.882896861187492e-05, "loss": 1.5921, "step": 2059 }, { "epoch": 0.18, "learning_rate": 1.8827625217154278e-05, "loss": 1.53, "step": 2060 }, { "epoch": 0.18, "learning_rate": 1.8826281100291684e-05, "loss": 1.6177, "step": 2061 }, { "epoch": 0.18, "learning_rate": 1.8824936261397083e-05, "loss": 1.5674, "step": 2062 }, { "epoch": 0.18, "learning_rate": 1.882359070058049e-05, "loss": 1.5597, "step": 2063 }, { "epoch": 0.18, "learning_rate": 1.8822244417951977e-05, "loss": 1.5325, "step": 2064 }, { "epoch": 0.18, "learning_rate": 1.8820897413621682e-05, "loss": 1.4526, "step": 2065 }, { "epoch": 0.18, "learning_rate": 1.881954968769979e-05, "loss": 1.5363, "step": 2066 }, { "epoch": 0.18, "learning_rate": 1.8818201240296557e-05, "loss": 1.6289, "step": 2067 }, { "epoch": 0.18, "learning_rate": 1.881685207152229e-05, "loss": 1.5983, "step": 2068 }, { "epoch": 0.18, "learning_rate": 1.8815502181487358e-05, "loss": 1.6255, "step": 2069 }, { "epoch": 0.18, "learning_rate": 1.8814151570302187e-05, "loss": 1.675, "step": 2070 }, { "epoch": 0.18, "learning_rate": 1.8812800238077265e-05, "loss": 1.5106, "step": 2071 }, { "epoch": 0.18, "learning_rate": 1.8811448184923137e-05, "loss": 1.5425, "step": 2072 }, { "epoch": 0.18, "learning_rate": 1.881009541095041e-05, "loss": 1.5712, "step": 2073 }, { "epoch": 0.18, "learning_rate": 1.880874191626974e-05, "loss": 1.4985, "step": 2074 }, { "epoch": 0.18, "learning_rate": 1.8807387700991857e-05, "loss": 1.499, "step": 2075 }, { "epoch": 0.18, "learning_rate": 1.8806032765227535e-05, "loss": 1.6401, "step": 2076 }, { "epoch": 0.18, "learning_rate": 1.8804677109087622e-05, "loss": 1.6266, "step": 2077 }, { "epoch": 0.18, "learning_rate": 1.8804677109087622e-05, "loss": 1.5166, "step": 2078 }, { "epoch": 0.18, "learning_rate": 1.8803320732683013e-05, "loss": 1.5646, "step": 2079 }, { "epoch": 0.18, "learning_rate": 1.8801963636124667e-05, "loss": 1.5562, "step": 2080 }, { "epoch": 0.18, "learning_rate": 1.8800605819523602e-05, "loss": 1.7195, "step": 2081 }, { "epoch": 0.18, "learning_rate": 1.8799247282990885e-05, "loss": 1.7121, "step": 2082 }, { "epoch": 0.18, "learning_rate": 1.8797888026637667e-05, "loss": 1.5628, "step": 2083 }, { "epoch": 0.18, "learning_rate": 1.879652805057513e-05, "loss": 1.5299, "step": 2084 }, { "epoch": 0.18, "learning_rate": 1.879516735491453e-05, "loss": 1.5017, "step": 2085 }, { "epoch": 0.18, "learning_rate": 1.8793805939767178e-05, "loss": 1.5144, "step": 2086 }, { "epoch": 0.18, "learning_rate": 1.8792443805244445e-05, "loss": 1.6082, "step": 2087 }, { "epoch": 0.18, "learning_rate": 1.879108095145776e-05, "loss": 1.5734, "step": 2088 }, { "epoch": 0.18, "learning_rate": 1.8789717378518604e-05, "loss": 1.6449, "step": 2089 }, { "epoch": 0.18, "learning_rate": 1.8788353086538537e-05, "loss": 1.6664, "step": 2090 }, { "epoch": 0.18, "learning_rate": 1.8786988075629158e-05, "loss": 1.5863, "step": 2091 }, { "epoch": 0.18, "learning_rate": 1.878562234590213e-05, "loss": 1.558, "step": 2092 }, { "epoch": 0.18, "learning_rate": 1.8784255897469174e-05, "loss": 1.5928, "step": 2093 }, { "epoch": 0.18, "learning_rate": 1.8782888730442077e-05, "loss": 1.5912, "step": 2094 }, { "epoch": 0.18, "learning_rate": 1.8781520844932683e-05, "loss": 1.6881, "step": 2095 }, { "epoch": 0.19, "learning_rate": 1.878015224105288e-05, "loss": 1.6143, "step": 2096 }, { "epoch": 0.19, "learning_rate": 1.8778782918914634e-05, "loss": 1.4963, "step": 2097 }, { "epoch": 0.19, "learning_rate": 1.8777412878629967e-05, "loss": 1.6127, "step": 2098 }, { "epoch": 0.19, "learning_rate": 1.8776042120310943e-05, "loss": 1.5547, "step": 2099 }, { "epoch": 0.19, "learning_rate": 1.8774670644069707e-05, "loss": 1.6656, "step": 2100 }, { "epoch": 0.19, "learning_rate": 1.8773298450018444e-05, "loss": 1.664, "step": 2101 }, { "epoch": 0.19, "learning_rate": 1.8771925538269415e-05, "loss": 1.5443, "step": 2102 }, { "epoch": 0.19, "learning_rate": 1.877055190893492e-05, "loss": 1.4709, "step": 2103 }, { "epoch": 0.19, "learning_rate": 1.8769177562127337e-05, "loss": 1.4308, "step": 2104 }, { "epoch": 0.19, "learning_rate": 1.876780249795909e-05, "loss": 1.6046, "step": 2105 }, { "epoch": 0.19, "learning_rate": 1.8766426716542668e-05, "loss": 1.6756, "step": 2106 }, { "epoch": 0.19, "learning_rate": 1.8765050217990618e-05, "loss": 1.5559, "step": 2107 }, { "epoch": 0.19, "learning_rate": 1.876367300241554e-05, "loss": 1.5385, "step": 2108 }, { "epoch": 0.19, "learning_rate": 1.87622950699301e-05, "loss": 1.4723, "step": 2109 }, { "epoch": 0.19, "learning_rate": 1.8760916420647016e-05, "loss": 1.5801, "step": 2110 }, { "epoch": 0.19, "learning_rate": 1.875953705467907e-05, "loss": 1.6264, "step": 2111 }, { "epoch": 0.19, "learning_rate": 1.87581569721391e-05, "loss": 1.5954, "step": 2112 }, { "epoch": 0.19, "learning_rate": 1.875677617314001e-05, "loss": 1.5437, "step": 2113 }, { "epoch": 0.19, "learning_rate": 1.8755394657794745e-05, "loss": 1.5118, "step": 2114 }, { "epoch": 0.19, "learning_rate": 1.8754012426216327e-05, "loss": 1.605, "step": 2115 }, { "epoch": 0.19, "learning_rate": 1.8752629478517826e-05, "loss": 1.4686, "step": 2116 }, { "epoch": 0.19, "learning_rate": 1.8751245814812377e-05, "loss": 1.5578, "step": 2117 }, { "epoch": 0.19, "learning_rate": 1.8749861435213163e-05, "loss": 1.737, "step": 2118 }, { "epoch": 0.19, "learning_rate": 1.874847633983344e-05, "loss": 1.6022, "step": 2119 }, { "epoch": 0.19, "learning_rate": 1.8747090528786516e-05, "loss": 1.6851, "step": 2120 }, { "epoch": 0.19, "learning_rate": 1.874570400218575e-05, "loss": 1.5494, "step": 2121 }, { "epoch": 0.19, "learning_rate": 1.8744316760144575e-05, "loss": 1.5046, "step": 2122 }, { "epoch": 0.19, "learning_rate": 1.874292880277647e-05, "loss": 1.5593, "step": 2123 }, { "epoch": 0.19, "learning_rate": 1.8741540130194974e-05, "loss": 1.546, "step": 2124 }, { "epoch": 0.19, "learning_rate": 1.8740150742513693e-05, "loss": 1.6181, "step": 2125 }, { "epoch": 0.19, "learning_rate": 1.8738760639846278e-05, "loss": 1.5276, "step": 2126 }, { "epoch": 0.19, "learning_rate": 1.873736982230645e-05, "loss": 1.6499, "step": 2127 }, { "epoch": 0.19, "learning_rate": 1.8735978290007985e-05, "loss": 1.53, "step": 2128 }, { "epoch": 0.19, "learning_rate": 1.873458604306472e-05, "loss": 1.545, "step": 2129 }, { "epoch": 0.19, "learning_rate": 1.873319308159054e-05, "loss": 1.6074, "step": 2130 }, { "epoch": 0.19, "learning_rate": 1.8731799405699403e-05, "loss": 1.5923, "step": 2131 }, { "epoch": 0.19, "learning_rate": 1.8730405015505314e-05, "loss": 1.4903, "step": 2132 }, { "epoch": 0.19, "learning_rate": 1.872900991112234e-05, "loss": 1.607, "step": 2133 }, { "epoch": 0.19, "learning_rate": 1.8727614092664613e-05, "loss": 1.6439, "step": 2134 }, { "epoch": 0.19, "learning_rate": 1.872621756024631e-05, "loss": 1.5298, "step": 2135 }, { "epoch": 0.19, "learning_rate": 1.8724820313981682e-05, "loss": 1.4168, "step": 2136 }, { "epoch": 0.19, "learning_rate": 1.8723422353985024e-05, "loss": 1.5172, "step": 2137 }, { "epoch": 0.19, "learning_rate": 1.87220236803707e-05, "loss": 1.5759, "step": 2138 }, { "epoch": 0.19, "learning_rate": 1.8720624293253124e-05, "loss": 1.581, "step": 2139 }, { "epoch": 0.19, "learning_rate": 1.8719224192746776e-05, "loss": 1.5322, "step": 2140 }, { "epoch": 0.19, "learning_rate": 1.8717823378966188e-05, "loss": 1.5248, "step": 2141 }, { "epoch": 0.19, "learning_rate": 1.8716421852025963e-05, "loss": 1.5765, "step": 2142 }, { "epoch": 0.19, "learning_rate": 1.8715019612040737e-05, "loss": 1.6478, "step": 2143 }, { "epoch": 0.19, "learning_rate": 1.8713616659125227e-05, "loss": 1.6486, "step": 2144 }, { "epoch": 0.19, "learning_rate": 1.8712212993394206e-05, "loss": 1.4963, "step": 2145 }, { "epoch": 0.19, "learning_rate": 1.8710808614962493e-05, "loss": 1.5653, "step": 2146 }, { "epoch": 0.19, "learning_rate": 1.870940352394498e-05, "loss": 1.5047, "step": 2147 }, { "epoch": 0.19, "learning_rate": 1.8707997720456602e-05, "loss": 1.5932, "step": 2148 }, { "epoch": 0.19, "learning_rate": 1.870659120461237e-05, "loss": 1.455, "step": 2149 }, { "epoch": 0.19, "learning_rate": 1.8705183976527336e-05, "loss": 1.5409, "step": 2150 }, { "epoch": 0.19, "learning_rate": 1.8703776036316622e-05, "loss": 1.5198, "step": 2151 }, { "epoch": 0.19, "learning_rate": 1.87023673840954e-05, "loss": 1.5186, "step": 2152 }, { "epoch": 0.19, "learning_rate": 1.8700958019978913e-05, "loss": 1.4756, "step": 2153 }, { "epoch": 0.19, "learning_rate": 1.8699547944082447e-05, "loss": 1.5708, "step": 2154 }, { "epoch": 0.19, "learning_rate": 1.869813715652135e-05, "loss": 1.605, "step": 2155 }, { "epoch": 0.19, "learning_rate": 1.8696725657411037e-05, "loss": 1.6497, "step": 2156 }, { "epoch": 0.19, "learning_rate": 1.8695313446866972e-05, "loss": 1.5741, "step": 2157 }, { "epoch": 0.19, "learning_rate": 1.8693900525004686e-05, "loss": 1.5978, "step": 2158 }, { "epoch": 0.19, "learning_rate": 1.8692486891939757e-05, "loss": 1.6238, "step": 2159 }, { "epoch": 0.19, "learning_rate": 1.8691072547787826e-05, "loss": 1.6234, "step": 2160 }, { "epoch": 0.19, "learning_rate": 1.8689657492664602e-05, "loss": 1.4304, "step": 2161 }, { "epoch": 0.19, "learning_rate": 1.8688241726685836e-05, "loss": 1.554, "step": 2162 }, { "epoch": 0.19, "learning_rate": 1.8686825249967345e-05, "loss": 1.5384, "step": 2163 }, { "epoch": 0.19, "learning_rate": 1.8685408062625002e-05, "loss": 1.4781, "step": 2164 }, { "epoch": 0.19, "learning_rate": 1.8683990164774744e-05, "loss": 1.5714, "step": 2165 }, { "epoch": 0.19, "learning_rate": 1.868257155653256e-05, "loss": 1.6227, "step": 2166 }, { "epoch": 0.19, "learning_rate": 1.86811522380145e-05, "loss": 1.6319, "step": 2167 }, { "epoch": 0.19, "learning_rate": 1.8679732209336665e-05, "loss": 1.4951, "step": 2168 }, { "epoch": 0.19, "learning_rate": 1.867831147061523e-05, "loss": 1.5735, "step": 2169 }, { "epoch": 0.19, "learning_rate": 1.867689002196641e-05, "loss": 1.477, "step": 2170 }, { "epoch": 0.19, "learning_rate": 1.8675467863506492e-05, "loss": 1.5359, "step": 2171 }, { "epoch": 0.19, "learning_rate": 1.867404499535181e-05, "loss": 1.4757, "step": 2172 }, { "epoch": 0.19, "learning_rate": 1.8672621417618765e-05, "loss": 1.6693, "step": 2173 }, { "epoch": 0.19, "learning_rate": 1.8671197130423816e-05, "loss": 1.5707, "step": 2174 }, { "epoch": 0.19, "learning_rate": 1.8669772133883466e-05, "loss": 1.6007, "step": 2175 }, { "epoch": 0.19, "learning_rate": 1.8668346428114295e-05, "loss": 1.5316, "step": 2176 }, { "epoch": 0.19, "learning_rate": 1.8666920013232932e-05, "loss": 1.4784, "step": 2177 }, { "epoch": 0.19, "learning_rate": 1.8665492889356064e-05, "loss": 1.651, "step": 2178 }, { "epoch": 0.19, "learning_rate": 1.8664065056600433e-05, "loss": 1.5584, "step": 2179 }, { "epoch": 0.19, "learning_rate": 1.8662636515082844e-05, "loss": 1.5969, "step": 2180 }, { "epoch": 0.19, "learning_rate": 1.8661207264920162e-05, "loss": 1.4627, "step": 2181 }, { "epoch": 0.19, "learning_rate": 1.8659777306229303e-05, "loss": 1.5625, "step": 2182 }, { "epoch": 0.19, "learning_rate": 1.8658346639127246e-05, "loss": 1.4607, "step": 2183 }, { "epoch": 0.19, "learning_rate": 1.8656915263731025e-05, "loss": 1.5683, "step": 2184 }, { "epoch": 0.19, "learning_rate": 1.8655483180157733e-05, "loss": 1.5583, "step": 2185 }, { "epoch": 0.19, "learning_rate": 1.8654050388524524e-05, "loss": 1.5974, "step": 2186 }, { "epoch": 0.19, "learning_rate": 1.865261688894861e-05, "loss": 1.4879, "step": 2187 }, { "epoch": 0.19, "learning_rate": 1.8651182681547246e-05, "loss": 1.6195, "step": 2188 }, { "epoch": 0.19, "learning_rate": 1.864974776643777e-05, "loss": 1.5232, "step": 2189 }, { "epoch": 0.19, "learning_rate": 1.8648312143737556e-05, "loss": 1.6056, "step": 2190 }, { "epoch": 0.19, "learning_rate": 1.8646875813564054e-05, "loss": 1.4763, "step": 2191 }, { "epoch": 0.19, "learning_rate": 1.864543877603475e-05, "loss": 1.5717, "step": 2192 }, { "epoch": 0.19, "learning_rate": 1.8644001031267214e-05, "loss": 1.5291, "step": 2193 }, { "epoch": 0.19, "learning_rate": 1.8642562579379053e-05, "loss": 1.5289, "step": 2194 }, { "epoch": 0.19, "learning_rate": 1.8641123420487938e-05, "loss": 1.5023, "step": 2195 }, { "epoch": 0.19, "learning_rate": 1.8639683554711605e-05, "loss": 1.5077, "step": 2196 }, { "epoch": 0.19, "learning_rate": 1.8638242982167837e-05, "loss": 1.6152, "step": 2197 }, { "epoch": 0.19, "learning_rate": 1.8636801702974483e-05, "loss": 1.6713, "step": 2198 }, { "epoch": 0.19, "learning_rate": 1.863535971724944e-05, "loss": 1.6817, "step": 2199 }, { "epoch": 0.19, "learning_rate": 1.8633917025110676e-05, "loss": 1.4918, "step": 2200 }, { "epoch": 0.19, "learning_rate": 1.8632473626676212e-05, "loss": 1.5907, "step": 2201 }, { "epoch": 0.19, "learning_rate": 1.8631029522064116e-05, "loss": 1.6186, "step": 2202 }, { "epoch": 0.19, "learning_rate": 1.8629584711392528e-05, "loss": 1.5438, "step": 2203 }, { "epoch": 0.19, "learning_rate": 1.8628139194779646e-05, "loss": 1.4817, "step": 2204 }, { "epoch": 0.19, "learning_rate": 1.862669297234371e-05, "loss": 1.5902, "step": 2205 }, { "epoch": 0.19, "learning_rate": 1.8625246044203033e-05, "loss": 1.5855, "step": 2206 }, { "epoch": 0.19, "learning_rate": 1.862379841047598e-05, "loss": 1.5233, "step": 2207 }, { "epoch": 0.19, "learning_rate": 1.862235007128097e-05, "loss": 1.6368, "step": 2208 }, { "epoch": 0.2, "learning_rate": 1.8620901026736493e-05, "loss": 1.6098, "step": 2209 }, { "epoch": 0.2, "learning_rate": 1.861945127696108e-05, "loss": 1.6223, "step": 2210 }, { "epoch": 0.2, "learning_rate": 1.8618000822073333e-05, "loss": 1.5779, "step": 2211 }, { "epoch": 0.2, "learning_rate": 1.8616549662191904e-05, "loss": 1.5278, "step": 2212 }, { "epoch": 0.2, "learning_rate": 1.8615097797435505e-05, "loss": 1.4584, "step": 2213 }, { "epoch": 0.2, "learning_rate": 1.8613645227922902e-05, "loss": 1.5673, "step": 2214 }, { "epoch": 0.2, "learning_rate": 1.861219195377293e-05, "loss": 1.6851, "step": 2215 }, { "epoch": 0.2, "learning_rate": 1.8610737975104463e-05, "loss": 1.5079, "step": 2216 }, { "epoch": 0.2, "learning_rate": 1.8609283292036454e-05, "loss": 1.5282, "step": 2217 }, { "epoch": 0.2, "learning_rate": 1.8607827904687897e-05, "loss": 1.5486, "step": 2218 }, { "epoch": 0.2, "learning_rate": 1.860637181317785e-05, "loss": 1.6062, "step": 2219 }, { "epoch": 0.2, "learning_rate": 1.860491501762543e-05, "loss": 1.5177, "step": 2220 }, { "epoch": 0.2, "learning_rate": 1.8603457518149808e-05, "loss": 1.5384, "step": 2221 }, { "epoch": 0.2, "learning_rate": 1.8601999314870213e-05, "loss": 1.528, "step": 2222 }, { "epoch": 0.2, "learning_rate": 1.860054040790594e-05, "loss": 1.608, "step": 2223 }, { "epoch": 0.2, "learning_rate": 1.859908079737633e-05, "loss": 1.4832, "step": 2224 }, { "epoch": 0.2, "learning_rate": 1.8597620483400782e-05, "loss": 1.6633, "step": 2225 }, { "epoch": 0.2, "learning_rate": 1.8596159466098766e-05, "loss": 1.5247, "step": 2226 }, { "epoch": 0.2, "learning_rate": 1.8594697745589793e-05, "loss": 1.5957, "step": 2227 }, { "epoch": 0.2, "learning_rate": 1.8593235321993438e-05, "loss": 1.4465, "step": 2228 }, { "epoch": 0.2, "learning_rate": 1.859177219542934e-05, "loss": 1.6573, "step": 2229 }, { "epoch": 0.2, "learning_rate": 1.8590308366017186e-05, "loss": 1.6707, "step": 2230 }, { "epoch": 0.2, "learning_rate": 1.8588843833876725e-05, "loss": 1.5676, "step": 2231 }, { "epoch": 0.2, "learning_rate": 1.8587378599127763e-05, "loss": 1.5694, "step": 2232 }, { "epoch": 0.2, "learning_rate": 1.8585912661890163e-05, "loss": 1.6384, "step": 2233 }, { "epoch": 0.2, "learning_rate": 1.8584446022283845e-05, "loss": 1.6126, "step": 2234 }, { "epoch": 0.2, "learning_rate": 1.8582978680428788e-05, "loss": 1.5626, "step": 2235 }, { "epoch": 0.2, "learning_rate": 1.858151063644503e-05, "loss": 1.5191, "step": 2236 }, { "epoch": 0.2, "learning_rate": 1.858004189045266e-05, "loss": 1.5023, "step": 2237 }, { "epoch": 0.2, "learning_rate": 1.8578572442571827e-05, "loss": 1.6325, "step": 2238 }, { "epoch": 0.2, "learning_rate": 1.8577102292922747e-05, "loss": 1.4843, "step": 2239 }, { "epoch": 0.2, "learning_rate": 1.8575631441625676e-05, "loss": 1.5636, "step": 2240 }, { "epoch": 0.2, "learning_rate": 1.857415988880094e-05, "loss": 1.5751, "step": 2241 }, { "epoch": 0.2, "learning_rate": 1.8572687634568922e-05, "loss": 1.556, "step": 2242 }, { "epoch": 0.2, "learning_rate": 1.8571214679050055e-05, "loss": 1.5875, "step": 2243 }, { "epoch": 0.2, "learning_rate": 1.856974102236484e-05, "loss": 1.6056, "step": 2244 }, { "epoch": 0.2, "learning_rate": 1.856826666463382e-05, "loss": 1.5384, "step": 2245 }, { "epoch": 0.2, "learning_rate": 1.856679160597761e-05, "loss": 1.7134, "step": 2246 }, { "epoch": 0.2, "learning_rate": 1.856531584651688e-05, "loss": 1.5238, "step": 2247 }, { "epoch": 0.2, "learning_rate": 1.8563839386372348e-05, "loss": 1.5277, "step": 2248 }, { "epoch": 0.2, "learning_rate": 1.85623622256648e-05, "loss": 1.6141, "step": 2249 }, { "epoch": 0.2, "learning_rate": 1.8560884364515072e-05, "loss": 1.4328, "step": 2250 }, { "epoch": 0.2, "learning_rate": 1.855940580304406e-05, "loss": 1.5361, "step": 2251 }, { "epoch": 0.2, "learning_rate": 1.855792654137272e-05, "loss": 1.6165, "step": 2252 }, { "epoch": 0.2, "learning_rate": 1.8556446579622064e-05, "loss": 1.5555, "step": 2253 }, { "epoch": 0.2, "learning_rate": 1.8554965917913154e-05, "loss": 1.4478, "step": 2254 }, { "epoch": 0.2, "learning_rate": 1.8553484556367117e-05, "loss": 1.5813, "step": 2255 }, { "epoch": 0.2, "learning_rate": 1.8552002495105137e-05, "loss": 1.7386, "step": 2256 }, { "epoch": 0.2, "learning_rate": 1.8550519734248454e-05, "loss": 1.5707, "step": 2257 }, { "epoch": 0.2, "learning_rate": 1.8549036273918366e-05, "loss": 1.5634, "step": 2258 }, { "epoch": 0.2, "learning_rate": 1.8547552114236224e-05, "loss": 1.6888, "step": 2259 }, { "epoch": 0.2, "learning_rate": 1.854606725532344e-05, "loss": 1.4984, "step": 2260 }, { "epoch": 0.2, "learning_rate": 1.8544581697301486e-05, "loss": 1.6689, "step": 2261 }, { "epoch": 0.2, "learning_rate": 1.854309544029188e-05, "loss": 1.5094, "step": 2262 }, { "epoch": 0.2, "learning_rate": 1.854160848441622e-05, "loss": 1.5908, "step": 2263 }, { "epoch": 0.2, "learning_rate": 1.8540120829796125e-05, "loss": 1.7627, "step": 2264 }, { "epoch": 0.2, "learning_rate": 1.853863247655331e-05, "loss": 1.5411, "step": 2265 }, { "epoch": 0.2, "learning_rate": 1.8537143424809524e-05, "loss": 1.6768, "step": 2266 }, { "epoch": 0.2, "learning_rate": 1.8535653674686575e-05, "loss": 1.4438, "step": 2267 }, { "epoch": 0.2, "learning_rate": 1.8534163226306334e-05, "loss": 1.5092, "step": 2268 }, { "epoch": 0.2, "learning_rate": 1.8532672079790727e-05, "loss": 1.541, "step": 2269 }, { "epoch": 0.2, "learning_rate": 1.853118023526174e-05, "loss": 1.564, "step": 2270 }, { "epoch": 0.2, "learning_rate": 1.8529687692841405e-05, "loss": 1.5434, "step": 2271 }, { "epoch": 0.2, "learning_rate": 1.8528194452651825e-05, "loss": 1.5143, "step": 2272 }, { "epoch": 0.2, "learning_rate": 1.8526700514815153e-05, "loss": 1.6374, "step": 2273 }, { "epoch": 0.2, "learning_rate": 1.85252058794536e-05, "loss": 1.561, "step": 2274 }, { "epoch": 0.2, "learning_rate": 1.852371054668944e-05, "loss": 1.5621, "step": 2275 }, { "epoch": 0.2, "learning_rate": 1.852221451664499e-05, "loss": 1.5957, "step": 2276 }, { "epoch": 0.2, "learning_rate": 1.8520717789442635e-05, "loss": 1.5146, "step": 2277 }, { "epoch": 0.2, "learning_rate": 1.8519220365204817e-05, "loss": 1.5132, "step": 2278 }, { "epoch": 0.2, "learning_rate": 1.851772224405403e-05, "loss": 1.4707, "step": 2279 }, { "epoch": 0.2, "learning_rate": 1.8516223426112826e-05, "loss": 1.6438, "step": 2280 }, { "epoch": 0.2, "learning_rate": 1.8514723911503817e-05, "loss": 1.6297, "step": 2281 }, { "epoch": 0.2, "learning_rate": 1.851322370034967e-05, "loss": 1.4572, "step": 2282 }, { "epoch": 0.2, "learning_rate": 1.8511722792773112e-05, "loss": 1.593, "step": 2283 }, { "epoch": 0.2, "learning_rate": 1.8510221188896924e-05, "loss": 1.6225, "step": 2284 }, { "epoch": 0.2, "learning_rate": 1.8508718888843944e-05, "loss": 1.4436, "step": 2285 }, { "epoch": 0.2, "learning_rate": 1.8507215892737065e-05, "loss": 1.6247, "step": 2286 }, { "epoch": 0.2, "learning_rate": 1.850571220069924e-05, "loss": 1.6835, "step": 2287 }, { "epoch": 0.2, "learning_rate": 1.8504207812853478e-05, "loss": 1.5817, "step": 2288 }, { "epoch": 0.2, "learning_rate": 1.850270272932285e-05, "loss": 1.6789, "step": 2289 }, { "epoch": 0.2, "learning_rate": 1.8501196950230472e-05, "loss": 1.5681, "step": 2290 }, { "epoch": 0.2, "learning_rate": 1.8499690475699525e-05, "loss": 1.6175, "step": 2291 }, { "epoch": 0.2, "learning_rate": 1.849818330585325e-05, "loss": 1.6278, "step": 2292 }, { "epoch": 0.2, "learning_rate": 1.849667544081494e-05, "loss": 1.6375, "step": 2293 }, { "epoch": 0.2, "learning_rate": 1.849516688070794e-05, "loss": 1.5165, "step": 2294 }, { "epoch": 0.2, "learning_rate": 1.8493657625655667e-05, "loss": 1.5083, "step": 2295 }, { "epoch": 0.2, "learning_rate": 1.8492147675781578e-05, "loss": 1.5311, "step": 2296 }, { "epoch": 0.2, "learning_rate": 1.8490637031209197e-05, "loss": 1.4683, "step": 2297 }, { "epoch": 0.2, "learning_rate": 1.8489125692062097e-05, "loss": 1.6217, "step": 2298 }, { "epoch": 0.2, "learning_rate": 1.848761365846392e-05, "loss": 1.5132, "step": 2299 }, { "epoch": 0.2, "learning_rate": 1.8486100930538354e-05, "loss": 1.4301, "step": 2300 }, { "epoch": 0.2, "learning_rate": 1.8484587508409145e-05, "loss": 1.6351, "step": 2301 }, { "epoch": 0.2, "learning_rate": 1.8483073392200102e-05, "loss": 1.7058, "step": 2302 }, { "epoch": 0.2, "learning_rate": 1.8481558582035086e-05, "loss": 1.6083, "step": 2303 }, { "epoch": 0.2, "learning_rate": 1.8480043078038016e-05, "loss": 1.5482, "step": 2304 }, { "epoch": 0.2, "learning_rate": 1.8478526880332866e-05, "loss": 1.4979, "step": 2305 }, { "epoch": 0.2, "learning_rate": 1.847700998904367e-05, "loss": 1.5323, "step": 2306 }, { "epoch": 0.2, "learning_rate": 1.8475492404294515e-05, "loss": 1.5673, "step": 2307 }, { "epoch": 0.2, "learning_rate": 1.8473974126209548e-05, "loss": 1.6658, "step": 2308 }, { "epoch": 0.2, "learning_rate": 1.8472455154912975e-05, "loss": 1.5194, "step": 2309 }, { "epoch": 0.2, "learning_rate": 1.8470935490529047e-05, "loss": 1.5943, "step": 2310 }, { "epoch": 0.2, "learning_rate": 1.8469415133182084e-05, "loss": 1.6341, "step": 2311 }, { "epoch": 0.2, "learning_rate": 1.846789408299646e-05, "loss": 1.5471, "step": 2312 }, { "epoch": 0.2, "learning_rate": 1.8466372340096603e-05, "loss": 1.6902, "step": 2313 }, { "epoch": 0.2, "learning_rate": 1.8464849904606997e-05, "loss": 1.6438, "step": 2314 }, { "epoch": 0.2, "learning_rate": 1.8463326776652188e-05, "loss": 1.4961, "step": 2315 }, { "epoch": 0.2, "learning_rate": 1.8461802956356775e-05, "loss": 1.5847, "step": 2316 }, { "epoch": 0.2, "learning_rate": 1.846027844384541e-05, "loss": 1.5336, "step": 2317 }, { "epoch": 0.2, "learning_rate": 1.8458753239242804e-05, "loss": 1.5523, "step": 2318 }, { "epoch": 0.2, "learning_rate": 1.8457227342673735e-05, "loss": 1.3341, "step": 2319 }, { "epoch": 0.2, "learning_rate": 1.8455700754263023e-05, "loss": 1.5041, "step": 2320 }, { "epoch": 0.2, "learning_rate": 1.8454173474135545e-05, "loss": 1.5378, "step": 2321 }, { "epoch": 0.21, "learning_rate": 1.845264550241625e-05, "loss": 1.462, "step": 2322 }, { "epoch": 0.21, "learning_rate": 1.8451116839230128e-05, "loss": 1.5479, "step": 2323 }, { "epoch": 0.21, "learning_rate": 1.8449587484702232e-05, "loss": 1.5963, "step": 2324 }, { "epoch": 0.21, "learning_rate": 1.844805743895767e-05, "loss": 1.6147, "step": 2325 }, { "epoch": 0.21, "learning_rate": 1.8446526702121607e-05, "loss": 1.5531, "step": 2326 }, { "epoch": 0.21, "learning_rate": 1.844499527431926e-05, "loss": 1.607, "step": 2327 }, { "epoch": 0.21, "learning_rate": 1.8443463155675917e-05, "loss": 1.5359, "step": 2328 }, { "epoch": 0.21, "learning_rate": 1.8441930346316907e-05, "loss": 1.4976, "step": 2329 }, { "epoch": 0.21, "learning_rate": 1.844039684636762e-05, "loss": 1.4984, "step": 2330 }, { "epoch": 0.21, "learning_rate": 1.8438862655953505e-05, "loss": 1.5293, "step": 2331 }, { "epoch": 0.21, "learning_rate": 1.843732777520007e-05, "loss": 1.4276, "step": 2332 }, { "epoch": 0.21, "learning_rate": 1.843579220423287e-05, "loss": 1.5842, "step": 2333 }, { "epoch": 0.21, "learning_rate": 1.8434255943177523e-05, "loss": 1.5251, "step": 2334 }, { "epoch": 0.21, "learning_rate": 1.8432718992159703e-05, "loss": 1.5381, "step": 2335 }, { "epoch": 0.21, "learning_rate": 1.8431181351305145e-05, "loss": 1.5224, "step": 2336 }, { "epoch": 0.21, "learning_rate": 1.8429643020739627e-05, "loss": 1.6017, "step": 2337 }, { "epoch": 0.21, "learning_rate": 1.8428104000588994e-05, "loss": 1.5024, "step": 2338 }, { "epoch": 0.21, "learning_rate": 1.842656429097915e-05, "loss": 1.5007, "step": 2339 }, { "epoch": 0.21, "learning_rate": 1.842502389203605e-05, "loss": 1.6303, "step": 2340 }, { "epoch": 0.21, "learning_rate": 1.84234828038857e-05, "loss": 1.5228, "step": 2341 }, { "epoch": 0.21, "learning_rate": 1.842194102665417e-05, "loss": 1.5884, "step": 2342 }, { "epoch": 0.21, "learning_rate": 1.8420398560467588e-05, "loss": 1.5333, "step": 2343 }, { "epoch": 0.21, "learning_rate": 1.8418855405452135e-05, "loss": 1.5095, "step": 2344 }, { "epoch": 0.21, "learning_rate": 1.8417311561734047e-05, "loss": 1.5851, "step": 2345 }, { "epoch": 0.21, "learning_rate": 1.8415767029439617e-05, "loss": 1.6396, "step": 2346 }, { "epoch": 0.21, "learning_rate": 1.8414221808695197e-05, "loss": 1.6003, "step": 2347 }, { "epoch": 0.21, "learning_rate": 1.841267589962719e-05, "loss": 1.6008, "step": 2348 }, { "epoch": 0.21, "learning_rate": 1.8411129302362065e-05, "loss": 1.5969, "step": 2349 }, { "epoch": 0.21, "learning_rate": 1.8409582017026336e-05, "loss": 1.5214, "step": 2350 }, { "epoch": 0.21, "learning_rate": 1.840803404374658e-05, "loss": 1.5987, "step": 2351 }, { "epoch": 0.21, "learning_rate": 1.8406485382649427e-05, "loss": 1.5734, "step": 2352 }, { "epoch": 0.21, "learning_rate": 1.8404936033861568e-05, "loss": 1.4573, "step": 2353 }, { "epoch": 0.21, "learning_rate": 1.8403385997509743e-05, "loss": 1.5545, "step": 2354 }, { "epoch": 0.21, "learning_rate": 1.8401835273720757e-05, "loss": 1.5506, "step": 2355 }, { "epoch": 0.21, "learning_rate": 1.840028386262146e-05, "loss": 1.6293, "step": 2356 }, { "epoch": 0.21, "learning_rate": 1.8398731764338773e-05, "loss": 1.5357, "step": 2357 }, { "epoch": 0.21, "learning_rate": 1.839717897899966e-05, "loss": 1.4836, "step": 2358 }, { "epoch": 0.21, "learning_rate": 1.839562550673115e-05, "loss": 1.4943, "step": 2359 }, { "epoch": 0.21, "learning_rate": 1.839407134766032e-05, "loss": 1.5486, "step": 2360 }, { "epoch": 0.21, "learning_rate": 1.839251650191431e-05, "loss": 1.5678, "step": 2361 }, { "epoch": 0.21, "learning_rate": 1.8390960969620314e-05, "loss": 1.6553, "step": 2362 }, { "epoch": 0.21, "learning_rate": 1.838940475090558e-05, "loss": 1.4734, "step": 2363 }, { "epoch": 0.21, "learning_rate": 1.8387847845897417e-05, "loss": 1.5387, "step": 2364 }, { "epoch": 0.21, "learning_rate": 1.8386290254723183e-05, "loss": 1.5233, "step": 2365 }, { "epoch": 0.21, "learning_rate": 1.8384731977510304e-05, "loss": 1.5991, "step": 2366 }, { "epoch": 0.21, "learning_rate": 1.8383173014386245e-05, "loss": 1.5279, "step": 2367 }, { "epoch": 0.21, "learning_rate": 1.8381613365478544e-05, "loss": 1.5638, "step": 2368 }, { "epoch": 0.21, "learning_rate": 1.8380053030914783e-05, "loss": 1.5188, "step": 2369 }, { "epoch": 0.21, "learning_rate": 1.837849201082261e-05, "loss": 1.6063, "step": 2370 }, { "epoch": 0.21, "learning_rate": 1.837693030532972e-05, "loss": 1.5165, "step": 2371 }, { "epoch": 0.21, "learning_rate": 1.837536791456387e-05, "loss": 1.5676, "step": 2372 }, { "epoch": 0.21, "learning_rate": 1.8373804838652865e-05, "loss": 1.534, "step": 2373 }, { "epoch": 0.21, "learning_rate": 1.8372241077724582e-05, "loss": 1.7443, "step": 2374 }, { "epoch": 0.21, "learning_rate": 1.8370676631906936e-05, "loss": 1.5554, "step": 2375 }, { "epoch": 0.21, "learning_rate": 1.836911150132791e-05, "loss": 1.4156, "step": 2376 }, { "epoch": 0.21, "learning_rate": 1.836754568611554e-05, "loss": 1.62, "step": 2377 }, { "epoch": 0.21, "learning_rate": 1.8365979186397914e-05, "loss": 1.5898, "step": 2378 }, { "epoch": 0.21, "learning_rate": 1.836441200230318e-05, "loss": 1.5544, "step": 2379 }, { "epoch": 0.21, "learning_rate": 1.8362844133959545e-05, "loss": 1.6404, "step": 2380 }, { "epoch": 0.21, "learning_rate": 1.8361275581495265e-05, "loss": 1.5164, "step": 2381 }, { "epoch": 0.21, "learning_rate": 1.8359706345038654e-05, "loss": 1.5114, "step": 2382 }, { "epoch": 0.21, "learning_rate": 1.8358136424718083e-05, "loss": 1.6144, "step": 2383 }, { "epoch": 0.21, "learning_rate": 1.8356565820661976e-05, "loss": 1.4525, "step": 2384 }, { "epoch": 0.21, "learning_rate": 1.8354994532998826e-05, "loss": 1.5802, "step": 2385 }, { "epoch": 0.21, "learning_rate": 1.8353422561857167e-05, "loss": 1.5884, "step": 2386 }, { "epoch": 0.21, "learning_rate": 1.835184990736559e-05, "loss": 1.587, "step": 2387 }, { "epoch": 0.21, "learning_rate": 1.835027656965275e-05, "loss": 1.5126, "step": 2388 }, { "epoch": 0.21, "learning_rate": 1.834870254884735e-05, "loss": 1.4798, "step": 2389 }, { "epoch": 0.21, "learning_rate": 1.8347127845078154e-05, "loss": 1.552, "step": 2390 }, { "epoch": 0.21, "learning_rate": 1.8345552458473982e-05, "loss": 1.4993, "step": 2391 }, { "epoch": 0.21, "learning_rate": 1.8343976389163708e-05, "loss": 1.5671, "step": 2392 }, { "epoch": 0.21, "learning_rate": 1.834239963727626e-05, "loss": 1.5362, "step": 2393 }, { "epoch": 0.21, "learning_rate": 1.8340822202940623e-05, "loss": 1.4731, "step": 2394 }, { "epoch": 0.21, "learning_rate": 1.8339244086285843e-05, "loss": 1.5352, "step": 2395 }, { "epoch": 0.21, "learning_rate": 1.833766528744101e-05, "loss": 1.6145, "step": 2396 }, { "epoch": 0.21, "learning_rate": 1.8336085806535287e-05, "loss": 1.5726, "step": 2397 }, { "epoch": 0.21, "learning_rate": 1.8334505643697876e-05, "loss": 1.6101, "step": 2398 }, { "epoch": 0.21, "learning_rate": 1.8332924799058042e-05, "loss": 1.5512, "step": 2399 }, { "epoch": 0.21, "learning_rate": 1.8331343272745113e-05, "loss": 1.5842, "step": 2400 }, { "epoch": 0.21, "learning_rate": 1.832976106488846e-05, "loss": 1.7306, "step": 2401 }, { "epoch": 0.21, "learning_rate": 1.8328178175617508e-05, "loss": 1.537, "step": 2402 }, { "epoch": 0.21, "learning_rate": 1.832659460506176e-05, "loss": 1.4807, "step": 2403 }, { "epoch": 0.21, "learning_rate": 1.8325010353350747e-05, "loss": 1.524, "step": 2404 }, { "epoch": 0.21, "learning_rate": 1.8323425420614074e-05, "loss": 1.4639, "step": 2405 }, { "epoch": 0.21, "learning_rate": 1.83218398069814e-05, "loss": 1.5111, "step": 2406 }, { "epoch": 0.21, "learning_rate": 1.8320253512582425e-05, "loss": 1.5909, "step": 2407 }, { "epoch": 0.21, "learning_rate": 1.8318666537546924e-05, "loss": 1.4753, "step": 2408 }, { "epoch": 0.21, "learning_rate": 1.8317078882004716e-05, "loss": 1.6066, "step": 2409 }, { "epoch": 0.21, "learning_rate": 1.831549054608568e-05, "loss": 1.6861, "step": 2410 }, { "epoch": 0.21, "learning_rate": 1.831390152991975e-05, "loss": 1.576, "step": 2411 }, { "epoch": 0.21, "learning_rate": 1.8312311833636912e-05, "loss": 1.5688, "step": 2412 }, { "epoch": 0.21, "learning_rate": 1.8310721457367215e-05, "loss": 1.4934, "step": 2413 }, { "epoch": 0.21, "learning_rate": 1.8309130401240757e-05, "loss": 1.6766, "step": 2414 }, { "epoch": 0.21, "learning_rate": 1.8307538665387695e-05, "loss": 1.5306, "step": 2415 }, { "epoch": 0.21, "learning_rate": 1.8305946249938236e-05, "loss": 1.5049, "step": 2416 }, { "epoch": 0.21, "learning_rate": 1.8304353155022654e-05, "loss": 1.6168, "step": 2417 }, { "epoch": 0.21, "learning_rate": 1.830275938077127e-05, "loss": 1.6406, "step": 2418 }, { "epoch": 0.21, "learning_rate": 1.8301164927314457e-05, "loss": 1.4908, "step": 2419 }, { "epoch": 0.21, "learning_rate": 1.8299569794782658e-05, "loss": 1.3975, "step": 2420 }, { "epoch": 0.21, "learning_rate": 1.8297973983306355e-05, "loss": 1.4772, "step": 2421 }, { "epoch": 0.21, "learning_rate": 1.8296377493016096e-05, "loss": 1.5801, "step": 2422 }, { "epoch": 0.21, "learning_rate": 1.829478032404248e-05, "loss": 1.6299, "step": 2423 }, { "epoch": 0.21, "learning_rate": 1.8293182476516166e-05, "loss": 1.5737, "step": 2424 }, { "epoch": 0.21, "learning_rate": 1.8291583950567864e-05, "loss": 1.5557, "step": 2425 }, { "epoch": 0.21, "learning_rate": 1.828998474632834e-05, "loss": 1.5873, "step": 2426 }, { "epoch": 0.21, "learning_rate": 1.8288384863928417e-05, "loss": 1.4689, "step": 2427 }, { "epoch": 0.21, "learning_rate": 1.8286784303498975e-05, "loss": 1.4972, "step": 2428 }, { "epoch": 0.21, "learning_rate": 1.8285183065170945e-05, "loss": 1.5853, "step": 2429 }, { "epoch": 0.21, "learning_rate": 1.8283581149075315e-05, "loss": 1.4962, "step": 2430 }, { "epoch": 0.21, "learning_rate": 1.8281978555343134e-05, "loss": 1.7096, "step": 2431 }, { "epoch": 0.21, "learning_rate": 1.82803752841055e-05, "loss": 1.4856, "step": 2432 }, { "epoch": 0.21, "learning_rate": 1.8278771335493565e-05, "loss": 1.5704, "step": 2433 }, { "epoch": 0.21, "learning_rate": 1.827716670963854e-05, "loss": 1.6446, "step": 2434 }, { "epoch": 0.22, "learning_rate": 1.8275561406671697e-05, "loss": 1.563, "step": 2435 }, { "epoch": 0.22, "learning_rate": 1.8273955426724352e-05, "loss": 1.5345, "step": 2436 }, { "epoch": 0.22, "learning_rate": 1.8272348769927883e-05, "loss": 1.7459, "step": 2437 }, { "epoch": 0.22, "learning_rate": 1.8270741436413722e-05, "loss": 1.443, "step": 2438 }, { "epoch": 0.22, "learning_rate": 1.826913342631336e-05, "loss": 1.5373, "step": 2439 }, { "epoch": 0.22, "learning_rate": 1.8267524739758332e-05, "loss": 1.575, "step": 2440 }, { "epoch": 0.22, "learning_rate": 1.8265915376880247e-05, "loss": 1.5332, "step": 2441 }, { "epoch": 0.22, "learning_rate": 1.826430533781075e-05, "loss": 1.5054, "step": 2442 }, { "epoch": 0.22, "learning_rate": 1.8262694622681554e-05, "loss": 1.6135, "step": 2443 }, { "epoch": 0.22, "learning_rate": 1.8261083231624423e-05, "loss": 1.6279, "step": 2444 }, { "epoch": 0.22, "learning_rate": 1.8259471164771172e-05, "loss": 1.5276, "step": 2445 }, { "epoch": 0.22, "learning_rate": 1.825785842225368e-05, "loss": 1.454, "step": 2446 }, { "epoch": 0.22, "learning_rate": 1.825624500420388e-05, "loss": 1.5316, "step": 2447 }, { "epoch": 0.22, "learning_rate": 1.825463091075375e-05, "loss": 1.4368, "step": 2448 }, { "epoch": 0.22, "learning_rate": 1.8253016142035336e-05, "loss": 1.5336, "step": 2449 }, { "epoch": 0.22, "learning_rate": 1.8251400698180733e-05, "loss": 1.5366, "step": 2450 }, { "epoch": 0.22, "learning_rate": 1.8249784579322088e-05, "loss": 1.5475, "step": 2451 }, { "epoch": 0.22, "learning_rate": 1.8248167785591616e-05, "loss": 1.5428, "step": 2452 }, { "epoch": 0.22, "learning_rate": 1.8246550317121565e-05, "loss": 1.607, "step": 2453 }, { "epoch": 0.22, "learning_rate": 1.8244932174044264e-05, "loss": 1.5688, "step": 2454 }, { "epoch": 0.22, "learning_rate": 1.8243313356492082e-05, "loss": 1.5553, "step": 2455 }, { "epoch": 0.22, "learning_rate": 1.824169386459744e-05, "loss": 1.4719, "step": 2456 }, { "epoch": 0.22, "learning_rate": 1.8240073698492826e-05, "loss": 1.5267, "step": 2457 }, { "epoch": 0.22, "learning_rate": 1.8238452858310775e-05, "loss": 1.5455, "step": 2458 }, { "epoch": 0.22, "learning_rate": 1.8236831344183882e-05, "loss": 1.6111, "step": 2459 }, { "epoch": 0.22, "learning_rate": 1.8235209156244794e-05, "loss": 1.6048, "step": 2460 }, { "epoch": 0.22, "learning_rate": 1.823358629462621e-05, "loss": 1.6172, "step": 2461 }, { "epoch": 0.22, "learning_rate": 1.823196275946089e-05, "loss": 1.5687, "step": 2462 }, { "epoch": 0.22, "learning_rate": 1.823033855088165e-05, "loss": 1.7236, "step": 2463 }, { "epoch": 0.22, "learning_rate": 1.8228713669021353e-05, "loss": 1.5888, "step": 2464 }, { "epoch": 0.22, "learning_rate": 1.8227088114012926e-05, "loss": 1.5135, "step": 2465 }, { "epoch": 0.22, "learning_rate": 1.8225461885989345e-05, "loss": 1.6009, "step": 2466 }, { "epoch": 0.22, "learning_rate": 1.8223834985083644e-05, "loss": 1.5005, "step": 2467 }, { "epoch": 0.22, "learning_rate": 1.8222207411428913e-05, "loss": 1.6342, "step": 2468 }, { "epoch": 0.22, "learning_rate": 1.822057916515829e-05, "loss": 1.5455, "step": 2469 }, { "epoch": 0.22, "learning_rate": 1.821895024640498e-05, "loss": 1.552, "step": 2470 }, { "epoch": 0.22, "learning_rate": 1.8217320655302232e-05, "loss": 1.5695, "step": 2471 }, { "epoch": 0.22, "learning_rate": 1.8215690391983356e-05, "loss": 1.5147, "step": 2472 }, { "epoch": 0.22, "learning_rate": 1.8214059456581713e-05, "loss": 1.4658, "step": 2473 }, { "epoch": 0.22, "learning_rate": 1.8212427849230726e-05, "loss": 1.5587, "step": 2474 }, { "epoch": 0.22, "learning_rate": 1.8210795570063865e-05, "loss": 1.49, "step": 2475 }, { "epoch": 0.22, "learning_rate": 1.820916261921466e-05, "loss": 1.4729, "step": 2476 }, { "epoch": 0.22, "learning_rate": 1.8207528996816692e-05, "loss": 1.5357, "step": 2477 }, { "epoch": 0.22, "learning_rate": 1.82058947030036e-05, "loss": 1.4797, "step": 2478 }, { "epoch": 0.22, "learning_rate": 1.8204259737909075e-05, "loss": 1.3465, "step": 2479 }, { "epoch": 0.22, "learning_rate": 1.8202624101666867e-05, "loss": 1.655, "step": 2480 }, { "epoch": 0.22, "learning_rate": 1.8200987794410777e-05, "loss": 1.4867, "step": 2481 }, { "epoch": 0.22, "learning_rate": 1.819935081627467e-05, "loss": 1.5367, "step": 2482 }, { "epoch": 0.22, "learning_rate": 1.819771316739245e-05, "loss": 1.4853, "step": 2483 }, { "epoch": 0.22, "learning_rate": 1.8196074847898092e-05, "loss": 1.5296, "step": 2484 }, { "epoch": 0.22, "learning_rate": 1.819443585792561e-05, "loss": 1.5398, "step": 2485 }, { "epoch": 0.22, "learning_rate": 1.819279619760909e-05, "loss": 1.5263, "step": 2486 }, { "epoch": 0.22, "learning_rate": 1.8191155867082654e-05, "loss": 1.6309, "step": 2487 }, { "epoch": 0.22, "learning_rate": 1.8189514866480496e-05, "loss": 1.6104, "step": 2488 }, { "epoch": 0.22, "learning_rate": 1.8187873195936856e-05, "loss": 1.5058, "step": 2489 }, { "epoch": 0.22, "learning_rate": 1.8186230855586033e-05, "loss": 1.5415, "step": 2490 }, { "epoch": 0.22, "learning_rate": 1.8184587845562372e-05, "loss": 1.4785, "step": 2491 }, { "epoch": 0.22, "learning_rate": 1.8182944166000286e-05, "loss": 1.6332, "step": 2492 }, { "epoch": 0.22, "learning_rate": 1.818129981703423e-05, "loss": 1.5541, "step": 2493 }, { "epoch": 0.22, "learning_rate": 1.8179654798798726e-05, "loss": 1.5256, "step": 2494 }, { "epoch": 0.22, "learning_rate": 1.8178009111428335e-05, "loss": 1.6437, "step": 2495 }, { "epoch": 0.22, "learning_rate": 1.8176362755057692e-05, "loss": 1.6199, "step": 2496 }, { "epoch": 0.22, "learning_rate": 1.8174715729821475e-05, "loss": 1.4986, "step": 2497 }, { "epoch": 0.22, "learning_rate": 1.8173068035854413e-05, "loss": 1.5426, "step": 2498 }, { "epoch": 0.22, "learning_rate": 1.8171419673291296e-05, "loss": 1.5444, "step": 2499 }, { "epoch": 0.22, "learning_rate": 1.8169770642266976e-05, "loss": 1.5628, "step": 2500 }, { "epoch": 0.22, "learning_rate": 1.8168120942916343e-05, "loss": 1.5214, "step": 2501 }, { "epoch": 0.22, "learning_rate": 1.8166470575374354e-05, "loss": 1.4684, "step": 2502 }, { "epoch": 0.22, "learning_rate": 1.8164819539776016e-05, "loss": 1.4447, "step": 2503 }, { "epoch": 0.22, "learning_rate": 1.8163167836256393e-05, "loss": 1.5371, "step": 2504 }, { "epoch": 0.22, "learning_rate": 1.81615154649506e-05, "loss": 1.444, "step": 2505 }, { "epoch": 0.22, "learning_rate": 1.815986242599381e-05, "loss": 1.5011, "step": 2506 }, { "epoch": 0.22, "learning_rate": 1.8158208719521248e-05, "loss": 1.5958, "step": 2507 }, { "epoch": 0.22, "learning_rate": 1.81565543456682e-05, "loss": 1.4949, "step": 2508 }, { "epoch": 0.22, "learning_rate": 1.8154899304569993e-05, "loss": 1.5732, "step": 2509 }, { "epoch": 0.22, "learning_rate": 1.815324359636203e-05, "loss": 1.6662, "step": 2510 }, { "epoch": 0.22, "learning_rate": 1.8151587221179742e-05, "loss": 1.5854, "step": 2511 }, { "epoch": 0.22, "learning_rate": 1.8149930179158638e-05, "loss": 1.5067, "step": 2512 }, { "epoch": 0.22, "learning_rate": 1.814827247043427e-05, "loss": 1.4845, "step": 2513 }, { "epoch": 0.22, "learning_rate": 1.8146614095142242e-05, "loss": 1.4873, "step": 2514 }, { "epoch": 0.22, "learning_rate": 1.8144955053418224e-05, "loss": 1.5596, "step": 2515 }, { "epoch": 0.22, "learning_rate": 1.8143295345397928e-05, "loss": 1.4604, "step": 2516 }, { "epoch": 0.22, "learning_rate": 1.8141634971217124e-05, "loss": 1.5441, "step": 2517 }, { "epoch": 0.22, "learning_rate": 1.8139973931011648e-05, "loss": 1.4066, "step": 2518 }, { "epoch": 0.22, "learning_rate": 1.8138312224917373e-05, "loss": 1.5679, "step": 2519 }, { "epoch": 0.22, "learning_rate": 1.8136649853070238e-05, "loss": 1.535, "step": 2520 }, { "epoch": 0.22, "learning_rate": 1.8134986815606233e-05, "loss": 1.4403, "step": 2521 }, { "epoch": 0.22, "learning_rate": 1.8133323112661397e-05, "loss": 1.6165, "step": 2522 }, { "epoch": 0.22, "learning_rate": 1.813165874437184e-05, "loss": 1.4592, "step": 2523 }, { "epoch": 0.22, "learning_rate": 1.8129993710873704e-05, "loss": 1.563, "step": 2524 }, { "epoch": 0.22, "learning_rate": 1.81283280123032e-05, "loss": 1.627, "step": 2525 }, { "epoch": 0.22, "learning_rate": 1.8126661648796596e-05, "loss": 1.5984, "step": 2526 }, { "epoch": 0.22, "learning_rate": 1.8124994620490198e-05, "loss": 1.5633, "step": 2527 }, { "epoch": 0.22, "learning_rate": 1.8123326927520387e-05, "loss": 1.4569, "step": 2528 }, { "epoch": 0.22, "learning_rate": 1.8121658570023583e-05, "loss": 1.6238, "step": 2529 }, { "epoch": 0.22, "learning_rate": 1.8119989548136266e-05, "loss": 1.5426, "step": 2530 }, { "epoch": 0.22, "learning_rate": 1.8118319861994972e-05, "loss": 1.4731, "step": 2531 }, { "epoch": 0.22, "learning_rate": 1.8116649511736286e-05, "loss": 1.4995, "step": 2532 }, { "epoch": 0.22, "learning_rate": 1.8114978497496855e-05, "loss": 1.491, "step": 2533 }, { "epoch": 0.22, "learning_rate": 1.8113306819413373e-05, "loss": 1.4877, "step": 2534 }, { "epoch": 0.22, "learning_rate": 1.811163447762259e-05, "loss": 1.5579, "step": 2535 }, { "epoch": 0.22, "learning_rate": 1.8109961472261313e-05, "loss": 1.516, "step": 2536 }, { "epoch": 0.22, "learning_rate": 1.8108287803466406e-05, "loss": 1.6288, "step": 2537 }, { "epoch": 0.22, "learning_rate": 1.8106613471374776e-05, "loss": 1.6327, "step": 2538 }, { "epoch": 0.22, "learning_rate": 1.8104938476123394e-05, "loss": 1.577, "step": 2539 }, { "epoch": 0.22, "learning_rate": 1.8103262817849287e-05, "loss": 1.528, "step": 2540 }, { "epoch": 0.22, "learning_rate": 1.8101586496689524e-05, "loss": 1.4692, "step": 2541 }, { "epoch": 0.22, "learning_rate": 1.809990951278124e-05, "loss": 1.5555, "step": 2542 }, { "epoch": 0.22, "learning_rate": 1.8098231866261626e-05, "loss": 1.6947, "step": 2543 }, { "epoch": 0.22, "learning_rate": 1.809655355726791e-05, "loss": 1.4596, "step": 2544 }, { "epoch": 0.22, "learning_rate": 1.8094874585937393e-05, "loss": 1.4571, "step": 2545 }, { "epoch": 0.22, "learning_rate": 1.8093194952407426e-05, "loss": 1.6951, "step": 2546 }, { "epoch": 0.22, "learning_rate": 1.80915146568154e-05, "loss": 1.5895, "step": 2547 }, { "epoch": 0.22, "learning_rate": 1.8089833699298784e-05, "loss": 1.638, "step": 2548 }, { "epoch": 0.23, "learning_rate": 1.8088152079995082e-05, "loss": 1.579, "step": 2549 }, { "epoch": 0.23, "learning_rate": 1.8086469799041855e-05, "loss": 1.6321, "step": 2550 }, { "epoch": 0.23, "learning_rate": 1.8084786856576727e-05, "loss": 1.5791, "step": 2551 }, { "epoch": 0.23, "learning_rate": 1.8083103252737367e-05, "loss": 1.5665, "step": 2552 }, { "epoch": 0.23, "learning_rate": 1.8081418987661506e-05, "loss": 1.3796, "step": 2553 }, { "epoch": 0.23, "learning_rate": 1.8079734061486923e-05, "loss": 1.4923, "step": 2554 }, { "epoch": 0.23, "learning_rate": 1.8078048474351454e-05, "loss": 1.4903, "step": 2555 }, { "epoch": 0.23, "learning_rate": 1.807636222639299e-05, "loss": 1.5296, "step": 2556 }, { "epoch": 0.23, "learning_rate": 1.8074675317749465e-05, "loss": 1.6358, "step": 2557 }, { "epoch": 0.23, "learning_rate": 1.8072987748558885e-05, "loss": 1.4825, "step": 2558 }, { "epoch": 0.23, "learning_rate": 1.8071299518959294e-05, "loss": 1.5761, "step": 2559 }, { "epoch": 0.23, "learning_rate": 1.8069610629088806e-05, "loss": 1.5988, "step": 2560 }, { "epoch": 0.23, "learning_rate": 1.8067921079085577e-05, "loss": 1.4518, "step": 2561 }, { "epoch": 0.23, "learning_rate": 1.8066230869087817e-05, "loss": 1.5367, "step": 2562 }, { "epoch": 0.23, "learning_rate": 1.8064539999233797e-05, "loss": 1.6056, "step": 2563 }, { "epoch": 0.23, "learning_rate": 1.8062848469661836e-05, "loss": 1.653, "step": 2564 }, { "epoch": 0.23, "learning_rate": 1.8061156280510314e-05, "loss": 1.555, "step": 2565 }, { "epoch": 0.23, "learning_rate": 1.8059463431917652e-05, "loss": 1.5646, "step": 2566 }, { "epoch": 0.23, "learning_rate": 1.805776992402234e-05, "loss": 1.5106, "step": 2567 }, { "epoch": 0.23, "learning_rate": 1.805607575696291e-05, "loss": 1.595, "step": 2568 }, { "epoch": 0.23, "learning_rate": 1.8054380930877953e-05, "loss": 1.4354, "step": 2569 }, { "epoch": 0.23, "learning_rate": 1.8052685445906122e-05, "loss": 1.505, "step": 2570 }, { "epoch": 0.23, "learning_rate": 1.805098930218611e-05, "loss": 1.5099, "step": 2571 }, { "epoch": 0.23, "learning_rate": 1.8049292499856665e-05, "loss": 1.7095, "step": 2572 }, { "epoch": 0.23, "learning_rate": 1.8047595039056604e-05, "loss": 1.5511, "step": 2573 }, { "epoch": 0.23, "learning_rate": 1.8045896919924776e-05, "loss": 1.5542, "step": 2574 }, { "epoch": 0.23, "learning_rate": 1.8044198142600104e-05, "loss": 1.3165, "step": 2575 }, { "epoch": 0.23, "learning_rate": 1.804249870722155e-05, "loss": 1.5908, "step": 2576 }, { "epoch": 0.23, "learning_rate": 1.8040798613928145e-05, "loss": 1.6592, "step": 2577 }, { "epoch": 0.23, "learning_rate": 1.803909786285895e-05, "loss": 1.5504, "step": 2578 }, { "epoch": 0.23, "learning_rate": 1.803739645415311e-05, "loss": 1.4623, "step": 2579 }, { "epoch": 0.23, "learning_rate": 1.8035694387949802e-05, "loss": 1.575, "step": 2580 }, { "epoch": 0.23, "learning_rate": 1.803399166438826e-05, "loss": 1.5498, "step": 2581 }, { "epoch": 0.23, "learning_rate": 1.8032288283607777e-05, "loss": 1.6381, "step": 2582 }, { "epoch": 0.23, "learning_rate": 1.8030584245747703e-05, "loss": 1.5572, "step": 2583 }, { "epoch": 0.23, "learning_rate": 1.8028879550947426e-05, "loss": 1.6206, "step": 2584 }, { "epoch": 0.23, "learning_rate": 1.802717419934641e-05, "loss": 1.5004, "step": 2585 }, { "epoch": 0.23, "learning_rate": 1.802546819108415e-05, "loss": 1.5151, "step": 2586 }, { "epoch": 0.23, "learning_rate": 1.8023761526300214e-05, "loss": 1.517, "step": 2587 }, { "epoch": 0.23, "learning_rate": 1.802205420513421e-05, "loss": 1.6164, "step": 2588 }, { "epoch": 0.23, "learning_rate": 1.8020346227725807e-05, "loss": 1.5981, "step": 2589 }, { "epoch": 0.23, "learning_rate": 1.8018637594214726e-05, "loss": 1.5759, "step": 2590 }, { "epoch": 0.23, "learning_rate": 1.801692830474074e-05, "loss": 1.4815, "step": 2591 }, { "epoch": 0.23, "learning_rate": 1.8015218359443682e-05, "loss": 1.6016, "step": 2592 }, { "epoch": 0.23, "learning_rate": 1.801350775846343e-05, "loss": 1.6831, "step": 2593 }, { "epoch": 0.23, "learning_rate": 1.8011796501939915e-05, "loss": 1.5971, "step": 2594 }, { "epoch": 0.23, "learning_rate": 1.8010084590013134e-05, "loss": 1.573, "step": 2595 }, { "epoch": 0.23, "learning_rate": 1.8008372022823125e-05, "loss": 1.4411, "step": 2596 }, { "epoch": 0.23, "learning_rate": 1.800665880050998e-05, "loss": 1.7138, "step": 2597 }, { "epoch": 0.23, "learning_rate": 1.8004944923213862e-05, "loss": 1.6319, "step": 2598 }, { "epoch": 0.23, "learning_rate": 1.800323039107496e-05, "loss": 1.6114, "step": 2599 }, { "epoch": 0.23, "learning_rate": 1.800151520423354e-05, "loss": 1.5878, "step": 2600 }, { "epoch": 0.23, "learning_rate": 1.7999799362829908e-05, "loss": 1.6333, "step": 2601 }, { "epoch": 0.23, "learning_rate": 1.7998082867004426e-05, "loss": 1.5869, "step": 2602 }, { "epoch": 0.23, "learning_rate": 1.7996365716897523e-05, "loss": 1.6156, "step": 2603 }, { "epoch": 0.23, "learning_rate": 1.7994647912649656e-05, "loss": 1.5913, "step": 2604 }, { "epoch": 0.23, "learning_rate": 1.799292945440136e-05, "loss": 1.45, "step": 2605 }, { "epoch": 0.23, "learning_rate": 1.799121034229321e-05, "loss": 1.4265, "step": 2606 }, { "epoch": 0.23, "learning_rate": 1.798949057646583e-05, "loss": 1.5049, "step": 2607 }, { "epoch": 0.23, "learning_rate": 1.7987770157059917e-05, "loss": 1.472, "step": 2608 }, { "epoch": 0.23, "learning_rate": 1.79860490842162e-05, "loss": 1.6812, "step": 2609 }, { "epoch": 0.23, "learning_rate": 1.798432735807548e-05, "loss": 1.6465, "step": 2610 }, { "epoch": 0.23, "learning_rate": 1.7982604978778592e-05, "loss": 1.6108, "step": 2611 }, { "epoch": 0.23, "learning_rate": 1.7980881946466447e-05, "loss": 1.6134, "step": 2612 }, { "epoch": 0.23, "learning_rate": 1.797915826127998e-05, "loss": 1.5079, "step": 2613 }, { "epoch": 0.23, "learning_rate": 1.797743392336022e-05, "loss": 1.4888, "step": 2614 }, { "epoch": 0.23, "learning_rate": 1.797570893284821e-05, "loss": 1.6, "step": 2615 }, { "epoch": 0.23, "learning_rate": 1.797398328988506e-05, "loss": 1.5648, "step": 2616 }, { "epoch": 0.23, "learning_rate": 1.797225699461195e-05, "loss": 1.6084, "step": 2617 }, { "epoch": 0.23, "learning_rate": 1.797053004717008e-05, "loss": 1.4613, "step": 2618 }, { "epoch": 0.23, "learning_rate": 1.7968802447700743e-05, "loss": 1.6662, "step": 2619 }, { "epoch": 0.23, "learning_rate": 1.7967074196345254e-05, "loss": 1.509, "step": 2620 }, { "epoch": 0.23, "learning_rate": 1.7965345293244994e-05, "loss": 1.6406, "step": 2621 }, { "epoch": 0.23, "learning_rate": 1.7963615738541393e-05, "loss": 1.6152, "step": 2622 }, { "epoch": 0.23, "learning_rate": 1.796188553237594e-05, "loss": 1.463, "step": 2623 }, { "epoch": 0.23, "learning_rate": 1.7960154674890172e-05, "loss": 1.6265, "step": 2624 }, { "epoch": 0.23, "learning_rate": 1.7958423166225683e-05, "loss": 1.542, "step": 2625 }, { "epoch": 0.23, "learning_rate": 1.7956691006524118e-05, "loss": 1.4596, "step": 2626 }, { "epoch": 0.23, "learning_rate": 1.7954958195927176e-05, "loss": 1.5284, "step": 2627 }, { "epoch": 0.23, "learning_rate": 1.7953224734576613e-05, "loss": 1.6695, "step": 2628 }, { "epoch": 0.23, "learning_rate": 1.7951490622614226e-05, "loss": 1.5278, "step": 2629 }, { "epoch": 0.23, "learning_rate": 1.7949755860181883e-05, "loss": 1.6313, "step": 2630 }, { "epoch": 0.23, "learning_rate": 1.794802044742149e-05, "loss": 1.5907, "step": 2631 }, { "epoch": 0.23, "learning_rate": 1.794628438447501e-05, "loss": 1.4552, "step": 2632 }, { "epoch": 0.23, "learning_rate": 1.794454767148447e-05, "loss": 1.4025, "step": 2633 }, { "epoch": 0.23, "learning_rate": 1.7942810308591932e-05, "loss": 1.5274, "step": 2634 }, { "epoch": 0.23, "learning_rate": 1.794107229593953e-05, "loss": 1.469, "step": 2635 }, { "epoch": 0.23, "learning_rate": 1.793933363366943e-05, "loss": 1.6249, "step": 2636 }, { "epoch": 0.23, "learning_rate": 1.7937594321923873e-05, "loss": 1.6255, "step": 2637 }, { "epoch": 0.23, "learning_rate": 1.793585436084514e-05, "loss": 1.5881, "step": 2638 }, { "epoch": 0.23, "learning_rate": 1.7934113750575567e-05, "loss": 1.5902, "step": 2639 }, { "epoch": 0.23, "learning_rate": 1.7932372491257545e-05, "loss": 1.6185, "step": 2640 }, { "epoch": 0.23, "learning_rate": 1.7930630583033516e-05, "loss": 1.6351, "step": 2641 }, { "epoch": 0.23, "learning_rate": 1.7928888026045976e-05, "loss": 1.4958, "step": 2642 }, { "epoch": 0.23, "learning_rate": 1.792714482043748e-05, "loss": 1.5808, "step": 2643 }, { "epoch": 0.23, "learning_rate": 1.7925400966350622e-05, "loss": 1.5096, "step": 2644 }, { "epoch": 0.23, "learning_rate": 1.7923656463928065e-05, "loss": 1.6821, "step": 2645 }, { "epoch": 0.23, "learning_rate": 1.7921911313312516e-05, "loss": 1.5443, "step": 2646 }, { "epoch": 0.23, "learning_rate": 1.7920165514646735e-05, "loss": 1.6017, "step": 2647 }, { "epoch": 0.23, "learning_rate": 1.7918419068073537e-05, "loss": 1.6361, "step": 2648 }, { "epoch": 0.23, "learning_rate": 1.7916671973735786e-05, "loss": 1.5922, "step": 2649 }, { "epoch": 0.23, "learning_rate": 1.791492423177641e-05, "loss": 1.4423, "step": 2650 }, { "epoch": 0.23, "learning_rate": 1.7913175842338377e-05, "loss": 1.5377, "step": 2651 }, { "epoch": 0.23, "learning_rate": 1.791142680556472e-05, "loss": 1.5969, "step": 2652 }, { "epoch": 0.23, "learning_rate": 1.790967712159851e-05, "loss": 1.4381, "step": 2653 }, { "epoch": 0.23, "learning_rate": 1.7907926790582887e-05, "loss": 1.6252, "step": 2654 }, { "epoch": 0.23, "learning_rate": 1.790617581266103e-05, "loss": 1.6062, "step": 2655 }, { "epoch": 0.23, "learning_rate": 1.7904424187976187e-05, "loss": 1.5264, "step": 2656 }, { "epoch": 0.23, "learning_rate": 1.7902671916671637e-05, "loss": 1.4953, "step": 2657 }, { "epoch": 0.23, "learning_rate": 1.7900918998890734e-05, "loss": 1.6101, "step": 2658 }, { "epoch": 0.23, "learning_rate": 1.789916543477687e-05, "loss": 1.4381, "step": 2659 }, { "epoch": 0.23, "learning_rate": 1.7897411224473495e-05, "loss": 1.4651, "step": 2660 }, { "epoch": 0.23, "learning_rate": 1.789565636812411e-05, "loss": 1.5742, "step": 2661 }, { "epoch": 0.24, "learning_rate": 1.789390086587228e-05, "loss": 1.5173, "step": 2662 }, { "epoch": 0.24, "learning_rate": 1.7892144717861603e-05, "loss": 1.5415, "step": 2663 }, { "epoch": 0.24, "learning_rate": 1.7890387924235747e-05, "loss": 1.5205, "step": 2664 }, { "epoch": 0.24, "learning_rate": 1.788863048513842e-05, "loss": 1.5214, "step": 2665 }, { "epoch": 0.24, "learning_rate": 1.7886872400713394e-05, "loss": 1.5614, "step": 2666 }, { "epoch": 0.24, "learning_rate": 1.7885113671104487e-05, "loss": 1.5288, "step": 2667 }, { "epoch": 0.24, "learning_rate": 1.788335429645557e-05, "loss": 1.6398, "step": 2668 }, { "epoch": 0.24, "learning_rate": 1.7881594276910577e-05, "loss": 1.5324, "step": 2669 }, { "epoch": 0.24, "learning_rate": 1.7879833612613474e-05, "loss": 1.5295, "step": 2670 }, { "epoch": 0.24, "learning_rate": 1.7878072303708296e-05, "loss": 1.4624, "step": 2671 }, { "epoch": 0.24, "learning_rate": 1.7876310350339126e-05, "loss": 1.5444, "step": 2672 }, { "epoch": 0.24, "learning_rate": 1.7874547752650104e-05, "loss": 1.537, "step": 2673 }, { "epoch": 0.24, "learning_rate": 1.7872784510785413e-05, "loss": 1.4861, "step": 2674 }, { "epoch": 0.24, "learning_rate": 1.78710206248893e-05, "loss": 1.5026, "step": 2675 }, { "epoch": 0.24, "learning_rate": 1.786925609510606e-05, "loss": 1.4639, "step": 2676 }, { "epoch": 0.24, "learning_rate": 1.7867490921580036e-05, "loss": 1.549, "step": 2677 }, { "epoch": 0.24, "learning_rate": 1.7865725104455624e-05, "loss": 1.5503, "step": 2678 }, { "epoch": 0.24, "learning_rate": 1.7863958643877288e-05, "loss": 1.4859, "step": 2679 }, { "epoch": 0.24, "learning_rate": 1.7862191539989523e-05, "loss": 1.4958, "step": 2680 }, { "epoch": 0.24, "learning_rate": 1.786042379293689e-05, "loss": 1.6246, "step": 2681 }, { "epoch": 0.24, "learning_rate": 1.7858655402863997e-05, "loss": 1.591, "step": 2682 }, { "epoch": 0.24, "learning_rate": 1.7856886369915507e-05, "loss": 1.5564, "step": 2683 }, { "epoch": 0.24, "learning_rate": 1.785511669423614e-05, "loss": 1.5435, "step": 2684 }, { "epoch": 0.24, "learning_rate": 1.785334637597066e-05, "loss": 1.7042, "step": 2685 }, { "epoch": 0.24, "learning_rate": 1.7851575415263888e-05, "loss": 1.5352, "step": 2686 }, { "epoch": 0.24, "learning_rate": 1.7849803812260698e-05, "loss": 1.6684, "step": 2687 }, { "epoch": 0.24, "learning_rate": 1.7848031567106015e-05, "loss": 1.5356, "step": 2688 }, { "epoch": 0.24, "learning_rate": 1.7846258679944816e-05, "loss": 1.5164, "step": 2689 }, { "epoch": 0.24, "learning_rate": 1.7844485150922132e-05, "loss": 1.5443, "step": 2690 }, { "epoch": 0.24, "learning_rate": 1.7842710980183052e-05, "loss": 1.6718, "step": 2691 }, { "epoch": 0.24, "learning_rate": 1.78409361678727e-05, "loss": 1.511, "step": 2692 }, { "epoch": 0.24, "learning_rate": 1.7839160714136274e-05, "loss": 1.5259, "step": 2693 }, { "epoch": 0.24, "learning_rate": 1.783738461911901e-05, "loss": 1.5201, "step": 2694 }, { "epoch": 0.24, "learning_rate": 1.7835607882966204e-05, "loss": 1.4518, "step": 2695 }, { "epoch": 0.24, "learning_rate": 1.78338305058232e-05, "loss": 1.5049, "step": 2696 }, { "epoch": 0.24, "learning_rate": 1.7832052487835396e-05, "loss": 1.519, "step": 2697 }, { "epoch": 0.24, "learning_rate": 1.783027382914824e-05, "loss": 1.5945, "step": 2698 }, { "epoch": 0.24, "learning_rate": 1.7828494529907238e-05, "loss": 1.571, "step": 2699 }, { "epoch": 0.24, "learning_rate": 1.7826714590257943e-05, "loss": 1.4936, "step": 2700 }, { "epoch": 0.24, "learning_rate": 1.7824934010345964e-05, "loss": 1.5549, "step": 2701 }, { "epoch": 0.24, "learning_rate": 1.782315279031696e-05, "loss": 1.6037, "step": 2702 }, { "epoch": 0.24, "learning_rate": 1.782137093031665e-05, "loss": 1.6618, "step": 2703 }, { "epoch": 0.24, "learning_rate": 1.7819588430490785e-05, "loss": 1.5909, "step": 2704 }, { "epoch": 0.24, "learning_rate": 1.7817805290985194e-05, "loss": 1.587, "step": 2705 }, { "epoch": 0.24, "learning_rate": 1.781602151194574e-05, "loss": 1.5591, "step": 2706 }, { "epoch": 0.24, "learning_rate": 1.7814237093518347e-05, "loss": 1.6164, "step": 2707 }, { "epoch": 0.24, "learning_rate": 1.781245203584899e-05, "loss": 1.6266, "step": 2708 }, { "epoch": 0.24, "learning_rate": 1.781066633908369e-05, "loss": 1.449, "step": 2709 }, { "epoch": 0.24, "learning_rate": 1.7808880003368535e-05, "loss": 1.5414, "step": 2710 }, { "epoch": 0.24, "learning_rate": 1.7807093028849646e-05, "loss": 1.5171, "step": 2711 }, { "epoch": 0.24, "learning_rate": 1.7805305415673214e-05, "loss": 1.5615, "step": 2712 }, { "epoch": 0.24, "learning_rate": 1.780351716398547e-05, "loss": 1.5109, "step": 2713 }, { "epoch": 0.24, "learning_rate": 1.78017282739327e-05, "loss": 1.7055, "step": 2714 }, { "epoch": 0.24, "learning_rate": 1.7799938745661246e-05, "loss": 1.5925, "step": 2715 }, { "epoch": 0.24, "learning_rate": 1.77981485793175e-05, "loss": 1.4434, "step": 2716 }, { "epoch": 0.24, "learning_rate": 1.7796357775047907e-05, "loss": 1.5399, "step": 2717 }, { "epoch": 0.24, "learning_rate": 1.779456633299896e-05, "loss": 1.6328, "step": 2718 }, { "epoch": 0.24, "learning_rate": 1.779277425331721e-05, "loss": 1.5931, "step": 2719 }, { "epoch": 0.24, "learning_rate": 1.779098153614926e-05, "loss": 1.3592, "step": 2720 }, { "epoch": 0.24, "learning_rate": 1.778918818164176e-05, "loss": 1.5324, "step": 2721 }, { "epoch": 0.24, "learning_rate": 1.7787394189941415e-05, "loss": 1.5351, "step": 2722 }, { "epoch": 0.24, "learning_rate": 1.7785599561194983e-05, "loss": 1.5592, "step": 2723 }, { "epoch": 0.24, "learning_rate": 1.7783804295549275e-05, "loss": 1.5661, "step": 2724 }, { "epoch": 0.24, "learning_rate": 1.7782008393151148e-05, "loss": 1.5411, "step": 2725 }, { "epoch": 0.24, "learning_rate": 1.7780211854147515e-05, "loss": 1.5437, "step": 2726 }, { "epoch": 0.24, "learning_rate": 1.7778414678685346e-05, "loss": 1.6548, "step": 2727 }, { "epoch": 0.24, "learning_rate": 1.7776616866911654e-05, "loss": 1.539, "step": 2728 }, { "epoch": 0.24, "learning_rate": 1.7774818418973518e-05, "loss": 1.5132, "step": 2729 }, { "epoch": 0.24, "learning_rate": 1.7773019335018047e-05, "loss": 1.6215, "step": 2730 }, { "epoch": 0.24, "learning_rate": 1.7771219615192422e-05, "loss": 1.5499, "step": 2731 }, { "epoch": 0.24, "learning_rate": 1.7769419259643864e-05, "loss": 1.5292, "step": 2732 }, { "epoch": 0.24, "learning_rate": 1.7767618268519658e-05, "loss": 1.4555, "step": 2733 }, { "epoch": 0.24, "learning_rate": 1.7765816641967127e-05, "loss": 1.5175, "step": 2734 }, { "epoch": 0.24, "learning_rate": 1.7764014380133655e-05, "loss": 1.5236, "step": 2735 }, { "epoch": 0.24, "learning_rate": 1.776221148316668e-05, "loss": 1.5281, "step": 2736 }, { "epoch": 0.24, "learning_rate": 1.776040795121368e-05, "loss": 1.4789, "step": 2737 }, { "epoch": 0.24, "learning_rate": 1.7758603784422198e-05, "loss": 1.4425, "step": 2738 }, { "epoch": 0.24, "learning_rate": 1.7756798982939818e-05, "loss": 1.5908, "step": 2739 }, { "epoch": 0.24, "learning_rate": 1.7754993546914184e-05, "loss": 1.593, "step": 2740 }, { "epoch": 0.24, "learning_rate": 1.7753187476492993e-05, "loss": 1.5412, "step": 2741 }, { "epoch": 0.24, "learning_rate": 1.7751380771823987e-05, "loss": 1.4652, "step": 2742 }, { "epoch": 0.24, "learning_rate": 1.7749573433054962e-05, "loss": 1.5095, "step": 2743 }, { "epoch": 0.24, "learning_rate": 1.774776546033377e-05, "loss": 1.4367, "step": 2744 }, { "epoch": 0.24, "learning_rate": 1.7745956853808312e-05, "loss": 1.3291, "step": 2745 }, { "epoch": 0.24, "learning_rate": 1.7744147613626534e-05, "loss": 1.5665, "step": 2746 }, { "epoch": 0.24, "learning_rate": 1.774233773993645e-05, "loss": 1.6061, "step": 2747 }, { "epoch": 0.24, "learning_rate": 1.7740527232886106e-05, "loss": 1.3787, "step": 2748 }, { "epoch": 0.24, "learning_rate": 1.773871609262362e-05, "loss": 1.5312, "step": 2749 }, { "epoch": 0.24, "learning_rate": 1.7736904319297147e-05, "loss": 1.5892, "step": 2750 }, { "epoch": 0.24, "learning_rate": 1.7735091913054898e-05, "loss": 1.5984, "step": 2751 }, { "epoch": 0.24, "learning_rate": 1.773327887404514e-05, "loss": 1.5508, "step": 2752 }, { "epoch": 0.24, "learning_rate": 1.7731465202416184e-05, "loss": 1.4175, "step": 2753 }, { "epoch": 0.24, "learning_rate": 1.7729650898316402e-05, "loss": 1.4655, "step": 2754 }, { "epoch": 0.24, "learning_rate": 1.7727835961894205e-05, "loss": 1.6545, "step": 2755 }, { "epoch": 0.24, "learning_rate": 1.7726020393298073e-05, "loss": 1.4062, "step": 2756 }, { "epoch": 0.24, "learning_rate": 1.7724204192676522e-05, "loss": 1.7121, "step": 2757 }, { "epoch": 0.24, "learning_rate": 1.7722387360178127e-05, "loss": 1.5111, "step": 2758 }, { "epoch": 0.24, "learning_rate": 1.7720569895951515e-05, "loss": 1.4528, "step": 2759 }, { "epoch": 0.24, "learning_rate": 1.7718751800145364e-05, "loss": 1.5277, "step": 2760 }, { "epoch": 0.24, "learning_rate": 1.77169330729084e-05, "loss": 1.4622, "step": 2761 }, { "epoch": 0.24, "learning_rate": 1.7715113714389402e-05, "loss": 1.5042, "step": 2762 }, { "epoch": 0.24, "learning_rate": 1.771329372473721e-05, "loss": 1.4934, "step": 2763 }, { "epoch": 0.24, "learning_rate": 1.77114731041007e-05, "loss": 1.5922, "step": 2764 }, { "epoch": 0.24, "learning_rate": 1.770965185262881e-05, "loss": 1.5488, "step": 2765 }, { "epoch": 0.24, "learning_rate": 1.7707829970470526e-05, "loss": 1.6355, "step": 2766 }, { "epoch": 0.24, "learning_rate": 1.7706007457774893e-05, "loss": 1.4879, "step": 2767 }, { "epoch": 0.24, "learning_rate": 1.770418431469099e-05, "loss": 1.4683, "step": 2768 }, { "epoch": 0.24, "learning_rate": 1.770236054136797e-05, "loss": 1.5801, "step": 2769 }, { "epoch": 0.24, "learning_rate": 1.7700536137955022e-05, "loss": 1.4996, "step": 2770 }, { "epoch": 0.24, "learning_rate": 1.769871110460139e-05, "loss": 1.4833, "step": 2771 }, { "epoch": 0.24, "learning_rate": 1.769688544145637e-05, "loss": 1.5342, "step": 2772 }, { "epoch": 0.24, "learning_rate": 1.7695059148669313e-05, "loss": 1.4463, "step": 2773 }, { "epoch": 0.24, "learning_rate": 1.7693232226389612e-05, "loss": 1.6186, "step": 2774 }, { "epoch": 0.25, "learning_rate": 1.7691404674766726e-05, "loss": 1.5021, "step": 2775 }, { "epoch": 0.25, "learning_rate": 1.7689576493950155e-05, "loss": 1.545, "step": 2776 }, { "epoch": 0.25, "learning_rate": 1.7687747684089448e-05, "loss": 1.5847, "step": 2777 }, { "epoch": 0.25, "learning_rate": 1.768591824533422e-05, "loss": 1.5645, "step": 2778 }, { "epoch": 0.25, "learning_rate": 1.7684088177834113e-05, "loss": 1.5534, "step": 2779 }, { "epoch": 0.25, "learning_rate": 1.7682257481738853e-05, "loss": 1.6041, "step": 2780 }, { "epoch": 0.25, "learning_rate": 1.7680426157198187e-05, "loss": 1.4908, "step": 2781 }, { "epoch": 0.25, "learning_rate": 1.7678594204361932e-05, "loss": 1.5997, "step": 2782 }, { "epoch": 0.25, "learning_rate": 1.7676761623379945e-05, "loss": 1.4419, "step": 2783 }, { "epoch": 0.25, "learning_rate": 1.7674928414402147e-05, "loss": 1.5255, "step": 2784 }, { "epoch": 0.25, "learning_rate": 1.76730945775785e-05, "loss": 1.5853, "step": 2785 }, { "epoch": 0.25, "learning_rate": 1.767126011305902e-05, "loss": 1.4307, "step": 2786 }, { "epoch": 0.25, "learning_rate": 1.7669425020993775e-05, "loss": 1.603, "step": 2787 }, { "epoch": 0.25, "learning_rate": 1.7667589301532884e-05, "loss": 1.4451, "step": 2788 }, { "epoch": 0.25, "learning_rate": 1.7665752954826518e-05, "loss": 1.4597, "step": 2789 }, { "epoch": 0.25, "learning_rate": 1.76639159810249e-05, "loss": 1.5593, "step": 2790 }, { "epoch": 0.25, "learning_rate": 1.7662078380278306e-05, "loss": 1.5546, "step": 2791 }, { "epoch": 0.25, "learning_rate": 1.7660240152737052e-05, "loss": 1.6243, "step": 2792 }, { "epoch": 0.25, "learning_rate": 1.765840129855152e-05, "loss": 1.536, "step": 2793 }, { "epoch": 0.25, "learning_rate": 1.7656561817872137e-05, "loss": 1.5214, "step": 2794 }, { "epoch": 0.25, "learning_rate": 1.7654721710849378e-05, "loss": 1.6713, "step": 2795 }, { "epoch": 0.25, "learning_rate": 1.765288097763378e-05, "loss": 1.4544, "step": 2796 }, { "epoch": 0.25, "learning_rate": 1.7651039618375913e-05, "loss": 1.5073, "step": 2797 }, { "epoch": 0.25, "learning_rate": 1.7649197633226418e-05, "loss": 1.5288, "step": 2798 }, { "epoch": 0.25, "learning_rate": 1.764735502233597e-05, "loss": 1.5504, "step": 2799 }, { "epoch": 0.25, "learning_rate": 1.7645511785855314e-05, "loss": 1.565, "step": 2800 }, { "epoch": 0.25, "learning_rate": 1.764366792393523e-05, "loss": 1.5025, "step": 2801 }, { "epoch": 0.25, "learning_rate": 1.764182343672655e-05, "loss": 1.4874, "step": 2802 }, { "epoch": 0.25, "learning_rate": 1.7639978324380168e-05, "loss": 1.5758, "step": 2803 }, { "epoch": 0.25, "learning_rate": 1.763813258704702e-05, "loss": 1.5151, "step": 2804 }, { "epoch": 0.25, "learning_rate": 1.76362862248781e-05, "loss": 1.4762, "step": 2805 }, { "epoch": 0.25, "learning_rate": 1.7634439238024446e-05, "loss": 1.3973, "step": 2806 }, { "epoch": 0.25, "learning_rate": 1.7632591626637153e-05, "loss": 1.5539, "step": 2807 }, { "epoch": 0.25, "learning_rate": 1.763074339086736e-05, "loss": 1.5125, "step": 2808 }, { "epoch": 0.25, "learning_rate": 1.7628894530866265e-05, "loss": 1.5064, "step": 2809 }, { "epoch": 0.25, "learning_rate": 1.762704504678511e-05, "loss": 1.5238, "step": 2810 }, { "epoch": 0.25, "learning_rate": 1.7625194938775196e-05, "loss": 1.3972, "step": 2811 }, { "epoch": 0.25, "learning_rate": 1.762334420698787e-05, "loss": 1.5598, "step": 2812 }, { "epoch": 0.25, "learning_rate": 1.762149285157453e-05, "loss": 1.3984, "step": 2813 }, { "epoch": 0.25, "learning_rate": 1.7619640872686627e-05, "loss": 1.5537, "step": 2814 }, { "epoch": 0.25, "learning_rate": 1.7617788270475655e-05, "loss": 1.5402, "step": 2815 }, { "epoch": 0.25, "learning_rate": 1.7615935045093172e-05, "loss": 1.3539, "step": 2816 }, { "epoch": 0.25, "learning_rate": 1.7614081196690784e-05, "loss": 1.5517, "step": 2817 }, { "epoch": 0.25, "learning_rate": 1.7612226725420136e-05, "loss": 1.5676, "step": 2818 }, { "epoch": 0.25, "learning_rate": 1.7610371631432936e-05, "loss": 1.6318, "step": 2819 }, { "epoch": 0.25, "learning_rate": 1.7608515914880944e-05, "loss": 1.5851, "step": 2820 }, { "epoch": 0.25, "learning_rate": 1.7606659575915963e-05, "loss": 1.506, "step": 2821 }, { "epoch": 0.25, "learning_rate": 1.7604802614689848e-05, "loss": 1.4515, "step": 2822 }, { "epoch": 0.25, "learning_rate": 1.760294503135451e-05, "loss": 1.561, "step": 2823 }, { "epoch": 0.25, "learning_rate": 1.7601086826061907e-05, "loss": 1.4914, "step": 2824 }, { "epoch": 0.25, "learning_rate": 1.7599227998964054e-05, "loss": 1.6793, "step": 2825 }, { "epoch": 0.25, "learning_rate": 1.7597368550213003e-05, "loss": 1.5308, "step": 2826 }, { "epoch": 0.25, "learning_rate": 1.7595508479960872e-05, "loss": 1.4972, "step": 2827 }, { "epoch": 0.25, "learning_rate": 1.759364778835982e-05, "loss": 1.5662, "step": 2828 }, { "epoch": 0.25, "learning_rate": 1.759178647556207e-05, "loss": 1.4608, "step": 2829 }, { "epoch": 0.25, "learning_rate": 1.7589924541719875e-05, "loss": 1.5837, "step": 2830 }, { "epoch": 0.25, "learning_rate": 1.7588061986985554e-05, "loss": 1.5883, "step": 2831 }, { "epoch": 0.25, "learning_rate": 1.758619881151147e-05, "loss": 1.5035, "step": 2832 }, { "epoch": 0.25, "learning_rate": 1.7584335015450046e-05, "loss": 1.5328, "step": 2833 }, { "epoch": 0.25, "learning_rate": 1.7582470598953746e-05, "loss": 1.5187, "step": 2834 }, { "epoch": 0.25, "learning_rate": 1.758060556217509e-05, "loss": 1.4437, "step": 2835 }, { "epoch": 0.25, "learning_rate": 1.757873990526664e-05, "loss": 1.4568, "step": 2836 }, { "epoch": 0.25, "learning_rate": 1.7576873628381025e-05, "loss": 1.5123, "step": 2837 }, { "epoch": 0.25, "learning_rate": 1.757500673167091e-05, "loss": 1.4661, "step": 2838 }, { "epoch": 0.25, "learning_rate": 1.7573139215289015e-05, "loss": 1.5789, "step": 2839 }, { "epoch": 0.25, "learning_rate": 1.757127107938812e-05, "loss": 1.4638, "step": 2840 }, { "epoch": 0.25, "learning_rate": 1.7569402324121037e-05, "loss": 1.4582, "step": 2841 }, { "epoch": 0.25, "learning_rate": 1.7567532949640647e-05, "loss": 1.6231, "step": 2842 }, { "epoch": 0.25, "learning_rate": 1.7565662956099867e-05, "loss": 1.5116, "step": 2843 }, { "epoch": 0.25, "learning_rate": 1.7563792343651676e-05, "loss": 1.5817, "step": 2844 }, { "epoch": 0.25, "learning_rate": 1.75619211124491e-05, "loss": 1.6103, "step": 2845 }, { "epoch": 0.25, "learning_rate": 1.756004926264521e-05, "loss": 1.6119, "step": 2846 }, { "epoch": 0.25, "learning_rate": 1.7558176794393134e-05, "loss": 1.5465, "step": 2847 }, { "epoch": 0.25, "learning_rate": 1.7556303707846052e-05, "loss": 1.5617, "step": 2848 }, { "epoch": 0.25, "learning_rate": 1.7554430003157188e-05, "loss": 1.4665, "step": 2849 }, { "epoch": 0.25, "learning_rate": 1.755255568047982e-05, "loss": 1.4547, "step": 2850 }, { "epoch": 0.25, "learning_rate": 1.755068073996728e-05, "loss": 1.4737, "step": 2851 }, { "epoch": 0.25, "learning_rate": 1.7548805181772943e-05, "loss": 1.5488, "step": 2852 }, { "epoch": 0.25, "learning_rate": 1.7546929006050244e-05, "loss": 1.4845, "step": 2853 }, { "epoch": 0.25, "learning_rate": 1.7545052212952653e-05, "loss": 1.4856, "step": 2854 }, { "epoch": 0.25, "learning_rate": 1.7543174802633714e-05, "loss": 1.467, "step": 2855 }, { "epoch": 0.25, "learning_rate": 1.7541296775247e-05, "loss": 1.618, "step": 2856 }, { "epoch": 0.25, "learning_rate": 1.7539418130946143e-05, "loss": 1.5065, "step": 2857 }, { "epoch": 0.25, "learning_rate": 1.7537538869884822e-05, "loss": 1.4168, "step": 2858 }, { "epoch": 0.25, "learning_rate": 1.7535658992216777e-05, "loss": 1.6767, "step": 2859 }, { "epoch": 0.25, "learning_rate": 1.7533778498095788e-05, "loss": 1.4752, "step": 2860 }, { "epoch": 0.25, "learning_rate": 1.753189738767569e-05, "loss": 1.5741, "step": 2861 }, { "epoch": 0.25, "learning_rate": 1.753001566111036e-05, "loss": 1.5539, "step": 2862 }, { "epoch": 0.25, "learning_rate": 1.7528133318553743e-05, "loss": 1.4738, "step": 2863 }, { "epoch": 0.25, "learning_rate": 1.752625036015981e-05, "loss": 1.485, "step": 2864 }, { "epoch": 0.25, "learning_rate": 1.7524366786082613e-05, "loss": 1.592, "step": 2865 }, { "epoch": 0.25, "learning_rate": 1.7522482596476222e-05, "loss": 1.5822, "step": 2866 }, { "epoch": 0.25, "learning_rate": 1.752059779149478e-05, "loss": 1.5695, "step": 2867 }, { "epoch": 0.25, "learning_rate": 1.751871237129247e-05, "loss": 1.6051, "step": 2868 }, { "epoch": 0.25, "learning_rate": 1.7516826336023533e-05, "loss": 1.5983, "step": 2869 }, { "epoch": 0.25, "learning_rate": 1.7514939685842253e-05, "loss": 1.5234, "step": 2870 }, { "epoch": 0.25, "learning_rate": 1.7513052420902966e-05, "loss": 1.4031, "step": 2871 }, { "epoch": 0.25, "learning_rate": 1.751116454136006e-05, "loss": 1.611, "step": 2872 }, { "epoch": 0.25, "learning_rate": 1.7509276047367973e-05, "loss": 1.5497, "step": 2873 }, { "epoch": 0.25, "learning_rate": 1.7507386939081193e-05, "loss": 1.5232, "step": 2874 }, { "epoch": 0.25, "learning_rate": 1.7505497216654255e-05, "loss": 1.5207, "step": 2875 }, { "epoch": 0.25, "learning_rate": 1.7503606880241754e-05, "loss": 1.5808, "step": 2876 }, { "epoch": 0.25, "learning_rate": 1.7501715929998323e-05, "loss": 1.5612, "step": 2877 }, { "epoch": 0.25, "learning_rate": 1.749982436607865e-05, "loss": 1.5861, "step": 2878 }, { "epoch": 0.25, "learning_rate": 1.749793218863748e-05, "loss": 1.5154, "step": 2879 }, { "epoch": 0.25, "learning_rate": 1.74960393978296e-05, "loss": 1.5017, "step": 2880 }, { "epoch": 0.25, "learning_rate": 1.7494145993809844e-05, "loss": 1.6018, "step": 2881 }, { "epoch": 0.25, "learning_rate": 1.7492251976733107e-05, "loss": 1.6031, "step": 2882 }, { "epoch": 0.25, "learning_rate": 1.7490357346754326e-05, "loss": 1.6522, "step": 2883 }, { "epoch": 0.25, "learning_rate": 1.7488462104028493e-05, "loss": 1.5998, "step": 2884 }, { "epoch": 0.25, "learning_rate": 1.748656624871065e-05, "loss": 1.413, "step": 2885 }, { "epoch": 0.25, "learning_rate": 1.7484669780955877e-05, "loss": 1.6132, "step": 2886 }, { "epoch": 0.25, "learning_rate": 1.7482772700919325e-05, "loss": 1.453, "step": 2887 }, { "epoch": 0.26, "learning_rate": 1.7480875008756178e-05, "loss": 1.4771, "step": 2888 }, { "epoch": 0.26, "learning_rate": 1.747897670462168e-05, "loss": 1.7248, "step": 2889 }, { "epoch": 0.26, "learning_rate": 1.7477077788671123e-05, "loss": 1.5268, "step": 2890 }, { "epoch": 0.26, "learning_rate": 1.747517826105984e-05, "loss": 1.6385, "step": 2891 }, { "epoch": 0.26, "learning_rate": 1.7473278121943228e-05, "loss": 1.4278, "step": 2892 }, { "epoch": 0.26, "learning_rate": 1.7471377371476725e-05, "loss": 1.5809, "step": 2893 }, { "epoch": 0.26, "learning_rate": 1.7469476009815818e-05, "loss": 1.487, "step": 2894 }, { "epoch": 0.26, "learning_rate": 1.7467574037116053e-05, "loss": 1.5376, "step": 2895 }, { "epoch": 0.26, "learning_rate": 1.746567145353302e-05, "loss": 1.5977, "step": 2896 }, { "epoch": 0.26, "learning_rate": 1.7463768259222358e-05, "loss": 1.6126, "step": 2897 }, { "epoch": 0.26, "learning_rate": 1.7461864454339753e-05, "loss": 1.4312, "step": 2898 }, { "epoch": 0.26, "learning_rate": 1.7459960039040954e-05, "loss": 1.7008, "step": 2899 }, { "epoch": 0.26, "learning_rate": 1.7458055013481744e-05, "loss": 1.5001, "step": 2900 }, { "epoch": 0.26, "learning_rate": 1.745614937781797e-05, "loss": 1.459, "step": 2901 }, { "epoch": 0.26, "learning_rate": 1.7454243132205516e-05, "loss": 1.5557, "step": 2902 }, { "epoch": 0.26, "learning_rate": 1.745233627680033e-05, "loss": 1.4267, "step": 2903 }, { "epoch": 0.26, "learning_rate": 1.7450428811758387e-05, "loss": 1.4362, "step": 2904 }, { "epoch": 0.26, "learning_rate": 1.7448520737235738e-05, "loss": 1.5115, "step": 2905 }, { "epoch": 0.26, "learning_rate": 1.7446612053388476e-05, "loss": 1.512, "step": 2906 }, { "epoch": 0.26, "learning_rate": 1.744470276037273e-05, "loss": 1.5437, "step": 2907 }, { "epoch": 0.26, "learning_rate": 1.7442792858344696e-05, "loss": 1.5138, "step": 2908 }, { "epoch": 0.26, "learning_rate": 1.744088234746061e-05, "loss": 1.6238, "step": 2909 }, { "epoch": 0.26, "learning_rate": 1.7438971227876766e-05, "loss": 1.4704, "step": 2910 }, { "epoch": 0.26, "learning_rate": 1.7437059499749497e-05, "loss": 1.6236, "step": 2911 }, { "epoch": 0.26, "learning_rate": 1.7435147163235196e-05, "loss": 1.5274, "step": 2912 }, { "epoch": 0.26, "learning_rate": 1.7433234218490293e-05, "loss": 1.5179, "step": 2913 }, { "epoch": 0.26, "learning_rate": 1.743132066567129e-05, "loss": 1.4981, "step": 2914 }, { "epoch": 0.26, "learning_rate": 1.742940650493471e-05, "loss": 1.4271, "step": 2915 }, { "epoch": 0.26, "learning_rate": 1.742749173643715e-05, "loss": 1.4766, "step": 2916 }, { "epoch": 0.26, "learning_rate": 1.7425576360335245e-05, "loss": 1.4996, "step": 2917 }, { "epoch": 0.26, "learning_rate": 1.7423660376785677e-05, "loss": 1.5554, "step": 2918 }, { "epoch": 0.26, "learning_rate": 1.7421743785945193e-05, "loss": 1.5251, "step": 2919 }, { "epoch": 0.26, "learning_rate": 1.7419826587970566e-05, "loss": 1.4305, "step": 2920 }, { "epoch": 0.26, "learning_rate": 1.7417908783018646e-05, "loss": 1.6005, "step": 2921 }, { "epoch": 0.26, "learning_rate": 1.741599037124631e-05, "loss": 1.5116, "step": 2922 }, { "epoch": 0.26, "learning_rate": 1.7414071352810494e-05, "loss": 1.6729, "step": 2923 }, { "epoch": 0.26, "learning_rate": 1.7412151727868187e-05, "loss": 1.5356, "step": 2924 }, { "epoch": 0.26, "learning_rate": 1.7410231496576416e-05, "loss": 1.5807, "step": 2925 }, { "epoch": 0.26, "learning_rate": 1.7408310659092272e-05, "loss": 1.5977, "step": 2926 }, { "epoch": 0.26, "learning_rate": 1.7406389215572887e-05, "loss": 1.442, "step": 2927 }, { "epoch": 0.26, "learning_rate": 1.7404467166175442e-05, "loss": 1.5529, "step": 2928 }, { "epoch": 0.26, "learning_rate": 1.740254451105717e-05, "loss": 1.5402, "step": 2929 }, { "epoch": 0.26, "learning_rate": 1.7400621250375358e-05, "loss": 1.53, "step": 2930 }, { "epoch": 0.26, "learning_rate": 1.7398697384287334e-05, "loss": 1.5383, "step": 2931 }, { "epoch": 0.26, "learning_rate": 1.739677291295048e-05, "loss": 1.5357, "step": 2932 }, { "epoch": 0.26, "learning_rate": 1.7394847836522227e-05, "loss": 1.5471, "step": 2933 }, { "epoch": 0.26, "learning_rate": 1.7392922155160055e-05, "loss": 1.5695, "step": 2934 }, { "epoch": 0.26, "learning_rate": 1.7390995869021495e-05, "loss": 1.5017, "step": 2935 }, { "epoch": 0.26, "learning_rate": 1.7389068978264126e-05, "loss": 1.448, "step": 2936 }, { "epoch": 0.26, "learning_rate": 1.7387141483045577e-05, "loss": 1.6174, "step": 2937 }, { "epoch": 0.26, "learning_rate": 1.7385213383523525e-05, "loss": 1.6369, "step": 2938 }, { "epoch": 0.26, "learning_rate": 1.7383284679855697e-05, "loss": 1.5495, "step": 2939 }, { "epoch": 0.26, "learning_rate": 1.7381355372199877e-05, "loss": 1.5564, "step": 2940 }, { "epoch": 0.26, "learning_rate": 1.7379425460713885e-05, "loss": 1.5025, "step": 2941 }, { "epoch": 0.26, "learning_rate": 1.7377494945555596e-05, "loss": 1.5416, "step": 2942 }, { "epoch": 0.26, "learning_rate": 1.737556382688294e-05, "loss": 1.5248, "step": 2943 }, { "epoch": 0.26, "learning_rate": 1.737363210485389e-05, "loss": 1.5439, "step": 2944 }, { "epoch": 0.26, "learning_rate": 1.7371699779626467e-05, "loss": 1.5448, "step": 2945 }, { "epoch": 0.26, "learning_rate": 1.7369766851358753e-05, "loss": 1.5231, "step": 2946 }, { "epoch": 0.26, "learning_rate": 1.736783332020886e-05, "loss": 1.5517, "step": 2947 }, { "epoch": 0.26, "learning_rate": 1.736589918633497e-05, "loss": 1.4969, "step": 2948 }, { "epoch": 0.26, "learning_rate": 1.7363964449895295e-05, "loss": 1.585, "step": 2949 }, { "epoch": 0.26, "learning_rate": 1.7362029111048113e-05, "loss": 1.4945, "step": 2950 }, { "epoch": 0.26, "learning_rate": 1.736009316995174e-05, "loss": 1.5592, "step": 2951 }, { "epoch": 0.26, "learning_rate": 1.7358156626764546e-05, "loss": 1.536, "step": 2952 }, { "epoch": 0.26, "learning_rate": 1.735621948164495e-05, "loss": 1.4331, "step": 2953 }, { "epoch": 0.26, "learning_rate": 1.735428173475142e-05, "loss": 1.4926, "step": 2954 }, { "epoch": 0.26, "learning_rate": 1.7352343386242475e-05, "loss": 1.5042, "step": 2955 }, { "epoch": 0.26, "learning_rate": 1.7350404436276676e-05, "loss": 1.6278, "step": 2956 }, { "epoch": 0.26, "learning_rate": 1.7348464885012644e-05, "loss": 1.5023, "step": 2957 }, { "epoch": 0.26, "learning_rate": 1.7346524732609035e-05, "loss": 1.5139, "step": 2958 }, { "epoch": 0.26, "learning_rate": 1.7344583979224575e-05, "loss": 1.5998, "step": 2959 }, { "epoch": 0.26, "learning_rate": 1.734264262501802e-05, "loss": 1.489, "step": 2960 }, { "epoch": 0.26, "learning_rate": 1.734070067014818e-05, "loss": 1.5511, "step": 2961 }, { "epoch": 0.26, "learning_rate": 1.733875811477392e-05, "loss": 1.588, "step": 2962 }, { "epoch": 0.26, "learning_rate": 1.7336814959054147e-05, "loss": 1.5664, "step": 2963 }, { "epoch": 0.26, "learning_rate": 1.7334871203147825e-05, "loss": 1.5541, "step": 2964 }, { "epoch": 0.26, "learning_rate": 1.7332926847213957e-05, "loss": 1.4877, "step": 2965 }, { "epoch": 0.26, "learning_rate": 1.7330981891411607e-05, "loss": 1.5667, "step": 2966 }, { "epoch": 0.26, "learning_rate": 1.7329036335899875e-05, "loss": 1.5123, "step": 2967 }, { "epoch": 0.26, "learning_rate": 1.732709018083792e-05, "loss": 1.5404, "step": 2968 }, { "epoch": 0.26, "learning_rate": 1.7325143426384953e-05, "loss": 1.6047, "step": 2969 }, { "epoch": 0.26, "learning_rate": 1.7323196072700214e-05, "loss": 1.4253, "step": 2970 }, { "epoch": 0.26, "learning_rate": 1.732124811994302e-05, "loss": 1.4474, "step": 2971 }, { "epoch": 0.26, "learning_rate": 1.7319299568272713e-05, "loss": 1.5649, "step": 2972 }, { "epoch": 0.26, "learning_rate": 1.7317350417848696e-05, "loss": 1.5164, "step": 2973 }, { "epoch": 0.26, "learning_rate": 1.7315400668830422e-05, "loss": 1.4057, "step": 2974 }, { "epoch": 0.26, "learning_rate": 1.7313450321377388e-05, "loss": 1.6139, "step": 2975 }, { "epoch": 0.26, "learning_rate": 1.731149937564914e-05, "loss": 1.6322, "step": 2976 }, { "epoch": 0.26, "learning_rate": 1.7309547831805276e-05, "loss": 1.5733, "step": 2977 }, { "epoch": 0.26, "learning_rate": 1.7307595690005446e-05, "loss": 1.4639, "step": 2978 }, { "epoch": 0.26, "learning_rate": 1.7305642950409335e-05, "loss": 1.5621, "step": 2979 }, { "epoch": 0.26, "learning_rate": 1.7303689613176697e-05, "loss": 1.4395, "step": 2980 }, { "epoch": 0.26, "learning_rate": 1.7301735678467316e-05, "loss": 1.3552, "step": 2981 }, { "epoch": 0.26, "learning_rate": 1.7299781146441037e-05, "loss": 1.4439, "step": 2982 }, { "epoch": 0.26, "learning_rate": 1.7297826017257748e-05, "loss": 1.5035, "step": 2983 }, { "epoch": 0.26, "learning_rate": 1.729587029107739e-05, "loss": 1.5645, "step": 2984 }, { "epoch": 0.26, "learning_rate": 1.7293913968059952e-05, "loss": 1.4752, "step": 2985 }, { "epoch": 0.26, "learning_rate": 1.7291957048365468e-05, "loss": 1.431, "step": 2986 }, { "epoch": 0.26, "learning_rate": 1.7289999532154022e-05, "loss": 1.6823, "step": 2987 }, { "epoch": 0.26, "learning_rate": 1.7288041419585753e-05, "loss": 1.4264, "step": 2988 }, { "epoch": 0.26, "learning_rate": 1.7286082710820838e-05, "loss": 1.5612, "step": 2989 }, { "epoch": 0.26, "learning_rate": 1.7284123406019516e-05, "loss": 1.4587, "step": 2990 }, { "epoch": 0.26, "learning_rate": 1.7282163505342063e-05, "loss": 1.5328, "step": 2991 }, { "epoch": 0.26, "learning_rate": 1.7280203008948807e-05, "loss": 1.5288, "step": 2992 }, { "epoch": 0.26, "learning_rate": 1.727824191700013e-05, "loss": 1.6489, "step": 2993 }, { "epoch": 0.26, "learning_rate": 1.727628022965645e-05, "loss": 1.4902, "step": 2994 }, { "epoch": 0.26, "learning_rate": 1.7274317947078256e-05, "loss": 1.5171, "step": 2995 }, { "epoch": 0.26, "learning_rate": 1.727235506942606e-05, "loss": 1.5894, "step": 2996 }, { "epoch": 0.26, "learning_rate": 1.7270391596860447e-05, "loss": 1.5178, "step": 2997 }, { "epoch": 0.26, "learning_rate": 1.7268427529542023e-05, "loss": 1.4587, "step": 2998 }, { "epoch": 0.26, "learning_rate": 1.726646286763147e-05, "loss": 1.61, "step": 2999 }, { "epoch": 0.26, "learning_rate": 1.7264497611289504e-05, "loss": 1.4313, "step": 3000 }, { "epoch": 0.26, "learning_rate": 1.726253176067689e-05, "loss": 1.4376, "step": 3001 }, { "epoch": 0.27, "learning_rate": 1.7260565315954446e-05, "loss": 1.5065, "step": 3002 }, { "epoch": 0.27, "learning_rate": 1.7258598277283034e-05, "loss": 1.4078, "step": 3003 }, { "epoch": 0.27, "learning_rate": 1.7256630644823573e-05, "loss": 1.5208, "step": 3004 }, { "epoch": 0.27, "learning_rate": 1.7254662418737014e-05, "loss": 1.5598, "step": 3005 }, { "epoch": 0.27, "learning_rate": 1.725269359918438e-05, "loss": 1.4708, "step": 3006 }, { "epoch": 0.27, "learning_rate": 1.7250724186326723e-05, "loss": 1.6264, "step": 3007 }, { "epoch": 0.27, "learning_rate": 1.724875418032515e-05, "loss": 1.477, "step": 3008 }, { "epoch": 0.27, "learning_rate": 1.7246783581340822e-05, "loss": 1.6158, "step": 3009 }, { "epoch": 0.27, "learning_rate": 1.7244812389534936e-05, "loss": 1.5759, "step": 3010 }, { "epoch": 0.27, "learning_rate": 1.7242840605068746e-05, "loss": 1.6115, "step": 3011 }, { "epoch": 0.27, "learning_rate": 1.7240868228103562e-05, "loss": 1.4841, "step": 3012 }, { "epoch": 0.27, "learning_rate": 1.7238895258800724e-05, "loss": 1.448, "step": 3013 }, { "epoch": 0.27, "learning_rate": 1.7236921697321637e-05, "loss": 1.5334, "step": 3014 }, { "epoch": 0.27, "learning_rate": 1.723494754382774e-05, "loss": 1.5685, "step": 3015 }, { "epoch": 0.27, "learning_rate": 1.723297279848054e-05, "loss": 1.6816, "step": 3016 }, { "epoch": 0.27, "learning_rate": 1.723099746144157e-05, "loss": 1.5755, "step": 3017 }, { "epoch": 0.27, "learning_rate": 1.722902153287243e-05, "loss": 1.514, "step": 3018 }, { "epoch": 0.27, "learning_rate": 1.722704501293475e-05, "loss": 1.4786, "step": 3019 }, { "epoch": 0.27, "learning_rate": 1.7225067901790228e-05, "loss": 1.5699, "step": 3020 }, { "epoch": 0.27, "learning_rate": 1.72230901996006e-05, "loss": 1.4517, "step": 3021 }, { "epoch": 0.27, "learning_rate": 1.7221111906527647e-05, "loss": 1.5931, "step": 3022 }, { "epoch": 0.27, "learning_rate": 1.7219133022733205e-05, "loss": 1.5167, "step": 3023 }, { "epoch": 0.27, "learning_rate": 1.721715354837916e-05, "loss": 1.6359, "step": 3024 }, { "epoch": 0.27, "learning_rate": 1.7215173483627438e-05, "loss": 1.5225, "step": 3025 }, { "epoch": 0.27, "learning_rate": 1.7213192828640014e-05, "loss": 1.583, "step": 3026 }, { "epoch": 0.27, "learning_rate": 1.7211211583578927e-05, "loss": 1.5873, "step": 3027 }, { "epoch": 0.27, "learning_rate": 1.7209229748606242e-05, "loss": 1.6124, "step": 3028 }, { "epoch": 0.27, "learning_rate": 1.7207247323884086e-05, "loss": 1.48, "step": 3029 }, { "epoch": 0.27, "learning_rate": 1.7205264309574633e-05, "loss": 1.584, "step": 3030 }, { "epoch": 0.27, "learning_rate": 1.7203280705840097e-05, "loss": 1.497, "step": 3031 }, { "epoch": 0.27, "learning_rate": 1.7201296512842752e-05, "loss": 1.4892, "step": 3032 }, { "epoch": 0.27, "learning_rate": 1.7199311730744912e-05, "loss": 1.532, "step": 3033 }, { "epoch": 0.27, "learning_rate": 1.7197326359708943e-05, "loss": 1.4232, "step": 3034 }, { "epoch": 0.27, "learning_rate": 1.7195340399897252e-05, "loss": 1.6967, "step": 3035 }, { "epoch": 0.27, "learning_rate": 1.7193353851472313e-05, "loss": 1.649, "step": 3036 }, { "epoch": 0.27, "learning_rate": 1.719136671459662e-05, "loss": 1.4042, "step": 3037 }, { "epoch": 0.27, "learning_rate": 1.7189378989432743e-05, "loss": 1.5414, "step": 3038 }, { "epoch": 0.27, "learning_rate": 1.7187390676143278e-05, "loss": 1.3765, "step": 3039 }, { "epoch": 0.27, "learning_rate": 1.7185401774890886e-05, "loss": 1.5271, "step": 3040 }, { "epoch": 0.27, "learning_rate": 1.7183412285838263e-05, "loss": 1.5448, "step": 3041 }, { "epoch": 0.27, "learning_rate": 1.718142220914816e-05, "loss": 1.5596, "step": 3042 }, { "epoch": 0.27, "learning_rate": 1.717943154498338e-05, "loss": 1.5332, "step": 3043 }, { "epoch": 0.27, "learning_rate": 1.717744029350676e-05, "loss": 1.5187, "step": 3044 }, { "epoch": 0.27, "learning_rate": 1.71754484548812e-05, "loss": 1.6288, "step": 3045 }, { "epoch": 0.27, "learning_rate": 1.717345602926964e-05, "loss": 1.685, "step": 3046 }, { "epoch": 0.27, "learning_rate": 1.717146301683507e-05, "loss": 1.6069, "step": 3047 }, { "epoch": 0.27, "learning_rate": 1.7169469417740526e-05, "loss": 1.4783, "step": 3048 }, { "epoch": 0.27, "learning_rate": 1.7167475232149098e-05, "loss": 1.4686, "step": 3049 }, { "epoch": 0.27, "learning_rate": 1.716548046022392e-05, "loss": 1.515, "step": 3050 }, { "epoch": 0.27, "learning_rate": 1.7163485102128167e-05, "loss": 1.4761, "step": 3051 }, { "epoch": 0.27, "learning_rate": 1.7161489158025077e-05, "loss": 1.5547, "step": 3052 }, { "epoch": 0.27, "learning_rate": 1.7159492628077924e-05, "loss": 1.6626, "step": 3053 }, { "epoch": 0.27, "learning_rate": 1.7157495512450033e-05, "loss": 1.4276, "step": 3054 }, { "epoch": 0.27, "learning_rate": 1.715549781130478e-05, "loss": 1.3417, "step": 3055 }, { "epoch": 0.27, "learning_rate": 1.7153499524805586e-05, "loss": 1.4921, "step": 3056 }, { "epoch": 0.27, "learning_rate": 1.715150065311592e-05, "loss": 1.333, "step": 3057 }, { "epoch": 0.27, "learning_rate": 1.7149501196399295e-05, "loss": 1.54, "step": 3058 }, { "epoch": 0.27, "learning_rate": 1.7147501154819287e-05, "loss": 1.5071, "step": 3059 }, { "epoch": 0.27, "learning_rate": 1.7145500528539496e-05, "loss": 1.559, "step": 3060 }, { "epoch": 0.27, "learning_rate": 1.7143499317723593e-05, "loss": 1.5336, "step": 3061 }, { "epoch": 0.27, "learning_rate": 1.714149752253528e-05, "loss": 1.455, "step": 3062 }, { "epoch": 0.27, "learning_rate": 1.7139495143138316e-05, "loss": 1.3942, "step": 3063 }, { "epoch": 0.27, "learning_rate": 1.7137492179696503e-05, "loss": 1.5143, "step": 3064 }, { "epoch": 0.27, "learning_rate": 1.71354886323737e-05, "loss": 1.5551, "step": 3065 }, { "epoch": 0.27, "learning_rate": 1.7133484501333798e-05, "loss": 1.5462, "step": 3066 }, { "epoch": 0.27, "learning_rate": 1.713147978674075e-05, "loss": 1.5705, "step": 3067 }, { "epoch": 0.27, "learning_rate": 1.7129474488758547e-05, "loss": 1.5515, "step": 3068 }, { "epoch": 0.27, "learning_rate": 1.7127468607551238e-05, "loss": 1.4689, "step": 3069 }, { "epoch": 0.27, "learning_rate": 1.712546214328291e-05, "loss": 1.5999, "step": 3070 }, { "epoch": 0.27, "learning_rate": 1.7123455096117697e-05, "loss": 1.5708, "step": 3071 }, { "epoch": 0.27, "learning_rate": 1.7121447466219796e-05, "loss": 1.5078, "step": 3072 }, { "epoch": 0.27, "learning_rate": 1.7119439253753433e-05, "loss": 1.58, "step": 3073 }, { "epoch": 0.27, "learning_rate": 1.7117430458882886e-05, "loss": 1.5406, "step": 3074 }, { "epoch": 0.27, "learning_rate": 1.7115421081772493e-05, "loss": 1.5628, "step": 3075 }, { "epoch": 0.27, "learning_rate": 1.7113411122586624e-05, "loss": 1.5456, "step": 3076 }, { "epoch": 0.27, "learning_rate": 1.711140058148971e-05, "loss": 1.4302, "step": 3077 }, { "epoch": 0.27, "learning_rate": 1.7109389458646216e-05, "loss": 1.6089, "step": 3078 }, { "epoch": 0.27, "learning_rate": 1.7107377754220663e-05, "loss": 1.5527, "step": 3079 }, { "epoch": 0.27, "learning_rate": 1.710536546837762e-05, "loss": 1.4928, "step": 3080 }, { "epoch": 0.27, "learning_rate": 1.71033526012817e-05, "loss": 1.5724, "step": 3081 }, { "epoch": 0.27, "learning_rate": 1.7101339153097568e-05, "loss": 1.4195, "step": 3082 }, { "epoch": 0.27, "learning_rate": 1.709932512398993e-05, "loss": 1.4728, "step": 3083 }, { "epoch": 0.27, "learning_rate": 1.7097310514123544e-05, "loss": 1.5535, "step": 3084 }, { "epoch": 0.27, "learning_rate": 1.7095295323663217e-05, "loss": 1.5157, "step": 3085 }, { "epoch": 0.27, "learning_rate": 1.7093279552773797e-05, "loss": 1.4621, "step": 3086 }, { "epoch": 0.27, "learning_rate": 1.709126320162019e-05, "loss": 1.594, "step": 3087 }, { "epoch": 0.27, "learning_rate": 1.7089246270367338e-05, "loss": 1.5579, "step": 3088 }, { "epoch": 0.27, "learning_rate": 1.708722875918024e-05, "loss": 1.5225, "step": 3089 }, { "epoch": 0.27, "learning_rate": 1.7085210668223927e-05, "loss": 1.4599, "step": 3090 }, { "epoch": 0.27, "learning_rate": 1.7083191997663504e-05, "loss": 1.517, "step": 3091 }, { "epoch": 0.27, "learning_rate": 1.70811727476641e-05, "loss": 1.56, "step": 3092 }, { "epoch": 0.27, "learning_rate": 1.70791529183909e-05, "loss": 1.514, "step": 3093 }, { "epoch": 0.27, "learning_rate": 1.7077132510009133e-05, "loss": 1.4214, "step": 3094 }, { "epoch": 0.27, "learning_rate": 1.707511152268408e-05, "loss": 1.6279, "step": 3095 }, { "epoch": 0.27, "learning_rate": 1.707308995658107e-05, "loss": 1.6395, "step": 3096 }, { "epoch": 0.27, "learning_rate": 1.7071067811865477e-05, "loss": 1.4879, "step": 3097 }, { "epoch": 0.27, "learning_rate": 1.706904508870272e-05, "loss": 1.465, "step": 3098 }, { "epoch": 0.27, "learning_rate": 1.7067021787258264e-05, "loss": 1.4664, "step": 3099 }, { "epoch": 0.27, "learning_rate": 1.706499790769763e-05, "loss": 1.4252, "step": 3100 }, { "epoch": 0.27, "learning_rate": 1.706297345018638e-05, "loss": 1.5563, "step": 3101 }, { "epoch": 0.27, "learning_rate": 1.7060948414890123e-05, "loss": 1.4107, "step": 3102 }, { "epoch": 0.27, "learning_rate": 1.7058922801974515e-05, "loss": 1.51, "step": 3103 }, { "epoch": 0.27, "learning_rate": 1.7056896611605264e-05, "loss": 1.467, "step": 3104 }, { "epoch": 0.27, "learning_rate": 1.7054869843948125e-05, "loss": 1.4663, "step": 3105 }, { "epoch": 0.27, "learning_rate": 1.7052842499168888e-05, "loss": 1.6149, "step": 3106 }, { "epoch": 0.27, "learning_rate": 1.7050814577433407e-05, "loss": 1.5868, "step": 3107 }, { "epoch": 0.27, "learning_rate": 1.7048786078907574e-05, "loss": 1.4584, "step": 3108 }, { "epoch": 0.27, "learning_rate": 1.7046757003757326e-05, "loss": 1.4775, "step": 3109 }, { "epoch": 0.27, "learning_rate": 1.704472735214866e-05, "loss": 1.4266, "step": 3110 }, { "epoch": 0.27, "learning_rate": 1.70426971242476e-05, "loss": 1.5281, "step": 3111 }, { "epoch": 0.27, "learning_rate": 1.7040666320220238e-05, "loss": 1.5806, "step": 3112 }, { "epoch": 0.27, "learning_rate": 1.7038634940232697e-05, "loss": 1.5064, "step": 3113 }, { "epoch": 0.27, "learning_rate": 1.7036602984451155e-05, "loss": 1.6112, "step": 3114 }, { "epoch": 0.28, "learning_rate": 1.7034570453041842e-05, "loss": 1.4325, "step": 3115 }, { "epoch": 0.28, "learning_rate": 1.703253734617102e-05, "loss": 1.5039, "step": 3116 }, { "epoch": 0.28, "learning_rate": 1.7030503664005008e-05, "loss": 1.6388, "step": 3117 }, { "epoch": 0.28, "learning_rate": 1.7028469406710178e-05, "loss": 1.4922, "step": 3118 }, { "epoch": 0.28, "learning_rate": 1.7026434574452934e-05, "loss": 1.6192, "step": 3119 }, { "epoch": 0.28, "learning_rate": 1.702439916739974e-05, "loss": 1.5113, "step": 3120 }, { "epoch": 0.28, "learning_rate": 1.70223631857171e-05, "loss": 1.4835, "step": 3121 }, { "epoch": 0.28, "learning_rate": 1.7020326629571568e-05, "loss": 1.5333, "step": 3122 }, { "epoch": 0.28, "learning_rate": 1.7018289499129742e-05, "loss": 1.5577, "step": 3123 }, { "epoch": 0.28, "learning_rate": 1.701625179455827e-05, "loss": 1.504, "step": 3124 }, { "epoch": 0.28, "learning_rate": 1.7014213516023848e-05, "loss": 1.4718, "step": 3125 }, { "epoch": 0.28, "learning_rate": 1.7012174663693216e-05, "loss": 1.4876, "step": 3126 }, { "epoch": 0.28, "learning_rate": 1.701013523773316e-05, "loss": 1.5314, "step": 3127 }, { "epoch": 0.28, "learning_rate": 1.7008095238310514e-05, "loss": 1.4957, "step": 3128 }, { "epoch": 0.28, "learning_rate": 1.7006054665592168e-05, "loss": 1.5028, "step": 3129 }, { "epoch": 0.28, "learning_rate": 1.7004013519745038e-05, "loss": 1.4591, "step": 3130 }, { "epoch": 0.28, "learning_rate": 1.700197180093611e-05, "loss": 1.5481, "step": 3131 }, { "epoch": 0.28, "learning_rate": 1.69999295093324e-05, "loss": 1.44, "step": 3132 }, { "epoch": 0.28, "learning_rate": 1.699788664510098e-05, "loss": 1.4606, "step": 3133 }, { "epoch": 0.28, "learning_rate": 1.6995843208408965e-05, "loss": 1.5949, "step": 3134 }, { "epoch": 0.28, "learning_rate": 1.6993799199423516e-05, "loss": 1.4973, "step": 3135 }, { "epoch": 0.28, "learning_rate": 1.699175461831185e-05, "loss": 1.5384, "step": 3136 }, { "epoch": 0.28, "learning_rate": 1.6989709465241214e-05, "loss": 1.498, "step": 3137 }, { "epoch": 0.28, "learning_rate": 1.6987663740378918e-05, "loss": 1.4688, "step": 3138 }, { "epoch": 0.28, "learning_rate": 1.698561744389231e-05, "loss": 1.5761, "step": 3139 }, { "epoch": 0.28, "learning_rate": 1.6983570575948786e-05, "loss": 1.5534, "step": 3140 }, { "epoch": 0.28, "learning_rate": 1.6981523136715793e-05, "loss": 1.5719, "step": 3141 }, { "epoch": 0.28, "learning_rate": 1.6979475126360816e-05, "loss": 1.4774, "step": 3142 }, { "epoch": 0.28, "learning_rate": 1.6977426545051396e-05, "loss": 1.5812, "step": 3143 }, { "epoch": 0.28, "learning_rate": 1.6975377392955115e-05, "loss": 1.4282, "step": 3144 }, { "epoch": 0.28, "learning_rate": 1.6973327670239603e-05, "loss": 1.4677, "step": 3145 }, { "epoch": 0.28, "learning_rate": 1.697127737707254e-05, "loss": 1.5666, "step": 3146 }, { "epoch": 0.28, "learning_rate": 1.6969226513621648e-05, "loss": 1.5513, "step": 3147 }, { "epoch": 0.28, "learning_rate": 1.6967175080054694e-05, "loss": 1.5594, "step": 3148 }, { "epoch": 0.28, "learning_rate": 1.69651230765395e-05, "loss": 1.5261, "step": 3149 }, { "epoch": 0.28, "learning_rate": 1.696307050324393e-05, "loss": 1.5225, "step": 3150 }, { "epoch": 0.28, "learning_rate": 1.696101736033589e-05, "loss": 1.4316, "step": 3151 }, { "epoch": 0.28, "learning_rate": 1.695896364798334e-05, "loss": 1.5062, "step": 3152 }, { "epoch": 0.28, "learning_rate": 1.6956909366354284e-05, "loss": 1.4269, "step": 3153 }, { "epoch": 0.28, "learning_rate": 1.6954854515616768e-05, "loss": 1.6438, "step": 3154 }, { "epoch": 0.28, "learning_rate": 1.6952799095938894e-05, "loss": 1.5524, "step": 3155 }, { "epoch": 0.28, "learning_rate": 1.69507431074888e-05, "loss": 1.5719, "step": 3156 }, { "epoch": 0.28, "learning_rate": 1.6948686550434682e-05, "loss": 1.5014, "step": 3157 }, { "epoch": 0.28, "learning_rate": 1.694662942494477e-05, "loss": 1.5658, "step": 3158 }, { "epoch": 0.28, "learning_rate": 1.694457173118735e-05, "loss": 1.466, "step": 3159 }, { "epoch": 0.28, "learning_rate": 1.694251346933075e-05, "loss": 1.5529, "step": 3160 }, { "epoch": 0.28, "learning_rate": 1.6940454639543346e-05, "loss": 1.4429, "step": 3161 }, { "epoch": 0.28, "learning_rate": 1.6938395241993562e-05, "loss": 1.5666, "step": 3162 }, { "epoch": 0.28, "learning_rate": 1.693633527684986e-05, "loss": 1.5366, "step": 3163 }, { "epoch": 0.28, "learning_rate": 1.6934274744280767e-05, "loss": 1.4887, "step": 3164 }, { "epoch": 0.28, "learning_rate": 1.6932213644454832e-05, "loss": 1.3507, "step": 3165 }, { "epoch": 0.28, "learning_rate": 1.693015197754067e-05, "loss": 1.4995, "step": 3166 }, { "epoch": 0.28, "learning_rate": 1.692808974370693e-05, "loss": 1.6009, "step": 3167 }, { "epoch": 0.28, "learning_rate": 1.6926026943122318e-05, "loss": 1.5448, "step": 3168 }, { "epoch": 0.28, "learning_rate": 1.6923963575955578e-05, "loss": 1.5035, "step": 3169 }, { "epoch": 0.28, "learning_rate": 1.6921899642375508e-05, "loss": 1.47, "step": 3170 }, { "epoch": 0.28, "learning_rate": 1.6919835142550938e-05, "loss": 1.4735, "step": 3171 }, { "epoch": 0.28, "learning_rate": 1.6917770076650763e-05, "loss": 1.4585, "step": 3172 }, { "epoch": 0.28, "learning_rate": 1.691570444484391e-05, "loss": 1.5552, "step": 3173 }, { "epoch": 0.28, "learning_rate": 1.691363824729936e-05, "loss": 1.4601, "step": 3174 }, { "epoch": 0.28, "learning_rate": 1.691157148418614e-05, "loss": 1.5042, "step": 3175 }, { "epoch": 0.28, "learning_rate": 1.6909504155673318e-05, "loss": 1.4266, "step": 3176 }, { "epoch": 0.28, "learning_rate": 1.6907436261930006e-05, "loss": 1.6461, "step": 3177 }, { "epoch": 0.28, "learning_rate": 1.6905367803125377e-05, "loss": 1.4618, "step": 3178 }, { "epoch": 0.28, "learning_rate": 1.690329877942864e-05, "loss": 1.5911, "step": 3179 }, { "epoch": 0.28, "learning_rate": 1.6901229191009046e-05, "loss": 1.5946, "step": 3180 }, { "epoch": 0.28, "learning_rate": 1.68991590380359e-05, "loss": 1.5482, "step": 3181 }, { "epoch": 0.28, "learning_rate": 1.6897088320678547e-05, "loss": 1.5495, "step": 3182 }, { "epoch": 0.28, "learning_rate": 1.6895017039106386e-05, "loss": 1.4527, "step": 3183 }, { "epoch": 0.28, "learning_rate": 1.689294519348886e-05, "loss": 1.4779, "step": 3184 }, { "epoch": 0.28, "learning_rate": 1.689087278399545e-05, "loss": 1.4776, "step": 3185 }, { "epoch": 0.28, "learning_rate": 1.688879981079569e-05, "loss": 1.4931, "step": 3186 }, { "epoch": 0.28, "learning_rate": 1.6886726274059157e-05, "loss": 1.5387, "step": 3187 }, { "epoch": 0.28, "learning_rate": 1.688465217395548e-05, "loss": 1.4875, "step": 3188 }, { "epoch": 0.28, "learning_rate": 1.6882577510654338e-05, "loss": 1.5655, "step": 3189 }, { "epoch": 0.28, "learning_rate": 1.6880502284325434e-05, "loss": 1.5012, "step": 3190 }, { "epoch": 0.28, "learning_rate": 1.687842649513854e-05, "loss": 1.5035, "step": 3191 }, { "epoch": 0.28, "learning_rate": 1.687635014326346e-05, "loss": 1.6129, "step": 3192 }, { "epoch": 0.28, "learning_rate": 1.6874273228870052e-05, "loss": 1.5428, "step": 3193 }, { "epoch": 0.28, "learning_rate": 1.687219575212822e-05, "loss": 1.5141, "step": 3194 }, { "epoch": 0.28, "learning_rate": 1.687011771320791e-05, "loss": 1.5215, "step": 3195 }, { "epoch": 0.28, "learning_rate": 1.6868039112279115e-05, "loss": 1.5105, "step": 3196 }, { "epoch": 0.28, "learning_rate": 1.686595994951187e-05, "loss": 1.4747, "step": 3197 }, { "epoch": 0.28, "learning_rate": 1.6863880225076273e-05, "loss": 1.611, "step": 3198 }, { "epoch": 0.28, "learning_rate": 1.6861799939142443e-05, "loss": 1.5015, "step": 3199 }, { "epoch": 0.28, "learning_rate": 1.6859719091880563e-05, "loss": 1.5782, "step": 3200 }, { "epoch": 0.28, "learning_rate": 1.6857637683460853e-05, "loss": 1.5828, "step": 3201 }, { "epoch": 0.28, "learning_rate": 1.6855555714053588e-05, "loss": 1.5099, "step": 3202 }, { "epoch": 0.28, "learning_rate": 1.6853473183829074e-05, "loss": 1.471, "step": 3203 }, { "epoch": 0.28, "learning_rate": 1.6851390092957684e-05, "loss": 1.5308, "step": 3204 }, { "epoch": 0.28, "learning_rate": 1.684930644160981e-05, "loss": 1.4502, "step": 3205 }, { "epoch": 0.28, "learning_rate": 1.684722222995592e-05, "loss": 1.4925, "step": 3206 }, { "epoch": 0.28, "learning_rate": 1.68451374581665e-05, "loss": 1.4278, "step": 3207 }, { "epoch": 0.28, "learning_rate": 1.6843052126412102e-05, "loss": 1.6307, "step": 3208 }, { "epoch": 0.28, "learning_rate": 1.6840966234863317e-05, "loss": 1.5568, "step": 3209 }, { "epoch": 0.28, "learning_rate": 1.6838879783690775e-05, "loss": 1.4918, "step": 3210 }, { "epoch": 0.28, "learning_rate": 1.6836792773065157e-05, "loss": 1.611, "step": 3211 }, { "epoch": 0.28, "learning_rate": 1.6834705203157197e-05, "loss": 1.3686, "step": 3212 }, { "epoch": 0.28, "learning_rate": 1.6832617074137662e-05, "loss": 1.4866, "step": 3213 }, { "epoch": 0.28, "learning_rate": 1.6830528386177375e-05, "loss": 1.4108, "step": 3214 }, { "epoch": 0.28, "learning_rate": 1.68284391394472e-05, "loss": 1.4411, "step": 3215 }, { "epoch": 0.28, "learning_rate": 1.6826349334118047e-05, "loss": 1.5462, "step": 3216 }, { "epoch": 0.28, "learning_rate": 1.6824258970360872e-05, "loss": 1.5516, "step": 3217 }, { "epoch": 0.28, "learning_rate": 1.6822168048346674e-05, "loss": 1.5341, "step": 3218 }, { "epoch": 0.28, "learning_rate": 1.6820076568246502e-05, "loss": 1.6068, "step": 3219 }, { "epoch": 0.28, "learning_rate": 1.6817984530231458e-05, "loss": 1.4897, "step": 3220 }, { "epoch": 0.28, "learning_rate": 1.6815891934472665e-05, "loss": 1.4414, "step": 3221 }, { "epoch": 0.28, "learning_rate": 1.681379878114132e-05, "loss": 1.5804, "step": 3222 }, { "epoch": 0.28, "learning_rate": 1.6811705070408645e-05, "loss": 1.5765, "step": 3223 }, { "epoch": 0.28, "learning_rate": 1.6809610802445922e-05, "loss": 1.4677, "step": 3224 }, { "epoch": 0.28, "learning_rate": 1.6807515977424466e-05, "loss": 1.5372, "step": 3225 }, { "epoch": 0.28, "learning_rate": 1.6805420595515646e-05, "loss": 1.4542, "step": 3226 }, { "epoch": 0.28, "learning_rate": 1.6803324656890877e-05, "loss": 1.5292, "step": 3227 }, { "epoch": 0.29, "learning_rate": 1.6801228161721613e-05, "loss": 1.4738, "step": 3228 }, { "epoch": 0.29, "learning_rate": 1.679913111017936e-05, "loss": 1.3957, "step": 3229 }, { "epoch": 0.29, "learning_rate": 1.6797033502435667e-05, "loss": 1.4877, "step": 3230 }, { "epoch": 0.29, "learning_rate": 1.6794935338662122e-05, "loss": 1.3921, "step": 3231 }, { "epoch": 0.29, "learning_rate": 1.679283661903037e-05, "loss": 1.4448, "step": 3232 }, { "epoch": 0.29, "learning_rate": 1.6790737343712105e-05, "loss": 1.5558, "step": 3233 }, { "epoch": 0.29, "learning_rate": 1.678863751287904e-05, "loss": 1.4981, "step": 3234 }, { "epoch": 0.29, "learning_rate": 1.6786537126702967e-05, "loss": 1.4569, "step": 3235 }, { "epoch": 0.29, "learning_rate": 1.6784436185355698e-05, "loss": 1.5152, "step": 3236 }, { "epoch": 0.29, "learning_rate": 1.67823346890091e-05, "loss": 1.4812, "step": 3237 }, { "epoch": 0.29, "learning_rate": 1.678023263783509e-05, "loss": 1.5254, "step": 3238 }, { "epoch": 0.29, "learning_rate": 1.6778130032005628e-05, "loss": 1.5752, "step": 3239 }, { "epoch": 0.29, "learning_rate": 1.6776026871692707e-05, "loss": 1.5645, "step": 3240 }, { "epoch": 0.29, "learning_rate": 1.6773923157068387e-05, "loss": 1.6075, "step": 3241 }, { "epoch": 0.29, "learning_rate": 1.6771818888304757e-05, "loss": 1.4667, "step": 3242 }, { "epoch": 0.29, "learning_rate": 1.676971406557395e-05, "loss": 1.4944, "step": 3243 }, { "epoch": 0.29, "learning_rate": 1.6767608689048164e-05, "loss": 1.5216, "step": 3244 }, { "epoch": 0.29, "learning_rate": 1.6765502758899616e-05, "loss": 1.4139, "step": 3245 }, { "epoch": 0.29, "learning_rate": 1.676339627530059e-05, "loss": 1.394, "step": 3246 }, { "epoch": 0.29, "learning_rate": 1.6761289238423405e-05, "loss": 1.524, "step": 3247 }, { "epoch": 0.29, "learning_rate": 1.675918164844042e-05, "loss": 1.5329, "step": 3248 }, { "epoch": 0.29, "learning_rate": 1.6757073505524054e-05, "loss": 1.4376, "step": 3249 }, { "epoch": 0.29, "learning_rate": 1.6754964809846757e-05, "loss": 1.4869, "step": 3250 }, { "epoch": 0.29, "learning_rate": 1.6752855561581035e-05, "loss": 1.5427, "step": 3251 }, { "epoch": 0.29, "learning_rate": 1.6750745760899432e-05, "loss": 1.6671, "step": 3252 }, { "epoch": 0.29, "learning_rate": 1.6748635407974544e-05, "loss": 1.6031, "step": 3253 }, { "epoch": 0.29, "learning_rate": 1.6746524502979005e-05, "loss": 1.4798, "step": 3254 }, { "epoch": 0.29, "learning_rate": 1.674441304608549e-05, "loss": 1.3994, "step": 3255 }, { "epoch": 0.29, "learning_rate": 1.674230103746674e-05, "loss": 1.5906, "step": 3256 }, { "epoch": 0.29, "learning_rate": 1.6740188477295517e-05, "loss": 1.367, "step": 3257 }, { "epoch": 0.29, "learning_rate": 1.6738075365744642e-05, "loss": 1.5712, "step": 3258 }, { "epoch": 0.29, "learning_rate": 1.6735961702986982e-05, "loss": 1.4886, "step": 3259 }, { "epoch": 0.29, "learning_rate": 1.6733847489195433e-05, "loss": 1.5154, "step": 3260 }, { "epoch": 0.29, "learning_rate": 1.6731732724542962e-05, "loss": 1.5428, "step": 3261 }, { "epoch": 0.29, "learning_rate": 1.6729617409202555e-05, "loss": 1.5688, "step": 3262 }, { "epoch": 0.29, "learning_rate": 1.6727501543347263e-05, "loss": 1.4794, "step": 3263 }, { "epoch": 0.29, "learning_rate": 1.6725385127150172e-05, "loss": 1.5441, "step": 3264 }, { "epoch": 0.29, "learning_rate": 1.672326816078441e-05, "loss": 1.4957, "step": 3265 }, { "epoch": 0.29, "learning_rate": 1.672115064442316e-05, "loss": 1.495, "step": 3266 }, { "epoch": 0.29, "learning_rate": 1.6719032578239643e-05, "loss": 1.5117, "step": 3267 }, { "epoch": 0.29, "learning_rate": 1.6716913962407134e-05, "loss": 1.4977, "step": 3268 }, { "epoch": 0.29, "learning_rate": 1.6714794797098934e-05, "loss": 1.6147, "step": 3269 }, { "epoch": 0.29, "learning_rate": 1.6712675082488407e-05, "loss": 1.5245, "step": 3270 }, { "epoch": 0.29, "learning_rate": 1.6710554818748955e-05, "loss": 1.5023, "step": 3271 }, { "epoch": 0.29, "learning_rate": 1.6708434006054026e-05, "loss": 1.4941, "step": 3272 }, { "epoch": 0.29, "learning_rate": 1.6706312644577113e-05, "loss": 1.5058, "step": 3273 }, { "epoch": 0.29, "learning_rate": 1.6704190734491753e-05, "loss": 1.4397, "step": 3274 }, { "epoch": 0.29, "learning_rate": 1.670206827597153e-05, "loss": 1.4955, "step": 3275 }, { "epoch": 0.29, "learning_rate": 1.6699945269190068e-05, "loss": 1.3923, "step": 3276 }, { "epoch": 0.29, "learning_rate": 1.669782171432104e-05, "loss": 1.3443, "step": 3277 }, { "epoch": 0.29, "learning_rate": 1.6695697611538166e-05, "loss": 1.5018, "step": 3278 }, { "epoch": 0.29, "learning_rate": 1.6693572961015205e-05, "loss": 1.4813, "step": 3279 }, { "epoch": 0.29, "learning_rate": 1.6691447762925966e-05, "loss": 1.4632, "step": 3280 }, { "epoch": 0.29, "learning_rate": 1.6689322017444294e-05, "loss": 1.4033, "step": 3281 }, { "epoch": 0.29, "learning_rate": 1.6687195724744095e-05, "loss": 1.5362, "step": 3282 }, { "epoch": 0.29, "learning_rate": 1.66850688849993e-05, "loss": 1.4486, "step": 3283 }, { "epoch": 0.29, "learning_rate": 1.66829414983839e-05, "loss": 1.386, "step": 3284 }, { "epoch": 0.29, "learning_rate": 1.668081356507193e-05, "loss": 1.5527, "step": 3285 }, { "epoch": 0.29, "learning_rate": 1.6678685085237453e-05, "loss": 1.5427, "step": 3286 }, { "epoch": 0.29, "learning_rate": 1.66765560590546e-05, "loss": 1.5241, "step": 3287 }, { "epoch": 0.29, "learning_rate": 1.667442648669753e-05, "loss": 1.4327, "step": 3288 }, { "epoch": 0.29, "learning_rate": 1.667229636834045e-05, "loss": 1.3744, "step": 3289 }, { "epoch": 0.29, "learning_rate": 1.667016570415762e-05, "loss": 1.6059, "step": 3290 }, { "epoch": 0.29, "learning_rate": 1.666803449432334e-05, "loss": 1.4512, "step": 3291 }, { "epoch": 0.29, "learning_rate": 1.6665902739011945e-05, "loss": 1.5549, "step": 3292 }, { "epoch": 0.29, "learning_rate": 1.6663770438397825e-05, "loss": 1.4831, "step": 3293 }, { "epoch": 0.29, "learning_rate": 1.6661637592655415e-05, "loss": 1.5277, "step": 3294 }, { "epoch": 0.29, "learning_rate": 1.665950420195919e-05, "loss": 1.4793, "step": 3295 }, { "epoch": 0.29, "learning_rate": 1.6657370266483676e-05, "loss": 1.5453, "step": 3296 }, { "epoch": 0.29, "learning_rate": 1.6655235786403432e-05, "loss": 1.4865, "step": 3297 }, { "epoch": 0.29, "learning_rate": 1.6653100761893076e-05, "loss": 1.4335, "step": 3298 }, { "epoch": 0.29, "learning_rate": 1.6650965193127254e-05, "loss": 1.425, "step": 3299 }, { "epoch": 0.29, "learning_rate": 1.664882908028068e-05, "loss": 1.4404, "step": 3300 }, { "epoch": 0.29, "learning_rate": 1.664669242352808e-05, "loss": 1.405, "step": 3301 }, { "epoch": 0.29, "learning_rate": 1.664455522304426e-05, "loss": 1.5747, "step": 3302 }, { "epoch": 0.29, "learning_rate": 1.6642417479004042e-05, "loss": 1.5898, "step": 3303 }, { "epoch": 0.29, "learning_rate": 1.6640279191582308e-05, "loss": 1.4203, "step": 3304 }, { "epoch": 0.29, "learning_rate": 1.6638140360953982e-05, "loss": 1.5497, "step": 3305 }, { "epoch": 0.29, "learning_rate": 1.6636000987294022e-05, "loss": 1.5583, "step": 3306 }, { "epoch": 0.29, "learning_rate": 1.663386107077745e-05, "loss": 1.2751, "step": 3307 }, { "epoch": 0.29, "learning_rate": 1.6631720611579316e-05, "loss": 1.4928, "step": 3308 }, { "epoch": 0.29, "learning_rate": 1.6629579609874717e-05, "loss": 1.3898, "step": 3309 }, { "epoch": 0.29, "learning_rate": 1.6627438065838804e-05, "loss": 1.466, "step": 3310 }, { "epoch": 0.29, "learning_rate": 1.6625295979646758e-05, "loss": 1.5814, "step": 3311 }, { "epoch": 0.29, "learning_rate": 1.662315335147382e-05, "loss": 1.4593, "step": 3312 }, { "epoch": 0.29, "learning_rate": 1.662101018149526e-05, "loss": 1.5605, "step": 3313 }, { "epoch": 0.29, "learning_rate": 1.6618866469886403e-05, "loss": 1.434, "step": 3314 }, { "epoch": 0.29, "learning_rate": 1.6616722216822616e-05, "loss": 1.4472, "step": 3315 }, { "epoch": 0.29, "learning_rate": 1.6614577422479304e-05, "loss": 1.484, "step": 3316 }, { "epoch": 0.29, "learning_rate": 1.661243208703193e-05, "loss": 1.5319, "step": 3317 }, { "epoch": 0.29, "learning_rate": 1.6610286210655984e-05, "loss": 1.6191, "step": 3318 }, { "epoch": 0.29, "learning_rate": 1.6608139793527016e-05, "loss": 1.3989, "step": 3319 }, { "epoch": 0.29, "learning_rate": 1.6605992835820604e-05, "loss": 1.5146, "step": 3320 }, { "epoch": 0.29, "learning_rate": 1.6603845337712393e-05, "loss": 1.4896, "step": 3321 }, { "epoch": 0.29, "learning_rate": 1.6601697299378045e-05, "loss": 1.6296, "step": 3322 }, { "epoch": 0.29, "learning_rate": 1.6599548720993286e-05, "loss": 1.4566, "step": 3323 }, { "epoch": 0.29, "learning_rate": 1.6597399602733885e-05, "loss": 1.4211, "step": 3324 }, { "epoch": 0.29, "learning_rate": 1.6595249944775637e-05, "loss": 1.5436, "step": 3325 }, { "epoch": 0.29, "learning_rate": 1.6593099747294407e-05, "loss": 1.4682, "step": 3326 }, { "epoch": 0.29, "learning_rate": 1.6590949010466086e-05, "loss": 1.4683, "step": 3327 }, { "epoch": 0.29, "learning_rate": 1.6588797734466617e-05, "loss": 1.538, "step": 3328 }, { "epoch": 0.29, "learning_rate": 1.658664591947198e-05, "loss": 1.4179, "step": 3329 }, { "epoch": 0.29, "learning_rate": 1.658449356565821e-05, "loss": 1.3352, "step": 3330 }, { "epoch": 0.29, "learning_rate": 1.6582340673201368e-05, "loss": 1.558, "step": 3331 }, { "epoch": 0.29, "learning_rate": 1.658018724227759e-05, "loss": 1.5156, "step": 3332 }, { "epoch": 0.29, "learning_rate": 1.657803327306302e-05, "loss": 1.5622, "step": 3333 }, { "epoch": 0.29, "learning_rate": 1.657587876573387e-05, "loss": 1.6442, "step": 3334 }, { "epoch": 0.29, "learning_rate": 1.6573723720466387e-05, "loss": 1.4916, "step": 3335 }, { "epoch": 0.29, "learning_rate": 1.6571568137436867e-05, "loss": 1.5247, "step": 3336 }, { "epoch": 0.29, "learning_rate": 1.6569412016821648e-05, "loss": 1.4391, "step": 3337 }, { "epoch": 0.29, "learning_rate": 1.6567255358797102e-05, "loss": 1.4875, "step": 3338 }, { "epoch": 0.29, "learning_rate": 1.6565098163539666e-05, "loss": 1.5441, "step": 3339 }, { "epoch": 0.29, "learning_rate": 1.6562940431225798e-05, "loss": 1.4764, "step": 3340 }, { "epoch": 0.3, "learning_rate": 1.6560782162032022e-05, "loss": 1.4849, "step": 3341 }, { "epoch": 0.3, "learning_rate": 1.655862335613488e-05, "loss": 1.6629, "step": 3342 }, { "epoch": 0.3, "learning_rate": 1.655646401371099e-05, "loss": 1.5047, "step": 3343 }, { "epoch": 0.3, "learning_rate": 1.6554304134936983e-05, "loss": 1.5195, "step": 3344 }, { "epoch": 0.3, "learning_rate": 1.6552143719989553e-05, "loss": 1.5314, "step": 3345 }, { "epoch": 0.3, "learning_rate": 1.654998276904543e-05, "loss": 1.6058, "step": 3346 }, { "epoch": 0.3, "learning_rate": 1.65478212822814e-05, "loss": 1.3762, "step": 3347 }, { "epoch": 0.3, "learning_rate": 1.6545659259874263e-05, "loss": 1.3859, "step": 3348 }, { "epoch": 0.3, "learning_rate": 1.65434967020009e-05, "loss": 1.3949, "step": 3349 }, { "epoch": 0.3, "learning_rate": 1.654133360883821e-05, "loss": 1.6661, "step": 3350 }, { "epoch": 0.3, "learning_rate": 1.653916998056315e-05, "loss": 1.3954, "step": 3351 }, { "epoch": 0.3, "learning_rate": 1.6537005817352714e-05, "loss": 1.4706, "step": 3352 }, { "epoch": 0.3, "learning_rate": 1.6534841119383936e-05, "loss": 1.5134, "step": 3353 }, { "epoch": 0.3, "learning_rate": 1.65326758868339e-05, "loss": 1.6022, "step": 3354 }, { "epoch": 0.3, "learning_rate": 1.653051011987974e-05, "loss": 1.4723, "step": 3355 }, { "epoch": 0.3, "learning_rate": 1.6528343818698614e-05, "loss": 1.4265, "step": 3356 }, { "epoch": 0.3, "learning_rate": 1.652617698346775e-05, "loss": 1.3848, "step": 3357 }, { "epoch": 0.3, "learning_rate": 1.6524009614364392e-05, "loss": 1.4266, "step": 3358 }, { "epoch": 0.3, "learning_rate": 1.652184171156585e-05, "loss": 1.4073, "step": 3359 }, { "epoch": 0.3, "learning_rate": 1.6519673275249465e-05, "loss": 1.4409, "step": 3360 }, { "epoch": 0.3, "learning_rate": 1.6517504305592625e-05, "loss": 1.4793, "step": 3361 }, { "epoch": 0.3, "learning_rate": 1.6515334802772766e-05, "loss": 1.5077, "step": 3362 }, { "epoch": 0.3, "learning_rate": 1.6513164766967356e-05, "loss": 1.6599, "step": 3363 }, { "epoch": 0.3, "learning_rate": 1.651099419835392e-05, "loss": 1.5258, "step": 3364 }, { "epoch": 0.3, "learning_rate": 1.650882309711002e-05, "loss": 1.5327, "step": 3365 }, { "epoch": 0.3, "learning_rate": 1.6506651463413265e-05, "loss": 1.6032, "step": 3366 }, { "epoch": 0.3, "learning_rate": 1.6504479297441305e-05, "loss": 1.6137, "step": 3367 }, { "epoch": 0.3, "learning_rate": 1.6502306599371826e-05, "loss": 1.3869, "step": 3368 }, { "epoch": 0.3, "learning_rate": 1.6500133369382573e-05, "loss": 1.4994, "step": 3369 }, { "epoch": 0.3, "learning_rate": 1.6497959607651324e-05, "loss": 1.5458, "step": 3370 }, { "epoch": 0.3, "learning_rate": 1.6495785314355903e-05, "loss": 1.4974, "step": 3371 }, { "epoch": 0.3, "learning_rate": 1.6493610489674177e-05, "loss": 1.5142, "step": 3372 }, { "epoch": 0.3, "learning_rate": 1.6491435133784058e-05, "loss": 1.4925, "step": 3373 }, { "epoch": 0.3, "learning_rate": 1.64892592468635e-05, "loss": 1.5445, "step": 3374 }, { "epoch": 0.3, "learning_rate": 1.6487082829090503e-05, "loss": 1.5695, "step": 3375 }, { "epoch": 0.3, "learning_rate": 1.6484905880643102e-05, "loss": 1.6021, "step": 3376 }, { "epoch": 0.3, "learning_rate": 1.6482728401699386e-05, "loss": 1.564, "step": 3377 }, { "epoch": 0.3, "learning_rate": 1.648055039243749e-05, "loss": 1.4855, "step": 3378 }, { "epoch": 0.3, "learning_rate": 1.6478371853035574e-05, "loss": 1.4563, "step": 3379 }, { "epoch": 0.3, "learning_rate": 1.6476192783671862e-05, "loss": 1.5319, "step": 3380 }, { "epoch": 0.3, "learning_rate": 1.6474013184524604e-05, "loss": 1.4768, "step": 3381 }, { "epoch": 0.3, "learning_rate": 1.6471833055772106e-05, "loss": 1.5095, "step": 3382 }, { "epoch": 0.3, "learning_rate": 1.6469652397592713e-05, "loss": 1.5771, "step": 3383 }, { "epoch": 0.3, "learning_rate": 1.6467471210164817e-05, "loss": 1.5753, "step": 3384 }, { "epoch": 0.3, "learning_rate": 1.646528949366684e-05, "loss": 1.5447, "step": 3385 }, { "epoch": 0.3, "learning_rate": 1.6463107248277267e-05, "loss": 1.496, "step": 3386 }, { "epoch": 0.3, "learning_rate": 1.646092447417461e-05, "loss": 1.5401, "step": 3387 }, { "epoch": 0.3, "learning_rate": 1.6458741171537434e-05, "loss": 1.3098, "step": 3388 }, { "epoch": 0.3, "learning_rate": 1.6456557340544338e-05, "loss": 1.5242, "step": 3389 }, { "epoch": 0.3, "learning_rate": 1.6454372981373978e-05, "loss": 1.4714, "step": 3390 }, { "epoch": 0.3, "learning_rate": 1.6452188094205038e-05, "loss": 1.6282, "step": 3391 }, { "epoch": 0.3, "learning_rate": 1.6450002679216254e-05, "loss": 1.3587, "step": 3392 }, { "epoch": 0.3, "learning_rate": 1.6447816736586406e-05, "loss": 1.5292, "step": 3393 }, { "epoch": 0.3, "learning_rate": 1.644563026649431e-05, "loss": 1.403, "step": 3394 }, { "epoch": 0.3, "learning_rate": 1.6443443269118834e-05, "loss": 1.5204, "step": 3395 }, { "epoch": 0.3, "learning_rate": 1.6441255744638883e-05, "loss": 1.5258, "step": 3396 }, { "epoch": 0.3, "learning_rate": 1.6439067693233407e-05, "loss": 1.49, "step": 3397 }, { "epoch": 0.3, "learning_rate": 1.64368791150814e-05, "loss": 1.4798, "step": 3398 }, { "epoch": 0.3, "learning_rate": 1.64346900103619e-05, "loss": 1.6378, "step": 3399 }, { "epoch": 0.3, "learning_rate": 1.6432500379253976e-05, "loss": 1.5844, "step": 3400 }, { "epoch": 0.3, "learning_rate": 1.6430310221936766e-05, "loss": 1.6374, "step": 3401 }, { "epoch": 0.3, "learning_rate": 1.642811953858942e-05, "loss": 1.5652, "step": 3402 }, { "epoch": 0.3, "learning_rate": 1.6425928329391162e-05, "loss": 1.6131, "step": 3403 }, { "epoch": 0.3, "learning_rate": 1.642373659452123e-05, "loss": 1.5347, "step": 3404 }, { "epoch": 0.3, "learning_rate": 1.642154433415892e-05, "loss": 1.4214, "step": 3405 }, { "epoch": 0.3, "learning_rate": 1.641935154848358e-05, "loss": 1.4941, "step": 3406 }, { "epoch": 0.3, "learning_rate": 1.6417158237674576e-05, "loss": 1.4832, "step": 3407 }, { "epoch": 0.3, "learning_rate": 1.6414964401911343e-05, "loss": 1.5784, "step": 3408 }, { "epoch": 0.3, "learning_rate": 1.6412770041373342e-05, "loss": 1.4942, "step": 3409 }, { "epoch": 0.3, "learning_rate": 1.6410575156240086e-05, "loss": 1.474, "step": 3410 }, { "epoch": 0.3, "learning_rate": 1.640837974669112e-05, "loss": 1.4527, "step": 3411 }, { "epoch": 0.3, "learning_rate": 1.6406183812906043e-05, "loss": 1.5436, "step": 3412 }, { "epoch": 0.3, "learning_rate": 1.6403987355064494e-05, "loss": 1.446, "step": 3413 }, { "epoch": 0.3, "learning_rate": 1.6401790373346155e-05, "loss": 1.4513, "step": 3414 }, { "epoch": 0.3, "learning_rate": 1.6399592867930744e-05, "loss": 1.5453, "step": 3415 }, { "epoch": 0.3, "learning_rate": 1.639739483899803e-05, "loss": 1.5815, "step": 3416 }, { "epoch": 0.3, "learning_rate": 1.6395196286727826e-05, "loss": 1.4507, "step": 3417 }, { "epoch": 0.3, "learning_rate": 1.639299721129998e-05, "loss": 1.3638, "step": 3418 }, { "epoch": 0.3, "learning_rate": 1.639079761289439e-05, "loss": 1.6078, "step": 3419 }, { "epoch": 0.3, "learning_rate": 1.6388597491690984e-05, "loss": 1.45, "step": 3420 }, { "epoch": 0.3, "learning_rate": 1.638639684786976e-05, "loss": 1.5833, "step": 3421 }, { "epoch": 0.3, "learning_rate": 1.6384195681610724e-05, "loss": 1.4863, "step": 3422 }, { "epoch": 0.3, "learning_rate": 1.638199399309395e-05, "loss": 1.5934, "step": 3423 }, { "epoch": 0.3, "learning_rate": 1.637979178249955e-05, "loss": 1.5925, "step": 3424 }, { "epoch": 0.3, "learning_rate": 1.637758905000767e-05, "loss": 1.545, "step": 3425 }, { "epoch": 0.3, "learning_rate": 1.6375385795798504e-05, "loss": 1.3739, "step": 3426 }, { "epoch": 0.3, "learning_rate": 1.637318202005229e-05, "loss": 1.4515, "step": 3427 }, { "epoch": 0.3, "learning_rate": 1.6370977722949312e-05, "loss": 1.4827, "step": 3428 }, { "epoch": 0.3, "learning_rate": 1.6368772904669882e-05, "loss": 1.5191, "step": 3429 }, { "epoch": 0.3, "learning_rate": 1.636656756539437e-05, "loss": 1.5586, "step": 3430 }, { "epoch": 0.3, "learning_rate": 1.6364361705303185e-05, "loss": 1.6034, "step": 3431 }, { "epoch": 0.3, "learning_rate": 1.6362155324576775e-05, "loss": 1.5446, "step": 3432 }, { "epoch": 0.3, "learning_rate": 1.6359948423395633e-05, "loss": 1.5448, "step": 3433 }, { "epoch": 0.3, "learning_rate": 1.635774100194029e-05, "loss": 1.437, "step": 3434 }, { "epoch": 0.3, "learning_rate": 1.6355533060391335e-05, "loss": 1.4701, "step": 3435 }, { "epoch": 0.3, "learning_rate": 1.6353324598929378e-05, "loss": 1.6115, "step": 3436 }, { "epoch": 0.3, "learning_rate": 1.6351115617735082e-05, "loss": 1.5395, "step": 3437 }, { "epoch": 0.3, "learning_rate": 1.6348906116989157e-05, "loss": 1.4288, "step": 3438 }, { "epoch": 0.3, "learning_rate": 1.634669609687235e-05, "loss": 1.5164, "step": 3439 }, { "epoch": 0.3, "learning_rate": 1.6344485557565443e-05, "loss": 1.3913, "step": 3440 }, { "epoch": 0.3, "learning_rate": 1.6342274499249283e-05, "loss": 1.5733, "step": 3441 }, { "epoch": 0.3, "learning_rate": 1.634006292210473e-05, "loss": 1.5871, "step": 3442 }, { "epoch": 0.3, "learning_rate": 1.6337850826312714e-05, "loss": 1.5674, "step": 3443 }, { "epoch": 0.3, "learning_rate": 1.6335638212054187e-05, "loss": 1.6068, "step": 3444 }, { "epoch": 0.3, "learning_rate": 1.633342507951016e-05, "loss": 1.4804, "step": 3445 }, { "epoch": 0.3, "learning_rate": 1.6331211428861665e-05, "loss": 1.4309, "step": 3446 }, { "epoch": 0.3, "learning_rate": 1.6328997260289804e-05, "loss": 1.5482, "step": 3447 }, { "epoch": 0.3, "learning_rate": 1.6326782573975693e-05, "loss": 1.451, "step": 3448 }, { "epoch": 0.3, "learning_rate": 1.6324567370100512e-05, "loss": 1.4525, "step": 3449 }, { "epoch": 0.3, "learning_rate": 1.6322351648845473e-05, "loss": 1.4736, "step": 3450 }, { "epoch": 0.3, "learning_rate": 1.6320135410391836e-05, "loss": 1.5703, "step": 3451 }, { "epoch": 0.3, "learning_rate": 1.631791865492089e-05, "loss": 1.6334, "step": 3452 }, { "epoch": 0.3, "learning_rate": 1.6315701382613988e-05, "loss": 1.472, "step": 3453 }, { "epoch": 0.3, "learning_rate": 1.6313483593652507e-05, "loss": 1.4855, "step": 3454 }, { "epoch": 0.31, "learning_rate": 1.6311265288217873e-05, "loss": 1.5484, "step": 3455 }, { "epoch": 0.31, "learning_rate": 1.630904646649156e-05, "loss": 1.5687, "step": 3456 }, { "epoch": 0.31, "learning_rate": 1.630682712865507e-05, "loss": 1.5197, "step": 3457 }, { "epoch": 0.31, "learning_rate": 1.630460727488996e-05, "loss": 1.6939, "step": 3458 }, { "epoch": 0.31, "learning_rate": 1.630238690537782e-05, "loss": 1.4236, "step": 3459 }, { "epoch": 0.31, "learning_rate": 1.6300166020300294e-05, "loss": 1.5122, "step": 3460 }, { "epoch": 0.31, "learning_rate": 1.6297944619839055e-05, "loss": 1.5188, "step": 3461 }, { "epoch": 0.31, "learning_rate": 1.629572270417583e-05, "loss": 1.6281, "step": 3462 }, { "epoch": 0.31, "learning_rate": 1.6293500273492372e-05, "loss": 1.5438, "step": 3463 }, { "epoch": 0.31, "learning_rate": 1.62912773279705e-05, "loss": 1.6449, "step": 3464 }, { "epoch": 0.31, "learning_rate": 1.6289053867792052e-05, "loss": 1.5393, "step": 3465 }, { "epoch": 0.31, "learning_rate": 1.6286829893138923e-05, "loss": 1.6148, "step": 3466 }, { "epoch": 0.31, "learning_rate": 1.628460540419304e-05, "loss": 1.5361, "step": 3467 }, { "epoch": 0.31, "learning_rate": 1.6282380401136383e-05, "loss": 1.4588, "step": 3468 }, { "epoch": 0.31, "learning_rate": 1.628015488415096e-05, "loss": 1.4893, "step": 3469 }, { "epoch": 0.31, "learning_rate": 1.6277928853418832e-05, "loss": 1.4102, "step": 3470 }, { "epoch": 0.31, "learning_rate": 1.6275702309122102e-05, "loss": 1.3578, "step": 3471 }, { "epoch": 0.31, "learning_rate": 1.627347525144291e-05, "loss": 1.586, "step": 3472 }, { "epoch": 0.31, "learning_rate": 1.627124768056344e-05, "loss": 1.4879, "step": 3473 }, { "epoch": 0.31, "learning_rate": 1.6269019596665917e-05, "loss": 1.4531, "step": 3474 }, { "epoch": 0.31, "learning_rate": 1.6266790999932614e-05, "loss": 1.5406, "step": 3475 }, { "epoch": 0.31, "learning_rate": 1.6264561890545833e-05, "loss": 1.497, "step": 3476 }, { "epoch": 0.31, "learning_rate": 1.626233226868793e-05, "loss": 1.4945, "step": 3477 }, { "epoch": 0.31, "learning_rate": 1.6260102134541296e-05, "loss": 1.5091, "step": 3478 }, { "epoch": 0.31, "learning_rate": 1.625787148828837e-05, "loss": 1.4354, "step": 3479 }, { "epoch": 0.31, "learning_rate": 1.625564033011163e-05, "loss": 1.6424, "step": 3480 }, { "epoch": 0.31, "learning_rate": 1.6253408660193594e-05, "loss": 1.5354, "step": 3481 }, { "epoch": 0.31, "learning_rate": 1.625117647871682e-05, "loss": 1.3126, "step": 3482 }, { "epoch": 0.31, "learning_rate": 1.624894378586392e-05, "loss": 1.55, "step": 3483 }, { "epoch": 0.31, "learning_rate": 1.6246710581817525e-05, "loss": 1.532, "step": 3484 }, { "epoch": 0.31, "learning_rate": 1.6244476866760336e-05, "loss": 1.5495, "step": 3485 }, { "epoch": 0.31, "learning_rate": 1.6242242640875075e-05, "loss": 1.5344, "step": 3486 }, { "epoch": 0.31, "learning_rate": 1.624000790434451e-05, "loss": 1.5425, "step": 3487 }, { "epoch": 0.31, "learning_rate": 1.623777265735146e-05, "loss": 1.4423, "step": 3488 }, { "epoch": 0.31, "learning_rate": 1.623553690007877e-05, "loss": 1.4379, "step": 3489 }, { "epoch": 0.31, "learning_rate": 1.6233300632709345e-05, "loss": 1.3852, "step": 3490 }, { "epoch": 0.31, "learning_rate": 1.6231063855426112e-05, "loss": 1.5192, "step": 3491 }, { "epoch": 0.31, "learning_rate": 1.622882656841206e-05, "loss": 1.5176, "step": 3492 }, { "epoch": 0.31, "learning_rate": 1.6226588771850207e-05, "loss": 1.5727, "step": 3493 }, { "epoch": 0.31, "learning_rate": 1.6224350465923612e-05, "loss": 1.3609, "step": 3494 }, { "epoch": 0.31, "learning_rate": 1.622211165081538e-05, "loss": 1.4459, "step": 3495 }, { "epoch": 0.31, "learning_rate": 1.6219872326708662e-05, "loss": 1.5781, "step": 3496 }, { "epoch": 0.31, "learning_rate": 1.6217632493786638e-05, "loss": 1.4341, "step": 3497 }, { "epoch": 0.31, "learning_rate": 1.6215392152232545e-05, "loss": 1.5396, "step": 3498 }, { "epoch": 0.31, "learning_rate": 1.6213151302229645e-05, "loss": 1.4487, "step": 3499 }, { "epoch": 0.31, "learning_rate": 1.6210909943961254e-05, "loss": 1.5535, "step": 3500 }, { "epoch": 0.31, "learning_rate": 1.6208668077610727e-05, "loss": 1.5681, "step": 3501 }, { "epoch": 0.31, "learning_rate": 1.620642570336146e-05, "loss": 1.5277, "step": 3502 }, { "epoch": 0.31, "learning_rate": 1.620418282139689e-05, "loss": 1.493, "step": 3503 }, { "epoch": 0.31, "learning_rate": 1.6201939431900492e-05, "loss": 1.4879, "step": 3504 }, { "epoch": 0.31, "learning_rate": 1.6199695535055788e-05, "loss": 1.5285, "step": 3505 }, { "epoch": 0.31, "learning_rate": 1.619745113104634e-05, "loss": 1.4933, "step": 3506 }, { "epoch": 0.31, "learning_rate": 1.619520622005575e-05, "loss": 1.4158, "step": 3507 }, { "epoch": 0.31, "learning_rate": 1.6192960802267662e-05, "loss": 1.5724, "step": 3508 }, { "epoch": 0.31, "learning_rate": 1.619071487786577e-05, "loss": 1.3958, "step": 3509 }, { "epoch": 0.31, "learning_rate": 1.6188468447033786e-05, "loss": 1.4512, "step": 3510 }, { "epoch": 0.31, "learning_rate": 1.618622150995549e-05, "loss": 1.5704, "step": 3511 }, { "epoch": 0.31, "learning_rate": 1.6183974066814686e-05, "loss": 1.5748, "step": 3512 }, { "epoch": 0.31, "learning_rate": 1.618172611779523e-05, "loss": 1.5932, "step": 3513 }, { "epoch": 0.31, "learning_rate": 1.6179477663081017e-05, "loss": 1.5292, "step": 3514 }, { "epoch": 0.31, "learning_rate": 1.6177228702855978e-05, "loss": 1.4377, "step": 3515 }, { "epoch": 0.31, "learning_rate": 1.617497923730409e-05, "loss": 1.4984, "step": 3516 }, { "epoch": 0.31, "learning_rate": 1.6172729266609364e-05, "loss": 1.4911, "step": 3517 }, { "epoch": 0.31, "learning_rate": 1.6170478790955866e-05, "loss": 1.5446, "step": 3518 }, { "epoch": 0.31, "learning_rate": 1.6168227810527693e-05, "loss": 1.5424, "step": 3519 }, { "epoch": 0.31, "learning_rate": 1.6165976325508987e-05, "loss": 1.508, "step": 3520 }, { "epoch": 0.31, "learning_rate": 1.6163724336083925e-05, "loss": 1.5075, "step": 3521 }, { "epoch": 0.31, "learning_rate": 1.616147184243674e-05, "loss": 1.5317, "step": 3522 }, { "epoch": 0.31, "learning_rate": 1.615921884475169e-05, "loss": 1.4578, "step": 3523 }, { "epoch": 0.31, "learning_rate": 1.615696534321308e-05, "loss": 1.484, "step": 3524 }, { "epoch": 0.31, "learning_rate": 1.6154711338005262e-05, "loss": 1.6017, "step": 3525 }, { "epoch": 0.31, "learning_rate": 1.615245682931262e-05, "loss": 1.5237, "step": 3526 }, { "epoch": 0.31, "learning_rate": 1.6150201817319586e-05, "loss": 1.3962, "step": 3527 }, { "epoch": 0.31, "learning_rate": 1.614794630221063e-05, "loss": 1.533, "step": 3528 }, { "epoch": 0.31, "learning_rate": 1.6145690284170267e-05, "loss": 1.466, "step": 3529 }, { "epoch": 0.31, "learning_rate": 1.6143433763383045e-05, "loss": 1.4062, "step": 3530 }, { "epoch": 0.31, "learning_rate": 1.614117674003356e-05, "loss": 1.4662, "step": 3531 }, { "epoch": 0.31, "learning_rate": 1.6138919214306448e-05, "loss": 1.5103, "step": 3532 }, { "epoch": 0.31, "learning_rate": 1.613666118638639e-05, "loss": 1.4994, "step": 3533 }, { "epoch": 0.31, "learning_rate": 1.613440265645809e-05, "loss": 1.5707, "step": 3534 }, { "epoch": 0.31, "learning_rate": 1.6132143624706323e-05, "loss": 1.5018, "step": 3535 }, { "epoch": 0.31, "learning_rate": 1.612988409131588e-05, "loss": 1.407, "step": 3536 }, { "epoch": 0.31, "learning_rate": 1.61276240564716e-05, "loss": 1.5648, "step": 3537 }, { "epoch": 0.31, "learning_rate": 1.612536352035837e-05, "loss": 1.5052, "step": 3538 }, { "epoch": 0.31, "learning_rate": 1.6123102483161105e-05, "loss": 1.6002, "step": 3539 }, { "epoch": 0.31, "learning_rate": 1.612084094506478e-05, "loss": 1.5117, "step": 3540 }, { "epoch": 0.31, "learning_rate": 1.6118578906254386e-05, "loss": 1.5566, "step": 3541 }, { "epoch": 0.31, "learning_rate": 1.6116316366914978e-05, "loss": 1.5772, "step": 3542 }, { "epoch": 0.31, "learning_rate": 1.6114053327231643e-05, "loss": 1.5314, "step": 3543 }, { "epoch": 0.31, "learning_rate": 1.6111789787389506e-05, "loss": 1.5299, "step": 3544 }, { "epoch": 0.31, "learning_rate": 1.6109525747573733e-05, "loss": 1.4681, "step": 3545 }, { "epoch": 0.31, "learning_rate": 1.6107261207969537e-05, "loss": 1.5088, "step": 3546 }, { "epoch": 0.31, "learning_rate": 1.6104996168762165e-05, "loss": 1.4951, "step": 3547 }, { "epoch": 0.31, "learning_rate": 1.610273063013691e-05, "loss": 1.581, "step": 3548 }, { "epoch": 0.31, "learning_rate": 1.6100464592279102e-05, "loss": 1.5852, "step": 3549 }, { "epoch": 0.31, "learning_rate": 1.6098198055374117e-05, "loss": 1.5, "step": 3550 }, { "epoch": 0.31, "learning_rate": 1.6095931019607367e-05, "loss": 1.4896, "step": 3551 }, { "epoch": 0.31, "learning_rate": 1.6093663485164304e-05, "loss": 1.6599, "step": 3552 }, { "epoch": 0.31, "learning_rate": 1.6091395452230426e-05, "loss": 1.5175, "step": 3553 }, { "epoch": 0.31, "learning_rate": 1.6089126920991273e-05, "loss": 1.5676, "step": 3554 }, { "epoch": 0.31, "learning_rate": 1.608685789163241e-05, "loss": 1.4393, "step": 3555 }, { "epoch": 0.31, "learning_rate": 1.6084588364339468e-05, "loss": 1.4936, "step": 3556 }, { "epoch": 0.31, "learning_rate": 1.6082318339298094e-05, "loss": 1.6246, "step": 3557 }, { "epoch": 0.31, "learning_rate": 1.6080047816693993e-05, "loss": 1.7129, "step": 3558 }, { "epoch": 0.31, "learning_rate": 1.6077776796712903e-05, "loss": 1.5381, "step": 3559 }, { "epoch": 0.31, "learning_rate": 1.6075505279540604e-05, "loss": 1.5635, "step": 3560 }, { "epoch": 0.31, "learning_rate": 1.6073233265362918e-05, "loss": 1.4655, "step": 3561 }, { "epoch": 0.31, "learning_rate": 1.6070960754365706e-05, "loss": 1.6333, "step": 3562 }, { "epoch": 0.31, "learning_rate": 1.606868774673487e-05, "loss": 1.5497, "step": 3563 }, { "epoch": 0.31, "learning_rate": 1.6066414242656353e-05, "loss": 1.498, "step": 3564 }, { "epoch": 0.31, "learning_rate": 1.6064140242316137e-05, "loss": 1.5549, "step": 3565 }, { "epoch": 0.31, "learning_rate": 1.6061865745900253e-05, "loss": 1.5562, "step": 3566 }, { "epoch": 0.31, "learning_rate": 1.6059590753594755e-05, "loss": 1.4191, "step": 3567 }, { "epoch": 0.32, "learning_rate": 1.6057315265585758e-05, "loss": 1.5769, "step": 3568 }, { "epoch": 0.32, "learning_rate": 1.6055039282059398e-05, "loss": 1.4378, "step": 3569 }, { "epoch": 0.32, "learning_rate": 1.6052762803201875e-05, "loss": 1.4448, "step": 3570 }, { "epoch": 0.32, "learning_rate": 1.60504858291994e-05, "loss": 1.4486, "step": 3571 }, { "epoch": 0.32, "learning_rate": 1.6048208360238252e-05, "loss": 1.4256, "step": 3572 }, { "epoch": 0.32, "learning_rate": 1.6045930396504736e-05, "loss": 1.4995, "step": 3573 }, { "epoch": 0.32, "learning_rate": 1.6043651938185195e-05, "loss": 1.4895, "step": 3574 }, { "epoch": 0.32, "learning_rate": 1.6041372985466024e-05, "loss": 1.5245, "step": 3575 }, { "epoch": 0.32, "learning_rate": 1.6039093538533652e-05, "loss": 1.5512, "step": 3576 }, { "epoch": 0.32, "learning_rate": 1.6036813597574545e-05, "loss": 1.5103, "step": 3577 }, { "epoch": 0.32, "learning_rate": 1.6034533162775217e-05, "loss": 1.5627, "step": 3578 }, { "epoch": 0.32, "learning_rate": 1.6032252234322213e-05, "loss": 1.4475, "step": 3579 }, { "epoch": 0.32, "learning_rate": 1.6029970812402133e-05, "loss": 1.4837, "step": 3580 }, { "epoch": 0.32, "learning_rate": 1.6027688897201597e-05, "loss": 1.5094, "step": 3581 }, { "epoch": 0.32, "learning_rate": 1.6025406488907286e-05, "loss": 1.5186, "step": 3582 }, { "epoch": 0.32, "learning_rate": 1.6023123587705908e-05, "loss": 1.377, "step": 3583 }, { "epoch": 0.32, "learning_rate": 1.602084019378421e-05, "loss": 1.634, "step": 3584 }, { "epoch": 0.32, "learning_rate": 1.6018556307328994e-05, "loss": 1.5931, "step": 3585 }, { "epoch": 0.32, "learning_rate": 1.6016271928527087e-05, "loss": 1.5789, "step": 3586 }, { "epoch": 0.32, "learning_rate": 1.601398705756537e-05, "loss": 1.5093, "step": 3587 }, { "epoch": 0.32, "learning_rate": 1.601170169463074e-05, "loss": 1.4836, "step": 3588 }, { "epoch": 0.32, "learning_rate": 1.6009415839910166e-05, "loss": 1.5306, "step": 3589 }, { "epoch": 0.32, "learning_rate": 1.600712949359064e-05, "loss": 1.5116, "step": 3590 }, { "epoch": 0.32, "learning_rate": 1.6004842655859184e-05, "loss": 1.4988, "step": 3591 }, { "epoch": 0.32, "learning_rate": 1.600255532690289e-05, "loss": 1.4298, "step": 3592 }, { "epoch": 0.32, "learning_rate": 1.600026750690886e-05, "loss": 1.4734, "step": 3593 }, { "epoch": 0.32, "learning_rate": 1.5997979196064252e-05, "loss": 1.5363, "step": 3594 }, { "epoch": 0.32, "learning_rate": 1.5995690394556257e-05, "loss": 1.4994, "step": 3595 }, { "epoch": 0.32, "learning_rate": 1.599340110257212e-05, "loss": 1.4501, "step": 3596 }, { "epoch": 0.32, "learning_rate": 1.5991111320299106e-05, "loss": 1.513, "step": 3597 }, { "epoch": 0.32, "learning_rate": 1.5988821047924537e-05, "loss": 1.3413, "step": 3598 }, { "epoch": 0.32, "learning_rate": 1.598653028563576e-05, "loss": 1.5445, "step": 3599 }, { "epoch": 0.32, "learning_rate": 1.598423903362018e-05, "loss": 1.4446, "step": 3600 }, { "epoch": 0.32, "learning_rate": 1.5981947292065226e-05, "loss": 1.5148, "step": 3601 }, { "epoch": 0.32, "learning_rate": 1.5979655061158373e-05, "loss": 1.5074, "step": 3602 }, { "epoch": 0.32, "learning_rate": 1.597736234108714e-05, "loss": 1.545, "step": 3603 }, { "epoch": 0.32, "learning_rate": 1.597506913203908e-05, "loss": 1.5257, "step": 3604 }, { "epoch": 0.32, "learning_rate": 1.5972775434201797e-05, "loss": 1.5666, "step": 3605 }, { "epoch": 0.32, "learning_rate": 1.597048124776291e-05, "loss": 1.478, "step": 3606 }, { "epoch": 0.32, "learning_rate": 1.596818657291011e-05, "loss": 1.5297, "step": 3607 }, { "epoch": 0.32, "learning_rate": 1.59658914098311e-05, "loss": 1.4224, "step": 3608 }, { "epoch": 0.32, "learning_rate": 1.5963595758713646e-05, "loss": 1.4904, "step": 3609 }, { "epoch": 0.32, "learning_rate": 1.5961299619745537e-05, "loss": 1.4031, "step": 3610 }, { "epoch": 0.32, "learning_rate": 1.595900299311461e-05, "loss": 1.6112, "step": 3611 }, { "epoch": 0.32, "learning_rate": 1.595670587900874e-05, "loss": 1.5206, "step": 3612 }, { "epoch": 0.32, "learning_rate": 1.5954408277615837e-05, "loss": 1.4796, "step": 3613 }, { "epoch": 0.32, "learning_rate": 1.5952110189123863e-05, "loss": 1.5372, "step": 3614 }, { "epoch": 0.32, "learning_rate": 1.5949811613720813e-05, "loss": 1.5964, "step": 3615 }, { "epoch": 0.32, "learning_rate": 1.5947512551594715e-05, "loss": 1.4016, "step": 3616 }, { "epoch": 0.32, "learning_rate": 1.5945213002933644e-05, "loss": 1.3755, "step": 3617 }, { "epoch": 0.32, "learning_rate": 1.5942912967925724e-05, "loss": 1.4712, "step": 3618 }, { "epoch": 0.32, "learning_rate": 1.5940612446759096e-05, "loss": 1.5558, "step": 3619 }, { "epoch": 0.32, "learning_rate": 1.5938311439621964e-05, "loss": 1.5825, "step": 3620 }, { "epoch": 0.32, "learning_rate": 1.593600994670255e-05, "loss": 1.5292, "step": 3621 }, { "epoch": 0.32, "learning_rate": 1.5933707968189137e-05, "loss": 1.5439, "step": 3622 }, { "epoch": 0.32, "learning_rate": 1.5931405504270035e-05, "loss": 1.5513, "step": 3623 }, { "epoch": 0.32, "learning_rate": 1.5929102555133598e-05, "loss": 1.5287, "step": 3624 }, { "epoch": 0.32, "learning_rate": 1.5926799120968213e-05, "loss": 1.5107, "step": 3625 }, { "epoch": 0.32, "learning_rate": 1.5924495201962317e-05, "loss": 1.4639, "step": 3626 }, { "epoch": 0.32, "learning_rate": 1.592219079830438e-05, "loss": 1.4906, "step": 3627 }, { "epoch": 0.32, "learning_rate": 1.5919885910182915e-05, "loss": 1.4861, "step": 3628 }, { "epoch": 0.32, "learning_rate": 1.5917580537786467e-05, "loss": 1.4249, "step": 3629 }, { "epoch": 0.32, "learning_rate": 1.5915274681303635e-05, "loss": 1.3953, "step": 3630 }, { "epoch": 0.32, "learning_rate": 1.591296834092304e-05, "loss": 1.4964, "step": 3631 }, { "epoch": 0.32, "learning_rate": 1.5910661516833358e-05, "loss": 1.4771, "step": 3632 }, { "epoch": 0.32, "learning_rate": 1.5908354209223297e-05, "loss": 1.5415, "step": 3633 }, { "epoch": 0.32, "learning_rate": 1.5906046418281606e-05, "loss": 1.5002, "step": 3634 }, { "epoch": 0.32, "learning_rate": 1.5903738144197075e-05, "loss": 1.4716, "step": 3635 }, { "epoch": 0.32, "learning_rate": 1.5901429387158526e-05, "loss": 1.5429, "step": 3636 }, { "epoch": 0.32, "learning_rate": 1.589912014735483e-05, "loss": 1.4021, "step": 3637 }, { "epoch": 0.32, "learning_rate": 1.5896810424974892e-05, "loss": 1.5133, "step": 3638 }, { "epoch": 0.32, "learning_rate": 1.5894500220207663e-05, "loss": 1.4735, "step": 3639 }, { "epoch": 0.32, "learning_rate": 1.5892189533242124e-05, "loss": 1.4651, "step": 3640 }, { "epoch": 0.32, "learning_rate": 1.5889878364267307e-05, "loss": 1.4477, "step": 3641 }, { "epoch": 0.32, "learning_rate": 1.5887566713472264e-05, "loss": 1.4825, "step": 3642 }, { "epoch": 0.32, "learning_rate": 1.5885254581046116e-05, "loss": 1.5041, "step": 3643 }, { "epoch": 0.32, "learning_rate": 1.588294196717799e-05, "loss": 1.4515, "step": 3644 }, { "epoch": 0.32, "learning_rate": 1.5880628872057077e-05, "loss": 1.6437, "step": 3645 }, { "epoch": 0.32, "learning_rate": 1.58783152958726e-05, "loss": 1.431, "step": 3646 }, { "epoch": 0.32, "learning_rate": 1.5876001238813822e-05, "loss": 1.5178, "step": 3647 }, { "epoch": 0.32, "learning_rate": 1.5873686701070036e-05, "loss": 1.5995, "step": 3648 }, { "epoch": 0.32, "learning_rate": 1.587137168283059e-05, "loss": 1.5608, "step": 3649 }, { "epoch": 0.32, "learning_rate": 1.5869056184284862e-05, "loss": 1.4727, "step": 3650 }, { "epoch": 0.32, "learning_rate": 1.5866740205622267e-05, "loss": 1.5637, "step": 3651 }, { "epoch": 0.32, "learning_rate": 1.5864423747032268e-05, "loss": 1.414, "step": 3652 }, { "epoch": 0.32, "learning_rate": 1.586210680870436e-05, "loss": 1.4797, "step": 3653 }, { "epoch": 0.32, "learning_rate": 1.585978939082808e-05, "loss": 1.5122, "step": 3654 }, { "epoch": 0.32, "learning_rate": 1.5857471493593003e-05, "loss": 1.5221, "step": 3655 }, { "epoch": 0.32, "learning_rate": 1.5855153117188745e-05, "loss": 1.7111, "step": 3656 }, { "epoch": 0.32, "learning_rate": 1.585283426180496e-05, "loss": 1.4794, "step": 3657 }, { "epoch": 0.32, "learning_rate": 1.585051492763134e-05, "loss": 1.4397, "step": 3658 }, { "epoch": 0.32, "learning_rate": 1.5848195114857622e-05, "loss": 1.5157, "step": 3659 }, { "epoch": 0.32, "learning_rate": 1.5845874823673573e-05, "loss": 1.5746, "step": 3660 }, { "epoch": 0.32, "learning_rate": 1.5843554054269007e-05, "loss": 1.3943, "step": 3661 }, { "epoch": 0.32, "learning_rate": 1.584123280683377e-05, "loss": 1.5201, "step": 3662 }, { "epoch": 0.32, "learning_rate": 1.583891108155776e-05, "loss": 1.4123, "step": 3663 }, { "epoch": 0.32, "learning_rate": 1.5836588878630892e-05, "loss": 1.6222, "step": 3664 }, { "epoch": 0.32, "learning_rate": 1.5834266198243147e-05, "loss": 1.4363, "step": 3665 }, { "epoch": 0.32, "learning_rate": 1.583194304058452e-05, "loss": 1.4551, "step": 3666 }, { "epoch": 0.32, "learning_rate": 1.5829619405845066e-05, "loss": 1.4448, "step": 3667 }, { "epoch": 0.32, "learning_rate": 1.582729529421486e-05, "loss": 1.4201, "step": 3668 }, { "epoch": 0.32, "learning_rate": 1.5824970705884034e-05, "loss": 1.5061, "step": 3669 }, { "epoch": 0.32, "learning_rate": 1.5822645641042744e-05, "loss": 1.5277, "step": 3670 }, { "epoch": 0.32, "learning_rate": 1.5820320099881196e-05, "loss": 1.4311, "step": 3671 }, { "epoch": 0.32, "learning_rate": 1.581799408258963e-05, "loss": 1.4403, "step": 3672 }, { "epoch": 0.32, "learning_rate": 1.581566758935832e-05, "loss": 1.4765, "step": 3673 }, { "epoch": 0.32, "learning_rate": 1.5813340620377592e-05, "loss": 1.6217, "step": 3674 }, { "epoch": 0.32, "learning_rate": 1.5811013175837796e-05, "loss": 1.4433, "step": 3675 }, { "epoch": 0.32, "learning_rate": 1.5808685255929334e-05, "loss": 1.5468, "step": 3676 }, { "epoch": 0.32, "learning_rate": 1.5806356860842638e-05, "loss": 1.4644, "step": 3677 }, { "epoch": 0.32, "learning_rate": 1.580402799076818e-05, "loss": 1.4273, "step": 3678 }, { "epoch": 0.32, "learning_rate": 1.5801698645896476e-05, "loss": 1.4267, "step": 3679 }, { "epoch": 0.32, "learning_rate": 1.579936882641808e-05, "loss": 1.5479, "step": 3680 }, { "epoch": 0.33, "learning_rate": 1.579703853252357e-05, "loss": 1.5703, "step": 3681 }, { "epoch": 0.33, "learning_rate": 1.579470776440359e-05, "loss": 1.3994, "step": 3682 }, { "epoch": 0.33, "learning_rate": 1.5792376522248805e-05, "loss": 1.4693, "step": 3683 }, { "epoch": 0.33, "learning_rate": 1.5790044806249913e-05, "loss": 1.5909, "step": 3684 }, { "epoch": 0.33, "learning_rate": 1.578771261659767e-05, "loss": 1.5614, "step": 3685 }, { "epoch": 0.33, "learning_rate": 1.578537995348285e-05, "loss": 1.4699, "step": 3686 }, { "epoch": 0.33, "learning_rate": 1.5783046817096286e-05, "loss": 1.5479, "step": 3687 }, { "epoch": 0.33, "learning_rate": 1.5780713207628834e-05, "loss": 1.4486, "step": 3688 }, { "epoch": 0.33, "learning_rate": 1.5778379125271396e-05, "loss": 1.4673, "step": 3689 }, { "epoch": 0.33, "learning_rate": 1.577604457021491e-05, "loss": 1.5206, "step": 3690 }, { "epoch": 0.33, "learning_rate": 1.5773709542650356e-05, "loss": 1.501, "step": 3691 }, { "epoch": 0.33, "learning_rate": 1.577137404276875e-05, "loss": 1.4695, "step": 3692 }, { "epoch": 0.33, "learning_rate": 1.5769038070761144e-05, "loss": 1.3317, "step": 3693 }, { "epoch": 0.33, "learning_rate": 1.5766701626818634e-05, "loss": 1.5114, "step": 3694 }, { "epoch": 0.33, "learning_rate": 1.5764364711132355e-05, "loss": 1.5346, "step": 3695 }, { "epoch": 0.33, "learning_rate": 1.576202732389347e-05, "loss": 1.4803, "step": 3696 }, { "epoch": 0.33, "learning_rate": 1.5759689465293197e-05, "loss": 1.3822, "step": 3697 }, { "epoch": 0.33, "learning_rate": 1.5757351135522776e-05, "loss": 1.4163, "step": 3698 }, { "epoch": 0.33, "learning_rate": 1.57550123347735e-05, "loss": 1.4307, "step": 3699 }, { "epoch": 0.33, "learning_rate": 1.5752673063236693e-05, "loss": 1.5649, "step": 3700 }, { "epoch": 0.33, "learning_rate": 1.5750333321103716e-05, "loss": 1.4573, "step": 3701 }, { "epoch": 0.33, "learning_rate": 1.5747993108565975e-05, "loss": 1.5592, "step": 3702 }, { "epoch": 0.33, "learning_rate": 1.5745652425814907e-05, "loss": 1.551, "step": 3703 }, { "epoch": 0.33, "learning_rate": 1.5743311273041994e-05, "loss": 1.5048, "step": 3704 }, { "epoch": 0.33, "learning_rate": 1.574096965043875e-05, "loss": 1.3913, "step": 3705 }, { "epoch": 0.33, "learning_rate": 1.5738627558196734e-05, "loss": 1.5537, "step": 3706 }, { "epoch": 0.33, "learning_rate": 1.5736284996507537e-05, "loss": 1.4252, "step": 3707 }, { "epoch": 0.33, "learning_rate": 1.5733941965562798e-05, "loss": 1.4315, "step": 3708 }, { "epoch": 0.33, "learning_rate": 1.5731598465554178e-05, "loss": 1.4755, "step": 3709 }, { "epoch": 0.33, "learning_rate": 1.5729254496673398e-05, "loss": 1.4266, "step": 3710 }, { "epoch": 0.33, "learning_rate": 1.5726910059112197e-05, "loss": 1.4793, "step": 3711 }, { "epoch": 0.33, "learning_rate": 1.5724565153062363e-05, "loss": 1.5741, "step": 3712 }, { "epoch": 0.33, "learning_rate": 1.5722219778715727e-05, "loss": 1.4256, "step": 3713 }, { "epoch": 0.33, "learning_rate": 1.5719873936264143e-05, "loss": 1.6066, "step": 3714 }, { "epoch": 0.33, "learning_rate": 1.5717527625899516e-05, "loss": 1.5204, "step": 3715 }, { "epoch": 0.33, "learning_rate": 1.571518084781379e-05, "loss": 1.5803, "step": 3716 }, { "epoch": 0.33, "learning_rate": 1.5712833602198934e-05, "loss": 1.5214, "step": 3717 }, { "epoch": 0.33, "learning_rate": 1.571048588924697e-05, "loss": 1.448, "step": 3718 }, { "epoch": 0.33, "learning_rate": 1.570813770914995e-05, "loss": 1.4735, "step": 3719 }, { "epoch": 0.33, "learning_rate": 1.5705789062099968e-05, "loss": 1.2943, "step": 3720 }, { "epoch": 0.33, "learning_rate": 1.570343994828915e-05, "loss": 1.5136, "step": 3721 }, { "epoch": 0.33, "learning_rate": 1.570109036790967e-05, "loss": 1.4642, "step": 3722 }, { "epoch": 0.33, "learning_rate": 1.569874032115373e-05, "loss": 1.4198, "step": 3723 }, { "epoch": 0.33, "learning_rate": 1.569638980821358e-05, "loss": 1.3169, "step": 3724 }, { "epoch": 0.33, "learning_rate": 1.56940388292815e-05, "loss": 1.5021, "step": 3725 }, { "epoch": 0.33, "learning_rate": 1.5691687384549812e-05, "loss": 1.5553, "step": 3726 }, { "epoch": 0.33, "learning_rate": 1.5689335474210882e-05, "loss": 1.5052, "step": 3727 }, { "epoch": 0.33, "learning_rate": 1.5686983098457095e-05, "loss": 1.3979, "step": 3728 }, { "epoch": 0.33, "learning_rate": 1.5684630257480893e-05, "loss": 1.4437, "step": 3729 }, { "epoch": 0.33, "learning_rate": 1.568227695147475e-05, "loss": 1.3949, "step": 3730 }, { "epoch": 0.33, "learning_rate": 1.5679923180631177e-05, "loss": 1.5627, "step": 3731 }, { "epoch": 0.33, "learning_rate": 1.5677568945142727e-05, "loss": 1.4893, "step": 3732 }, { "epoch": 0.33, "learning_rate": 1.567521424520198e-05, "loss": 1.5777, "step": 3733 }, { "epoch": 0.33, "learning_rate": 1.567285908100157e-05, "loss": 1.4887, "step": 3734 }, { "epoch": 0.33, "learning_rate": 1.5670503452734157e-05, "loss": 1.4298, "step": 3735 }, { "epoch": 0.33, "learning_rate": 1.5668147360592444e-05, "loss": 1.4713, "step": 3736 }, { "epoch": 0.33, "learning_rate": 1.5665790804769162e-05, "loss": 1.5086, "step": 3737 }, { "epoch": 0.33, "learning_rate": 1.56634337854571e-05, "loss": 1.5825, "step": 3738 }, { "epoch": 0.33, "learning_rate": 1.566107630284907e-05, "loss": 1.5653, "step": 3739 }, { "epoch": 0.33, "learning_rate": 1.5658718357137926e-05, "loss": 1.505, "step": 3740 }, { "epoch": 0.33, "learning_rate": 1.5656359948516557e-05, "loss": 1.5374, "step": 3741 }, { "epoch": 0.33, "learning_rate": 1.5654001077177894e-05, "loss": 1.4902, "step": 3742 }, { "epoch": 0.33, "learning_rate": 1.56516417433149e-05, "loss": 1.3855, "step": 3743 }, { "epoch": 0.33, "learning_rate": 1.5649281947120587e-05, "loss": 1.4084, "step": 3744 }, { "epoch": 0.33, "learning_rate": 1.564692168878799e-05, "loss": 1.5449, "step": 3745 }, { "epoch": 0.33, "learning_rate": 1.5644560968510194e-05, "loss": 1.3972, "step": 3746 }, { "epoch": 0.33, "learning_rate": 1.5642199786480315e-05, "loss": 1.501, "step": 3747 }, { "epoch": 0.33, "learning_rate": 1.5639838142891513e-05, "loss": 1.5116, "step": 3748 }, { "epoch": 0.33, "learning_rate": 1.5637476037936978e-05, "loss": 1.4866, "step": 3749 }, { "epoch": 0.33, "learning_rate": 1.5635113471809936e-05, "loss": 1.5639, "step": 3750 }, { "epoch": 0.33, "learning_rate": 1.5632750444703665e-05, "loss": 1.4541, "step": 3751 }, { "epoch": 0.33, "learning_rate": 1.5630386956811474e-05, "loss": 1.5485, "step": 3752 }, { "epoch": 0.33, "learning_rate": 1.56280230083267e-05, "loss": 1.5371, "step": 3753 }, { "epoch": 0.33, "learning_rate": 1.562565859944273e-05, "loss": 1.4523, "step": 3754 }, { "epoch": 0.33, "learning_rate": 1.562329373035298e-05, "loss": 1.5051, "step": 3755 }, { "epoch": 0.33, "learning_rate": 1.562092840125091e-05, "loss": 1.5047, "step": 3756 }, { "epoch": 0.33, "learning_rate": 1.5618562612330017e-05, "loss": 1.5709, "step": 3757 }, { "epoch": 0.33, "learning_rate": 1.561619636378383e-05, "loss": 1.5616, "step": 3758 }, { "epoch": 0.33, "learning_rate": 1.561382965580592e-05, "loss": 1.4891, "step": 3759 }, { "epoch": 0.33, "learning_rate": 1.56114624885899e-05, "loss": 1.5939, "step": 3760 }, { "epoch": 0.33, "learning_rate": 1.5609094862329406e-05, "loss": 1.4693, "step": 3761 }, { "epoch": 0.33, "learning_rate": 1.5606726777218132e-05, "loss": 1.4242, "step": 3762 }, { "epoch": 0.33, "learning_rate": 1.5604358233449796e-05, "loss": 1.4403, "step": 3763 }, { "epoch": 0.33, "learning_rate": 1.560198923121815e-05, "loss": 1.4282, "step": 3764 }, { "epoch": 0.33, "learning_rate": 1.5599619770716997e-05, "loss": 1.6041, "step": 3765 }, { "epoch": 0.33, "learning_rate": 1.559724985214016e-05, "loss": 1.6434, "step": 3766 }, { "epoch": 0.33, "learning_rate": 1.5594879475681524e-05, "loss": 1.557, "step": 3767 }, { "epoch": 0.33, "learning_rate": 1.5592508641534984e-05, "loss": 1.4819, "step": 3768 }, { "epoch": 0.33, "learning_rate": 1.5590137349894497e-05, "loss": 1.4216, "step": 3769 }, { "epoch": 0.33, "learning_rate": 1.5587765600954034e-05, "loss": 1.346, "step": 3770 }, { "epoch": 0.33, "learning_rate": 1.558539339490763e-05, "loss": 1.4596, "step": 3771 }, { "epoch": 0.33, "learning_rate": 1.5583020731949326e-05, "loss": 1.3961, "step": 3772 }, { "epoch": 0.33, "learning_rate": 1.558064761227323e-05, "loss": 1.4867, "step": 3773 }, { "epoch": 0.33, "learning_rate": 1.5578274036073468e-05, "loss": 1.5251, "step": 3774 }, { "epoch": 0.33, "learning_rate": 1.5575900003544214e-05, "loss": 1.3093, "step": 3775 }, { "epoch": 0.33, "learning_rate": 1.5573525514879674e-05, "loss": 1.289, "step": 3776 }, { "epoch": 0.33, "learning_rate": 1.557115057027409e-05, "loss": 1.4848, "step": 3777 }, { "epoch": 0.33, "learning_rate": 1.5568775169921745e-05, "loss": 1.5052, "step": 3778 }, { "epoch": 0.33, "learning_rate": 1.556639931401696e-05, "loss": 1.4822, "step": 3779 }, { "epoch": 0.33, "learning_rate": 1.5564023002754095e-05, "loss": 1.4548, "step": 3780 }, { "epoch": 0.33, "learning_rate": 1.556164623632753e-05, "loss": 1.5568, "step": 3781 }, { "epoch": 0.33, "learning_rate": 1.555926901493171e-05, "loss": 1.3775, "step": 3782 }, { "epoch": 0.33, "learning_rate": 1.5556891338761095e-05, "loss": 1.4056, "step": 3783 }, { "epoch": 0.33, "learning_rate": 1.5554513208010198e-05, "loss": 1.4355, "step": 3784 }, { "epoch": 0.33, "learning_rate": 1.5552134622873548e-05, "loss": 1.5292, "step": 3785 }, { "epoch": 0.33, "learning_rate": 1.554975558354574e-05, "loss": 1.515, "step": 3786 }, { "epoch": 0.33, "learning_rate": 1.5547376090221383e-05, "loss": 1.4216, "step": 3787 }, { "epoch": 0.33, "learning_rate": 1.554499614309513e-05, "loss": 1.4671, "step": 3788 }, { "epoch": 0.33, "learning_rate": 1.5542615742361674e-05, "loss": 1.5804, "step": 3789 }, { "epoch": 0.33, "learning_rate": 1.5540234888215744e-05, "loss": 1.5313, "step": 3790 }, { "epoch": 0.33, "learning_rate": 1.5537853580852105e-05, "loss": 1.5607, "step": 3791 }, { "epoch": 0.33, "learning_rate": 1.5535471820465557e-05, "loss": 1.4613, "step": 3792 }, { "epoch": 0.33, "learning_rate": 1.5533089607250942e-05, "loss": 1.3768, "step": 3793 }, { "epoch": 0.34, "learning_rate": 1.5530706941403134e-05, "loss": 1.549, "step": 3794 }, { "epoch": 0.34, "learning_rate": 1.552832382311705e-05, "loss": 1.5935, "step": 3795 }, { "epoch": 0.34, "learning_rate": 1.5525940252587636e-05, "loss": 1.3984, "step": 3796 }, { "epoch": 0.34, "learning_rate": 1.5523556230009884e-05, "loss": 1.4606, "step": 3797 }, { "epoch": 0.34, "learning_rate": 1.5521171755578814e-05, "loss": 1.4795, "step": 3798 }, { "epoch": 0.34, "learning_rate": 1.5518786829489495e-05, "loss": 1.3579, "step": 3799 }, { "epoch": 0.34, "learning_rate": 1.5516401451937013e-05, "loss": 1.5337, "step": 3800 }, { "epoch": 0.34, "learning_rate": 1.5514015623116513e-05, "loss": 1.5234, "step": 3801 }, { "epoch": 0.34, "learning_rate": 1.5511629343223166e-05, "loss": 1.438, "step": 3802 }, { "epoch": 0.34, "learning_rate": 1.550924261245218e-05, "loss": 1.5257, "step": 3803 }, { "epoch": 0.34, "learning_rate": 1.55068554309988e-05, "loss": 1.4662, "step": 3804 }, { "epoch": 0.34, "learning_rate": 1.5504467799058306e-05, "loss": 1.4741, "step": 3805 }, { "epoch": 0.34, "learning_rate": 1.5502079716826027e-05, "loss": 1.5712, "step": 3806 }, { "epoch": 0.34, "learning_rate": 1.549969118449731e-05, "loss": 1.4468, "step": 3807 }, { "epoch": 0.34, "learning_rate": 1.5497302202267555e-05, "loss": 1.5615, "step": 3808 }, { "epoch": 0.34, "learning_rate": 1.5494912770332185e-05, "loss": 1.4389, "step": 3809 }, { "epoch": 0.34, "learning_rate": 1.5492522888886673e-05, "loss": 1.4643, "step": 3810 }, { "epoch": 0.34, "learning_rate": 1.5490132558126526e-05, "loss": 1.4737, "step": 3811 }, { "epoch": 0.34, "learning_rate": 1.5487741778247275e-05, "loss": 1.4243, "step": 3812 }, { "epoch": 0.34, "learning_rate": 1.5485350549444502e-05, "loss": 1.5245, "step": 3813 }, { "epoch": 0.34, "learning_rate": 1.548295887191382e-05, "loss": 1.4885, "step": 3814 }, { "epoch": 0.34, "learning_rate": 1.5480566745850883e-05, "loss": 1.6642, "step": 3815 }, { "epoch": 0.34, "learning_rate": 1.5478174171451374e-05, "loss": 1.435, "step": 3816 }, { "epoch": 0.34, "learning_rate": 1.5475781148911018e-05, "loss": 1.3983, "step": 3817 }, { "epoch": 0.34, "learning_rate": 1.5473387678425575e-05, "loss": 1.4326, "step": 3818 }, { "epoch": 0.34, "learning_rate": 1.5470993760190846e-05, "loss": 1.4793, "step": 3819 }, { "epoch": 0.34, "learning_rate": 1.5468599394402664e-05, "loss": 1.4487, "step": 3820 }, { "epoch": 0.34, "learning_rate": 1.5466204581256897e-05, "loss": 1.5305, "step": 3821 }, { "epoch": 0.34, "learning_rate": 1.5463809320949454e-05, "loss": 1.4536, "step": 3822 }, { "epoch": 0.34, "learning_rate": 1.5461413613676278e-05, "loss": 1.5374, "step": 3823 }, { "epoch": 0.34, "learning_rate": 1.545901745963335e-05, "loss": 1.5112, "step": 3824 }, { "epoch": 0.34, "learning_rate": 1.5456620859016683e-05, "loss": 1.5285, "step": 3825 }, { "epoch": 0.34, "learning_rate": 1.5454223812022336e-05, "loss": 1.5097, "step": 3826 }, { "epoch": 0.34, "learning_rate": 1.5451826318846398e-05, "loss": 1.5286, "step": 3827 }, { "epoch": 0.34, "learning_rate": 1.5449428379684993e-05, "loss": 1.5028, "step": 3828 }, { "epoch": 0.34, "learning_rate": 1.5447029994734286e-05, "loss": 1.4182, "step": 3829 }, { "epoch": 0.34, "learning_rate": 1.5444631164190472e-05, "loss": 1.3776, "step": 3830 }, { "epoch": 0.34, "learning_rate": 1.5442231888249795e-05, "loss": 1.3692, "step": 3831 }, { "epoch": 0.34, "learning_rate": 1.543983216710852e-05, "loss": 1.5969, "step": 3832 }, { "epoch": 0.34, "learning_rate": 1.5437432000962958e-05, "loss": 1.3813, "step": 3833 }, { "epoch": 0.34, "learning_rate": 1.5435031390009455e-05, "loss": 1.4746, "step": 3834 }, { "epoch": 0.34, "learning_rate": 1.543263033444439e-05, "loss": 1.5149, "step": 3835 }, { "epoch": 0.34, "learning_rate": 1.5430228834464185e-05, "loss": 1.5585, "step": 3836 }, { "epoch": 0.34, "learning_rate": 1.5427826890265288e-05, "loss": 1.5407, "step": 3837 }, { "epoch": 0.34, "learning_rate": 1.5425424502044196e-05, "loss": 1.4422, "step": 3838 }, { "epoch": 0.34, "learning_rate": 1.542302166999743e-05, "loss": 1.5661, "step": 3839 }, { "epoch": 0.34, "learning_rate": 1.5420618394321557e-05, "loss": 1.4338, "step": 3840 }, { "epoch": 0.34, "learning_rate": 1.541821467521318e-05, "loss": 1.5887, "step": 3841 }, { "epoch": 0.34, "learning_rate": 1.5415810512868925e-05, "loss": 1.4423, "step": 3842 }, { "epoch": 0.34, "learning_rate": 1.541340590748547e-05, "loss": 1.47, "step": 3843 }, { "epoch": 0.34, "learning_rate": 1.5411000859259524e-05, "loss": 1.5732, "step": 3844 }, { "epoch": 0.34, "learning_rate": 1.5408595368387832e-05, "loss": 1.5551, "step": 3845 }, { "epoch": 0.34, "learning_rate": 1.5406189435067165e-05, "loss": 1.4113, "step": 3846 }, { "epoch": 0.34, "learning_rate": 1.5403783059494358e-05, "loss": 1.4412, "step": 3847 }, { "epoch": 0.34, "learning_rate": 1.5401376241866245e-05, "loss": 1.5074, "step": 3848 }, { "epoch": 0.34, "learning_rate": 1.5398968982379727e-05, "loss": 1.6174, "step": 3849 }, { "epoch": 0.34, "learning_rate": 1.5396561281231724e-05, "loss": 1.4619, "step": 3850 }, { "epoch": 0.34, "learning_rate": 1.5394153138619202e-05, "loss": 1.4878, "step": 3851 }, { "epoch": 0.34, "learning_rate": 1.5391744554739157e-05, "loss": 1.4933, "step": 3852 }, { "epoch": 0.34, "learning_rate": 1.538933552978862e-05, "loss": 1.5016, "step": 3853 }, { "epoch": 0.34, "learning_rate": 1.538692606396466e-05, "loss": 1.424, "step": 3854 }, { "epoch": 0.34, "learning_rate": 1.5384516157464392e-05, "loss": 1.5279, "step": 3855 }, { "epoch": 0.34, "learning_rate": 1.5382105810484943e-05, "loss": 1.5011, "step": 3856 }, { "epoch": 0.34, "learning_rate": 1.5379695023223503e-05, "loss": 1.4472, "step": 3857 }, { "epoch": 0.34, "learning_rate": 1.5377283795877286e-05, "loss": 1.5544, "step": 3858 }, { "epoch": 0.34, "learning_rate": 1.537487212864353e-05, "loss": 1.4192, "step": 3859 }, { "epoch": 0.34, "learning_rate": 1.5372460021719534e-05, "loss": 1.466, "step": 3860 }, { "epoch": 0.34, "learning_rate": 1.5370047475302613e-05, "loss": 1.5803, "step": 3861 }, { "epoch": 0.34, "learning_rate": 1.5367634489590127e-05, "loss": 1.426, "step": 3862 }, { "epoch": 0.34, "learning_rate": 1.5365221064779468e-05, "loss": 1.4314, "step": 3863 }, { "epoch": 0.34, "learning_rate": 1.536280720106807e-05, "loss": 1.6033, "step": 3864 }, { "epoch": 0.34, "learning_rate": 1.5360392898653396e-05, "loss": 1.4212, "step": 3865 }, { "epoch": 0.34, "learning_rate": 1.5357978157732945e-05, "loss": 1.5411, "step": 3866 }, { "epoch": 0.34, "learning_rate": 1.5355562978504253e-05, "loss": 1.4915, "step": 3867 }, { "epoch": 0.34, "learning_rate": 1.5353147361164904e-05, "loss": 1.5453, "step": 3868 }, { "epoch": 0.34, "learning_rate": 1.5350731305912495e-05, "loss": 1.4989, "step": 3869 }, { "epoch": 0.34, "learning_rate": 1.534831481294468e-05, "loss": 1.4761, "step": 3870 }, { "epoch": 0.34, "learning_rate": 1.534589788245913e-05, "loss": 1.5329, "step": 3871 }, { "epoch": 0.34, "learning_rate": 1.5343480514653568e-05, "loss": 1.5884, "step": 3872 }, { "epoch": 0.34, "learning_rate": 1.5341062709725746e-05, "loss": 1.528, "step": 3873 }, { "epoch": 0.34, "learning_rate": 1.5338644467873454e-05, "loss": 1.4956, "step": 3874 }, { "epoch": 0.34, "learning_rate": 1.533622578929451e-05, "loss": 1.5479, "step": 3875 }, { "epoch": 0.34, "learning_rate": 1.5333806674186776e-05, "loss": 1.5413, "step": 3876 }, { "epoch": 0.34, "learning_rate": 1.5331387122748147e-05, "loss": 1.4761, "step": 3877 }, { "epoch": 0.34, "learning_rate": 1.5328967135176553e-05, "loss": 1.4457, "step": 3878 }, { "epoch": 0.34, "learning_rate": 1.5326546711669966e-05, "loss": 1.4845, "step": 3879 }, { "epoch": 0.34, "learning_rate": 1.532412585242638e-05, "loss": 1.4615, "step": 3880 }, { "epoch": 0.34, "learning_rate": 1.532170455764384e-05, "loss": 1.5177, "step": 3881 }, { "epoch": 0.34, "learning_rate": 1.5319282827520417e-05, "loss": 1.4837, "step": 3882 }, { "epoch": 0.34, "learning_rate": 1.5316860662254217e-05, "loss": 1.407, "step": 3883 }, { "epoch": 0.34, "learning_rate": 1.5314438062043386e-05, "loss": 1.5288, "step": 3884 }, { "epoch": 0.34, "learning_rate": 1.5312015027086107e-05, "loss": 1.5396, "step": 3885 }, { "epoch": 0.34, "learning_rate": 1.5309591557580595e-05, "loss": 1.5904, "step": 3886 }, { "epoch": 0.34, "learning_rate": 1.5307167653725105e-05, "loss": 1.503, "step": 3887 }, { "epoch": 0.34, "learning_rate": 1.5304743315717915e-05, "loss": 1.4244, "step": 3888 }, { "epoch": 0.34, "learning_rate": 1.530231854375735e-05, "loss": 1.3589, "step": 3889 }, { "epoch": 0.34, "learning_rate": 1.5299893338041772e-05, "loss": 1.5213, "step": 3890 }, { "epoch": 0.34, "learning_rate": 1.5297467698769578e-05, "loss": 1.6227, "step": 3891 }, { "epoch": 0.34, "learning_rate": 1.5295041626139183e-05, "loss": 1.4385, "step": 3892 }, { "epoch": 0.34, "learning_rate": 1.5292615120349065e-05, "loss": 1.5365, "step": 3893 }, { "epoch": 0.34, "learning_rate": 1.529018818159772e-05, "loss": 1.5001, "step": 3894 }, { "epoch": 0.34, "learning_rate": 1.5287760810083676e-05, "loss": 1.3658, "step": 3895 }, { "epoch": 0.34, "learning_rate": 1.5285333006005512e-05, "loss": 1.5587, "step": 3896 }, { "epoch": 0.34, "learning_rate": 1.5282904769561833e-05, "loss": 1.4796, "step": 3897 }, { "epoch": 0.34, "learning_rate": 1.5280476100951272e-05, "loss": 1.4759, "step": 3898 }, { "epoch": 0.34, "learning_rate": 1.5278047000372522e-05, "loss": 1.5066, "step": 3899 }, { "epoch": 0.34, "learning_rate": 1.527561746802428e-05, "loss": 1.4955, "step": 3900 }, { "epoch": 0.34, "learning_rate": 1.52731875041053e-05, "loss": 1.5375, "step": 3901 }, { "epoch": 0.34, "learning_rate": 1.5270757108814364e-05, "loss": 1.419, "step": 3902 }, { "epoch": 0.34, "learning_rate": 1.5268326282350286e-05, "loss": 1.4458, "step": 3903 }, { "epoch": 0.34, "learning_rate": 1.526589502491193e-05, "loss": 1.4732, "step": 3904 }, { "epoch": 0.34, "learning_rate": 1.526346333669817e-05, "loss": 1.5188, "step": 3905 }, { "epoch": 0.34, "learning_rate": 1.526103121790794e-05, "loss": 1.5834, "step": 3906 }, { "epoch": 0.34, "learning_rate": 1.5258598668740194e-05, "loss": 1.3921, "step": 3907 }, { "epoch": 0.35, "learning_rate": 1.525616568939393e-05, "loss": 1.3157, "step": 3908 }, { "epoch": 0.35, "learning_rate": 1.5253732280068175e-05, "loss": 1.4815, "step": 3909 }, { "epoch": 0.35, "learning_rate": 1.525129844096199e-05, "loss": 1.503, "step": 3910 }, { "epoch": 0.35, "learning_rate": 1.5248864172274483e-05, "loss": 1.4844, "step": 3911 }, { "epoch": 0.35, "learning_rate": 1.5246429474204785e-05, "loss": 1.4782, "step": 3912 }, { "epoch": 0.35, "learning_rate": 1.5243994346952063e-05, "loss": 1.465, "step": 3913 }, { "epoch": 0.35, "learning_rate": 1.5241558790715526e-05, "loss": 1.4229, "step": 3914 }, { "epoch": 0.35, "learning_rate": 1.5239122805694409e-05, "loss": 1.4928, "step": 3915 }, { "epoch": 0.35, "learning_rate": 1.5236686392087993e-05, "loss": 1.4843, "step": 3916 }, { "epoch": 0.35, "learning_rate": 1.5234249550095586e-05, "loss": 1.5549, "step": 3917 }, { "epoch": 0.35, "learning_rate": 1.5231812279916533e-05, "loss": 1.5093, "step": 3918 }, { "epoch": 0.35, "learning_rate": 1.5229374581750213e-05, "loss": 1.4427, "step": 3919 }, { "epoch": 0.35, "learning_rate": 1.5226936455796049e-05, "loss": 1.5083, "step": 3920 }, { "epoch": 0.35, "learning_rate": 1.522449790225348e-05, "loss": 1.5469, "step": 3921 }, { "epoch": 0.35, "learning_rate": 1.5222058921321999e-05, "loss": 1.5012, "step": 3922 }, { "epoch": 0.35, "learning_rate": 1.5219619513201123e-05, "loss": 1.5284, "step": 3923 }, { "epoch": 0.35, "learning_rate": 1.5217179678090406e-05, "loss": 1.531, "step": 3924 }, { "epoch": 0.35, "learning_rate": 1.5214739416189446e-05, "loss": 1.4729, "step": 3925 }, { "epoch": 0.35, "learning_rate": 1.5212298727697858e-05, "loss": 1.3021, "step": 3926 }, { "epoch": 0.35, "learning_rate": 1.5209857612815309e-05, "loss": 1.4999, "step": 3927 }, { "epoch": 0.35, "learning_rate": 1.520741607174149e-05, "loss": 1.6263, "step": 3928 }, { "epoch": 0.35, "learning_rate": 1.5204974104676132e-05, "loss": 1.5105, "step": 3929 }, { "epoch": 0.35, "learning_rate": 1.5202531711819001e-05, "loss": 1.4818, "step": 3930 }, { "epoch": 0.35, "learning_rate": 1.5200088893369892e-05, "loss": 1.3319, "step": 3931 }, { "epoch": 0.35, "learning_rate": 1.5197645649528645e-05, "loss": 1.5153, "step": 3932 }, { "epoch": 0.35, "learning_rate": 1.5195201980495129e-05, "loss": 1.5137, "step": 3933 }, { "epoch": 0.35, "learning_rate": 1.5192757886469243e-05, "loss": 1.4878, "step": 3934 }, { "epoch": 0.35, "learning_rate": 1.5190313367650927e-05, "loss": 1.6088, "step": 3935 }, { "epoch": 0.35, "learning_rate": 1.5187868424240155e-05, "loss": 1.4338, "step": 3936 }, { "epoch": 0.35, "learning_rate": 1.5185423056436936e-05, "loss": 1.4894, "step": 3937 }, { "epoch": 0.35, "learning_rate": 1.5182977264441312e-05, "loss": 1.4676, "step": 3938 }, { "epoch": 0.35, "learning_rate": 1.5180531048453362e-05, "loss": 1.5627, "step": 3939 }, { "epoch": 0.35, "learning_rate": 1.5178084408673193e-05, "loss": 1.5687, "step": 3940 }, { "epoch": 0.35, "learning_rate": 1.5175637345300957e-05, "loss": 1.4157, "step": 3941 }, { "epoch": 0.35, "learning_rate": 1.5173189858536836e-05, "loss": 1.5689, "step": 3942 }, { "epoch": 0.35, "learning_rate": 1.5170741948581045e-05, "loss": 1.653, "step": 3943 }, { "epoch": 0.35, "learning_rate": 1.5168293615633828e-05, "loss": 1.4693, "step": 3944 }, { "epoch": 0.35, "learning_rate": 1.5165844859895481e-05, "loss": 1.4026, "step": 3945 }, { "epoch": 0.35, "learning_rate": 1.5163395681566319e-05, "loss": 1.4009, "step": 3946 }, { "epoch": 0.35, "learning_rate": 1.5160946080846694e-05, "loss": 1.4536, "step": 3947 }, { "epoch": 0.35, "learning_rate": 1.5158496057937e-05, "loss": 1.6554, "step": 3948 }, { "epoch": 0.35, "learning_rate": 1.5156045613037656e-05, "loss": 1.3965, "step": 3949 }, { "epoch": 0.35, "learning_rate": 1.5153594746349124e-05, "loss": 1.4333, "step": 3950 }, { "epoch": 0.35, "learning_rate": 1.5151143458071898e-05, "loss": 1.4376, "step": 3951 }, { "epoch": 0.35, "learning_rate": 1.5148691748406496e-05, "loss": 1.6134, "step": 3952 }, { "epoch": 0.35, "learning_rate": 1.514623961755349e-05, "loss": 1.4705, "step": 3953 }, { "epoch": 0.35, "learning_rate": 1.5143787065713468e-05, "loss": 1.5157, "step": 3954 }, { "epoch": 0.35, "learning_rate": 1.5141334093087067e-05, "loss": 1.3981, "step": 3955 }, { "epoch": 0.35, "learning_rate": 1.5138880699874949e-05, "loss": 1.523, "step": 3956 }, { "epoch": 0.35, "learning_rate": 1.5136426886277812e-05, "loss": 1.5204, "step": 3957 }, { "epoch": 0.35, "learning_rate": 1.513397265249639e-05, "loss": 1.5366, "step": 3958 }, { "epoch": 0.35, "learning_rate": 1.5131517998731458e-05, "loss": 1.3853, "step": 3959 }, { "epoch": 0.35, "learning_rate": 1.5129062925183808e-05, "loss": 1.399, "step": 3960 }, { "epoch": 0.35, "learning_rate": 1.512660743205428e-05, "loss": 1.4654, "step": 3961 }, { "epoch": 0.35, "learning_rate": 1.5124151519543748e-05, "loss": 1.4725, "step": 3962 }, { "epoch": 0.35, "learning_rate": 1.5121695187853118e-05, "loss": 1.582, "step": 3963 }, { "epoch": 0.35, "learning_rate": 1.5119238437183326e-05, "loss": 1.5593, "step": 3964 }, { "epoch": 0.35, "learning_rate": 1.5116781267735348e-05, "loss": 1.4819, "step": 3965 }, { "epoch": 0.35, "learning_rate": 1.5114323679710187e-05, "loss": 1.4974, "step": 3966 }, { "epoch": 0.35, "learning_rate": 1.5111865673308895e-05, "loss": 1.443, "step": 3967 }, { "epoch": 0.35, "learning_rate": 1.5109407248732542e-05, "loss": 1.4578, "step": 3968 }, { "epoch": 0.35, "learning_rate": 1.510694840618224e-05, "loss": 1.4928, "step": 3969 }, { "epoch": 0.35, "learning_rate": 1.5104489145859135e-05, "loss": 1.4329, "step": 3970 }, { "epoch": 0.35, "learning_rate": 1.5102029467964407e-05, "loss": 1.5125, "step": 3971 }, { "epoch": 0.35, "learning_rate": 1.509956937269927e-05, "loss": 1.4769, "step": 3972 }, { "epoch": 0.35, "learning_rate": 1.5097108860264966e-05, "loss": 1.3867, "step": 3973 }, { "epoch": 0.35, "learning_rate": 1.5094647930862783e-05, "loss": 1.5714, "step": 3974 }, { "epoch": 0.35, "learning_rate": 1.5092186584694032e-05, "loss": 1.5098, "step": 3975 }, { "epoch": 0.35, "learning_rate": 1.508972482196007e-05, "loss": 1.3548, "step": 3976 }, { "epoch": 0.35, "learning_rate": 1.5087262642862272e-05, "loss": 1.354, "step": 3977 }, { "epoch": 0.35, "learning_rate": 1.5084800047602059e-05, "loss": 1.5055, "step": 3978 }, { "epoch": 0.35, "learning_rate": 1.508233703638089e-05, "loss": 1.5813, "step": 3979 }, { "epoch": 0.35, "learning_rate": 1.5079873609400241e-05, "loss": 1.5653, "step": 3980 }, { "epoch": 0.35, "learning_rate": 1.507740976686164e-05, "loss": 1.4679, "step": 3981 }, { "epoch": 0.35, "learning_rate": 1.5074945508966635e-05, "loss": 1.5865, "step": 3982 }, { "epoch": 0.35, "learning_rate": 1.5072480835916818e-05, "loss": 1.494, "step": 3983 }, { "epoch": 0.35, "learning_rate": 1.5070015747913806e-05, "loss": 1.3698, "step": 3984 }, { "epoch": 0.35, "learning_rate": 1.5067550245159264e-05, "loss": 1.5707, "step": 3985 }, { "epoch": 0.35, "learning_rate": 1.5065084327854875e-05, "loss": 1.3162, "step": 3986 }, { "epoch": 0.35, "learning_rate": 1.5062617996202365e-05, "loss": 1.4971, "step": 3987 }, { "epoch": 0.35, "learning_rate": 1.506015125040349e-05, "loss": 1.526, "step": 3988 }, { "epoch": 0.35, "learning_rate": 1.5057684090660045e-05, "loss": 1.6101, "step": 3989 }, { "epoch": 0.35, "learning_rate": 1.505521651717385e-05, "loss": 1.5451, "step": 3990 }, { "epoch": 0.35, "learning_rate": 1.5052748530146768e-05, "loss": 1.6051, "step": 3991 }, { "epoch": 0.35, "learning_rate": 1.5050280129780692e-05, "loss": 1.5167, "step": 3992 }, { "epoch": 0.35, "learning_rate": 1.5047811316277549e-05, "loss": 1.4888, "step": 3993 }, { "epoch": 0.35, "learning_rate": 1.5045342089839304e-05, "loss": 1.4637, "step": 3994 }, { "epoch": 0.35, "learning_rate": 1.504287245066794e-05, "loss": 1.3818, "step": 3995 }, { "epoch": 0.35, "learning_rate": 1.5040402398965496e-05, "loss": 1.4733, "step": 3996 }, { "epoch": 0.35, "learning_rate": 1.5037931934934032e-05, "loss": 1.4453, "step": 3997 }, { "epoch": 0.35, "learning_rate": 1.503546105877564e-05, "loss": 1.4594, "step": 3998 }, { "epoch": 0.35, "learning_rate": 1.5032989770692448e-05, "loss": 1.4417, "step": 3999 }, { "epoch": 0.35, "learning_rate": 1.5030518070886628e-05, "loss": 1.3792, "step": 4000 }, { "epoch": 0.35, "learning_rate": 1.5028045959560369e-05, "loss": 1.4446, "step": 4001 }, { "epoch": 0.35, "learning_rate": 1.5025573436915908e-05, "loss": 1.3719, "step": 4002 }, { "epoch": 0.35, "learning_rate": 1.5023100503155501e-05, "loss": 1.501, "step": 4003 }, { "epoch": 0.35, "learning_rate": 1.502062715848145e-05, "loss": 1.5533, "step": 4004 }, { "epoch": 0.35, "learning_rate": 1.5018153403096093e-05, "loss": 1.5144, "step": 4005 }, { "epoch": 0.35, "learning_rate": 1.5015679237201782e-05, "loss": 1.4111, "step": 4006 }, { "epoch": 0.35, "learning_rate": 1.5013204661000924e-05, "loss": 1.4069, "step": 4007 }, { "epoch": 0.35, "learning_rate": 1.5010729674695947e-05, "loss": 1.501, "step": 4008 }, { "epoch": 0.35, "learning_rate": 1.5008254278489323e-05, "loss": 1.4073, "step": 4009 }, { "epoch": 0.35, "learning_rate": 1.5005778472583545e-05, "loss": 1.3865, "step": 4010 }, { "epoch": 0.35, "learning_rate": 1.500330225718115e-05, "loss": 1.598, "step": 4011 }, { "epoch": 0.35, "learning_rate": 1.5000825632484701e-05, "loss": 1.4844, "step": 4012 }, { "epoch": 0.35, "learning_rate": 1.4998348598696802e-05, "loss": 1.4691, "step": 4013 }, { "epoch": 0.35, "learning_rate": 1.4995871156020078e-05, "loss": 1.5435, "step": 4014 }, { "epoch": 0.35, "learning_rate": 1.4993393304657206e-05, "loss": 1.4296, "step": 4015 }, { "epoch": 0.35, "learning_rate": 1.4990915044810877e-05, "loss": 1.4639, "step": 4016 }, { "epoch": 0.35, "learning_rate": 1.4988436376683831e-05, "loss": 1.5517, "step": 4017 }, { "epoch": 0.35, "learning_rate": 1.4985957300478834e-05, "loss": 1.4622, "step": 4018 }, { "epoch": 0.35, "learning_rate": 1.4983477816398681e-05, "loss": 1.5986, "step": 4019 }, { "epoch": 0.35, "learning_rate": 1.4980997924646211e-05, "loss": 1.539, "step": 4020 }, { "epoch": 0.36, "learning_rate": 1.4978517625424289e-05, "loss": 1.4315, "step": 4021 }, { "epoch": 0.36, "learning_rate": 1.497603691893582e-05, "loss": 1.4421, "step": 4022 }, { "epoch": 0.36, "learning_rate": 1.4973555805383728e-05, "loss": 1.4639, "step": 4023 }, { "epoch": 0.36, "learning_rate": 1.4971074284970988e-05, "loss": 1.4524, "step": 4024 }, { "epoch": 0.36, "learning_rate": 1.4968592357900596e-05, "loss": 1.4676, "step": 4025 }, { "epoch": 0.36, "learning_rate": 1.4966110024375588e-05, "loss": 1.5619, "step": 4026 }, { "epoch": 0.36, "learning_rate": 1.4963627284599029e-05, "loss": 1.5786, "step": 4027 }, { "epoch": 0.36, "learning_rate": 1.4961144138774023e-05, "loss": 1.5099, "step": 4028 }, { "epoch": 0.36, "learning_rate": 1.4958660587103695e-05, "loss": 1.4631, "step": 4029 }, { "epoch": 0.36, "learning_rate": 1.4956176629791221e-05, "loss": 1.4283, "step": 4030 }, { "epoch": 0.36, "learning_rate": 1.4953692267039795e-05, "loss": 1.4412, "step": 4031 }, { "epoch": 0.36, "learning_rate": 1.4951207499052648e-05, "loss": 1.4354, "step": 4032 }, { "epoch": 0.36, "learning_rate": 1.494872232603305e-05, "loss": 1.3086, "step": 4033 }, { "epoch": 0.36, "learning_rate": 1.4946236748184302e-05, "loss": 1.3132, "step": 4034 }, { "epoch": 0.36, "learning_rate": 1.494375076570973e-05, "loss": 1.5342, "step": 4035 }, { "epoch": 0.36, "learning_rate": 1.4941264378812702e-05, "loss": 1.5218, "step": 4036 }, { "epoch": 0.36, "learning_rate": 1.4938777587696618e-05, "loss": 1.4074, "step": 4037 }, { "epoch": 0.36, "learning_rate": 1.4936290392564908e-05, "loss": 1.5118, "step": 4038 }, { "epoch": 0.36, "learning_rate": 1.4933802793621035e-05, "loss": 1.5303, "step": 4039 }, { "epoch": 0.36, "learning_rate": 1.4931314791068498e-05, "loss": 1.433, "step": 4040 }, { "epoch": 0.36, "learning_rate": 1.492882638511083e-05, "loss": 1.4672, "step": 4041 }, { "epoch": 0.36, "learning_rate": 1.4926337575951588e-05, "loss": 1.4516, "step": 4042 }, { "epoch": 0.36, "learning_rate": 1.4923848363794375e-05, "loss": 1.3967, "step": 4043 }, { "epoch": 0.36, "learning_rate": 1.4921358748842819e-05, "loss": 1.4838, "step": 4044 }, { "epoch": 0.36, "learning_rate": 1.4918868731300577e-05, "loss": 1.6195, "step": 4045 }, { "epoch": 0.36, "learning_rate": 1.4916378311371351e-05, "loss": 1.4281, "step": 4046 }, { "epoch": 0.36, "learning_rate": 1.4913887489258865e-05, "loss": 1.5097, "step": 4047 }, { "epoch": 0.36, "learning_rate": 1.4911396265166886e-05, "loss": 1.4212, "step": 4048 }, { "epoch": 0.36, "learning_rate": 1.4908904639299203e-05, "loss": 1.487, "step": 4049 }, { "epoch": 0.36, "learning_rate": 1.490641261185964e-05, "loss": 1.5228, "step": 4050 }, { "epoch": 0.36, "learning_rate": 1.4903920183052063e-05, "loss": 1.5049, "step": 4051 }, { "epoch": 0.36, "learning_rate": 1.4901427353080365e-05, "loss": 1.4297, "step": 4052 }, { "epoch": 0.36, "learning_rate": 1.4898934122148465e-05, "loss": 1.4396, "step": 4053 }, { "epoch": 0.36, "learning_rate": 1.4896440490460326e-05, "loss": 1.5445, "step": 4054 }, { "epoch": 0.36, "learning_rate": 1.489394645821994e-05, "loss": 1.3596, "step": 4055 }, { "epoch": 0.36, "learning_rate": 1.4891452025631327e-05, "loss": 1.5096, "step": 4056 }, { "epoch": 0.36, "learning_rate": 1.4888957192898545e-05, "loss": 1.5301, "step": 4057 }, { "epoch": 0.36, "learning_rate": 1.4886461960225682e-05, "loss": 1.4003, "step": 4058 }, { "epoch": 0.36, "learning_rate": 1.4883966327816864e-05, "loss": 1.4341, "step": 4059 }, { "epoch": 0.36, "learning_rate": 1.4881470295876241e-05, "loss": 1.3921, "step": 4060 }, { "epoch": 0.36, "learning_rate": 1.4878973864608006e-05, "loss": 1.5069, "step": 4061 }, { "epoch": 0.36, "learning_rate": 1.4876477034216373e-05, "loss": 1.6012, "step": 4062 }, { "epoch": 0.36, "learning_rate": 1.4873979804905595e-05, "loss": 1.5333, "step": 4063 }, { "epoch": 0.36, "learning_rate": 1.4871482176879963e-05, "loss": 1.4699, "step": 4064 }, { "epoch": 0.36, "learning_rate": 1.4868984150343792e-05, "loss": 1.5405, "step": 4065 }, { "epoch": 0.36, "learning_rate": 1.4866485725501429e-05, "loss": 1.5262, "step": 4066 }, { "epoch": 0.36, "learning_rate": 1.4863986902557262e-05, "loss": 1.49, "step": 4067 }, { "epoch": 0.36, "learning_rate": 1.4861487681715703e-05, "loss": 1.5274, "step": 4068 }, { "epoch": 0.36, "learning_rate": 1.4858988063181204e-05, "loss": 1.41, "step": 4069 }, { "epoch": 0.36, "learning_rate": 1.4856488047158242e-05, "loss": 1.4088, "step": 4070 }, { "epoch": 0.36, "learning_rate": 1.485398763385133e-05, "loss": 1.6327, "step": 4071 }, { "epoch": 0.36, "learning_rate": 1.4851486823465018e-05, "loss": 1.3911, "step": 4072 }, { "epoch": 0.36, "learning_rate": 1.484898561620388e-05, "loss": 1.5115, "step": 4073 }, { "epoch": 0.36, "learning_rate": 1.4846484012272532e-05, "loss": 1.5795, "step": 4074 }, { "epoch": 0.36, "learning_rate": 1.4843982011875613e-05, "loss": 1.3739, "step": 4075 }, { "epoch": 0.36, "learning_rate": 1.4841479615217798e-05, "loss": 1.5687, "step": 4076 }, { "epoch": 0.36, "learning_rate": 1.4838976822503797e-05, "loss": 1.4968, "step": 4077 }, { "epoch": 0.36, "learning_rate": 1.483647363393835e-05, "loss": 1.4637, "step": 4078 }, { "epoch": 0.36, "learning_rate": 1.483397004972623e-05, "loss": 1.4919, "step": 4079 }, { "epoch": 0.36, "learning_rate": 1.4831466070072238e-05, "loss": 1.4529, "step": 4080 }, { "epoch": 0.36, "learning_rate": 1.4828961695181219e-05, "loss": 1.6855, "step": 4081 }, { "epoch": 0.36, "learning_rate": 1.482645692525804e-05, "loss": 1.4901, "step": 4082 }, { "epoch": 0.36, "learning_rate": 1.48239517605076e-05, "loss": 1.4699, "step": 4083 }, { "epoch": 0.36, "learning_rate": 1.4821446201134837e-05, "loss": 1.4991, "step": 4084 }, { "epoch": 0.36, "learning_rate": 1.4818940247344716e-05, "loss": 1.4592, "step": 4085 }, { "epoch": 0.36, "learning_rate": 1.4816433899342232e-05, "loss": 1.6096, "step": 4086 }, { "epoch": 0.36, "learning_rate": 1.4813927157332429e-05, "loss": 1.6473, "step": 4087 }, { "epoch": 0.36, "learning_rate": 1.4811420021520358e-05, "loss": 1.4871, "step": 4088 }, { "epoch": 0.36, "learning_rate": 1.4808912492111119e-05, "loss": 1.4807, "step": 4089 }, { "epoch": 0.36, "learning_rate": 1.4806404569309838e-05, "loss": 1.5948, "step": 4090 }, { "epoch": 0.36, "learning_rate": 1.480389625332168e-05, "loss": 1.4724, "step": 4091 }, { "epoch": 0.36, "learning_rate": 1.4801387544351832e-05, "loss": 1.4712, "step": 4092 }, { "epoch": 0.36, "learning_rate": 1.4798878442605523e-05, "loss": 1.437, "step": 4093 }, { "epoch": 0.36, "learning_rate": 1.4796368948288005e-05, "loss": 1.521, "step": 4094 }, { "epoch": 0.36, "learning_rate": 1.4793859061604573e-05, "loss": 1.451, "step": 4095 }, { "epoch": 0.36, "learning_rate": 1.4791348782760538e-05, "loss": 1.5408, "step": 4096 }, { "epoch": 0.36, "learning_rate": 1.478883811196126e-05, "loss": 1.4517, "step": 4097 }, { "epoch": 0.36, "learning_rate": 1.478632704941212e-05, "loss": 1.4611, "step": 4098 }, { "epoch": 0.36, "learning_rate": 1.4783815595318538e-05, "loss": 1.4671, "step": 4099 }, { "epoch": 0.36, "learning_rate": 1.4781303749885964e-05, "loss": 1.4886, "step": 4100 }, { "epoch": 0.36, "learning_rate": 1.4778791513319876e-05, "loss": 1.4824, "step": 4101 }, { "epoch": 0.36, "learning_rate": 1.4776278885825788e-05, "loss": 1.5992, "step": 4102 }, { "epoch": 0.36, "learning_rate": 1.4773765867609244e-05, "loss": 1.4743, "step": 4103 }, { "epoch": 0.36, "learning_rate": 1.4771252458875824e-05, "loss": 1.4572, "step": 4104 }, { "epoch": 0.36, "learning_rate": 1.4768738659831133e-05, "loss": 1.4667, "step": 4105 }, { "epoch": 0.36, "learning_rate": 1.4766224470680811e-05, "loss": 1.3724, "step": 4106 }, { "epoch": 0.36, "learning_rate": 1.4763709891630538e-05, "loss": 1.4499, "step": 4107 }, { "epoch": 0.36, "learning_rate": 1.4761194922886014e-05, "loss": 1.5929, "step": 4108 }, { "epoch": 0.36, "learning_rate": 1.4758679564652972e-05, "loss": 1.4795, "step": 4109 }, { "epoch": 0.36, "learning_rate": 1.4756163817137182e-05, "loss": 1.4232, "step": 4110 }, { "epoch": 0.36, "learning_rate": 1.4753647680544448e-05, "loss": 1.4501, "step": 4111 }, { "epoch": 0.36, "learning_rate": 1.47511311550806e-05, "loss": 1.4351, "step": 4112 }, { "epoch": 0.36, "learning_rate": 1.4748614240951501e-05, "loss": 1.3964, "step": 4113 }, { "epoch": 0.36, "learning_rate": 1.4746096938363048e-05, "loss": 1.4642, "step": 4114 }, { "epoch": 0.36, "learning_rate": 1.4743579247521167e-05, "loss": 1.5075, "step": 4115 }, { "epoch": 0.36, "learning_rate": 1.4741061168631815e-05, "loss": 1.5993, "step": 4116 }, { "epoch": 0.36, "learning_rate": 1.473854270190099e-05, "loss": 1.5224, "step": 4117 }, { "epoch": 0.36, "learning_rate": 1.4736023847534708e-05, "loss": 1.4877, "step": 4118 }, { "epoch": 0.36, "learning_rate": 1.4733504605739027e-05, "loss": 1.4931, "step": 4119 }, { "epoch": 0.36, "learning_rate": 1.4730984976720029e-05, "loss": 1.4308, "step": 4120 }, { "epoch": 0.36, "learning_rate": 1.4728464960683835e-05, "loss": 1.525, "step": 4121 }, { "epoch": 0.36, "learning_rate": 1.4725944557836593e-05, "loss": 1.4337, "step": 4122 }, { "epoch": 0.36, "learning_rate": 1.4723423768384486e-05, "loss": 1.4907, "step": 4123 }, { "epoch": 0.36, "learning_rate": 1.4720902592533727e-05, "loss": 1.4353, "step": 4124 }, { "epoch": 0.36, "learning_rate": 1.4718381030490554e-05, "loss": 1.478, "step": 4125 }, { "epoch": 0.36, "learning_rate": 1.4715859082461253e-05, "loss": 1.3453, "step": 4126 }, { "epoch": 0.36, "learning_rate": 1.471333674865212e-05, "loss": 1.506, "step": 4127 }, { "epoch": 0.36, "learning_rate": 1.4710814029269503e-05, "loss": 1.4887, "step": 4128 }, { "epoch": 0.36, "learning_rate": 1.4708290924519767e-05, "loss": 1.4652, "step": 4129 }, { "epoch": 0.36, "learning_rate": 1.470576743460932e-05, "loss": 1.5034, "step": 4130 }, { "epoch": 0.36, "learning_rate": 1.4703243559744589e-05, "loss": 1.3594, "step": 4131 }, { "epoch": 0.36, "learning_rate": 1.4700719300132042e-05, "loss": 1.4035, "step": 4132 }, { "epoch": 0.36, "learning_rate": 1.4698194655978176e-05, "loss": 1.3507, "step": 4133 }, { "epoch": 0.37, "learning_rate": 1.469566962748952e-05, "loss": 1.4564, "step": 4134 }, { "epoch": 0.37, "learning_rate": 1.4693144214872628e-05, "loss": 1.4816, "step": 4135 }, { "epoch": 0.37, "learning_rate": 1.4690618418334094e-05, "loss": 1.6068, "step": 4136 }, { "epoch": 0.37, "learning_rate": 1.4688092238080545e-05, "loss": 1.502, "step": 4137 }, { "epoch": 0.37, "learning_rate": 1.4685565674318628e-05, "loss": 1.3884, "step": 4138 }, { "epoch": 0.37, "learning_rate": 1.4683038727255032e-05, "loss": 1.5266, "step": 4139 }, { "epoch": 0.37, "learning_rate": 1.4680511397096469e-05, "loss": 1.5024, "step": 4140 }, { "epoch": 0.37, "learning_rate": 1.467798368404969e-05, "loss": 1.5336, "step": 4141 }, { "epoch": 0.37, "learning_rate": 1.4675455588321476e-05, "loss": 1.4798, "step": 4142 }, { "epoch": 0.37, "learning_rate": 1.4672927110118634e-05, "loss": 1.4957, "step": 4143 }, { "epoch": 0.37, "learning_rate": 1.4670398249648007e-05, "loss": 1.3804, "step": 4144 }, { "epoch": 0.37, "learning_rate": 1.4667869007116465e-05, "loss": 1.4839, "step": 4145 }, { "epoch": 0.37, "learning_rate": 1.4665339382730914e-05, "loss": 1.4068, "step": 4146 }, { "epoch": 0.37, "learning_rate": 1.4662809376698292e-05, "loss": 1.4772, "step": 4147 }, { "epoch": 0.37, "learning_rate": 1.4660278989225565e-05, "loss": 1.5041, "step": 4148 }, { "epoch": 0.37, "learning_rate": 1.4657748220519725e-05, "loss": 1.5085, "step": 4149 }, { "epoch": 0.37, "learning_rate": 1.4655217070787807e-05, "loss": 1.4063, "step": 4150 }, { "epoch": 0.37, "learning_rate": 1.4652685540236869e-05, "loss": 1.4598, "step": 4151 }, { "epoch": 0.37, "learning_rate": 1.4650153629074005e-05, "loss": 1.4507, "step": 4152 }, { "epoch": 0.37, "learning_rate": 1.4647621337506328e-05, "loss": 1.4208, "step": 4153 }, { "epoch": 0.37, "learning_rate": 1.4645088665741005e-05, "loss": 1.5735, "step": 4154 }, { "epoch": 0.37, "learning_rate": 1.4642555613985215e-05, "loss": 1.4444, "step": 4155 }, { "epoch": 0.37, "learning_rate": 1.464002218244617e-05, "loss": 1.4655, "step": 4156 }, { "epoch": 0.37, "learning_rate": 1.4637488371331118e-05, "loss": 1.4781, "step": 4157 }, { "epoch": 0.37, "learning_rate": 1.4634954180847345e-05, "loss": 1.5472, "step": 4158 }, { "epoch": 0.37, "learning_rate": 1.4632419611202149e-05, "loss": 1.4106, "step": 4159 }, { "epoch": 0.37, "learning_rate": 1.4629884662602878e-05, "loss": 1.5605, "step": 4160 }, { "epoch": 0.37, "learning_rate": 1.4627349335256893e-05, "loss": 1.3972, "step": 4161 }, { "epoch": 0.37, "learning_rate": 1.4624813629371607e-05, "loss": 1.5065, "step": 4162 }, { "epoch": 0.37, "learning_rate": 1.4622277545154448e-05, "loss": 1.5262, "step": 4163 }, { "epoch": 0.37, "learning_rate": 1.4619741082812879e-05, "loss": 1.6259, "step": 4164 }, { "epoch": 0.37, "learning_rate": 1.4617204242554399e-05, "loss": 1.4114, "step": 4165 }, { "epoch": 0.37, "learning_rate": 1.4614667024586526e-05, "loss": 1.5364, "step": 4166 }, { "epoch": 0.37, "learning_rate": 1.4612129429116823e-05, "loss": 1.3348, "step": 4167 }, { "epoch": 0.37, "learning_rate": 1.4609591456352877e-05, "loss": 1.481, "step": 4168 }, { "epoch": 0.37, "learning_rate": 1.4607053106502305e-05, "loss": 1.5276, "step": 4169 }, { "epoch": 0.37, "learning_rate": 1.460451437977275e-05, "loss": 1.4883, "step": 4170 }, { "epoch": 0.37, "learning_rate": 1.4601975276371904e-05, "loss": 1.3489, "step": 4171 }, { "epoch": 0.37, "learning_rate": 1.4599435796507471e-05, "loss": 1.5526, "step": 4172 }, { "epoch": 0.37, "learning_rate": 1.4596895940387193e-05, "loss": 1.3957, "step": 4173 }, { "epoch": 0.37, "learning_rate": 1.4594355708218839e-05, "loss": 1.5163, "step": 4174 }, { "epoch": 0.37, "learning_rate": 1.459181510021022e-05, "loss": 1.5456, "step": 4175 }, { "epoch": 0.37, "learning_rate": 1.4589274116569164e-05, "loss": 1.4204, "step": 4176 }, { "epoch": 0.37, "learning_rate": 1.4586732757503536e-05, "loss": 1.4828, "step": 4177 }, { "epoch": 0.37, "learning_rate": 1.4584191023221234e-05, "loss": 1.4526, "step": 4178 }, { "epoch": 0.37, "learning_rate": 1.4581648913930184e-05, "loss": 1.4478, "step": 4179 }, { "epoch": 0.37, "learning_rate": 1.4579106429838342e-05, "loss": 1.5048, "step": 4180 }, { "epoch": 0.37, "learning_rate": 1.4576563571153692e-05, "loss": 1.4263, "step": 4181 }, { "epoch": 0.37, "learning_rate": 1.4574020338084259e-05, "loss": 1.5049, "step": 4182 }, { "epoch": 0.37, "learning_rate": 1.4571476730838081e-05, "loss": 1.4883, "step": 4183 }, { "epoch": 0.37, "learning_rate": 1.4568932749623248e-05, "loss": 1.4499, "step": 4184 }, { "epoch": 0.37, "learning_rate": 1.4566388394647863e-05, "loss": 1.4389, "step": 4185 }, { "epoch": 0.37, "learning_rate": 1.4563843666120073e-05, "loss": 1.4785, "step": 4186 }, { "epoch": 0.37, "learning_rate": 1.4561298564248038e-05, "loss": 1.4883, "step": 4187 }, { "epoch": 0.37, "learning_rate": 1.4558753089239974e-05, "loss": 1.4987, "step": 4188 }, { "epoch": 0.37, "learning_rate": 1.4556207241304102e-05, "loss": 1.4529, "step": 4189 }, { "epoch": 0.37, "learning_rate": 1.4553661020648689e-05, "loss": 1.5937, "step": 4190 }, { "epoch": 0.37, "learning_rate": 1.455111442748202e-05, "loss": 1.4835, "step": 4191 }, { "epoch": 0.37, "learning_rate": 1.454856746201243e-05, "loss": 1.4519, "step": 4192 }, { "epoch": 0.37, "learning_rate": 1.4546020124448272e-05, "loss": 1.461, "step": 4193 }, { "epoch": 0.37, "learning_rate": 1.4543472414997923e-05, "loss": 1.4173, "step": 4194 }, { "epoch": 0.37, "learning_rate": 1.4540924333869804e-05, "loss": 1.552, "step": 4195 }, { "epoch": 0.37, "learning_rate": 1.4538375881272355e-05, "loss": 1.4541, "step": 4196 }, { "epoch": 0.37, "learning_rate": 1.4535827057414056e-05, "loss": 1.3208, "step": 4197 }, { "epoch": 0.37, "learning_rate": 1.4533277862503412e-05, "loss": 1.4347, "step": 4198 }, { "epoch": 0.37, "learning_rate": 1.4530728296748956e-05, "loss": 1.4627, "step": 4199 }, { "epoch": 0.37, "learning_rate": 1.452817836035926e-05, "loss": 1.4351, "step": 4200 }, { "epoch": 0.37, "learning_rate": 1.4525628053542918e-05, "loss": 1.4015, "step": 4201 }, { "epoch": 0.37, "learning_rate": 1.4523077376508556e-05, "loss": 1.5038, "step": 4202 }, { "epoch": 0.37, "learning_rate": 1.4520526329464838e-05, "loss": 1.6303, "step": 4203 }, { "epoch": 0.37, "learning_rate": 1.451797491262044e-05, "loss": 1.4863, "step": 4204 }, { "epoch": 0.37, "learning_rate": 1.451542312618409e-05, "loss": 1.4418, "step": 4205 }, { "epoch": 0.37, "learning_rate": 1.4512870970364537e-05, "loss": 1.4156, "step": 4206 }, { "epoch": 0.37, "learning_rate": 1.4510318445370552e-05, "loss": 1.4792, "step": 4207 }, { "epoch": 0.37, "learning_rate": 1.4507765551410949e-05, "loss": 1.3914, "step": 4208 }, { "epoch": 0.37, "learning_rate": 1.4505212288694565e-05, "loss": 1.5115, "step": 4209 }, { "epoch": 0.37, "learning_rate": 1.4502658657430273e-05, "loss": 1.4195, "step": 4210 }, { "epoch": 0.37, "learning_rate": 1.4500104657826967e-05, "loss": 1.3665, "step": 4211 }, { "epoch": 0.37, "learning_rate": 1.4497550290093578e-05, "loss": 1.4415, "step": 4212 }, { "epoch": 0.37, "learning_rate": 1.4494995554439067e-05, "loss": 1.4711, "step": 4213 }, { "epoch": 0.37, "learning_rate": 1.4492440451072422e-05, "loss": 1.4444, "step": 4214 }, { "epoch": 0.37, "learning_rate": 1.4489884980202664e-05, "loss": 1.4718, "step": 4215 }, { "epoch": 0.37, "learning_rate": 1.4487329142038842e-05, "loss": 1.4911, "step": 4216 }, { "epoch": 0.37, "learning_rate": 1.4484772936790034e-05, "loss": 1.3908, "step": 4217 }, { "epoch": 0.37, "learning_rate": 1.4482216364665353e-05, "loss": 1.6028, "step": 4218 }, { "epoch": 0.37, "learning_rate": 1.4479659425873938e-05, "loss": 1.4591, "step": 4219 }, { "epoch": 0.37, "learning_rate": 1.4477102120624957e-05, "loss": 1.3261, "step": 4220 }, { "epoch": 0.37, "learning_rate": 1.4474544449127613e-05, "loss": 1.4484, "step": 4221 }, { "epoch": 0.37, "learning_rate": 1.447198641159113e-05, "loss": 1.438, "step": 4222 }, { "epoch": 0.37, "learning_rate": 1.4469428008224777e-05, "loss": 1.4113, "step": 4223 }, { "epoch": 0.37, "learning_rate": 1.4466869239237834e-05, "loss": 1.4976, "step": 4224 }, { "epoch": 0.37, "learning_rate": 1.4464310104839625e-05, "loss": 1.4566, "step": 4225 }, { "epoch": 0.37, "learning_rate": 1.4461750605239499e-05, "loss": 1.3551, "step": 4226 }, { "epoch": 0.37, "learning_rate": 1.4459190740646839e-05, "loss": 1.3457, "step": 4227 }, { "epoch": 0.37, "learning_rate": 1.4456630511271049e-05, "loss": 1.45, "step": 4228 }, { "epoch": 0.37, "learning_rate": 1.4454069917321567e-05, "loss": 1.4432, "step": 4229 }, { "epoch": 0.37, "learning_rate": 1.4451508959007864e-05, "loss": 1.571, "step": 4230 }, { "epoch": 0.37, "learning_rate": 1.4448947636539443e-05, "loss": 1.5783, "step": 4231 }, { "epoch": 0.37, "learning_rate": 1.4446385950125827e-05, "loss": 1.6375, "step": 4232 }, { "epoch": 0.37, "learning_rate": 1.4443823899976574e-05, "loss": 1.4704, "step": 4233 }, { "epoch": 0.37, "learning_rate": 1.4441261486301276e-05, "loss": 1.3626, "step": 4234 }, { "epoch": 0.37, "learning_rate": 1.4438698709309546e-05, "loss": 1.3968, "step": 4235 }, { "epoch": 0.37, "learning_rate": 1.4436135569211036e-05, "loss": 1.3882, "step": 4236 }, { "epoch": 0.37, "learning_rate": 1.443357206621542e-05, "loss": 1.4177, "step": 4237 }, { "epoch": 0.37, "learning_rate": 1.4431008200532406e-05, "loss": 1.4159, "step": 4238 }, { "epoch": 0.37, "learning_rate": 1.4428443972371731e-05, "loss": 1.5115, "step": 4239 }, { "epoch": 0.37, "learning_rate": 1.442587938194316e-05, "loss": 1.6206, "step": 4240 }, { "epoch": 0.37, "learning_rate": 1.4423314429456488e-05, "loss": 1.4656, "step": 4241 }, { "epoch": 0.37, "learning_rate": 1.4420749115121542e-05, "loss": 1.4596, "step": 4242 }, { "epoch": 0.37, "learning_rate": 1.4418183439148174e-05, "loss": 1.5535, "step": 4243 }, { "epoch": 0.37, "learning_rate": 1.4415617401746271e-05, "loss": 1.4353, "step": 4244 }, { "epoch": 0.37, "learning_rate": 1.4413051003125753e-05, "loss": 1.5436, "step": 4245 }, { "epoch": 0.37, "learning_rate": 1.4410484243496554e-05, "loss": 1.45, "step": 4246 }, { "epoch": 0.38, "learning_rate": 1.4407917123068649e-05, "loss": 1.4673, "step": 4247 }, { "epoch": 0.38, "learning_rate": 1.4405349642052044e-05, "loss": 1.4586, "step": 4248 }, { "epoch": 0.38, "learning_rate": 1.4402781800656774e-05, "loss": 1.4989, "step": 4249 }, { "epoch": 0.38, "learning_rate": 1.4400213599092892e-05, "loss": 1.4825, "step": 4250 }, { "epoch": 0.38, "learning_rate": 1.4397645037570494e-05, "loss": 1.6316, "step": 4251 }, { "epoch": 0.38, "learning_rate": 1.4395076116299702e-05, "loss": 1.5181, "step": 4252 }, { "epoch": 0.38, "learning_rate": 1.4392506835490664e-05, "loss": 1.5123, "step": 4253 }, { "epoch": 0.38, "learning_rate": 1.4389937195353562e-05, "loss": 1.3392, "step": 4254 }, { "epoch": 0.38, "learning_rate": 1.43873671960986e-05, "loss": 1.4631, "step": 4255 }, { "epoch": 0.38, "learning_rate": 1.4384796837936019e-05, "loss": 1.447, "step": 4256 }, { "epoch": 0.38, "learning_rate": 1.4382226121076088e-05, "loss": 1.4661, "step": 4257 }, { "epoch": 0.38, "learning_rate": 1.4379655045729104e-05, "loss": 1.4585, "step": 4258 }, { "epoch": 0.38, "learning_rate": 1.437708361210539e-05, "loss": 1.3984, "step": 4259 }, { "epoch": 0.38, "learning_rate": 1.4374511820415302e-05, "loss": 1.498, "step": 4260 }, { "epoch": 0.38, "learning_rate": 1.4371939670869227e-05, "loss": 1.3818, "step": 4261 }, { "epoch": 0.38, "learning_rate": 1.4369367163677582e-05, "loss": 1.4172, "step": 4262 }, { "epoch": 0.38, "learning_rate": 1.43667942990508e-05, "loss": 1.4335, "step": 4263 }, { "epoch": 0.38, "learning_rate": 1.4364221077199365e-05, "loss": 1.5003, "step": 4264 }, { "epoch": 0.38, "learning_rate": 1.4361647498333772e-05, "loss": 1.474, "step": 4265 }, { "epoch": 0.38, "learning_rate": 1.4359073562664558e-05, "loss": 1.4225, "step": 4266 }, { "epoch": 0.38, "learning_rate": 1.4356499270402278e-05, "loss": 1.5104, "step": 4267 }, { "epoch": 0.38, "learning_rate": 1.4353924621757522e-05, "loss": 1.5797, "step": 4268 }, { "epoch": 0.38, "learning_rate": 1.4351349616940908e-05, "loss": 1.4765, "step": 4269 }, { "epoch": 0.38, "learning_rate": 1.434877425616309e-05, "loss": 1.5474, "step": 4270 }, { "epoch": 0.38, "learning_rate": 1.4346198539634738e-05, "loss": 1.469, "step": 4271 }, { "epoch": 0.38, "learning_rate": 1.4343622467566559e-05, "loss": 1.6049, "step": 4272 }, { "epoch": 0.38, "learning_rate": 1.4341046040169287e-05, "loss": 1.4406, "step": 4273 }, { "epoch": 0.38, "learning_rate": 1.433846925765369e-05, "loss": 1.4336, "step": 4274 }, { "epoch": 0.38, "learning_rate": 1.4335892120230561e-05, "loss": 1.5058, "step": 4275 }, { "epoch": 0.38, "learning_rate": 1.433331462811072e-05, "loss": 1.4232, "step": 4276 }, { "epoch": 0.38, "learning_rate": 1.4330736781505016e-05, "loss": 1.3918, "step": 4277 }, { "epoch": 0.38, "learning_rate": 1.4328158580624334e-05, "loss": 1.4316, "step": 4278 }, { "epoch": 0.38, "learning_rate": 1.4325580025679582e-05, "loss": 1.4007, "step": 4279 }, { "epoch": 0.38, "learning_rate": 1.4323001116881695e-05, "loss": 1.4896, "step": 4280 }, { "epoch": 0.38, "learning_rate": 1.4320421854441641e-05, "loss": 1.5243, "step": 4281 }, { "epoch": 0.38, "learning_rate": 1.4317842238570418e-05, "loss": 1.5814, "step": 4282 }, { "epoch": 0.38, "learning_rate": 1.4315262269479054e-05, "loss": 1.4886, "step": 4283 }, { "epoch": 0.38, "learning_rate": 1.4312681947378595e-05, "loss": 1.3841, "step": 4284 }, { "epoch": 0.38, "learning_rate": 1.4310101272480126e-05, "loss": 1.4433, "step": 4285 }, { "epoch": 0.38, "learning_rate": 1.4307520244994764e-05, "loss": 1.5443, "step": 4286 }, { "epoch": 0.38, "learning_rate": 1.4304938865133645e-05, "loss": 1.4994, "step": 4287 }, { "epoch": 0.38, "learning_rate": 1.4302357133107937e-05, "loss": 1.3183, "step": 4288 }, { "epoch": 0.38, "learning_rate": 1.429977504912884e-05, "loss": 1.3949, "step": 4289 }, { "epoch": 0.38, "learning_rate": 1.429719261340758e-05, "loss": 1.5602, "step": 4290 }, { "epoch": 0.38, "learning_rate": 1.4294609826155415e-05, "loss": 1.5431, "step": 4291 }, { "epoch": 0.38, "learning_rate": 1.4292026687583629e-05, "loss": 1.3597, "step": 4292 }, { "epoch": 0.38, "learning_rate": 1.4289443197903529e-05, "loss": 1.3681, "step": 4293 }, { "epoch": 0.38, "learning_rate": 1.4286859357326463e-05, "loss": 1.4539, "step": 4294 }, { "epoch": 0.38, "learning_rate": 1.42842751660638e-05, "loss": 1.4785, "step": 4295 }, { "epoch": 0.38, "learning_rate": 1.4281690624326942e-05, "loss": 1.3877, "step": 4296 }, { "epoch": 0.38, "learning_rate": 1.4279105732327312e-05, "loss": 1.5603, "step": 4297 }, { "epoch": 0.38, "learning_rate": 1.4276520490276364e-05, "loss": 1.4794, "step": 4298 }, { "epoch": 0.38, "learning_rate": 1.4273934898385596e-05, "loss": 1.5461, "step": 4299 }, { "epoch": 0.38, "learning_rate": 1.4271348956866508e-05, "loss": 1.5523, "step": 4300 }, { "epoch": 0.38, "learning_rate": 1.4268762665930651e-05, "loss": 1.4547, "step": 4301 }, { "epoch": 0.38, "learning_rate": 1.4266176025789588e-05, "loss": 1.4075, "step": 4302 }, { "epoch": 0.38, "learning_rate": 1.4263589036654927e-05, "loss": 1.4776, "step": 4303 }, { "epoch": 0.38, "learning_rate": 1.4261001698738292e-05, "loss": 1.4102, "step": 4304 }, { "epoch": 0.38, "learning_rate": 1.4258414012251343e-05, "loss": 1.5063, "step": 4305 }, { "epoch": 0.38, "learning_rate": 1.4255825977405758e-05, "loss": 1.4195, "step": 4306 }, { "epoch": 0.38, "learning_rate": 1.4253237594413255e-05, "loss": 1.452, "step": 4307 }, { "epoch": 0.38, "learning_rate": 1.4250648863485576e-05, "loss": 1.4855, "step": 4308 }, { "epoch": 0.38, "learning_rate": 1.4248059784834496e-05, "loss": 1.6876, "step": 4309 }, { "epoch": 0.38, "learning_rate": 1.4245470358671804e-05, "loss": 1.4608, "step": 4310 }, { "epoch": 0.38, "learning_rate": 1.4242880585209333e-05, "loss": 1.3302, "step": 4311 }, { "epoch": 0.38, "learning_rate": 1.4240290464658943e-05, "loss": 1.4826, "step": 4312 }, { "epoch": 0.38, "learning_rate": 1.423769999723251e-05, "loss": 1.4078, "step": 4313 }, { "epoch": 0.38, "learning_rate": 1.4235109183141955e-05, "loss": 1.4196, "step": 4314 }, { "epoch": 0.38, "learning_rate": 1.4232518022599209e-05, "loss": 1.4819, "step": 4315 }, { "epoch": 0.38, "learning_rate": 1.422992651581625e-05, "loss": 1.6122, "step": 4316 }, { "epoch": 0.38, "learning_rate": 1.4227334663005075e-05, "loss": 1.5092, "step": 4317 }, { "epoch": 0.38, "learning_rate": 1.4224742464377705e-05, "loss": 1.5, "step": 4318 }, { "epoch": 0.38, "learning_rate": 1.4222149920146195e-05, "loss": 1.519, "step": 4319 }, { "epoch": 0.38, "learning_rate": 1.4219557030522633e-05, "loss": 1.4484, "step": 4320 }, { "epoch": 0.38, "learning_rate": 1.4216963795719126e-05, "loss": 1.5468, "step": 4321 }, { "epoch": 0.38, "learning_rate": 1.4214370215947813e-05, "loss": 1.4947, "step": 4322 }, { "epoch": 0.38, "learning_rate": 1.4211776291420858e-05, "loss": 1.4808, "step": 4323 }, { "epoch": 0.38, "learning_rate": 1.420918202235046e-05, "loss": 1.463, "step": 4324 }, { "epoch": 0.38, "learning_rate": 1.4206587408948847e-05, "loss": 1.4931, "step": 4325 }, { "epoch": 0.38, "learning_rate": 1.420399245142826e-05, "loss": 1.3703, "step": 4326 }, { "epoch": 0.38, "learning_rate": 1.4201397150000993e-05, "loss": 1.5003, "step": 4327 }, { "epoch": 0.38, "learning_rate": 1.4198801504879338e-05, "loss": 1.503, "step": 4328 }, { "epoch": 0.38, "learning_rate": 1.4196205516275645e-05, "loss": 1.3647, "step": 4329 }, { "epoch": 0.38, "learning_rate": 1.419360918440227e-05, "loss": 1.3849, "step": 4330 }, { "epoch": 0.38, "learning_rate": 1.4191012509471612e-05, "loss": 1.6245, "step": 4331 }, { "epoch": 0.38, "learning_rate": 1.4188415491696082e-05, "loss": 1.4384, "step": 4332 }, { "epoch": 0.38, "learning_rate": 1.4185818131288138e-05, "loss": 1.3629, "step": 4333 }, { "epoch": 0.38, "learning_rate": 1.4183220428460249e-05, "loss": 1.5075, "step": 4334 }, { "epoch": 0.38, "learning_rate": 1.4180622383424927e-05, "loss": 1.4181, "step": 4335 }, { "epoch": 0.38, "learning_rate": 1.4178023996394697e-05, "loss": 1.461, "step": 4336 }, { "epoch": 0.38, "learning_rate": 1.4175425267582121e-05, "loss": 1.447, "step": 4337 }, { "epoch": 0.38, "learning_rate": 1.4172826197199796e-05, "loss": 1.4078, "step": 4338 }, { "epoch": 0.38, "learning_rate": 1.4170226785460326e-05, "loss": 1.5235, "step": 4339 }, { "epoch": 0.38, "learning_rate": 1.4167627032576364e-05, "loss": 1.3969, "step": 4340 }, { "epoch": 0.38, "learning_rate": 1.4165026938760578e-05, "loss": 1.4209, "step": 4341 }, { "epoch": 0.38, "learning_rate": 1.4162426504225672e-05, "loss": 1.4565, "step": 4342 }, { "epoch": 0.38, "learning_rate": 1.4159825729184369e-05, "loss": 1.4228, "step": 4343 }, { "epoch": 0.38, "learning_rate": 1.4157224613849427e-05, "loss": 1.5758, "step": 4344 }, { "epoch": 0.38, "learning_rate": 1.4154623158433631e-05, "loss": 1.4732, "step": 4345 }, { "epoch": 0.38, "learning_rate": 1.4152021363149793e-05, "loss": 1.4951, "step": 4346 }, { "epoch": 0.38, "learning_rate": 1.4149419228210748e-05, "loss": 1.436, "step": 4347 }, { "epoch": 0.38, "learning_rate": 1.4146816753829367e-05, "loss": 1.402, "step": 4348 }, { "epoch": 0.38, "learning_rate": 1.4144213940218541e-05, "loss": 1.4938, "step": 4349 }, { "epoch": 0.38, "learning_rate": 1.41416107875912e-05, "loss": 1.3958, "step": 4350 }, { "epoch": 0.38, "learning_rate": 1.4139007296160285e-05, "loss": 1.3463, "step": 4351 }, { "epoch": 0.38, "learning_rate": 1.4136403466138782e-05, "loss": 1.4767, "step": 4352 }, { "epoch": 0.38, "learning_rate": 1.413379929773969e-05, "loss": 1.5236, "step": 4353 }, { "epoch": 0.38, "learning_rate": 1.4131194791176049e-05, "loss": 1.4868, "step": 4354 }, { "epoch": 0.38, "learning_rate": 1.4128589946660916e-05, "loss": 1.361, "step": 4355 }, { "epoch": 0.38, "learning_rate": 1.4125984764407381e-05, "loss": 1.4368, "step": 4356 }, { "epoch": 0.38, "learning_rate": 1.4123379244628557e-05, "loss": 1.4108, "step": 4357 }, { "epoch": 0.38, "learning_rate": 1.4120773387537594e-05, "loss": 1.5149, "step": 4358 }, { "epoch": 0.38, "learning_rate": 1.411816719334766e-05, "loss": 1.4682, "step": 4359 }, { "epoch": 0.38, "learning_rate": 1.4115560662271954e-05, "loss": 1.5428, "step": 4360 }, { "epoch": 0.39, "learning_rate": 1.4112953794523704e-05, "loss": 1.5689, "step": 4361 }, { "epoch": 0.39, "learning_rate": 1.4110346590316164e-05, "loss": 1.5639, "step": 4362 }, { "epoch": 0.39, "learning_rate": 1.4107739049862616e-05, "loss": 1.383, "step": 4363 }, { "epoch": 0.39, "learning_rate": 1.4105131173376367e-05, "loss": 1.4562, "step": 4364 }, { "epoch": 0.39, "learning_rate": 1.4102522961070758e-05, "loss": 1.428, "step": 4365 }, { "epoch": 0.39, "learning_rate": 1.4099914413159148e-05, "loss": 1.417, "step": 4366 }, { "epoch": 0.39, "learning_rate": 1.4097305529854933e-05, "loss": 1.5562, "step": 4367 }, { "epoch": 0.39, "learning_rate": 1.409469631137153e-05, "loss": 1.5792, "step": 4368 }, { "epoch": 0.39, "learning_rate": 1.4092086757922384e-05, "loss": 1.5493, "step": 4369 }, { "epoch": 0.39, "learning_rate": 1.4089476869720973e-05, "loss": 1.3948, "step": 4370 }, { "epoch": 0.39, "learning_rate": 1.4086866646980796e-05, "loss": 1.4699, "step": 4371 }, { "epoch": 0.39, "learning_rate": 1.4084256089915385e-05, "loss": 1.4764, "step": 4372 }, { "epoch": 0.39, "learning_rate": 1.4081645198738289e-05, "loss": 1.5282, "step": 4373 }, { "epoch": 0.39, "learning_rate": 1.4079033973663096e-05, "loss": 1.5045, "step": 4374 }, { "epoch": 0.39, "learning_rate": 1.4076422414903419e-05, "loss": 1.587, "step": 4375 }, { "epoch": 0.39, "learning_rate": 1.4073810522672893e-05, "loss": 1.4311, "step": 4376 }, { "epoch": 0.39, "learning_rate": 1.4071198297185182e-05, "loss": 1.6519, "step": 4377 }, { "epoch": 0.39, "learning_rate": 1.4068585738653979e-05, "loss": 1.4376, "step": 4378 }, { "epoch": 0.39, "learning_rate": 1.4065972847293006e-05, "loss": 1.3732, "step": 4379 }, { "epoch": 0.39, "learning_rate": 1.4063359623316006e-05, "loss": 1.4949, "step": 4380 }, { "epoch": 0.39, "learning_rate": 1.4060746066936762e-05, "loss": 1.3409, "step": 4381 }, { "epoch": 0.39, "learning_rate": 1.4058132178369066e-05, "loss": 1.4843, "step": 4382 }, { "epoch": 0.39, "learning_rate": 1.405551795782675e-05, "loss": 1.5001, "step": 4383 }, { "epoch": 0.39, "learning_rate": 1.4052903405523668e-05, "loss": 1.5377, "step": 4384 }, { "epoch": 0.39, "learning_rate": 1.405028852167371e-05, "loss": 1.4987, "step": 4385 }, { "epoch": 0.39, "learning_rate": 1.4047673306490777e-05, "loss": 1.3695, "step": 4386 }, { "epoch": 0.39, "learning_rate": 1.404505776018881e-05, "loss": 1.3918, "step": 4387 }, { "epoch": 0.39, "learning_rate": 1.4042441882981773e-05, "loss": 1.6395, "step": 4388 }, { "epoch": 0.39, "learning_rate": 1.4039825675083659e-05, "loss": 1.4103, "step": 4389 }, { "epoch": 0.39, "learning_rate": 1.4037209136708481e-05, "loss": 1.4552, "step": 4390 }, { "epoch": 0.39, "learning_rate": 1.4034592268070289e-05, "loss": 1.5038, "step": 4391 }, { "epoch": 0.39, "learning_rate": 1.4031975069383154e-05, "loss": 1.5485, "step": 4392 }, { "epoch": 0.39, "learning_rate": 1.4029357540861176e-05, "loss": 1.4425, "step": 4393 }, { "epoch": 0.39, "learning_rate": 1.4026739682718483e-05, "loss": 1.4788, "step": 4394 }, { "epoch": 0.39, "learning_rate": 1.4024121495169224e-05, "loss": 1.4879, "step": 4395 }, { "epoch": 0.39, "learning_rate": 1.4021502978427583e-05, "loss": 1.5763, "step": 4396 }, { "epoch": 0.39, "learning_rate": 1.4018884132707766e-05, "loss": 1.578, "step": 4397 }, { "epoch": 0.39, "learning_rate": 1.4016264958224009e-05, "loss": 1.4758, "step": 4398 }, { "epoch": 0.39, "learning_rate": 1.4013645455190568e-05, "loss": 1.496, "step": 4399 }, { "epoch": 0.39, "learning_rate": 1.4011025623821733e-05, "loss": 1.4397, "step": 4400 }, { "epoch": 0.39, "learning_rate": 1.400840546433182e-05, "loss": 1.4678, "step": 4401 }, { "epoch": 0.39, "learning_rate": 1.400578497693517e-05, "loss": 1.6952, "step": 4402 }, { "epoch": 0.39, "learning_rate": 1.4003164161846153e-05, "loss": 1.3956, "step": 4403 }, { "epoch": 0.39, "learning_rate": 1.400054301927916e-05, "loss": 1.4057, "step": 4404 }, { "epoch": 0.39, "learning_rate": 1.3997921549448617e-05, "loss": 1.3822, "step": 4405 }, { "epoch": 0.39, "learning_rate": 1.3995299752568972e-05, "loss": 1.5518, "step": 4406 }, { "epoch": 0.39, "learning_rate": 1.39926776288547e-05, "loss": 1.3631, "step": 4407 }, { "epoch": 0.39, "learning_rate": 1.3990055178520303e-05, "loss": 1.5226, "step": 4408 }, { "epoch": 0.39, "learning_rate": 1.3987432401780309e-05, "loss": 1.4855, "step": 4409 }, { "epoch": 0.39, "learning_rate": 1.3984809298849275e-05, "loss": 1.5666, "step": 4410 }, { "epoch": 0.39, "learning_rate": 1.3982185869941787e-05, "loss": 1.4583, "step": 4411 }, { "epoch": 0.39, "learning_rate": 1.3979562115272446e-05, "loss": 1.3257, "step": 4412 }, { "epoch": 0.39, "learning_rate": 1.3976938035055894e-05, "loss": 1.4625, "step": 4413 }, { "epoch": 0.39, "learning_rate": 1.397431362950679e-05, "loss": 1.5283, "step": 4414 }, { "epoch": 0.39, "learning_rate": 1.3971688898839828e-05, "loss": 1.3766, "step": 4415 }, { "epoch": 0.39, "learning_rate": 1.3969063843269714e-05, "loss": 1.4689, "step": 4416 }, { "epoch": 0.39, "learning_rate": 1.3966438463011199e-05, "loss": 1.4008, "step": 4417 }, { "epoch": 0.39, "learning_rate": 1.3963812758279047e-05, "loss": 1.4485, "step": 4418 }, { "epoch": 0.39, "learning_rate": 1.3961186729288057e-05, "loss": 1.4904, "step": 4419 }, { "epoch": 0.39, "learning_rate": 1.3958560376253048e-05, "loss": 1.4241, "step": 4420 }, { "epoch": 0.39, "learning_rate": 1.3955933699388867e-05, "loss": 1.5358, "step": 4421 }, { "epoch": 0.39, "learning_rate": 1.3953306698910389e-05, "loss": 1.4622, "step": 4422 }, { "epoch": 0.39, "learning_rate": 1.3950679375032519e-05, "loss": 1.548, "step": 4423 }, { "epoch": 0.39, "learning_rate": 1.3948051727970182e-05, "loss": 1.3811, "step": 4424 }, { "epoch": 0.39, "learning_rate": 1.3945423757938332e-05, "loss": 1.4131, "step": 4425 }, { "epoch": 0.39, "learning_rate": 1.394279546515195e-05, "loss": 1.466, "step": 4426 }, { "epoch": 0.39, "learning_rate": 1.3940166849826043e-05, "loss": 1.371, "step": 4427 }, { "epoch": 0.39, "learning_rate": 1.3937537912175645e-05, "loss": 1.4278, "step": 4428 }, { "epoch": 0.39, "learning_rate": 1.3934908652415813e-05, "loss": 1.4653, "step": 4429 }, { "epoch": 0.39, "learning_rate": 1.3932279070761633e-05, "loss": 1.4664, "step": 4430 }, { "epoch": 0.39, "learning_rate": 1.3929649167428222e-05, "loss": 1.3646, "step": 4431 }, { "epoch": 0.39, "learning_rate": 1.3927018942630715e-05, "loss": 1.4547, "step": 4432 }, { "epoch": 0.39, "learning_rate": 1.392438839658428e-05, "loss": 1.5085, "step": 4433 }, { "epoch": 0.39, "learning_rate": 1.3921757529504102e-05, "loss": 1.4499, "step": 4434 }, { "epoch": 0.39, "learning_rate": 1.3919126341605407e-05, "loss": 1.3988, "step": 4435 }, { "epoch": 0.39, "learning_rate": 1.3916494833103431e-05, "loss": 1.5286, "step": 4436 }, { "epoch": 0.39, "learning_rate": 1.3913863004213448e-05, "loss": 1.4559, "step": 4437 }, { "epoch": 0.39, "learning_rate": 1.3911230855150755e-05, "loss": 1.5558, "step": 4438 }, { "epoch": 0.39, "learning_rate": 1.390859838613067e-05, "loss": 1.5174, "step": 4439 }, { "epoch": 0.39, "learning_rate": 1.3905965597368544e-05, "loss": 1.4063, "step": 4440 }, { "epoch": 0.39, "learning_rate": 1.3903332489079757e-05, "loss": 1.4529, "step": 4441 }, { "epoch": 0.39, "learning_rate": 1.39006990614797e-05, "loss": 1.5159, "step": 4442 }, { "epoch": 0.39, "learning_rate": 1.3898065314783805e-05, "loss": 1.5083, "step": 4443 }, { "epoch": 0.39, "learning_rate": 1.3895431249207527e-05, "loss": 1.4105, "step": 4444 }, { "epoch": 0.39, "learning_rate": 1.3892796864966343e-05, "loss": 1.3829, "step": 4445 }, { "epoch": 0.39, "learning_rate": 1.389016216227576e-05, "loss": 1.4989, "step": 4446 }, { "epoch": 0.39, "learning_rate": 1.3887527141351302e-05, "loss": 1.4639, "step": 4447 }, { "epoch": 0.39, "learning_rate": 1.3884891802408538e-05, "loss": 1.3441, "step": 4448 }, { "epoch": 0.39, "learning_rate": 1.3882256145663042e-05, "loss": 1.3881, "step": 4449 }, { "epoch": 0.39, "learning_rate": 1.387962017133043e-05, "loss": 1.4612, "step": 4450 }, { "epoch": 0.39, "learning_rate": 1.387698387962633e-05, "loss": 1.5249, "step": 4451 }, { "epoch": 0.39, "learning_rate": 1.387434727076641e-05, "loss": 1.5629, "step": 4452 }, { "epoch": 0.39, "learning_rate": 1.3871710344966354e-05, "loss": 1.4436, "step": 4453 }, { "epoch": 0.39, "learning_rate": 1.3869073102441874e-05, "loss": 1.4196, "step": 4454 }, { "epoch": 0.39, "learning_rate": 1.3866435543408711e-05, "loss": 1.3076, "step": 4455 }, { "epoch": 0.39, "learning_rate": 1.386379766808263e-05, "loss": 1.4182, "step": 4456 }, { "epoch": 0.39, "learning_rate": 1.3861159476679422e-05, "loss": 1.4236, "step": 4457 }, { "epoch": 0.39, "learning_rate": 1.3858520969414901e-05, "loss": 1.5652, "step": 4458 }, { "epoch": 0.39, "learning_rate": 1.3855882146504916e-05, "loss": 1.5251, "step": 4459 }, { "epoch": 0.39, "learning_rate": 1.3853243008165324e-05, "loss": 1.335, "step": 4460 }, { "epoch": 0.39, "learning_rate": 1.3850603554612031e-05, "loss": 1.4897, "step": 4461 }, { "epoch": 0.39, "learning_rate": 1.3847963786060952e-05, "loss": 1.4816, "step": 4462 }, { "epoch": 0.39, "learning_rate": 1.3845323702728032e-05, "loss": 1.2994, "step": 4463 }, { "epoch": 0.39, "learning_rate": 1.384268330482924e-05, "loss": 1.5029, "step": 4464 }, { "epoch": 0.39, "learning_rate": 1.384004259258058e-05, "loss": 1.3698, "step": 4465 }, { "epoch": 0.39, "learning_rate": 1.383740156619807e-05, "loss": 1.5606, "step": 4466 }, { "epoch": 0.39, "learning_rate": 1.3834760225897764e-05, "loss": 1.4711, "step": 4467 }, { "epoch": 0.39, "learning_rate": 1.3832118571895723e-05, "loss": 1.5469, "step": 4468 }, { "epoch": 0.39, "learning_rate": 1.3829476604408064e-05, "loss": 1.42, "step": 4469 }, { "epoch": 0.39, "learning_rate": 1.3826834323650899e-05, "loss": 1.4645, "step": 4470 }, { "epoch": 0.39, "learning_rate": 1.3824191729840387e-05, "loss": 1.5089, "step": 4471 }, { "epoch": 0.39, "learning_rate": 1.3821548823192703e-05, "loss": 1.4351, "step": 4472 }, { "epoch": 0.39, "learning_rate": 1.3818905603924047e-05, "loss": 1.5348, "step": 4473 }, { "epoch": 0.4, "learning_rate": 1.3816262072250653e-05, "loss": 1.5124, "step": 4474 }, { "epoch": 0.4, "learning_rate": 1.3813618228388765e-05, "loss": 1.418, "step": 4475 }, { "epoch": 0.4, "learning_rate": 1.3810974072554673e-05, "loss": 1.5563, "step": 4476 }, { "epoch": 0.4, "learning_rate": 1.3808329604964672e-05, "loss": 1.425, "step": 4477 }, { "epoch": 0.4, "learning_rate": 1.3805684825835098e-05, "loss": 1.4973, "step": 4478 }, { "epoch": 0.4, "learning_rate": 1.3803039735382304e-05, "loss": 1.393, "step": 4479 }, { "epoch": 0.4, "learning_rate": 1.3800394333822674e-05, "loss": 1.5534, "step": 4480 }, { "epoch": 0.4, "learning_rate": 1.3797748621372609e-05, "loss": 1.4524, "step": 4481 }, { "epoch": 0.4, "learning_rate": 1.3795102598248547e-05, "loss": 1.4827, "step": 4482 }, { "epoch": 0.4, "learning_rate": 1.379245626466694e-05, "loss": 1.4443, "step": 4483 }, { "epoch": 0.4, "learning_rate": 1.3789809620844275e-05, "loss": 1.4058, "step": 4484 }, { "epoch": 0.4, "learning_rate": 1.3787162666997058e-05, "loss": 1.4988, "step": 4485 }, { "epoch": 0.4, "learning_rate": 1.3784515403341824e-05, "loss": 1.488, "step": 4486 }, { "epoch": 0.4, "learning_rate": 1.378186783009513e-05, "loss": 1.4478, "step": 4487 }, { "epoch": 0.4, "learning_rate": 1.377921994747356e-05, "loss": 1.4652, "step": 4488 }, { "epoch": 0.4, "learning_rate": 1.3776571755693726e-05, "loss": 1.4888, "step": 4489 }, { "epoch": 0.4, "learning_rate": 1.3773923254972259e-05, "loss": 1.3512, "step": 4490 }, { "epoch": 0.4, "learning_rate": 1.3771274445525824e-05, "loss": 1.429, "step": 4491 }, { "epoch": 0.4, "learning_rate": 1.37686253275711e-05, "loss": 1.5648, "step": 4492 }, { "epoch": 0.4, "learning_rate": 1.3765975901324806e-05, "loss": 1.4448, "step": 4493 }, { "epoch": 0.4, "learning_rate": 1.376332616700367e-05, "loss": 1.4641, "step": 4494 }, { "epoch": 0.4, "learning_rate": 1.3760676124824457e-05, "loss": 1.5495, "step": 4495 }, { "epoch": 0.4, "learning_rate": 1.375802577500395e-05, "loss": 1.5017, "step": 4496 }, { "epoch": 0.4, "learning_rate": 1.3755375117758961e-05, "loss": 1.4028, "step": 4497 }, { "epoch": 0.4, "learning_rate": 1.375272415330633e-05, "loss": 1.3974, "step": 4498 }, { "epoch": 0.4, "learning_rate": 1.3750072881862912e-05, "loss": 1.487, "step": 4499 }, { "epoch": 0.4, "learning_rate": 1.3747421303645603e-05, "loss": 1.5116, "step": 4500 }, { "epoch": 0.4, "learning_rate": 1.3744769418871307e-05, "loss": 1.3903, "step": 4501 }, { "epoch": 0.4, "learning_rate": 1.3742117227756963e-05, "loss": 1.3655, "step": 4502 }, { "epoch": 0.4, "learning_rate": 1.3739464730519535e-05, "loss": 1.466, "step": 4503 }, { "epoch": 0.4, "learning_rate": 1.3736811927376008e-05, "loss": 1.4353, "step": 4504 }, { "epoch": 0.4, "learning_rate": 1.3734158818543392e-05, "loss": 1.4619, "step": 4505 }, { "epoch": 0.4, "learning_rate": 1.3731505404238728e-05, "loss": 1.4974, "step": 4506 }, { "epoch": 0.4, "learning_rate": 1.3728851684679077e-05, "loss": 1.4421, "step": 4507 }, { "epoch": 0.4, "learning_rate": 1.3726197660081523e-05, "loss": 1.4611, "step": 4508 }, { "epoch": 0.4, "learning_rate": 1.3723543330663183e-05, "loss": 1.4855, "step": 4509 }, { "epoch": 0.4, "learning_rate": 1.3720888696641189e-05, "loss": 1.6333, "step": 4510 }, { "epoch": 0.4, "learning_rate": 1.3718233758232704e-05, "loss": 1.4544, "step": 4511 }, { "epoch": 0.4, "learning_rate": 1.3715578515654915e-05, "loss": 1.4627, "step": 4512 }, { "epoch": 0.4, "learning_rate": 1.3712922969125038e-05, "loss": 1.5259, "step": 4513 }, { "epoch": 0.4, "learning_rate": 1.3710267118860301e-05, "loss": 1.4629, "step": 4514 }, { "epoch": 0.4, "learning_rate": 1.370761096507797e-05, "loss": 1.3949, "step": 4515 }, { "epoch": 0.4, "learning_rate": 1.3704954507995331e-05, "loss": 1.5448, "step": 4516 }, { "epoch": 0.4, "learning_rate": 1.3702297747829698e-05, "loss": 1.4825, "step": 4517 }, { "epoch": 0.4, "learning_rate": 1.36996406847984e-05, "loss": 1.3429, "step": 4518 }, { "epoch": 0.4, "learning_rate": 1.36969833191188e-05, "loss": 1.4006, "step": 4519 }, { "epoch": 0.4, "learning_rate": 1.3694325651008285e-05, "loss": 1.5418, "step": 4520 }, { "epoch": 0.4, "learning_rate": 1.3691667680684266e-05, "loss": 1.4853, "step": 4521 }, { "epoch": 0.4, "learning_rate": 1.3689009408364174e-05, "loss": 1.4926, "step": 4522 }, { "epoch": 0.4, "learning_rate": 1.3686350834265467e-05, "loss": 1.5236, "step": 4523 }, { "epoch": 0.4, "learning_rate": 1.3683691958605634e-05, "loss": 1.4012, "step": 4524 }, { "epoch": 0.4, "learning_rate": 1.3681032781602182e-05, "loss": 1.5556, "step": 4525 }, { "epoch": 0.4, "learning_rate": 1.3678373303472646e-05, "loss": 1.3942, "step": 4526 }, { "epoch": 0.4, "learning_rate": 1.367571352443458e-05, "loss": 1.4664, "step": 4527 }, { "epoch": 0.4, "learning_rate": 1.3673053444705568e-05, "loss": 1.353, "step": 4528 }, { "epoch": 0.4, "learning_rate": 1.3670393064503219e-05, "loss": 1.4829, "step": 4529 }, { "epoch": 0.4, "learning_rate": 1.3667732384045164e-05, "loss": 1.4517, "step": 4530 }, { "epoch": 0.4, "learning_rate": 1.3665071403549062e-05, "loss": 1.3903, "step": 4531 }, { "epoch": 0.4, "learning_rate": 1.3662410123232588e-05, "loss": 1.4537, "step": 4532 }, { "epoch": 0.4, "learning_rate": 1.365974854331345e-05, "loss": 1.5372, "step": 4533 }, { "epoch": 0.4, "learning_rate": 1.3657086664009382e-05, "loss": 1.4475, "step": 4534 }, { "epoch": 0.4, "learning_rate": 1.3654424485538132e-05, "loss": 1.5465, "step": 4535 }, { "epoch": 0.4, "learning_rate": 1.3651762008117482e-05, "loss": 1.4929, "step": 4536 }, { "epoch": 0.4, "learning_rate": 1.3649099231965238e-05, "loss": 1.4585, "step": 4537 }, { "epoch": 0.4, "learning_rate": 1.3646436157299223e-05, "loss": 1.4685, "step": 4538 }, { "epoch": 0.4, "learning_rate": 1.3643772784337293e-05, "loss": 1.3323, "step": 4539 }, { "epoch": 0.4, "learning_rate": 1.3641109113297321e-05, "loss": 1.4912, "step": 4540 }, { "epoch": 0.4, "learning_rate": 1.3638445144397213e-05, "loss": 1.498, "step": 4541 }, { "epoch": 0.4, "learning_rate": 1.3635780877854887e-05, "loss": 1.54, "step": 4542 }, { "epoch": 0.4, "learning_rate": 1.3633116313888301e-05, "loss": 1.4533, "step": 4543 }, { "epoch": 0.4, "learning_rate": 1.3630451452715426e-05, "loss": 1.3981, "step": 4544 }, { "epoch": 0.4, "learning_rate": 1.3627786294554258e-05, "loss": 1.5174, "step": 4545 }, { "epoch": 0.4, "learning_rate": 1.3625120839622821e-05, "loss": 1.5132, "step": 4546 }, { "epoch": 0.4, "learning_rate": 1.3622455088139166e-05, "loss": 1.3137, "step": 4547 }, { "epoch": 0.4, "learning_rate": 1.3619789040321356e-05, "loss": 1.4707, "step": 4548 }, { "epoch": 0.4, "learning_rate": 1.3617122696387493e-05, "loss": 1.4486, "step": 4549 }, { "epoch": 0.4, "learning_rate": 1.3614456056555693e-05, "loss": 1.4861, "step": 4550 }, { "epoch": 0.4, "learning_rate": 1.3611789121044104e-05, "loss": 1.4546, "step": 4551 }, { "epoch": 0.4, "learning_rate": 1.3609121890070894e-05, "loss": 1.5007, "step": 4552 }, { "epoch": 0.4, "learning_rate": 1.360645436385425e-05, "loss": 1.4622, "step": 4553 }, { "epoch": 0.4, "learning_rate": 1.360378654261239e-05, "loss": 1.299, "step": 4554 }, { "epoch": 0.4, "learning_rate": 1.3601118426563558e-05, "loss": 1.3305, "step": 4555 }, { "epoch": 0.4, "learning_rate": 1.3598450015926017e-05, "loss": 1.3101, "step": 4556 }, { "epoch": 0.4, "learning_rate": 1.3595781310918052e-05, "loss": 1.5017, "step": 4557 }, { "epoch": 0.4, "learning_rate": 1.359311231175798e-05, "loss": 1.4263, "step": 4558 }, { "epoch": 0.4, "learning_rate": 1.359044301866414e-05, "loss": 1.3358, "step": 4559 }, { "epoch": 0.4, "learning_rate": 1.358777343185489e-05, "loss": 1.4479, "step": 4560 }, { "epoch": 0.4, "learning_rate": 1.3585103551548614e-05, "loss": 1.5565, "step": 4561 }, { "epoch": 0.4, "learning_rate": 1.358243337796372e-05, "loss": 1.47, "step": 4562 }, { "epoch": 0.4, "learning_rate": 1.3579762911318644e-05, "loss": 1.4926, "step": 4563 }, { "epoch": 0.4, "learning_rate": 1.357709215183184e-05, "loss": 1.5718, "step": 4564 }, { "epoch": 0.4, "learning_rate": 1.3574421099721794e-05, "loss": 1.4429, "step": 4565 }, { "epoch": 0.4, "learning_rate": 1.3571749755207008e-05, "loss": 1.5384, "step": 4566 }, { "epoch": 0.4, "learning_rate": 1.3569078118506008e-05, "loss": 1.4027, "step": 4567 }, { "epoch": 0.4, "learning_rate": 1.3566406189837349e-05, "loss": 1.4488, "step": 4568 }, { "epoch": 0.4, "learning_rate": 1.356373396941961e-05, "loss": 1.4752, "step": 4569 }, { "epoch": 0.4, "learning_rate": 1.3561061457471386e-05, "loss": 1.5367, "step": 4570 }, { "epoch": 0.4, "learning_rate": 1.3558388654211307e-05, "loss": 1.439, "step": 4571 }, { "epoch": 0.4, "learning_rate": 1.3555715559858016e-05, "loss": 1.4579, "step": 4572 }, { "epoch": 0.4, "learning_rate": 1.355304217463019e-05, "loss": 1.4333, "step": 4573 }, { "epoch": 0.4, "learning_rate": 1.3550368498746517e-05, "loss": 1.3217, "step": 4574 }, { "epoch": 0.4, "learning_rate": 1.3547694532425726e-05, "loss": 1.5484, "step": 4575 }, { "epoch": 0.4, "learning_rate": 1.3545020275886554e-05, "loss": 1.4651, "step": 4576 }, { "epoch": 0.4, "learning_rate": 1.3542345729347767e-05, "loss": 1.4667, "step": 4577 }, { "epoch": 0.4, "learning_rate": 1.3539670893028162e-05, "loss": 1.4256, "step": 4578 }, { "epoch": 0.4, "learning_rate": 1.353699576714655e-05, "loss": 1.4725, "step": 4579 }, { "epoch": 0.4, "learning_rate": 1.3534320351921766e-05, "loss": 1.5245, "step": 4580 }, { "epoch": 0.4, "learning_rate": 1.3531644647572675e-05, "loss": 1.4171, "step": 4581 }, { "epoch": 0.4, "learning_rate": 1.3528968654318164e-05, "loss": 1.4348, "step": 4582 }, { "epoch": 0.4, "learning_rate": 1.3526292372377138e-05, "loss": 1.3654, "step": 4583 }, { "epoch": 0.4, "learning_rate": 1.3523615801968534e-05, "loss": 1.3867, "step": 4584 }, { "epoch": 0.4, "learning_rate": 1.3520938943311303e-05, "loss": 1.4624, "step": 4585 }, { "epoch": 0.4, "learning_rate": 1.3518261796624432e-05, "loss": 1.4144, "step": 4586 }, { "epoch": 0.41, "learning_rate": 1.3515584362126917e-05, "loss": 1.4295, "step": 4587 }, { "epoch": 0.41, "learning_rate": 1.3512906640037789e-05, "loss": 1.4204, "step": 4588 }, { "epoch": 0.41, "learning_rate": 1.3510228630576095e-05, "loss": 1.4303, "step": 4589 }, { "epoch": 0.41, "learning_rate": 1.3507550333960913e-05, "loss": 1.582, "step": 4590 }, { "epoch": 0.41, "learning_rate": 1.350487175041134e-05, "loss": 1.2843, "step": 4591 }, { "epoch": 0.41, "learning_rate": 1.3502192880146495e-05, "loss": 1.3855, "step": 4592 }, { "epoch": 0.41, "learning_rate": 1.3499513723385525e-05, "loss": 1.5898, "step": 4593 }, { "epoch": 0.41, "learning_rate": 1.3496834280347594e-05, "loss": 1.5616, "step": 4594 }, { "epoch": 0.41, "learning_rate": 1.34941545512519e-05, "loss": 1.4208, "step": 4595 }, { "epoch": 0.41, "learning_rate": 1.3491474536317649e-05, "loss": 1.3947, "step": 4596 }, { "epoch": 0.41, "learning_rate": 1.3488794235764083e-05, "loss": 1.513, "step": 4597 }, { "epoch": 0.41, "learning_rate": 1.3486113649810463e-05, "loss": 1.3886, "step": 4598 }, { "epoch": 0.41, "learning_rate": 1.3483432778676078e-05, "loss": 1.4962, "step": 4599 }, { "epoch": 0.41, "learning_rate": 1.3480751622580228e-05, "loss": 1.4069, "step": 4600 }, { "epoch": 0.41, "learning_rate": 1.3478070181742254e-05, "loss": 1.4351, "step": 4601 }, { "epoch": 0.41, "learning_rate": 1.34753884563815e-05, "loss": 1.3698, "step": 4602 }, { "epoch": 0.41, "learning_rate": 1.3472706446717354e-05, "loss": 1.4872, "step": 4603 }, { "epoch": 0.41, "learning_rate": 1.3470024152969209e-05, "loss": 1.555, "step": 4604 }, { "epoch": 0.41, "learning_rate": 1.3467341575356492e-05, "loss": 1.4642, "step": 4605 }, { "epoch": 0.41, "learning_rate": 1.3464658714098658e-05, "loss": 1.3122, "step": 4606 }, { "epoch": 0.41, "learning_rate": 1.3461975569415167e-05, "loss": 1.4348, "step": 4607 }, { "epoch": 0.41, "learning_rate": 1.3459292141525519e-05, "loss": 1.4554, "step": 4608 }, { "epoch": 0.41, "learning_rate": 1.3456608430649227e-05, "loss": 1.4743, "step": 4609 }, { "epoch": 0.41, "learning_rate": 1.3453924437005838e-05, "loss": 1.3044, "step": 4610 }, { "epoch": 0.41, "learning_rate": 1.3451240160814911e-05, "loss": 1.4383, "step": 4611 }, { "epoch": 0.41, "learning_rate": 1.3448555602296035e-05, "loss": 1.5818, "step": 4612 }, { "epoch": 0.41, "learning_rate": 1.3445870761668815e-05, "loss": 1.4767, "step": 4613 }, { "epoch": 0.41, "learning_rate": 1.344318563915289e-05, "loss": 1.4253, "step": 4614 }, { "epoch": 0.41, "learning_rate": 1.3440500234967913e-05, "loss": 1.3845, "step": 4615 }, { "epoch": 0.41, "learning_rate": 1.3437814549333564e-05, "loss": 1.3125, "step": 4616 }, { "epoch": 0.41, "learning_rate": 1.343512858246954e-05, "loss": 1.3944, "step": 4617 }, { "epoch": 0.41, "learning_rate": 1.3432442334595565e-05, "loss": 1.4765, "step": 4618 }, { "epoch": 0.41, "learning_rate": 1.3429755805931402e-05, "loss": 1.2592, "step": 4619 }, { "epoch": 0.41, "learning_rate": 1.3427068996696806e-05, "loss": 1.4904, "step": 4620 }, { "epoch": 0.41, "learning_rate": 1.3424381907111579e-05, "loss": 1.4889, "step": 4621 }, { "epoch": 0.41, "learning_rate": 1.342169453739553e-05, "loss": 1.586, "step": 4622 }, { "epoch": 0.41, "learning_rate": 1.341900688776851e-05, "loss": 1.4454, "step": 4623 }, { "epoch": 0.41, "learning_rate": 1.341631895845037e-05, "loss": 1.5929, "step": 4624 }, { "epoch": 0.41, "learning_rate": 1.3413630749661006e-05, "loss": 1.4963, "step": 4625 }, { "epoch": 0.41, "learning_rate": 1.3410942261620314e-05, "loss": 1.474, "step": 4626 }, { "epoch": 0.41, "learning_rate": 1.340825349454824e-05, "loss": 1.4688, "step": 4627 }, { "epoch": 0.41, "learning_rate": 1.3405564448664727e-05, "loss": 1.3213, "step": 4628 }, { "epoch": 0.41, "learning_rate": 1.3402875124189758e-05, "loss": 1.3689, "step": 4629 }, { "epoch": 0.41, "learning_rate": 1.3400185521343322e-05, "loss": 1.5017, "step": 4630 }, { "epoch": 0.41, "learning_rate": 1.3397495640345458e-05, "loss": 1.4634, "step": 4631 }, { "epoch": 0.41, "learning_rate": 1.3394805481416198e-05, "loss": 1.4172, "step": 4632 }, { "epoch": 0.41, "learning_rate": 1.3392115044775613e-05, "loss": 1.4419, "step": 4633 }, { "epoch": 0.41, "learning_rate": 1.3389424330643798e-05, "loss": 1.4199, "step": 4634 }, { "epoch": 0.41, "learning_rate": 1.338673333924086e-05, "loss": 1.4673, "step": 4635 }, { "epoch": 0.41, "learning_rate": 1.3384042070786942e-05, "loss": 1.5313, "step": 4636 }, { "epoch": 0.41, "learning_rate": 1.3381350525502197e-05, "loss": 1.4098, "step": 4637 }, { "epoch": 0.41, "learning_rate": 1.3378658703606808e-05, "loss": 1.3952, "step": 4638 }, { "epoch": 0.41, "learning_rate": 1.3375966605320977e-05, "loss": 1.4177, "step": 4639 }, { "epoch": 0.41, "learning_rate": 1.3373274230864934e-05, "loss": 1.5267, "step": 4640 }, { "epoch": 0.41, "learning_rate": 1.3370581580458926e-05, "loss": 1.417, "step": 4641 }, { "epoch": 0.41, "learning_rate": 1.3367888654323226e-05, "loss": 1.5818, "step": 4642 }, { "epoch": 0.41, "learning_rate": 1.3365195452678124e-05, "loss": 1.4873, "step": 4643 }, { "epoch": 0.41, "learning_rate": 1.336250197574394e-05, "loss": 1.4543, "step": 4644 }, { "epoch": 0.41, "learning_rate": 1.335980822374102e-05, "loss": 1.4893, "step": 4645 }, { "epoch": 0.41, "learning_rate": 1.3357114196889713e-05, "loss": 1.5395, "step": 4646 }, { "epoch": 0.41, "learning_rate": 1.3354419895410411e-05, "loss": 1.4832, "step": 4647 }, { "epoch": 0.41, "learning_rate": 1.3351725319523521e-05, "loss": 1.3916, "step": 4648 }, { "epoch": 0.41, "learning_rate": 1.3349030469449469e-05, "loss": 1.424, "step": 4649 }, { "epoch": 0.41, "learning_rate": 1.3346335345408708e-05, "loss": 1.519, "step": 4650 }, { "epoch": 0.41, "learning_rate": 1.3343639947621712e-05, "loss": 1.6149, "step": 4651 }, { "epoch": 0.41, "learning_rate": 1.3340944276308978e-05, "loss": 1.378, "step": 4652 }, { "epoch": 0.41, "learning_rate": 1.3338248331691026e-05, "loss": 1.4562, "step": 4653 }, { "epoch": 0.41, "learning_rate": 1.3335552113988392e-05, "loss": 1.5593, "step": 4654 }, { "epoch": 0.41, "learning_rate": 1.3332855623421647e-05, "loss": 1.3897, "step": 4655 }, { "epoch": 0.41, "learning_rate": 1.3330158860211369e-05, "loss": 1.6374, "step": 4656 }, { "epoch": 0.41, "learning_rate": 1.3327461824578169e-05, "loss": 1.4237, "step": 4657 }, { "epoch": 0.41, "learning_rate": 1.3324764516742683e-05, "loss": 1.4386, "step": 4658 }, { "epoch": 0.41, "learning_rate": 1.3322066936925557e-05, "loss": 1.4523, "step": 4659 }, { "epoch": 0.41, "learning_rate": 1.3319369085347468e-05, "loss": 1.4692, "step": 4660 }, { "epoch": 0.41, "learning_rate": 1.3316670962229112e-05, "loss": 1.3679, "step": 4661 }, { "epoch": 0.41, "learning_rate": 1.3313972567791211e-05, "loss": 1.56, "step": 4662 }, { "epoch": 0.41, "learning_rate": 1.3311273902254507e-05, "loss": 1.5425, "step": 4663 }, { "epoch": 0.41, "learning_rate": 1.3308574965839757e-05, "loss": 1.4843, "step": 4664 }, { "epoch": 0.41, "learning_rate": 1.3305875758767754e-05, "loss": 1.3562, "step": 4665 }, { "epoch": 0.41, "learning_rate": 1.3303176281259305e-05, "loss": 1.4527, "step": 4666 }, { "epoch": 0.41, "learning_rate": 1.3300476533535239e-05, "loss": 1.515, "step": 4667 }, { "epoch": 0.41, "learning_rate": 1.3297776515816408e-05, "loss": 1.5487, "step": 4668 }, { "epoch": 0.41, "learning_rate": 1.3295076228323684e-05, "loss": 1.4286, "step": 4669 }, { "epoch": 0.41, "learning_rate": 1.3292375671277966e-05, "loss": 1.5574, "step": 4670 }, { "epoch": 0.41, "learning_rate": 1.3289674844900176e-05, "loss": 1.4349, "step": 4671 }, { "epoch": 0.41, "learning_rate": 1.328697374941125e-05, "loss": 1.3317, "step": 4672 }, { "epoch": 0.41, "learning_rate": 1.3284272385032151e-05, "loss": 1.5632, "step": 4673 }, { "epoch": 0.41, "learning_rate": 1.3281570751983863e-05, "loss": 1.507, "step": 4674 }, { "epoch": 0.41, "learning_rate": 1.3278868850487397e-05, "loss": 1.4257, "step": 4675 }, { "epoch": 0.41, "learning_rate": 1.3276166680763773e-05, "loss": 1.4649, "step": 4676 }, { "epoch": 0.41, "learning_rate": 1.3273464243034048e-05, "loss": 1.3909, "step": 4677 }, { "epoch": 0.41, "learning_rate": 1.3270761537519294e-05, "loss": 1.4068, "step": 4678 }, { "epoch": 0.41, "learning_rate": 1.3268058564440604e-05, "loss": 1.3828, "step": 4679 }, { "epoch": 0.41, "learning_rate": 1.3265355324019097e-05, "loss": 1.4313, "step": 4680 }, { "epoch": 0.41, "learning_rate": 1.3262651816475902e-05, "loss": 1.478, "step": 4681 }, { "epoch": 0.41, "learning_rate": 1.3259948042032188e-05, "loss": 1.4216, "step": 4682 }, { "epoch": 0.41, "learning_rate": 1.3257244000909135e-05, "loss": 1.4808, "step": 4683 }, { "epoch": 0.41, "learning_rate": 1.3254539693327943e-05, "loss": 1.3221, "step": 4684 }, { "epoch": 0.41, "learning_rate": 1.325183511950984e-05, "loss": 1.3383, "step": 4685 }, { "epoch": 0.41, "learning_rate": 1.3249130279676073e-05, "loss": 1.5017, "step": 4686 }, { "epoch": 0.41, "learning_rate": 1.324642517404791e-05, "loss": 1.4938, "step": 4687 }, { "epoch": 0.41, "learning_rate": 1.3243719802846644e-05, "loss": 1.5016, "step": 4688 }, { "epoch": 0.41, "learning_rate": 1.3241014166293581e-05, "loss": 1.4272, "step": 4689 }, { "epoch": 0.41, "learning_rate": 1.323830826461006e-05, "loss": 1.4032, "step": 4690 }, { "epoch": 0.41, "learning_rate": 1.323560209801744e-05, "loss": 1.4562, "step": 4691 }, { "epoch": 0.41, "learning_rate": 1.3232895666737092e-05, "loss": 1.4472, "step": 4692 }, { "epoch": 0.41, "learning_rate": 1.3230188970990417e-05, "loss": 1.4376, "step": 4693 }, { "epoch": 0.41, "learning_rate": 1.3227482010998835e-05, "loss": 1.4173, "step": 4694 }, { "epoch": 0.41, "learning_rate": 1.322477478698379e-05, "loss": 1.5102, "step": 4695 }, { "epoch": 0.41, "learning_rate": 1.3222067299166748e-05, "loss": 1.4454, "step": 4696 }, { "epoch": 0.41, "learning_rate": 1.3219359547769189e-05, "loss": 1.4264, "step": 4697 }, { "epoch": 0.41, "learning_rate": 1.321665153301262e-05, "loss": 1.4142, "step": 4698 }, { "epoch": 0.41, "learning_rate": 1.3213943255118575e-05, "loss": 1.4711, "step": 4699 }, { "epoch": 0.42, "learning_rate": 1.3211234714308601e-05, "loss": 1.4614, "step": 4700 }, { "epoch": 0.42, "learning_rate": 1.3208525910804271e-05, "loss": 1.3153, "step": 4701 }, { "epoch": 0.42, "learning_rate": 1.3205816844827175e-05, "loss": 1.3464, "step": 4702 }, { "epoch": 0.42, "learning_rate": 1.320310751659893e-05, "loss": 1.4875, "step": 4703 }, { "epoch": 0.42, "learning_rate": 1.320039792634117e-05, "loss": 1.4442, "step": 4704 }, { "epoch": 0.42, "learning_rate": 1.3197688074275558e-05, "loss": 1.5166, "step": 4705 }, { "epoch": 0.42, "learning_rate": 1.3194977960623767e-05, "loss": 1.4458, "step": 4706 }, { "epoch": 0.42, "learning_rate": 1.31922675856075e-05, "loss": 1.5895, "step": 4707 }, { "epoch": 0.42, "learning_rate": 1.3189556949448475e-05, "loss": 1.4386, "step": 4708 }, { "epoch": 0.42, "learning_rate": 1.318684605236844e-05, "loss": 1.3877, "step": 4709 }, { "epoch": 0.42, "learning_rate": 1.3184134894589154e-05, "loss": 1.4709, "step": 4710 }, { "epoch": 0.42, "learning_rate": 1.3181423476332408e-05, "loss": 1.5374, "step": 4711 }, { "epoch": 0.42, "learning_rate": 1.3178711797820006e-05, "loss": 1.3141, "step": 4712 }, { "epoch": 0.42, "learning_rate": 1.3175999859273776e-05, "loss": 1.4929, "step": 4713 }, { "epoch": 0.42, "learning_rate": 1.3173287660915573e-05, "loss": 1.4125, "step": 4714 }, { "epoch": 0.42, "learning_rate": 1.3170575202967257e-05, "loss": 1.519, "step": 4715 }, { "epoch": 0.42, "learning_rate": 1.3167862485650729e-05, "loss": 1.5557, "step": 4716 }, { "epoch": 0.42, "learning_rate": 1.31651495091879e-05, "loss": 1.3225, "step": 4717 }, { "epoch": 0.42, "learning_rate": 1.3162436273800705e-05, "loss": 1.33, "step": 4718 }, { "epoch": 0.42, "learning_rate": 1.3159722779711098e-05, "loss": 1.4142, "step": 4719 }, { "epoch": 0.42, "learning_rate": 1.3157009027141054e-05, "loss": 1.4377, "step": 4720 }, { "epoch": 0.42, "learning_rate": 1.3154295016312576e-05, "loss": 1.4916, "step": 4721 }, { "epoch": 0.42, "learning_rate": 1.3151580747447682e-05, "loss": 1.4704, "step": 4722 }, { "epoch": 0.42, "learning_rate": 1.3148866220768407e-05, "loss": 1.3878, "step": 4723 }, { "epoch": 0.42, "learning_rate": 1.3146151436496817e-05, "loss": 1.5111, "step": 4724 }, { "epoch": 0.42, "learning_rate": 1.3143436394854992e-05, "loss": 1.5348, "step": 4725 }, { "epoch": 0.42, "learning_rate": 1.3140721096065039e-05, "loss": 1.4241, "step": 4726 }, { "epoch": 0.42, "learning_rate": 1.3138005540349079e-05, "loss": 1.4369, "step": 4727 }, { "epoch": 0.42, "learning_rate": 1.3135289727929259e-05, "loss": 1.477, "step": 4728 }, { "epoch": 0.42, "learning_rate": 1.3132573659027743e-05, "loss": 1.3595, "step": 4729 }, { "epoch": 0.42, "learning_rate": 1.3129857333866722e-05, "loss": 1.3746, "step": 4730 }, { "epoch": 0.42, "learning_rate": 1.3127140752668404e-05, "loss": 1.5095, "step": 4731 }, { "epoch": 0.42, "learning_rate": 1.3124423915655014e-05, "loss": 1.5186, "step": 4732 }, { "epoch": 0.42, "learning_rate": 1.3121706823048808e-05, "loss": 1.4068, "step": 4733 }, { "epoch": 0.42, "learning_rate": 1.3118989475072053e-05, "loss": 1.5291, "step": 4734 }, { "epoch": 0.42, "learning_rate": 1.3116271871947043e-05, "loss": 1.5486, "step": 4735 }, { "epoch": 0.42, "learning_rate": 1.311355401389609e-05, "loss": 1.485, "step": 4736 }, { "epoch": 0.42, "learning_rate": 1.3110835901141528e-05, "loss": 1.506, "step": 4737 }, { "epoch": 0.42, "learning_rate": 1.3108117533905713e-05, "loss": 1.5106, "step": 4738 }, { "epoch": 0.42, "learning_rate": 1.3105398912411019e-05, "loss": 1.4363, "step": 4739 }, { "epoch": 0.42, "learning_rate": 1.3102680036879844e-05, "loss": 1.441, "step": 4740 }, { "epoch": 0.42, "learning_rate": 1.3099960907534602e-05, "loss": 1.3406, "step": 4741 }, { "epoch": 0.42, "learning_rate": 1.3097241524597733e-05, "loss": 1.4404, "step": 4742 }, { "epoch": 0.42, "learning_rate": 1.309452188829169e-05, "loss": 1.3943, "step": 4743 }, { "epoch": 0.42, "learning_rate": 1.3091801998838966e-05, "loss": 1.4475, "step": 4744 }, { "epoch": 0.42, "learning_rate": 1.3089081856462046e-05, "loss": 1.4755, "step": 4745 }, { "epoch": 0.42, "learning_rate": 1.3086361461383458e-05, "loss": 1.6048, "step": 4746 }, { "epoch": 0.42, "learning_rate": 1.3083640813825745e-05, "loss": 1.4589, "step": 4747 }, { "epoch": 0.42, "learning_rate": 1.3080919914011463e-05, "loss": 1.3505, "step": 4748 }, { "epoch": 0.42, "learning_rate": 1.30781987621632e-05, "loss": 1.4591, "step": 4749 }, { "epoch": 0.42, "learning_rate": 1.3075477358503552e-05, "loss": 1.5021, "step": 4750 }, { "epoch": 0.42, "learning_rate": 1.3072755703255154e-05, "loss": 1.4852, "step": 4751 }, { "epoch": 0.42, "learning_rate": 1.307003379664064e-05, "loss": 1.3961, "step": 4752 }, { "epoch": 0.42, "learning_rate": 1.3067311638882682e-05, "loss": 1.3674, "step": 4753 }, { "epoch": 0.42, "learning_rate": 1.3064589230203958e-05, "loss": 1.4767, "step": 4754 }, { "epoch": 0.42, "learning_rate": 1.3061866570827183e-05, "loss": 1.3691, "step": 4755 }, { "epoch": 0.42, "learning_rate": 1.305914366097508e-05, "loss": 1.309, "step": 4756 }, { "epoch": 0.42, "learning_rate": 1.3056420500870394e-05, "loss": 1.419, "step": 4757 }, { "epoch": 0.42, "learning_rate": 1.3053697090735892e-05, "loss": 1.4544, "step": 4758 }, { "epoch": 0.42, "learning_rate": 1.3050973430794367e-05, "loss": 1.5168, "step": 4759 }, { "epoch": 0.42, "learning_rate": 1.3048249521268623e-05, "loss": 1.4949, "step": 4760 }, { "epoch": 0.42, "learning_rate": 1.304552536238149e-05, "loss": 1.5872, "step": 4761 }, { "epoch": 0.42, "learning_rate": 1.3042800954355816e-05, "loss": 1.3729, "step": 4762 }, { "epoch": 0.42, "learning_rate": 1.3040076297414474e-05, "loss": 1.5199, "step": 4763 }, { "epoch": 0.42, "learning_rate": 1.303735139178035e-05, "loss": 1.5236, "step": 4764 }, { "epoch": 0.42, "learning_rate": 1.3034626237676357e-05, "loss": 1.4979, "step": 4765 }, { "epoch": 0.42, "learning_rate": 1.3031900835325427e-05, "loss": 1.432, "step": 4766 }, { "epoch": 0.42, "learning_rate": 1.3029175184950505e-05, "loss": 1.513, "step": 4767 }, { "epoch": 0.42, "learning_rate": 1.302644928677457e-05, "loss": 1.4899, "step": 4768 }, { "epoch": 0.42, "learning_rate": 1.3023723141020606e-05, "loss": 1.4604, "step": 4769 }, { "epoch": 0.42, "learning_rate": 1.3020996747911632e-05, "loss": 1.5879, "step": 4770 }, { "epoch": 0.42, "learning_rate": 1.3018270107670673e-05, "loss": 1.5503, "step": 4771 }, { "epoch": 0.42, "learning_rate": 1.3015543220520786e-05, "loss": 1.4707, "step": 4772 }, { "epoch": 0.42, "learning_rate": 1.301281608668504e-05, "loss": 1.454, "step": 4773 }, { "epoch": 0.42, "learning_rate": 1.3010088706386533e-05, "loss": 1.2713, "step": 4774 }, { "epoch": 0.42, "learning_rate": 1.3007361079848368e-05, "loss": 1.4012, "step": 4775 }, { "epoch": 0.42, "learning_rate": 1.3004633207293692e-05, "loss": 1.3518, "step": 4776 }, { "epoch": 0.42, "learning_rate": 1.3001905088945646e-05, "loss": 1.5014, "step": 4777 }, { "epoch": 0.42, "learning_rate": 1.2999176725027404e-05, "loss": 1.5412, "step": 4778 }, { "epoch": 0.42, "learning_rate": 1.299644811576217e-05, "loss": 1.4661, "step": 4779 }, { "epoch": 0.42, "learning_rate": 1.2993719261373144e-05, "loss": 1.4681, "step": 4780 }, { "epoch": 0.42, "learning_rate": 1.2990990162083567e-05, "loss": 1.5129, "step": 4781 }, { "epoch": 0.42, "learning_rate": 1.2988260818116692e-05, "loss": 1.4841, "step": 4782 }, { "epoch": 0.42, "learning_rate": 1.2985531229695792e-05, "loss": 1.4521, "step": 4783 }, { "epoch": 0.42, "learning_rate": 1.2982801397044153e-05, "loss": 1.4827, "step": 4784 }, { "epoch": 0.42, "learning_rate": 1.2980071320385102e-05, "loss": 1.479, "step": 4785 }, { "epoch": 0.42, "learning_rate": 1.2977340999941959e-05, "loss": 1.4226, "step": 4786 }, { "epoch": 0.42, "learning_rate": 1.2974610435938088e-05, "loss": 1.3196, "step": 4787 }, { "epoch": 0.42, "learning_rate": 1.2971879628596853e-05, "loss": 1.4235, "step": 4788 }, { "epoch": 0.42, "learning_rate": 1.2969148578141654e-05, "loss": 1.3962, "step": 4789 }, { "epoch": 0.42, "learning_rate": 1.2966417284795902e-05, "loss": 1.4588, "step": 4790 }, { "epoch": 0.42, "learning_rate": 1.2963685748783027e-05, "loss": 1.5246, "step": 4791 }, { "epoch": 0.42, "learning_rate": 1.2960953970326487e-05, "loss": 1.6143, "step": 4792 }, { "epoch": 0.42, "learning_rate": 1.295822194964975e-05, "loss": 1.351, "step": 4793 }, { "epoch": 0.42, "learning_rate": 1.295548968697631e-05, "loss": 1.4957, "step": 4794 }, { "epoch": 0.42, "learning_rate": 1.295275718252968e-05, "loss": 1.3876, "step": 4795 }, { "epoch": 0.42, "learning_rate": 1.295002443653339e-05, "loss": 1.3886, "step": 4796 }, { "epoch": 0.42, "learning_rate": 1.2947291449210995e-05, "loss": 1.4863, "step": 4797 }, { "epoch": 0.42, "learning_rate": 1.2944558220786063e-05, "loss": 1.3829, "step": 4798 }, { "epoch": 0.42, "learning_rate": 1.2941824751482185e-05, "loss": 1.4736, "step": 4799 }, { "epoch": 0.42, "learning_rate": 1.2939091041522975e-05, "loss": 1.6057, "step": 4800 }, { "epoch": 0.42, "learning_rate": 1.2936357091132059e-05, "loss": 1.5238, "step": 4801 }, { "epoch": 0.42, "learning_rate": 1.2933622900533094e-05, "loss": 1.4411, "step": 4802 }, { "epoch": 0.42, "learning_rate": 1.2930888469949743e-05, "loss": 1.5457, "step": 4803 }, { "epoch": 0.42, "learning_rate": 1.2928153799605698e-05, "loss": 1.4254, "step": 4804 }, { "epoch": 0.42, "learning_rate": 1.292541888972467e-05, "loss": 1.4765, "step": 4805 }, { "epoch": 0.42, "learning_rate": 1.2922683740530386e-05, "loss": 1.5059, "step": 4806 }, { "epoch": 0.42, "learning_rate": 1.2919948352246596e-05, "loss": 1.3254, "step": 4807 }, { "epoch": 0.42, "learning_rate": 1.2917212725097064e-05, "loss": 1.4569, "step": 4808 }, { "epoch": 0.42, "learning_rate": 1.291447685930558e-05, "loss": 1.3821, "step": 4809 }, { "epoch": 0.42, "learning_rate": 1.291174075509595e-05, "loss": 1.3617, "step": 4810 }, { "epoch": 0.42, "learning_rate": 1.2909004412692006e-05, "loss": 1.4285, "step": 4811 }, { "epoch": 0.42, "learning_rate": 1.2906267832317583e-05, "loss": 1.4402, "step": 4812 }, { "epoch": 0.42, "learning_rate": 1.2903531014196553e-05, "loss": 1.4704, "step": 4813 }, { "epoch": 0.43, "learning_rate": 1.2900793958552801e-05, "loss": 1.4143, "step": 4814 }, { "epoch": 0.43, "learning_rate": 1.2898056665610232e-05, "loss": 1.3771, "step": 4815 }, { "epoch": 0.43, "learning_rate": 1.2895319135592765e-05, "loss": 1.4623, "step": 4816 }, { "epoch": 0.43, "learning_rate": 1.2892581368724349e-05, "loss": 1.6069, "step": 4817 }, { "epoch": 0.43, "learning_rate": 1.2889843365228942e-05, "loss": 1.4844, "step": 4818 }, { "epoch": 0.43, "learning_rate": 1.2887105125330526e-05, "loss": 1.5021, "step": 4819 }, { "epoch": 0.43, "learning_rate": 1.2884366649253106e-05, "loss": 1.353, "step": 4820 }, { "epoch": 0.43, "learning_rate": 1.2881627937220698e-05, "loss": 1.3972, "step": 4821 }, { "epoch": 0.43, "learning_rate": 1.2878888989457343e-05, "loss": 1.433, "step": 4822 }, { "epoch": 0.43, "learning_rate": 1.2876149806187102e-05, "loss": 1.5233, "step": 4823 }, { "epoch": 0.43, "learning_rate": 1.287341038763405e-05, "loss": 1.4001, "step": 4824 }, { "epoch": 0.43, "learning_rate": 1.2870670734022287e-05, "loss": 1.4808, "step": 4825 }, { "epoch": 0.43, "learning_rate": 1.2867930845575932e-05, "loss": 1.4449, "step": 4826 }, { "epoch": 0.43, "learning_rate": 1.2865190722519114e-05, "loss": 1.5321, "step": 4827 }, { "epoch": 0.43, "learning_rate": 1.2862450365075995e-05, "loss": 1.472, "step": 4828 }, { "epoch": 0.43, "learning_rate": 1.2859709773470748e-05, "loss": 1.4042, "step": 4829 }, { "epoch": 0.43, "learning_rate": 1.2856968947927561e-05, "loss": 1.4781, "step": 4830 }, { "epoch": 0.43, "learning_rate": 1.2854227888670651e-05, "loss": 1.4616, "step": 4831 }, { "epoch": 0.43, "learning_rate": 1.2851486595924252e-05, "loss": 1.4964, "step": 4832 }, { "epoch": 0.43, "learning_rate": 1.2848745069912613e-05, "loss": 1.5922, "step": 4833 }, { "epoch": 0.43, "learning_rate": 1.2846003310860003e-05, "loss": 1.4905, "step": 4834 }, { "epoch": 0.43, "learning_rate": 1.284326131899071e-05, "loss": 1.4323, "step": 4835 }, { "epoch": 0.43, "learning_rate": 1.2840519094529043e-05, "loss": 1.4095, "step": 4836 }, { "epoch": 0.43, "learning_rate": 1.283777663769933e-05, "loss": 1.4307, "step": 4837 }, { "epoch": 0.43, "learning_rate": 1.2835033948725917e-05, "loss": 1.434, "step": 4838 }, { "epoch": 0.43, "learning_rate": 1.2832291027833168e-05, "loss": 1.3874, "step": 4839 }, { "epoch": 0.43, "learning_rate": 1.2829547875245467e-05, "loss": 1.3882, "step": 4840 }, { "epoch": 0.43, "learning_rate": 1.2826804491187221e-05, "loss": 1.416, "step": 4841 }, { "epoch": 0.43, "learning_rate": 1.2824060875882847e-05, "loss": 1.3924, "step": 4842 }, { "epoch": 0.43, "learning_rate": 1.2821317029556786e-05, "loss": 1.3867, "step": 4843 }, { "epoch": 0.43, "learning_rate": 1.2818572952433498e-05, "loss": 1.4009, "step": 4844 }, { "epoch": 0.43, "learning_rate": 1.2815828644737464e-05, "loss": 1.5297, "step": 4845 }, { "epoch": 0.43, "learning_rate": 1.2813084106693184e-05, "loss": 1.5651, "step": 4846 }, { "epoch": 0.43, "learning_rate": 1.2810339338525166e-05, "loss": 1.3221, "step": 4847 }, { "epoch": 0.43, "learning_rate": 1.2807594340457952e-05, "loss": 1.5159, "step": 4848 }, { "epoch": 0.43, "learning_rate": 1.2804849112716093e-05, "loss": 1.4374, "step": 4849 }, { "epoch": 0.43, "learning_rate": 1.2802103655524165e-05, "loss": 1.4462, "step": 4850 }, { "epoch": 0.43, "learning_rate": 1.2799357969106753e-05, "loss": 1.4431, "step": 4851 }, { "epoch": 0.43, "learning_rate": 1.2796612053688473e-05, "loss": 1.4142, "step": 4852 }, { "epoch": 0.43, "learning_rate": 1.2793865909493954e-05, "loss": 1.4853, "step": 4853 }, { "epoch": 0.43, "learning_rate": 1.2791119536747842e-05, "loss": 1.4659, "step": 4854 }, { "epoch": 0.43, "learning_rate": 1.2788372935674802e-05, "loss": 1.3414, "step": 4855 }, { "epoch": 0.43, "learning_rate": 1.278562610649952e-05, "loss": 1.4397, "step": 4856 }, { "epoch": 0.43, "learning_rate": 1.2782879049446702e-05, "loss": 1.4824, "step": 4857 }, { "epoch": 0.43, "learning_rate": 1.2780131764741067e-05, "loss": 1.4081, "step": 4858 }, { "epoch": 0.43, "learning_rate": 1.2777384252607361e-05, "loss": 1.5089, "step": 4859 }, { "epoch": 0.43, "learning_rate": 1.2774636513270338e-05, "loss": 1.5089, "step": 4860 }, { "epoch": 0.43, "learning_rate": 1.2771888546954777e-05, "loss": 1.5014, "step": 4861 }, { "epoch": 0.43, "learning_rate": 1.2769140353885479e-05, "loss": 1.3364, "step": 4862 }, { "epoch": 0.43, "learning_rate": 1.2766391934287259e-05, "loss": 1.4815, "step": 4863 }, { "epoch": 0.43, "learning_rate": 1.2763643288384946e-05, "loss": 1.4669, "step": 4864 }, { "epoch": 0.43, "learning_rate": 1.2760894416403393e-05, "loss": 1.455, "step": 4865 }, { "epoch": 0.43, "learning_rate": 1.2758145318567476e-05, "loss": 1.4269, "step": 4866 }, { "epoch": 0.43, "learning_rate": 1.2755395995102082e-05, "loss": 1.4191, "step": 4867 }, { "epoch": 0.43, "learning_rate": 1.2752646446232115e-05, "loss": 1.5308, "step": 4868 }, { "epoch": 0.43, "learning_rate": 1.2749896672182506e-05, "loss": 1.3784, "step": 4869 }, { "epoch": 0.43, "learning_rate": 1.2747146673178196e-05, "loss": 1.5738, "step": 4870 }, { "epoch": 0.43, "learning_rate": 1.2744396449444154e-05, "loss": 1.4375, "step": 4871 }, { "epoch": 0.43, "learning_rate": 1.2741646001205356e-05, "loss": 1.4687, "step": 4872 }, { "epoch": 0.43, "learning_rate": 1.2738895328686802e-05, "loss": 1.3642, "step": 4873 }, { "epoch": 0.43, "learning_rate": 1.2736144432113513e-05, "loss": 1.4547, "step": 4874 }, { "epoch": 0.43, "learning_rate": 1.2733393311710522e-05, "loss": 1.6, "step": 4875 }, { "epoch": 0.43, "learning_rate": 1.2730641967702887e-05, "loss": 1.3773, "step": 4876 }, { "epoch": 0.43, "learning_rate": 1.272789040031568e-05, "loss": 1.5116, "step": 4877 }, { "epoch": 0.43, "learning_rate": 1.2725138609773993e-05, "loss": 1.4511, "step": 4878 }, { "epoch": 0.43, "learning_rate": 1.2722386596302933e-05, "loss": 1.4532, "step": 4879 }, { "epoch": 0.43, "learning_rate": 1.2719634360127632e-05, "loss": 1.3512, "step": 4880 }, { "epoch": 0.43, "learning_rate": 1.2716881901473235e-05, "loss": 1.4708, "step": 4881 }, { "epoch": 0.43, "learning_rate": 1.2714129220564901e-05, "loss": 1.438, "step": 4882 }, { "epoch": 0.43, "learning_rate": 1.2711376317627819e-05, "loss": 1.3562, "step": 4883 }, { "epoch": 0.43, "learning_rate": 1.2708623192887187e-05, "loss": 1.3753, "step": 4884 }, { "epoch": 0.43, "learning_rate": 1.2705869846568224e-05, "loss": 1.6055, "step": 4885 }, { "epoch": 0.43, "learning_rate": 1.2703116278896166e-05, "loss": 1.501, "step": 4886 }, { "epoch": 0.43, "learning_rate": 1.270036249009627e-05, "loss": 1.3135, "step": 4887 }, { "epoch": 0.43, "learning_rate": 1.2697608480393808e-05, "loss": 1.4102, "step": 4888 }, { "epoch": 0.43, "learning_rate": 1.269485425001407e-05, "loss": 1.4093, "step": 4889 }, { "epoch": 0.43, "learning_rate": 1.269209979918237e-05, "loss": 1.5063, "step": 4890 }, { "epoch": 0.43, "learning_rate": 1.2689345128124027e-05, "loss": 1.4523, "step": 4891 }, { "epoch": 0.43, "learning_rate": 1.268659023706439e-05, "loss": 1.5135, "step": 4892 }, { "epoch": 0.43, "learning_rate": 1.268383512622883e-05, "loss": 1.5332, "step": 4893 }, { "epoch": 0.43, "learning_rate": 1.2681079795842717e-05, "loss": 1.4466, "step": 4894 }, { "epoch": 0.43, "learning_rate": 1.2678324246131453e-05, "loss": 1.3733, "step": 4895 }, { "epoch": 0.43, "learning_rate": 1.2675568477320459e-05, "loss": 1.5182, "step": 4896 }, { "epoch": 0.43, "learning_rate": 1.2672812489635167e-05, "loss": 1.3239, "step": 4897 }, { "epoch": 0.43, "learning_rate": 1.267005628330103e-05, "loss": 1.2851, "step": 4898 }, { "epoch": 0.43, "learning_rate": 1.2667299858543517e-05, "loss": 1.4341, "step": 4899 }, { "epoch": 0.43, "learning_rate": 1.2664543215588128e-05, "loss": 1.6482, "step": 4900 }, { "epoch": 0.43, "learning_rate": 1.2661786354660355e-05, "loss": 1.4805, "step": 4901 }, { "epoch": 0.43, "learning_rate": 1.2659029275985731e-05, "loss": 1.4857, "step": 4902 }, { "epoch": 0.43, "learning_rate": 1.2656271979789793e-05, "loss": 1.2565, "step": 4903 }, { "epoch": 0.43, "learning_rate": 1.2653514466298105e-05, "loss": 1.444, "step": 4904 }, { "epoch": 0.43, "learning_rate": 1.2650756735736242e-05, "loss": 1.6025, "step": 4905 }, { "epoch": 0.43, "learning_rate": 1.2647998788329803e-05, "loss": 1.4218, "step": 4906 }, { "epoch": 0.43, "learning_rate": 1.2645240624304399e-05, "loss": 1.5048, "step": 4907 }, { "epoch": 0.43, "learning_rate": 1.2642482243885658e-05, "loss": 1.3734, "step": 4908 }, { "epoch": 0.43, "learning_rate": 1.2639723647299236e-05, "loss": 1.4289, "step": 4909 }, { "epoch": 0.43, "learning_rate": 1.2636964834770793e-05, "loss": 1.4518, "step": 4910 }, { "epoch": 0.43, "learning_rate": 1.2634205806526016e-05, "loss": 1.436, "step": 4911 }, { "epoch": 0.43, "learning_rate": 1.2631446562790603e-05, "loss": 1.4821, "step": 4912 }, { "epoch": 0.43, "learning_rate": 1.2628687103790282e-05, "loss": 1.4385, "step": 4913 }, { "epoch": 0.43, "learning_rate": 1.2625927429750781e-05, "loss": 1.4173, "step": 4914 }, { "epoch": 0.43, "learning_rate": 1.262316754089786e-05, "loss": 1.5113, "step": 4915 }, { "epoch": 0.43, "learning_rate": 1.2620407437457287e-05, "loss": 1.5203, "step": 4916 }, { "epoch": 0.43, "learning_rate": 1.2617647119654857e-05, "loss": 1.5344, "step": 4917 }, { "epoch": 0.43, "learning_rate": 1.2614886587716369e-05, "loss": 1.4125, "step": 4918 }, { "epoch": 0.43, "learning_rate": 1.2612125841867658e-05, "loss": 1.3516, "step": 4919 }, { "epoch": 0.43, "learning_rate": 1.2609364882334555e-05, "loss": 1.5204, "step": 4920 }, { "epoch": 0.43, "learning_rate": 1.2606603709342928e-05, "loss": 1.3531, "step": 4921 }, { "epoch": 0.43, "learning_rate": 1.260384232311865e-05, "loss": 1.4741, "step": 4922 }, { "epoch": 0.43, "learning_rate": 1.2601080723887624e-05, "loss": 1.3146, "step": 4923 }, { "epoch": 0.43, "learning_rate": 1.2598318911875749e-05, "loss": 1.4905, "step": 4924 }, { "epoch": 0.43, "learning_rate": 1.2595556887308959e-05, "loss": 1.513, "step": 4925 }, { "epoch": 0.43, "learning_rate": 1.259279465041321e-05, "loss": 1.4849, "step": 4926 }, { "epoch": 0.44, "learning_rate": 1.2590032201414452e-05, "loss": 1.4701, "step": 4927 }, { "epoch": 0.44, "learning_rate": 1.2587269540538677e-05, "loss": 1.4499, "step": 4928 }, { "epoch": 0.44, "learning_rate": 1.2584506668011876e-05, "loss": 1.3727, "step": 4929 }, { "epoch": 0.44, "learning_rate": 1.2581743584060076e-05, "loss": 1.5014, "step": 4930 }, { "epoch": 0.44, "learning_rate": 1.2578980288909297e-05, "loss": 1.5947, "step": 4931 }, { "epoch": 0.44, "learning_rate": 1.2576216782785602e-05, "loss": 1.4677, "step": 4932 }, { "epoch": 0.44, "learning_rate": 1.2573453065915047e-05, "loss": 1.4723, "step": 4933 }, { "epoch": 0.44, "learning_rate": 1.257068913852373e-05, "loss": 1.5411, "step": 4934 }, { "epoch": 0.44, "learning_rate": 1.2567925000837745e-05, "loss": 1.4625, "step": 4935 }, { "epoch": 0.44, "learning_rate": 1.2565160653083214e-05, "loss": 1.3573, "step": 4936 }, { "epoch": 0.44, "learning_rate": 1.2562396095486274e-05, "loss": 1.4826, "step": 4937 }, { "epoch": 0.44, "learning_rate": 1.255963132827308e-05, "loss": 1.3886, "step": 4938 }, { "epoch": 0.44, "learning_rate": 1.25568663516698e-05, "loss": 1.6024, "step": 4939 }, { "epoch": 0.44, "learning_rate": 1.2554101165902624e-05, "loss": 1.4371, "step": 4940 }, { "epoch": 0.44, "learning_rate": 1.2551335771197761e-05, "loss": 1.4596, "step": 4941 }, { "epoch": 0.44, "learning_rate": 1.2548570167781426e-05, "loss": 1.4154, "step": 4942 }, { "epoch": 0.44, "learning_rate": 1.2545804355879867e-05, "loss": 1.4337, "step": 4943 }, { "epoch": 0.44, "learning_rate": 1.2543038335719336e-05, "loss": 1.38, "step": 4944 }, { "epoch": 0.44, "learning_rate": 1.2540272107526105e-05, "loss": 1.5948, "step": 4945 }, { "epoch": 0.44, "learning_rate": 1.2537505671526467e-05, "loss": 1.5364, "step": 4946 }, { "epoch": 0.44, "learning_rate": 1.2534739027946731e-05, "loss": 1.5789, "step": 4947 }, { "epoch": 0.44, "learning_rate": 1.2531972177013221e-05, "loss": 1.4855, "step": 4948 }, { "epoch": 0.44, "learning_rate": 1.2529205118952276e-05, "loss": 1.477, "step": 4949 }, { "epoch": 0.44, "learning_rate": 1.2526437853990254e-05, "loss": 1.5405, "step": 4950 }, { "epoch": 0.44, "learning_rate": 1.2523670382353537e-05, "loss": 1.4872, "step": 4951 }, { "epoch": 0.44, "learning_rate": 1.2520902704268513e-05, "loss": 1.342, "step": 4952 }, { "epoch": 0.44, "learning_rate": 1.2518134819961594e-05, "loss": 1.3549, "step": 4953 }, { "epoch": 0.44, "learning_rate": 1.25153667296592e-05, "loss": 1.535, "step": 4954 }, { "epoch": 0.44, "learning_rate": 1.2512598433587777e-05, "loss": 1.4488, "step": 4955 }, { "epoch": 0.44, "learning_rate": 1.2509829931973792e-05, "loss": 1.44, "step": 4956 }, { "epoch": 0.44, "learning_rate": 1.2507061225043709e-05, "loss": 1.4891, "step": 4957 }, { "epoch": 0.44, "learning_rate": 1.250429231302403e-05, "loss": 1.5442, "step": 4958 }, { "epoch": 0.44, "learning_rate": 1.2501523196141263e-05, "loss": 1.5405, "step": 4959 }, { "epoch": 0.44, "learning_rate": 1.2498753874621938e-05, "loss": 1.4526, "step": 4960 }, { "epoch": 0.44, "learning_rate": 1.2495984348692593e-05, "loss": 1.4069, "step": 4961 }, { "epoch": 0.44, "learning_rate": 1.2493214618579794e-05, "loss": 1.5914, "step": 4962 }, { "epoch": 0.44, "learning_rate": 1.249044468451011e-05, "loss": 1.4747, "step": 4963 }, { "epoch": 0.44, "learning_rate": 1.2487674546710144e-05, "loss": 1.4604, "step": 4964 }, { "epoch": 0.44, "learning_rate": 1.2484904205406505e-05, "loss": 1.3852, "step": 4965 }, { "epoch": 0.44, "learning_rate": 1.2482133660825816e-05, "loss": 1.375, "step": 4966 }, { "epoch": 0.44, "learning_rate": 1.2479362913194722e-05, "loss": 1.391, "step": 4967 }, { "epoch": 0.44, "learning_rate": 1.2476591962739883e-05, "loss": 1.4619, "step": 4968 }, { "epoch": 0.44, "learning_rate": 1.2473820809687982e-05, "loss": 1.5337, "step": 4969 }, { "epoch": 0.44, "learning_rate": 1.2471049454265705e-05, "loss": 1.4328, "step": 4970 }, { "epoch": 0.44, "learning_rate": 1.2468277896699764e-05, "loss": 1.5523, "step": 4971 }, { "epoch": 0.44, "learning_rate": 1.2465506137216887e-05, "loss": 1.4839, "step": 4972 }, { "epoch": 0.44, "learning_rate": 1.246273417604382e-05, "loss": 1.5448, "step": 4973 }, { "epoch": 0.44, "learning_rate": 1.2459962013407316e-05, "loss": 1.4218, "step": 4974 }, { "epoch": 0.44, "learning_rate": 1.2457189649534156e-05, "loss": 1.4169, "step": 4975 }, { "epoch": 0.44, "learning_rate": 1.2454417084651131e-05, "loss": 1.4245, "step": 4976 }, { "epoch": 0.44, "learning_rate": 1.245164431898505e-05, "loss": 1.3901, "step": 4977 }, { "epoch": 0.44, "learning_rate": 1.2448871352762742e-05, "loss": 1.3887, "step": 4978 }, { "epoch": 0.44, "learning_rate": 1.2446098186211043e-05, "loss": 1.3054, "step": 4979 }, { "epoch": 0.44, "learning_rate": 1.2443324819556812e-05, "loss": 1.546, "step": 4980 }, { "epoch": 0.44, "learning_rate": 1.2440551253026929e-05, "loss": 1.4736, "step": 4981 }, { "epoch": 0.44, "learning_rate": 1.2437777486848282e-05, "loss": 1.5293, "step": 4982 }, { "epoch": 0.44, "learning_rate": 1.2435003521247777e-05, "loss": 1.3677, "step": 4983 }, { "epoch": 0.44, "learning_rate": 1.2432229356452337e-05, "loss": 1.5061, "step": 4984 }, { "epoch": 0.44, "learning_rate": 1.2429454992688905e-05, "loss": 1.4851, "step": 4985 }, { "epoch": 0.44, "learning_rate": 1.2426680430184438e-05, "loss": 1.4285, "step": 4986 }, { "epoch": 0.44, "learning_rate": 1.2423905669165905e-05, "loss": 1.5145, "step": 4987 }, { "epoch": 0.44, "learning_rate": 1.2421130709860294e-05, "loss": 1.5326, "step": 4988 }, { "epoch": 0.44, "learning_rate": 1.2418355552494614e-05, "loss": 1.2603, "step": 4989 }, { "epoch": 0.44, "learning_rate": 1.2415580197295882e-05, "loss": 1.329, "step": 4990 }, { "epoch": 0.44, "learning_rate": 1.241280464449114e-05, "loss": 1.3975, "step": 4991 }, { "epoch": 0.44, "learning_rate": 1.241002889430744e-05, "loss": 1.532, "step": 4992 }, { "epoch": 0.44, "learning_rate": 1.2407252946971848e-05, "loss": 1.4435, "step": 4993 }, { "epoch": 0.44, "learning_rate": 1.240447680271145e-05, "loss": 1.3839, "step": 4994 }, { "epoch": 0.44, "learning_rate": 1.2401700461753357e-05, "loss": 1.3706, "step": 4995 }, { "epoch": 0.44, "learning_rate": 1.2398923924324677e-05, "loss": 1.4061, "step": 4996 }, { "epoch": 0.44, "learning_rate": 1.2396147190652546e-05, "loss": 1.4244, "step": 4997 }, { "epoch": 0.44, "learning_rate": 1.2393370260964116e-05, "loss": 1.5028, "step": 4998 }, { "epoch": 0.44, "learning_rate": 1.2390593135486554e-05, "loss": 1.3962, "step": 4999 }, { "epoch": 0.44, "learning_rate": 1.2387815814447039e-05, "loss": 1.508, "step": 5000 }, { "epoch": 0.44, "learning_rate": 1.2385038298072771e-05, "loss": 1.4821, "step": 5001 }, { "epoch": 0.44, "learning_rate": 1.2382260586590964e-05, "loss": 1.4254, "step": 5002 }, { "epoch": 0.44, "learning_rate": 1.237948268022885e-05, "loss": 1.4747, "step": 5003 }, { "epoch": 0.44, "learning_rate": 1.237670457921367e-05, "loss": 1.3972, "step": 5004 }, { "epoch": 0.44, "learning_rate": 1.237392628377269e-05, "loss": 1.4344, "step": 5005 }, { "epoch": 0.44, "learning_rate": 1.2371147794133188e-05, "loss": 1.4825, "step": 5006 }, { "epoch": 0.44, "learning_rate": 1.2368369110522453e-05, "loss": 1.4803, "step": 5007 }, { "epoch": 0.44, "learning_rate": 1.2365590233167804e-05, "loss": 1.4486, "step": 5008 }, { "epoch": 0.44, "learning_rate": 1.2362811162296557e-05, "loss": 1.5284, "step": 5009 }, { "epoch": 0.44, "learning_rate": 1.2360031898136056e-05, "loss": 1.4523, "step": 5010 }, { "epoch": 0.44, "learning_rate": 1.2357252440913661e-05, "loss": 1.3294, "step": 5011 }, { "epoch": 0.44, "learning_rate": 1.2354472790856745e-05, "loss": 1.4873, "step": 5012 }, { "epoch": 0.44, "learning_rate": 1.235169294819269e-05, "loss": 1.3938, "step": 5013 }, { "epoch": 0.44, "learning_rate": 1.2348912913148908e-05, "loss": 1.4845, "step": 5014 }, { "epoch": 0.44, "learning_rate": 1.2346132685952811e-05, "loss": 1.4093, "step": 5015 }, { "epoch": 0.44, "learning_rate": 1.2343352266831848e-05, "loss": 1.4945, "step": 5016 }, { "epoch": 0.44, "learning_rate": 1.2340571656013457e-05, "loss": 1.3846, "step": 5017 }, { "epoch": 0.44, "learning_rate": 1.2337790853725114e-05, "loss": 1.5036, "step": 5018 }, { "epoch": 0.44, "learning_rate": 1.2335009860194295e-05, "loss": 1.4054, "step": 5019 }, { "epoch": 0.44, "learning_rate": 1.2332228675648506e-05, "loss": 1.543, "step": 5020 }, { "epoch": 0.44, "learning_rate": 1.2329447300315258e-05, "loss": 1.3665, "step": 5021 }, { "epoch": 0.44, "learning_rate": 1.2326665734422076e-05, "loss": 1.4515, "step": 5022 }, { "epoch": 0.44, "learning_rate": 1.2323883978196513e-05, "loss": 1.3431, "step": 5023 }, { "epoch": 0.44, "learning_rate": 1.2321102031866125e-05, "loss": 1.3761, "step": 5024 }, { "epoch": 0.44, "learning_rate": 1.2318319895658495e-05, "loss": 1.3255, "step": 5025 }, { "epoch": 0.44, "learning_rate": 1.2315537569801205e-05, "loss": 1.4217, "step": 5026 }, { "epoch": 0.44, "learning_rate": 1.231275505452187e-05, "loss": 1.5287, "step": 5027 }, { "epoch": 0.44, "learning_rate": 1.2309972350048111e-05, "loss": 1.4303, "step": 5028 }, { "epoch": 0.44, "learning_rate": 1.2307189456607569e-05, "loss": 1.501, "step": 5029 }, { "epoch": 0.44, "learning_rate": 1.230440637442789e-05, "loss": 1.3793, "step": 5030 }, { "epoch": 0.44, "learning_rate": 1.2301623103736752e-05, "loss": 1.5215, "step": 5031 }, { "epoch": 0.44, "learning_rate": 1.2298839644761835e-05, "loss": 1.3933, "step": 5032 }, { "epoch": 0.44, "learning_rate": 1.2296055997730843e-05, "loss": 1.4053, "step": 5033 }, { "epoch": 0.44, "learning_rate": 1.229327216287149e-05, "loss": 1.4227, "step": 5034 }, { "epoch": 0.44, "learning_rate": 1.2290488140411506e-05, "loss": 1.4822, "step": 5035 }, { "epoch": 0.44, "learning_rate": 1.2287703930578637e-05, "loss": 1.4421, "step": 5036 }, { "epoch": 0.44, "learning_rate": 1.2284919533600645e-05, "loss": 1.4229, "step": 5037 }, { "epoch": 0.44, "learning_rate": 1.2282134949705311e-05, "loss": 1.3927, "step": 5038 }, { "epoch": 0.44, "learning_rate": 1.2279350179120422e-05, "loss": 1.51, "step": 5039 }, { "epoch": 0.45, "learning_rate": 1.2276565222073786e-05, "loss": 1.4786, "step": 5040 }, { "epoch": 0.45, "learning_rate": 1.2273780078793227e-05, "loss": 1.5717, "step": 5041 }, { "epoch": 0.45, "learning_rate": 1.2270994749506589e-05, "loss": 1.495, "step": 5042 }, { "epoch": 0.45, "learning_rate": 1.2268209234441715e-05, "loss": 1.4092, "step": 5043 }, { "epoch": 0.45, "learning_rate": 1.2265423533826475e-05, "loss": 1.4223, "step": 5044 }, { "epoch": 0.45, "learning_rate": 1.226263764788876e-05, "loss": 1.4574, "step": 5045 }, { "epoch": 0.45, "learning_rate": 1.2259851576856463e-05, "loss": 1.3964, "step": 5046 }, { "epoch": 0.45, "learning_rate": 1.2257065320957502e-05, "loss": 1.4294, "step": 5047 }, { "epoch": 0.45, "learning_rate": 1.2254278880419802e-05, "loss": 1.4957, "step": 5048 }, { "epoch": 0.45, "learning_rate": 1.2251492255471306e-05, "loss": 1.4691, "step": 5049 }, { "epoch": 0.45, "learning_rate": 1.2248705446339978e-05, "loss": 1.4144, "step": 5050 }, { "epoch": 0.45, "learning_rate": 1.2245918453253793e-05, "loss": 1.3376, "step": 5051 }, { "epoch": 0.45, "learning_rate": 1.2243131276440733e-05, "loss": 1.3239, "step": 5052 }, { "epoch": 0.45, "learning_rate": 1.2240343916128809e-05, "loss": 1.4736, "step": 5053 }, { "epoch": 0.45, "learning_rate": 1.2237556372546038e-05, "loss": 1.387, "step": 5054 }, { "epoch": 0.45, "learning_rate": 1.2234768645920457e-05, "loss": 1.5129, "step": 5055 }, { "epoch": 0.45, "learning_rate": 1.223198073648011e-05, "loss": 1.5839, "step": 5056 }, { "epoch": 0.45, "learning_rate": 1.2229192644453062e-05, "loss": 1.5352, "step": 5057 }, { "epoch": 0.45, "learning_rate": 1.2226404370067402e-05, "loss": 1.4595, "step": 5058 }, { "epoch": 0.45, "learning_rate": 1.2223615913551212e-05, "loss": 1.482, "step": 5059 }, { "epoch": 0.45, "learning_rate": 1.2220827275132609e-05, "loss": 1.451, "step": 5060 }, { "epoch": 0.45, "learning_rate": 1.2218038455039709e-05, "loss": 1.4314, "step": 5061 }, { "epoch": 0.45, "learning_rate": 1.221524945350066e-05, "loss": 1.4673, "step": 5062 }, { "epoch": 0.45, "learning_rate": 1.221246027074361e-05, "loss": 1.4887, "step": 5063 }, { "epoch": 0.45, "learning_rate": 1.2209670906996731e-05, "loss": 1.4439, "step": 5064 }, { "epoch": 0.45, "learning_rate": 1.2206881362488198e-05, "loss": 1.4345, "step": 5065 }, { "epoch": 0.45, "learning_rate": 1.2204091637446223e-05, "loss": 1.5103, "step": 5066 }, { "epoch": 0.45, "learning_rate": 1.2201301732099007e-05, "loss": 1.4221, "step": 5067 }, { "epoch": 0.45, "learning_rate": 1.2198511646674785e-05, "loss": 1.5129, "step": 5068 }, { "epoch": 0.45, "learning_rate": 1.219572138140179e-05, "loss": 1.3622, "step": 5069 }, { "epoch": 0.45, "learning_rate": 1.2192930936508287e-05, "loss": 1.3451, "step": 5070 }, { "epoch": 0.45, "learning_rate": 1.219014031222255e-05, "loss": 1.3827, "step": 5071 }, { "epoch": 0.45, "learning_rate": 1.2187349508772857e-05, "loss": 1.5328, "step": 5072 }, { "epoch": 0.45, "learning_rate": 1.2184558526387517e-05, "loss": 1.4069, "step": 5073 }, { "epoch": 0.45, "learning_rate": 1.2181767365294836e-05, "loss": 1.4655, "step": 5074 }, { "epoch": 0.45, "learning_rate": 1.2178976025723156e-05, "loss": 1.4082, "step": 5075 }, { "epoch": 0.45, "learning_rate": 1.2176184507900815e-05, "loss": 1.3821, "step": 5076 }, { "epoch": 0.45, "learning_rate": 1.2173392812056174e-05, "loss": 1.2905, "step": 5077 }, { "epoch": 0.45, "learning_rate": 1.2170600938417606e-05, "loss": 1.3836, "step": 5078 }, { "epoch": 0.45, "learning_rate": 1.2167808887213501e-05, "loss": 1.5526, "step": 5079 }, { "epoch": 0.45, "learning_rate": 1.216501665867226e-05, "loss": 1.5034, "step": 5080 }, { "epoch": 0.45, "learning_rate": 1.2162224253022305e-05, "loss": 1.376, "step": 5081 }, { "epoch": 0.45, "learning_rate": 1.215943167049206e-05, "loss": 1.4255, "step": 5082 }, { "epoch": 0.45, "learning_rate": 1.2156638911309984e-05, "loss": 1.5695, "step": 5083 }, { "epoch": 0.45, "learning_rate": 1.2153845975704528e-05, "loss": 1.4296, "step": 5084 }, { "epoch": 0.45, "learning_rate": 1.215105286390417e-05, "loss": 1.504, "step": 5085 }, { "epoch": 0.45, "learning_rate": 1.2148259576137402e-05, "loss": 1.4236, "step": 5086 }, { "epoch": 0.45, "learning_rate": 1.2145466112632725e-05, "loss": 1.5143, "step": 5087 }, { "epoch": 0.45, "learning_rate": 1.214267247361866e-05, "loss": 1.4637, "step": 5088 }, { "epoch": 0.45, "learning_rate": 1.2139878659323738e-05, "loss": 1.3732, "step": 5089 }, { "epoch": 0.45, "learning_rate": 1.2137084669976513e-05, "loss": 1.5183, "step": 5090 }, { "epoch": 0.45, "learning_rate": 1.2134290505805536e-05, "loss": 1.3765, "step": 5091 }, { "epoch": 0.45, "learning_rate": 1.2131496167039391e-05, "loss": 1.4122, "step": 5092 }, { "epoch": 0.45, "learning_rate": 1.2128701653906664e-05, "loss": 1.4766, "step": 5093 }, { "epoch": 0.45, "learning_rate": 1.2125906966635964e-05, "loss": 1.5142, "step": 5094 }, { "epoch": 0.45, "learning_rate": 1.2123112105455904e-05, "loss": 1.4242, "step": 5095 }, { "epoch": 0.45, "learning_rate": 1.2120317070595122e-05, "loss": 1.4476, "step": 5096 }, { "epoch": 0.45, "learning_rate": 1.2117521862282264e-05, "loss": 1.6134, "step": 5097 }, { "epoch": 0.45, "learning_rate": 1.211472648074599e-05, "loss": 1.3276, "step": 5098 }, { "epoch": 0.45, "learning_rate": 1.2111930926214975e-05, "loss": 1.4548, "step": 5099 }, { "epoch": 0.45, "learning_rate": 1.210913519891791e-05, "loss": 1.48, "step": 5100 }, { "epoch": 0.45, "learning_rate": 1.2106339299083503e-05, "loss": 1.4461, "step": 5101 }, { "epoch": 0.45, "learning_rate": 1.2103543226940463e-05, "loss": 1.4404, "step": 5102 }, { "epoch": 0.45, "learning_rate": 1.2100746982717529e-05, "loss": 1.4636, "step": 5103 }, { "epoch": 0.45, "learning_rate": 1.2097950566643444e-05, "loss": 1.3391, "step": 5104 }, { "epoch": 0.45, "learning_rate": 1.2095153978946972e-05, "loss": 1.516, "step": 5105 }, { "epoch": 0.45, "learning_rate": 1.2092357219856883e-05, "loss": 1.5012, "step": 5106 }, { "epoch": 0.45, "learning_rate": 1.208956028960197e-05, "loss": 1.4394, "step": 5107 }, { "epoch": 0.45, "learning_rate": 1.2086763188411025e-05, "loss": 1.4695, "step": 5108 }, { "epoch": 0.45, "learning_rate": 1.2083965916512877e-05, "loss": 1.431, "step": 5109 }, { "epoch": 0.45, "learning_rate": 1.208116847413635e-05, "loss": 1.5406, "step": 5110 }, { "epoch": 0.45, "learning_rate": 1.2078370861510286e-05, "loss": 1.4531, "step": 5111 }, { "epoch": 0.45, "learning_rate": 1.2075573078863548e-05, "loss": 1.5227, "step": 5112 }, { "epoch": 0.45, "learning_rate": 1.2072775126425004e-05, "loss": 1.4105, "step": 5113 }, { "epoch": 0.45, "learning_rate": 1.2069977004423543e-05, "loss": 1.4945, "step": 5114 }, { "epoch": 0.45, "learning_rate": 1.2067178713088063e-05, "loss": 1.4306, "step": 5115 }, { "epoch": 0.45, "learning_rate": 1.2064380252647481e-05, "loss": 1.444, "step": 5116 }, { "epoch": 0.45, "learning_rate": 1.2061581623330719e-05, "loss": 1.2698, "step": 5117 }, { "epoch": 0.45, "learning_rate": 1.2058782825366722e-05, "loss": 1.4404, "step": 5118 }, { "epoch": 0.45, "learning_rate": 1.2055983858984441e-05, "loss": 1.4097, "step": 5119 }, { "epoch": 0.45, "learning_rate": 1.2053184724412849e-05, "loss": 1.3079, "step": 5120 }, { "epoch": 0.45, "learning_rate": 1.2050385421880927e-05, "loss": 1.5734, "step": 5121 }, { "epoch": 0.45, "learning_rate": 1.2047585951617676e-05, "loss": 1.5518, "step": 5122 }, { "epoch": 0.45, "learning_rate": 1.2044786313852097e-05, "loss": 1.5661, "step": 5123 }, { "epoch": 0.45, "learning_rate": 1.204198650881322e-05, "loss": 1.4437, "step": 5124 }, { "epoch": 0.45, "learning_rate": 1.203918653673008e-05, "loss": 1.4359, "step": 5125 }, { "epoch": 0.45, "learning_rate": 1.203638639783173e-05, "loss": 1.2909, "step": 5126 }, { "epoch": 0.45, "learning_rate": 1.2033586092347233e-05, "loss": 1.4807, "step": 5127 }, { "epoch": 0.45, "learning_rate": 1.203078562050567e-05, "loss": 1.553, "step": 5128 }, { "epoch": 0.45, "learning_rate": 1.202798498253613e-05, "loss": 1.3408, "step": 5129 }, { "epoch": 0.45, "learning_rate": 1.2025184178667717e-05, "loss": 1.5303, "step": 5130 }, { "epoch": 0.45, "learning_rate": 1.2022383209129559e-05, "loss": 1.4345, "step": 5131 }, { "epoch": 0.45, "learning_rate": 1.2019582074150777e-05, "loss": 1.3612, "step": 5132 }, { "epoch": 0.45, "learning_rate": 1.2016780773960526e-05, "loss": 1.3675, "step": 5133 }, { "epoch": 0.45, "learning_rate": 1.201397930878796e-05, "loss": 1.554, "step": 5134 }, { "epoch": 0.45, "learning_rate": 1.2011177678862258e-05, "loss": 1.3646, "step": 5135 }, { "epoch": 0.45, "learning_rate": 1.20083758844126e-05, "loss": 1.2709, "step": 5136 }, { "epoch": 0.45, "learning_rate": 1.200557392566819e-05, "loss": 1.5052, "step": 5137 }, { "epoch": 0.45, "learning_rate": 1.2002771802858245e-05, "loss": 1.4883, "step": 5138 }, { "epoch": 0.45, "learning_rate": 1.1999969516211984e-05, "loss": 1.612, "step": 5139 }, { "epoch": 0.45, "learning_rate": 1.1997167065958654e-05, "loss": 1.3541, "step": 5140 }, { "epoch": 0.45, "learning_rate": 1.1994364452327508e-05, "loss": 1.523, "step": 5141 }, { "epoch": 0.45, "learning_rate": 1.1991561675547809e-05, "loss": 1.5696, "step": 5142 }, { "epoch": 0.45, "learning_rate": 1.1988758735848842e-05, "loss": 1.4941, "step": 5143 }, { "epoch": 0.45, "learning_rate": 1.19859556334599e-05, "loss": 1.4058, "step": 5144 }, { "epoch": 0.45, "learning_rate": 1.198315236861029e-05, "loss": 1.5607, "step": 5145 }, { "epoch": 0.45, "learning_rate": 1.198034894152933e-05, "loss": 1.4866, "step": 5146 }, { "epoch": 0.45, "learning_rate": 1.1977545352446354e-05, "loss": 1.384, "step": 5147 }, { "epoch": 0.45, "learning_rate": 1.1974741601590715e-05, "loss": 1.4465, "step": 5148 }, { "epoch": 0.45, "learning_rate": 1.1971937689191765e-05, "loss": 1.4162, "step": 5149 }, { "epoch": 0.45, "learning_rate": 1.1969133615478885e-05, "loss": 1.3823, "step": 5150 }, { "epoch": 0.45, "learning_rate": 1.1966329380681454e-05, "loss": 1.5332, "step": 5151 }, { "epoch": 0.45, "learning_rate": 1.1963524985028879e-05, "loss": 1.5026, "step": 5152 }, { "epoch": 0.46, "learning_rate": 1.196072042875057e-05, "loss": 1.4596, "step": 5153 }, { "epoch": 0.46, "learning_rate": 1.1957915712075952e-05, "loss": 1.3431, "step": 5154 }, { "epoch": 0.46, "learning_rate": 1.1955110835234462e-05, "loss": 1.4441, "step": 5155 }, { "epoch": 0.46, "learning_rate": 1.195230579845556e-05, "loss": 1.4932, "step": 5156 }, { "epoch": 0.46, "learning_rate": 1.1949500601968707e-05, "loss": 1.4687, "step": 5157 }, { "epoch": 0.46, "learning_rate": 1.1946695246003377e-05, "loss": 1.4177, "step": 5158 }, { "epoch": 0.46, "learning_rate": 1.1943889730789069e-05, "loss": 1.3396, "step": 5159 }, { "epoch": 0.46, "learning_rate": 1.1941084056555283e-05, "loss": 1.5053, "step": 5160 }, { "epoch": 0.46, "learning_rate": 1.1938278223531541e-05, "loss": 1.3921, "step": 5161 }, { "epoch": 0.46, "learning_rate": 1.1935472231947367e-05, "loss": 1.4771, "step": 5162 }, { "epoch": 0.46, "learning_rate": 1.1932666082032312e-05, "loss": 1.4867, "step": 5163 }, { "epoch": 0.46, "learning_rate": 1.1929859774015924e-05, "loss": 1.4665, "step": 5164 }, { "epoch": 0.46, "learning_rate": 1.1927053308127778e-05, "loss": 1.3789, "step": 5165 }, { "epoch": 0.46, "learning_rate": 1.192424668459746e-05, "loss": 1.3774, "step": 5166 }, { "epoch": 0.46, "learning_rate": 1.1921439903654558e-05, "loss": 1.4085, "step": 5167 }, { "epoch": 0.46, "learning_rate": 1.191863296552868e-05, "loss": 1.3987, "step": 5168 }, { "epoch": 0.46, "learning_rate": 1.1915825870449451e-05, "loss": 1.3428, "step": 5169 }, { "epoch": 0.46, "learning_rate": 1.1913018618646509e-05, "loss": 1.4552, "step": 5170 }, { "epoch": 0.46, "learning_rate": 1.1910211210349487e-05, "loss": 1.4874, "step": 5171 }, { "epoch": 0.46, "learning_rate": 1.1907403645788057e-05, "loss": 1.4493, "step": 5172 }, { "epoch": 0.46, "learning_rate": 1.1904595925191883e-05, "loss": 1.4968, "step": 5173 }, { "epoch": 0.46, "learning_rate": 1.1901788048790659e-05, "loss": 1.5599, "step": 5174 }, { "epoch": 0.46, "learning_rate": 1.1898980016814074e-05, "loss": 1.4046, "step": 5175 }, { "epoch": 0.46, "learning_rate": 1.1896171829491842e-05, "loss": 1.4607, "step": 5176 }, { "epoch": 0.46, "learning_rate": 1.1893363487053686e-05, "loss": 1.4728, "step": 5177 }, { "epoch": 0.46, "learning_rate": 1.1890554989729344e-05, "loss": 1.3128, "step": 5178 }, { "epoch": 0.46, "learning_rate": 1.1887746337748562e-05, "loss": 1.3496, "step": 5179 }, { "epoch": 0.46, "learning_rate": 1.1884937531341101e-05, "loss": 1.3979, "step": 5180 }, { "epoch": 0.46, "learning_rate": 1.1882128570736733e-05, "loss": 1.6529, "step": 5181 }, { "epoch": 0.46, "learning_rate": 1.1879319456165249e-05, "loss": 1.5531, "step": 5182 }, { "epoch": 0.46, "learning_rate": 1.1876510187856447e-05, "loss": 1.4573, "step": 5183 }, { "epoch": 0.46, "learning_rate": 1.1873700766040134e-05, "loss": 1.4325, "step": 5184 }, { "epoch": 0.46, "learning_rate": 1.1870891190946138e-05, "loss": 1.4247, "step": 5185 }, { "epoch": 0.46, "learning_rate": 1.1868081462804297e-05, "loss": 1.4484, "step": 5186 }, { "epoch": 0.46, "learning_rate": 1.186527158184446e-05, "loss": 1.5907, "step": 5187 }, { "epoch": 0.46, "learning_rate": 1.1862461548296484e-05, "loss": 1.4858, "step": 5188 }, { "epoch": 0.46, "learning_rate": 1.1859651362390245e-05, "loss": 1.5144, "step": 5189 }, { "epoch": 0.46, "learning_rate": 1.1856841024355635e-05, "loss": 1.4841, "step": 5190 }, { "epoch": 0.46, "learning_rate": 1.1854030534422545e-05, "loss": 1.4405, "step": 5191 }, { "epoch": 0.46, "learning_rate": 1.1851219892820893e-05, "loss": 1.474, "step": 5192 }, { "epoch": 0.46, "learning_rate": 1.18484090997806e-05, "loss": 1.426, "step": 5193 }, { "epoch": 0.46, "learning_rate": 1.1845598155531602e-05, "loss": 1.4763, "step": 5194 }, { "epoch": 0.46, "learning_rate": 1.1842787060303849e-05, "loss": 1.4198, "step": 5195 }, { "epoch": 0.46, "learning_rate": 1.1839975814327303e-05, "loss": 1.4703, "step": 5196 }, { "epoch": 0.46, "learning_rate": 1.1837164417831933e-05, "loss": 1.4143, "step": 5197 }, { "epoch": 0.46, "learning_rate": 1.1834352871047727e-05, "loss": 1.3519, "step": 5198 }, { "epoch": 0.46, "learning_rate": 1.1831541174204686e-05, "loss": 1.5592, "step": 5199 }, { "epoch": 0.46, "learning_rate": 1.1828729327532818e-05, "loss": 1.465, "step": 5200 }, { "epoch": 0.46, "learning_rate": 1.1825917331262144e-05, "loss": 1.3932, "step": 5201 }, { "epoch": 0.46, "learning_rate": 1.1823105185622699e-05, "loss": 1.6289, "step": 5202 }, { "epoch": 0.46, "learning_rate": 1.1820292890844531e-05, "loss": 1.4218, "step": 5203 }, { "epoch": 0.46, "learning_rate": 1.18174804471577e-05, "loss": 1.3186, "step": 5204 }, { "epoch": 0.46, "learning_rate": 1.181466785479228e-05, "loss": 1.3678, "step": 5205 }, { "epoch": 0.46, "learning_rate": 1.1811855113978344e-05, "loss": 1.4968, "step": 5206 }, { "epoch": 0.46, "learning_rate": 1.1809042224946e-05, "loss": 1.5152, "step": 5207 }, { "epoch": 0.46, "learning_rate": 1.1806229187925347e-05, "loss": 1.5159, "step": 5208 }, { "epoch": 0.46, "learning_rate": 1.180341600314651e-05, "loss": 1.4114, "step": 5209 }, { "epoch": 0.46, "learning_rate": 1.1800602670839619e-05, "loss": 1.4227, "step": 5210 }, { "epoch": 0.46, "learning_rate": 1.1797789191234816e-05, "loss": 1.4225, "step": 5211 }, { "epoch": 0.46, "learning_rate": 1.1794975564562262e-05, "loss": 1.4628, "step": 5212 }, { "epoch": 0.46, "learning_rate": 1.1792161791052123e-05, "loss": 1.3546, "step": 5213 }, { "epoch": 0.46, "learning_rate": 1.1789347870934578e-05, "loss": 1.6246, "step": 5214 }, { "epoch": 0.46, "learning_rate": 1.178653380443982e-05, "loss": 1.4082, "step": 5215 }, { "epoch": 0.46, "learning_rate": 1.1783719591798048e-05, "loss": 1.4965, "step": 5216 }, { "epoch": 0.46, "learning_rate": 1.1780905233239488e-05, "loss": 1.4328, "step": 5217 }, { "epoch": 0.46, "learning_rate": 1.1778090728994362e-05, "loss": 1.4226, "step": 5218 }, { "epoch": 0.46, "learning_rate": 1.177527607929291e-05, "loss": 1.3434, "step": 5219 }, { "epoch": 0.46, "learning_rate": 1.1772461284365386e-05, "loss": 1.3965, "step": 5220 }, { "epoch": 0.46, "learning_rate": 1.1769646344442049e-05, "loss": 1.5019, "step": 5221 }, { "epoch": 0.46, "learning_rate": 1.1766831259753182e-05, "loss": 1.4754, "step": 5222 }, { "epoch": 0.46, "learning_rate": 1.1764016030529065e-05, "loss": 1.4088, "step": 5223 }, { "epoch": 0.46, "learning_rate": 1.1761200657000003e-05, "loss": 1.371, "step": 5224 }, { "epoch": 0.46, "learning_rate": 1.1758385139396302e-05, "loss": 1.3541, "step": 5225 }, { "epoch": 0.46, "learning_rate": 1.175556947794829e-05, "loss": 1.5458, "step": 5226 }, { "epoch": 0.46, "learning_rate": 1.1752753672886294e-05, "loss": 1.555, "step": 5227 }, { "epoch": 0.46, "learning_rate": 1.174993772444067e-05, "loss": 1.4171, "step": 5228 }, { "epoch": 0.46, "learning_rate": 1.174712163284177e-05, "loss": 1.375, "step": 5229 }, { "epoch": 0.46, "learning_rate": 1.1744305398319961e-05, "loss": 1.4897, "step": 5230 }, { "epoch": 0.46, "learning_rate": 1.1741489021105634e-05, "loss": 1.4264, "step": 5231 }, { "epoch": 0.46, "learning_rate": 1.1738672501429171e-05, "loss": 1.4238, "step": 5232 }, { "epoch": 0.46, "learning_rate": 1.1735855839520989e-05, "loss": 1.4727, "step": 5233 }, { "epoch": 0.46, "learning_rate": 1.1733039035611494e-05, "loss": 1.3546, "step": 5234 }, { "epoch": 0.46, "learning_rate": 1.1730222089931118e-05, "loss": 1.4341, "step": 5235 }, { "epoch": 0.46, "learning_rate": 1.17274050027103e-05, "loss": 1.479, "step": 5236 }, { "epoch": 0.46, "learning_rate": 1.1724587774179492e-05, "loss": 1.4637, "step": 5237 }, { "epoch": 0.46, "learning_rate": 1.1721770404569154e-05, "loss": 1.4373, "step": 5238 }, { "epoch": 0.46, "learning_rate": 1.1718952894109767e-05, "loss": 1.418, "step": 5239 }, { "epoch": 0.46, "learning_rate": 1.1716135243031808e-05, "loss": 1.371, "step": 5240 }, { "epoch": 0.46, "learning_rate": 1.1713317451565782e-05, "loss": 1.4609, "step": 5241 }, { "epoch": 0.46, "learning_rate": 1.1710499519942196e-05, "loss": 1.405, "step": 5242 }, { "epoch": 0.46, "learning_rate": 1.1707681448391566e-05, "loss": 1.419, "step": 5243 }, { "epoch": 0.46, "learning_rate": 1.1704863237144427e-05, "loss": 1.4158, "step": 5244 }, { "epoch": 0.46, "learning_rate": 1.1702044886431319e-05, "loss": 1.4533, "step": 5245 }, { "epoch": 0.46, "learning_rate": 1.1699226396482806e-05, "loss": 1.5395, "step": 5246 }, { "epoch": 0.46, "learning_rate": 1.1696407767529443e-05, "loss": 1.3266, "step": 5247 }, { "epoch": 0.46, "learning_rate": 1.1693588999801812e-05, "loss": 1.4016, "step": 5248 }, { "epoch": 0.46, "learning_rate": 1.16907700935305e-05, "loss": 1.4004, "step": 5249 }, { "epoch": 0.46, "learning_rate": 1.168795104894611e-05, "loss": 1.5839, "step": 5250 }, { "epoch": 0.46, "learning_rate": 1.1685131866279253e-05, "loss": 1.5818, "step": 5251 }, { "epoch": 0.46, "learning_rate": 1.1682312545760551e-05, "loss": 1.5449, "step": 5252 }, { "epoch": 0.46, "learning_rate": 1.1679493087620632e-05, "loss": 1.4754, "step": 5253 }, { "epoch": 0.46, "learning_rate": 1.1676673492090151e-05, "loss": 1.3709, "step": 5254 }, { "epoch": 0.46, "learning_rate": 1.1673853759399759e-05, "loss": 1.4453, "step": 5255 }, { "epoch": 0.46, "learning_rate": 1.1671033889780126e-05, "loss": 1.5407, "step": 5256 }, { "epoch": 0.46, "learning_rate": 1.1668213883461923e-05, "loss": 1.4236, "step": 5257 }, { "epoch": 0.46, "learning_rate": 1.166539374067585e-05, "loss": 1.5567, "step": 5258 }, { "epoch": 0.46, "learning_rate": 1.1662573461652604e-05, "loss": 1.3771, "step": 5259 }, { "epoch": 0.46, "learning_rate": 1.1659753046622899e-05, "loss": 1.3898, "step": 5260 }, { "epoch": 0.46, "learning_rate": 1.1656932495817457e-05, "loss": 1.4467, "step": 5261 }, { "epoch": 0.46, "learning_rate": 1.165411180946701e-05, "loss": 1.4199, "step": 5262 }, { "epoch": 0.46, "learning_rate": 1.165129098780231e-05, "loss": 1.3775, "step": 5263 }, { "epoch": 0.46, "learning_rate": 1.1648470031054108e-05, "loss": 1.2532, "step": 5264 }, { "epoch": 0.46, "learning_rate": 1.1645648939453173e-05, "loss": 1.3977, "step": 5265 }, { "epoch": 0.46, "learning_rate": 1.1642827713230285e-05, "loss": 1.5232, "step": 5266 }, { "epoch": 0.47, "learning_rate": 1.1640006352616235e-05, "loss": 1.5636, "step": 5267 }, { "epoch": 0.47, "learning_rate": 1.1637184857841822e-05, "loss": 1.4692, "step": 5268 }, { "epoch": 0.47, "learning_rate": 1.1634363229137857e-05, "loss": 1.5042, "step": 5269 }, { "epoch": 0.47, "learning_rate": 1.1631541466735162e-05, "loss": 1.4421, "step": 5270 }, { "epoch": 0.47, "learning_rate": 1.1628719570864573e-05, "loss": 1.4464, "step": 5271 }, { "epoch": 0.47, "learning_rate": 1.1625897541756937e-05, "loss": 1.514, "step": 5272 }, { "epoch": 0.47, "learning_rate": 1.1623075379643103e-05, "loss": 1.3778, "step": 5273 }, { "epoch": 0.47, "learning_rate": 1.1620253084753941e-05, "loss": 1.4333, "step": 5274 }, { "epoch": 0.47, "learning_rate": 1.1617430657320327e-05, "loss": 1.4329, "step": 5275 }, { "epoch": 0.47, "learning_rate": 1.1614608097573155e-05, "loss": 1.5279, "step": 5276 }, { "epoch": 0.47, "learning_rate": 1.1611785405743314e-05, "loss": 1.536, "step": 5277 }, { "epoch": 0.47, "learning_rate": 1.1608962582061718e-05, "loss": 1.3336, "step": 5278 }, { "epoch": 0.47, "learning_rate": 1.1606139626759289e-05, "loss": 1.4895, "step": 5279 }, { "epoch": 0.47, "learning_rate": 1.160331654006696e-05, "loss": 1.4462, "step": 5280 }, { "epoch": 0.47, "learning_rate": 1.160049332221567e-05, "loss": 1.4976, "step": 5281 }, { "epoch": 0.47, "learning_rate": 1.159766997343637e-05, "loss": 1.3836, "step": 5282 }, { "epoch": 0.47, "learning_rate": 1.1594846493960023e-05, "loss": 1.3737, "step": 5283 }, { "epoch": 0.47, "learning_rate": 1.1592022884017608e-05, "loss": 1.3262, "step": 5284 }, { "epoch": 0.47, "learning_rate": 1.1589199143840111e-05, "loss": 1.3946, "step": 5285 }, { "epoch": 0.47, "learning_rate": 1.1586375273658518e-05, "loss": 1.4201, "step": 5286 }, { "epoch": 0.47, "learning_rate": 1.1583551273703843e-05, "loss": 1.535, "step": 5287 }, { "epoch": 0.47, "learning_rate": 1.1580727144207103e-05, "loss": 1.515, "step": 5288 }, { "epoch": 0.47, "learning_rate": 1.1577902885399322e-05, "loss": 1.3441, "step": 5289 }, { "epoch": 0.47, "learning_rate": 1.1575078497511536e-05, "loss": 1.45, "step": 5290 }, { "epoch": 0.47, "learning_rate": 1.1572253980774797e-05, "loss": 1.5391, "step": 5291 }, { "epoch": 0.47, "learning_rate": 1.1569429335420164e-05, "loss": 1.3679, "step": 5292 }, { "epoch": 0.47, "learning_rate": 1.1566604561678707e-05, "loss": 1.4986, "step": 5293 }, { "epoch": 0.47, "learning_rate": 1.1563779659781502e-05, "loss": 1.3501, "step": 5294 }, { "epoch": 0.47, "learning_rate": 1.1563779659781502e-05, "loss": 1.3809, "step": 5295 }, { "epoch": 0.47, "learning_rate": 1.1560954629959644e-05, "loss": 1.4216, "step": 5296 }, { "epoch": 0.47, "learning_rate": 1.155812947244423e-05, "loss": 1.4375, "step": 5297 }, { "epoch": 0.47, "learning_rate": 1.1555304187466374e-05, "loss": 1.4382, "step": 5298 }, { "epoch": 0.47, "learning_rate": 1.1552478775257201e-05, "loss": 1.49, "step": 5299 }, { "epoch": 0.47, "learning_rate": 1.1549653236047835e-05, "loss": 1.4611, "step": 5300 }, { "epoch": 0.47, "learning_rate": 1.1546827570069426e-05, "loss": 1.4302, "step": 5301 }, { "epoch": 0.47, "learning_rate": 1.1544001777553123e-05, "loss": 1.4551, "step": 5302 }, { "epoch": 0.47, "learning_rate": 1.1541175858730092e-05, "loss": 1.4432, "step": 5303 }, { "epoch": 0.47, "learning_rate": 1.1538349813831502e-05, "loss": 1.4369, "step": 5304 }, { "epoch": 0.47, "learning_rate": 1.1535523643088541e-05, "loss": 1.4184, "step": 5305 }, { "epoch": 0.47, "learning_rate": 1.1532697346732404e-05, "loss": 1.4209, "step": 5306 }, { "epoch": 0.47, "learning_rate": 1.1529870924994296e-05, "loss": 1.5505, "step": 5307 }, { "epoch": 0.47, "learning_rate": 1.1527044378105425e-05, "loss": 1.4693, "step": 5308 }, { "epoch": 0.47, "learning_rate": 1.1524217706297022e-05, "loss": 1.4838, "step": 5309 }, { "epoch": 0.47, "learning_rate": 1.1521390909800324e-05, "loss": 1.3085, "step": 5310 }, { "epoch": 0.47, "learning_rate": 1.151856398884657e-05, "loss": 1.4825, "step": 5311 }, { "epoch": 0.47, "learning_rate": 1.1515736943667024e-05, "loss": 1.4702, "step": 5312 }, { "epoch": 0.47, "learning_rate": 1.1512909774492941e-05, "loss": 1.6136, "step": 5313 }, { "epoch": 0.47, "learning_rate": 1.1510082481555607e-05, "loss": 1.5142, "step": 5314 }, { "epoch": 0.47, "learning_rate": 1.1507255065086302e-05, "loss": 1.3952, "step": 5315 }, { "epoch": 0.47, "learning_rate": 1.1504427525316329e-05, "loss": 1.3209, "step": 5316 }, { "epoch": 0.47, "learning_rate": 1.1501599862476986e-05, "loss": 1.5504, "step": 5317 }, { "epoch": 0.47, "learning_rate": 1.1498772076799592e-05, "loss": 1.4591, "step": 5318 }, { "epoch": 0.47, "learning_rate": 1.1495944168515477e-05, "loss": 1.4987, "step": 5319 }, { "epoch": 0.47, "learning_rate": 1.1493116137855973e-05, "loss": 1.3577, "step": 5320 }, { "epoch": 0.47, "learning_rate": 1.1490287985052428e-05, "loss": 1.4357, "step": 5321 }, { "epoch": 0.47, "learning_rate": 1.14874597103362e-05, "loss": 1.4668, "step": 5322 }, { "epoch": 0.47, "learning_rate": 1.1484631313938652e-05, "loss": 1.4845, "step": 5323 }, { "epoch": 0.47, "learning_rate": 1.1481802796091164e-05, "loss": 1.4364, "step": 5324 }, { "epoch": 0.47, "learning_rate": 1.147897415702512e-05, "loss": 1.4365, "step": 5325 }, { "epoch": 0.47, "learning_rate": 1.1476145396971918e-05, "loss": 1.4379, "step": 5326 }, { "epoch": 0.47, "learning_rate": 1.147331651616296e-05, "loss": 1.3133, "step": 5327 }, { "epoch": 0.47, "learning_rate": 1.1470487514829667e-05, "loss": 1.4142, "step": 5328 }, { "epoch": 0.47, "learning_rate": 1.1467658393203465e-05, "loss": 1.4023, "step": 5329 }, { "epoch": 0.47, "learning_rate": 1.1464829151515783e-05, "loss": 1.4552, "step": 5330 }, { "epoch": 0.47, "learning_rate": 1.1461999789998072e-05, "loss": 1.4505, "step": 5331 }, { "epoch": 0.47, "learning_rate": 1.1459170308881789e-05, "loss": 1.329, "step": 5332 }, { "epoch": 0.47, "learning_rate": 1.1456340708398394e-05, "loss": 1.3881, "step": 5333 }, { "epoch": 0.47, "learning_rate": 1.1453510988779366e-05, "loss": 1.4028, "step": 5334 }, { "epoch": 0.47, "learning_rate": 1.1450681150256186e-05, "loss": 1.3811, "step": 5335 }, { "epoch": 0.47, "learning_rate": 1.144785119306035e-05, "loss": 1.4309, "step": 5336 }, { "epoch": 0.47, "learning_rate": 1.1445021117423365e-05, "loss": 1.5399, "step": 5337 }, { "epoch": 0.47, "learning_rate": 1.144219092357674e-05, "loss": 1.4859, "step": 5338 }, { "epoch": 0.47, "learning_rate": 1.1439360611751997e-05, "loss": 1.433, "step": 5339 }, { "epoch": 0.47, "learning_rate": 1.1436530182180676e-05, "loss": 1.3858, "step": 5340 }, { "epoch": 0.47, "learning_rate": 1.1433699635094315e-05, "loss": 1.3977, "step": 5341 }, { "epoch": 0.47, "learning_rate": 1.1430868970724469e-05, "loss": 1.4528, "step": 5342 }, { "epoch": 0.47, "learning_rate": 1.1428038189302697e-05, "loss": 1.457, "step": 5343 }, { "epoch": 0.47, "learning_rate": 1.1425207291060571e-05, "loss": 1.5134, "step": 5344 }, { "epoch": 0.47, "learning_rate": 1.1422376276229672e-05, "loss": 1.3295, "step": 5345 }, { "epoch": 0.47, "learning_rate": 1.1419545145041593e-05, "loss": 1.4826, "step": 5346 }, { "epoch": 0.47, "learning_rate": 1.1416713897727931e-05, "loss": 1.5081, "step": 5347 }, { "epoch": 0.47, "learning_rate": 1.1413882534520297e-05, "loss": 1.5032, "step": 5348 }, { "epoch": 0.47, "learning_rate": 1.1411051055650311e-05, "loss": 1.4166, "step": 5349 }, { "epoch": 0.47, "learning_rate": 1.1408219461349599e-05, "loss": 1.5191, "step": 5350 }, { "epoch": 0.47, "learning_rate": 1.14053877518498e-05, "loss": 1.4798, "step": 5351 }, { "epoch": 0.47, "learning_rate": 1.1402555927382562e-05, "loss": 1.3655, "step": 5352 }, { "epoch": 0.47, "learning_rate": 1.1399723988179543e-05, "loss": 1.4282, "step": 5353 }, { "epoch": 0.47, "learning_rate": 1.1396891934472405e-05, "loss": 1.3414, "step": 5354 }, { "epoch": 0.47, "learning_rate": 1.139405976649283e-05, "loss": 1.3485, "step": 5355 }, { "epoch": 0.47, "learning_rate": 1.1391227484472496e-05, "loss": 1.433, "step": 5356 }, { "epoch": 0.47, "learning_rate": 1.13883950886431e-05, "loss": 1.4101, "step": 5357 }, { "epoch": 0.47, "learning_rate": 1.1385562579236348e-05, "loss": 1.5087, "step": 5358 }, { "epoch": 0.47, "learning_rate": 1.138272995648395e-05, "loss": 1.3908, "step": 5359 }, { "epoch": 0.47, "learning_rate": 1.137989722061763e-05, "loss": 1.4612, "step": 5360 }, { "epoch": 0.47, "learning_rate": 1.1377064371869115e-05, "loss": 1.3269, "step": 5361 }, { "epoch": 0.47, "learning_rate": 1.137423141047015e-05, "loss": 1.3907, "step": 5362 }, { "epoch": 0.47, "learning_rate": 1.1371398336652486e-05, "loss": 1.4276, "step": 5363 }, { "epoch": 0.47, "learning_rate": 1.1368565150647877e-05, "loss": 1.4934, "step": 5364 }, { "epoch": 0.47, "learning_rate": 1.1365731852688092e-05, "loss": 1.4142, "step": 5365 }, { "epoch": 0.47, "learning_rate": 1.136289844300491e-05, "loss": 1.3913, "step": 5366 }, { "epoch": 0.47, "learning_rate": 1.136006492183012e-05, "loss": 1.4614, "step": 5367 }, { "epoch": 0.47, "learning_rate": 1.1357231289395517e-05, "loss": 1.3103, "step": 5368 }, { "epoch": 0.47, "learning_rate": 1.1354397545932897e-05, "loss": 1.5282, "step": 5369 }, { "epoch": 0.47, "learning_rate": 1.1351563691674086e-05, "loss": 1.3845, "step": 5370 }, { "epoch": 0.47, "learning_rate": 1.1348729726850898e-05, "loss": 1.5133, "step": 5371 }, { "epoch": 0.47, "learning_rate": 1.1345895651695171e-05, "loss": 1.4905, "step": 5372 }, { "epoch": 0.47, "learning_rate": 1.1343061466438738e-05, "loss": 1.3223, "step": 5373 }, { "epoch": 0.47, "learning_rate": 1.134022717131346e-05, "loss": 1.4485, "step": 5374 }, { "epoch": 0.47, "learning_rate": 1.1337392766551188e-05, "loss": 1.3892, "step": 5375 }, { "epoch": 0.47, "learning_rate": 1.1334558252383792e-05, "loss": 1.4402, "step": 5376 }, { "epoch": 0.47, "learning_rate": 1.1331723629043145e-05, "loss": 1.4846, "step": 5377 }, { "epoch": 0.47, "learning_rate": 1.1328888896761137e-05, "loss": 1.435, "step": 5378 }, { "epoch": 0.47, "learning_rate": 1.1326054055769666e-05, "loss": 1.5256, "step": 5379 }, { "epoch": 0.48, "learning_rate": 1.1323219106300626e-05, "loss": 1.4346, "step": 5380 }, { "epoch": 0.48, "learning_rate": 1.1320384048585942e-05, "loss": 1.4258, "step": 5381 }, { "epoch": 0.48, "learning_rate": 1.131754888285752e-05, "loss": 1.4871, "step": 5382 }, { "epoch": 0.48, "learning_rate": 1.1314713609347306e-05, "loss": 1.5172, "step": 5383 }, { "epoch": 0.48, "learning_rate": 1.1311878228287225e-05, "loss": 1.3188, "step": 5384 }, { "epoch": 0.48, "learning_rate": 1.1309042739909235e-05, "loss": 1.4241, "step": 5385 }, { "epoch": 0.48, "learning_rate": 1.1306207144445285e-05, "loss": 1.5039, "step": 5386 }, { "epoch": 0.48, "learning_rate": 1.130337144212735e-05, "loss": 1.3816, "step": 5387 }, { "epoch": 0.48, "learning_rate": 1.1300535633187391e-05, "loss": 1.466, "step": 5388 }, { "epoch": 0.48, "learning_rate": 1.1297699717857401e-05, "loss": 1.4758, "step": 5389 }, { "epoch": 0.48, "learning_rate": 1.1294863696369365e-05, "loss": 1.3794, "step": 5390 }, { "epoch": 0.48, "learning_rate": 1.1292027568955287e-05, "loss": 1.331, "step": 5391 }, { "epoch": 0.48, "learning_rate": 1.1289191335847178e-05, "loss": 1.5547, "step": 5392 }, { "epoch": 0.48, "learning_rate": 1.128635499727705e-05, "loss": 1.3802, "step": 5393 }, { "epoch": 0.48, "learning_rate": 1.128351855347693e-05, "loss": 1.4208, "step": 5394 }, { "epoch": 0.48, "learning_rate": 1.1280682004678853e-05, "loss": 1.437, "step": 5395 }, { "epoch": 0.48, "learning_rate": 1.127784535111487e-05, "loss": 1.4452, "step": 5396 }, { "epoch": 0.48, "learning_rate": 1.1275008593017018e-05, "loss": 1.3933, "step": 5397 }, { "epoch": 0.48, "learning_rate": 1.1272171730617372e-05, "loss": 1.3805, "step": 5398 }, { "epoch": 0.48, "learning_rate": 1.1269334764147987e-05, "loss": 1.4094, "step": 5399 }, { "epoch": 0.48, "learning_rate": 1.1266497693840957e-05, "loss": 1.3362, "step": 5400 }, { "epoch": 0.48, "learning_rate": 1.1263660519928351e-05, "loss": 1.4789, "step": 5401 }, { "epoch": 0.48, "learning_rate": 1.1260823242642276e-05, "loss": 1.3767, "step": 5402 }, { "epoch": 0.48, "learning_rate": 1.1257985862214826e-05, "loss": 1.4017, "step": 5403 }, { "epoch": 0.48, "learning_rate": 1.125514837887812e-05, "loss": 1.5021, "step": 5404 }, { "epoch": 0.48, "learning_rate": 1.1252310792864272e-05, "loss": 1.401, "step": 5405 }, { "epoch": 0.48, "learning_rate": 1.124947310440541e-05, "loss": 1.4099, "step": 5406 }, { "epoch": 0.48, "learning_rate": 1.1246635313733674e-05, "loss": 1.4383, "step": 5407 }, { "epoch": 0.48, "learning_rate": 1.1243797421081209e-05, "loss": 1.3863, "step": 5408 }, { "epoch": 0.48, "learning_rate": 1.1240959426680166e-05, "loss": 1.4872, "step": 5409 }, { "epoch": 0.48, "learning_rate": 1.1238121330762706e-05, "loss": 1.3614, "step": 5410 }, { "epoch": 0.48, "learning_rate": 1.1235283133561e-05, "loss": 1.2973, "step": 5411 }, { "epoch": 0.48, "learning_rate": 1.1232444835307226e-05, "loss": 1.3685, "step": 5412 }, { "epoch": 0.48, "learning_rate": 1.1229606436233574e-05, "loss": 1.4735, "step": 5413 }, { "epoch": 0.48, "learning_rate": 1.122676793657223e-05, "loss": 1.4487, "step": 5414 }, { "epoch": 0.48, "learning_rate": 1.1223929336555406e-05, "loss": 1.5284, "step": 5415 }, { "epoch": 0.48, "learning_rate": 1.1221090636415304e-05, "loss": 1.3857, "step": 5416 }, { "epoch": 0.48, "learning_rate": 1.121825183638415e-05, "loss": 1.4442, "step": 5417 }, { "epoch": 0.48, "learning_rate": 1.1215412936694172e-05, "loss": 1.3723, "step": 5418 }, { "epoch": 0.48, "learning_rate": 1.12125739375776e-05, "loss": 1.4791, "step": 5419 }, { "epoch": 0.48, "learning_rate": 1.1209734839266682e-05, "loss": 1.4206, "step": 5420 }, { "epoch": 0.48, "learning_rate": 1.120689564199367e-05, "loss": 1.3747, "step": 5421 }, { "epoch": 0.48, "learning_rate": 1.120405634599082e-05, "loss": 1.4545, "step": 5422 }, { "epoch": 0.48, "learning_rate": 1.1201216951490406e-05, "loss": 1.3805, "step": 5423 }, { "epoch": 0.48, "learning_rate": 1.1198377458724699e-05, "loss": 1.4614, "step": 5424 }, { "epoch": 0.48, "learning_rate": 1.1195537867925983e-05, "loss": 1.4739, "step": 5425 }, { "epoch": 0.48, "learning_rate": 1.1192698179326555e-05, "loss": 1.5025, "step": 5426 }, { "epoch": 0.48, "learning_rate": 1.118985839315871e-05, "loss": 1.5012, "step": 5427 }, { "epoch": 0.48, "learning_rate": 1.1187018509654756e-05, "loss": 1.3907, "step": 5428 }, { "epoch": 0.48, "learning_rate": 1.1184178529047014e-05, "loss": 1.4414, "step": 5429 }, { "epoch": 0.48, "learning_rate": 1.1181338451567804e-05, "loss": 1.4599, "step": 5430 }, { "epoch": 0.48, "learning_rate": 1.1178498277449461e-05, "loss": 1.3457, "step": 5431 }, { "epoch": 0.48, "learning_rate": 1.1175658006924319e-05, "loss": 1.4105, "step": 5432 }, { "epoch": 0.48, "learning_rate": 1.1172817640224731e-05, "loss": 1.3208, "step": 5433 }, { "epoch": 0.48, "learning_rate": 1.1169977177583053e-05, "loss": 1.3146, "step": 5434 }, { "epoch": 0.48, "learning_rate": 1.1167136619231646e-05, "loss": 1.435, "step": 5435 }, { "epoch": 0.48, "learning_rate": 1.116429596540288e-05, "loss": 1.4138, "step": 5436 }, { "epoch": 0.48, "learning_rate": 1.1161455216329138e-05, "loss": 1.3687, "step": 5437 }, { "epoch": 0.48, "learning_rate": 1.1158614372242801e-05, "loss": 1.3969, "step": 5438 }, { "epoch": 0.48, "learning_rate": 1.1155773433376273e-05, "loss": 1.4026, "step": 5439 }, { "epoch": 0.48, "learning_rate": 1.1152932399961946e-05, "loss": 1.4705, "step": 5440 }, { "epoch": 0.48, "learning_rate": 1.1150091272232238e-05, "loss": 1.4018, "step": 5441 }, { "epoch": 0.48, "learning_rate": 1.1147250050419562e-05, "loss": 1.3176, "step": 5442 }, { "epoch": 0.48, "learning_rate": 1.1144408734756346e-05, "loss": 1.321, "step": 5443 }, { "epoch": 0.48, "learning_rate": 1.1141567325475022e-05, "loss": 1.2633, "step": 5444 }, { "epoch": 0.48, "learning_rate": 1.113872582280803e-05, "loss": 1.4241, "step": 5445 }, { "epoch": 0.48, "learning_rate": 1.113588422698782e-05, "loss": 1.364, "step": 5446 }, { "epoch": 0.48, "learning_rate": 1.113304253824685e-05, "loss": 1.3773, "step": 5447 }, { "epoch": 0.48, "learning_rate": 1.1130200756817581e-05, "loss": 1.4129, "step": 5448 }, { "epoch": 0.48, "learning_rate": 1.1127358882932483e-05, "loss": 1.4012, "step": 5449 }, { "epoch": 0.48, "learning_rate": 1.112451691682404e-05, "loss": 1.308, "step": 5450 }, { "epoch": 0.48, "learning_rate": 1.1121674858724734e-05, "loss": 1.4991, "step": 5451 }, { "epoch": 0.48, "learning_rate": 1.1118832708867061e-05, "loss": 1.4431, "step": 5452 }, { "epoch": 0.48, "learning_rate": 1.1115990467483519e-05, "loss": 1.4151, "step": 5453 }, { "epoch": 0.48, "learning_rate": 1.1113148134806622e-05, "loss": 1.5053, "step": 5454 }, { "epoch": 0.48, "learning_rate": 1.1110305711068883e-05, "loss": 1.4152, "step": 5455 }, { "epoch": 0.48, "learning_rate": 1.1107463196502828e-05, "loss": 1.4881, "step": 5456 }, { "epoch": 0.48, "learning_rate": 1.1104620591340987e-05, "loss": 1.4972, "step": 5457 }, { "epoch": 0.48, "learning_rate": 1.1101777895815895e-05, "loss": 1.4229, "step": 5458 }, { "epoch": 0.48, "learning_rate": 1.1098935110160105e-05, "loss": 1.5161, "step": 5459 }, { "epoch": 0.48, "learning_rate": 1.1096092234606168e-05, "loss": 1.4475, "step": 5460 }, { "epoch": 0.48, "learning_rate": 1.1093249269386645e-05, "loss": 1.3873, "step": 5461 }, { "epoch": 0.48, "learning_rate": 1.1090406214734102e-05, "loss": 1.3287, "step": 5462 }, { "epoch": 0.48, "learning_rate": 1.1087563070881115e-05, "loss": 1.4253, "step": 5463 }, { "epoch": 0.48, "learning_rate": 1.1084719838060267e-05, "loss": 1.419, "step": 5464 }, { "epoch": 0.48, "learning_rate": 1.1081876516504153e-05, "loss": 1.4524, "step": 5465 }, { "epoch": 0.48, "learning_rate": 1.1079033106445362e-05, "loss": 1.5149, "step": 5466 }, { "epoch": 0.48, "learning_rate": 1.1076189608116504e-05, "loss": 1.4858, "step": 5467 }, { "epoch": 0.48, "learning_rate": 1.1073346021750187e-05, "loss": 1.5088, "step": 5468 }, { "epoch": 0.48, "learning_rate": 1.1070502347579035e-05, "loss": 1.4203, "step": 5469 }, { "epoch": 0.48, "learning_rate": 1.106765858583567e-05, "loss": 1.4015, "step": 5470 }, { "epoch": 0.48, "learning_rate": 1.1064814736752726e-05, "loss": 1.4458, "step": 5471 }, { "epoch": 0.48, "learning_rate": 1.1061970800562844e-05, "loss": 1.4935, "step": 5472 }, { "epoch": 0.48, "learning_rate": 1.1059126777498674e-05, "loss": 1.3722, "step": 5473 }, { "epoch": 0.48, "learning_rate": 1.1056282667792868e-05, "loss": 1.5118, "step": 5474 }, { "epoch": 0.48, "learning_rate": 1.1053438471678087e-05, "loss": 1.4154, "step": 5475 }, { "epoch": 0.48, "learning_rate": 1.1050594189387e-05, "loss": 1.5874, "step": 5476 }, { "epoch": 0.48, "learning_rate": 1.1047749821152284e-05, "loss": 1.3703, "step": 5477 }, { "epoch": 0.48, "learning_rate": 1.1044905367206626e-05, "loss": 1.4205, "step": 5478 }, { "epoch": 0.48, "learning_rate": 1.1042060827782708e-05, "loss": 1.5749, "step": 5479 }, { "epoch": 0.48, "learning_rate": 1.1039216203113232e-05, "loss": 1.3447, "step": 5480 }, { "epoch": 0.48, "learning_rate": 1.1036371493430899e-05, "loss": 1.4797, "step": 5481 }, { "epoch": 0.48, "learning_rate": 1.1033526698968425e-05, "loss": 1.57, "step": 5482 }, { "epoch": 0.48, "learning_rate": 1.1030681819958522e-05, "loss": 1.3817, "step": 5483 }, { "epoch": 0.48, "learning_rate": 1.1027836856633915e-05, "loss": 1.2927, "step": 5484 }, { "epoch": 0.48, "learning_rate": 1.1024991809227338e-05, "loss": 1.4653, "step": 5485 }, { "epoch": 0.48, "learning_rate": 1.102214667797153e-05, "loss": 1.4734, "step": 5486 }, { "epoch": 0.48, "learning_rate": 1.1019301463099239e-05, "loss": 1.4326, "step": 5487 }, { "epoch": 0.48, "learning_rate": 1.101645616484321e-05, "loss": 1.5281, "step": 5488 }, { "epoch": 0.48, "learning_rate": 1.1013610783436204e-05, "loss": 1.3428, "step": 5489 }, { "epoch": 0.48, "learning_rate": 1.101076531911099e-05, "loss": 1.549, "step": 5490 }, { "epoch": 0.48, "learning_rate": 1.1007919772100342e-05, "loss": 1.4034, "step": 5491 }, { "epoch": 0.48, "learning_rate": 1.1005074142637033e-05, "loss": 1.5163, "step": 5492 }, { "epoch": 0.49, "learning_rate": 1.1002228430953853e-05, "loss": 1.4221, "step": 5493 }, { "epoch": 0.49, "learning_rate": 1.0999382637283594e-05, "loss": 1.4983, "step": 5494 }, { "epoch": 0.49, "learning_rate": 1.0996536761859055e-05, "loss": 1.3614, "step": 5495 }, { "epoch": 0.49, "learning_rate": 1.0993690804913043e-05, "loss": 1.2632, "step": 5496 }, { "epoch": 0.49, "learning_rate": 1.0990844766678372e-05, "loss": 1.2543, "step": 5497 }, { "epoch": 0.49, "learning_rate": 1.0987998647387861e-05, "loss": 1.503, "step": 5498 }, { "epoch": 0.49, "learning_rate": 1.0985152447274335e-05, "loss": 1.4204, "step": 5499 }, { "epoch": 0.49, "learning_rate": 1.098230616657063e-05, "loss": 1.4162, "step": 5500 }, { "epoch": 0.49, "learning_rate": 1.0979459805509578e-05, "loss": 1.4263, "step": 5501 }, { "epoch": 0.49, "learning_rate": 1.097661336432403e-05, "loss": 1.4433, "step": 5502 }, { "epoch": 0.49, "learning_rate": 1.097376684324684e-05, "loss": 1.4493, "step": 5503 }, { "epoch": 0.49, "learning_rate": 1.0970920242510867e-05, "loss": 1.4584, "step": 5504 }, { "epoch": 0.49, "learning_rate": 1.0968073562348974e-05, "loss": 1.3849, "step": 5505 }, { "epoch": 0.49, "learning_rate": 1.0965226802994031e-05, "loss": 1.3705, "step": 5506 }, { "epoch": 0.49, "learning_rate": 1.0962379964678921e-05, "loss": 1.4122, "step": 5507 }, { "epoch": 0.49, "learning_rate": 1.095953304763653e-05, "loss": 1.4223, "step": 5508 }, { "epoch": 0.49, "learning_rate": 1.0956686052099743e-05, "loss": 1.4711, "step": 5509 }, { "epoch": 0.49, "learning_rate": 1.0953838978301464e-05, "loss": 1.4577, "step": 5510 }, { "epoch": 0.49, "learning_rate": 1.0950991826474593e-05, "loss": 1.4823, "step": 5511 }, { "epoch": 0.49, "learning_rate": 1.0948144596852044e-05, "loss": 1.4043, "step": 5512 }, { "epoch": 0.49, "learning_rate": 1.0945297289666732e-05, "loss": 1.3732, "step": 5513 }, { "epoch": 0.49, "learning_rate": 1.094244990515158e-05, "loss": 1.4063, "step": 5514 }, { "epoch": 0.49, "learning_rate": 1.0939602443539521e-05, "loss": 1.4133, "step": 5515 }, { "epoch": 0.49, "learning_rate": 1.0936754905063485e-05, "loss": 1.42, "step": 5516 }, { "epoch": 0.49, "learning_rate": 1.0933907289956423e-05, "loss": 1.4527, "step": 5517 }, { "epoch": 0.49, "learning_rate": 1.0931059598451275e-05, "loss": 1.4078, "step": 5518 }, { "epoch": 0.49, "learning_rate": 1.0928211830780997e-05, "loss": 1.347, "step": 5519 }, { "epoch": 0.49, "learning_rate": 1.0925363987178554e-05, "loss": 1.3667, "step": 5520 }, { "epoch": 0.49, "learning_rate": 1.0922516067876914e-05, "loss": 1.4259, "step": 5521 }, { "epoch": 0.49, "learning_rate": 1.0919668073109044e-05, "loss": 1.4563, "step": 5522 }, { "epoch": 0.49, "learning_rate": 1.0916820003107928e-05, "loss": 1.4492, "step": 5523 }, { "epoch": 0.49, "learning_rate": 1.0913971858106551e-05, "loss": 1.4982, "step": 5524 }, { "epoch": 0.49, "learning_rate": 1.0911123638337906e-05, "loss": 1.357, "step": 5525 }, { "epoch": 0.49, "learning_rate": 1.0908275344034993e-05, "loss": 1.4002, "step": 5526 }, { "epoch": 0.49, "learning_rate": 1.0905426975430806e-05, "loss": 1.5112, "step": 5527 }, { "epoch": 0.49, "learning_rate": 1.0902578532758368e-05, "loss": 1.4383, "step": 5528 }, { "epoch": 0.49, "learning_rate": 1.0899730016250687e-05, "loss": 1.462, "step": 5529 }, { "epoch": 0.49, "learning_rate": 1.089688142614079e-05, "loss": 1.4485, "step": 5530 }, { "epoch": 0.49, "learning_rate": 1.08940327626617e-05, "loss": 1.458, "step": 5531 }, { "epoch": 0.49, "learning_rate": 1.0891184026046458e-05, "loss": 1.4553, "step": 5532 }, { "epoch": 0.49, "learning_rate": 1.0888335216528099e-05, "loss": 1.509, "step": 5533 }, { "epoch": 0.49, "learning_rate": 1.088548633433967e-05, "loss": 1.4209, "step": 5534 }, { "epoch": 0.49, "learning_rate": 1.0882637379714222e-05, "loss": 1.395, "step": 5535 }, { "epoch": 0.49, "learning_rate": 1.087978835288482e-05, "loss": 1.2942, "step": 5536 }, { "epoch": 0.49, "learning_rate": 1.087693925408452e-05, "loss": 1.4094, "step": 5537 }, { "epoch": 0.49, "learning_rate": 1.0874090083546396e-05, "loss": 1.5132, "step": 5538 }, { "epoch": 0.49, "learning_rate": 1.0871240841503526e-05, "loss": 1.4457, "step": 5539 }, { "epoch": 0.49, "learning_rate": 1.0868391528188981e-05, "loss": 1.308, "step": 5540 }, { "epoch": 0.49, "learning_rate": 1.0865542143835864e-05, "loss": 1.3272, "step": 5541 }, { "epoch": 0.49, "learning_rate": 1.0862692688677254e-05, "loss": 1.5571, "step": 5542 }, { "epoch": 0.49, "learning_rate": 1.0859843162946262e-05, "loss": 1.3642, "step": 5543 }, { "epoch": 0.49, "learning_rate": 1.085699356687598e-05, "loss": 1.3177, "step": 5544 }, { "epoch": 0.49, "learning_rate": 1.0854143900699533e-05, "loss": 1.4816, "step": 5545 }, { "epoch": 0.49, "learning_rate": 1.0851294164650026e-05, "loss": 1.4321, "step": 5546 }, { "epoch": 0.49, "learning_rate": 1.0848444358960586e-05, "loss": 1.4413, "step": 5547 }, { "epoch": 0.49, "learning_rate": 1.0845594483864337e-05, "loss": 1.5014, "step": 5548 }, { "epoch": 0.49, "learning_rate": 1.084274453959442e-05, "loss": 1.5259, "step": 5549 }, { "epoch": 0.49, "learning_rate": 1.0839894526383968e-05, "loss": 1.4279, "step": 5550 }, { "epoch": 0.49, "learning_rate": 1.0837044444466124e-05, "loss": 1.5055, "step": 5551 }, { "epoch": 0.49, "learning_rate": 1.0834194294074043e-05, "loss": 1.3929, "step": 5552 }, { "epoch": 0.49, "learning_rate": 1.0831344075440879e-05, "loss": 1.4429, "step": 5553 }, { "epoch": 0.49, "learning_rate": 1.0828493788799797e-05, "loss": 1.5095, "step": 5554 }, { "epoch": 0.49, "learning_rate": 1.0825643434383957e-05, "loss": 1.4435, "step": 5555 }, { "epoch": 0.49, "learning_rate": 1.0822793012426537e-05, "loss": 1.4243, "step": 5556 }, { "epoch": 0.49, "learning_rate": 1.0819942523160714e-05, "loss": 1.3971, "step": 5557 }, { "epoch": 0.49, "learning_rate": 1.0817091966819671e-05, "loss": 1.4445, "step": 5558 }, { "epoch": 0.49, "learning_rate": 1.0814241343636596e-05, "loss": 1.4525, "step": 5559 }, { "epoch": 0.49, "learning_rate": 1.0811390653844689e-05, "loss": 1.4076, "step": 5560 }, { "epoch": 0.49, "learning_rate": 1.0808539897677139e-05, "loss": 1.505, "step": 5561 }, { "epoch": 0.49, "learning_rate": 1.0805689075367164e-05, "loss": 1.3836, "step": 5562 }, { "epoch": 0.49, "learning_rate": 1.080283818714797e-05, "loss": 1.45, "step": 5563 }, { "epoch": 0.49, "learning_rate": 1.0799987233252773e-05, "loss": 1.4761, "step": 5564 }, { "epoch": 0.49, "learning_rate": 1.079713621391479e-05, "loss": 1.4375, "step": 5565 }, { "epoch": 0.49, "learning_rate": 1.0794285129367252e-05, "loss": 1.4278, "step": 5566 }, { "epoch": 0.49, "learning_rate": 1.0791433979843397e-05, "loss": 1.4252, "step": 5567 }, { "epoch": 0.49, "learning_rate": 1.0788582765576456e-05, "loss": 1.461, "step": 5568 }, { "epoch": 0.49, "learning_rate": 1.0785731486799671e-05, "loss": 1.4812, "step": 5569 }, { "epoch": 0.49, "learning_rate": 1.0782880143746296e-05, "loss": 1.4959, "step": 5570 }, { "epoch": 0.49, "learning_rate": 1.0780028736649579e-05, "loss": 1.4014, "step": 5571 }, { "epoch": 0.49, "learning_rate": 1.0777177265742782e-05, "loss": 1.3891, "step": 5572 }, { "epoch": 0.49, "learning_rate": 1.0774325731259166e-05, "loss": 1.3128, "step": 5573 }, { "epoch": 0.49, "learning_rate": 1.0771474133432003e-05, "loss": 1.5162, "step": 5574 }, { "epoch": 0.49, "learning_rate": 1.0768622472494567e-05, "loss": 1.4152, "step": 5575 }, { "epoch": 0.49, "learning_rate": 1.0765770748680137e-05, "loss": 1.5019, "step": 5576 }, { "epoch": 0.49, "learning_rate": 1.0762918962222e-05, "loss": 1.4793, "step": 5577 }, { "epoch": 0.49, "learning_rate": 1.0760067113353436e-05, "loss": 1.3648, "step": 5578 }, { "epoch": 0.49, "learning_rate": 1.0757215202307752e-05, "loss": 1.3349, "step": 5579 }, { "epoch": 0.49, "learning_rate": 1.0754363229318243e-05, "loss": 1.4518, "step": 5580 }, { "epoch": 0.49, "learning_rate": 1.0751511194618214e-05, "loss": 1.452, "step": 5581 }, { "epoch": 0.49, "learning_rate": 1.0748659098440975e-05, "loss": 1.4048, "step": 5582 }, { "epoch": 0.49, "learning_rate": 1.0745806941019841e-05, "loss": 1.4021, "step": 5583 }, { "epoch": 0.49, "learning_rate": 1.0742954722588133e-05, "loss": 1.3317, "step": 5584 }, { "epoch": 0.49, "learning_rate": 1.0740102443379174e-05, "loss": 1.3616, "step": 5585 }, { "epoch": 0.49, "learning_rate": 1.0737250103626298e-05, "loss": 1.5036, "step": 5586 }, { "epoch": 0.49, "learning_rate": 1.0734397703562836e-05, "loss": 1.3777, "step": 5587 }, { "epoch": 0.49, "learning_rate": 1.073154524342213e-05, "loss": 1.3284, "step": 5588 }, { "epoch": 0.49, "learning_rate": 1.0728692723437526e-05, "loss": 1.4383, "step": 5589 }, { "epoch": 0.49, "learning_rate": 1.072584014384237e-05, "loss": 1.4939, "step": 5590 }, { "epoch": 0.49, "learning_rate": 1.0722987504870019e-05, "loss": 1.3744, "step": 5591 }, { "epoch": 0.49, "learning_rate": 1.0720134806753833e-05, "loss": 1.5001, "step": 5592 }, { "epoch": 0.49, "learning_rate": 1.0717282049727177e-05, "loss": 1.3934, "step": 5593 }, { "epoch": 0.49, "learning_rate": 1.0714429234023417e-05, "loss": 1.5632, "step": 5594 }, { "epoch": 0.49, "learning_rate": 1.071157635987593e-05, "loss": 1.2733, "step": 5595 }, { "epoch": 0.49, "learning_rate": 1.0708723427518095e-05, "loss": 1.4336, "step": 5596 }, { "epoch": 0.49, "learning_rate": 1.0705870437183294e-05, "loss": 1.3258, "step": 5597 }, { "epoch": 0.49, "learning_rate": 1.0703017389104915e-05, "loss": 1.504, "step": 5598 }, { "epoch": 0.49, "learning_rate": 1.070016428351635e-05, "loss": 1.538, "step": 5599 }, { "epoch": 0.49, "learning_rate": 1.0697311120650997e-05, "loss": 1.385, "step": 5600 }, { "epoch": 0.49, "step": 5600, "total_flos": 4.871701069844972e+18, "train_loss": 0.0, "train_runtime": 988.1616, "train_samples_per_second": 0.001, "train_steps_per_second": 0.001 } ], "logging_steps": 1.0, "max_steps": 1, "num_train_epochs": 1, "save_steps": 100, "total_flos": 4.871701069844972e+18, "trial_name": null, "trial_params": null }