{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9999615399407715, "global_step": 3250, "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": 1.8412, "step": 1 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 2.4423, "step": 2 }, { "epoch": 0.0, "learning_rate": 2.0408163265306121e-07, "loss": 2.1726, "step": 3 }, { "epoch": 0.0, "learning_rate": 4.0816326530612243e-07, "loss": 2.6872, "step": 4 }, { "epoch": 0.0, "learning_rate": 6.122448979591837e-07, "loss": 2.1498, "step": 5 }, { "epoch": 0.0, "learning_rate": 8.163265306122449e-07, "loss": 2.6042, "step": 6 }, { "epoch": 0.0, "learning_rate": 1.0204081632653063e-06, "loss": 2.3096, "step": 7 }, { "epoch": 0.0, "learning_rate": 1.2244897959183673e-06, "loss": 2.4673, "step": 8 }, { "epoch": 0.0, "learning_rate": 1.4285714285714286e-06, "loss": 2.2869, "step": 9 }, { "epoch": 0.0, "learning_rate": 1.6326530612244897e-06, "loss": 2.2953, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.8367346938775512e-06, "loss": 1.9851, "step": 11 }, { "epoch": 0.0, "learning_rate": 2.0408163265306125e-06, "loss": 1.9965, "step": 12 }, { "epoch": 0.0, "learning_rate": 2.244897959183674e-06, "loss": 1.7586, "step": 13 }, { "epoch": 0.0, "learning_rate": 2.4489795918367347e-06, "loss": 2.4944, "step": 14 }, { "epoch": 0.0, "learning_rate": 2.6530612244897964e-06, "loss": 2.0767, "step": 15 }, { "epoch": 0.0, "learning_rate": 2.8571428571428573e-06, "loss": 2.6923, "step": 16 }, { "epoch": 0.01, "learning_rate": 3.0612244897959185e-06, "loss": 2.099, "step": 17 }, { "epoch": 0.01, "learning_rate": 3.2653061224489794e-06, "loss": 1.956, "step": 18 }, { "epoch": 0.01, "learning_rate": 3.469387755102041e-06, "loss": 1.7529, "step": 19 }, { "epoch": 0.01, "learning_rate": 3.6734693877551024e-06, "loss": 1.487, "step": 20 }, { "epoch": 0.01, "learning_rate": 3.877551020408164e-06, "loss": 1.518, "step": 21 }, { "epoch": 0.01, "learning_rate": 4.081632653061225e-06, "loss": 1.447, "step": 22 }, { "epoch": 0.01, "learning_rate": 4.2857142857142855e-06, "loss": 1.7908, "step": 23 }, { "epoch": 0.01, "learning_rate": 4.489795918367348e-06, "loss": 1.5428, "step": 24 }, { "epoch": 0.01, "learning_rate": 4.693877551020409e-06, "loss": 1.4308, "step": 25 }, { "epoch": 0.01, "learning_rate": 4.897959183673469e-06, "loss": 1.3911, "step": 26 }, { "epoch": 0.01, "learning_rate": 5.1020408163265315e-06, "loss": 1.6423, "step": 27 }, { "epoch": 0.01, "learning_rate": 5.306122448979593e-06, "loss": 1.6353, "step": 28 }, { "epoch": 0.01, "learning_rate": 5.510204081632653e-06, "loss": 1.9413, "step": 29 }, { "epoch": 0.01, "learning_rate": 5.7142857142857145e-06, "loss": 1.5212, "step": 30 }, { "epoch": 0.01, "learning_rate": 5.918367346938776e-06, "loss": 1.5408, "step": 31 }, { "epoch": 0.01, "learning_rate": 6.122448979591837e-06, "loss": 1.6166, "step": 32 }, { "epoch": 0.01, "learning_rate": 6.326530612244899e-06, "loss": 1.8031, "step": 33 }, { "epoch": 0.01, "learning_rate": 6.530612244897959e-06, "loss": 1.934, "step": 34 }, { "epoch": 0.01, "learning_rate": 6.734693877551021e-06, "loss": 1.4541, "step": 35 }, { "epoch": 0.01, "learning_rate": 6.938775510204082e-06, "loss": 1.5663, "step": 36 }, { "epoch": 0.01, "learning_rate": 7.1428571428571436e-06, "loss": 1.4314, "step": 37 }, { "epoch": 0.01, "learning_rate": 7.346938775510205e-06, "loss": 1.4307, "step": 38 }, { "epoch": 0.01, "learning_rate": 7.551020408163265e-06, "loss": 1.7589, "step": 39 }, { "epoch": 0.01, "learning_rate": 7.755102040816327e-06, "loss": 1.7896, "step": 40 }, { "epoch": 0.01, "learning_rate": 7.959183673469388e-06, "loss": 1.4308, "step": 41 }, { "epoch": 0.01, "learning_rate": 8.16326530612245e-06, "loss": 1.7288, "step": 42 }, { "epoch": 0.01, "learning_rate": 8.36734693877551e-06, "loss": 1.6473, "step": 43 }, { "epoch": 0.01, "learning_rate": 8.571428571428571e-06, "loss": 1.757, "step": 44 }, { "epoch": 0.01, "learning_rate": 8.775510204081633e-06, "loss": 1.369, "step": 45 }, { "epoch": 0.01, "learning_rate": 8.979591836734695e-06, "loss": 1.827, "step": 46 }, { "epoch": 0.01, "learning_rate": 9.183673469387756e-06, "loss": 2.2143, "step": 47 }, { "epoch": 0.01, "learning_rate": 9.387755102040818e-06, "loss": 1.781, "step": 48 }, { "epoch": 0.02, "learning_rate": 9.591836734693878e-06, "loss": 1.6623, "step": 49 }, { "epoch": 0.02, "learning_rate": 9.795918367346939e-06, "loss": 1.6225, "step": 50 }, { "epoch": 0.02, "learning_rate": 1e-05, "loss": 1.7036, "step": 51 }, { "epoch": 0.02, "learning_rate": 1.0204081632653063e-05, "loss": 1.4179, "step": 52 }, { "epoch": 0.02, "learning_rate": 1.0408163265306123e-05, "loss": 1.4681, "step": 53 }, { "epoch": 0.02, "learning_rate": 1.0612244897959186e-05, "loss": 1.4616, "step": 54 }, { "epoch": 0.02, "learning_rate": 1.0816326530612246e-05, "loss": 1.6009, "step": 55 }, { "epoch": 0.02, "learning_rate": 1.1020408163265306e-05, "loss": 1.2498, "step": 56 }, { "epoch": 0.02, "learning_rate": 1.1224489795918367e-05, "loss": 1.7834, "step": 57 }, { "epoch": 0.02, "learning_rate": 1.1428571428571429e-05, "loss": 1.5594, "step": 58 }, { "epoch": 0.02, "learning_rate": 1.1632653061224491e-05, "loss": 1.7175, "step": 59 }, { "epoch": 0.02, "learning_rate": 1.1836734693877552e-05, "loss": 1.5544, "step": 60 }, { "epoch": 0.02, "learning_rate": 1.2040816326530614e-05, "loss": 2.25, "step": 61 }, { "epoch": 0.02, "learning_rate": 1.2244897959183674e-05, "loss": 1.5575, "step": 62 }, { "epoch": 0.02, "learning_rate": 1.2448979591836736e-05, "loss": 1.6405, "step": 63 }, { "epoch": 0.02, "learning_rate": 1.2653061224489798e-05, "loss": 1.5264, "step": 64 }, { "epoch": 0.02, "learning_rate": 1.2857142857142859e-05, "loss": 1.2849, "step": 65 }, { "epoch": 0.02, "learning_rate": 1.3061224489795918e-05, "loss": 1.5249, "step": 66 }, { "epoch": 0.02, "learning_rate": 1.326530612244898e-05, "loss": 1.5535, "step": 67 }, { "epoch": 0.02, "learning_rate": 1.3469387755102042e-05, "loss": 1.6305, "step": 68 }, { "epoch": 0.02, "learning_rate": 1.3673469387755102e-05, "loss": 1.4558, "step": 69 }, { "epoch": 0.02, "learning_rate": 1.3877551020408165e-05, "loss": 1.1902, "step": 70 }, { "epoch": 0.02, "learning_rate": 1.4081632653061225e-05, "loss": 1.3432, "step": 71 }, { "epoch": 0.02, "learning_rate": 1.4285714285714287e-05, "loss": 1.593, "step": 72 }, { "epoch": 0.02, "learning_rate": 1.448979591836735e-05, "loss": 1.5448, "step": 73 }, { "epoch": 0.02, "learning_rate": 1.469387755102041e-05, "loss": 1.5227, "step": 74 }, { "epoch": 0.02, "learning_rate": 1.4897959183673472e-05, "loss": 1.8562, "step": 75 }, { "epoch": 0.02, "learning_rate": 1.510204081632653e-05, "loss": 1.7151, "step": 76 }, { "epoch": 0.02, "learning_rate": 1.530612244897959e-05, "loss": 1.5046, "step": 77 }, { "epoch": 0.02, "learning_rate": 1.5510204081632655e-05, "loss": 1.6402, "step": 78 }, { "epoch": 0.02, "learning_rate": 1.5714285714285715e-05, "loss": 1.5564, "step": 79 }, { "epoch": 0.02, "learning_rate": 1.5918367346938776e-05, "loss": 1.8259, "step": 80 }, { "epoch": 0.02, "learning_rate": 1.612244897959184e-05, "loss": 1.6569, "step": 81 }, { "epoch": 0.03, "learning_rate": 1.63265306122449e-05, "loss": 1.5577, "step": 82 }, { "epoch": 0.03, "learning_rate": 1.653061224489796e-05, "loss": 1.7821, "step": 83 }, { "epoch": 0.03, "learning_rate": 1.673469387755102e-05, "loss": 1.4537, "step": 84 }, { "epoch": 0.03, "learning_rate": 1.6938775510204085e-05, "loss": 2.3595, "step": 85 }, { "epoch": 0.03, "learning_rate": 1.7142857142857142e-05, "loss": 1.7363, "step": 86 }, { "epoch": 0.03, "learning_rate": 1.7346938775510206e-05, "loss": 1.6054, "step": 87 }, { "epoch": 0.03, "learning_rate": 1.7551020408163266e-05, "loss": 1.343, "step": 88 }, { "epoch": 0.03, "learning_rate": 1.7755102040816327e-05, "loss": 1.822, "step": 89 }, { "epoch": 0.03, "learning_rate": 1.795918367346939e-05, "loss": 1.3541, "step": 90 }, { "epoch": 0.03, "learning_rate": 1.816326530612245e-05, "loss": 1.6902, "step": 91 }, { "epoch": 0.03, "learning_rate": 1.836734693877551e-05, "loss": 1.9583, "step": 92 }, { "epoch": 0.03, "learning_rate": 1.8571428571428575e-05, "loss": 1.9069, "step": 93 }, { "epoch": 0.03, "learning_rate": 1.8775510204081636e-05, "loss": 1.472, "step": 94 }, { "epoch": 0.03, "learning_rate": 1.8979591836734696e-05, "loss": 1.7204, "step": 95 }, { "epoch": 0.03, "learning_rate": 1.9183673469387756e-05, "loss": 1.6665, "step": 96 }, { "epoch": 0.03, "learning_rate": 1.9387755102040817e-05, "loss": 1.4465, "step": 97 }, { "epoch": 0.03, "learning_rate": 1.9591836734693877e-05, "loss": 1.8124, "step": 98 }, { "epoch": 0.03, "learning_rate": 1.979591836734694e-05, "loss": 1.7005, "step": 99 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 1.5046, "step": 100 }, { "epoch": 0.03, "learning_rate": 1.9999995032964135e-05, "loss": 1.3994, "step": 101 }, { "epoch": 0.03, "learning_rate": 1.999998013186146e-05, "loss": 2.3198, "step": 102 }, { "epoch": 0.03, "learning_rate": 1.9999955296706797e-05, "loss": 1.9909, "step": 103 }, { "epoch": 0.03, "learning_rate": 1.99999205275248e-05, "loss": 1.495, "step": 104 }, { "epoch": 0.03, "learning_rate": 1.9999875824350018e-05, "loss": 1.6305, "step": 105 }, { "epoch": 0.03, "learning_rate": 1.9999821187226853e-05, "loss": 1.594, "step": 106 }, { "epoch": 0.03, "learning_rate": 1.999975661620959e-05, "loss": 1.4535, "step": 107 }, { "epoch": 0.03, "learning_rate": 1.9999682111362368e-05, "loss": 1.8067, "step": 108 }, { "epoch": 0.03, "learning_rate": 1.9999597672759204e-05, "loss": 1.3976, "step": 109 }, { "epoch": 0.03, "learning_rate": 1.9999503300483978e-05, "loss": 1.42, "step": 110 }, { "epoch": 0.03, "learning_rate": 1.9999398994630442e-05, "loss": 1.5618, "step": 111 }, { "epoch": 0.03, "learning_rate": 1.9999284755302213e-05, "loss": 1.8627, "step": 112 }, { "epoch": 0.03, "learning_rate": 1.9999160582612775e-05, "loss": 1.8137, "step": 113 }, { "epoch": 0.04, "learning_rate": 1.9999026476685484e-05, "loss": 1.7382, "step": 114 }, { "epoch": 0.04, "learning_rate": 1.9998882437653565e-05, "loss": 1.7163, "step": 115 }, { "epoch": 0.04, "learning_rate": 1.9998728465660105e-05, "loss": 1.7374, "step": 116 }, { "epoch": 0.04, "learning_rate": 1.999856456085806e-05, "loss": 1.5717, "step": 117 }, { "epoch": 0.04, "learning_rate": 1.999839072341025e-05, "loss": 1.4653, "step": 118 }, { "epoch": 0.04, "learning_rate": 1.9998206953489374e-05, "loss": 1.5283, "step": 119 }, { "epoch": 0.04, "learning_rate": 1.9998013251277988e-05, "loss": 1.399, "step": 120 }, { "epoch": 0.04, "learning_rate": 1.9997809616968514e-05, "loss": 1.8169, "step": 121 }, { "epoch": 0.04, "learning_rate": 1.9997596050763253e-05, "loss": 1.7379, "step": 122 }, { "epoch": 0.04, "learning_rate": 1.999737255287435e-05, "loss": 1.2735, "step": 123 }, { "epoch": 0.04, "learning_rate": 1.999713912352384e-05, "loss": 1.6284, "step": 124 }, { "epoch": 0.04, "learning_rate": 1.9996895762943604e-05, "loss": 1.8106, "step": 125 }, { "epoch": 0.04, "learning_rate": 1.9996642471375406e-05, "loss": 1.628, "step": 126 }, { "epoch": 0.04, "learning_rate": 1.9996379249070865e-05, "loss": 2.1126, "step": 127 }, { "epoch": 0.04, "learning_rate": 1.999610609629147e-05, "loss": 1.5752, "step": 128 }, { "epoch": 0.04, "learning_rate": 1.9995823013308567e-05, "loss": 1.6581, "step": 129 }, { "epoch": 0.04, "learning_rate": 1.9995530000403378e-05, "loss": 2.675, "step": 130 }, { "epoch": 0.04, "learning_rate": 1.9995227057866985e-05, "loss": 1.3908, "step": 131 }, { "epoch": 0.04, "learning_rate": 1.999491418600033e-05, "loss": 1.7088, "step": 132 }, { "epoch": 0.04, "learning_rate": 1.9994591385114218e-05, "loss": 1.7882, "step": 133 }, { "epoch": 0.04, "learning_rate": 1.9994258655529336e-05, "loss": 1.5781, "step": 134 }, { "epoch": 0.04, "learning_rate": 1.9993915997576204e-05, "loss": 1.7004, "step": 135 }, { "epoch": 0.04, "learning_rate": 1.999356341159523e-05, "loss": 1.5753, "step": 136 }, { "epoch": 0.04, "learning_rate": 1.9993200897936673e-05, "loss": 1.7941, "step": 137 }, { "epoch": 0.04, "learning_rate": 1.9992828456960654e-05, "loss": 1.6074, "step": 138 }, { "epoch": 0.04, "learning_rate": 1.9992446089037164e-05, "loss": 1.3368, "step": 139 }, { "epoch": 0.04, "learning_rate": 1.999205379454605e-05, "loss": 1.6603, "step": 140 }, { "epoch": 0.04, "learning_rate": 1.9991651573877012e-05, "loss": 2.8197, "step": 141 }, { "epoch": 0.04, "learning_rate": 1.999123942742963e-05, "loss": 1.837, "step": 142 }, { "epoch": 0.04, "learning_rate": 1.9990817355613323e-05, "loss": 1.7987, "step": 143 }, { "epoch": 0.04, "learning_rate": 1.9990385358847385e-05, "loss": 1.4613, "step": 144 }, { "epoch": 0.04, "learning_rate": 1.998994343756097e-05, "loss": 2.5077, "step": 145 }, { "epoch": 0.04, "learning_rate": 1.9989491592193074e-05, "loss": 1.6775, "step": 146 }, { "epoch": 0.05, "learning_rate": 1.9989029823192575e-05, "loss": 1.6271, "step": 147 }, { "epoch": 0.05, "learning_rate": 1.9988558131018188e-05, "loss": 1.6006, "step": 148 }, { "epoch": 0.05, "learning_rate": 1.99880765161385e-05, "loss": 1.5418, "step": 149 }, { "epoch": 0.05, "learning_rate": 1.9987584979031954e-05, "loss": 1.4503, "step": 150 }, { "epoch": 0.05, "learning_rate": 1.998708352018684e-05, "loss": 1.7547, "step": 151 }, { "epoch": 0.05, "learning_rate": 1.9986572140101312e-05, "loss": 1.5231, "step": 152 }, { "epoch": 0.05, "learning_rate": 1.9986050839283383e-05, "loss": 1.536, "step": 153 }, { "epoch": 0.05, "learning_rate": 1.9985519618250912e-05, "loss": 1.578, "step": 154 }, { "epoch": 0.05, "learning_rate": 1.9984978477531625e-05, "loss": 1.4663, "step": 155 }, { "epoch": 0.05, "learning_rate": 1.9984427417663085e-05, "loss": 1.5846, "step": 156 }, { "epoch": 0.05, "learning_rate": 1.9983866439192726e-05, "loss": 1.4892, "step": 157 }, { "epoch": 0.05, "learning_rate": 1.998329554267783e-05, "loss": 1.4006, "step": 158 }, { "epoch": 0.05, "learning_rate": 1.998271472868552e-05, "loss": 1.5786, "step": 159 }, { "epoch": 0.05, "learning_rate": 1.998212399779279e-05, "loss": 1.6252, "step": 160 }, { "epoch": 0.05, "learning_rate": 1.998152335058647e-05, "loss": 1.5301, "step": 161 }, { "epoch": 0.05, "learning_rate": 1.9980912787663253e-05, "loss": 1.8721, "step": 162 }, { "epoch": 0.05, "learning_rate": 1.998029230962967e-05, "loss": 1.7937, "step": 163 }, { "epoch": 0.05, "learning_rate": 1.9979661917102116e-05, "loss": 1.685, "step": 164 }, { "epoch": 0.05, "learning_rate": 1.9979021610706822e-05, "loss": 2.0472, "step": 165 }, { "epoch": 0.05, "learning_rate": 1.9978371391079874e-05, "loss": 1.5465, "step": 166 }, { "epoch": 0.05, "learning_rate": 1.9978371391079874e-05, "loss": 1.3668, "step": 167 }, { "epoch": 0.05, "learning_rate": 1.9977711258867204e-05, "loss": 1.4535, "step": 168 }, { "epoch": 0.05, "learning_rate": 1.9977041214724594e-05, "loss": 1.6723, "step": 169 }, { "epoch": 0.05, "learning_rate": 1.9976361259317674e-05, "loss": 1.7369, "step": 170 }, { "epoch": 0.05, "learning_rate": 1.997567139332191e-05, "loss": 1.6297, "step": 171 }, { "epoch": 0.05, "learning_rate": 1.9974971617422622e-05, "loss": 1.5693, "step": 172 }, { "epoch": 0.05, "learning_rate": 1.997426193231497e-05, "loss": 1.7087, "step": 173 }, { "epoch": 0.05, "learning_rate": 1.9973542338703965e-05, "loss": 1.3286, "step": 174 }, { "epoch": 0.05, "learning_rate": 1.9972812837304454e-05, "loss": 1.7632, "step": 175 }, { "epoch": 0.05, "learning_rate": 1.9972073428841134e-05, "loss": 1.8314, "step": 176 }, { "epoch": 0.05, "learning_rate": 1.9971324114048533e-05, "loss": 1.4059, "step": 177 }, { "epoch": 0.05, "learning_rate": 1.997056489367102e-05, "loss": 1.3944, "step": 178 }, { "epoch": 0.06, "learning_rate": 1.9969795768462824e-05, "loss": 1.6226, "step": 179 }, { "epoch": 0.06, "learning_rate": 1.996901673918799e-05, "loss": 1.5514, "step": 180 }, { "epoch": 0.06, "learning_rate": 1.9968227806620413e-05, "loss": 1.7222, "step": 181 }, { "epoch": 0.06, "learning_rate": 1.9967428971543824e-05, "loss": 1.3574, "step": 182 }, { "epoch": 0.06, "learning_rate": 1.9966620234751794e-05, "loss": 1.396, "step": 183 }, { "epoch": 0.06, "learning_rate": 1.9965801597047724e-05, "loss": 1.5345, "step": 184 }, { "epoch": 0.06, "learning_rate": 1.996497305924486e-05, "loss": 1.5218, "step": 185 }, { "epoch": 0.06, "learning_rate": 1.9964134622166273e-05, "loss": 1.7672, "step": 186 }, { "epoch": 0.06, "learning_rate": 1.9963286286644875e-05, "loss": 1.8566, "step": 187 }, { "epoch": 0.06, "learning_rate": 1.9962428053523406e-05, "loss": 1.9358, "step": 188 }, { "epoch": 0.06, "learning_rate": 1.996155992365444e-05, "loss": 1.828, "step": 189 }, { "epoch": 0.06, "learning_rate": 1.9960681897900388e-05, "loss": 1.6402, "step": 190 }, { "epoch": 0.06, "learning_rate": 1.9959793977133483e-05, "loss": 1.5794, "step": 191 }, { "epoch": 0.06, "learning_rate": 1.9958896162235794e-05, "loss": 1.7885, "step": 192 }, { "epoch": 0.06, "learning_rate": 1.995798845409922e-05, "loss": 1.6629, "step": 193 }, { "epoch": 0.06, "learning_rate": 1.9957070853625476e-05, "loss": 1.6692, "step": 194 }, { "epoch": 0.06, "learning_rate": 1.9956143361726123e-05, "loss": 1.4805, "step": 195 }, { "epoch": 0.06, "learning_rate": 1.9955205979322528e-05, "loss": 1.6437, "step": 196 }, { "epoch": 0.06, "learning_rate": 1.9954258707345903e-05, "loss": 1.9134, "step": 197 }, { "epoch": 0.06, "learning_rate": 1.9953301546737267e-05, "loss": 1.4309, "step": 198 }, { "epoch": 0.06, "learning_rate": 1.9952334498447473e-05, "loss": 1.5045, "step": 199 }, { "epoch": 0.06, "learning_rate": 1.9951357563437198e-05, "loss": 1.5969, "step": 200 }, { "epoch": 0.06, "learning_rate": 1.995037074267693e-05, "loss": 1.2856, "step": 201 }, { "epoch": 0.06, "learning_rate": 1.9949374037146984e-05, "loss": 1.6973, "step": 202 }, { "epoch": 0.06, "learning_rate": 1.9948367447837503e-05, "loss": 1.8148, "step": 203 }, { "epoch": 0.06, "learning_rate": 1.9947350975748428e-05, "loss": 1.7634, "step": 204 }, { "epoch": 0.06, "learning_rate": 1.9946324621889538e-05, "loss": 1.6316, "step": 205 }, { "epoch": 0.06, "learning_rate": 1.994528838728042e-05, "loss": 1.5419, "step": 206 }, { "epoch": 0.06, "learning_rate": 1.9944242272950468e-05, "loss": 1.4158, "step": 207 }, { "epoch": 0.06, "learning_rate": 1.9943186279938908e-05, "loss": 1.7323, "step": 208 }, { "epoch": 0.06, "learning_rate": 1.9942120409294768e-05, "loss": 1.3954, "step": 209 }, { "epoch": 0.06, "learning_rate": 1.9941044662076898e-05, "loss": 1.7536, "step": 210 }, { "epoch": 0.06, "learning_rate": 1.9939959039353947e-05, "loss": 1.5485, "step": 211 }, { "epoch": 0.07, "learning_rate": 1.993886354220438e-05, "loss": 2.001, "step": 212 }, { "epoch": 0.07, "learning_rate": 1.9937758171716468e-05, "loss": 1.5152, "step": 213 }, { "epoch": 0.07, "learning_rate": 1.9936642928988304e-05, "loss": 2.101, "step": 214 }, { "epoch": 0.07, "learning_rate": 1.993551781512777e-05, "loss": 1.6773, "step": 215 }, { "epoch": 0.07, "learning_rate": 1.9934382831252565e-05, "loss": 2.0094, "step": 216 }, { "epoch": 0.07, "learning_rate": 1.9933237978490193e-05, "loss": 1.5861, "step": 217 }, { "epoch": 0.07, "learning_rate": 1.9932083257977954e-05, "loss": 1.6079, "step": 218 }, { "epoch": 0.07, "learning_rate": 1.993091867086296e-05, "loss": 1.3814, "step": 219 }, { "epoch": 0.07, "learning_rate": 1.9929744218302116e-05, "loss": 1.8781, "step": 220 }, { "epoch": 0.07, "learning_rate": 1.992855990146213e-05, "loss": 1.5523, "step": 221 }, { "epoch": 0.07, "learning_rate": 1.992736572151952e-05, "loss": 1.4769, "step": 222 }, { "epoch": 0.07, "learning_rate": 1.9926161679660588e-05, "loss": 1.6965, "step": 223 }, { "epoch": 0.07, "learning_rate": 1.9924947777081434e-05, "loss": 1.3362, "step": 224 }, { "epoch": 0.07, "learning_rate": 1.9923724014987963e-05, "loss": 1.5268, "step": 225 }, { "epoch": 0.07, "learning_rate": 1.9922490394595862e-05, "loss": 1.6003, "step": 226 }, { "epoch": 0.07, "learning_rate": 1.9921246917130627e-05, "loss": 1.3486, "step": 227 }, { "epoch": 0.07, "learning_rate": 1.991999358382753e-05, "loss": 1.4856, "step": 228 }, { "epoch": 0.07, "learning_rate": 1.9918730395931648e-05, "loss": 1.3157, "step": 229 }, { "epoch": 0.07, "learning_rate": 1.9917457354697838e-05, "loss": 1.5973, "step": 230 }, { "epoch": 0.07, "learning_rate": 1.991617446139075e-05, "loss": 1.4162, "step": 231 }, { "epoch": 0.07, "learning_rate": 1.9914881717284812e-05, "loss": 1.5909, "step": 232 }, { "epoch": 0.07, "learning_rate": 1.9913579123664253e-05, "loss": 1.4641, "step": 233 }, { "epoch": 0.07, "learning_rate": 1.9912266681823077e-05, "loss": 1.4476, "step": 234 }, { "epoch": 0.07, "learning_rate": 1.9910944393065073e-05, "loss": 1.4394, "step": 235 }, { "epoch": 0.07, "learning_rate": 1.990961225870381e-05, "loss": 1.7175, "step": 236 }, { "epoch": 0.07, "learning_rate": 1.9908270280062643e-05, "loss": 1.7933, "step": 237 }, { "epoch": 0.07, "learning_rate": 1.99069184584747e-05, "loss": 1.3454, "step": 238 }, { "epoch": 0.07, "learning_rate": 1.9905556795282894e-05, "loss": 1.2857, "step": 239 }, { "epoch": 0.07, "learning_rate": 1.990418529183991e-05, "loss": 1.7622, "step": 240 }, { "epoch": 0.07, "learning_rate": 1.9902803949508206e-05, "loss": 1.6376, "step": 241 }, { "epoch": 0.07, "learning_rate": 1.9901412769660023e-05, "loss": 1.5477, "step": 242 }, { "epoch": 0.07, "learning_rate": 1.9900011753677363e-05, "loss": 1.8393, "step": 243 }, { "epoch": 0.08, "learning_rate": 1.989860090295201e-05, "loss": 1.592, "step": 244 }, { "epoch": 0.08, "learning_rate": 1.989718021888551e-05, "loss": 1.3824, "step": 245 }, { "epoch": 0.08, "learning_rate": 1.989574970288918e-05, "loss": 1.7632, "step": 246 }, { "epoch": 0.08, "learning_rate": 1.989430935638411e-05, "loss": 1.5889, "step": 247 }, { "epoch": 0.08, "learning_rate": 1.9892859180801146e-05, "loss": 1.4855, "step": 248 }, { "epoch": 0.08, "learning_rate": 1.989139917758091e-05, "loss": 1.9986, "step": 249 }, { "epoch": 0.08, "learning_rate": 1.9889929348173767e-05, "loss": 1.7831, "step": 250 }, { "epoch": 0.08, "learning_rate": 1.9888449694039866e-05, "loss": 1.7295, "step": 251 }, { "epoch": 0.08, "learning_rate": 1.9886960216649102e-05, "loss": 2.1687, "step": 252 }, { "epoch": 0.08, "learning_rate": 1.9885460917481137e-05, "loss": 1.6867, "step": 253 }, { "epoch": 0.08, "learning_rate": 1.9883951798025377e-05, "loss": 1.6013, "step": 254 }, { "epoch": 0.08, "learning_rate": 1.9882432859781002e-05, "loss": 1.5046, "step": 255 }, { "epoch": 0.08, "learning_rate": 1.9880904104256927e-05, "loss": 1.5368, "step": 256 }, { "epoch": 0.08, "learning_rate": 1.987936553297184e-05, "loss": 1.3318, "step": 257 }, { "epoch": 0.08, "learning_rate": 1.9877817147454156e-05, "loss": 1.6516, "step": 258 }, { "epoch": 0.08, "learning_rate": 1.987625894924206e-05, "loss": 1.4754, "step": 259 }, { "epoch": 0.08, "learning_rate": 1.9874690939883473e-05, "loss": 1.8631, "step": 260 }, { "epoch": 0.08, "learning_rate": 1.9873113120936074e-05, "loss": 1.7939, "step": 261 }, { "epoch": 0.08, "learning_rate": 1.9871525493967273e-05, "loss": 1.715, "step": 262 }, { "epoch": 0.08, "learning_rate": 1.9869928060554234e-05, "loss": 1.7277, "step": 263 }, { "epoch": 0.08, "learning_rate": 1.9868320822283853e-05, "loss": 1.1737, "step": 264 }, { "epoch": 0.08, "learning_rate": 1.9866703780752777e-05, "loss": 1.6055, "step": 265 }, { "epoch": 0.08, "learning_rate": 1.9865076937567384e-05, "loss": 1.5523, "step": 266 }, { "epoch": 0.08, "learning_rate": 1.9863440294343793e-05, "loss": 1.4681, "step": 267 }, { "epoch": 0.08, "learning_rate": 1.986179385270786e-05, "loss": 1.8599, "step": 268 }, { "epoch": 0.08, "learning_rate": 1.986013761429517e-05, "loss": 1.7851, "step": 269 }, { "epoch": 0.08, "learning_rate": 1.985847158075104e-05, "loss": 1.663, "step": 270 }, { "epoch": 0.08, "learning_rate": 1.985679575373052e-05, "loss": 1.8522, "step": 271 }, { "epoch": 0.08, "learning_rate": 1.985511013489839e-05, "loss": 1.7483, "step": 272 }, { "epoch": 0.08, "learning_rate": 1.9853414725929157e-05, "loss": 1.6165, "step": 273 }, { "epoch": 0.08, "learning_rate": 1.9851709528507047e-05, "loss": 2.2397, "step": 274 }, { "epoch": 0.08, "learning_rate": 1.984999454432602e-05, "loss": 1.605, "step": 275 }, { "epoch": 0.08, "learning_rate": 1.9848269775089756e-05, "loss": 1.7188, "step": 276 }, { "epoch": 0.09, "learning_rate": 1.9846535222511648e-05, "loss": 1.7159, "step": 277 }, { "epoch": 0.09, "learning_rate": 1.9844790888314812e-05, "loss": 1.4937, "step": 278 }, { "epoch": 0.09, "learning_rate": 1.984303677423209e-05, "loss": 1.4489, "step": 279 }, { "epoch": 0.09, "learning_rate": 1.9841272882006026e-05, "loss": 1.4188, "step": 280 }, { "epoch": 0.09, "learning_rate": 1.983949921338888e-05, "loss": 1.6416, "step": 281 }, { "epoch": 0.09, "learning_rate": 1.9837715770142633e-05, "loss": 2.0655, "step": 282 }, { "epoch": 0.09, "learning_rate": 1.983592255403897e-05, "loss": 1.6055, "step": 283 }, { "epoch": 0.09, "learning_rate": 1.983411956685928e-05, "loss": 1.7311, "step": 284 }, { "epoch": 0.09, "learning_rate": 1.9832306810394665e-05, "loss": 1.8048, "step": 285 }, { "epoch": 0.09, "learning_rate": 1.9830484286445935e-05, "loss": 2.0122, "step": 286 }, { "epoch": 0.09, "learning_rate": 1.9828651996823593e-05, "loss": 1.6216, "step": 287 }, { "epoch": 0.09, "learning_rate": 1.982680994334785e-05, "loss": 1.6082, "step": 288 }, { "epoch": 0.09, "learning_rate": 1.9824958127848618e-05, "loss": 1.3639, "step": 289 }, { "epoch": 0.09, "learning_rate": 1.98230965521655e-05, "loss": 1.4936, "step": 290 }, { "epoch": 0.09, "learning_rate": 1.98212252181478e-05, "loss": 1.4603, "step": 291 }, { "epoch": 0.09, "learning_rate": 1.9819344127654515e-05, "loss": 1.6137, "step": 292 }, { "epoch": 0.09, "learning_rate": 1.9817453282554334e-05, "loss": 1.2439, "step": 293 }, { "epoch": 0.09, "learning_rate": 1.9815552684725635e-05, "loss": 1.8232, "step": 294 }, { "epoch": 0.09, "learning_rate": 1.9813642336056488e-05, "loss": 1.6075, "step": 295 }, { "epoch": 0.09, "learning_rate": 1.9811722238444642e-05, "loss": 1.7471, "step": 296 }, { "epoch": 0.09, "learning_rate": 1.9809792393797543e-05, "loss": 1.5875, "step": 297 }, { "epoch": 0.09, "learning_rate": 1.9807852804032306e-05, "loss": 1.6445, "step": 298 }, { "epoch": 0.09, "learning_rate": 1.9805903471075735e-05, "loss": 1.4694, "step": 299 }, { "epoch": 0.09, "learning_rate": 1.9803944396864313e-05, "loss": 2.347, "step": 300 }, { "epoch": 0.09, "learning_rate": 1.98019755833442e-05, "loss": 1.6418, "step": 301 }, { "epoch": 0.09, "learning_rate": 1.9799997032471223e-05, "loss": 1.6859, "step": 302 }, { "epoch": 0.09, "learning_rate": 1.9798008746210894e-05, "loss": 1.5495, "step": 303 }, { "epoch": 0.09, "learning_rate": 1.9796010726538393e-05, "loss": 1.5832, "step": 304 }, { "epoch": 0.09, "learning_rate": 1.979400297543856e-05, "loss": 1.6645, "step": 305 }, { "epoch": 0.09, "learning_rate": 1.9791985494905913e-05, "loss": 1.631, "step": 306 }, { "epoch": 0.09, "learning_rate": 1.9789958286944632e-05, "loss": 1.4156, "step": 307 }, { "epoch": 0.09, "learning_rate": 1.978792135356856e-05, "loss": 1.9984, "step": 308 }, { "epoch": 0.1, "learning_rate": 1.97858746968012e-05, "loss": 2.0683, "step": 309 }, { "epoch": 0.1, "learning_rate": 1.97858746968012e-05, "loss": 1.6231, "step": 310 }, { "epoch": 0.1, "learning_rate": 1.978381831867572e-05, "loss": 1.734, "step": 311 }, { "epoch": 0.1, "learning_rate": 1.9781752221234932e-05, "loss": 1.3379, "step": 312 }, { "epoch": 0.1, "learning_rate": 1.977967640653132e-05, "loss": 1.6322, "step": 313 }, { "epoch": 0.1, "learning_rate": 1.9777590876627012e-05, "loss": 1.4378, "step": 314 }, { "epoch": 0.1, "learning_rate": 1.977549563359378e-05, "loss": 1.6712, "step": 315 }, { "epoch": 0.1, "learning_rate": 1.9773390679513068e-05, "loss": 1.6888, "step": 316 }, { "epoch": 0.1, "learning_rate": 1.977127601647594e-05, "loss": 1.7305, "step": 317 }, { "epoch": 0.1, "learning_rate": 1.9769151646583122e-05, "loss": 1.5078, "step": 318 }, { "epoch": 0.1, "learning_rate": 1.9767017571944976e-05, "loss": 1.416, "step": 319 }, { "epoch": 0.1, "learning_rate": 1.9764873794681514e-05, "loss": 1.5982, "step": 320 }, { "epoch": 0.1, "learning_rate": 1.9762720316922372e-05, "loss": 1.6363, "step": 321 }, { "epoch": 0.1, "learning_rate": 1.9760557140806838e-05, "loss": 2.1026, "step": 322 }, { "epoch": 0.1, "learning_rate": 1.9758384268483816e-05, "loss": 1.3741, "step": 323 }, { "epoch": 0.1, "learning_rate": 1.9756201702111863e-05, "loss": 1.4933, "step": 324 }, { "epoch": 0.1, "learning_rate": 1.975400944385915e-05, "loss": 1.3125, "step": 325 }, { "epoch": 0.1, "learning_rate": 1.9751807495903484e-05, "loss": 1.6205, "step": 326 }, { "epoch": 0.1, "learning_rate": 1.9749595860432298e-05, "loss": 1.5026, "step": 327 }, { "epoch": 0.1, "learning_rate": 1.9747374539642643e-05, "loss": 1.5848, "step": 328 }, { "epoch": 0.1, "learning_rate": 1.97451435357412e-05, "loss": 1.852, "step": 329 }, { "epoch": 0.1, "learning_rate": 1.9742902850944257e-05, "loss": 1.5626, "step": 330 }, { "epoch": 0.1, "learning_rate": 1.974065248747773e-05, "loss": 1.6991, "step": 331 }, { "epoch": 0.1, "learning_rate": 1.9738392447577148e-05, "loss": 1.4958, "step": 332 }, { "epoch": 0.1, "learning_rate": 1.9736122733487648e-05, "loss": 1.5429, "step": 333 }, { "epoch": 0.1, "learning_rate": 1.9733843347463982e-05, "loss": 1.5443, "step": 334 }, { "epoch": 0.1, "learning_rate": 1.973155429177051e-05, "loss": 1.5045, "step": 335 }, { "epoch": 0.1, "learning_rate": 1.972925556868119e-05, "loss": 1.419, "step": 336 }, { "epoch": 0.1, "learning_rate": 1.9726947180479598e-05, "loss": 1.8166, "step": 337 }, { "epoch": 0.1, "learning_rate": 1.97246291294589e-05, "loss": 1.6045, "step": 338 }, { "epoch": 0.1, "learning_rate": 1.9722301417921864e-05, "loss": 1.526, "step": 339 }, { "epoch": 0.1, "learning_rate": 1.9719964048180853e-05, "loss": 1.4242, "step": 340 }, { "epoch": 0.1, "learning_rate": 1.9717617022557828e-05, "loss": 1.6578, "step": 341 }, { "epoch": 0.11, "learning_rate": 1.971526034338435e-05, "loss": 1.508, "step": 342 }, { "epoch": 0.11, "learning_rate": 1.971289401300155e-05, "loss": 1.4866, "step": 343 }, { "epoch": 0.11, "learning_rate": 1.9710518033760158e-05, "loss": 1.5905, "step": 344 }, { "epoch": 0.11, "learning_rate": 1.9708132408020493e-05, "loss": 1.6059, "step": 345 }, { "epoch": 0.11, "learning_rate": 1.970573713815245e-05, "loss": 1.5395, "step": 346 }, { "epoch": 0.11, "learning_rate": 1.9703332226535515e-05, "loss": 1.7586, "step": 347 }, { "epoch": 0.11, "learning_rate": 1.9700917675558733e-05, "loss": 1.3932, "step": 348 }, { "epoch": 0.11, "learning_rate": 1.9698493487620742e-05, "loss": 1.5638, "step": 349 }, { "epoch": 0.11, "learning_rate": 1.969605966512975e-05, "loss": 1.5222, "step": 350 }, { "epoch": 0.11, "learning_rate": 1.9693616210503528e-05, "loss": 1.322, "step": 351 }, { "epoch": 0.11, "learning_rate": 1.9691163126169428e-05, "loss": 1.636, "step": 352 }, { "epoch": 0.11, "learning_rate": 1.9688700414564355e-05, "loss": 1.6224, "step": 353 }, { "epoch": 0.11, "learning_rate": 1.968622807813479e-05, "loss": 1.4956, "step": 354 }, { "epoch": 0.11, "learning_rate": 1.9683746119336763e-05, "loss": 1.5849, "step": 355 }, { "epoch": 0.11, "learning_rate": 1.9681254540635877e-05, "loss": 1.4047, "step": 356 }, { "epoch": 0.11, "learning_rate": 1.967875334450728e-05, "loss": 1.4244, "step": 357 }, { "epoch": 0.11, "learning_rate": 1.967624253343568e-05, "loss": 1.8103, "step": 358 }, { "epoch": 0.11, "learning_rate": 1.9673722109915333e-05, "loss": 1.4233, "step": 359 }, { "epoch": 0.11, "learning_rate": 1.9671192076450047e-05, "loss": 1.5318, "step": 360 }, { "epoch": 0.11, "learning_rate": 1.9668652435553173e-05, "loss": 1.2261, "step": 361 }, { "epoch": 0.11, "learning_rate": 1.966610318974761e-05, "loss": 1.4746, "step": 362 }, { "epoch": 0.11, "learning_rate": 1.96635443415658e-05, "loss": 1.8807, "step": 363 }, { "epoch": 0.11, "learning_rate": 1.9660975893549722e-05, "loss": 1.5223, "step": 364 }, { "epoch": 0.11, "learning_rate": 1.9658397848250882e-05, "loss": 1.582, "step": 365 }, { "epoch": 0.11, "learning_rate": 1.9655810208230334e-05, "loss": 1.8018, "step": 366 }, { "epoch": 0.11, "learning_rate": 1.9653212976058664e-05, "loss": 1.5699, "step": 367 }, { "epoch": 0.11, "learning_rate": 1.9650606154315967e-05, "loss": 1.7314, "step": 368 }, { "epoch": 0.11, "learning_rate": 1.964798974559189e-05, "loss": 1.56, "step": 369 }, { "epoch": 0.11, "learning_rate": 1.9645363752485594e-05, "loss": 1.4964, "step": 370 }, { "epoch": 0.11, "learning_rate": 1.9642728177605747e-05, "loss": 1.7369, "step": 371 }, { "epoch": 0.11, "learning_rate": 1.964008302357056e-05, "loss": 1.2725, "step": 372 }, { "epoch": 0.11, "learning_rate": 1.9637428293007744e-05, "loss": 1.449, "step": 373 }, { "epoch": 0.12, "learning_rate": 1.963476398855452e-05, "loss": 1.6289, "step": 374 }, { "epoch": 0.12, "learning_rate": 1.963209011285764e-05, "loss": 0.988, "step": 375 }, { "epoch": 0.12, "learning_rate": 1.9629406668573346e-05, "loss": 1.6609, "step": 376 }, { "epoch": 0.12, "learning_rate": 1.9626713658367385e-05, "loss": 1.7869, "step": 377 }, { "epoch": 0.12, "learning_rate": 1.962401108491502e-05, "loss": 1.4871, "step": 378 }, { "epoch": 0.12, "learning_rate": 1.9621298950901005e-05, "loss": 1.3658, "step": 379 }, { "epoch": 0.12, "learning_rate": 1.9618577259019595e-05, "loss": 1.6384, "step": 380 }, { "epoch": 0.12, "learning_rate": 1.961584601197453e-05, "loss": 1.6409, "step": 381 }, { "epoch": 0.12, "learning_rate": 1.961310521247906e-05, "loss": 1.9667, "step": 382 }, { "epoch": 0.12, "learning_rate": 1.9610354863255914e-05, "loss": 1.3878, "step": 383 }, { "epoch": 0.12, "learning_rate": 1.96075949670373e-05, "loss": 1.5849, "step": 384 }, { "epoch": 0.12, "learning_rate": 1.960482552656493e-05, "loss": 1.4441, "step": 385 }, { "epoch": 0.12, "learning_rate": 1.960204654458998e-05, "loss": 1.8731, "step": 386 }, { "epoch": 0.12, "learning_rate": 1.9599258023873108e-05, "loss": 1.6578, "step": 387 }, { "epoch": 0.12, "learning_rate": 1.9596459967184458e-05, "loss": 1.4602, "step": 388 }, { "epoch": 0.12, "learning_rate": 1.959365237730363e-05, "loss": 1.8843, "step": 389 }, { "epoch": 0.12, "learning_rate": 1.9590835257019715e-05, "loss": 1.685, "step": 390 }, { "epoch": 0.12, "learning_rate": 1.9588008609131253e-05, "loss": 1.6099, "step": 391 }, { "epoch": 0.12, "learning_rate": 1.958517243644626e-05, "loss": 1.4345, "step": 392 }, { "epoch": 0.12, "learning_rate": 1.9582326741782204e-05, "loss": 1.5144, "step": 393 }, { "epoch": 0.12, "learning_rate": 1.9579471527966026e-05, "loss": 1.5208, "step": 394 }, { "epoch": 0.12, "learning_rate": 1.9576606797834114e-05, "loss": 1.493, "step": 395 }, { "epoch": 0.12, "learning_rate": 1.957373255423231e-05, "loss": 1.3919, "step": 396 }, { "epoch": 0.12, "learning_rate": 1.9570848800015904e-05, "loss": 1.518, "step": 397 }, { "epoch": 0.12, "learning_rate": 1.9567955538049643e-05, "loss": 1.4975, "step": 398 }, { "epoch": 0.12, "learning_rate": 1.956505277120772e-05, "loss": 1.6597, "step": 399 }, { "epoch": 0.12, "learning_rate": 1.9562140502373753e-05, "loss": 1.7864, "step": 400 }, { "epoch": 0.12, "learning_rate": 1.9559218734440818e-05, "loss": 1.6085, "step": 401 }, { "epoch": 0.12, "learning_rate": 1.9556287470311418e-05, "loss": 1.5447, "step": 402 }, { "epoch": 0.12, "learning_rate": 1.955334671289749e-05, "loss": 1.4193, "step": 403 }, { "epoch": 0.12, "learning_rate": 1.955039646512041e-05, "loss": 1.4759, "step": 404 }, { "epoch": 0.12, "learning_rate": 1.9547436729910968e-05, "loss": 1.4027, "step": 405 }, { "epoch": 0.12, "learning_rate": 1.954446751020939e-05, "loss": 1.5694, "step": 406 }, { "epoch": 0.13, "learning_rate": 1.9541488808965318e-05, "loss": 1.3939, "step": 407 }, { "epoch": 0.13, "learning_rate": 1.9538500629137814e-05, "loss": 1.3507, "step": 408 }, { "epoch": 0.13, "learning_rate": 1.9535502973695362e-05, "loss": 1.916, "step": 409 }, { "epoch": 0.13, "learning_rate": 1.9532495845615854e-05, "loss": 2.342, "step": 410 }, { "epoch": 0.13, "learning_rate": 1.952947924788659e-05, "loss": 1.3617, "step": 411 }, { "epoch": 0.13, "learning_rate": 1.952645318350428e-05, "loss": 1.4452, "step": 412 }, { "epoch": 0.13, "learning_rate": 1.952341765547504e-05, "loss": 1.5346, "step": 413 }, { "epoch": 0.13, "learning_rate": 1.952037266681438e-05, "loss": 1.6285, "step": 414 }, { "epoch": 0.13, "learning_rate": 1.951731822054722e-05, "loss": 1.8198, "step": 415 }, { "epoch": 0.13, "learning_rate": 1.9514254319707868e-05, "loss": 1.7929, "step": 416 }, { "epoch": 0.13, "learning_rate": 1.951118096734002e-05, "loss": 1.199, "step": 417 }, { "epoch": 0.13, "learning_rate": 1.950809816649677e-05, "loss": 1.7565, "step": 418 }, { "epoch": 0.13, "learning_rate": 1.9505005920240596e-05, "loss": 1.3858, "step": 419 }, { "epoch": 0.13, "learning_rate": 1.9501904231643353e-05, "loss": 1.974, "step": 420 }, { "epoch": 0.13, "learning_rate": 1.9498793103786288e-05, "loss": 1.4981, "step": 421 }, { "epoch": 0.13, "learning_rate": 1.949567253976001e-05, "loss": 1.9566, "step": 422 }, { "epoch": 0.13, "learning_rate": 1.949254254266451e-05, "loss": 1.9694, "step": 423 }, { "epoch": 0.13, "learning_rate": 1.9489403115609157e-05, "loss": 1.6272, "step": 424 }, { "epoch": 0.13, "learning_rate": 1.9486254261712677e-05, "loss": 1.465, "step": 425 }, { "epoch": 0.13, "learning_rate": 1.948309598410316e-05, "loss": 1.6619, "step": 426 }, { "epoch": 0.13, "learning_rate": 1.9479928285918065e-05, "loss": 1.3405, "step": 427 }, { "epoch": 0.13, "learning_rate": 1.9476751170304206e-05, "loss": 1.6428, "step": 428 }, { "epoch": 0.13, "learning_rate": 1.947356464041775e-05, "loss": 1.7289, "step": 429 }, { "epoch": 0.13, "learning_rate": 1.947036869942422e-05, "loss": 1.6445, "step": 430 }, { "epoch": 0.13, "learning_rate": 1.946716335049849e-05, "loss": 1.503, "step": 431 }, { "epoch": 0.13, "learning_rate": 1.9463948596824774e-05, "loss": 1.845, "step": 432 }, { "epoch": 0.13, "learning_rate": 1.946072444159663e-05, "loss": 1.7573, "step": 433 }, { "epoch": 0.13, "learning_rate": 1.9457490888016962e-05, "loss": 1.6144, "step": 434 }, { "epoch": 0.13, "learning_rate": 1.9454247939297996e-05, "loss": 1.7371, "step": 435 }, { "epoch": 0.13, "learning_rate": 1.945099559866131e-05, "loss": 1.5602, "step": 436 }, { "epoch": 0.13, "learning_rate": 1.94477338693378e-05, "loss": 1.5315, "step": 437 }, { "epoch": 0.13, "learning_rate": 1.9444462754567682e-05, "loss": 1.5651, "step": 438 }, { "epoch": 0.14, "learning_rate": 1.944118225760052e-05, "loss": 1.2868, "step": 439 }, { "epoch": 0.14, "learning_rate": 1.9437892381695174e-05, "loss": 1.589, "step": 440 }, { "epoch": 0.14, "learning_rate": 1.9434593130119835e-05, "loss": 1.5082, "step": 441 }, { "epoch": 0.14, "learning_rate": 1.9431284506151996e-05, "loss": 1.7603, "step": 442 }, { "epoch": 0.14, "learning_rate": 1.9427966513078476e-05, "loss": 1.5296, "step": 443 }, { "epoch": 0.14, "learning_rate": 1.9424639154195386e-05, "loss": 1.8436, "step": 444 }, { "epoch": 0.14, "learning_rate": 1.9421302432808153e-05, "loss": 1.5024, "step": 445 }, { "epoch": 0.14, "learning_rate": 1.94179563522315e-05, "loss": 1.4501, "step": 446 }, { "epoch": 0.14, "learning_rate": 1.9414600915789446e-05, "loss": 1.4316, "step": 447 }, { "epoch": 0.14, "learning_rate": 1.9411236126815306e-05, "loss": 1.4682, "step": 448 }, { "epoch": 0.14, "learning_rate": 1.9407861988651683e-05, "loss": 1.3392, "step": 449 }, { "epoch": 0.14, "learning_rate": 1.9404478504650473e-05, "loss": 1.8718, "step": 450 }, { "epoch": 0.14, "learning_rate": 1.9401085678172852e-05, "loss": 1.7281, "step": 451 }, { "epoch": 0.14, "learning_rate": 1.9397683512589282e-05, "loss": 1.5543, "step": 452 }, { "epoch": 0.14, "learning_rate": 1.9394272011279492e-05, "loss": 1.4684, "step": 453 }, { "epoch": 0.14, "learning_rate": 1.9390851177632496e-05, "loss": 1.5275, "step": 454 }, { "epoch": 0.14, "learning_rate": 1.9387421015046576e-05, "loss": 1.8237, "step": 455 }, { "epoch": 0.14, "learning_rate": 1.9383981526929277e-05, "loss": 1.5748, "step": 456 }, { "epoch": 0.14, "learning_rate": 1.9380532716697412e-05, "loss": 1.5865, "step": 457 }, { "epoch": 0.14, "learning_rate": 1.937707458777705e-05, "loss": 1.5173, "step": 458 }, { "epoch": 0.14, "learning_rate": 1.9373607143603533e-05, "loss": 1.5135, "step": 459 }, { "epoch": 0.14, "learning_rate": 1.937013038762143e-05, "loss": 1.5442, "step": 460 }, { "epoch": 0.14, "learning_rate": 1.9366644323284584e-05, "loss": 2.0023, "step": 461 }, { "epoch": 0.14, "learning_rate": 1.9363148954056077e-05, "loss": 1.4716, "step": 462 }, { "epoch": 0.14, "learning_rate": 1.9359644283408233e-05, "loss": 1.9954, "step": 463 }, { "epoch": 0.14, "learning_rate": 1.935613031482261e-05, "loss": 2.2567, "step": 464 }, { "epoch": 0.14, "learning_rate": 1.935260705179002e-05, "loss": 1.7179, "step": 465 }, { "epoch": 0.14, "learning_rate": 1.934907449781049e-05, "loss": 1.6664, "step": 466 }, { "epoch": 0.14, "learning_rate": 1.934553265639329e-05, "loss": 1.4733, "step": 467 }, { "epoch": 0.14, "learning_rate": 1.9341981531056905e-05, "loss": 1.563, "step": 468 }, { "epoch": 0.14, "learning_rate": 1.9338421125329053e-05, "loss": 1.4747, "step": 469 }, { "epoch": 0.14, "learning_rate": 1.9334851442746665e-05, "loss": 1.6416, "step": 470 }, { "epoch": 0.14, "learning_rate": 1.933127248685589e-05, "loss": 1.7337, "step": 471 }, { "epoch": 0.15, "learning_rate": 1.9327684261212084e-05, "loss": 1.5642, "step": 472 }, { "epoch": 0.15, "learning_rate": 1.9324086769379825e-05, "loss": 1.528, "step": 473 }, { "epoch": 0.15, "learning_rate": 1.932048001493288e-05, "loss": 1.876, "step": 474 }, { "epoch": 0.15, "learning_rate": 1.9316864001454226e-05, "loss": 1.4909, "step": 475 }, { "epoch": 0.15, "learning_rate": 1.931323873253604e-05, "loss": 1.6337, "step": 476 }, { "epoch": 0.15, "learning_rate": 1.9309604211779684e-05, "loss": 2.5694, "step": 477 }, { "epoch": 0.15, "learning_rate": 1.930596044279572e-05, "loss": 1.6391, "step": 478 }, { "epoch": 0.15, "learning_rate": 1.9302307429203902e-05, "loss": 1.4738, "step": 479 }, { "epoch": 0.15, "learning_rate": 1.9298645174633147e-05, "loss": 1.6618, "step": 480 }, { "epoch": 0.15, "learning_rate": 1.9294973682721573e-05, "loss": 1.7738, "step": 481 }, { "epoch": 0.15, "learning_rate": 1.929129295711646e-05, "loss": 1.6075, "step": 482 }, { "epoch": 0.15, "learning_rate": 1.9287603001474274e-05, "loss": 1.3121, "step": 483 }, { "epoch": 0.15, "learning_rate": 1.9283903819460644e-05, "loss": 1.4098, "step": 484 }, { "epoch": 0.15, "learning_rate": 1.928019541475036e-05, "loss": 1.8231, "step": 485 }, { "epoch": 0.15, "learning_rate": 1.9276477791027374e-05, "loss": 1.6571, "step": 486 }, { "epoch": 0.15, "learning_rate": 1.9272750951984808e-05, "loss": 1.5591, "step": 487 }, { "epoch": 0.15, "learning_rate": 1.9269014901324926e-05, "loss": 1.3717, "step": 488 }, { "epoch": 0.15, "learning_rate": 1.926526964275915e-05, "loss": 1.4766, "step": 489 }, { "epoch": 0.15, "learning_rate": 1.9261515180008047e-05, "loss": 1.4282, "step": 490 }, { "epoch": 0.15, "learning_rate": 1.9257751516801323e-05, "loss": 1.8362, "step": 491 }, { "epoch": 0.15, "learning_rate": 1.925397865687783e-05, "loss": 1.4829, "step": 492 }, { "epoch": 0.15, "learning_rate": 1.9250196603985557e-05, "loss": 1.4105, "step": 493 }, { "epoch": 0.15, "learning_rate": 1.9246405361881623e-05, "loss": 1.6449, "step": 494 }, { "epoch": 0.15, "learning_rate": 1.9242604934332268e-05, "loss": 2.0773, "step": 495 }, { "epoch": 0.15, "learning_rate": 1.9238795325112867e-05, "loss": 1.6207, "step": 496 }, { "epoch": 0.15, "learning_rate": 1.923497653800792e-05, "loss": 1.4271, "step": 497 }, { "epoch": 0.15, "learning_rate": 1.923114857681103e-05, "loss": 1.6192, "step": 498 }, { "epoch": 0.15, "learning_rate": 1.9227311445324924e-05, "loss": 1.2114, "step": 499 }, { "epoch": 0.15, "learning_rate": 1.9223465147361433e-05, "loss": 1.3277, "step": 500 }, { "epoch": 0.15, "learning_rate": 1.92196096867415e-05, "loss": 1.5036, "step": 501 }, { "epoch": 0.15, "learning_rate": 1.9215745067295168e-05, "loss": 1.5056, "step": 502 }, { "epoch": 0.15, "learning_rate": 1.9211871292861576e-05, "loss": 1.4721, "step": 503 }, { "epoch": 0.16, "learning_rate": 1.9207988367288956e-05, "loss": 1.8662, "step": 504 }, { "epoch": 0.16, "learning_rate": 1.920409629443464e-05, "loss": 1.231, "step": 505 }, { "epoch": 0.16, "learning_rate": 1.920019507816504e-05, "loss": 1.8711, "step": 506 }, { "epoch": 0.16, "learning_rate": 1.9196284722355648e-05, "loss": 1.7091, "step": 507 }, { "epoch": 0.16, "learning_rate": 1.9192365230891044e-05, "loss": 1.5687, "step": 508 }, { "epoch": 0.16, "learning_rate": 1.9188436607664878e-05, "loss": 1.4181, "step": 509 }, { "epoch": 0.16, "learning_rate": 1.918449885657987e-05, "loss": 1.4208, "step": 510 }, { "epoch": 0.16, "learning_rate": 1.9180551981547813e-05, "loss": 1.7203, "step": 511 }, { "epoch": 0.16, "learning_rate": 1.9176595986489558e-05, "loss": 1.6528, "step": 512 }, { "epoch": 0.16, "learning_rate": 1.9172630875335024e-05, "loss": 1.6635, "step": 513 }, { "epoch": 0.16, "learning_rate": 1.9168656652023176e-05, "loss": 1.5474, "step": 514 }, { "epoch": 0.16, "learning_rate": 1.916467332050204e-05, "loss": 1.4327, "step": 515 }, { "epoch": 0.16, "learning_rate": 1.916068088472868e-05, "loss": 1.2963, "step": 516 }, { "epoch": 0.16, "learning_rate": 1.9156679348669215e-05, "loss": 1.694, "step": 517 }, { "epoch": 0.16, "learning_rate": 1.91526687162988e-05, "loss": 1.4347, "step": 518 }, { "epoch": 0.16, "learning_rate": 1.9148648991601623e-05, "loss": 1.8322, "step": 519 }, { "epoch": 0.16, "learning_rate": 1.914462017857091e-05, "loss": 1.995, "step": 520 }, { "epoch": 0.16, "learning_rate": 1.914058228120891e-05, "loss": 1.3905, "step": 521 }, { "epoch": 0.16, "learning_rate": 1.9136535303526906e-05, "loss": 1.738, "step": 522 }, { "epoch": 0.16, "learning_rate": 1.9132479249545186e-05, "loss": 1.7038, "step": 523 }, { "epoch": 0.16, "learning_rate": 1.9128414123293065e-05, "loss": 1.387, "step": 524 }, { "epoch": 0.16, "learning_rate": 1.912433992880887e-05, "loss": 1.5046, "step": 525 }, { "epoch": 0.16, "learning_rate": 1.9120256670139942e-05, "loss": 1.6666, "step": 526 }, { "epoch": 0.16, "learning_rate": 1.911616435134261e-05, "loss": 1.5341, "step": 527 }, { "epoch": 0.16, "learning_rate": 1.9112062976482213e-05, "loss": 1.5286, "step": 528 }, { "epoch": 0.16, "learning_rate": 1.910795254963309e-05, "loss": 1.47, "step": 529 }, { "epoch": 0.16, "learning_rate": 1.9103833074878565e-05, "loss": 2.0051, "step": 530 }, { "epoch": 0.16, "learning_rate": 1.9099704556310962e-05, "loss": 1.4829, "step": 531 }, { "epoch": 0.16, "learning_rate": 1.909556699803157e-05, "loss": 1.4724, "step": 532 }, { "epoch": 0.16, "learning_rate": 1.9091420404150674e-05, "loss": 2.0598, "step": 533 }, { "epoch": 0.16, "learning_rate": 1.9087264778787534e-05, "loss": 1.6841, "step": 534 }, { "epoch": 0.16, "learning_rate": 1.9083100126070375e-05, "loss": 1.4274, "step": 535 }, { "epoch": 0.16, "learning_rate": 1.907892645013639e-05, "loss": 1.5352, "step": 536 }, { "epoch": 0.17, "learning_rate": 1.9074743755131743e-05, "loss": 1.7034, "step": 537 }, { "epoch": 0.17, "learning_rate": 1.9070552045211544e-05, "loss": 1.8087, "step": 538 }, { "epoch": 0.17, "learning_rate": 1.906635132453988e-05, "loss": 1.2675, "step": 539 }, { "epoch": 0.17, "learning_rate": 1.9062141597289773e-05, "loss": 1.3889, "step": 540 }, { "epoch": 0.17, "learning_rate": 1.9057922867643196e-05, "loss": 1.4754, "step": 541 }, { "epoch": 0.17, "learning_rate": 1.9053695139791058e-05, "loss": 1.6518, "step": 542 }, { "epoch": 0.17, "learning_rate": 1.9049458417933225e-05, "loss": 1.5358, "step": 543 }, { "epoch": 0.17, "learning_rate": 1.9045212706278485e-05, "loss": 1.6428, "step": 544 }, { "epoch": 0.17, "learning_rate": 1.904095800904455e-05, "loss": 1.408, "step": 545 }, { "epoch": 0.17, "learning_rate": 1.9036694330458077e-05, "loss": 2.1169, "step": 546 }, { "epoch": 0.17, "learning_rate": 1.903242167475463e-05, "loss": 2.1308, "step": 547 }, { "epoch": 0.17, "learning_rate": 1.9028140046178696e-05, "loss": 1.5411, "step": 548 }, { "epoch": 0.17, "learning_rate": 1.9023849448983674e-05, "loss": 1.5414, "step": 549 }, { "epoch": 0.17, "learning_rate": 1.901954988743188e-05, "loss": 1.3376, "step": 550 }, { "epoch": 0.17, "learning_rate": 1.901524136579452e-05, "loss": 1.2951, "step": 551 }, { "epoch": 0.17, "learning_rate": 1.9010923888351717e-05, "loss": 2.0171, "step": 552 }, { "epoch": 0.17, "learning_rate": 1.9006597459392483e-05, "loss": 1.5098, "step": 553 }, { "epoch": 0.17, "learning_rate": 1.9002262083214728e-05, "loss": 1.6109, "step": 554 }, { "epoch": 0.17, "learning_rate": 1.8997917764125233e-05, "loss": 1.3259, "step": 555 }, { "epoch": 0.17, "learning_rate": 1.8993564506439687e-05, "loss": 1.756, "step": 556 }, { "epoch": 0.17, "learning_rate": 1.8989202314482645e-05, "loss": 1.4023, "step": 557 }, { "epoch": 0.17, "learning_rate": 1.898483119258754e-05, "loss": 1.2758, "step": 558 }, { "epoch": 0.17, "learning_rate": 1.898045114509667e-05, "loss": 1.4951, "step": 559 }, { "epoch": 0.17, "learning_rate": 1.8976062176361213e-05, "loss": 1.7553, "step": 560 }, { "epoch": 0.17, "learning_rate": 1.89716642907412e-05, "loss": 1.6671, "step": 561 }, { "epoch": 0.17, "learning_rate": 1.8967257492605522e-05, "loss": 1.5865, "step": 562 }, { "epoch": 0.17, "learning_rate": 1.8962841786331923e-05, "loss": 1.3889, "step": 563 }, { "epoch": 0.17, "learning_rate": 1.8958417176307e-05, "loss": 1.5483, "step": 564 }, { "epoch": 0.17, "learning_rate": 1.8953983666926182e-05, "loss": 1.9848, "step": 565 }, { "epoch": 0.17, "learning_rate": 1.8949541262593764e-05, "loss": 1.4827, "step": 566 }, { "epoch": 0.17, "learning_rate": 1.8945089967722853e-05, "loss": 2.1598, "step": 567 }, { "epoch": 0.17, "learning_rate": 1.89406297867354e-05, "loss": 1.2254, "step": 568 }, { "epoch": 0.18, "learning_rate": 1.8936160724062182e-05, "loss": 1.6542, "step": 569 }, { "epoch": 0.18, "learning_rate": 1.8931682784142792e-05, "loss": 1.8275, "step": 570 }, { "epoch": 0.18, "learning_rate": 1.8927195971425655e-05, "loss": 1.6244, "step": 571 }, { "epoch": 0.18, "learning_rate": 1.8922700290367998e-05, "loss": 1.6699, "step": 572 }, { "epoch": 0.18, "learning_rate": 1.8918195745435866e-05, "loss": 1.4768, "step": 573 }, { "epoch": 0.18, "learning_rate": 1.8913682341104107e-05, "loss": 1.7655, "step": 574 }, { "epoch": 0.18, "learning_rate": 1.8909160081856365e-05, "loss": 1.4945, "step": 575 }, { "epoch": 0.18, "learning_rate": 1.8904628972185086e-05, "loss": 1.5845, "step": 576 }, { "epoch": 0.18, "learning_rate": 1.890008901659151e-05, "loss": 1.3189, "step": 577 }, { "epoch": 0.18, "learning_rate": 1.8895540219585656e-05, "loss": 2.177, "step": 578 }, { "epoch": 0.18, "learning_rate": 1.889098258568634e-05, "loss": 2.121, "step": 579 }, { "epoch": 0.18, "learning_rate": 1.888641611942114e-05, "loss": 1.397, "step": 580 }, { "epoch": 0.18, "learning_rate": 1.8881840825326416e-05, "loss": 1.6375, "step": 581 }, { "epoch": 0.18, "learning_rate": 1.8877256707947308e-05, "loss": 1.2365, "step": 582 }, { "epoch": 0.18, "learning_rate": 1.88726637718377e-05, "loss": 1.2064, "step": 583 }, { "epoch": 0.18, "learning_rate": 1.8868062021560254e-05, "loss": 1.3946, "step": 584 }, { "epoch": 0.18, "learning_rate": 1.8863451461686376e-05, "loss": 1.5162, "step": 585 }, { "epoch": 0.18, "learning_rate": 1.885883209679624e-05, "loss": 1.3973, "step": 586 }, { "epoch": 0.18, "learning_rate": 1.8854203931478742e-05, "loss": 1.8254, "step": 587 }, { "epoch": 0.18, "learning_rate": 1.884956697033155e-05, "loss": 1.6801, "step": 588 }, { "epoch": 0.18, "learning_rate": 1.884492121796104e-05, "loss": 1.3981, "step": 589 }, { "epoch": 0.18, "learning_rate": 1.8840266678982343e-05, "loss": 1.8412, "step": 590 }, { "epoch": 0.18, "learning_rate": 1.8835603358019313e-05, "loss": 1.675, "step": 591 }, { "epoch": 0.18, "learning_rate": 1.883093125970452e-05, "loss": 2.1703, "step": 592 }, { "epoch": 0.18, "learning_rate": 1.882625038867927e-05, "loss": 1.2109, "step": 593 }, { "epoch": 0.18, "learning_rate": 1.8821560749593564e-05, "loss": 1.5727, "step": 594 }, { "epoch": 0.18, "learning_rate": 1.8816862347106125e-05, "loss": 1.9582, "step": 595 }, { "epoch": 0.18, "learning_rate": 1.8812155185884385e-05, "loss": 1.484, "step": 596 }, { "epoch": 0.18, "learning_rate": 1.880743927060447e-05, "loss": 1.6283, "step": 597 }, { "epoch": 0.18, "learning_rate": 1.88027146059512e-05, "loss": 1.7687, "step": 598 }, { "epoch": 0.18, "learning_rate": 1.8797981196618095e-05, "loss": 1.586, "step": 599 }, { "epoch": 0.18, "learning_rate": 1.8793239047307358e-05, "loss": 1.5589, "step": 600 }, { "epoch": 0.18, "learning_rate": 1.8788488162729868e-05, "loss": 1.625, "step": 601 }, { "epoch": 0.19, "learning_rate": 1.8783728547605192e-05, "loss": 1.4686, "step": 602 }, { "epoch": 0.19, "learning_rate": 1.8778960206661567e-05, "loss": 1.3376, "step": 603 }, { "epoch": 0.19, "learning_rate": 1.8774183144635894e-05, "loss": 1.5638, "step": 604 }, { "epoch": 0.19, "learning_rate": 1.876939736627375e-05, "loss": 1.6563, "step": 605 }, { "epoch": 0.19, "learning_rate": 1.8764602876329346e-05, "loss": 1.8871, "step": 606 }, { "epoch": 0.19, "learning_rate": 1.8759799679565572e-05, "loss": 1.6298, "step": 607 }, { "epoch": 0.19, "learning_rate": 1.875498778075396e-05, "loss": 1.4083, "step": 608 }, { "epoch": 0.19, "learning_rate": 1.8750167184674682e-05, "loss": 1.3672, "step": 609 }, { "epoch": 0.19, "learning_rate": 1.874533789611655e-05, "loss": 1.859, "step": 610 }, { "epoch": 0.19, "learning_rate": 1.8740499919877017e-05, "loss": 1.3689, "step": 611 }, { "epoch": 0.19, "learning_rate": 1.8735653260762165e-05, "loss": 1.4764, "step": 612 }, { "epoch": 0.19, "learning_rate": 1.8730797923586696e-05, "loss": 1.2668, "step": 613 }, { "epoch": 0.19, "learning_rate": 1.872593391317394e-05, "loss": 1.7082, "step": 614 }, { "epoch": 0.19, "learning_rate": 1.872106123435584e-05, "loss": 1.4882, "step": 615 }, { "epoch": 0.19, "learning_rate": 1.8716179891972942e-05, "loss": 1.3451, "step": 616 }, { "epoch": 0.19, "learning_rate": 1.8711289890874415e-05, "loss": 1.742, "step": 617 }, { "epoch": 0.19, "learning_rate": 1.8706391235918022e-05, "loss": 1.7175, "step": 618 }, { "epoch": 0.19, "learning_rate": 1.870148393197012e-05, "loss": 1.3208, "step": 619 }, { "epoch": 0.19, "learning_rate": 1.8696567983905655e-05, "loss": 1.6317, "step": 620 }, { "epoch": 0.19, "learning_rate": 1.869164339660817e-05, "loss": 1.4781, "step": 621 }, { "epoch": 0.19, "learning_rate": 1.8686710174969786e-05, "loss": 1.383, "step": 622 }, { "epoch": 0.19, "learning_rate": 1.8681768323891196e-05, "loss": 1.5551, "step": 623 }, { "epoch": 0.19, "learning_rate": 1.8676817848281677e-05, "loss": 1.2818, "step": 624 }, { "epoch": 0.19, "learning_rate": 1.867185875305906e-05, "loss": 1.3895, "step": 625 }, { "epoch": 0.19, "learning_rate": 1.8666891043149757e-05, "loss": 1.9428, "step": 626 }, { "epoch": 0.19, "learning_rate": 1.866191472348871e-05, "loss": 1.4577, "step": 627 }, { "epoch": 0.19, "learning_rate": 1.8656929799019444e-05, "loss": 1.4175, "step": 628 }, { "epoch": 0.19, "learning_rate": 1.8651936274694013e-05, "loss": 1.3815, "step": 629 }, { "epoch": 0.19, "learning_rate": 1.8646934155473025e-05, "loss": 1.4734, "step": 630 }, { "epoch": 0.19, "learning_rate": 1.8641923446325615e-05, "loss": 1.6124, "step": 631 }, { "epoch": 0.19, "learning_rate": 1.8636904152229455e-05, "loss": 1.3308, "step": 632 }, { "epoch": 0.19, "learning_rate": 1.8631876278170758e-05, "loss": 1.5259, "step": 633 }, { "epoch": 0.2, "learning_rate": 1.8626839829144244e-05, "loss": 1.5456, "step": 634 }, { "epoch": 0.2, "learning_rate": 1.8621794810153155e-05, "loss": 1.691, "step": 635 }, { "epoch": 0.2, "learning_rate": 1.8616741226209256e-05, "loss": 1.3078, "step": 636 }, { "epoch": 0.2, "learning_rate": 1.8611679082332808e-05, "loss": 1.3482, "step": 637 }, { "epoch": 0.2, "learning_rate": 1.8606608383552583e-05, "loss": 1.8421, "step": 638 }, { "epoch": 0.2, "learning_rate": 1.860152913490585e-05, "loss": 1.2417, "step": 639 }, { "epoch": 0.2, "learning_rate": 1.859644134143837e-05, "loss": 1.6078, "step": 640 }, { "epoch": 0.2, "learning_rate": 1.8591345008204394e-05, "loss": 1.3818, "step": 641 }, { "epoch": 0.2, "learning_rate": 1.8586240140266655e-05, "loss": 2.0646, "step": 642 }, { "epoch": 0.2, "learning_rate": 1.858112674269637e-05, "loss": 1.7184, "step": 643 }, { "epoch": 0.2, "learning_rate": 1.8576004820573217e-05, "loss": 1.7486, "step": 644 }, { "epoch": 0.2, "learning_rate": 1.8570874378985357e-05, "loss": 1.6406, "step": 645 }, { "epoch": 0.2, "learning_rate": 1.8565735423029406e-05, "loss": 1.7162, "step": 646 }, { "epoch": 0.2, "learning_rate": 1.8560587957810434e-05, "loss": 1.7694, "step": 647 }, { "epoch": 0.2, "learning_rate": 1.8555431988441977e-05, "loss": 1.6327, "step": 648 }, { "epoch": 0.2, "learning_rate": 1.855026752004601e-05, "loss": 1.4946, "step": 649 }, { "epoch": 0.2, "learning_rate": 1.854509455775295e-05, "loss": 1.5323, "step": 650 }, { "epoch": 0.2, "learning_rate": 1.8539913106701663e-05, "loss": 1.3565, "step": 651 }, { "epoch": 0.2, "learning_rate": 1.853472317203943e-05, "loss": 1.6078, "step": 652 }, { "epoch": 0.2, "learning_rate": 1.8529524758921973e-05, "loss": 1.7001, "step": 653 }, { "epoch": 0.2, "learning_rate": 1.8524317872513434e-05, "loss": 1.809, "step": 654 }, { "epoch": 0.2, "learning_rate": 1.851910251798637e-05, "loss": 1.8589, "step": 655 }, { "epoch": 0.2, "learning_rate": 1.8513878700521752e-05, "loss": 1.46, "step": 656 }, { "epoch": 0.2, "learning_rate": 1.8508646425308958e-05, "loss": 1.6257, "step": 657 }, { "epoch": 0.2, "learning_rate": 1.8503405697545767e-05, "loss": 1.5721, "step": 658 }, { "epoch": 0.2, "learning_rate": 1.8498156522438356e-05, "loss": 1.4022, "step": 659 }, { "epoch": 0.2, "learning_rate": 1.8492898905201297e-05, "loss": 1.6251, "step": 660 }, { "epoch": 0.2, "learning_rate": 1.8487632851057533e-05, "loss": 1.4298, "step": 661 }, { "epoch": 0.2, "learning_rate": 1.8482358365238414e-05, "loss": 1.7755, "step": 662 }, { "epoch": 0.2, "learning_rate": 1.8477075452983643e-05, "loss": 1.6327, "step": 663 }, { "epoch": 0.2, "learning_rate": 1.847178411954131e-05, "loss": 1.754, "step": 664 }, { "epoch": 0.2, "learning_rate": 1.8466484370167855e-05, "loss": 1.5889, "step": 665 }, { "epoch": 0.2, "learning_rate": 1.8461176210128095e-05, "loss": 1.7053, "step": 666 }, { "epoch": 0.21, "learning_rate": 1.8455859644695186e-05, "loss": 1.5127, "step": 667 }, { "epoch": 0.21, "learning_rate": 1.845053467915065e-05, "loss": 1.4308, "step": 668 }, { "epoch": 0.21, "learning_rate": 1.8445201318784343e-05, "loss": 1.8029, "step": 669 }, { "epoch": 0.21, "learning_rate": 1.8439859568894464e-05, "loss": 1.8827, "step": 670 }, { "epoch": 0.21, "learning_rate": 1.8434509434787544e-05, "loss": 1.3758, "step": 671 }, { "epoch": 0.21, "learning_rate": 1.8429150921778448e-05, "loss": 1.404, "step": 672 }, { "epoch": 0.21, "learning_rate": 1.8423784035190354e-05, "loss": 1.485, "step": 673 }, { "epoch": 0.21, "learning_rate": 1.8418408780354776e-05, "loss": 1.54, "step": 674 }, { "epoch": 0.21, "learning_rate": 1.8413025162611525e-05, "loss": 1.6564, "step": 675 }, { "epoch": 0.21, "learning_rate": 1.8407633187308722e-05, "loss": 1.6289, "step": 676 }, { "epoch": 0.21, "learning_rate": 1.84022328598028e-05, "loss": 1.7564, "step": 677 }, { "epoch": 0.21, "learning_rate": 1.839682418545848e-05, "loss": 1.2386, "step": 678 }, { "epoch": 0.21, "learning_rate": 1.8391407169648777e-05, "loss": 1.7526, "step": 679 }, { "epoch": 0.21, "learning_rate": 1.8385981817754997e-05, "loss": 2.0724, "step": 680 }, { "epoch": 0.21, "learning_rate": 1.8380548135166725e-05, "loss": 1.9446, "step": 681 }, { "epoch": 0.21, "learning_rate": 1.837510612728181e-05, "loss": 1.5604, "step": 682 }, { "epoch": 0.21, "learning_rate": 1.8369655799506394e-05, "loss": 1.7632, "step": 683 }, { "epoch": 0.21, "learning_rate": 1.8364197157254863e-05, "loss": 1.3337, "step": 684 }, { "epoch": 0.21, "learning_rate": 1.835873020594988e-05, "loss": 1.5513, "step": 685 }, { "epoch": 0.21, "learning_rate": 1.8353254951022342e-05, "loss": 1.5122, "step": 686 }, { "epoch": 0.21, "learning_rate": 1.8347771397911414e-05, "loss": 1.9496, "step": 687 }, { "epoch": 0.21, "learning_rate": 1.83422795520645e-05, "loss": 1.3472, "step": 688 }, { "epoch": 0.21, "learning_rate": 1.833677941893723e-05, "loss": 1.2167, "step": 689 }, { "epoch": 0.21, "learning_rate": 1.833127100399348e-05, "loss": 2.1145, "step": 690 }, { "epoch": 0.21, "learning_rate": 1.8325754312705354e-05, "loss": 1.837, "step": 691 }, { "epoch": 0.21, "learning_rate": 1.8320229350553165e-05, "loss": 1.3419, "step": 692 }, { "epoch": 0.21, "learning_rate": 1.8314696123025456e-05, "loss": 1.6396, "step": 693 }, { "epoch": 0.21, "learning_rate": 1.8309154635618967e-05, "loss": 1.3863, "step": 694 }, { "epoch": 0.21, "learning_rate": 1.830360489383866e-05, "loss": 1.5589, "step": 695 }, { "epoch": 0.21, "learning_rate": 1.8298046903197685e-05, "loss": 1.1815, "step": 696 }, { "epoch": 0.21, "learning_rate": 1.8292480669217387e-05, "loss": 1.5832, "step": 697 }, { "epoch": 0.21, "learning_rate": 1.828690619742731e-05, "loss": 1.5901, "step": 698 }, { "epoch": 0.22, "learning_rate": 1.8281323493365165e-05, "loss": 1.657, "step": 699 }, { "epoch": 0.22, "learning_rate": 1.8275732562576857e-05, "loss": 1.3439, "step": 700 }, { "epoch": 0.22, "learning_rate": 1.8270133410616453e-05, "loss": 1.8008, "step": 701 }, { "epoch": 0.22, "learning_rate": 1.8264526043046193e-05, "loss": 1.3573, "step": 702 }, { "epoch": 0.22, "learning_rate": 1.8258910465436476e-05, "loss": 1.4711, "step": 703 }, { "epoch": 0.22, "learning_rate": 1.8253286683365855e-05, "loss": 1.2365, "step": 704 }, { "epoch": 0.22, "learning_rate": 1.8247654702421037e-05, "loss": 1.6896, "step": 705 }, { "epoch": 0.22, "learning_rate": 1.8242014528196875e-05, "loss": 1.6622, "step": 706 }, { "epoch": 0.22, "learning_rate": 1.8236366166296356e-05, "loss": 1.7212, "step": 707 }, { "epoch": 0.22, "learning_rate": 1.8230709622330596e-05, "loss": 1.7019, "step": 708 }, { "epoch": 0.22, "learning_rate": 1.8225044901918862e-05, "loss": 1.8005, "step": 709 }, { "epoch": 0.22, "learning_rate": 1.8219372010688516e-05, "loss": 1.586, "step": 710 }, { "epoch": 0.22, "learning_rate": 1.821369095427505e-05, "loss": 1.8295, "step": 711 }, { "epoch": 0.22, "learning_rate": 1.820800173832207e-05, "loss": 1.2133, "step": 712 }, { "epoch": 0.22, "learning_rate": 1.8202304368481285e-05, "loss": 1.6126, "step": 713 }, { "epoch": 0.22, "learning_rate": 1.8196598850412496e-05, "loss": 1.5056, "step": 714 }, { "epoch": 0.22, "learning_rate": 1.819088518978361e-05, "loss": 1.4861, "step": 715 }, { "epoch": 0.22, "learning_rate": 1.8185163392270623e-05, "loss": 1.5699, "step": 716 }, { "epoch": 0.22, "learning_rate": 1.8179433463557603e-05, "loss": 1.6072, "step": 717 }, { "epoch": 0.22, "learning_rate": 1.8173695409336703e-05, "loss": 1.5056, "step": 718 }, { "epoch": 0.22, "learning_rate": 1.8167949235308148e-05, "loss": 1.5162, "step": 719 }, { "epoch": 0.22, "learning_rate": 1.8162194947180228e-05, "loss": 1.6402, "step": 720 }, { "epoch": 0.22, "learning_rate": 1.8156432550669298e-05, "loss": 1.6683, "step": 721 }, { "epoch": 0.22, "learning_rate": 1.815066205149976e-05, "loss": 1.6646, "step": 722 }, { "epoch": 0.22, "learning_rate": 1.8144883455404072e-05, "loss": 1.959, "step": 723 }, { "epoch": 0.22, "learning_rate": 1.8139096768122727e-05, "loss": 1.3213, "step": 724 }, { "epoch": 0.22, "learning_rate": 1.813330199540427e-05, "loss": 1.3985, "step": 725 }, { "epoch": 0.22, "learning_rate": 1.8127499143005266e-05, "loss": 1.4718, "step": 726 }, { "epoch": 0.22, "learning_rate": 1.812168821669031e-05, "loss": 1.2316, "step": 727 }, { "epoch": 0.22, "learning_rate": 1.8115869222232015e-05, "loss": 2.1794, "step": 728 }, { "epoch": 0.22, "learning_rate": 1.8110042165411016e-05, "loss": 1.6209, "step": 729 }, { "epoch": 0.22, "learning_rate": 1.8104207052015952e-05, "loss": 1.7872, "step": 730 }, { "epoch": 0.22, "learning_rate": 1.8098363887843467e-05, "loss": 1.4011, "step": 731 }, { "epoch": 0.23, "learning_rate": 1.80925126786982e-05, "loss": 1.5551, "step": 732 }, { "epoch": 0.23, "learning_rate": 1.8086653430392784e-05, "loss": 1.7005, "step": 733 }, { "epoch": 0.23, "learning_rate": 1.8080786148747842e-05, "loss": 1.366, "step": 734 }, { "epoch": 0.23, "learning_rate": 1.8074910839591972e-05, "loss": 1.555, "step": 735 }, { "epoch": 0.23, "learning_rate": 1.8069027508761747e-05, "loss": 1.5576, "step": 736 }, { "epoch": 0.23, "learning_rate": 1.8063136162101715e-05, "loss": 1.5748, "step": 737 }, { "epoch": 0.23, "learning_rate": 1.805723680546437e-05, "loss": 1.8539, "step": 738 }, { "epoch": 0.23, "learning_rate": 1.8051329444710187e-05, "loss": 1.4886, "step": 739 }, { "epoch": 0.23, "learning_rate": 1.8045414085707578e-05, "loss": 1.3864, "step": 740 }, { "epoch": 0.23, "learning_rate": 1.80394907343329e-05, "loss": 1.4891, "step": 741 }, { "epoch": 0.23, "learning_rate": 1.8033559396470455e-05, "loss": 1.7299, "step": 742 }, { "epoch": 0.23, "learning_rate": 1.8027620078012474e-05, "loss": 1.4252, "step": 743 }, { "epoch": 0.23, "learning_rate": 1.802167278485912e-05, "loss": 1.5474, "step": 744 }, { "epoch": 0.23, "learning_rate": 1.801571752291848e-05, "loss": 1.4145, "step": 745 }, { "epoch": 0.23, "learning_rate": 1.8009754298106553e-05, "loss": 1.5132, "step": 746 }, { "epoch": 0.23, "learning_rate": 1.8003783116347245e-05, "loss": 2.0066, "step": 747 }, { "epoch": 0.23, "learning_rate": 1.7997803983572372e-05, "loss": 1.62, "step": 748 }, { "epoch": 0.23, "learning_rate": 1.7991816905721654e-05, "loss": 1.4211, "step": 749 }, { "epoch": 0.23, "learning_rate": 1.7985821888742687e-05, "loss": 1.5473, "step": 750 }, { "epoch": 0.23, "learning_rate": 1.7979818938590967e-05, "loss": 1.5114, "step": 751 }, { "epoch": 0.23, "learning_rate": 1.7973808061229873e-05, "loss": 1.3383, "step": 752 }, { "epoch": 0.23, "learning_rate": 1.7967789262630647e-05, "loss": 1.616, "step": 753 }, { "epoch": 0.23, "learning_rate": 1.7961762548772413e-05, "loss": 1.7893, "step": 754 }, { "epoch": 0.23, "learning_rate": 1.7955727925642147e-05, "loss": 1.4786, "step": 755 }, { "epoch": 0.23, "learning_rate": 1.794968539923469e-05, "loss": 1.6082, "step": 756 }, { "epoch": 0.23, "learning_rate": 1.794363497555273e-05, "loss": 1.5085, "step": 757 }, { "epoch": 0.23, "learning_rate": 1.79375766606068e-05, "loss": 1.2891, "step": 758 }, { "epoch": 0.23, "learning_rate": 1.7931510460415272e-05, "loss": 1.6728, "step": 759 }, { "epoch": 0.23, "learning_rate": 1.7925436381004356e-05, "loss": 1.4013, "step": 760 }, { "epoch": 0.23, "learning_rate": 1.7919354428408086e-05, "loss": 1.8139, "step": 761 }, { "epoch": 0.23, "learning_rate": 1.7913264608668316e-05, "loss": 1.4135, "step": 762 }, { "epoch": 0.23, "learning_rate": 1.7907166927834717e-05, "loss": 1.4739, "step": 763 }, { "epoch": 0.24, "learning_rate": 1.790106139196477e-05, "loss": 1.303, "step": 764 }, { "epoch": 0.24, "learning_rate": 1.789494800712376e-05, "loss": 1.6892, "step": 765 }, { "epoch": 0.24, "learning_rate": 1.788882677938476e-05, "loss": 1.3484, "step": 766 }, { "epoch": 0.24, "learning_rate": 1.7882697714828647e-05, "loss": 1.5391, "step": 767 }, { "epoch": 0.24, "learning_rate": 1.7876560819544077e-05, "loss": 1.3175, "step": 768 }, { "epoch": 0.24, "learning_rate": 1.7870416099627488e-05, "loss": 1.5041, "step": 769 }, { "epoch": 0.24, "learning_rate": 1.7864263561183085e-05, "loss": 1.7336, "step": 770 }, { "epoch": 0.24, "learning_rate": 1.785810321032284e-05, "loss": 1.6633, "step": 771 }, { "epoch": 0.24, "learning_rate": 1.7851935053166503e-05, "loss": 1.6733, "step": 772 }, { "epoch": 0.24, "learning_rate": 1.7845759095841553e-05, "loss": 1.9953, "step": 773 }, { "epoch": 0.24, "learning_rate": 1.7839575344483237e-05, "loss": 1.371, "step": 774 }, { "epoch": 0.24, "learning_rate": 1.7833383805234536e-05, "loss": 1.3293, "step": 775 }, { "epoch": 0.24, "learning_rate": 1.782718448424617e-05, "loss": 1.4687, "step": 776 }, { "epoch": 0.24, "learning_rate": 1.7820977387676583e-05, "loss": 1.587, "step": 777 }, { "epoch": 0.24, "learning_rate": 1.781476252169196e-05, "loss": 1.3963, "step": 778 }, { "epoch": 0.24, "learning_rate": 1.7808539892466186e-05, "loss": 1.3674, "step": 779 }, { "epoch": 0.24, "learning_rate": 1.7802309506180867e-05, "loss": 1.4119, "step": 780 }, { "epoch": 0.24, "learning_rate": 1.779607136902531e-05, "loss": 1.4019, "step": 781 }, { "epoch": 0.24, "learning_rate": 1.7789825487196538e-05, "loss": 1.4235, "step": 782 }, { "epoch": 0.24, "learning_rate": 1.7783571866899242e-05, "loss": 1.6729, "step": 783 }, { "epoch": 0.24, "learning_rate": 1.7777310514345816e-05, "loss": 2.3911, "step": 784 }, { "epoch": 0.24, "learning_rate": 1.7771041435756332e-05, "loss": 1.5293, "step": 785 }, { "epoch": 0.24, "learning_rate": 1.7764764637358545e-05, "loss": 1.5793, "step": 786 }, { "epoch": 0.24, "learning_rate": 1.775848012538786e-05, "loss": 1.7615, "step": 787 }, { "epoch": 0.24, "learning_rate": 1.7752187906087366e-05, "loss": 1.8085, "step": 788 }, { "epoch": 0.24, "learning_rate": 1.7745887985707796e-05, "loss": 1.6918, "step": 789 }, { "epoch": 0.24, "learning_rate": 1.7739580370507533e-05, "loss": 1.8163, "step": 790 }, { "epoch": 0.24, "learning_rate": 1.7733265066752606e-05, "loss": 1.7952, "step": 791 }, { "epoch": 0.24, "learning_rate": 1.7726942080716693e-05, "loss": 1.5377, "step": 792 }, { "epoch": 0.24, "learning_rate": 1.7720611418681082e-05, "loss": 1.4194, "step": 793 }, { "epoch": 0.24, "learning_rate": 1.7714273086934705e-05, "loss": 1.4771, "step": 794 }, { "epoch": 0.24, "learning_rate": 1.7707927091774103e-05, "loss": 1.5729, "step": 795 }, { "epoch": 0.24, "learning_rate": 1.7701573439503437e-05, "loss": 2.0678, "step": 796 }, { "epoch": 0.25, "learning_rate": 1.769521213643447e-05, "loss": 1.1961, "step": 797 }, { "epoch": 0.25, "learning_rate": 1.768884318888656e-05, "loss": 2.1071, "step": 798 }, { "epoch": 0.25, "learning_rate": 1.768246660318667e-05, "loss": 1.6993, "step": 799 }, { "epoch": 0.25, "learning_rate": 1.7676082385669343e-05, "loss": 1.6004, "step": 800 }, { "epoch": 0.25, "learning_rate": 1.7669690542676713e-05, "loss": 1.4411, "step": 801 }, { "epoch": 0.25, "learning_rate": 1.7663291080558477e-05, "loss": 1.7872, "step": 802 }, { "epoch": 0.25, "learning_rate": 1.7656884005671912e-05, "loss": 1.7064, "step": 803 }, { "epoch": 0.25, "learning_rate": 1.7650469324381843e-05, "loss": 1.4028, "step": 804 }, { "epoch": 0.25, "learning_rate": 1.764404704306067e-05, "loss": 1.5014, "step": 805 }, { "epoch": 0.25, "learning_rate": 1.7637617168088327e-05, "loss": 1.4999, "step": 806 }, { "epoch": 0.25, "learning_rate": 1.76311797058523e-05, "loss": 1.5959, "step": 807 }, { "epoch": 0.25, "learning_rate": 1.762473466274761e-05, "loss": 1.6539, "step": 808 }, { "epoch": 0.25, "learning_rate": 1.7618282045176812e-05, "loss": 1.3852, "step": 809 }, { "epoch": 0.25, "learning_rate": 1.7611821859549977e-05, "loss": 1.5617, "step": 810 }, { "epoch": 0.25, "learning_rate": 1.76053541122847e-05, "loss": 1.3193, "step": 811 }, { "epoch": 0.25, "learning_rate": 1.7598878809806095e-05, "loss": 1.3467, "step": 812 }, { "epoch": 0.25, "learning_rate": 1.7592395958546767e-05, "loss": 1.4551, "step": 813 }, { "epoch": 0.25, "learning_rate": 1.758590556494683e-05, "loss": 1.4005, "step": 814 }, { "epoch": 0.25, "learning_rate": 1.7579407635453884e-05, "loss": 1.4037, "step": 815 }, { "epoch": 0.25, "learning_rate": 1.7572902176523023e-05, "loss": 1.5949, "step": 816 }, { "epoch": 0.25, "learning_rate": 1.7566389194616816e-05, "loss": 1.4334, "step": 817 }, { "epoch": 0.25, "learning_rate": 1.7559868696205304e-05, "loss": 1.9066, "step": 818 }, { "epoch": 0.25, "learning_rate": 1.7553340687765996e-05, "loss": 1.6524, "step": 819 }, { "epoch": 0.25, "learning_rate": 1.7546805175783866e-05, "loss": 1.8763, "step": 820 }, { "epoch": 0.25, "learning_rate": 1.7540262166751336e-05, "loss": 1.458, "step": 821 }, { "epoch": 0.25, "learning_rate": 1.753371166716828e-05, "loss": 1.3904, "step": 822 }, { "epoch": 0.25, "learning_rate": 1.7527153683542007e-05, "loss": 1.4798, "step": 823 }, { "epoch": 0.25, "learning_rate": 1.752058822238727e-05, "loss": 1.3741, "step": 824 }, { "epoch": 0.25, "learning_rate": 1.7514015290226244e-05, "loss": 2.6011, "step": 825 }, { "epoch": 0.25, "learning_rate": 1.7507434893588525e-05, "loss": 1.5167, "step": 826 }, { "epoch": 0.25, "learning_rate": 1.7500847039011128e-05, "loss": 1.4155, "step": 827 }, { "epoch": 0.25, "learning_rate": 1.7494251733038473e-05, "loss": 1.4533, "step": 828 }, { "epoch": 0.26, "learning_rate": 1.748764898222239e-05, "loss": 1.5373, "step": 829 }, { "epoch": 0.26, "learning_rate": 1.748103879312209e-05, "loss": 1.5178, "step": 830 }, { "epoch": 0.26, "learning_rate": 1.7474421172304188e-05, "loss": 1.244, "step": 831 }, { "epoch": 0.26, "learning_rate": 1.7467796126342677e-05, "loss": 1.6269, "step": 832 }, { "epoch": 0.26, "learning_rate": 1.7461163661818917e-05, "loss": 1.4472, "step": 833 }, { "epoch": 0.26, "learning_rate": 1.745452378532166e-05, "loss": 1.652, "step": 834 }, { "epoch": 0.26, "learning_rate": 1.7447876503446996e-05, "loss": 1.8686, "step": 835 }, { "epoch": 0.26, "learning_rate": 1.7441221822798385e-05, "loss": 1.5792, "step": 836 }, { "epoch": 0.26, "learning_rate": 1.7434559749986633e-05, "loss": 1.3301, "step": 837 }, { "epoch": 0.26, "learning_rate": 1.7427890291629895e-05, "loss": 1.4879, "step": 838 }, { "epoch": 0.26, "learning_rate": 1.7421213454353657e-05, "loss": 1.7992, "step": 839 }, { "epoch": 0.26, "learning_rate": 1.7414529244790733e-05, "loss": 1.3126, "step": 840 }, { "epoch": 0.26, "learning_rate": 1.7407837669581274e-05, "loss": 1.6731, "step": 841 }, { "epoch": 0.26, "learning_rate": 1.740113873537273e-05, "loss": 1.5731, "step": 842 }, { "epoch": 0.26, "learning_rate": 1.7394432448819874e-05, "loss": 1.7039, "step": 843 }, { "epoch": 0.26, "learning_rate": 1.7387718816584775e-05, "loss": 1.5279, "step": 844 }, { "epoch": 0.26, "learning_rate": 1.738099784533681e-05, "loss": 1.4238, "step": 845 }, { "epoch": 0.26, "learning_rate": 1.737426954175264e-05, "loss": 1.3752, "step": 846 }, { "epoch": 0.26, "learning_rate": 1.7367533912516203e-05, "loss": 1.452, "step": 847 }, { "epoch": 0.26, "learning_rate": 1.7360790964318727e-05, "loss": 1.3888, "step": 848 }, { "epoch": 0.26, "learning_rate": 1.7354040703858707e-05, "loss": 1.7618, "step": 849 }, { "epoch": 0.26, "learning_rate": 1.734728313784189e-05, "loss": 1.9281, "step": 850 }, { "epoch": 0.26, "learning_rate": 1.7340518272981304e-05, "loss": 1.5503, "step": 851 }, { "epoch": 0.26, "learning_rate": 1.7333746115997204e-05, "loss": 1.5289, "step": 852 }, { "epoch": 0.26, "learning_rate": 1.7326966673617102e-05, "loss": 1.1858, "step": 853 }, { "epoch": 0.26, "learning_rate": 1.732017995257575e-05, "loss": 1.2547, "step": 854 }, { "epoch": 0.26, "learning_rate": 1.731338595961512e-05, "loss": 1.5849, "step": 855 }, { "epoch": 0.26, "learning_rate": 1.7306584701484414e-05, "loss": 1.832, "step": 856 }, { "epoch": 0.26, "learning_rate": 1.729977618494005e-05, "loss": 1.524, "step": 857 }, { "epoch": 0.26, "learning_rate": 1.7292960416745657e-05, "loss": 1.5202, "step": 858 }, { "epoch": 0.26, "learning_rate": 1.728613740367207e-05, "loss": 1.396, "step": 859 }, { "epoch": 0.26, "learning_rate": 1.7279307152497316e-05, "loss": 1.7042, "step": 860 }, { "epoch": 0.26, "learning_rate": 1.7272469670006625e-05, "loss": 1.5634, "step": 861 }, { "epoch": 0.27, "learning_rate": 1.726562496299239e-05, "loss": 2.8179, "step": 862 }, { "epoch": 0.27, "learning_rate": 1.7258773038254194e-05, "loss": 1.6097, "step": 863 }, { "epoch": 0.27, "learning_rate": 1.7251913902598795e-05, "loss": 1.5126, "step": 864 }, { "epoch": 0.27, "learning_rate": 1.72450475628401e-05, "loss": 1.2417, "step": 865 }, { "epoch": 0.27, "learning_rate": 1.7238174025799183e-05, "loss": 1.848, "step": 866 }, { "epoch": 0.27, "learning_rate": 1.7231293298304265e-05, "loss": 1.4678, "step": 867 }, { "epoch": 0.27, "learning_rate": 1.722440538719071e-05, "loss": 2.2675, "step": 868 }, { "epoch": 0.27, "learning_rate": 1.721751029930102e-05, "loss": 1.1174, "step": 869 }, { "epoch": 0.27, "learning_rate": 1.721060804148482e-05, "loss": 1.6167, "step": 870 }, { "epoch": 0.27, "learning_rate": 1.720369862059887e-05, "loss": 1.4021, "step": 871 }, { "epoch": 0.27, "learning_rate": 1.719678204350703e-05, "loss": 1.8564, "step": 872 }, { "epoch": 0.27, "learning_rate": 1.718985831708028e-05, "loss": 1.4963, "step": 873 }, { "epoch": 0.27, "learning_rate": 1.7182927448196705e-05, "loss": 1.3176, "step": 874 }, { "epoch": 0.27, "learning_rate": 1.7175989443741472e-05, "loss": 1.6269, "step": 875 }, { "epoch": 0.27, "learning_rate": 1.716904431060685e-05, "loss": 1.3289, "step": 876 }, { "epoch": 0.27, "learning_rate": 1.716209205569218e-05, "loss": 1.897, "step": 877 }, { "epoch": 0.27, "learning_rate": 1.7155132685903888e-05, "loss": 1.7436, "step": 878 }, { "epoch": 0.27, "learning_rate": 1.7148166208155453e-05, "loss": 1.5632, "step": 879 }, { "epoch": 0.27, "learning_rate": 1.714119262936743e-05, "loss": 1.454, "step": 880 }, { "epoch": 0.27, "learning_rate": 1.7134211956467425e-05, "loss": 1.6819, "step": 881 }, { "epoch": 0.27, "learning_rate": 1.712722419639008e-05, "loss": 1.491, "step": 882 }, { "epoch": 0.27, "learning_rate": 1.7120229356077095e-05, "loss": 1.9001, "step": 883 }, { "epoch": 0.27, "learning_rate": 1.7113227442477182e-05, "loss": 1.4566, "step": 884 }, { "epoch": 0.27, "learning_rate": 1.7106218462546107e-05, "loss": 1.6, "step": 885 }, { "epoch": 0.27, "learning_rate": 1.7099202423246632e-05, "loss": 1.4392, "step": 886 }, { "epoch": 0.27, "learning_rate": 1.7092179331548545e-05, "loss": 1.8749, "step": 887 }, { "epoch": 0.27, "learning_rate": 1.7085149194428628e-05, "loss": 1.3524, "step": 888 }, { "epoch": 0.27, "learning_rate": 1.707811201887068e-05, "loss": 1.5274, "step": 889 }, { "epoch": 0.27, "learning_rate": 1.7071067811865477e-05, "loss": 1.7023, "step": 890 }, { "epoch": 0.27, "learning_rate": 1.7064016580410782e-05, "loss": 1.4988, "step": 891 }, { "epoch": 0.27, "learning_rate": 1.7056958331511343e-05, "loss": 1.5491, "step": 892 }, { "epoch": 0.27, "learning_rate": 1.7049893072178874e-05, "loss": 1.6425, "step": 893 }, { "epoch": 0.28, "learning_rate": 1.7042820809432057e-05, "loss": 1.3709, "step": 894 }, { "epoch": 0.28, "learning_rate": 1.7035741550296522e-05, "loss": 1.5589, "step": 895 }, { "epoch": 0.28, "learning_rate": 1.7028655301804858e-05, "loss": 1.3701, "step": 896 }, { "epoch": 0.28, "learning_rate": 1.7021562070996597e-05, "loss": 1.4948, "step": 897 }, { "epoch": 0.28, "learning_rate": 1.7014461864918205e-05, "loss": 1.439, "step": 898 }, { "epoch": 0.28, "learning_rate": 1.7007354690623082e-05, "loss": 1.3605, "step": 899 }, { "epoch": 0.28, "learning_rate": 1.7000240555171536e-05, "loss": 1.7092, "step": 900 }, { "epoch": 0.28, "learning_rate": 1.6993119465630806e-05, "loss": 1.6787, "step": 901 }, { "epoch": 0.28, "learning_rate": 1.6985991429075038e-05, "loss": 1.4352, "step": 902 }, { "epoch": 0.28, "learning_rate": 1.6978856452585266e-05, "loss": 1.1501, "step": 903 }, { "epoch": 0.28, "learning_rate": 1.697171454324943e-05, "loss": 1.6532, "step": 904 }, { "epoch": 0.28, "learning_rate": 1.6964565708162354e-05, "loss": 1.9605, "step": 905 }, { "epoch": 0.28, "learning_rate": 1.6957409954425747e-05, "loss": 1.4744, "step": 906 }, { "epoch": 0.28, "learning_rate": 1.6950247289148176e-05, "loss": 1.6303, "step": 907 }, { "epoch": 0.28, "learning_rate": 1.694307771944509e-05, "loss": 1.3183, "step": 908 }, { "epoch": 0.28, "learning_rate": 1.6935901252438794e-05, "loss": 1.6299, "step": 909 }, { "epoch": 0.28, "learning_rate": 1.692871789525844e-05, "loss": 1.6264, "step": 910 }, { "epoch": 0.28, "learning_rate": 1.692152765504002e-05, "loss": 1.6717, "step": 911 }, { "epoch": 0.28, "learning_rate": 1.6914330538926377e-05, "loss": 1.7613, "step": 912 }, { "epoch": 0.28, "learning_rate": 1.6907126554067175e-05, "loss": 1.4282, "step": 913 }, { "epoch": 0.28, "learning_rate": 1.6899915707618905e-05, "loss": 1.2316, "step": 914 }, { "epoch": 0.28, "learning_rate": 1.6892698006744876e-05, "loss": 1.5535, "step": 915 }, { "epoch": 0.28, "learning_rate": 1.6885473458615204e-05, "loss": 1.514, "step": 916 }, { "epoch": 0.28, "learning_rate": 1.6878242070406797e-05, "loss": 1.4689, "step": 917 }, { "epoch": 0.28, "learning_rate": 1.687100384930338e-05, "loss": 1.58, "step": 918 }, { "epoch": 0.28, "learning_rate": 1.686375880249545e-05, "loss": 1.5177, "step": 919 }, { "epoch": 0.28, "learning_rate": 1.685650693718029e-05, "loss": 1.3925, "step": 920 }, { "epoch": 0.28, "learning_rate": 1.6849248260561947e-05, "loss": 1.3467, "step": 921 }, { "epoch": 0.28, "learning_rate": 1.6841982779851254e-05, "loss": 1.4362, "step": 922 }, { "epoch": 0.28, "learning_rate": 1.683471050226578e-05, "loss": 2.2228, "step": 923 }, { "epoch": 0.28, "learning_rate": 1.682743143502987e-05, "loss": 1.5136, "step": 924 }, { "epoch": 0.28, "learning_rate": 1.6820145585374588e-05, "loss": 1.7973, "step": 925 }, { "epoch": 0.28, "learning_rate": 1.6812852960537763e-05, "loss": 1.4953, "step": 926 }, { "epoch": 0.29, "learning_rate": 1.680555356776393e-05, "loss": 1.5916, "step": 927 }, { "epoch": 0.29, "learning_rate": 1.6798247414304363e-05, "loss": 2.7003, "step": 928 }, { "epoch": 0.29, "learning_rate": 1.679093450741705e-05, "loss": 1.3493, "step": 929 }, { "epoch": 0.29, "learning_rate": 1.678361485436668e-05, "loss": 1.7298, "step": 930 }, { "epoch": 0.29, "learning_rate": 1.677628846242465e-05, "loss": 1.5449, "step": 931 }, { "epoch": 0.29, "learning_rate": 1.676895533886905e-05, "loss": 1.4672, "step": 932 }, { "epoch": 0.29, "learning_rate": 1.676161549098466e-05, "loss": 1.5761, "step": 933 }, { "epoch": 0.29, "learning_rate": 1.6754268926062936e-05, "loss": 1.4477, "step": 934 }, { "epoch": 0.29, "learning_rate": 1.6746915651402006e-05, "loss": 1.5947, "step": 935 }, { "epoch": 0.29, "learning_rate": 1.6739555674306675e-05, "loss": 1.5135, "step": 936 }, { "epoch": 0.29, "learning_rate": 1.6732189002088386e-05, "loss": 1.698, "step": 937 }, { "epoch": 0.29, "learning_rate": 1.6724815642065245e-05, "loss": 1.3135, "step": 938 }, { "epoch": 0.29, "learning_rate": 1.671743560156201e-05, "loss": 1.4586, "step": 939 }, { "epoch": 0.29, "learning_rate": 1.6710048887910058e-05, "loss": 1.6851, "step": 940 }, { "epoch": 0.29, "learning_rate": 1.6702655508447402e-05, "loss": 1.4078, "step": 941 }, { "epoch": 0.29, "learning_rate": 1.6695255470518687e-05, "loss": 1.4554, "step": 942 }, { "epoch": 0.29, "learning_rate": 1.6687848781475156e-05, "loss": 1.6047, "step": 943 }, { "epoch": 0.29, "learning_rate": 1.6680435448674673e-05, "loss": 1.4915, "step": 944 }, { "epoch": 0.29, "learning_rate": 1.6673015479481688e-05, "loss": 1.7276, "step": 945 }, { "epoch": 0.29, "learning_rate": 1.666558888126726e-05, "loss": 1.425, "step": 946 }, { "epoch": 0.29, "learning_rate": 1.6658155661409017e-05, "loss": 1.4183, "step": 947 }, { "epoch": 0.29, "learning_rate": 1.6650715827291178e-05, "loss": 1.4812, "step": 948 }, { "epoch": 0.29, "learning_rate": 1.6643269386304526e-05, "loss": 1.382, "step": 949 }, { "epoch": 0.29, "learning_rate": 1.6635816345846413e-05, "loss": 1.5144, "step": 950 }, { "epoch": 0.29, "learning_rate": 1.6628356713320737e-05, "loss": 1.4955, "step": 951 }, { "epoch": 0.29, "learning_rate": 1.6620890496137953e-05, "loss": 1.5889, "step": 952 }, { "epoch": 0.29, "learning_rate": 1.6613417701715056e-05, "loss": 1.3152, "step": 953 }, { "epoch": 0.29, "learning_rate": 1.660593833747557e-05, "loss": 1.7075, "step": 954 }, { "epoch": 0.29, "learning_rate": 1.659845241084955e-05, "loss": 1.7521, "step": 955 }, { "epoch": 0.29, "learning_rate": 1.6590959929273574e-05, "loss": 1.7181, "step": 956 }, { "epoch": 0.29, "learning_rate": 1.658346090019072e-05, "loss": 1.6183, "step": 957 }, { "epoch": 0.29, "learning_rate": 1.6575955331050585e-05, "loss": 1.6797, "step": 958 }, { "epoch": 0.3, "learning_rate": 1.6568443229309247e-05, "loss": 1.66, "step": 959 }, { "epoch": 0.3, "learning_rate": 1.656092460242929e-05, "loss": 1.5973, "step": 960 }, { "epoch": 0.3, "learning_rate": 1.655339945787976e-05, "loss": 1.8312, "step": 961 }, { "epoch": 0.3, "learning_rate": 1.6545867803136203e-05, "loss": 1.4946, "step": 962 }, { "epoch": 0.3, "learning_rate": 1.6538329645680612e-05, "loss": 1.4063, "step": 963 }, { "epoch": 0.3, "learning_rate": 1.6530784993001448e-05, "loss": 1.3378, "step": 964 }, { "epoch": 0.3, "learning_rate": 1.6523233852593622e-05, "loss": 1.5978, "step": 965 }, { "epoch": 0.3, "learning_rate": 1.6515676231958488e-05, "loss": 1.4422, "step": 966 }, { "epoch": 0.3, "learning_rate": 1.650811213860385e-05, "loss": 2.9863, "step": 967 }, { "epoch": 0.3, "learning_rate": 1.650054158004392e-05, "loss": 1.4816, "step": 968 }, { "epoch": 0.3, "learning_rate": 1.649296456379936e-05, "loss": 1.3404, "step": 969 }, { "epoch": 0.3, "learning_rate": 1.6485381097397223e-05, "loss": 1.7107, "step": 970 }, { "epoch": 0.3, "learning_rate": 1.6477791188370978e-05, "loss": 1.4126, "step": 971 }, { "epoch": 0.3, "learning_rate": 1.6470194844260497e-05, "loss": 1.4862, "step": 972 }, { "epoch": 0.3, "learning_rate": 1.6462592072612044e-05, "loss": 1.4365, "step": 973 }, { "epoch": 0.3, "learning_rate": 1.645498288097827e-05, "loss": 1.7108, "step": 974 }, { "epoch": 0.3, "learning_rate": 1.644736727691819e-05, "loss": 1.4595, "step": 975 }, { "epoch": 0.3, "learning_rate": 1.643974526799721e-05, "loss": 1.6194, "step": 976 }, { "epoch": 0.3, "learning_rate": 1.643211686178708e-05, "loss": 1.5961, "step": 977 }, { "epoch": 0.3, "learning_rate": 1.6424482065865925e-05, "loss": 1.5277, "step": 978 }, { "epoch": 0.3, "learning_rate": 1.6416840887818195e-05, "loss": 1.5029, "step": 979 }, { "epoch": 0.3, "learning_rate": 1.6409193335234698e-05, "loss": 1.7965, "step": 980 }, { "epoch": 0.3, "learning_rate": 1.640153941571256e-05, "loss": 1.2555, "step": 981 }, { "epoch": 0.3, "learning_rate": 1.6393879136855247e-05, "loss": 1.7234, "step": 982 }, { "epoch": 0.3, "learning_rate": 1.6386212506272533e-05, "loss": 1.5772, "step": 983 }, { "epoch": 0.3, "learning_rate": 1.6378539531580502e-05, "loss": 1.4364, "step": 984 }, { "epoch": 0.3, "learning_rate": 1.6370860220401547e-05, "loss": 1.3704, "step": 985 }, { "epoch": 0.3, "learning_rate": 1.636317458036434e-05, "loss": 1.1916, "step": 986 }, { "epoch": 0.3, "learning_rate": 1.6355482619103866e-05, "loss": 1.4641, "step": 987 }, { "epoch": 0.3, "learning_rate": 1.6347784344261365e-05, "loss": 1.4288, "step": 988 }, { "epoch": 0.3, "learning_rate": 1.6340079763484357e-05, "loss": 1.7138, "step": 989 }, { "epoch": 0.3, "learning_rate": 1.633236888442663e-05, "loss": 1.9642, "step": 990 }, { "epoch": 0.3, "learning_rate": 1.6324651714748226e-05, "loss": 1.6564, "step": 991 }, { "epoch": 0.31, "learning_rate": 1.6316928262115443e-05, "loss": 1.6127, "step": 992 }, { "epoch": 0.31, "learning_rate": 1.6309198534200807e-05, "loss": 1.3114, "step": 993 }, { "epoch": 0.31, "learning_rate": 1.6301462538683087e-05, "loss": 1.9242, "step": 994 }, { "epoch": 0.31, "learning_rate": 1.6293720283247276e-05, "loss": 2.0184, "step": 995 }, { "epoch": 0.31, "learning_rate": 1.6285971775584587e-05, "loss": 1.5211, "step": 996 }, { "epoch": 0.31, "learning_rate": 1.6278217023392447e-05, "loss": 1.2897, "step": 997 }, { "epoch": 0.31, "learning_rate": 1.6270456034374477e-05, "loss": 1.6037, "step": 998 }, { "epoch": 0.31, "learning_rate": 1.62626888162405e-05, "loss": 1.7091, "step": 999 }, { "epoch": 0.31, "learning_rate": 1.6254915376706526e-05, "loss": 1.7262, "step": 1000 }, { "epoch": 0.31, "learning_rate": 1.624713572349475e-05, "loss": 1.5866, "step": 1001 }, { "epoch": 0.31, "learning_rate": 1.6239349864333527e-05, "loss": 1.532, "step": 1002 }, { "epoch": 0.31, "learning_rate": 1.623155780695739e-05, "loss": 1.3556, "step": 1003 }, { "epoch": 0.31, "learning_rate": 1.6223759559107028e-05, "loss": 1.5362, "step": 1004 }, { "epoch": 0.31, "learning_rate": 1.6215955128529273e-05, "loss": 1.68, "step": 1005 }, { "epoch": 0.31, "learning_rate": 1.62081445229771e-05, "loss": 1.285, "step": 1006 }, { "epoch": 0.31, "learning_rate": 1.620032775020963e-05, "loss": 1.5264, "step": 1007 }, { "epoch": 0.31, "learning_rate": 1.6192504817992086e-05, "loss": 1.4866, "step": 1008 }, { "epoch": 0.31, "learning_rate": 1.6184675734095838e-05, "loss": 1.6023, "step": 1009 }, { "epoch": 0.31, "learning_rate": 1.6176840506298345e-05, "loss": 1.5414, "step": 1010 }, { "epoch": 0.31, "learning_rate": 1.6168999142383187e-05, "loss": 1.6073, "step": 1011 }, { "epoch": 0.31, "learning_rate": 1.6161151650140027e-05, "loss": 1.4433, "step": 1012 }, { "epoch": 0.31, "learning_rate": 1.6153298037364617e-05, "loss": 1.6553, "step": 1013 }, { "epoch": 0.31, "learning_rate": 1.61454383118588e-05, "loss": 1.4277, "step": 1014 }, { "epoch": 0.31, "learning_rate": 1.6137572481430474e-05, "loss": 1.4453, "step": 1015 }, { "epoch": 0.31, "learning_rate": 1.612970055389362e-05, "loss": 1.3933, "step": 1016 }, { "epoch": 0.31, "learning_rate": 1.6121822537068263e-05, "loss": 1.6303, "step": 1017 }, { "epoch": 0.31, "learning_rate": 1.611393843878048e-05, "loss": 1.9715, "step": 1018 }, { "epoch": 0.31, "learning_rate": 1.6106048266862395e-05, "loss": 1.383, "step": 1019 }, { "epoch": 0.31, "learning_rate": 1.609815202915216e-05, "loss": 1.6517, "step": 1020 }, { "epoch": 0.31, "learning_rate": 1.6090249733493952e-05, "loss": 1.5899, "step": 1021 }, { "epoch": 0.31, "learning_rate": 1.608234138773797e-05, "loss": 1.3725, "step": 1022 }, { "epoch": 0.31, "learning_rate": 1.6074426999740422e-05, "loss": 1.651, "step": 1023 }, { "epoch": 0.32, "learning_rate": 1.6066506577363518e-05, "loss": 1.7959, "step": 1024 }, { "epoch": 0.32, "learning_rate": 1.6058580128475458e-05, "loss": 1.5919, "step": 1025 }, { "epoch": 0.32, "learning_rate": 1.605064766095044e-05, "loss": 1.3681, "step": 1026 }, { "epoch": 0.32, "learning_rate": 1.6042709182668628e-05, "loss": 2.5635, "step": 1027 }, { "epoch": 0.32, "learning_rate": 1.603476470151617e-05, "loss": 1.5505, "step": 1028 }, { "epoch": 0.32, "learning_rate": 1.6026814225385163e-05, "loss": 1.4891, "step": 1029 }, { "epoch": 0.32, "learning_rate": 1.6018857762173672e-05, "loss": 1.5213, "step": 1030 }, { "epoch": 0.32, "learning_rate": 1.60108953197857e-05, "loss": 1.7509, "step": 1031 }, { "epoch": 0.32, "learning_rate": 1.6002926906131202e-05, "loss": 1.4602, "step": 1032 }, { "epoch": 0.32, "learning_rate": 1.599495252912605e-05, "loss": 1.6008, "step": 1033 }, { "epoch": 0.32, "learning_rate": 1.5986972196692053e-05, "loss": 1.4041, "step": 1034 }, { "epoch": 0.32, "learning_rate": 1.5978985916756925e-05, "loss": 1.4373, "step": 1035 }, { "epoch": 0.32, "learning_rate": 1.5970993697254294e-05, "loss": 1.6824, "step": 1036 }, { "epoch": 0.32, "learning_rate": 1.5962995546123696e-05, "loss": 1.463, "step": 1037 }, { "epoch": 0.32, "learning_rate": 1.5954991471310542e-05, "loss": 1.6836, "step": 1038 }, { "epoch": 0.32, "learning_rate": 1.5946981480766146e-05, "loss": 1.3966, "step": 1039 }, { "epoch": 0.32, "learning_rate": 1.5938965582447683e-05, "loss": 1.0201, "step": 1040 }, { "epoch": 0.32, "learning_rate": 1.593094378431821e-05, "loss": 1.6136, "step": 1041 }, { "epoch": 0.32, "learning_rate": 1.5922916094346633e-05, "loss": 1.3993, "step": 1042 }, { "epoch": 0.32, "learning_rate": 1.5914882520507716e-05, "loss": 2.2084, "step": 1043 }, { "epoch": 0.32, "learning_rate": 1.5906843070782077e-05, "loss": 1.3329, "step": 1044 }, { "epoch": 0.32, "learning_rate": 1.5898797753156156e-05, "loss": 1.3613, "step": 1045 }, { "epoch": 0.32, "learning_rate": 1.589074657562223e-05, "loss": 1.5029, "step": 1046 }, { "epoch": 0.32, "learning_rate": 1.58826895461784e-05, "loss": 1.2901, "step": 1047 }, { "epoch": 0.32, "learning_rate": 1.5874626672828573e-05, "loss": 1.564, "step": 1048 }, { "epoch": 0.32, "learning_rate": 1.586655796358247e-05, "loss": 1.3343, "step": 1049 }, { "epoch": 0.32, "learning_rate": 1.58584834264556e-05, "loss": 1.7643, "step": 1050 }, { "epoch": 0.32, "learning_rate": 1.585040306946927e-05, "loss": 1.864, "step": 1051 }, { "epoch": 0.32, "learning_rate": 1.584231690065056e-05, "loss": 1.7747, "step": 1052 }, { "epoch": 0.32, "learning_rate": 1.583422492803233e-05, "loss": 1.9124, "step": 1053 }, { "epoch": 0.32, "learning_rate": 1.5826127159653203e-05, "loss": 1.5124, "step": 1054 }, { "epoch": 0.32, "learning_rate": 1.581802360355756e-05, "loss": 1.8209, "step": 1055 }, { "epoch": 0.32, "learning_rate": 1.5809914267795542e-05, "loss": 1.6177, "step": 1056 }, { "epoch": 0.33, "learning_rate": 1.5801799160423004e-05, "loss": 1.2172, "step": 1057 }, { "epoch": 0.33, "learning_rate": 1.5793678289501564e-05, "loss": 1.1287, "step": 1058 }, { "epoch": 0.33, "learning_rate": 1.5785551663098546e-05, "loss": 1.5319, "step": 1059 }, { "epoch": 0.33, "learning_rate": 1.5777419289287008e-05, "loss": 1.5833, "step": 1060 }, { "epoch": 0.33, "learning_rate": 1.57692811761457e-05, "loss": 1.5145, "step": 1061 }, { "epoch": 0.33, "learning_rate": 1.5761137331759084e-05, "loss": 1.556, "step": 1062 }, { "epoch": 0.33, "learning_rate": 1.5752987764217317e-05, "loss": 1.3611, "step": 1063 }, { "epoch": 0.33, "learning_rate": 1.5744832481616235e-05, "loss": 1.5687, "step": 1064 }, { "epoch": 0.33, "learning_rate": 1.5736671492057355e-05, "loss": 1.6609, "step": 1065 }, { "epoch": 0.33, "learning_rate": 1.5728504803647857e-05, "loss": 1.9621, "step": 1066 }, { "epoch": 0.33, "learning_rate": 1.57203324245006e-05, "loss": 1.5601, "step": 1067 }, { "epoch": 0.33, "learning_rate": 1.5712154362734073e-05, "loss": 1.3043, "step": 1068 }, { "epoch": 0.33, "learning_rate": 1.5703970626472424e-05, "loss": 1.3164, "step": 1069 }, { "epoch": 0.33, "learning_rate": 1.5695781223845442e-05, "loss": 1.3817, "step": 1070 }, { "epoch": 0.33, "learning_rate": 1.568758616298853e-05, "loss": 1.4874, "step": 1071 }, { "epoch": 0.33, "learning_rate": 1.5679385452042724e-05, "loss": 1.3455, "step": 1072 }, { "epoch": 0.33, "learning_rate": 1.567117909915467e-05, "loss": 1.8071, "step": 1073 }, { "epoch": 0.33, "learning_rate": 1.5662967112476613e-05, "loss": 1.3661, "step": 1074 }, { "epoch": 0.33, "learning_rate": 1.56547495001664e-05, "loss": 1.7409, "step": 1075 }, { "epoch": 0.33, "learning_rate": 1.5646526270387473e-05, "loss": 1.55, "step": 1076 }, { "epoch": 0.33, "learning_rate": 1.5638297431308846e-05, "loss": 1.31, "step": 1077 }, { "epoch": 0.33, "learning_rate": 1.56300629911051e-05, "loss": 1.6774, "step": 1078 }, { "epoch": 0.33, "learning_rate": 1.562182295795639e-05, "loss": 1.6859, "step": 1079 }, { "epoch": 0.33, "learning_rate": 1.5613577340048423e-05, "loss": 1.8374, "step": 1080 }, { "epoch": 0.33, "learning_rate": 1.560532614557246e-05, "loss": 2.1071, "step": 1081 }, { "epoch": 0.33, "learning_rate": 1.5597069382725293e-05, "loss": 1.1481, "step": 1082 }, { "epoch": 0.33, "learning_rate": 1.558880705970925e-05, "loss": 1.5379, "step": 1083 }, { "epoch": 0.33, "learning_rate": 1.5580539184732188e-05, "loss": 1.2301, "step": 1084 }, { "epoch": 0.33, "learning_rate": 1.557226576600746e-05, "loss": 1.4623, "step": 1085 }, { "epoch": 0.33, "learning_rate": 1.5563986811753948e-05, "loss": 1.4972, "step": 1086 }, { "epoch": 0.33, "learning_rate": 1.5555702330196024e-05, "loss": 1.512, "step": 1087 }, { "epoch": 0.33, "learning_rate": 1.554741232956355e-05, "loss": 1.4794, "step": 1088 }, { "epoch": 0.34, "learning_rate": 1.5539116818091873e-05, "loss": 1.6693, "step": 1089 }, { "epoch": 0.34, "learning_rate": 1.553081580402182e-05, "loss": 1.439, "step": 1090 }, { "epoch": 0.34, "learning_rate": 1.5522509295599662e-05, "loss": 1.4569, "step": 1091 }, { "epoch": 0.34, "learning_rate": 1.551419730107716e-05, "loss": 1.4267, "step": 1092 }, { "epoch": 0.34, "learning_rate": 1.5505879828711495e-05, "loss": 1.9064, "step": 1093 }, { "epoch": 0.34, "learning_rate": 1.5497556886765316e-05, "loss": 1.5052, "step": 1094 }, { "epoch": 0.34, "learning_rate": 1.548922848350669e-05, "loss": 1.5837, "step": 1095 }, { "epoch": 0.34, "learning_rate": 1.548089462720911e-05, "loss": 1.4804, "step": 1096 }, { "epoch": 0.34, "learning_rate": 1.5472555326151488e-05, "loss": 1.1238, "step": 1097 }, { "epoch": 0.34, "learning_rate": 1.5464210588618148e-05, "loss": 1.4484, "step": 1098 }, { "epoch": 0.34, "learning_rate": 1.5455860422898813e-05, "loss": 1.6398, "step": 1099 }, { "epoch": 0.34, "learning_rate": 1.544750483728859e-05, "loss": 1.9449, "step": 1100 }, { "epoch": 0.34, "learning_rate": 1.543914384008799e-05, "loss": 1.6514, "step": 1101 }, { "epoch": 0.34, "learning_rate": 1.5430777439602875e-05, "loss": 1.4899, "step": 1102 }, { "epoch": 0.34, "learning_rate": 1.54224056441445e-05, "loss": 1.3312, "step": 1103 }, { "epoch": 0.34, "learning_rate": 1.5414028462029454e-05, "loss": 1.6147, "step": 1104 }, { "epoch": 0.34, "learning_rate": 1.54056459015797e-05, "loss": 1.2962, "step": 1105 }, { "epoch": 0.34, "learning_rate": 1.5397257971122527e-05, "loss": 1.382, "step": 1106 }, { "epoch": 0.34, "learning_rate": 1.5388864678990568e-05, "loss": 2.0166, "step": 1107 }, { "epoch": 0.34, "learning_rate": 1.5380466033521783e-05, "loss": 1.2341, "step": 1108 }, { "epoch": 0.34, "learning_rate": 1.5372062043059436e-05, "loss": 1.6534, "step": 1109 }, { "epoch": 0.34, "learning_rate": 1.536365271595212e-05, "loss": 1.5405, "step": 1110 }, { "epoch": 0.34, "learning_rate": 1.5355238060553727e-05, "loss": 1.6008, "step": 1111 }, { "epoch": 0.34, "learning_rate": 1.5346818085223423e-05, "loss": 1.6252, "step": 1112 }, { "epoch": 0.34, "learning_rate": 1.5338392798325673e-05, "loss": 1.4996, "step": 1113 }, { "epoch": 0.34, "learning_rate": 1.532996220823023e-05, "loss": 1.4685, "step": 1114 }, { "epoch": 0.34, "learning_rate": 1.532152632331209e-05, "loss": 1.5744, "step": 1115 }, { "epoch": 0.34, "learning_rate": 1.5313085151951527e-05, "loss": 1.3161, "step": 1116 }, { "epoch": 0.34, "learning_rate": 1.5304638702534056e-05, "loss": 1.5155, "step": 1117 }, { "epoch": 0.34, "learning_rate": 1.529618698345045e-05, "loss": 1.4947, "step": 1118 }, { "epoch": 0.34, "learning_rate": 1.52877300030967e-05, "loss": 1.3002, "step": 1119 }, { "epoch": 0.34, "learning_rate": 1.5279267769874034e-05, "loss": 1.2234, "step": 1120 }, { "epoch": 0.34, "learning_rate": 1.527080029218889e-05, "loss": 1.342, "step": 1121 }, { "epoch": 0.35, "learning_rate": 1.5262327578452927e-05, "loss": 1.5391, "step": 1122 }, { "epoch": 0.35, "learning_rate": 1.5253849637082997e-05, "loss": 1.3691, "step": 1123 }, { "epoch": 0.35, "learning_rate": 1.5245366476501151e-05, "loss": 1.6297, "step": 1124 }, { "epoch": 0.35, "learning_rate": 1.5236878105134617e-05, "loss": 1.5292, "step": 1125 }, { "epoch": 0.35, "learning_rate": 1.5228384531415809e-05, "loss": 1.2432, "step": 1126 }, { "epoch": 0.35, "learning_rate": 1.5219885763782301e-05, "loss": 1.5855, "step": 1127 }, { "epoch": 0.35, "learning_rate": 1.5211381810676826e-05, "loss": 1.6148, "step": 1128 }, { "epoch": 0.35, "learning_rate": 1.5202872680547276e-05, "loss": 1.4231, "step": 1129 }, { "epoch": 0.35, "learning_rate": 1.5194358381846686e-05, "loss": 1.3389, "step": 1130 }, { "epoch": 0.35, "learning_rate": 1.5185838923033216e-05, "loss": 1.7355, "step": 1131 }, { "epoch": 0.35, "learning_rate": 1.5177314312570155e-05, "loss": 1.7297, "step": 1132 }, { "epoch": 0.35, "learning_rate": 1.516878455892592e-05, "loss": 1.5918, "step": 1133 }, { "epoch": 0.35, "learning_rate": 1.5160249670574026e-05, "loss": 1.6845, "step": 1134 }, { "epoch": 0.35, "learning_rate": 1.515170965599309e-05, "loss": 1.3584, "step": 1135 }, { "epoch": 0.35, "learning_rate": 1.5143164523666824e-05, "loss": 1.7132, "step": 1136 }, { "epoch": 0.35, "learning_rate": 1.5134614282084025e-05, "loss": 1.4241, "step": 1137 }, { "epoch": 0.35, "learning_rate": 1.5126058939738566e-05, "loss": 1.4952, "step": 1138 }, { "epoch": 0.35, "learning_rate": 1.5117498505129384e-05, "loss": 1.3988, "step": 1139 }, { "epoch": 0.35, "learning_rate": 1.510893298676048e-05, "loss": 1.4212, "step": 1140 }, { "epoch": 0.35, "learning_rate": 1.5100362393140891e-05, "loss": 1.5234, "step": 1141 }, { "epoch": 0.35, "learning_rate": 1.5091786732784717e-05, "loss": 2.0149, "step": 1142 }, { "epoch": 0.35, "learning_rate": 1.5083206014211071e-05, "loss": 1.5635, "step": 1143 }, { "epoch": 0.35, "learning_rate": 1.5074620245944112e-05, "loss": 1.5342, "step": 1144 }, { "epoch": 0.35, "learning_rate": 1.5066029436512992e-05, "loss": 1.4296, "step": 1145 }, { "epoch": 0.35, "learning_rate": 1.5057433594451891e-05, "loss": 1.6421, "step": 1146 }, { "epoch": 0.35, "learning_rate": 1.5048832728299978e-05, "loss": 1.2417, "step": 1147 }, { "epoch": 0.35, "learning_rate": 1.5040226846601413e-05, "loss": 1.6521, "step": 1148 }, { "epoch": 0.35, "learning_rate": 1.5031615957905343e-05, "loss": 1.5733, "step": 1149 }, { "epoch": 0.35, "learning_rate": 1.5023000070765886e-05, "loss": 1.4894, "step": 1150 }, { "epoch": 0.35, "learning_rate": 1.5014379193742123e-05, "loss": 1.6537, "step": 1151 }, { "epoch": 0.35, "learning_rate": 1.50057533353981e-05, "loss": 1.9206, "step": 1152 }, { "epoch": 0.35, "learning_rate": 1.4997122504302803e-05, "loss": 1.5018, "step": 1153 }, { "epoch": 0.36, "learning_rate": 1.4988486709030166e-05, "loss": 1.5288, "step": 1154 }, { "epoch": 0.36, "learning_rate": 1.4979845958159045e-05, "loss": 1.5589, "step": 1155 }, { "epoch": 0.36, "learning_rate": 1.4971200260273223e-05, "loss": 1.2415, "step": 1156 }, { "epoch": 0.36, "learning_rate": 1.4962549623961401e-05, "loss": 1.7756, "step": 1157 }, { "epoch": 0.36, "learning_rate": 1.495389405781719e-05, "loss": 1.7568, "step": 1158 }, { "epoch": 0.36, "learning_rate": 1.4945233570439079e-05, "loss": 1.4947, "step": 1159 }, { "epoch": 0.36, "learning_rate": 1.4936568170430465e-05, "loss": 1.6302, "step": 1160 }, { "epoch": 0.36, "learning_rate": 1.4927897866399619e-05, "loss": 1.6416, "step": 1161 }, { "epoch": 0.36, "learning_rate": 1.491922266695968e-05, "loss": 1.3903, "step": 1162 }, { "epoch": 0.36, "learning_rate": 1.4910542580728656e-05, "loss": 1.4154, "step": 1163 }, { "epoch": 0.36, "learning_rate": 1.4901857616329405e-05, "loss": 1.6626, "step": 1164 }, { "epoch": 0.36, "learning_rate": 1.4893167782389635e-05, "loss": 1.3699, "step": 1165 }, { "epoch": 0.36, "learning_rate": 1.488447308754189e-05, "loss": 1.2841, "step": 1166 }, { "epoch": 0.36, "learning_rate": 1.4875773540423538e-05, "loss": 1.8124, "step": 1167 }, { "epoch": 0.36, "learning_rate": 1.4867069149676774e-05, "loss": 1.6663, "step": 1168 }, { "epoch": 0.36, "learning_rate": 1.4858359923948604e-05, "loss": 1.1577, "step": 1169 }, { "epoch": 0.36, "learning_rate": 1.4849645871890832e-05, "loss": 1.7759, "step": 1170 }, { "epoch": 0.36, "learning_rate": 1.4840927002160062e-05, "loss": 1.2419, "step": 1171 }, { "epoch": 0.36, "learning_rate": 1.4832203323417679e-05, "loss": 1.3198, "step": 1172 }, { "epoch": 0.36, "learning_rate": 1.4823474844329852e-05, "loss": 1.5915, "step": 1173 }, { "epoch": 0.36, "learning_rate": 1.4814741573567514e-05, "loss": 1.5611, "step": 1174 }, { "epoch": 0.36, "learning_rate": 1.4806003519806363e-05, "loss": 1.842, "step": 1175 }, { "epoch": 0.36, "learning_rate": 1.4797260691726831e-05, "loss": 1.4088, "step": 1176 }, { "epoch": 0.36, "learning_rate": 1.4788513098014114e-05, "loss": 1.5827, "step": 1177 }, { "epoch": 0.36, "learning_rate": 1.4779760747358141e-05, "loss": 1.7068, "step": 1178 }, { "epoch": 0.36, "learning_rate": 1.4771003648453548e-05, "loss": 1.7779, "step": 1179 }, { "epoch": 0.36, "learning_rate": 1.4762241809999704e-05, "loss": 1.3589, "step": 1180 }, { "epoch": 0.36, "learning_rate": 1.475347524070069e-05, "loss": 1.7083, "step": 1181 }, { "epoch": 0.36, "learning_rate": 1.4744703949265268e-05, "loss": 1.4768, "step": 1182 }, { "epoch": 0.36, "learning_rate": 1.4735927944406909e-05, "loss": 1.5142, "step": 1183 }, { "epoch": 0.36, "learning_rate": 1.4727147234843752e-05, "loss": 1.8684, "step": 1184 }, { "epoch": 0.36, "learning_rate": 1.4718361829298626e-05, "loss": 1.4434, "step": 1185 }, { "epoch": 0.36, "learning_rate": 1.4709571736499008e-05, "loss": 1.508, "step": 1186 }, { "epoch": 0.37, "learning_rate": 1.4700776965177044e-05, "loss": 1.7898, "step": 1187 }, { "epoch": 0.37, "learning_rate": 1.4691977524069518e-05, "loss": 1.5498, "step": 1188 }, { "epoch": 0.37, "learning_rate": 1.4683173421917865e-05, "loss": 1.5826, "step": 1189 }, { "epoch": 0.37, "learning_rate": 1.467436466746814e-05, "loss": 1.973, "step": 1190 }, { "epoch": 0.37, "learning_rate": 1.4665551269471021e-05, "loss": 1.8257, "step": 1191 }, { "epoch": 0.37, "learning_rate": 1.4656733236681802e-05, "loss": 1.7267, "step": 1192 }, { "epoch": 0.37, "learning_rate": 1.4647910577860378e-05, "loss": 1.7979, "step": 1193 }, { "epoch": 0.37, "learning_rate": 1.4639083301771246e-05, "loss": 1.9642, "step": 1194 }, { "epoch": 0.37, "learning_rate": 1.4630251417183483e-05, "loss": 1.1606, "step": 1195 }, { "epoch": 0.37, "learning_rate": 1.4621414932870746e-05, "loss": 2.0086, "step": 1196 }, { "epoch": 0.37, "learning_rate": 1.4612573857611265e-05, "loss": 1.6921, "step": 1197 }, { "epoch": 0.37, "learning_rate": 1.4603728200187823e-05, "loss": 1.868, "step": 1198 }, { "epoch": 0.37, "learning_rate": 1.4594877969387761e-05, "loss": 1.6571, "step": 1199 }, { "epoch": 0.37, "learning_rate": 1.4586023174002961e-05, "loss": 1.5016, "step": 1200 }, { "epoch": 0.37, "learning_rate": 1.4577163822829845e-05, "loss": 1.5835, "step": 1201 }, { "epoch": 0.37, "learning_rate": 1.4568299924669351e-05, "loss": 1.893, "step": 1202 }, { "epoch": 0.37, "learning_rate": 1.4559431488326943e-05, "loss": 1.5338, "step": 1203 }, { "epoch": 0.37, "learning_rate": 1.4550558522612584e-05, "loss": 1.2956, "step": 1204 }, { "epoch": 0.37, "learning_rate": 1.4541681036340744e-05, "loss": 1.4081, "step": 1205 }, { "epoch": 0.37, "learning_rate": 1.4532799038330385e-05, "loss": 1.6298, "step": 1206 }, { "epoch": 0.37, "learning_rate": 1.4523912537404945e-05, "loss": 1.1129, "step": 1207 }, { "epoch": 0.37, "learning_rate": 1.4515021542392335e-05, "loss": 1.7905, "step": 1208 }, { "epoch": 0.37, "learning_rate": 1.4506126062124936e-05, "loss": 1.076, "step": 1209 }, { "epoch": 0.37, "learning_rate": 1.4497226105439586e-05, "loss": 1.2592, "step": 1210 }, { "epoch": 0.37, "learning_rate": 1.4488321681177558e-05, "loss": 1.4406, "step": 1211 }, { "epoch": 0.37, "learning_rate": 1.4479412798184576e-05, "loss": 1.6368, "step": 1212 }, { "epoch": 0.37, "learning_rate": 1.4470499465310786e-05, "loss": 1.5226, "step": 1213 }, { "epoch": 0.37, "learning_rate": 1.4461581691410757e-05, "loss": 1.418, "step": 1214 }, { "epoch": 0.37, "learning_rate": 1.4452659485343473e-05, "loss": 1.4452, "step": 1215 }, { "epoch": 0.37, "learning_rate": 1.4443732855972307e-05, "loss": 1.5885, "step": 1216 }, { "epoch": 0.37, "learning_rate": 1.4434801812165052e-05, "loss": 1.7341, "step": 1217 }, { "epoch": 0.37, "learning_rate": 1.4425866362793863e-05, "loss": 1.5881, "step": 1218 }, { "epoch": 0.38, "learning_rate": 1.4416926516735274e-05, "loss": 1.5061, "step": 1219 }, { "epoch": 0.38, "learning_rate": 1.4407982282870203e-05, "loss": 1.3805, "step": 1220 }, { "epoch": 0.38, "learning_rate": 1.4399033670083911e-05, "loss": 1.2547, "step": 1221 }, { "epoch": 0.38, "learning_rate": 1.4390080687266013e-05, "loss": 1.4346, "step": 1222 }, { "epoch": 0.38, "learning_rate": 1.4381123343310468e-05, "loss": 1.6429, "step": 1223 }, { "epoch": 0.38, "learning_rate": 1.4372161647115563e-05, "loss": 1.716, "step": 1224 }, { "epoch": 0.38, "learning_rate": 1.4363195607583919e-05, "loss": 1.6579, "step": 1225 }, { "epoch": 0.38, "learning_rate": 1.4354225233622457e-05, "loss": 1.4573, "step": 1226 }, { "epoch": 0.38, "learning_rate": 1.4345250534142408e-05, "loss": 1.443, "step": 1227 }, { "epoch": 0.38, "learning_rate": 1.433627151805931e-05, "loss": 1.7023, "step": 1228 }, { "epoch": 0.38, "learning_rate": 1.4327288194292978e-05, "loss": 1.1989, "step": 1229 }, { "epoch": 0.38, "learning_rate": 1.4318300571767514e-05, "loss": 1.4702, "step": 1230 }, { "epoch": 0.38, "learning_rate": 1.4309308659411282e-05, "loss": 2.2549, "step": 1231 }, { "epoch": 0.38, "learning_rate": 1.4300312466156917e-05, "loss": 1.369, "step": 1232 }, { "epoch": 0.38, "learning_rate": 1.42913120009413e-05, "loss": 1.2485, "step": 1233 }, { "epoch": 0.38, "learning_rate": 1.4282307272705555e-05, "loss": 1.8576, "step": 1234 }, { "epoch": 0.38, "learning_rate": 1.4273298290395043e-05, "loss": 1.5776, "step": 1235 }, { "epoch": 0.38, "learning_rate": 1.4264285062959358e-05, "loss": 1.5215, "step": 1236 }, { "epoch": 0.38, "learning_rate": 1.4255267599352301e-05, "loss": 1.4871, "step": 1237 }, { "epoch": 0.38, "learning_rate": 1.4246245908531883e-05, "loss": 1.4785, "step": 1238 }, { "epoch": 0.38, "learning_rate": 1.4237219999460323e-05, "loss": 1.5851, "step": 1239 }, { "epoch": 0.38, "learning_rate": 1.4228189881104016e-05, "loss": 1.5691, "step": 1240 }, { "epoch": 0.38, "learning_rate": 1.4219155562433551e-05, "loss": 1.4772, "step": 1241 }, { "epoch": 0.38, "learning_rate": 1.4210117052423683e-05, "loss": 1.6606, "step": 1242 }, { "epoch": 0.38, "learning_rate": 1.4201074360053332e-05, "loss": 1.7508, "step": 1243 }, { "epoch": 0.38, "learning_rate": 1.4192027494305578e-05, "loss": 1.7585, "step": 1244 }, { "epoch": 0.38, "learning_rate": 1.4182976464167637e-05, "loss": 1.712, "step": 1245 }, { "epoch": 0.38, "learning_rate": 1.4173921278630874e-05, "loss": 1.4973, "step": 1246 }, { "epoch": 0.38, "learning_rate": 1.4164861946690765e-05, "loss": 1.4005, "step": 1247 }, { "epoch": 0.38, "learning_rate": 1.4155798477346922e-05, "loss": 1.4932, "step": 1248 }, { "epoch": 0.38, "learning_rate": 1.4146730879603063e-05, "loss": 1.4863, "step": 1249 }, { "epoch": 0.38, "learning_rate": 1.4137659162466999e-05, "loss": 1.6575, "step": 1250 }, { "epoch": 0.38, "learning_rate": 1.4128583334950638e-05, "loss": 1.2822, "step": 1251 }, { "epoch": 0.39, "learning_rate": 1.411950340606998e-05, "loss": 1.8933, "step": 1252 }, { "epoch": 0.39, "learning_rate": 1.4110419384845086e-05, "loss": 1.678, "step": 1253 }, { "epoch": 0.39, "learning_rate": 1.410133128030009e-05, "loss": 1.7881, "step": 1254 }, { "epoch": 0.39, "learning_rate": 1.4092239101463174e-05, "loss": 1.5135, "step": 1255 }, { "epoch": 0.39, "learning_rate": 1.408314285736658e-05, "loss": 1.4438, "step": 1256 }, { "epoch": 0.39, "learning_rate": 1.4074042557046584e-05, "loss": 1.8626, "step": 1257 }, { "epoch": 0.39, "learning_rate": 1.4064938209543483e-05, "loss": 1.4916, "step": 1258 }, { "epoch": 0.39, "learning_rate": 1.4055829823901604e-05, "loss": 1.3214, "step": 1259 }, { "epoch": 0.39, "learning_rate": 1.4046717409169288e-05, "loss": 1.202, "step": 1260 }, { "epoch": 0.39, "learning_rate": 1.4037600974398863e-05, "loss": 1.3464, "step": 1261 }, { "epoch": 0.39, "learning_rate": 1.4028480528646669e-05, "loss": 1.493, "step": 1262 }, { "epoch": 0.39, "learning_rate": 1.4019356080973017e-05, "loss": 1.3624, "step": 1263 }, { "epoch": 0.39, "learning_rate": 1.4010227640442202e-05, "loss": 1.7877, "step": 1264 }, { "epoch": 0.39, "learning_rate": 1.400109521612248e-05, "loss": 1.2741, "step": 1265 }, { "epoch": 0.39, "learning_rate": 1.3991958817086068e-05, "loss": 1.6242, "step": 1266 }, { "epoch": 0.39, "learning_rate": 1.3982818452409132e-05, "loss": 1.5145, "step": 1267 }, { "epoch": 0.39, "learning_rate": 1.3973674131171775e-05, "loss": 1.5785, "step": 1268 }, { "epoch": 0.39, "learning_rate": 1.3964525862458029e-05, "loss": 1.5778, "step": 1269 }, { "epoch": 0.39, "learning_rate": 1.3955373655355852e-05, "loss": 1.1616, "step": 1270 }, { "epoch": 0.39, "learning_rate": 1.394621751895711e-05, "loss": 1.96, "step": 1271 }, { "epoch": 0.39, "learning_rate": 1.3937057462357578e-05, "loss": 1.4356, "step": 1272 }, { "epoch": 0.39, "learning_rate": 1.392789349465692e-05, "loss": 1.7084, "step": 1273 }, { "epoch": 0.39, "learning_rate": 1.3918725624958684e-05, "loss": 1.2878, "step": 1274 }, { "epoch": 0.39, "learning_rate": 1.3909553862370305e-05, "loss": 1.3199, "step": 1275 }, { "epoch": 0.39, "learning_rate": 1.3900378216003073e-05, "loss": 1.7056, "step": 1276 }, { "epoch": 0.39, "learning_rate": 1.3891198694972139e-05, "loss": 1.581, "step": 1277 }, { "epoch": 0.39, "learning_rate": 1.3882015308396508e-05, "loss": 1.5768, "step": 1278 }, { "epoch": 0.39, "learning_rate": 1.3872828065399023e-05, "loss": 1.198, "step": 1279 }, { "epoch": 0.39, "learning_rate": 1.3863636975106354e-05, "loss": 1.7849, "step": 1280 }, { "epoch": 0.39, "learning_rate": 1.3854442046649002e-05, "loss": 1.2813, "step": 1281 }, { "epoch": 0.39, "learning_rate": 1.3845243289161263e-05, "loss": 1.2747, "step": 1282 }, { "epoch": 0.39, "learning_rate": 1.3836040711781263e-05, "loss": 1.8253, "step": 1283 }, { "epoch": 0.4, "learning_rate": 1.3826834323650899e-05, "loss": 1.4281, "step": 1284 }, { "epoch": 0.4, "learning_rate": 1.3817624133915865e-05, "loss": 1.3276, "step": 1285 }, { "epoch": 0.4, "learning_rate": 1.3808410151725633e-05, "loss": 1.4582, "step": 1286 }, { "epoch": 0.4, "learning_rate": 1.3799192386233434e-05, "loss": 1.592, "step": 1287 }, { "epoch": 0.4, "learning_rate": 1.3789970846596268e-05, "loss": 1.2669, "step": 1288 }, { "epoch": 0.4, "learning_rate": 1.3780745541974871e-05, "loss": 1.3246, "step": 1289 }, { "epoch": 0.4, "learning_rate": 1.3771516481533733e-05, "loss": 1.9186, "step": 1290 }, { "epoch": 0.4, "learning_rate": 1.3762283674441066e-05, "loss": 1.5179, "step": 1291 }, { "epoch": 0.4, "learning_rate": 1.3753047129868808e-05, "loss": 1.4538, "step": 1292 }, { "epoch": 0.4, "learning_rate": 1.374380685699261e-05, "loss": 1.4632, "step": 1293 }, { "epoch": 0.4, "learning_rate": 1.3734562864991823e-05, "loss": 1.7178, "step": 1294 }, { "epoch": 0.4, "learning_rate": 1.3725315163049495e-05, "loss": 1.8489, "step": 1295 }, { "epoch": 0.4, "learning_rate": 1.3716063760352359e-05, "loss": 1.4246, "step": 1296 }, { "epoch": 0.4, "learning_rate": 1.3706808666090827e-05, "loss": 1.6744, "step": 1297 }, { "epoch": 0.4, "learning_rate": 1.3697549889458974e-05, "loss": 1.5797, "step": 1298 }, { "epoch": 0.4, "learning_rate": 1.3688287439654537e-05, "loss": 1.4504, "step": 1299 }, { "epoch": 0.4, "learning_rate": 1.3679021325878901e-05, "loss": 1.4808, "step": 1300 }, { "epoch": 0.4, "learning_rate": 1.3669751557337085e-05, "loss": 1.5905, "step": 1301 }, { "epoch": 0.4, "learning_rate": 1.3660478143237748e-05, "loss": 1.5264, "step": 1302 }, { "epoch": 0.4, "learning_rate": 1.3651201092793164e-05, "loss": 1.5836, "step": 1303 }, { "epoch": 0.4, "learning_rate": 1.3641920415219224e-05, "loss": 1.5105, "step": 1304 }, { "epoch": 0.4, "learning_rate": 1.3632636119735416e-05, "loss": 1.2502, "step": 1305 }, { "epoch": 0.4, "learning_rate": 1.3623348215564827e-05, "loss": 1.6512, "step": 1306 }, { "epoch": 0.4, "learning_rate": 1.361405671193413e-05, "loss": 1.4239, "step": 1307 }, { "epoch": 0.4, "learning_rate": 1.3604761618073567e-05, "loss": 1.3912, "step": 1308 }, { "epoch": 0.4, "learning_rate": 1.3595462943216959e-05, "loss": 1.613, "step": 1309 }, { "epoch": 0.4, "learning_rate": 1.3586160696601667e-05, "loss": 1.3493, "step": 1310 }, { "epoch": 0.4, "learning_rate": 1.3576854887468614e-05, "loss": 2.049, "step": 1311 }, { "epoch": 0.4, "learning_rate": 1.3567545525062256e-05, "loss": 1.7991, "step": 1312 }, { "epoch": 0.4, "learning_rate": 1.3558232618630582e-05, "loss": 1.1772, "step": 1313 }, { "epoch": 0.4, "learning_rate": 1.3548916177425095e-05, "loss": 1.5329, "step": 1314 }, { "epoch": 0.4, "learning_rate": 1.3539596210700825e-05, "loss": 1.4228, "step": 1315 }, { "epoch": 0.4, "learning_rate": 1.3530272727716286e-05, "loss": 1.2907, "step": 1316 }, { "epoch": 0.41, "learning_rate": 1.3520945737733496e-05, "loss": 1.4106, "step": 1317 }, { "epoch": 0.41, "learning_rate": 1.3511615250017948e-05, "loss": 1.6078, "step": 1318 }, { "epoch": 0.41, "learning_rate": 1.3502281273838622e-05, "loss": 1.1787, "step": 1319 }, { "epoch": 0.41, "learning_rate": 1.3492943818467956e-05, "loss": 1.8198, "step": 1320 }, { "epoch": 0.41, "learning_rate": 1.3483602893181841e-05, "loss": 1.4558, "step": 1321 }, { "epoch": 0.41, "learning_rate": 1.3474258507259623e-05, "loss": 1.4652, "step": 1322 }, { "epoch": 0.41, "learning_rate": 1.3464910669984084e-05, "loss": 1.9823, "step": 1323 }, { "epoch": 0.41, "learning_rate": 1.3455559390641426e-05, "loss": 1.1311, "step": 1324 }, { "epoch": 0.41, "learning_rate": 1.3446204678521281e-05, "loss": 1.453, "step": 1325 }, { "epoch": 0.41, "learning_rate": 1.3436846542916686e-05, "loss": 1.4574, "step": 1326 }, { "epoch": 0.41, "learning_rate": 1.3427484993124082e-05, "loss": 1.4486, "step": 1327 }, { "epoch": 0.41, "learning_rate": 1.34181200384433e-05, "loss": 1.3743, "step": 1328 }, { "epoch": 0.41, "learning_rate": 1.3408751688177551e-05, "loss": 1.6277, "step": 1329 }, { "epoch": 0.41, "learning_rate": 1.339937995163342e-05, "loss": 1.2064, "step": 1330 }, { "epoch": 0.41, "learning_rate": 1.3390004838120861e-05, "loss": 1.8015, "step": 1331 }, { "epoch": 0.41, "learning_rate": 1.3380626356953175e-05, "loss": 1.6031, "step": 1332 }, { "epoch": 0.41, "learning_rate": 1.3371244517447017e-05, "loss": 1.5908, "step": 1333 }, { "epoch": 0.41, "learning_rate": 1.3361859328922368e-05, "loss": 1.4192, "step": 1334 }, { "epoch": 0.41, "learning_rate": 1.3352470800702548e-05, "loss": 2.5847, "step": 1335 }, { "epoch": 0.41, "learning_rate": 1.3343078942114185e-05, "loss": 1.5099, "step": 1336 }, { "epoch": 0.41, "learning_rate": 1.333368376248722e-05, "loss": 1.6909, "step": 1337 }, { "epoch": 0.41, "learning_rate": 1.3324285271154889e-05, "loss": 1.8992, "step": 1338 }, { "epoch": 0.41, "learning_rate": 1.3314883477453725e-05, "loss": 1.4883, "step": 1339 }, { "epoch": 0.41, "learning_rate": 1.3305478390723532e-05, "loss": 1.5553, "step": 1340 }, { "epoch": 0.41, "learning_rate": 1.3296070020307394e-05, "loss": 1.5106, "step": 1341 }, { "epoch": 0.41, "learning_rate": 1.3286658375551654e-05, "loss": 1.7311, "step": 1342 }, { "epoch": 0.41, "learning_rate": 1.3277243465805905e-05, "loss": 1.3575, "step": 1343 }, { "epoch": 0.41, "learning_rate": 1.3267825300422991e-05, "loss": 1.3254, "step": 1344 }, { "epoch": 0.41, "learning_rate": 1.3258403888758978e-05, "loss": 1.3226, "step": 1345 }, { "epoch": 0.41, "learning_rate": 1.3248979240173167e-05, "loss": 1.4751, "step": 1346 }, { "epoch": 0.41, "learning_rate": 1.3239551364028076e-05, "loss": 1.4943, "step": 1347 }, { "epoch": 0.41, "learning_rate": 1.3230120269689417e-05, "loss": 1.3, "step": 1348 }, { "epoch": 0.42, "learning_rate": 1.3220685966526112e-05, "loss": 1.4543, "step": 1349 }, { "epoch": 0.42, "learning_rate": 1.3211248463910263e-05, "loss": 2.2739, "step": 1350 }, { "epoch": 0.42, "learning_rate": 1.3201807771217156e-05, "loss": 1.7298, "step": 1351 }, { "epoch": 0.42, "learning_rate": 1.319236389782524e-05, "loss": 1.3881, "step": 1352 }, { "epoch": 0.42, "learning_rate": 1.3182916853116124e-05, "loss": 1.432, "step": 1353 }, { "epoch": 0.42, "learning_rate": 1.3173466646474575e-05, "loss": 1.8247, "step": 1354 }, { "epoch": 0.42, "learning_rate": 1.3164013287288495e-05, "loss": 1.1211, "step": 1355 }, { "epoch": 0.42, "learning_rate": 1.3154556784948918e-05, "loss": 1.3836, "step": 1356 }, { "epoch": 0.42, "learning_rate": 1.3145097148850002e-05, "loss": 1.1664, "step": 1357 }, { "epoch": 0.42, "learning_rate": 1.3135634388389016e-05, "loss": 1.6141, "step": 1358 }, { "epoch": 0.42, "learning_rate": 1.3126168512966333e-05, "loss": 1.841, "step": 1359 }, { "epoch": 0.42, "learning_rate": 1.3116699531985426e-05, "loss": 1.7351, "step": 1360 }, { "epoch": 0.42, "learning_rate": 1.3107227454852841e-05, "loss": 1.7628, "step": 1361 }, { "epoch": 0.42, "learning_rate": 1.3097752290978219e-05, "loss": 1.3661, "step": 1362 }, { "epoch": 0.42, "learning_rate": 1.3088274049774245e-05, "loss": 1.5343, "step": 1363 }, { "epoch": 0.42, "learning_rate": 1.3078792740656679e-05, "loss": 1.6005, "step": 1364 }, { "epoch": 0.42, "learning_rate": 1.3069308373044315e-05, "loss": 1.43, "step": 1365 }, { "epoch": 0.42, "learning_rate": 1.3059820956358998e-05, "loss": 1.4833, "step": 1366 }, { "epoch": 0.42, "learning_rate": 1.3050330500025592e-05, "loss": 1.3415, "step": 1367 }, { "epoch": 0.42, "learning_rate": 1.3040837013471985e-05, "loss": 1.6342, "step": 1368 }, { "epoch": 0.42, "learning_rate": 1.3031340506129074e-05, "loss": 1.481, "step": 1369 }, { "epoch": 0.42, "learning_rate": 1.3021840987430761e-05, "loss": 1.4388, "step": 1370 }, { "epoch": 0.42, "learning_rate": 1.301233846681393e-05, "loss": 1.4746, "step": 1371 }, { "epoch": 0.42, "learning_rate": 1.3002832953718462e-05, "loss": 1.4783, "step": 1372 }, { "epoch": 0.42, "learning_rate": 1.2993324457587194e-05, "loss": 1.3375, "step": 1373 }, { "epoch": 0.42, "learning_rate": 1.2983812987865936e-05, "loss": 1.4069, "step": 1374 }, { "epoch": 0.42, "learning_rate": 1.2974298554003451e-05, "loss": 1.8389, "step": 1375 }, { "epoch": 0.42, "learning_rate": 1.2964781165451448e-05, "loss": 1.0991, "step": 1376 }, { "epoch": 0.42, "learning_rate": 1.2955260831664564e-05, "loss": 1.3493, "step": 1377 }, { "epoch": 0.42, "learning_rate": 1.2945737562100373e-05, "loss": 2.1661, "step": 1378 }, { "epoch": 0.42, "learning_rate": 1.2936211366219359e-05, "loss": 1.268, "step": 1379 }, { "epoch": 0.42, "learning_rate": 1.2926682253484906e-05, "loss": 1.2769, "step": 1380 }, { "epoch": 0.42, "learning_rate": 1.2917150233363307e-05, "loss": 1.7214, "step": 1381 }, { "epoch": 0.43, "learning_rate": 1.290761531532374e-05, "loss": 1.44, "step": 1382 }, { "epoch": 0.43, "learning_rate": 1.2898077508838263e-05, "loss": 1.596, "step": 1383 }, { "epoch": 0.43, "learning_rate": 1.2888536823381799e-05, "loss": 1.4986, "step": 1384 }, { "epoch": 0.43, "learning_rate": 1.287899326843213e-05, "loss": 1.4707, "step": 1385 }, { "epoch": 0.43, "learning_rate": 1.2869446853469898e-05, "loss": 1.6921, "step": 1386 }, { "epoch": 0.43, "learning_rate": 1.285989758797858e-05, "loss": 1.3802, "step": 1387 }, { "epoch": 0.43, "learning_rate": 1.285034548144448e-05, "loss": 1.5127, "step": 1388 }, { "epoch": 0.43, "learning_rate": 1.284079054335673e-05, "loss": 1.6045, "step": 1389 }, { "epoch": 0.43, "learning_rate": 1.2831232783207278e-05, "loss": 1.35, "step": 1390 }, { "epoch": 0.43, "learning_rate": 1.2821672210490865e-05, "loss": 1.4313, "step": 1391 }, { "epoch": 0.43, "learning_rate": 1.281210883470504e-05, "loss": 1.369, "step": 1392 }, { "epoch": 0.43, "learning_rate": 1.2802542665350124e-05, "loss": 1.2239, "step": 1393 }, { "epoch": 0.43, "learning_rate": 1.2792973711929222e-05, "loss": 1.5115, "step": 1394 }, { "epoch": 0.43, "learning_rate": 1.2783401983948198e-05, "loss": 1.5038, "step": 1395 }, { "epoch": 0.43, "learning_rate": 1.2773827490915678e-05, "loss": 1.3529, "step": 1396 }, { "epoch": 0.43, "learning_rate": 1.2764250242343025e-05, "loss": 1.3804, "step": 1397 }, { "epoch": 0.43, "learning_rate": 1.2754670247744353e-05, "loss": 1.35, "step": 1398 }, { "epoch": 0.43, "learning_rate": 1.2745087516636501e-05, "loss": 1.4144, "step": 1399 }, { "epoch": 0.43, "learning_rate": 1.2735502058539014e-05, "loss": 1.6021, "step": 1400 }, { "epoch": 0.43, "learning_rate": 1.272591388297416e-05, "loss": 1.1659, "step": 1401 }, { "epoch": 0.43, "learning_rate": 1.27163229994669e-05, "loss": 1.5967, "step": 1402 }, { "epoch": 0.43, "learning_rate": 1.2706729417544885e-05, "loss": 1.4683, "step": 1403 }, { "epoch": 0.43, "learning_rate": 1.2697133146738452e-05, "loss": 2.2997, "step": 1404 }, { "epoch": 0.43, "learning_rate": 1.2687534196580597e-05, "loss": 1.4074, "step": 1405 }, { "epoch": 0.43, "learning_rate": 1.2677932576606998e-05, "loss": 1.7564, "step": 1406 }, { "epoch": 0.43, "learning_rate": 1.2668328296355964e-05, "loss": 1.4242, "step": 1407 }, { "epoch": 0.43, "learning_rate": 1.2658721365368462e-05, "loss": 1.2521, "step": 1408 }, { "epoch": 0.43, "learning_rate": 1.264911179318808e-05, "loss": 1.7296, "step": 1409 }, { "epoch": 0.43, "learning_rate": 1.2639499589361041e-05, "loss": 1.6688, "step": 1410 }, { "epoch": 0.43, "learning_rate": 1.2629884763436176e-05, "loss": 1.8353, "step": 1411 }, { "epoch": 0.43, "learning_rate": 1.2620267324964918e-05, "loss": 1.4325, "step": 1412 }, { "epoch": 0.43, "learning_rate": 1.2610647283501303e-05, "loss": 1.4561, "step": 1413 }, { "epoch": 0.44, "learning_rate": 1.260102464860195e-05, "loss": 1.9443, "step": 1414 }, { "epoch": 0.44, "learning_rate": 1.2591399429826054e-05, "loss": 2.1286, "step": 1415 }, { "epoch": 0.44, "learning_rate": 1.258177163673537e-05, "loss": 1.1668, "step": 1416 }, { "epoch": 0.44, "learning_rate": 1.2572141278894226e-05, "loss": 1.3574, "step": 1417 }, { "epoch": 0.44, "learning_rate": 1.2562508365869484e-05, "loss": 1.3362, "step": 1418 }, { "epoch": 0.44, "learning_rate": 1.2552872907230547e-05, "loss": 1.8117, "step": 1419 }, { "epoch": 0.44, "learning_rate": 1.2543234912549352e-05, "loss": 1.7966, "step": 1420 }, { "epoch": 0.44, "learning_rate": 1.2533594391400347e-05, "loss": 1.3343, "step": 1421 }, { "epoch": 0.44, "learning_rate": 1.2523951353360504e-05, "loss": 1.7886, "step": 1422 }, { "epoch": 0.44, "learning_rate": 1.2514305808009279e-05, "loss": 1.3105, "step": 1423 }, { "epoch": 0.44, "learning_rate": 1.2504657764928626e-05, "loss": 1.3934, "step": 1424 }, { "epoch": 0.44, "learning_rate": 1.2495007233702982e-05, "loss": 1.7103, "step": 1425 }, { "epoch": 0.44, "learning_rate": 1.2485354223919255e-05, "loss": 1.5023, "step": 1426 }, { "epoch": 0.44, "learning_rate": 1.2475698745166813e-05, "loss": 1.4954, "step": 1427 }, { "epoch": 0.44, "learning_rate": 1.2466040807037478e-05, "loss": 1.4154, "step": 1428 }, { "epoch": 0.44, "learning_rate": 1.2456380419125516e-05, "loss": 1.617, "step": 1429 }, { "epoch": 0.44, "learning_rate": 1.2446717591027624e-05, "loss": 1.8304, "step": 1430 }, { "epoch": 0.44, "learning_rate": 1.2437052332342923e-05, "loss": 1.5548, "step": 1431 }, { "epoch": 0.44, "learning_rate": 1.2427384652672954e-05, "loss": 1.391, "step": 1432 }, { "epoch": 0.44, "learning_rate": 1.2417714561621656e-05, "loss": 1.6611, "step": 1433 }, { "epoch": 0.44, "learning_rate": 1.2408042068795373e-05, "loss": 1.4653, "step": 1434 }, { "epoch": 0.44, "learning_rate": 1.2398367183802822e-05, "loss": 1.233, "step": 1435 }, { "epoch": 0.44, "learning_rate": 1.2388689916255106e-05, "loss": 1.2863, "step": 1436 }, { "epoch": 0.44, "learning_rate": 1.2379010275765693e-05, "loss": 1.5886, "step": 1437 }, { "epoch": 0.44, "learning_rate": 1.2369328271950404e-05, "loss": 1.5959, "step": 1438 }, { "epoch": 0.44, "learning_rate": 1.2359643914427415e-05, "loss": 1.3312, "step": 1439 }, { "epoch": 0.44, "learning_rate": 1.2349957212817229e-05, "loss": 1.7169, "step": 1440 }, { "epoch": 0.44, "learning_rate": 1.2340268176742696e-05, "loss": 1.6262, "step": 1441 }, { "epoch": 0.44, "learning_rate": 1.2330576815828964e-05, "loss": 1.2585, "step": 1442 }, { "epoch": 0.44, "learning_rate": 1.2320883139703512e-05, "loss": 2.1559, "step": 1443 }, { "epoch": 0.44, "learning_rate": 1.2311187157996091e-05, "loss": 1.6384, "step": 1444 }, { "epoch": 0.44, "learning_rate": 1.2301488880338772e-05, "loss": 1.5301, "step": 1445 }, { "epoch": 0.44, "learning_rate": 1.229178831636589e-05, "loss": 1.4363, "step": 1446 }, { "epoch": 0.45, "learning_rate": 1.2282085475714052e-05, "loss": 1.5623, "step": 1447 }, { "epoch": 0.45, "learning_rate": 1.2272380368022133e-05, "loss": 1.4032, "step": 1448 }, { "epoch": 0.45, "learning_rate": 1.2262673002931254e-05, "loss": 1.4183, "step": 1449 }, { "epoch": 0.45, "learning_rate": 1.2252963390084784e-05, "loss": 1.5666, "step": 1450 }, { "epoch": 0.45, "learning_rate": 1.224325153912832e-05, "loss": 1.9528, "step": 1451 }, { "epoch": 0.45, "learning_rate": 1.2233537459709683e-05, "loss": 1.4767, "step": 1452 }, { "epoch": 0.45, "learning_rate": 1.2223821161478912e-05, "loss": 1.7994, "step": 1453 }, { "epoch": 0.45, "learning_rate": 1.2214102654088248e-05, "loss": 1.4598, "step": 1454 }, { "epoch": 0.45, "learning_rate": 1.2204381947192123e-05, "loss": 1.3264, "step": 1455 }, { "epoch": 0.45, "learning_rate": 1.2194659050447158e-05, "loss": 1.4388, "step": 1456 }, { "epoch": 0.45, "learning_rate": 1.2184933973512152e-05, "loss": 1.3978, "step": 1457 }, { "epoch": 0.45, "learning_rate": 1.2175206726048058e-05, "loss": 2.2029, "step": 1458 }, { "epoch": 0.45, "learning_rate": 1.2165477317718002e-05, "loss": 1.8447, "step": 1459 }, { "epoch": 0.45, "learning_rate": 1.215574575818724e-05, "loss": 1.3005, "step": 1460 }, { "epoch": 0.45, "learning_rate": 1.214601205712318e-05, "loss": 1.9965, "step": 1461 }, { "epoch": 0.45, "learning_rate": 1.2136276224195349e-05, "loss": 1.6336, "step": 1462 }, { "epoch": 0.45, "learning_rate": 1.2126538269075392e-05, "loss": 1.3078, "step": 1463 }, { "epoch": 0.45, "learning_rate": 1.2116798201437061e-05, "loss": 1.3533, "step": 1464 }, { "epoch": 0.45, "learning_rate": 1.2107056030956214e-05, "loss": 1.2577, "step": 1465 }, { "epoch": 0.45, "learning_rate": 1.2097311767310791e-05, "loss": 1.4717, "step": 1466 }, { "epoch": 0.45, "learning_rate": 1.208756542018081e-05, "loss": 1.5975, "step": 1467 }, { "epoch": 0.45, "learning_rate": 1.2077816999248366e-05, "loss": 1.3437, "step": 1468 }, { "epoch": 0.45, "learning_rate": 1.2068066514197611e-05, "loss": 1.4219, "step": 1469 }, { "epoch": 0.45, "learning_rate": 1.2058313974714746e-05, "loss": 1.3091, "step": 1470 }, { "epoch": 0.45, "learning_rate": 1.2048559390488013e-05, "loss": 1.5736, "step": 1471 }, { "epoch": 0.45, "learning_rate": 1.2038802771207686e-05, "loss": 1.2382, "step": 1472 }, { "epoch": 0.45, "learning_rate": 1.2029044126566058e-05, "loss": 1.3758, "step": 1473 }, { "epoch": 0.45, "learning_rate": 1.2019283466257441e-05, "loss": 1.4917, "step": 1474 }, { "epoch": 0.45, "learning_rate": 1.2009520799978143e-05, "loss": 1.4151, "step": 1475 }, { "epoch": 0.45, "learning_rate": 1.1999756137426465e-05, "loss": 1.6299, "step": 1476 }, { "epoch": 0.45, "learning_rate": 1.1989989488302695e-05, "loss": 1.4298, "step": 1477 }, { "epoch": 0.45, "learning_rate": 1.1980220862309097e-05, "loss": 1.5317, "step": 1478 }, { "epoch": 0.46, "learning_rate": 1.1970450269149882e-05, "loss": 1.9429, "step": 1479 }, { "epoch": 0.46, "learning_rate": 1.196067771853124e-05, "loss": 1.5051, "step": 1480 }, { "epoch": 0.46, "learning_rate": 1.1950903220161286e-05, "loss": 1.6433, "step": 1481 }, { "epoch": 0.46, "learning_rate": 1.1941126783750075e-05, "loss": 1.5357, "step": 1482 }, { "epoch": 0.46, "learning_rate": 1.1931348419009594e-05, "loss": 1.9704, "step": 1483 }, { "epoch": 0.46, "learning_rate": 1.1921568135653738e-05, "loss": 1.6505, "step": 1484 }, { "epoch": 0.46, "learning_rate": 1.1911785943398316e-05, "loss": 1.4378, "step": 1485 }, { "epoch": 0.46, "learning_rate": 1.1902001851961019e-05, "loss": 1.7249, "step": 1486 }, { "epoch": 0.46, "learning_rate": 1.1892215871061436e-05, "loss": 1.2966, "step": 1487 }, { "epoch": 0.46, "learning_rate": 1.1882428010421036e-05, "loss": 1.5901, "step": 1488 }, { "epoch": 0.46, "learning_rate": 1.1872638279763147e-05, "loss": 1.7083, "step": 1489 }, { "epoch": 0.46, "learning_rate": 1.1862846688812956e-05, "loss": 1.2796, "step": 1490 }, { "epoch": 0.46, "learning_rate": 1.1853053247297503e-05, "loss": 1.4548, "step": 1491 }, { "epoch": 0.46, "learning_rate": 1.184325796494566e-05, "loss": 1.6324, "step": 1492 }, { "epoch": 0.46, "learning_rate": 1.1833460851488132e-05, "loss": 2.0364, "step": 1493 }, { "epoch": 0.46, "learning_rate": 1.1823661916657441e-05, "loss": 1.5063, "step": 1494 }, { "epoch": 0.46, "learning_rate": 1.1813861170187919e-05, "loss": 2.2661, "step": 1495 }, { "epoch": 0.46, "learning_rate": 1.1804058621815699e-05, "loss": 1.6608, "step": 1496 }, { "epoch": 0.46, "learning_rate": 1.1794254281278704e-05, "loss": 1.8123, "step": 1497 }, { "epoch": 0.46, "learning_rate": 1.1784448158316633e-05, "loss": 1.5114, "step": 1498 }, { "epoch": 0.46, "learning_rate": 1.1774640262670962e-05, "loss": 1.5714, "step": 1499 }, { "epoch": 0.46, "learning_rate": 1.1764830604084923e-05, "loss": 1.2827, "step": 1500 }, { "epoch": 0.46, "learning_rate": 1.1755019192303503e-05, "loss": 1.3728, "step": 1501 }, { "epoch": 0.46, "learning_rate": 1.1745206037073426e-05, "loss": 1.451, "step": 1502 }, { "epoch": 0.46, "learning_rate": 1.1735391148143149e-05, "loss": 1.4311, "step": 1503 }, { "epoch": 0.46, "learning_rate": 1.1725574535262863e-05, "loss": 1.4488, "step": 1504 }, { "epoch": 0.46, "learning_rate": 1.171575620818445e-05, "loss": 1.577, "step": 1505 }, { "epoch": 0.46, "learning_rate": 1.1705936176661516e-05, "loss": 1.4405, "step": 1506 }, { "epoch": 0.46, "learning_rate": 1.1696114450449343e-05, "loss": 1.5442, "step": 1507 }, { "epoch": 0.46, "learning_rate": 1.168629103930491e-05, "loss": 1.5151, "step": 1508 }, { "epoch": 0.46, "learning_rate": 1.167646595298686e-05, "loss": 1.4906, "step": 1509 }, { "epoch": 0.46, "learning_rate": 1.1666639201255507e-05, "loss": 1.2289, "step": 1510 }, { "epoch": 0.46, "learning_rate": 1.1656810793872813e-05, "loss": 1.7102, "step": 1511 }, { "epoch": 0.47, "learning_rate": 1.1646980740602394e-05, "loss": 1.4379, "step": 1512 }, { "epoch": 0.47, "learning_rate": 1.1637149051209496e-05, "loss": 1.2132, "step": 1513 }, { "epoch": 0.47, "learning_rate": 1.162731573546098e-05, "loss": 1.4962, "step": 1514 }, { "epoch": 0.47, "learning_rate": 1.1617480803125338e-05, "loss": 1.503, "step": 1515 }, { "epoch": 0.47, "learning_rate": 1.1607644263972669e-05, "loss": 1.4258, "step": 1516 }, { "epoch": 0.47, "learning_rate": 1.1597806127774651e-05, "loss": 1.5131, "step": 1517 }, { "epoch": 0.47, "learning_rate": 1.1587966404304565e-05, "loss": 1.5341, "step": 1518 }, { "epoch": 0.47, "learning_rate": 1.157812510333726e-05, "loss": 1.3106, "step": 1519 }, { "epoch": 0.47, "learning_rate": 1.156828223464916e-05, "loss": 1.3401, "step": 1520 }, { "epoch": 0.47, "learning_rate": 1.1558437808018238e-05, "loss": 1.4217, "step": 1521 }, { "epoch": 0.47, "learning_rate": 1.1548591833224015e-05, "loss": 1.4464, "step": 1522 }, { "epoch": 0.47, "learning_rate": 1.1538744320047554e-05, "loss": 1.4141, "step": 1523 }, { "epoch": 0.47, "learning_rate": 1.152889527827145e-05, "loss": 1.3032, "step": 1524 }, { "epoch": 0.47, "learning_rate": 1.151904471767981e-05, "loss": 2.1309, "step": 1525 }, { "epoch": 0.47, "learning_rate": 1.150919264805825e-05, "loss": 1.8071, "step": 1526 }, { "epoch": 0.47, "learning_rate": 1.1499339079193886e-05, "loss": 1.4872, "step": 1527 }, { "epoch": 0.47, "learning_rate": 1.1489484020875325e-05, "loss": 1.7761, "step": 1528 }, { "epoch": 0.47, "learning_rate": 1.1479627482892655e-05, "loss": 1.1814, "step": 1529 }, { "epoch": 0.47, "learning_rate": 1.1469769475037427e-05, "loss": 1.8832, "step": 1530 }, { "epoch": 0.47, "learning_rate": 1.1459910007102657e-05, "loss": 1.4938, "step": 1531 }, { "epoch": 0.47, "learning_rate": 1.1450049088882816e-05, "loss": 1.6288, "step": 1532 }, { "epoch": 0.47, "learning_rate": 1.1440186730173808e-05, "loss": 1.5206, "step": 1533 }, { "epoch": 0.47, "learning_rate": 1.143032294077297e-05, "loss": 1.2526, "step": 1534 }, { "epoch": 0.47, "learning_rate": 1.1420457730479061e-05, "loss": 1.7994, "step": 1535 }, { "epoch": 0.47, "learning_rate": 1.1410591109092257e-05, "loss": 1.6462, "step": 1536 }, { "epoch": 0.47, "learning_rate": 1.1400723086414122e-05, "loss": 1.6014, "step": 1537 }, { "epoch": 0.47, "learning_rate": 1.1390853672247629e-05, "loss": 1.1764, "step": 1538 }, { "epoch": 0.47, "learning_rate": 1.1380982876397115e-05, "loss": 1.8971, "step": 1539 }, { "epoch": 0.47, "learning_rate": 1.1371110708668307e-05, "loss": 1.4256, "step": 1540 }, { "epoch": 0.47, "learning_rate": 1.1361237178868288e-05, "loss": 1.5668, "step": 1541 }, { "epoch": 0.47, "learning_rate": 1.1351362296805487e-05, "loss": 1.4795, "step": 1542 }, { "epoch": 0.47, "learning_rate": 1.1341486072289686e-05, "loss": 1.5851, "step": 1543 }, { "epoch": 0.48, "learning_rate": 1.1331608515132001e-05, "loss": 1.4929, "step": 1544 }, { "epoch": 0.48, "learning_rate": 1.1321729635144862e-05, "loss": 1.5407, "step": 1545 }, { "epoch": 0.48, "learning_rate": 1.1311849442142025e-05, "loss": 1.6635, "step": 1546 }, { "epoch": 0.48, "learning_rate": 1.1301967945938538e-05, "loss": 1.6153, "step": 1547 }, { "epoch": 0.48, "learning_rate": 1.1292085156350756e-05, "loss": 1.4714, "step": 1548 }, { "epoch": 0.48, "learning_rate": 1.128220108319631e-05, "loss": 1.4397, "step": 1549 }, { "epoch": 0.48, "learning_rate": 1.1272315736294108e-05, "loss": 1.4616, "step": 1550 }, { "epoch": 0.48, "learning_rate": 1.1262429125464328e-05, "loss": 1.5294, "step": 1551 }, { "epoch": 0.48, "learning_rate": 1.1252541260528398e-05, "loss": 1.8233, "step": 1552 }, { "epoch": 0.48, "learning_rate": 1.1242652151308997e-05, "loss": 1.5038, "step": 1553 }, { "epoch": 0.48, "learning_rate": 1.1232761807630032e-05, "loss": 1.4738, "step": 1554 }, { "epoch": 0.48, "learning_rate": 1.1222870239316645e-05, "loss": 1.3698, "step": 1555 }, { "epoch": 0.48, "learning_rate": 1.121297745619519e-05, "loss": 1.5445, "step": 1556 }, { "epoch": 0.48, "learning_rate": 1.120308346809323e-05, "loss": 1.667, "step": 1557 }, { "epoch": 0.48, "learning_rate": 1.1193188284839518e-05, "loss": 1.5548, "step": 1558 }, { "epoch": 0.48, "learning_rate": 1.1183291916264007e-05, "loss": 1.4341, "step": 1559 }, { "epoch": 0.48, "learning_rate": 1.117339437219782e-05, "loss": 1.3633, "step": 1560 }, { "epoch": 0.48, "learning_rate": 1.1163495662473242e-05, "loss": 1.254, "step": 1561 }, { "epoch": 0.48, "learning_rate": 1.115359579692373e-05, "loss": 2.2963, "step": 1562 }, { "epoch": 0.48, "learning_rate": 1.1143694785383877e-05, "loss": 1.367, "step": 1563 }, { "epoch": 0.48, "learning_rate": 1.113379263768942e-05, "loss": 1.5674, "step": 1564 }, { "epoch": 0.48, "learning_rate": 1.1123889363677223e-05, "loss": 1.5677, "step": 1565 }, { "epoch": 0.48, "learning_rate": 1.1113984973185268e-05, "loss": 1.4457, "step": 1566 }, { "epoch": 0.48, "learning_rate": 1.1104079476052649e-05, "loss": 1.3479, "step": 1567 }, { "epoch": 0.48, "learning_rate": 1.1094172882119562e-05, "loss": 1.5166, "step": 1568 }, { "epoch": 0.48, "learning_rate": 1.1084265201227284e-05, "loss": 1.3685, "step": 1569 }, { "epoch": 0.48, "learning_rate": 1.1074356443218175e-05, "loss": 1.5713, "step": 1570 }, { "epoch": 0.48, "learning_rate": 1.106444661793567e-05, "loss": 1.5827, "step": 1571 }, { "epoch": 0.48, "learning_rate": 1.1054535735224257e-05, "loss": 1.9172, "step": 1572 }, { "epoch": 0.48, "learning_rate": 1.1044623804929481e-05, "loss": 1.4493, "step": 1573 }, { "epoch": 0.48, "learning_rate": 1.1034710836897922e-05, "loss": 1.3679, "step": 1574 }, { "epoch": 0.48, "learning_rate": 1.1024796840977196e-05, "loss": 2.3108, "step": 1575 }, { "epoch": 0.48, "learning_rate": 1.1014881827015939e-05, "loss": 1.7419, "step": 1576 }, { "epoch": 0.49, "learning_rate": 1.1004965804863795e-05, "loss": 1.3969, "step": 1577 }, { "epoch": 0.49, "learning_rate": 1.0995048784371405e-05, "loss": 1.3786, "step": 1578 }, { "epoch": 0.49, "learning_rate": 1.098513077539042e-05, "loss": 1.8025, "step": 1579 }, { "epoch": 0.49, "learning_rate": 1.0975211787773455e-05, "loss": 1.0797, "step": 1580 }, { "epoch": 0.49, "learning_rate": 1.0965291831374104e-05, "loss": 1.4982, "step": 1581 }, { "epoch": 0.49, "learning_rate": 1.095537091604692e-05, "loss": 1.3096, "step": 1582 }, { "epoch": 0.49, "learning_rate": 1.0945449051647417e-05, "loss": 1.2556, "step": 1583 }, { "epoch": 0.49, "learning_rate": 1.0935526248032047e-05, "loss": 1.4918, "step": 1584 }, { "epoch": 0.49, "learning_rate": 1.0925602515058187e-05, "loss": 1.6028, "step": 1585 }, { "epoch": 0.49, "learning_rate": 1.0915677862584149e-05, "loss": 1.4136, "step": 1586 }, { "epoch": 0.49, "learning_rate": 1.0905752300469155e-05, "loss": 1.224, "step": 1587 }, { "epoch": 0.49, "learning_rate": 1.0895825838573326e-05, "loss": 1.4274, "step": 1588 }, { "epoch": 0.49, "learning_rate": 1.0885898486757686e-05, "loss": 1.5501, "step": 1589 }, { "epoch": 0.49, "learning_rate": 1.087597025488413e-05, "loss": 1.5916, "step": 1590 }, { "epoch": 0.49, "learning_rate": 1.0866041152815443e-05, "loss": 1.4807, "step": 1591 }, { "epoch": 0.49, "learning_rate": 1.0856111190415262e-05, "loss": 1.4673, "step": 1592 }, { "epoch": 0.49, "learning_rate": 1.0846180377548082e-05, "loss": 1.4942, "step": 1593 }, { "epoch": 0.49, "learning_rate": 1.0836248724079241e-05, "loss": 1.4996, "step": 1594 }, { "epoch": 0.49, "learning_rate": 1.0826316239874923e-05, "loss": 1.4299, "step": 1595 }, { "epoch": 0.49, "learning_rate": 1.0816382934802123e-05, "loss": 1.815, "step": 1596 }, { "epoch": 0.49, "learning_rate": 1.080644881872866e-05, "loss": 1.5384, "step": 1597 }, { "epoch": 0.49, "learning_rate": 1.0796513901523156e-05, "loss": 1.484, "step": 1598 }, { "epoch": 0.49, "learning_rate": 1.0786578193055029e-05, "loss": 1.4599, "step": 1599 }, { "epoch": 0.49, "learning_rate": 1.077664170319448e-05, "loss": 1.2585, "step": 1600 }, { "epoch": 0.49, "learning_rate": 1.0766704441812493e-05, "loss": 1.4028, "step": 1601 }, { "epoch": 0.49, "learning_rate": 1.0756766418780811e-05, "loss": 0.9765, "step": 1602 }, { "epoch": 0.49, "learning_rate": 1.0746827643971942e-05, "loss": 1.6094, "step": 1603 }, { "epoch": 0.49, "learning_rate": 1.0736888127259132e-05, "loss": 1.36, "step": 1604 }, { "epoch": 0.49, "learning_rate": 1.0726947878516373e-05, "loss": 1.4325, "step": 1605 }, { "epoch": 0.49, "learning_rate": 1.0717006907618377e-05, "loss": 1.6546, "step": 1606 }, { "epoch": 0.49, "learning_rate": 1.0707065224440575e-05, "loss": 1.5306, "step": 1607 }, { "epoch": 0.49, "learning_rate": 1.0697122838859106e-05, "loss": 1.5791, "step": 1608 }, { "epoch": 0.5, "learning_rate": 1.0687179760750808e-05, "loss": 1.9094, "step": 1609 }, { "epoch": 0.5, "learning_rate": 1.0677235999993205e-05, "loss": 1.4735, "step": 1610 }, { "epoch": 0.5, "learning_rate": 1.0667291566464504e-05, "loss": 1.4291, "step": 1611 }, { "epoch": 0.5, "learning_rate": 1.0657346470043575e-05, "loss": 1.892, "step": 1612 }, { "epoch": 0.5, "learning_rate": 1.0647400720609942e-05, "loss": 1.6397, "step": 1613 }, { "epoch": 0.5, "learning_rate": 1.0637454328043792e-05, "loss": 1.6173, "step": 1614 }, { "epoch": 0.5, "learning_rate": 1.0627507302225938e-05, "loss": 1.6717, "step": 1615 }, { "epoch": 0.5, "learning_rate": 1.0617559653037833e-05, "loss": 1.4193, "step": 1616 }, { "epoch": 0.5, "learning_rate": 1.0607611390361534e-05, "loss": 1.254, "step": 1617 }, { "epoch": 0.5, "learning_rate": 1.0597662524079721e-05, "loss": 1.3787, "step": 1618 }, { "epoch": 0.5, "learning_rate": 1.0587713064075673e-05, "loss": 1.7384, "step": 1619 }, { "epoch": 0.5, "learning_rate": 1.057776302023325e-05, "loss": 1.6016, "step": 1620 }, { "epoch": 0.5, "learning_rate": 1.0567812402436895e-05, "loss": 1.396, "step": 1621 }, { "epoch": 0.5, "learning_rate": 1.0557861220571626e-05, "loss": 1.406, "step": 1622 }, { "epoch": 0.5, "learning_rate": 1.0547909484523023e-05, "loss": 1.3675, "step": 1623 }, { "epoch": 0.5, "learning_rate": 1.0537957204177204e-05, "loss": 1.6323, "step": 1624 }, { "epoch": 0.5, "learning_rate": 1.052800438942084e-05, "loss": 1.5804, "step": 1625 }, { "epoch": 0.5, "learning_rate": 1.051805105014113e-05, "loss": 1.7015, "step": 1626 }, { "epoch": 0.5, "learning_rate": 1.0508097196225787e-05, "loss": 1.6947, "step": 1627 }, { "epoch": 0.5, "learning_rate": 1.0498142837563048e-05, "loss": 1.679, "step": 1628 }, { "epoch": 0.5, "learning_rate": 1.0488187984041636e-05, "loss": 1.6727, "step": 1629 }, { "epoch": 0.5, "learning_rate": 1.0478232645550784e-05, "loss": 1.7717, "step": 1630 }, { "epoch": 0.5, "learning_rate": 1.0468276831980186e-05, "loss": 2.1289, "step": 1631 }, { "epoch": 0.5, "learning_rate": 1.045832055322003e-05, "loss": 1.39, "step": 1632 }, { "epoch": 0.5, "learning_rate": 1.0448363819160946e-05, "loss": 1.4587, "step": 1633 }, { "epoch": 0.5, "learning_rate": 1.0438406639694025e-05, "loss": 1.8243, "step": 1634 }, { "epoch": 0.5, "learning_rate": 1.0428449024710805e-05, "loss": 1.3176, "step": 1635 }, { "epoch": 0.5, "learning_rate": 1.041849098410325e-05, "loss": 1.4693, "step": 1636 }, { "epoch": 0.5, "learning_rate": 1.0408532527763749e-05, "loss": 1.3979, "step": 1637 }, { "epoch": 0.5, "learning_rate": 1.0398573665585105e-05, "loss": 1.4386, "step": 1638 }, { "epoch": 0.5, "learning_rate": 1.0388614407460523e-05, "loss": 1.3095, "step": 1639 }, { "epoch": 0.5, "learning_rate": 1.0378654763283602e-05, "loss": 1.2712, "step": 1640 }, { "epoch": 0.5, "learning_rate": 1.036869474294832e-05, "loss": 1.3695, "step": 1641 }, { "epoch": 0.51, "learning_rate": 1.0358734356349035e-05, "loss": 1.5895, "step": 1642 }, { "epoch": 0.51, "learning_rate": 1.034877361338047e-05, "loss": 1.342, "step": 1643 }, { "epoch": 0.51, "learning_rate": 1.0338812523937694e-05, "loss": 1.7784, "step": 1644 }, { "epoch": 0.51, "learning_rate": 1.0328851097916124e-05, "loss": 1.3724, "step": 1645 }, { "epoch": 0.51, "learning_rate": 1.0318889345211519e-05, "loss": 1.4914, "step": 1646 }, { "epoch": 0.51, "learning_rate": 1.0308927275719949e-05, "loss": 1.323, "step": 1647 }, { "epoch": 0.51, "learning_rate": 1.0298964899337807e-05, "loss": 1.6419, "step": 1648 }, { "epoch": 0.51, "learning_rate": 1.0289002225961788e-05, "loss": 1.5935, "step": 1649 }, { "epoch": 0.51, "learning_rate": 1.0279039265488885e-05, "loss": 1.4547, "step": 1650 }, { "epoch": 0.51, "learning_rate": 1.0269076027816376e-05, "loss": 1.3306, "step": 1651 }, { "epoch": 0.51, "learning_rate": 1.0259112522841808e-05, "loss": 1.6222, "step": 1652 }, { "epoch": 0.51, "learning_rate": 1.0249148760463001e-05, "loss": 1.9939, "step": 1653 }, { "epoch": 0.51, "learning_rate": 1.023918475057803e-05, "loss": 1.4909, "step": 1654 }, { "epoch": 0.51, "learning_rate": 1.0229220503085214e-05, "loss": 1.4362, "step": 1655 }, { "epoch": 0.51, "learning_rate": 1.0219256027883102e-05, "loss": 1.4717, "step": 1656 }, { "epoch": 0.51, "learning_rate": 1.0209291334870479e-05, "loss": 1.3928, "step": 1657 }, { "epoch": 0.51, "learning_rate": 1.0199326433946345e-05, "loss": 1.5495, "step": 1658 }, { "epoch": 0.51, "learning_rate": 1.01893613350099e-05, "loss": 1.3461, "step": 1659 }, { "epoch": 0.51, "learning_rate": 1.0179396047960546e-05, "loss": 1.6597, "step": 1660 }, { "epoch": 0.51, "learning_rate": 1.0169430582697873e-05, "loss": 1.2357, "step": 1661 }, { "epoch": 0.51, "learning_rate": 1.0159464949121642e-05, "loss": 1.3447, "step": 1662 }, { "epoch": 0.51, "learning_rate": 1.0149499157131788e-05, "loss": 1.5661, "step": 1663 }, { "epoch": 0.51, "learning_rate": 1.0139533216628399e-05, "loss": 1.3869, "step": 1664 }, { "epoch": 0.51, "learning_rate": 1.012956713751171e-05, "loss": 1.34, "step": 1665 }, { "epoch": 0.51, "learning_rate": 1.0119600929682096e-05, "loss": 1.6007, "step": 1666 }, { "epoch": 0.51, "learning_rate": 1.0109634603040062e-05, "loss": 1.5608, "step": 1667 }, { "epoch": 0.51, "learning_rate": 1.0099668167486227e-05, "loss": 1.0849, "step": 1668 }, { "epoch": 0.51, "learning_rate": 1.0089701632921321e-05, "loss": 1.2905, "step": 1669 }, { "epoch": 0.51, "learning_rate": 1.0079735009246168e-05, "loss": 1.5287, "step": 1670 }, { "epoch": 0.51, "learning_rate": 1.0069768306361685e-05, "loss": 1.7288, "step": 1671 }, { "epoch": 0.51, "learning_rate": 1.0059801534168868e-05, "loss": 1.4597, "step": 1672 }, { "epoch": 0.51, "learning_rate": 1.0049834702568772e-05, "loss": 1.6025, "step": 1673 }, { "epoch": 0.52, "learning_rate": 1.0039867821462529e-05, "loss": 1.2934, "step": 1674 }, { "epoch": 0.52, "learning_rate": 1.0029900900751308e-05, "loss": 1.0656, "step": 1675 }, { "epoch": 0.52, "learning_rate": 1.0019933950336314e-05, "loss": 1.4362, "step": 1676 }, { "epoch": 0.52, "learning_rate": 1.0009966980118792e-05, "loss": 1.5876, "step": 1677 }, { "epoch": 0.52, "learning_rate": 1e-05, "loss": 1.4996, "step": 1678 }, { "epoch": 0.52, "learning_rate": 9.99003301988121e-06, "loss": 1.608, "step": 1679 }, { "epoch": 0.52, "learning_rate": 9.98006604966369e-06, "loss": 1.1425, "step": 1680 }, { "epoch": 0.52, "learning_rate": 9.970099099248694e-06, "loss": 1.4318, "step": 1681 }, { "epoch": 0.52, "learning_rate": 9.960132178537471e-06, "loss": 1.54, "step": 1682 }, { "epoch": 0.52, "learning_rate": 9.95016529743123e-06, "loss": 1.6719, "step": 1683 }, { "epoch": 0.52, "learning_rate": 9.940198465831137e-06, "loss": 1.1922, "step": 1684 }, { "epoch": 0.52, "learning_rate": 9.930231693638316e-06, "loss": 1.6652, "step": 1685 }, { "epoch": 0.52, "learning_rate": 9.920264990753837e-06, "loss": 1.341, "step": 1686 }, { "epoch": 0.52, "learning_rate": 9.910298367078684e-06, "loss": 2.004, "step": 1687 }, { "epoch": 0.52, "learning_rate": 9.900331832513775e-06, "loss": 1.7567, "step": 1688 }, { "epoch": 0.52, "learning_rate": 9.89036539695994e-06, "loss": 1.59, "step": 1689 }, { "epoch": 0.52, "learning_rate": 9.880399070317907e-06, "loss": 1.3059, "step": 1690 }, { "epoch": 0.52, "learning_rate": 9.870432862488295e-06, "loss": 1.317, "step": 1691 }, { "epoch": 0.52, "learning_rate": 9.860466783371605e-06, "loss": 1.228, "step": 1692 }, { "epoch": 0.52, "learning_rate": 9.850500842868215e-06, "loss": 1.2998, "step": 1693 }, { "epoch": 0.52, "learning_rate": 9.84053505087836e-06, "loss": 1.2889, "step": 1694 }, { "epoch": 0.52, "learning_rate": 9.83056941730213e-06, "loss": 1.5742, "step": 1695 }, { "epoch": 0.52, "learning_rate": 9.820603952039456e-06, "loss": 1.8903, "step": 1696 }, { "epoch": 0.52, "learning_rate": 9.810638664990101e-06, "loss": 1.5518, "step": 1697 }, { "epoch": 0.52, "learning_rate": 9.800673566053657e-06, "loss": 1.8494, "step": 1698 }, { "epoch": 0.52, "learning_rate": 9.790708665129521e-06, "loss": 1.511, "step": 1699 }, { "epoch": 0.52, "learning_rate": 9.780743972116903e-06, "loss": 1.6458, "step": 1700 }, { "epoch": 0.52, "learning_rate": 9.770779496914791e-06, "loss": 1.8114, "step": 1701 }, { "epoch": 0.52, "learning_rate": 9.760815249421973e-06, "loss": 1.2366, "step": 1702 }, { "epoch": 0.52, "learning_rate": 9.750851239537002e-06, "loss": 1.3515, "step": 1703 }, { "epoch": 0.52, "learning_rate": 9.740887477158194e-06, "loss": 1.5392, "step": 1704 }, { "epoch": 0.52, "learning_rate": 9.730923972183629e-06, "loss": 1.1979, "step": 1705 }, { "epoch": 0.52, "learning_rate": 9.720960734511118e-06, "loss": 1.6623, "step": 1706 }, { "epoch": 0.53, "learning_rate": 9.710997774038215e-06, "loss": 1.6344, "step": 1707 }, { "epoch": 0.53, "learning_rate": 9.701035100662195e-06, "loss": 1.4079, "step": 1708 }, { "epoch": 0.53, "learning_rate": 9.691072724280053e-06, "loss": 1.396, "step": 1709 }, { "epoch": 0.53, "learning_rate": 9.681110654788483e-06, "loss": 1.3516, "step": 1710 }, { "epoch": 0.53, "learning_rate": 9.671148902083876e-06, "loss": 1.6512, "step": 1711 }, { "epoch": 0.53, "learning_rate": 9.661187476062308e-06, "loss": 1.9343, "step": 1712 }, { "epoch": 0.53, "learning_rate": 9.65122638661953e-06, "loss": 1.5428, "step": 1713 }, { "epoch": 0.53, "learning_rate": 9.641265643650968e-06, "loss": 1.5049, "step": 1714 }, { "epoch": 0.53, "learning_rate": 9.631305257051683e-06, "loss": 1.0889, "step": 1715 }, { "epoch": 0.53, "learning_rate": 9.621345236716403e-06, "loss": 1.412, "step": 1716 }, { "epoch": 0.53, "learning_rate": 9.61138559253948e-06, "loss": 1.1771, "step": 1717 }, { "epoch": 0.53, "learning_rate": 9.601426334414898e-06, "loss": 1.3467, "step": 1718 }, { "epoch": 0.53, "learning_rate": 9.591467472236255e-06, "loss": 1.3944, "step": 1719 }, { "epoch": 0.53, "learning_rate": 9.581509015896751e-06, "loss": 1.5119, "step": 1720 }, { "epoch": 0.53, "learning_rate": 9.571550975289197e-06, "loss": 1.4827, "step": 1721 }, { "epoch": 0.53, "learning_rate": 9.561593360305977e-06, "loss": 1.427, "step": 1722 }, { "epoch": 0.53, "learning_rate": 9.55163618083906e-06, "loss": 1.3517, "step": 1723 }, { "epoch": 0.53, "learning_rate": 9.541679446779972e-06, "loss": 1.3801, "step": 1724 }, { "epoch": 0.53, "learning_rate": 9.531723168019814e-06, "loss": 1.3607, "step": 1725 }, { "epoch": 0.53, "learning_rate": 9.521767354449218e-06, "loss": 1.1911, "step": 1726 }, { "epoch": 0.53, "learning_rate": 9.511812015958364e-06, "loss": 1.1936, "step": 1727 }, { "epoch": 0.53, "learning_rate": 9.501857162436957e-06, "loss": 1.7146, "step": 1728 }, { "epoch": 0.53, "learning_rate": 9.491902803774216e-06, "loss": 1.4931, "step": 1729 }, { "epoch": 0.53, "learning_rate": 9.481948949858876e-06, "loss": 1.5108, "step": 1730 }, { "epoch": 0.53, "learning_rate": 9.471995610579161e-06, "loss": 1.4073, "step": 1731 }, { "epoch": 0.53, "learning_rate": 9.462042795822799e-06, "loss": 1.4956, "step": 1732 }, { "epoch": 0.53, "learning_rate": 9.452090515476982e-06, "loss": 1.5541, "step": 1733 }, { "epoch": 0.53, "learning_rate": 9.442138779428376e-06, "loss": 1.7594, "step": 1734 }, { "epoch": 0.53, "learning_rate": 9.43218759756311e-06, "loss": 1.2847, "step": 1735 }, { "epoch": 0.53, "learning_rate": 9.422236979766755e-06, "loss": 1.4494, "step": 1736 }, { "epoch": 0.53, "learning_rate": 9.41228693592433e-06, "loss": 1.4843, "step": 1737 }, { "epoch": 0.53, "learning_rate": 9.40233747592028e-06, "loss": 1.2377, "step": 1738 }, { "epoch": 0.54, "learning_rate": 9.39238860963847e-06, "loss": 1.7476, "step": 1739 }, { "epoch": 0.54, "learning_rate": 9.382440346962169e-06, "loss": 1.0276, "step": 1740 }, { "epoch": 0.54, "learning_rate": 9.372492697774062e-06, "loss": 1.4952, "step": 1741 }, { "epoch": 0.54, "learning_rate": 9.362545671956213e-06, "loss": 2.1685, "step": 1742 }, { "epoch": 0.54, "learning_rate": 9.352599279390062e-06, "loss": 1.3387, "step": 1743 }, { "epoch": 0.54, "learning_rate": 9.342653529956432e-06, "loss": 1.3846, "step": 1744 }, { "epoch": 0.54, "learning_rate": 9.3327084335355e-06, "loss": 1.4131, "step": 1745 }, { "epoch": 0.54, "learning_rate": 9.322764000006798e-06, "loss": 2.6065, "step": 1746 }, { "epoch": 0.54, "learning_rate": 9.312820239249194e-06, "loss": 1.3596, "step": 1747 }, { "epoch": 0.54, "learning_rate": 9.302877161140897e-06, "loss": 1.381, "step": 1748 }, { "epoch": 0.54, "learning_rate": 9.292934775559428e-06, "loss": 1.4341, "step": 1749 }, { "epoch": 0.54, "learning_rate": 9.282993092381626e-06, "loss": 1.3982, "step": 1750 }, { "epoch": 0.54, "learning_rate": 9.273052121483628e-06, "loss": 1.4022, "step": 1751 }, { "epoch": 0.54, "learning_rate": 9.263111872740866e-06, "loss": 1.4594, "step": 1752 }, { "epoch": 0.54, "learning_rate": 9.253172356028058e-06, "loss": 1.641, "step": 1753 }, { "epoch": 0.54, "learning_rate": 9.24323358121919e-06, "loss": 1.6621, "step": 1754 }, { "epoch": 0.54, "learning_rate": 9.23329555818751e-06, "loss": 1.4911, "step": 1755 }, { "epoch": 0.54, "learning_rate": 9.223358296805526e-06, "loss": 1.1983, "step": 1756 }, { "epoch": 0.54, "learning_rate": 9.213421806944978e-06, "loss": 1.1931, "step": 1757 }, { "epoch": 0.54, "learning_rate": 9.20348609847685e-06, "loss": 1.5871, "step": 1758 }, { "epoch": 0.54, "learning_rate": 9.193551181271341e-06, "loss": 1.3482, "step": 1759 }, { "epoch": 0.54, "learning_rate": 9.183617065197879e-06, "loss": 1.4159, "step": 1760 }, { "epoch": 0.54, "learning_rate": 9.17368376012508e-06, "loss": 1.627, "step": 1761 }, { "epoch": 0.54, "learning_rate": 9.163751275920762e-06, "loss": 1.6991, "step": 1762 }, { "epoch": 0.54, "learning_rate": 9.153819622451921e-06, "loss": 1.6957, "step": 1763 }, { "epoch": 0.54, "learning_rate": 9.143888809584741e-06, "loss": 1.3581, "step": 1764 }, { "epoch": 0.54, "learning_rate": 9.133958847184559e-06, "loss": 1.8417, "step": 1765 }, { "epoch": 0.54, "learning_rate": 9.124029745115872e-06, "loss": 1.6578, "step": 1766 }, { "epoch": 0.54, "learning_rate": 9.114101513242317e-06, "loss": 1.4522, "step": 1767 }, { "epoch": 0.54, "learning_rate": 9.104174161426674e-06, "loss": 1.6761, "step": 1768 }, { "epoch": 0.54, "learning_rate": 9.094247699530845e-06, "loss": 1.5325, "step": 1769 }, { "epoch": 0.54, "learning_rate": 9.084322137415855e-06, "loss": 1.3307, "step": 1770 }, { "epoch": 0.54, "learning_rate": 9.074397484941816e-06, "loss": 1.45, "step": 1771 }, { "epoch": 0.55, "learning_rate": 9.064473751967958e-06, "loss": 1.976, "step": 1772 }, { "epoch": 0.55, "learning_rate": 9.054550948352587e-06, "loss": 1.5559, "step": 1773 }, { "epoch": 0.55, "learning_rate": 9.044629083953082e-06, "loss": 1.2764, "step": 1774 }, { "epoch": 0.55, "learning_rate": 9.0347081686259e-06, "loss": 1.5768, "step": 1775 }, { "epoch": 0.55, "learning_rate": 9.024788212226549e-06, "loss": 1.3257, "step": 1776 }, { "epoch": 0.55, "learning_rate": 9.014869224609584e-06, "loss": 1.3006, "step": 1777 }, { "epoch": 0.55, "learning_rate": 9.004951215628597e-06, "loss": 1.2039, "step": 1778 }, { "epoch": 0.55, "learning_rate": 8.995034195136209e-06, "loss": 1.5623, "step": 1779 }, { "epoch": 0.55, "learning_rate": 8.985118172984063e-06, "loss": 1.5361, "step": 1780 }, { "epoch": 0.55, "learning_rate": 8.975203159022804e-06, "loss": 1.2436, "step": 1781 }, { "epoch": 0.55, "learning_rate": 8.96528916310208e-06, "loss": 1.5689, "step": 1782 }, { "epoch": 0.55, "learning_rate": 8.95537619507052e-06, "loss": 1.7321, "step": 1783 }, { "epoch": 0.55, "learning_rate": 8.945464264775748e-06, "loss": 1.8205, "step": 1784 }, { "epoch": 0.55, "learning_rate": 8.935553382064336e-06, "loss": 1.574, "step": 1785 }, { "epoch": 0.55, "learning_rate": 8.925643556781828e-06, "loss": 1.5428, "step": 1786 }, { "epoch": 0.55, "learning_rate": 8.915734798772721e-06, "loss": 2.2525, "step": 1787 }, { "epoch": 0.55, "learning_rate": 8.905827117880441e-06, "loss": 1.4458, "step": 1788 }, { "epoch": 0.55, "learning_rate": 8.895920523947353e-06, "loss": 1.4314, "step": 1789 }, { "epoch": 0.55, "learning_rate": 8.886015026814736e-06, "loss": 1.3173, "step": 1790 }, { "epoch": 0.55, "learning_rate": 8.87611063632278e-06, "loss": 1.3583, "step": 1791 }, { "epoch": 0.55, "learning_rate": 8.866207362310583e-06, "loss": 1.4705, "step": 1792 }, { "epoch": 0.55, "learning_rate": 8.856305214616125e-06, "loss": 1.2249, "step": 1793 }, { "epoch": 0.55, "learning_rate": 8.846404203076273e-06, "loss": 1.7282, "step": 1794 }, { "epoch": 0.55, "learning_rate": 8.836504337526758e-06, "loss": 1.3321, "step": 1795 }, { "epoch": 0.55, "learning_rate": 8.826605627802182e-06, "loss": 1.4131, "step": 1796 }, { "epoch": 0.55, "learning_rate": 8.816708083735993e-06, "loss": 1.4614, "step": 1797 }, { "epoch": 0.55, "learning_rate": 8.806811715160485e-06, "loss": 0.9892, "step": 1798 }, { "epoch": 0.55, "learning_rate": 8.796916531906776e-06, "loss": 1.2521, "step": 1799 }, { "epoch": 0.55, "learning_rate": 8.787022543804814e-06, "loss": 1.1811, "step": 1800 }, { "epoch": 0.55, "learning_rate": 8.77712976068336e-06, "loss": 1.4258, "step": 1801 }, { "epoch": 0.55, "learning_rate": 8.76723819236997e-06, "loss": 1.515, "step": 1802 }, { "epoch": 0.55, "learning_rate": 8.757347848691006e-06, "loss": 1.9053, "step": 1803 }, { "epoch": 0.56, "learning_rate": 8.747458739471603e-06, "loss": 1.2192, "step": 1804 }, { "epoch": 0.56, "learning_rate": 8.737570874535674e-06, "loss": 1.4117, "step": 1805 }, { "epoch": 0.56, "learning_rate": 8.727684263705896e-06, "loss": 1.6294, "step": 1806 }, { "epoch": 0.56, "learning_rate": 8.717798916803692e-06, "loss": 1.4099, "step": 1807 }, { "epoch": 0.56, "learning_rate": 8.707914843649247e-06, "loss": 1.6191, "step": 1808 }, { "epoch": 0.56, "learning_rate": 8.698032054061464e-06, "loss": 1.2496, "step": 1809 }, { "epoch": 0.56, "learning_rate": 8.688150557857979e-06, "loss": 1.3684, "step": 1810 }, { "epoch": 0.56, "learning_rate": 8.678270364855137e-06, "loss": 1.6995, "step": 1811 }, { "epoch": 0.56, "learning_rate": 8.668391484868004e-06, "loss": 1.8171, "step": 1812 }, { "epoch": 0.56, "learning_rate": 8.658513927710317e-06, "loss": 1.2929, "step": 1813 }, { "epoch": 0.56, "learning_rate": 8.648637703194515e-06, "loss": 1.7156, "step": 1814 }, { "epoch": 0.56, "learning_rate": 8.638762821131717e-06, "loss": 1.4532, "step": 1815 }, { "epoch": 0.56, "learning_rate": 8.628889291331696e-06, "loss": 1.4685, "step": 1816 }, { "epoch": 0.56, "learning_rate": 8.619017123602888e-06, "loss": 1.1877, "step": 1817 }, { "epoch": 0.56, "learning_rate": 8.609146327752376e-06, "loss": 1.4762, "step": 1818 }, { "epoch": 0.56, "learning_rate": 8.59927691358588e-06, "loss": 1.3398, "step": 1819 }, { "epoch": 0.56, "learning_rate": 8.589408890907746e-06, "loss": 1.5266, "step": 1820 }, { "epoch": 0.56, "learning_rate": 8.57954226952094e-06, "loss": 1.57, "step": 1821 }, { "epoch": 0.56, "learning_rate": 8.569677059227033e-06, "loss": 1.5417, "step": 1822 }, { "epoch": 0.56, "learning_rate": 8.559813269826194e-06, "loss": 1.5402, "step": 1823 }, { "epoch": 0.56, "learning_rate": 8.549950911117186e-06, "loss": 1.4313, "step": 1824 }, { "epoch": 0.56, "learning_rate": 8.540089992897343e-06, "loss": 1.3216, "step": 1825 }, { "epoch": 0.56, "learning_rate": 8.530230524962578e-06, "loss": 1.2815, "step": 1826 }, { "epoch": 0.56, "learning_rate": 8.520372517107352e-06, "loss": 1.4336, "step": 1827 }, { "epoch": 0.56, "learning_rate": 8.51051597912468e-06, "loss": 1.4191, "step": 1828 }, { "epoch": 0.56, "learning_rate": 8.500660920806119e-06, "loss": 1.3227, "step": 1829 }, { "epoch": 0.56, "learning_rate": 8.490807351941753e-06, "loss": 1.6915, "step": 1830 }, { "epoch": 0.56, "learning_rate": 8.480955282320192e-06, "loss": 1.6988, "step": 1831 }, { "epoch": 0.56, "learning_rate": 8.471104721728551e-06, "loss": 1.4046, "step": 1832 }, { "epoch": 0.56, "learning_rate": 8.461255679952448e-06, "loss": 1.2138, "step": 1833 }, { "epoch": 0.56, "learning_rate": 8.451408166775987e-06, "loss": 1.5134, "step": 1834 }, { "epoch": 0.56, "learning_rate": 8.441562191981766e-06, "loss": 1.2223, "step": 1835 }, { "epoch": 0.56, "learning_rate": 8.431717765350842e-06, "loss": 1.573, "step": 1836 }, { "epoch": 0.57, "learning_rate": 8.42187489666274e-06, "loss": 1.9228, "step": 1837 }, { "epoch": 0.57, "learning_rate": 8.412033595695436e-06, "loss": 1.3358, "step": 1838 }, { "epoch": 0.57, "learning_rate": 8.40219387222535e-06, "loss": 1.2591, "step": 1839 }, { "epoch": 0.57, "learning_rate": 8.392355736027338e-06, "loss": 1.372, "step": 1840 }, { "epoch": 0.57, "learning_rate": 8.382519196874664e-06, "loss": 1.5483, "step": 1841 }, { "epoch": 0.57, "learning_rate": 8.372684264539025e-06, "loss": 1.2932, "step": 1842 }, { "epoch": 0.57, "learning_rate": 8.36285094879051e-06, "loss": 1.6244, "step": 1843 }, { "epoch": 0.57, "learning_rate": 8.353019259397609e-06, "loss": 1.7016, "step": 1844 }, { "epoch": 0.57, "learning_rate": 8.343189206127189e-06, "loss": 1.5351, "step": 1845 }, { "epoch": 0.57, "learning_rate": 8.333360798744496e-06, "loss": 1.7508, "step": 1846 }, { "epoch": 0.57, "learning_rate": 8.323534047013141e-06, "loss": 1.2294, "step": 1847 }, { "epoch": 0.57, "learning_rate": 8.313708960695094e-06, "loss": 1.3116, "step": 1848 }, { "epoch": 0.57, "learning_rate": 8.30388554955066e-06, "loss": 1.6935, "step": 1849 }, { "epoch": 0.57, "learning_rate": 8.294063823338486e-06, "loss": 1.533, "step": 1850 }, { "epoch": 0.57, "learning_rate": 8.284243791815549e-06, "loss": 1.6415, "step": 1851 }, { "epoch": 0.57, "learning_rate": 8.274425464737139e-06, "loss": 1.1968, "step": 1852 }, { "epoch": 0.57, "learning_rate": 8.26460885185685e-06, "loss": 1.3317, "step": 1853 }, { "epoch": 0.57, "learning_rate": 8.254793962926579e-06, "loss": 1.4101, "step": 1854 }, { "epoch": 0.57, "learning_rate": 8.244980807696504e-06, "loss": 1.6742, "step": 1855 }, { "epoch": 0.57, "learning_rate": 8.23516939591508e-06, "loss": 1.0346, "step": 1856 }, { "epoch": 0.57, "learning_rate": 8.22535973732904e-06, "loss": 1.6658, "step": 1857 }, { "epoch": 0.57, "learning_rate": 8.215551841683369e-06, "loss": 1.5758, "step": 1858 }, { "epoch": 0.57, "learning_rate": 8.2057457187213e-06, "loss": 1.5242, "step": 1859 }, { "epoch": 0.57, "learning_rate": 8.195941378184303e-06, "loss": 1.87, "step": 1860 }, { "epoch": 0.57, "learning_rate": 8.186138829812085e-06, "loss": 1.1414, "step": 1861 }, { "epoch": 0.57, "learning_rate": 8.17633808334256e-06, "loss": 1.1334, "step": 1862 }, { "epoch": 0.57, "learning_rate": 8.166539148511871e-06, "loss": 1.8316, "step": 1863 }, { "epoch": 0.57, "learning_rate": 8.156742035054343e-06, "loss": 1.6207, "step": 1864 }, { "epoch": 0.57, "learning_rate": 8.1469467527025e-06, "loss": 1.2682, "step": 1865 }, { "epoch": 0.57, "learning_rate": 8.137153311187044e-06, "loss": 1.7005, "step": 1866 }, { "epoch": 0.57, "learning_rate": 8.127361720236854e-06, "loss": 1.6663, "step": 1867 }, { "epoch": 0.57, "learning_rate": 8.117571989578969e-06, "loss": 1.264, "step": 1868 }, { "epoch": 0.58, "learning_rate": 8.107784128938566e-06, "loss": 1.9644, "step": 1869 }, { "epoch": 0.58, "learning_rate": 8.097998148038986e-06, "loss": 1.2632, "step": 1870 }, { "epoch": 0.58, "learning_rate": 8.08821405660169e-06, "loss": 1.5973, "step": 1871 }, { "epoch": 0.58, "learning_rate": 8.078431864346265e-06, "loss": 1.5157, "step": 1872 }, { "epoch": 0.58, "learning_rate": 8.068651580990407e-06, "loss": 1.4383, "step": 1873 }, { "epoch": 0.58, "learning_rate": 8.058873216249927e-06, "loss": 1.3048, "step": 1874 }, { "epoch": 0.58, "learning_rate": 8.04909677983872e-06, "loss": 1.7001, "step": 1875 }, { "epoch": 0.58, "learning_rate": 8.039322281468765e-06, "loss": 1.6673, "step": 1876 }, { "epoch": 0.58, "learning_rate": 8.02954973085012e-06, "loss": 1.4748, "step": 1877 }, { "epoch": 0.58, "learning_rate": 8.019779137690906e-06, "loss": 1.6206, "step": 1878 }, { "epoch": 0.58, "learning_rate": 8.010010511697303e-06, "loss": 1.5354, "step": 1879 }, { "epoch": 0.58, "learning_rate": 8.000243862573536e-06, "loss": 1.4034, "step": 1880 }, { "epoch": 0.58, "learning_rate": 7.990479200021858e-06, "loss": 1.5352, "step": 1881 }, { "epoch": 0.58, "learning_rate": 7.980716533742558e-06, "loss": 1.5687, "step": 1882 }, { "epoch": 0.58, "learning_rate": 7.970955873433947e-06, "loss": 1.3477, "step": 1883 }, { "epoch": 0.58, "learning_rate": 7.96119722879232e-06, "loss": 1.1569, "step": 1884 }, { "epoch": 0.58, "learning_rate": 7.95144060951199e-06, "loss": 1.6488, "step": 1885 }, { "epoch": 0.58, "learning_rate": 7.941686025285257e-06, "loss": 1.3721, "step": 1886 }, { "epoch": 0.58, "learning_rate": 7.93193348580239e-06, "loss": 1.387, "step": 1887 }, { "epoch": 0.58, "learning_rate": 7.922183000751635e-06, "loss": 1.423, "step": 1888 }, { "epoch": 0.58, "learning_rate": 7.912434579819191e-06, "loss": 1.6236, "step": 1889 }, { "epoch": 0.58, "learning_rate": 7.902688232689212e-06, "loss": 1.1951, "step": 1890 }, { "epoch": 0.58, "learning_rate": 7.892943969043787e-06, "loss": 1.5847, "step": 1891 }, { "epoch": 0.58, "learning_rate": 7.88320179856294e-06, "loss": 1.4674, "step": 1892 }, { "epoch": 0.58, "learning_rate": 7.873461730924612e-06, "loss": 1.2043, "step": 1893 }, { "epoch": 0.58, "learning_rate": 7.863723775804651e-06, "loss": 1.506, "step": 1894 }, { "epoch": 0.58, "learning_rate": 7.85398794287682e-06, "loss": 1.4507, "step": 1895 }, { "epoch": 0.58, "learning_rate": 7.84425424181276e-06, "loss": 1.8841, "step": 1896 }, { "epoch": 0.58, "learning_rate": 7.834522682282003e-06, "loss": 1.422, "step": 1897 }, { "epoch": 0.58, "learning_rate": 7.824793273951945e-06, "loss": 1.138, "step": 1898 }, { "epoch": 0.58, "learning_rate": 7.815066026487855e-06, "loss": 1.423, "step": 1899 }, { "epoch": 0.58, "learning_rate": 7.805340949552846e-06, "loss": 1.2234, "step": 1900 }, { "epoch": 0.58, "learning_rate": 7.795618052807879e-06, "loss": 1.4055, "step": 1901 }, { "epoch": 0.59, "learning_rate": 7.785897345911755e-06, "loss": 1.5733, "step": 1902 }, { "epoch": 0.59, "learning_rate": 7.77617883852109e-06, "loss": 1.7592, "step": 1903 }, { "epoch": 0.59, "learning_rate": 7.76646254029032e-06, "loss": 1.4252, "step": 1904 }, { "epoch": 0.59, "learning_rate": 7.756748460871681e-06, "loss": 1.2516, "step": 1905 }, { "epoch": 0.59, "learning_rate": 7.747036609915217e-06, "loss": 1.6656, "step": 1906 }, { "epoch": 0.59, "learning_rate": 7.737326997068748e-06, "loss": 1.4984, "step": 1907 }, { "epoch": 0.59, "learning_rate": 7.727619631977869e-06, "loss": 1.5011, "step": 1908 }, { "epoch": 0.59, "learning_rate": 7.71791452428595e-06, "loss": 1.5291, "step": 1909 }, { "epoch": 0.59, "learning_rate": 7.708211683634112e-06, "loss": 1.1616, "step": 1910 }, { "epoch": 0.59, "learning_rate": 7.698511119661233e-06, "loss": 1.3613, "step": 1911 }, { "epoch": 0.59, "learning_rate": 7.68881284200391e-06, "loss": 1.1526, "step": 1912 }, { "epoch": 0.59, "learning_rate": 7.679116860296493e-06, "loss": 2.1535, "step": 1913 }, { "epoch": 0.59, "learning_rate": 7.669423184171037e-06, "loss": 1.5564, "step": 1914 }, { "epoch": 0.59, "learning_rate": 7.659731823257309e-06, "loss": 1.3777, "step": 1915 }, { "epoch": 0.59, "learning_rate": 7.650042787182773e-06, "loss": 1.7495, "step": 1916 }, { "epoch": 0.59, "learning_rate": 7.64035608557259e-06, "loss": 1.4611, "step": 1917 }, { "epoch": 0.59, "learning_rate": 7.6306717280496e-06, "loss": 2.4728, "step": 1918 }, { "epoch": 0.59, "learning_rate": 7.620989724234311e-06, "loss": 1.6802, "step": 1919 }, { "epoch": 0.59, "learning_rate": 7.611310083744897e-06, "loss": 1.1031, "step": 1920 }, { "epoch": 0.59, "learning_rate": 7.601632816197179e-06, "loss": 1.3549, "step": 1921 }, { "epoch": 0.59, "learning_rate": 7.591957931204629e-06, "loss": 1.6642, "step": 1922 }, { "epoch": 0.59, "learning_rate": 7.582285438378344e-06, "loss": 1.2271, "step": 1923 }, { "epoch": 0.59, "learning_rate": 7.572615347327048e-06, "loss": 1.5193, "step": 1924 }, { "epoch": 0.59, "learning_rate": 7.562947667657081e-06, "loss": 1.4017, "step": 1925 }, { "epoch": 0.59, "learning_rate": 7.553282408972382e-06, "loss": 1.5452, "step": 1926 }, { "epoch": 0.59, "learning_rate": 7.543619580874489e-06, "loss": 1.5713, "step": 1927 }, { "epoch": 0.59, "learning_rate": 7.533959192962525e-06, "loss": 1.5599, "step": 1928 }, { "epoch": 0.59, "learning_rate": 7.52430125483319e-06, "loss": 1.2469, "step": 1929 }, { "epoch": 0.59, "learning_rate": 7.514645776080747e-06, "loss": 1.4363, "step": 1930 }, { "epoch": 0.59, "learning_rate": 7.50499276629702e-06, "loss": 1.2099, "step": 1931 }, { "epoch": 0.59, "learning_rate": 7.495342235071377e-06, "loss": 1.7733, "step": 1932 }, { "epoch": 0.59, "learning_rate": 7.485694191990724e-06, "loss": 1.4292, "step": 1933 }, { "epoch": 0.6, "learning_rate": 7.476048646639498e-06, "loss": 1.6801, "step": 1934 }, { "epoch": 0.6, "learning_rate": 7.466405608599653e-06, "loss": 1.5798, "step": 1935 }, { "epoch": 0.6, "learning_rate": 7.456765087450652e-06, "loss": 1.2582, "step": 1936 }, { "epoch": 0.6, "learning_rate": 7.447127092769453e-06, "loss": 1.2624, "step": 1937 }, { "epoch": 0.6, "learning_rate": 7.437491634130518e-06, "loss": 1.563, "step": 1938 }, { "epoch": 0.6, "learning_rate": 7.4278587211057785e-06, "loss": 1.1712, "step": 1939 }, { "epoch": 0.6, "learning_rate": 7.418228363264631e-06, "loss": 1.3578, "step": 1940 }, { "epoch": 0.6, "learning_rate": 7.408600570173951e-06, "loss": 1.3691, "step": 1941 }, { "epoch": 0.6, "learning_rate": 7.398975351398054e-06, "loss": 1.6906, "step": 1942 }, { "epoch": 0.6, "learning_rate": 7.389352716498701e-06, "loss": 2.3196, "step": 1943 }, { "epoch": 0.6, "learning_rate": 7.379732675035085e-06, "loss": 1.7796, "step": 1944 }, { "epoch": 0.6, "learning_rate": 7.370115236563828e-06, "loss": 1.4547, "step": 1945 }, { "epoch": 0.6, "learning_rate": 7.360500410638962e-06, "loss": 1.5591, "step": 1946 }, { "epoch": 0.6, "learning_rate": 7.350888206811923e-06, "loss": 1.4449, "step": 1947 }, { "epoch": 0.6, "learning_rate": 7.3412786346315415e-06, "loss": 1.3225, "step": 1948 }, { "epoch": 0.6, "learning_rate": 7.3316717036440356e-06, "loss": 0.8577, "step": 1949 }, { "epoch": 0.6, "learning_rate": 7.322067423393002e-06, "loss": 1.4257, "step": 1950 }, { "epoch": 0.6, "learning_rate": 7.312465803419402e-06, "loss": 1.7571, "step": 1951 }, { "epoch": 0.6, "learning_rate": 7.302866853261552e-06, "loss": 1.6365, "step": 1952 }, { "epoch": 0.6, "learning_rate": 7.29327058245512e-06, "loss": 1.3949, "step": 1953 }, { "epoch": 0.6, "learning_rate": 7.283677000533106e-06, "loss": 1.2676, "step": 1954 }, { "epoch": 0.6, "learning_rate": 7.2740861170258445e-06, "loss": 1.2759, "step": 1955 }, { "epoch": 0.6, "learning_rate": 7.264497941460988e-06, "loss": 1.559, "step": 1956 }, { "epoch": 0.6, "learning_rate": 7.254912483363503e-06, "loss": 1.0111, "step": 1957 }, { "epoch": 0.6, "learning_rate": 7.2453297522556475e-06, "loss": 1.2686, "step": 1958 }, { "epoch": 0.6, "learning_rate": 7.235749757656978e-06, "loss": 1.3747, "step": 1959 }, { "epoch": 0.6, "learning_rate": 7.226172509084326e-06, "loss": 1.0332, "step": 1960 }, { "epoch": 0.6, "learning_rate": 7.216598016051804e-06, "loss": 1.1655, "step": 1961 }, { "epoch": 0.6, "learning_rate": 7.20702628807078e-06, "loss": 1.52, "step": 1962 }, { "epoch": 0.6, "learning_rate": 7.197457334649878e-06, "loss": 1.3386, "step": 1963 }, { "epoch": 0.6, "learning_rate": 7.1878911652949626e-06, "loss": 1.2893, "step": 1964 }, { "epoch": 0.6, "learning_rate": 7.178327789509134e-06, "loss": 1.3839, "step": 1965 }, { "epoch": 0.6, "learning_rate": 7.168767216792725e-06, "loss": 1.6415, "step": 1966 }, { "epoch": 0.61, "learning_rate": 7.1592094566432745e-06, "loss": 1.1777, "step": 1967 }, { "epoch": 0.61, "learning_rate": 7.149654518555526e-06, "loss": 1.2382, "step": 1968 }, { "epoch": 0.61, "learning_rate": 7.140102412021426e-06, "loss": 1.3861, "step": 1969 }, { "epoch": 0.61, "learning_rate": 7.130553146530105e-06, "loss": 1.4882, "step": 1970 }, { "epoch": 0.61, "learning_rate": 7.121006731567873e-06, "loss": 1.5152, "step": 1971 }, { "epoch": 0.61, "learning_rate": 7.111463176618205e-06, "loss": 1.5381, "step": 1972 }, { "epoch": 0.61, "learning_rate": 7.101922491161739e-06, "loss": 1.5232, "step": 1973 }, { "epoch": 0.61, "learning_rate": 7.092384684676263e-06, "loss": 1.5662, "step": 1974 }, { "epoch": 0.61, "learning_rate": 7.0828497666366965e-06, "loss": 1.2246, "step": 1975 }, { "epoch": 0.61, "learning_rate": 7.073317746515097e-06, "loss": 1.3451, "step": 1976 }, { "epoch": 0.61, "learning_rate": 7.063788633780646e-06, "loss": 1.4567, "step": 1977 }, { "epoch": 0.61, "learning_rate": 7.054262437899628e-06, "loss": 1.5154, "step": 1978 }, { "epoch": 0.61, "learning_rate": 7.044739168335436e-06, "loss": 1.3573, "step": 1979 }, { "epoch": 0.61, "learning_rate": 7.035218834548554e-06, "loss": 1.2748, "step": 1980 }, { "epoch": 0.61, "learning_rate": 7.025701445996553e-06, "loss": 1.7599, "step": 1981 }, { "epoch": 0.61, "learning_rate": 7.016187012134069e-06, "loss": 1.0552, "step": 1982 }, { "epoch": 0.61, "learning_rate": 7.006675542412811e-06, "loss": 1.2412, "step": 1983 }, { "epoch": 0.61, "learning_rate": 6.9971670462815425e-06, "loss": 1.5353, "step": 1984 }, { "epoch": 0.61, "learning_rate": 6.987661533186072e-06, "loss": 1.7036, "step": 1985 }, { "epoch": 0.61, "learning_rate": 6.978159012569243e-06, "loss": 1.9498, "step": 1986 }, { "epoch": 0.61, "learning_rate": 6.96865949387093e-06, "loss": 1.4701, "step": 1987 }, { "epoch": 0.61, "learning_rate": 6.959162986528019e-06, "loss": 1.6037, "step": 1988 }, { "epoch": 0.61, "learning_rate": 6.949669499974412e-06, "loss": 1.2395, "step": 1989 }, { "epoch": 0.61, "learning_rate": 6.940179043641005e-06, "loss": 1.9126, "step": 1990 }, { "epoch": 0.61, "learning_rate": 6.930691626955688e-06, "loss": 1.4737, "step": 1991 }, { "epoch": 0.61, "learning_rate": 6.921207259343323e-06, "loss": 1.3704, "step": 1992 }, { "epoch": 0.61, "learning_rate": 6.911725950225756e-06, "loss": 1.6618, "step": 1993 }, { "epoch": 0.61, "learning_rate": 6.902247709021783e-06, "loss": 1.3129, "step": 1994 }, { "epoch": 0.61, "learning_rate": 6.89277254514716e-06, "loss": 1.6496, "step": 1995 }, { "epoch": 0.61, "learning_rate": 6.88330046801458e-06, "loss": 1.6016, "step": 1996 }, { "epoch": 0.61, "learning_rate": 6.873831487033671e-06, "loss": 1.5081, "step": 1997 }, { "epoch": 0.61, "learning_rate": 6.86436561161099e-06, "loss": 1.5485, "step": 1998 }, { "epoch": 0.62, "learning_rate": 6.8549028511500016e-06, "loss": 1.3337, "step": 1999 }, { "epoch": 0.62, "learning_rate": 6.845443215051084e-06, "loss": 1.5207, "step": 2000 }, { "epoch": 0.62, "learning_rate": 6.835986712711507e-06, "loss": 1.6363, "step": 2001 }, { "epoch": 0.62, "learning_rate": 6.826533353525427e-06, "loss": 1.5806, "step": 2002 }, { "epoch": 0.62, "learning_rate": 6.81708314688388e-06, "loss": 2.0782, "step": 2003 }, { "epoch": 0.62, "learning_rate": 6.807636102174764e-06, "loss": 1.1988, "step": 2004 }, { "epoch": 0.62, "learning_rate": 6.798192228782847e-06, "loss": 1.31, "step": 2005 }, { "epoch": 0.62, "learning_rate": 6.788751536089739e-06, "loss": 1.7434, "step": 2006 }, { "epoch": 0.62, "learning_rate": 6.77931403347389e-06, "loss": 1.2421, "step": 2007 }, { "epoch": 0.62, "learning_rate": 6.769879730310583e-06, "loss": 1.4289, "step": 2008 }, { "epoch": 0.62, "learning_rate": 6.7604486359719305e-06, "loss": 1.6098, "step": 2009 }, { "epoch": 0.62, "learning_rate": 6.751020759826836e-06, "loss": 1.3215, "step": 2010 }, { "epoch": 0.62, "learning_rate": 6.741596111241026e-06, "loss": 1.2353, "step": 2011 }, { "epoch": 0.62, "learning_rate": 6.7321746995770144e-06, "loss": 1.2478, "step": 2012 }, { "epoch": 0.62, "learning_rate": 6.722756534194098e-06, "loss": 1.6369, "step": 2013 }, { "epoch": 0.62, "learning_rate": 6.713341624448351e-06, "loss": 1.5766, "step": 2014 }, { "epoch": 0.62, "learning_rate": 6.7039299796926086e-06, "loss": 1.3183, "step": 2015 }, { "epoch": 0.62, "learning_rate": 6.694521609276471e-06, "loss": 1.7239, "step": 2016 }, { "epoch": 0.62, "learning_rate": 6.68511652254628e-06, "loss": 1.3676, "step": 2017 }, { "epoch": 0.62, "learning_rate": 6.675714728845114e-06, "loss": 1.4914, "step": 2018 }, { "epoch": 0.62, "learning_rate": 6.666316237512784e-06, "loss": 1.3917, "step": 2019 }, { "epoch": 0.62, "learning_rate": 6.656921057885815e-06, "loss": 1.4206, "step": 2020 }, { "epoch": 0.62, "learning_rate": 6.6475291992974534e-06, "loss": 1.3812, "step": 2021 }, { "epoch": 0.62, "learning_rate": 6.638140671077633e-06, "loss": 1.59, "step": 2022 }, { "epoch": 0.62, "learning_rate": 6.628755482552989e-06, "loss": 1.0776, "step": 2023 }, { "epoch": 0.62, "learning_rate": 6.619373643046829e-06, "loss": 1.4771, "step": 2024 }, { "epoch": 0.62, "learning_rate": 6.609995161879146e-06, "loss": 1.3245, "step": 2025 }, { "epoch": 0.62, "learning_rate": 6.600620048366586e-06, "loss": 0.9801, "step": 2026 }, { "epoch": 0.62, "learning_rate": 6.591248311822454e-06, "loss": 1.7605, "step": 2027 }, { "epoch": 0.62, "learning_rate": 6.581879961556703e-06, "loss": 1.458, "step": 2028 }, { "epoch": 0.62, "learning_rate": 6.572515006875919e-06, "loss": 1.4061, "step": 2029 }, { "epoch": 0.62, "learning_rate": 6.563153457083315e-06, "loss": 1.4054, "step": 2030 }, { "epoch": 0.62, "learning_rate": 6.553795321478719e-06, "loss": 1.5378, "step": 2031 }, { "epoch": 0.63, "learning_rate": 6.544440609358575e-06, "loss": 1.4288, "step": 2032 }, { "epoch": 0.63, "learning_rate": 6.535089330015919e-06, "loss": 1.9646, "step": 2033 }, { "epoch": 0.63, "learning_rate": 6.525741492740378e-06, "loss": 1.3538, "step": 2034 }, { "epoch": 0.63, "learning_rate": 6.5163971068181596e-06, "loss": 1.519, "step": 2035 }, { "epoch": 0.63, "learning_rate": 6.507056181532045e-06, "loss": 1.4911, "step": 2036 }, { "epoch": 0.63, "learning_rate": 6.497718726161382e-06, "loss": 1.3094, "step": 2037 }, { "epoch": 0.63, "learning_rate": 6.488384749982054e-06, "loss": 1.5947, "step": 2038 }, { "epoch": 0.63, "learning_rate": 6.479054262266509e-06, "loss": 1.1979, "step": 2039 }, { "epoch": 0.63, "learning_rate": 6.469727272283717e-06, "loss": 1.1595, "step": 2040 }, { "epoch": 0.63, "learning_rate": 6.460403789299178e-06, "loss": 1.3664, "step": 2041 }, { "epoch": 0.63, "learning_rate": 6.451083822574907e-06, "loss": 1.454, "step": 2042 }, { "epoch": 0.63, "learning_rate": 6.441767381369422e-06, "loss": 1.1066, "step": 2043 }, { "epoch": 0.63, "learning_rate": 6.432454474937747e-06, "loss": 1.2649, "step": 2044 }, { "epoch": 0.63, "learning_rate": 6.42314511253139e-06, "loss": 1.2013, "step": 2045 }, { "epoch": 0.63, "learning_rate": 6.413839303398337e-06, "loss": 1.3546, "step": 2046 }, { "epoch": 0.63, "learning_rate": 6.404537056783042e-06, "loss": 1.7463, "step": 2047 }, { "epoch": 0.63, "learning_rate": 6.395238381926432e-06, "loss": 1.6325, "step": 2048 }, { "epoch": 0.63, "learning_rate": 6.385943288065871e-06, "loss": 1.3127, "step": 2049 }, { "epoch": 0.63, "learning_rate": 6.376651784435174e-06, "loss": 1.1262, "step": 2050 }, { "epoch": 0.63, "learning_rate": 6.367363880264587e-06, "loss": 1.6402, "step": 2051 }, { "epoch": 0.63, "learning_rate": 6.358079584780781e-06, "loss": 1.3605, "step": 2052 }, { "epoch": 0.63, "learning_rate": 6.34879890720684e-06, "loss": 1.372, "step": 2053 }, { "epoch": 0.63, "learning_rate": 6.339521856762254e-06, "loss": 1.2671, "step": 2054 }, { "epoch": 0.63, "learning_rate": 6.330248442662919e-06, "loss": 1.3776, "step": 2055 }, { "epoch": 0.63, "learning_rate": 6.320978674121103e-06, "loss": 1.6665, "step": 2056 }, { "epoch": 0.63, "learning_rate": 6.3117125603454665e-06, "loss": 2.2293, "step": 2057 }, { "epoch": 0.63, "learning_rate": 6.3024501105410295e-06, "loss": 1.6166, "step": 2058 }, { "epoch": 0.63, "learning_rate": 6.2931913339091755e-06, "loss": 1.5341, "step": 2059 }, { "epoch": 0.63, "learning_rate": 6.283936239647643e-06, "loss": 1.588, "step": 2060 }, { "epoch": 0.63, "learning_rate": 6.274684836950509e-06, "loss": 1.2892, "step": 2061 }, { "epoch": 0.63, "learning_rate": 6.265437135008181e-06, "loss": 1.5453, "step": 2062 }, { "epoch": 0.63, "learning_rate": 6.256193143007392e-06, "loss": 1.7253, "step": 2063 }, { "epoch": 0.64, "learning_rate": 6.2469528701311935e-06, "loss": 2.1678, "step": 2064 }, { "epoch": 0.64, "learning_rate": 6.237716325558939e-06, "loss": 1.3906, "step": 2065 }, { "epoch": 0.64, "learning_rate": 6.228483518466271e-06, "loss": 1.3488, "step": 2066 }, { "epoch": 0.64, "learning_rate": 6.219254458025132e-06, "loss": 1.6074, "step": 2067 }, { "epoch": 0.64, "learning_rate": 6.210029153403738e-06, "loss": 1.3285, "step": 2068 }, { "epoch": 0.64, "learning_rate": 6.2008076137665694e-06, "loss": 1.2593, "step": 2069 }, { "epoch": 0.64, "learning_rate": 6.191589848274369e-06, "loss": 1.4621, "step": 2070 }, { "epoch": 0.64, "learning_rate": 6.1823758660841356e-06, "loss": 1.3235, "step": 2071 }, { "epoch": 0.64, "learning_rate": 6.173165676349103e-06, "loss": 1.5771, "step": 2072 }, { "epoch": 0.64, "learning_rate": 6.163959288218741e-06, "loss": 1.3293, "step": 2073 }, { "epoch": 0.64, "learning_rate": 6.154756710838739e-06, "loss": 1.1208, "step": 2074 }, { "epoch": 0.64, "learning_rate": 6.145557953351002e-06, "loss": 1.5575, "step": 2075 }, { "epoch": 0.64, "learning_rate": 6.136363024893647e-06, "loss": 1.0037, "step": 2076 }, { "epoch": 0.64, "learning_rate": 6.12717193460098e-06, "loss": 1.3171, "step": 2077 }, { "epoch": 0.64, "learning_rate": 6.117984691603493e-06, "loss": 1.5036, "step": 2078 }, { "epoch": 0.64, "learning_rate": 6.108801305027863e-06, "loss": 1.5682, "step": 2079 }, { "epoch": 0.64, "learning_rate": 6.099621783996934e-06, "loss": 1.3691, "step": 2080 }, { "epoch": 0.64, "learning_rate": 6.0904461376297e-06, "loss": 1.6378, "step": 2081 }, { "epoch": 0.64, "learning_rate": 6.081274375041317e-06, "loss": 1.4463, "step": 2082 }, { "epoch": 0.64, "learning_rate": 6.072106505343085e-06, "loss": 1.3154, "step": 2083 }, { "epoch": 0.64, "learning_rate": 6.062942537642427e-06, "loss": 2.0535, "step": 2084 }, { "epoch": 0.64, "learning_rate": 6.053782481042894e-06, "loss": 1.464, "step": 2085 }, { "epoch": 0.64, "learning_rate": 6.044626344644151e-06, "loss": 1.4481, "step": 2086 }, { "epoch": 0.64, "learning_rate": 6.035474137541974e-06, "loss": 1.6842, "step": 2087 }, { "epoch": 0.64, "learning_rate": 6.026325868828228e-06, "loss": 1.5216, "step": 2088 }, { "epoch": 0.64, "learning_rate": 6.017181547590869e-06, "loss": 1.2619, "step": 2089 }, { "epoch": 0.64, "learning_rate": 6.008041182913933e-06, "loss": 1.5203, "step": 2090 }, { "epoch": 0.64, "learning_rate": 5.998904783877521e-06, "loss": 1.4824, "step": 2091 }, { "epoch": 0.64, "learning_rate": 5.989772359557799e-06, "loss": 1.4267, "step": 2092 }, { "epoch": 0.64, "learning_rate": 5.980643919026985e-06, "loss": 1.8765, "step": 2093 }, { "epoch": 0.64, "learning_rate": 5.971519471353335e-06, "loss": 1.3318, "step": 2094 }, { "epoch": 0.64, "learning_rate": 5.96239902560114e-06, "loss": 1.4291, "step": 2095 }, { "epoch": 0.64, "learning_rate": 5.953282590830718e-06, "loss": 1.4627, "step": 2096 }, { "epoch": 0.65, "learning_rate": 5.944170176098398e-06, "loss": 1.3275, "step": 2097 }, { "epoch": 0.65, "learning_rate": 5.935061790456519e-06, "loss": 1.3707, "step": 2098 }, { "epoch": 0.65, "learning_rate": 5.9259574429534195e-06, "loss": 1.4946, "step": 2099 }, { "epoch": 0.65, "learning_rate": 5.916857142633421e-06, "loss": 1.6703, "step": 2100 }, { "epoch": 0.65, "learning_rate": 5.907760898536828e-06, "loss": 1.4886, "step": 2101 }, { "epoch": 0.65, "learning_rate": 5.898668719699914e-06, "loss": 1.3178, "step": 2102 }, { "epoch": 0.65, "learning_rate": 5.889580615154915e-06, "loss": 1.3641, "step": 2103 }, { "epoch": 0.65, "learning_rate": 5.880496593930021e-06, "loss": 1.0599, "step": 2104 }, { "epoch": 0.65, "learning_rate": 5.871416665049361e-06, "loss": 1.5125, "step": 2105 }, { "epoch": 0.65, "learning_rate": 5.8623408375330035e-06, "loss": 1.1797, "step": 2106 }, { "epoch": 0.65, "learning_rate": 5.853269120396938e-06, "loss": 1.2804, "step": 2107 }, { "epoch": 0.65, "learning_rate": 5.844201522653081e-06, "loss": 1.4735, "step": 2108 }, { "epoch": 0.65, "learning_rate": 5.835138053309239e-06, "loss": 1.3389, "step": 2109 }, { "epoch": 0.65, "learning_rate": 5.82607872136913e-06, "loss": 1.2159, "step": 2110 }, { "epoch": 0.65, "learning_rate": 5.8170235358323665e-06, "loss": 1.3314, "step": 2111 }, { "epoch": 0.65, "learning_rate": 5.8079725056944245e-06, "loss": 1.4456, "step": 2112 }, { "epoch": 0.65, "learning_rate": 5.798925639946669e-06, "loss": 1.3746, "step": 2113 }, { "epoch": 0.65, "learning_rate": 5.789882947576321e-06, "loss": 1.6103, "step": 2114 }, { "epoch": 0.65, "learning_rate": 5.780844437566451e-06, "loss": 1.2023, "step": 2115 }, { "epoch": 0.65, "learning_rate": 5.771810118895984e-06, "loss": 1.123, "step": 2116 }, { "epoch": 0.65, "learning_rate": 5.762780000539681e-06, "loss": 1.7342, "step": 2117 }, { "epoch": 0.65, "learning_rate": 5.753754091468116e-06, "loss": 1.2948, "step": 2118 }, { "epoch": 0.65, "learning_rate": 5.7447324006477016e-06, "loss": 1.6341, "step": 2119 }, { "epoch": 0.65, "learning_rate": 5.735714937040642e-06, "loss": 1.6557, "step": 2120 }, { "epoch": 0.65, "learning_rate": 5.726701709604955e-06, "loss": 0.9759, "step": 2121 }, { "epoch": 0.65, "learning_rate": 5.717692727294453e-06, "loss": 1.3825, "step": 2122 }, { "epoch": 0.65, "learning_rate": 5.7086879990587064e-06, "loss": 1.4902, "step": 2123 }, { "epoch": 0.65, "learning_rate": 5.699687533843086e-06, "loss": 1.6783, "step": 2124 }, { "epoch": 0.65, "learning_rate": 5.690691340588721e-06, "loss": 1.4209, "step": 2125 }, { "epoch": 0.65, "learning_rate": 5.681699428232488e-06, "loss": 1.3473, "step": 2126 }, { "epoch": 0.65, "learning_rate": 5.6727118057070255e-06, "loss": 1.3774, "step": 2127 }, { "epoch": 0.65, "learning_rate": 5.6637284819406936e-06, "loss": 1.5165, "step": 2128 }, { "epoch": 0.66, "learning_rate": 5.654749465857594e-06, "loss": 1.3847, "step": 2129 }, { "epoch": 0.66, "learning_rate": 5.64577476637755e-06, "loss": 1.4085, "step": 2130 }, { "epoch": 0.66, "learning_rate": 5.636804392416084e-06, "loss": 1.5003, "step": 2131 }, { "epoch": 0.66, "learning_rate": 5.627838352884435e-06, "loss": 1.4249, "step": 2132 }, { "epoch": 0.66, "learning_rate": 5.618876656689534e-06, "loss": 1.4116, "step": 2133 }, { "epoch": 0.66, "learning_rate": 5.609919312733987e-06, "loss": 1.7253, "step": 2134 }, { "epoch": 0.66, "learning_rate": 5.600966329916091e-06, "loss": 1.4766, "step": 2135 }, { "epoch": 0.66, "learning_rate": 5.592017717129801e-06, "loss": 1.6466, "step": 2136 }, { "epoch": 0.66, "learning_rate": 5.583073483264727e-06, "loss": 1.7244, "step": 2137 }, { "epoch": 0.66, "learning_rate": 5.574133637206144e-06, "loss": 1.36, "step": 2138 }, { "epoch": 0.66, "learning_rate": 5.565198187834952e-06, "loss": 2.148, "step": 2139 }, { "epoch": 0.66, "learning_rate": 5.556267144027692e-06, "loss": 1.3596, "step": 2140 }, { "epoch": 0.66, "learning_rate": 5.547340514656534e-06, "loss": 1.3419, "step": 2141 }, { "epoch": 0.66, "learning_rate": 5.538418308589246e-06, "loss": 1.5568, "step": 2142 }, { "epoch": 0.66, "learning_rate": 5.529500534689216e-06, "loss": 1.6058, "step": 2143 }, { "epoch": 0.66, "learning_rate": 5.520587201815428e-06, "loss": 1.4502, "step": 2144 }, { "epoch": 0.66, "learning_rate": 5.511678318822443e-06, "loss": 1.4745, "step": 2145 }, { "epoch": 0.66, "learning_rate": 5.502773894560418e-06, "loss": 1.8227, "step": 2146 }, { "epoch": 0.66, "learning_rate": 5.493873937875063e-06, "loss": 1.4998, "step": 2147 }, { "epoch": 0.66, "learning_rate": 5.484978457607664e-06, "loss": 1.413, "step": 2148 }, { "epoch": 0.66, "learning_rate": 5.476087462595058e-06, "loss": 1.6073, "step": 2149 }, { "epoch": 0.66, "learning_rate": 5.467200961669619e-06, "loss": 1.3238, "step": 2150 }, { "epoch": 0.66, "learning_rate": 5.458318963659257e-06, "loss": 1.3572, "step": 2151 }, { "epoch": 0.66, "learning_rate": 5.4494414773874205e-06, "loss": 1.3029, "step": 2152 }, { "epoch": 0.66, "learning_rate": 5.440568511673061e-06, "loss": 1.2352, "step": 2153 }, { "epoch": 0.66, "learning_rate": 5.431700075330654e-06, "loss": 1.9073, "step": 2154 }, { "epoch": 0.66, "learning_rate": 5.422836177170159e-06, "loss": 1.6075, "step": 2155 }, { "epoch": 0.66, "learning_rate": 5.4139768259970385e-06, "loss": 1.2912, "step": 2156 }, { "epoch": 0.66, "learning_rate": 5.405122030612243e-06, "loss": 1.4296, "step": 2157 }, { "epoch": 0.66, "learning_rate": 5.39627179981218e-06, "loss": 1.1789, "step": 2158 }, { "epoch": 0.66, "learning_rate": 5.387426142388736e-06, "loss": 1.2714, "step": 2159 }, { "epoch": 0.66, "learning_rate": 5.378585067129256e-06, "loss": 1.6964, "step": 2160 }, { "epoch": 0.66, "learning_rate": 5.369748582816517e-06, "loss": 1.5475, "step": 2161 }, { "epoch": 0.67, "learning_rate": 5.360916698228756e-06, "loss": 1.6773, "step": 2162 }, { "epoch": 0.67, "learning_rate": 5.3520894221396214e-06, "loss": 1.3107, "step": 2163 }, { "epoch": 0.67, "learning_rate": 5.343266763318202e-06, "loss": 1.5463, "step": 2164 }, { "epoch": 0.67, "learning_rate": 5.334448730528985e-06, "loss": 1.3822, "step": 2165 }, { "epoch": 0.67, "learning_rate": 5.325635332531864e-06, "loss": 1.6067, "step": 2166 }, { "epoch": 0.67, "learning_rate": 5.316826578082136e-06, "loss": 1.4894, "step": 2167 }, { "epoch": 0.67, "learning_rate": 5.308022475930484e-06, "loss": 1.4188, "step": 2168 }, { "epoch": 0.67, "learning_rate": 5.299223034822959e-06, "loss": 1.558, "step": 2169 }, { "epoch": 0.67, "learning_rate": 5.290428263500996e-06, "loss": 1.3369, "step": 2170 }, { "epoch": 0.67, "learning_rate": 5.281638170701378e-06, "loss": 1.2599, "step": 2171 }, { "epoch": 0.67, "learning_rate": 5.2728527651562475e-06, "loss": 1.3512, "step": 2172 }, { "epoch": 0.67, "learning_rate": 5.264072055593094e-06, "loss": 1.5191, "step": 2173 }, { "epoch": 0.67, "learning_rate": 5.255296050734733e-06, "loss": 1.6249, "step": 2174 }, { "epoch": 0.67, "learning_rate": 5.2465247592993095e-06, "loss": 1.3065, "step": 2175 }, { "epoch": 0.67, "learning_rate": 5.237758190000295e-06, "loss": 1.6039, "step": 2176 }, { "epoch": 0.67, "learning_rate": 5.2289963515464525e-06, "loss": 1.4283, "step": 2177 }, { "epoch": 0.67, "learning_rate": 5.220239252641863e-06, "loss": 1.4831, "step": 2178 }, { "epoch": 0.67, "learning_rate": 5.211486901985889e-06, "loss": 1.2216, "step": 2179 }, { "epoch": 0.67, "learning_rate": 5.202739308273172e-06, "loss": 1.4167, "step": 2180 }, { "epoch": 0.67, "learning_rate": 5.193996480193646e-06, "loss": 0.9553, "step": 2181 }, { "epoch": 0.67, "learning_rate": 5.1852584264324864e-06, "loss": 1.4596, "step": 2182 }, { "epoch": 0.67, "learning_rate": 5.176525155670146e-06, "loss": 1.5661, "step": 2183 }, { "epoch": 0.67, "learning_rate": 5.1677966765823226e-06, "loss": 1.2616, "step": 2184 }, { "epoch": 0.67, "learning_rate": 5.15907299783994e-06, "loss": 1.3346, "step": 2185 }, { "epoch": 0.67, "learning_rate": 5.150354128109173e-06, "loss": 1.1332, "step": 2186 }, { "epoch": 0.67, "learning_rate": 5.1416400760514e-06, "loss": 1.2076, "step": 2187 }, { "epoch": 0.67, "learning_rate": 5.132930850323228e-06, "loss": 1.6124, "step": 2188 }, { "epoch": 0.67, "learning_rate": 5.1242264595764664e-06, "loss": 1.6916, "step": 2189 }, { "epoch": 0.67, "learning_rate": 5.115526912458113e-06, "loss": 1.2544, "step": 2190 }, { "epoch": 0.67, "learning_rate": 5.1068322176103646e-06, "loss": 1.1686, "step": 2191 }, { "epoch": 0.67, "learning_rate": 5.0981423836706e-06, "loss": 1.3855, "step": 2192 }, { "epoch": 0.67, "learning_rate": 5.089457419271349e-06, "loss": 1.4128, "step": 2193 }, { "epoch": 0.68, "learning_rate": 5.080777333040323e-06, "loss": 1.6512, "step": 2194 }, { "epoch": 0.68, "learning_rate": 5.072102133600386e-06, "loss": 1.181, "step": 2195 }, { "epoch": 0.68, "learning_rate": 5.063431829569537e-06, "loss": 1.3364, "step": 2196 }, { "epoch": 0.68, "learning_rate": 5.054766429560927e-06, "loss": 1.6372, "step": 2197 }, { "epoch": 0.68, "learning_rate": 5.046105942182815e-06, "loss": 1.4139, "step": 2198 }, { "epoch": 0.68, "learning_rate": 5.037450376038598e-06, "loss": 1.34, "step": 2199 }, { "epoch": 0.68, "learning_rate": 5.02879973972678e-06, "loss": 1.6916, "step": 2200 }, { "epoch": 0.68, "learning_rate": 5.020154041840959e-06, "loss": 1.5294, "step": 2201 }, { "epoch": 0.68, "learning_rate": 5.011513290969839e-06, "loss": 1.4231, "step": 2202 }, { "epoch": 0.68, "learning_rate": 5.002877495697198e-06, "loss": 1.2681, "step": 2203 }, { "epoch": 0.68, "learning_rate": 4.9942466646019e-06, "loss": 1.5514, "step": 2204 }, { "epoch": 0.68, "learning_rate": 4.98562080625788e-06, "loss": 1.7546, "step": 2205 }, { "epoch": 0.68, "learning_rate": 4.97699992923412e-06, "loss": 1.6773, "step": 2206 }, { "epoch": 0.68, "learning_rate": 4.96838404209466e-06, "loss": 1.2192, "step": 2207 }, { "epoch": 0.68, "learning_rate": 4.9597731533985915e-06, "loss": 1.4569, "step": 2208 }, { "epoch": 0.68, "learning_rate": 4.951167271700026e-06, "loss": 1.463, "step": 2209 }, { "epoch": 0.68, "learning_rate": 4.942566405548109e-06, "loss": 1.3425, "step": 2210 }, { "epoch": 0.68, "learning_rate": 4.933970563487011e-06, "loss": 1.2888, "step": 2211 }, { "epoch": 0.68, "learning_rate": 4.925379754055891e-06, "loss": 1.8507, "step": 2212 }, { "epoch": 0.68, "learning_rate": 4.916793985788931e-06, "loss": 1.406, "step": 2213 }, { "epoch": 0.68, "learning_rate": 4.908213267215287e-06, "loss": 1.4564, "step": 2214 }, { "epoch": 0.68, "learning_rate": 4.89963760685911e-06, "loss": 1.425, "step": 2215 }, { "epoch": 0.68, "learning_rate": 4.891067013239526e-06, "loss": 1.5078, "step": 2216 }, { "epoch": 0.68, "learning_rate": 4.882501494870616e-06, "loss": 1.624, "step": 2217 }, { "epoch": 0.68, "learning_rate": 4.873941060261437e-06, "loss": 1.5035, "step": 2218 }, { "epoch": 0.68, "learning_rate": 4.865385717915976e-06, "loss": 1.4894, "step": 2219 }, { "epoch": 0.68, "learning_rate": 4.856835476333181e-06, "loss": 1.4498, "step": 2220 }, { "epoch": 0.68, "learning_rate": 4.848290344006917e-06, "loss": 1.2851, "step": 2221 }, { "epoch": 0.68, "learning_rate": 4.839750329425979e-06, "loss": 1.6096, "step": 2222 }, { "epoch": 0.68, "learning_rate": 4.8312154410740815e-06, "loss": 1.3143, "step": 2223 }, { "epoch": 0.68, "learning_rate": 4.822685687429848e-06, "loss": 1.4599, "step": 2224 }, { "epoch": 0.68, "learning_rate": 4.8141610769667876e-06, "loss": 1.4214, "step": 2225 }, { "epoch": 0.68, "learning_rate": 4.805641618153316e-06, "loss": 1.3106, "step": 2226 }, { "epoch": 0.69, "learning_rate": 4.7971273194527254e-06, "loss": 1.3005, "step": 2227 }, { "epoch": 0.69, "learning_rate": 4.788618189323176e-06, "loss": 1.1305, "step": 2228 }, { "epoch": 0.69, "learning_rate": 4.780114236217706e-06, "loss": 1.1596, "step": 2229 }, { "epoch": 0.69, "learning_rate": 4.771615468584194e-06, "loss": 1.2986, "step": 2230 }, { "epoch": 0.69, "learning_rate": 4.763121894865382e-06, "loss": 1.7865, "step": 2231 }, { "epoch": 0.69, "learning_rate": 4.7546335234988515e-06, "loss": 1.404, "step": 2232 }, { "epoch": 0.69, "learning_rate": 4.746150362917003e-06, "loss": 1.6623, "step": 2233 }, { "epoch": 0.69, "learning_rate": 4.737672421547076e-06, "loss": 1.6717, "step": 2234 }, { "epoch": 0.69, "learning_rate": 4.7291997078111165e-06, "loss": 1.4524, "step": 2235 }, { "epoch": 0.69, "learning_rate": 4.720732230125973e-06, "loss": 1.3729, "step": 2236 }, { "epoch": 0.69, "learning_rate": 4.712269996903307e-06, "loss": 1.4386, "step": 2237 }, { "epoch": 0.69, "learning_rate": 4.703813016549555e-06, "loss": 1.5615, "step": 2238 }, { "epoch": 0.69, "learning_rate": 4.695361297465946e-06, "loss": 1.45, "step": 2239 }, { "epoch": 0.69, "learning_rate": 4.6869148480484795e-06, "loss": 1.3076, "step": 2240 }, { "epoch": 0.69, "learning_rate": 4.6784736766879136e-06, "loss": 1.5083, "step": 2241 }, { "epoch": 0.69, "learning_rate": 4.670037791769772e-06, "loss": 1.2165, "step": 2242 }, { "epoch": 0.69, "learning_rate": 4.661607201674328e-06, "loss": 1.1603, "step": 2243 }, { "epoch": 0.69, "learning_rate": 4.65318191477658e-06, "loss": 1.2037, "step": 2244 }, { "epoch": 0.69, "learning_rate": 4.644761939446277e-06, "loss": 1.6643, "step": 2245 }, { "epoch": 0.69, "learning_rate": 4.636347284047878e-06, "loss": 1.6561, "step": 2246 }, { "epoch": 0.69, "learning_rate": 4.6279379569405626e-06, "loss": 1.4538, "step": 2247 }, { "epoch": 0.69, "learning_rate": 4.619533966478225e-06, "loss": 1.5816, "step": 2248 }, { "epoch": 0.69, "learning_rate": 4.6111353210094346e-06, "loss": 1.3687, "step": 2249 }, { "epoch": 0.69, "learning_rate": 4.602742028877475e-06, "loss": 1.1537, "step": 2250 }, { "epoch": 0.69, "learning_rate": 4.594354098420305e-06, "loss": 1.7346, "step": 2251 }, { "epoch": 0.69, "learning_rate": 4.5859715379705475e-06, "loss": 1.4227, "step": 2252 }, { "epoch": 0.69, "learning_rate": 4.577594355855506e-06, "loss": 2.1561, "step": 2253 }, { "epoch": 0.69, "learning_rate": 4.569222560397126e-06, "loss": 1.4633, "step": 2254 }, { "epoch": 0.69, "learning_rate": 4.560856159912013e-06, "loss": 1.7257, "step": 2255 }, { "epoch": 0.69, "learning_rate": 4.552495162711412e-06, "loss": 1.3568, "step": 2256 }, { "epoch": 0.69, "learning_rate": 4.544139577101191e-06, "loss": 1.2608, "step": 2257 }, { "epoch": 0.69, "learning_rate": 4.535789411381852e-06, "loss": 1.4341, "step": 2258 }, { "epoch": 0.7, "learning_rate": 4.527444673848514e-06, "loss": 1.3042, "step": 2259 }, { "epoch": 0.7, "learning_rate": 4.519105372790892e-06, "loss": 1.2717, "step": 2260 }, { "epoch": 0.7, "learning_rate": 4.510771516493313e-06, "loss": 1.4883, "step": 2261 }, { "epoch": 0.7, "learning_rate": 4.502443113234688e-06, "loss": 2.1037, "step": 2262 }, { "epoch": 0.7, "learning_rate": 4.4941201712885065e-06, "loss": 1.3394, "step": 2263 }, { "epoch": 0.7, "learning_rate": 4.485802698922849e-06, "loss": 1.3954, "step": 2264 }, { "epoch": 0.7, "learning_rate": 4.4774907044003426e-06, "loss": 1.3984, "step": 2265 }, { "epoch": 0.7, "learning_rate": 4.4691841959781865e-06, "loss": 1.3015, "step": 2266 }, { "epoch": 0.7, "learning_rate": 4.46088318190813e-06, "loss": 1.231, "step": 2267 }, { "epoch": 0.7, "learning_rate": 4.452587670436451e-06, "loss": 1.3133, "step": 2268 }, { "epoch": 0.7, "learning_rate": 4.444297669803981e-06, "loss": 1.8337, "step": 2269 }, { "epoch": 0.7, "learning_rate": 4.436013188246056e-06, "loss": 1.4048, "step": 2270 }, { "epoch": 0.7, "learning_rate": 4.427734233992542e-06, "loss": 1.6339, "step": 2271 }, { "epoch": 0.7, "learning_rate": 4.419460815267818e-06, "loss": 1.5258, "step": 2272 }, { "epoch": 0.7, "learning_rate": 4.41119294029075e-06, "loss": 1.5853, "step": 2273 }, { "epoch": 0.7, "learning_rate": 4.402930617274704e-06, "loss": 1.2298, "step": 2274 }, { "epoch": 0.7, "learning_rate": 4.394673854427541e-06, "loss": 1.2447, "step": 2275 }, { "epoch": 0.7, "learning_rate": 4.386422659951575e-06, "loss": 1.4458, "step": 2276 }, { "epoch": 0.7, "learning_rate": 4.378177042043613e-06, "loss": 1.3232, "step": 2277 }, { "epoch": 0.7, "learning_rate": 4.369937008894906e-06, "loss": 1.3119, "step": 2278 }, { "epoch": 0.7, "learning_rate": 4.361702568691159e-06, "loss": 1.342, "step": 2279 }, { "epoch": 0.7, "learning_rate": 4.353473729612531e-06, "loss": 1.4981, "step": 2280 }, { "epoch": 0.7, "learning_rate": 4.3452504998336005e-06, "loss": 1.465, "step": 2281 }, { "epoch": 0.7, "learning_rate": 4.33703288752339e-06, "loss": 1.7319, "step": 2282 }, { "epoch": 0.7, "learning_rate": 4.328820900845336e-06, "loss": 1.513, "step": 2283 }, { "epoch": 0.7, "learning_rate": 4.320614547957278e-06, "loss": 1.512, "step": 2284 }, { "epoch": 0.7, "learning_rate": 4.31241383701147e-06, "loss": 1.4747, "step": 2285 }, { "epoch": 0.7, "learning_rate": 4.30421877615456e-06, "loss": 1.4082, "step": 2286 }, { "epoch": 0.7, "learning_rate": 4.296029373527573e-06, "loss": 1.6727, "step": 2287 }, { "epoch": 0.7, "learning_rate": 4.2878456372659295e-06, "loss": 1.6795, "step": 2288 }, { "epoch": 0.7, "learning_rate": 4.279667575499402e-06, "loss": 1.4107, "step": 2289 }, { "epoch": 0.7, "learning_rate": 4.271495196352141e-06, "loss": 1.4648, "step": 2290 }, { "epoch": 0.7, "learning_rate": 4.263328507942653e-06, "loss": 1.0956, "step": 2291 }, { "epoch": 0.71, "learning_rate": 4.255167518383771e-06, "loss": 1.1659, "step": 2292 }, { "epoch": 0.71, "learning_rate": 4.247012235782686e-06, "loss": 1.44, "step": 2293 }, { "epoch": 0.71, "learning_rate": 4.23886266824092e-06, "loss": 1.2127, "step": 2294 }, { "epoch": 0.71, "learning_rate": 4.230718823854304e-06, "loss": 1.3666, "step": 2295 }, { "epoch": 0.71, "learning_rate": 4.222580710712998e-06, "loss": 1.5255, "step": 2296 }, { "epoch": 0.71, "learning_rate": 4.2144483369014555e-06, "loss": 1.299, "step": 2297 }, { "epoch": 0.71, "learning_rate": 4.2063217104984374e-06, "loss": 1.1362, "step": 2298 }, { "epoch": 0.71, "learning_rate": 4.198200839576999e-06, "loss": 1.1963, "step": 2299 }, { "epoch": 0.71, "learning_rate": 4.190085732204462e-06, "loss": 1.5483, "step": 2300 }, { "epoch": 0.71, "learning_rate": 4.181976396442436e-06, "loss": 1.3075, "step": 2301 }, { "epoch": 0.71, "learning_rate": 4.173872840346798e-06, "loss": 1.069, "step": 2302 }, { "epoch": 0.71, "learning_rate": 4.1657750719676715e-06, "loss": 1.3717, "step": 2303 }, { "epoch": 0.71, "learning_rate": 4.157683099349444e-06, "loss": 1.5581, "step": 2304 }, { "epoch": 0.71, "learning_rate": 4.149596930530736e-06, "loss": 1.0952, "step": 2305 }, { "epoch": 0.71, "learning_rate": 4.1415165735444016e-06, "loss": 1.2826, "step": 2306 }, { "epoch": 0.71, "learning_rate": 4.133442036417534e-06, "loss": 1.3499, "step": 2307 }, { "epoch": 0.71, "learning_rate": 4.125373327171428e-06, "loss": 1.4579, "step": 2308 }, { "epoch": 0.71, "learning_rate": 4.117310453821602e-06, "loss": 1.2177, "step": 2309 }, { "epoch": 0.71, "learning_rate": 4.109253424377773e-06, "loss": 1.8603, "step": 2310 }, { "epoch": 0.71, "learning_rate": 4.101202246843846e-06, "loss": 1.3527, "step": 2311 }, { "epoch": 0.71, "learning_rate": 4.093156929217929e-06, "loss": 1.6952, "step": 2312 }, { "epoch": 0.71, "learning_rate": 4.085117479492285e-06, "loss": 1.3367, "step": 2313 }, { "epoch": 0.71, "learning_rate": 4.07708390565337e-06, "loss": 1.4601, "step": 2314 }, { "epoch": 0.71, "learning_rate": 4.069056215681794e-06, "loss": 1.5856, "step": 2315 }, { "epoch": 0.71, "learning_rate": 4.0610344175523175e-06, "loss": 1.3092, "step": 2316 }, { "epoch": 0.71, "learning_rate": 4.053018519233853e-06, "loss": 1.2615, "step": 2317 }, { "epoch": 0.71, "learning_rate": 4.045008528689457e-06, "loss": 1.7463, "step": 2318 }, { "epoch": 0.71, "learning_rate": 4.037004453876308e-06, "loss": 1.1891, "step": 2319 }, { "epoch": 0.71, "learning_rate": 4.029006302745706e-06, "loss": 1.5644, "step": 2320 }, { "epoch": 0.71, "learning_rate": 4.02101408324308e-06, "loss": 1.5895, "step": 2321 }, { "epoch": 0.71, "learning_rate": 4.01302780330795e-06, "loss": 1.4587, "step": 2322 }, { "epoch": 0.71, "learning_rate": 4.005047470873954e-06, "loss": 1.4265, "step": 2323 }, { "epoch": 0.72, "learning_rate": 3.997073093868802e-06, "loss": 1.6216, "step": 2324 }, { "epoch": 0.72, "learning_rate": 3.9891046802143e-06, "loss": 1.6914, "step": 2325 }, { "epoch": 0.72, "learning_rate": 3.981142237826332e-06, "loss": 1.4617, "step": 2326 }, { "epoch": 0.72, "learning_rate": 3.97318577461484e-06, "loss": 1.6414, "step": 2327 }, { "epoch": 0.72, "learning_rate": 3.965235298483835e-06, "loss": 1.636, "step": 2328 }, { "epoch": 0.72, "learning_rate": 3.957290817331373e-06, "loss": 1.6273, "step": 2329 }, { "epoch": 0.72, "learning_rate": 3.949352339049561e-06, "loss": 1.2289, "step": 2330 }, { "epoch": 0.72, "learning_rate": 3.941419871524542e-06, "loss": 1.7793, "step": 2331 }, { "epoch": 0.72, "learning_rate": 3.933493422636483e-06, "loss": 1.3266, "step": 2332 }, { "epoch": 0.72, "learning_rate": 3.92557300025958e-06, "loss": 1.2637, "step": 2333 }, { "epoch": 0.72, "learning_rate": 3.917658612262033e-06, "loss": 1.3153, "step": 2334 }, { "epoch": 0.72, "learning_rate": 3.9097502665060506e-06, "loss": 1.3495, "step": 2335 }, { "epoch": 0.72, "learning_rate": 3.901847970847842e-06, "loss": 2.0786, "step": 2336 }, { "epoch": 0.72, "learning_rate": 3.893951733137608e-06, "loss": 1.1121, "step": 2337 }, { "epoch": 0.72, "learning_rate": 3.886061561219521e-06, "loss": 1.5521, "step": 2338 }, { "epoch": 0.72, "learning_rate": 3.8781774629317415e-06, "loss": 1.3826, "step": 2339 }, { "epoch": 0.72, "learning_rate": 3.870299446106382e-06, "loss": 1.4512, "step": 2340 }, { "epoch": 0.72, "learning_rate": 3.862427518569526e-06, "loss": 1.2351, "step": 2341 }, { "epoch": 0.72, "learning_rate": 3.854561688141205e-06, "loss": 1.6359, "step": 2342 }, { "epoch": 0.72, "learning_rate": 3.846701962635383e-06, "loss": 1.4687, "step": 2343 }, { "epoch": 0.72, "learning_rate": 3.838848349859977e-06, "loss": 1.4153, "step": 2344 }, { "epoch": 0.72, "learning_rate": 3.831000857616815e-06, "loss": 1.3847, "step": 2345 }, { "epoch": 0.72, "learning_rate": 3.823159493701654e-06, "loss": 1.3268, "step": 2346 }, { "epoch": 0.72, "learning_rate": 3.815324265904169e-06, "loss": 1.3804, "step": 2347 }, { "epoch": 0.72, "learning_rate": 3.8074951820079186e-06, "loss": 1.5267, "step": 2348 }, { "epoch": 0.72, "learning_rate": 3.799672249790376e-06, "loss": 1.1626, "step": 2349 }, { "epoch": 0.72, "learning_rate": 3.791855477022903e-06, "loss": 1.5135, "step": 2350 }, { "epoch": 0.72, "learning_rate": 3.78404487147073e-06, "loss": 1.8566, "step": 2351 }, { "epoch": 0.72, "learning_rate": 3.776240440892973e-06, "loss": 1.637, "step": 2352 }, { "epoch": 0.72, "learning_rate": 3.768442193042612e-06, "loss": 1.6293, "step": 2353 }, { "epoch": 0.72, "learning_rate": 3.760650135666476e-06, "loss": 1.2378, "step": 2354 }, { "epoch": 0.72, "learning_rate": 3.752864276505257e-06, "loss": 1.2596, "step": 2355 }, { "epoch": 0.72, "learning_rate": 3.7450846232934768e-06, "loss": 1.5485, "step": 2356 }, { "epoch": 0.73, "learning_rate": 3.737311183759502e-06, "loss": 1.3137, "step": 2357 }, { "epoch": 0.73, "learning_rate": 3.729543965625526e-06, "loss": 1.3654, "step": 2358 }, { "epoch": 0.73, "learning_rate": 3.721782976607554e-06, "loss": 1.6538, "step": 2359 }, { "epoch": 0.73, "learning_rate": 3.7140282244154136e-06, "loss": 1.4991, "step": 2360 }, { "epoch": 0.73, "learning_rate": 3.7062797167527274e-06, "loss": 1.3429, "step": 2361 }, { "epoch": 0.73, "learning_rate": 3.6985374613169166e-06, "loss": 1.6156, "step": 2362 }, { "epoch": 0.73, "learning_rate": 3.6908014657991985e-06, "loss": 1.3656, "step": 2363 }, { "epoch": 0.73, "learning_rate": 3.6830717378845614e-06, "loss": 1.4568, "step": 2364 }, { "epoch": 0.73, "learning_rate": 3.675348285251774e-06, "loss": 1.5126, "step": 2365 }, { "epoch": 0.73, "learning_rate": 3.6676311155733745e-06, "loss": 1.6582, "step": 2366 }, { "epoch": 0.73, "learning_rate": 3.6599202365156483e-06, "loss": 1.1857, "step": 2367 }, { "epoch": 0.73, "learning_rate": 3.652215655738639e-06, "loss": 1.8582, "step": 2368 }, { "epoch": 0.73, "learning_rate": 3.6445173808961375e-06, "loss": 1.4223, "step": 2369 }, { "epoch": 0.73, "learning_rate": 3.6368254196356576e-06, "loss": 1.6014, "step": 2370 }, { "epoch": 0.73, "learning_rate": 3.6291397795984564e-06, "loss": 1.1939, "step": 2371 }, { "epoch": 0.73, "learning_rate": 3.6214604684194974e-06, "loss": 1.6545, "step": 2372 }, { "epoch": 0.73, "learning_rate": 3.6137874937274652e-06, "loss": 1.5132, "step": 2373 }, { "epoch": 0.73, "learning_rate": 3.606120863144753e-06, "loss": 1.2304, "step": 2374 }, { "epoch": 0.73, "learning_rate": 3.598460584287442e-06, "loss": 1.5432, "step": 2375 }, { "epoch": 0.73, "learning_rate": 3.590806664765305e-06, "loss": 1.6947, "step": 2376 }, { "epoch": 0.73, "learning_rate": 3.5831591121818086e-06, "loss": 2.0035, "step": 2377 }, { "epoch": 0.73, "learning_rate": 3.5755179341340772e-06, "loss": 0.7274, "step": 2378 }, { "epoch": 0.73, "learning_rate": 3.567883138212922e-06, "loss": 2.0919, "step": 2379 }, { "epoch": 0.73, "learning_rate": 3.5602547320027937e-06, "loss": 1.4689, "step": 2380 }, { "epoch": 0.73, "learning_rate": 3.5526327230818114e-06, "loss": 1.3267, "step": 2381 }, { "epoch": 0.73, "learning_rate": 3.5450171190217364e-06, "loss": 1.2935, "step": 2382 }, { "epoch": 0.73, "learning_rate": 3.537407927387957e-06, "loss": 1.2245, "step": 2383 }, { "epoch": 0.73, "learning_rate": 3.529805155739502e-06, "loss": 1.0391, "step": 2384 }, { "epoch": 0.73, "learning_rate": 3.5222088116290243e-06, "loss": 1.2109, "step": 2385 }, { "epoch": 0.73, "learning_rate": 3.5146189026027787e-06, "loss": 1.2736, "step": 2386 }, { "epoch": 0.73, "learning_rate": 3.507035436200642e-06, "loss": 1.3463, "step": 2387 }, { "epoch": 0.73, "learning_rate": 3.4994584199560767e-06, "loss": 1.3743, "step": 2388 }, { "epoch": 0.74, "learning_rate": 3.491887861396154e-06, "loss": 1.3068, "step": 2389 }, { "epoch": 0.74, "learning_rate": 3.4843237680415153e-06, "loss": 1.4922, "step": 2390 }, { "epoch": 0.74, "learning_rate": 3.476766147406384e-06, "loss": 1.2816, "step": 2391 }, { "epoch": 0.74, "learning_rate": 3.4692150069985563e-06, "loss": 1.4652, "step": 2392 }, { "epoch": 0.74, "learning_rate": 3.461670354319393e-06, "loss": 1.5029, "step": 2393 }, { "epoch": 0.74, "learning_rate": 3.4541321968637995e-06, "loss": 1.3108, "step": 2394 }, { "epoch": 0.74, "learning_rate": 3.446600542120243e-06, "loss": 1.208, "step": 2395 }, { "epoch": 0.74, "learning_rate": 3.439075397570716e-06, "loss": 1.3612, "step": 2396 }, { "epoch": 0.74, "learning_rate": 3.431556770690755e-06, "loss": 1.4533, "step": 2397 }, { "epoch": 0.74, "learning_rate": 3.42404466894942e-06, "loss": 1.3458, "step": 2398 }, { "epoch": 0.74, "learning_rate": 3.4165390998092806e-06, "loss": 1.3477, "step": 2399 }, { "epoch": 0.74, "learning_rate": 3.409040070726427e-06, "loss": 1.3303, "step": 2400 }, { "epoch": 0.74, "learning_rate": 3.4015475891504514e-06, "loss": 1.4927, "step": 2401 }, { "epoch": 0.74, "learning_rate": 3.3940616625244314e-06, "loss": 1.3388, "step": 2402 }, { "epoch": 0.74, "learning_rate": 3.3865822982849485e-06, "loss": 1.6998, "step": 2403 }, { "epoch": 0.74, "learning_rate": 3.3791095038620515e-06, "loss": 1.3615, "step": 2404 }, { "epoch": 0.74, "learning_rate": 3.371643286679266e-06, "loss": 1.1296, "step": 2405 }, { "epoch": 0.74, "learning_rate": 3.364183654153592e-06, "loss": 1.5067, "step": 2406 }, { "epoch": 0.74, "learning_rate": 3.356730613695476e-06, "loss": 1.4794, "step": 2407 }, { "epoch": 0.74, "learning_rate": 3.349284172708823e-06, "loss": 1.5227, "step": 2408 }, { "epoch": 0.74, "learning_rate": 3.341844338590987e-06, "loss": 1.4886, "step": 2409 }, { "epoch": 0.74, "learning_rate": 3.334411118732744e-06, "loss": 1.3697, "step": 2410 }, { "epoch": 0.74, "learning_rate": 3.326984520518313e-06, "loss": 1.4566, "step": 2411 }, { "epoch": 0.74, "learning_rate": 3.3195645513253316e-06, "loss": 1.8796, "step": 2412 }, { "epoch": 0.74, "learning_rate": 3.3121512185248437e-06, "loss": 1.4035, "step": 2413 }, { "epoch": 0.74, "learning_rate": 3.304744529481315e-06, "loss": 1.3376, "step": 2414 }, { "epoch": 0.74, "learning_rate": 3.2973444915525966e-06, "loss": 1.3358, "step": 2415 }, { "epoch": 0.74, "learning_rate": 3.289951112089943e-06, "loss": 0.9312, "step": 2416 }, { "epoch": 0.74, "learning_rate": 3.282564398437995e-06, "loss": 1.2696, "step": 2417 }, { "epoch": 0.74, "learning_rate": 3.2751843579347562e-06, "loss": 1.2902, "step": 2418 }, { "epoch": 0.74, "learning_rate": 3.267810997911617e-06, "loss": 1.448, "step": 2419 }, { "epoch": 0.74, "learning_rate": 3.2604443256933307e-06, "loss": 1.4241, "step": 2420 }, { "epoch": 0.74, "learning_rate": 3.2530843485979936e-06, "loss": 1.2873, "step": 2421 }, { "epoch": 0.75, "learning_rate": 3.2457310739370684e-06, "loss": 1.1503, "step": 2422 }, { "epoch": 0.75, "learning_rate": 3.2383845090153434e-06, "loss": 1.058, "step": 2423 }, { "epoch": 0.75, "learning_rate": 3.231044661130951e-06, "loss": 1.54, "step": 2424 }, { "epoch": 0.75, "learning_rate": 3.2237115375753547e-06, "loss": 1.2798, "step": 2425 }, { "epoch": 0.75, "learning_rate": 3.216385145633324e-06, "loss": 1.5906, "step": 2426 }, { "epoch": 0.75, "learning_rate": 3.209065492582951e-06, "loss": 1.3133, "step": 2427 }, { "epoch": 0.75, "learning_rate": 3.201752585695638e-06, "loss": 1.5513, "step": 2428 }, { "epoch": 0.75, "learning_rate": 3.1944464322360704e-06, "loss": 1.413, "step": 2429 }, { "epoch": 0.75, "learning_rate": 3.1871470394622407e-06, "loss": 1.3208, "step": 2430 }, { "epoch": 0.75, "learning_rate": 3.179854414625415e-06, "loss": 1.4812, "step": 2431 }, { "epoch": 0.75, "learning_rate": 3.1725685649701354e-06, "loss": 1.4869, "step": 2432 }, { "epoch": 0.75, "learning_rate": 3.165289497734223e-06, "loss": 1.2375, "step": 2433 }, { "epoch": 0.75, "learning_rate": 3.158017220148751e-06, "loss": 1.794, "step": 2434 }, { "epoch": 0.75, "learning_rate": 3.1507517394380536e-06, "loss": 1.7377, "step": 2435 }, { "epoch": 0.75, "learning_rate": 3.1434930628197145e-06, "loss": 1.3471, "step": 2436 }, { "epoch": 0.75, "learning_rate": 3.1362411975045504e-06, "loss": 1.3891, "step": 2437 }, { "epoch": 0.75, "learning_rate": 3.1289961506966217e-06, "loss": 1.2586, "step": 2438 }, { "epoch": 0.75, "learning_rate": 3.121757929593203e-06, "loss": 1.2273, "step": 2439 }, { "epoch": 0.75, "learning_rate": 3.114526541384799e-06, "loss": 1.3378, "step": 2440 }, { "epoch": 0.75, "learning_rate": 3.107301993255126e-06, "loss": 1.8218, "step": 2441 }, { "epoch": 0.75, "learning_rate": 3.100084292381095e-06, "loss": 1.6571, "step": 2442 }, { "epoch": 0.75, "learning_rate": 3.092873445932825e-06, "loss": 1.4724, "step": 2443 }, { "epoch": 0.75, "learning_rate": 3.085669461073626e-06, "loss": 1.0988, "step": 2444 }, { "epoch": 0.75, "learning_rate": 3.078472344959985e-06, "loss": 1.5087, "step": 2445 }, { "epoch": 0.75, "learning_rate": 3.0712821047415655e-06, "loss": 1.5175, "step": 2446 }, { "epoch": 0.75, "learning_rate": 3.0640987475612104e-06, "loss": 1.317, "step": 2447 }, { "epoch": 0.75, "learning_rate": 3.0569222805549114e-06, "loss": 1.6835, "step": 2448 }, { "epoch": 0.75, "learning_rate": 3.0497527108518286e-06, "loss": 1.3147, "step": 2449 }, { "epoch": 0.75, "learning_rate": 3.0425900455742584e-06, "loss": 1.5031, "step": 2450 }, { "epoch": 0.75, "learning_rate": 3.0354342918376467e-06, "loss": 1.4818, "step": 2451 }, { "epoch": 0.75, "learning_rate": 3.0282854567505737e-06, "loss": 1.6095, "step": 2452 }, { "epoch": 0.75, "learning_rate": 3.0211435474147364e-06, "loss": 1.3944, "step": 2453 }, { "epoch": 0.76, "learning_rate": 3.0140085709249666e-06, "loss": 1.2271, "step": 2454 }, { "epoch": 0.76, "learning_rate": 3.0068805343691944e-06, "loss": 1.3591, "step": 2455 }, { "epoch": 0.76, "learning_rate": 2.9997594448284638e-06, "loss": 1.3668, "step": 2456 }, { "epoch": 0.76, "learning_rate": 2.9926453093769215e-06, "loss": 1.3199, "step": 2457 }, { "epoch": 0.76, "learning_rate": 2.9855381350817937e-06, "loss": 1.335, "step": 2458 }, { "epoch": 0.76, "learning_rate": 2.9784379290034038e-06, "loss": 1.2559, "step": 2459 }, { "epoch": 0.76, "learning_rate": 2.9713446981951466e-06, "loss": 1.3688, "step": 2460 }, { "epoch": 0.76, "learning_rate": 2.964258449703482e-06, "loss": 1.7332, "step": 2461 }, { "epoch": 0.76, "learning_rate": 2.957179190567947e-06, "loss": 1.5644, "step": 2462 }, { "epoch": 0.76, "learning_rate": 2.950106927821128e-06, "loss": 1.426, "step": 2463 }, { "epoch": 0.76, "learning_rate": 2.9430416684886575e-06, "loss": 1.6684, "step": 2464 }, { "epoch": 0.76, "learning_rate": 2.93598341958922e-06, "loss": 1.4267, "step": 2465 }, { "epoch": 0.76, "learning_rate": 2.9289321881345257e-06, "loss": 1.4776, "step": 2466 }, { "epoch": 0.76, "learning_rate": 2.9218879811293198e-06, "loss": 1.4945, "step": 2467 }, { "epoch": 0.76, "learning_rate": 2.9148508055713733e-06, "loss": 1.3943, "step": 2468 }, { "epoch": 0.76, "learning_rate": 2.9078206684514575e-06, "loss": 1.3681, "step": 2469 }, { "epoch": 0.76, "learning_rate": 2.9007975767533714e-06, "loss": 1.5331, "step": 2470 }, { "epoch": 0.76, "learning_rate": 2.893781537453895e-06, "loss": 1.3095, "step": 2471 }, { "epoch": 0.76, "learning_rate": 2.8867725575228166e-06, "loss": 1.212, "step": 2472 }, { "epoch": 0.76, "learning_rate": 2.87977064392291e-06, "loss": 1.2286, "step": 2473 }, { "epoch": 0.76, "learning_rate": 2.872775803609924e-06, "loss": 1.3955, "step": 2474 }, { "epoch": 0.76, "learning_rate": 2.8657880435325793e-06, "loss": 1.7652, "step": 2475 }, { "epoch": 0.76, "learning_rate": 2.8588073706325726e-06, "loss": 1.3642, "step": 2476 }, { "epoch": 0.76, "learning_rate": 2.8518337918445484e-06, "loss": 1.3104, "step": 2477 }, { "epoch": 0.76, "learning_rate": 2.844867314096115e-06, "loss": 1.2454, "step": 2478 }, { "epoch": 0.76, "learning_rate": 2.8379079443078215e-06, "loss": 1.3106, "step": 2479 }, { "epoch": 0.76, "learning_rate": 2.8309556893931522e-06, "loss": 1.51, "step": 2480 }, { "epoch": 0.76, "learning_rate": 2.8240105562585307e-06, "loss": 1.3334, "step": 2481 }, { "epoch": 0.76, "learning_rate": 2.8170725518032983e-06, "loss": 1.0288, "step": 2482 }, { "epoch": 0.76, "learning_rate": 2.8101416829197204e-06, "loss": 1.6592, "step": 2483 }, { "epoch": 0.76, "learning_rate": 2.803217956492974e-06, "loss": 1.5159, "step": 2484 }, { "epoch": 0.76, "learning_rate": 2.796301379401134e-06, "loss": 1.8982, "step": 2485 }, { "epoch": 0.76, "learning_rate": 2.789391958515183e-06, "loss": 1.3649, "step": 2486 }, { "epoch": 0.77, "learning_rate": 2.782489700698984e-06, "loss": 1.392, "step": 2487 }, { "epoch": 0.77, "learning_rate": 2.7755946128092935e-06, "loss": 1.5098, "step": 2488 }, { "epoch": 0.77, "learning_rate": 2.7687067016957413e-06, "loss": 1.6033, "step": 2489 }, { "epoch": 0.77, "learning_rate": 2.7618259742008226e-06, "loss": 1.5718, "step": 2490 }, { "epoch": 0.77, "learning_rate": 2.7549524371599033e-06, "loss": 1.5207, "step": 2491 }, { "epoch": 0.77, "learning_rate": 2.74808609740121e-06, "loss": 1.5259, "step": 2492 }, { "epoch": 0.77, "learning_rate": 2.7412269617458074e-06, "loss": 1.2681, "step": 2493 }, { "epoch": 0.77, "learning_rate": 2.7343750370076127e-06, "loss": 1.6076, "step": 2494 }, { "epoch": 0.77, "learning_rate": 2.7275303299933786e-06, "loss": 1.6076, "step": 2495 }, { "epoch": 0.77, "learning_rate": 2.7206928475026828e-06, "loss": 1.2696, "step": 2496 }, { "epoch": 0.77, "learning_rate": 2.713862596327933e-06, "loss": 1.4948, "step": 2497 }, { "epoch": 0.77, "learning_rate": 2.707039583254346e-06, "loss": 1.5384, "step": 2498 }, { "epoch": 0.77, "learning_rate": 2.7002238150599527e-06, "loss": 1.3518, "step": 2499 }, { "epoch": 0.77, "learning_rate": 2.6934152985155904e-06, "loss": 1.5482, "step": 2500 }, { "epoch": 0.77, "learning_rate": 2.686614040384883e-06, "loss": 1.3428, "step": 2501 }, { "epoch": 0.77, "learning_rate": 2.679820047424253e-06, "loss": 1.3813, "step": 2502 }, { "epoch": 0.77, "learning_rate": 2.6730333263829e-06, "loss": 1.2026, "step": 2503 }, { "epoch": 0.77, "learning_rate": 2.6662538840027984e-06, "loss": 1.2863, "step": 2504 }, { "epoch": 0.77, "learning_rate": 2.6594817270187012e-06, "loss": 1.5945, "step": 2505 }, { "epoch": 0.77, "learning_rate": 2.6527168621581114e-06, "loss": 1.2914, "step": 2506 }, { "epoch": 0.77, "learning_rate": 2.6459592961412973e-06, "loss": 1.4734, "step": 2507 }, { "epoch": 0.77, "learning_rate": 2.639209035681276e-06, "loss": 1.5494, "step": 2508 }, { "epoch": 0.77, "learning_rate": 2.6324660874837982e-06, "loss": 1.4894, "step": 2509 }, { "epoch": 0.77, "learning_rate": 2.625730458247362e-06, "loss": 1.8397, "step": 2510 }, { "epoch": 0.77, "learning_rate": 2.6190021546631915e-06, "loss": 1.1767, "step": 2511 }, { "epoch": 0.77, "learning_rate": 2.6122811834152253e-06, "loss": 1.5503, "step": 2512 }, { "epoch": 0.77, "learning_rate": 2.6055675511801315e-06, "loss": 1.5331, "step": 2513 }, { "epoch": 0.77, "learning_rate": 2.5988612646272735e-06, "loss": 1.8818, "step": 2514 }, { "epoch": 0.77, "learning_rate": 2.592162330418728e-06, "loss": 1.351, "step": 2515 }, { "epoch": 0.77, "learning_rate": 2.5854707552092705e-06, "loss": 1.4186, "step": 2516 }, { "epoch": 0.77, "learning_rate": 2.578786545646348e-06, "loss": 1.3413, "step": 2517 }, { "epoch": 0.77, "learning_rate": 2.5721097083701085e-06, "loss": 1.9233, "step": 2518 }, { "epoch": 0.78, "learning_rate": 2.5654402500133714e-06, "loss": 1.3252, "step": 2519 }, { "epoch": 0.78, "learning_rate": 2.5587781772016196e-06, "loss": 1.5766, "step": 2520 }, { "epoch": 0.78, "learning_rate": 2.5521234965530094e-06, "loss": 1.5627, "step": 2521 }, { "epoch": 0.78, "learning_rate": 2.545476214678344e-06, "loss": 1.5373, "step": 2522 }, { "epoch": 0.78, "learning_rate": 2.538836338181082e-06, "loss": 1.2069, "step": 2523 }, { "epoch": 0.78, "learning_rate": 2.5322038736573273e-06, "loss": 1.4028, "step": 2524 }, { "epoch": 0.78, "learning_rate": 2.525578827695814e-06, "loss": 1.751, "step": 2525 }, { "epoch": 0.78, "learning_rate": 2.518961206877911e-06, "loss": 1.3764, "step": 2526 }, { "epoch": 0.78, "learning_rate": 2.5123510177776143e-06, "loss": 1.4346, "step": 2527 }, { "epoch": 0.78, "learning_rate": 2.5057482669615263e-06, "loss": 1.3396, "step": 2528 }, { "epoch": 0.78, "learning_rate": 2.4991529609888743e-06, "loss": 1.3525, "step": 2529 }, { "epoch": 0.78, "learning_rate": 2.4925651064114788e-06, "loss": 1.4713, "step": 2530 }, { "epoch": 0.78, "learning_rate": 2.485984709773759e-06, "loss": 1.2568, "step": 2531 }, { "epoch": 0.78, "learning_rate": 2.4794117776127337e-06, "loss": 1.8404, "step": 2532 }, { "epoch": 0.78, "learning_rate": 2.472846316457995e-06, "loss": 1.625, "step": 2533 }, { "epoch": 0.78, "learning_rate": 2.466288332831722e-06, "loss": 1.4472, "step": 2534 }, { "epoch": 0.78, "learning_rate": 2.4597378332486675e-06, "loss": 1.9582, "step": 2535 }, { "epoch": 0.78, "learning_rate": 2.4531948242161365e-06, "loss": 1.4819, "step": 2536 }, { "epoch": 0.78, "learning_rate": 2.4466593122340043e-06, "loss": 1.5532, "step": 2537 }, { "epoch": 0.78, "learning_rate": 2.4401313037946993e-06, "loss": 1.5089, "step": 2538 }, { "epoch": 0.78, "learning_rate": 2.4336108053831863e-06, "loss": 1.3625, "step": 2539 }, { "epoch": 0.78, "learning_rate": 2.4270978234769794e-06, "loss": 1.4871, "step": 2540 }, { "epoch": 0.78, "learning_rate": 2.4205923645461173e-06, "loss": 1.3577, "step": 2541 }, { "epoch": 0.78, "learning_rate": 2.4140944350531714e-06, "loss": 1.4938, "step": 2542 }, { "epoch": 0.78, "learning_rate": 2.4076040414532352e-06, "loss": 1.352, "step": 2543 }, { "epoch": 0.78, "learning_rate": 2.401121190193909e-06, "loss": 1.3946, "step": 2544 }, { "epoch": 0.78, "learning_rate": 2.394645887715301e-06, "loss": 1.0848, "step": 2545 }, { "epoch": 0.78, "learning_rate": 2.3881781404500284e-06, "loss": 1.4984, "step": 2546 }, { "epoch": 0.78, "learning_rate": 2.381717954823193e-06, "loss": 1.2543, "step": 2547 }, { "epoch": 0.78, "learning_rate": 2.3752653372523937e-06, "loss": 1.3804, "step": 2548 }, { "epoch": 0.78, "learning_rate": 2.3688202941477033e-06, "loss": 1.6574, "step": 2549 }, { "epoch": 0.78, "learning_rate": 2.362382831911675e-06, "loss": 1.0916, "step": 2550 }, { "epoch": 0.78, "learning_rate": 2.3559529569393337e-06, "loss": 1.474, "step": 2551 }, { "epoch": 0.79, "learning_rate": 2.349530675618158e-06, "loss": 1.5092, "step": 2552 }, { "epoch": 0.79, "learning_rate": 2.34311599432809e-06, "loss": 1.3123, "step": 2553 }, { "epoch": 0.79, "learning_rate": 2.3367089194415226e-06, "loss": 1.6283, "step": 2554 }, { "epoch": 0.79, "learning_rate": 2.3303094573232865e-06, "loss": 1.4334, "step": 2555 }, { "epoch": 0.79, "learning_rate": 2.3239176143306576e-06, "loss": 1.0077, "step": 2556 }, { "epoch": 0.79, "learning_rate": 2.317533396813332e-06, "loss": 1.1655, "step": 2557 }, { "epoch": 0.79, "learning_rate": 2.311156811113444e-06, "loss": 1.6687, "step": 2558 }, { "epoch": 0.79, "learning_rate": 2.3047878635655373e-06, "loss": 1.0468, "step": 2559 }, { "epoch": 0.79, "learning_rate": 2.298426560496566e-06, "loss": 1.4683, "step": 2560 }, { "epoch": 0.79, "learning_rate": 2.2920729082258974e-06, "loss": 1.8363, "step": 2561 }, { "epoch": 0.79, "learning_rate": 2.2857269130652983e-06, "loss": 1.3517, "step": 2562 }, { "epoch": 0.79, "learning_rate": 2.27938858131892e-06, "loss": 1.3948, "step": 2563 }, { "epoch": 0.79, "learning_rate": 2.2730579192833124e-06, "loss": 1.5454, "step": 2564 }, { "epoch": 0.79, "learning_rate": 2.2667349332473944e-06, "loss": 1.2572, "step": 2565 }, { "epoch": 0.79, "learning_rate": 2.2604196294924696e-06, "loss": 1.439, "step": 2566 }, { "epoch": 0.79, "learning_rate": 2.254112014292209e-06, "loss": 1.0789, "step": 2567 }, { "epoch": 0.79, "learning_rate": 2.254112014292209e-06, "loss": 1.4932, "step": 2568 }, { "epoch": 0.79, "learning_rate": 2.2478120939126357e-06, "loss": 1.4678, "step": 2569 }, { "epoch": 0.79, "learning_rate": 2.2415198746121392e-06, "loss": 1.4271, "step": 2570 }, { "epoch": 0.79, "learning_rate": 2.235235362641458e-06, "loss": 1.2138, "step": 2571 }, { "epoch": 0.79, "learning_rate": 2.2289585642436675e-06, "loss": 1.3228, "step": 2572 }, { "epoch": 0.79, "learning_rate": 2.222689485654187e-06, "loss": 1.2213, "step": 2573 }, { "epoch": 0.79, "learning_rate": 2.216428133100763e-06, "loss": 1.4364, "step": 2574 }, { "epoch": 0.79, "learning_rate": 2.2101745128034657e-06, "loss": 1.5871, "step": 2575 }, { "epoch": 0.79, "learning_rate": 2.203928630974691e-06, "loss": 1.494, "step": 2576 }, { "epoch": 0.79, "learning_rate": 2.1976904938191366e-06, "loss": 1.295, "step": 2577 }, { "epoch": 0.79, "learning_rate": 2.191460107533817e-06, "loss": 1.6491, "step": 2578 }, { "epoch": 0.79, "learning_rate": 2.1852374783080443e-06, "loss": 1.5261, "step": 2579 }, { "epoch": 0.79, "learning_rate": 2.179022612323418e-06, "loss": 1.4721, "step": 2580 }, { "epoch": 0.79, "learning_rate": 2.1728155157538367e-06, "loss": 1.2618, "step": 2581 }, { "epoch": 0.79, "learning_rate": 2.1666161947654673e-06, "loss": 1.5919, "step": 2582 }, { "epoch": 0.79, "learning_rate": 2.160424655516764e-06, "loss": 1.4995, "step": 2583 }, { "epoch": 0.8, "learning_rate": 2.1542409041584487e-06, "loss": 1.4501, "step": 2584 }, { "epoch": 0.8, "learning_rate": 2.148064946833498e-06, "loss": 1.3278, "step": 2585 }, { "epoch": 0.8, "learning_rate": 2.141896789677157e-06, "loss": 1.585, "step": 2586 }, { "epoch": 0.8, "learning_rate": 2.135736438816921e-06, "loss": 1.5089, "step": 2587 }, { "epoch": 0.8, "learning_rate": 2.129583900372516e-06, "loss": 1.514, "step": 2588 }, { "epoch": 0.8, "learning_rate": 2.123439180455924e-06, "loss": 1.3724, "step": 2589 }, { "epoch": 0.8, "learning_rate": 2.1173022851713554e-06, "loss": 1.3589, "step": 2590 }, { "epoch": 0.8, "learning_rate": 2.1111732206152424e-06, "loss": 1.7269, "step": 2591 }, { "epoch": 0.8, "learning_rate": 2.105051992876245e-06, "loss": 1.3547, "step": 2592 }, { "epoch": 0.8, "learning_rate": 2.0989386080352304e-06, "loss": 1.4995, "step": 2593 }, { "epoch": 0.8, "learning_rate": 2.092833072165282e-06, "loss": 1.309, "step": 2594 }, { "epoch": 0.8, "learning_rate": 2.086735391331686e-06, "loss": 1.0108, "step": 2595 }, { "epoch": 0.8, "learning_rate": 2.080645571591915e-06, "loss": 0.9986, "step": 2596 }, { "epoch": 0.8, "learning_rate": 2.0745636189956465e-06, "loss": 1.3846, "step": 2597 }, { "epoch": 0.8, "learning_rate": 2.0684895395847306e-06, "loss": 1.2547, "step": 2598 }, { "epoch": 0.8, "learning_rate": 2.0624233393932024e-06, "loss": 1.3957, "step": 2599 }, { "epoch": 0.8, "learning_rate": 2.0563650244472733e-06, "loss": 1.2948, "step": 2600 }, { "epoch": 0.8, "learning_rate": 2.050314600765313e-06, "loss": 1.2872, "step": 2601 }, { "epoch": 0.8, "learning_rate": 2.044272074357854e-06, "loss": 1.5883, "step": 2602 }, { "epoch": 0.8, "learning_rate": 2.038237451227589e-06, "loss": 1.4656, "step": 2603 }, { "epoch": 0.8, "learning_rate": 2.0322107373693535e-06, "loss": 1.6607, "step": 2604 }, { "epoch": 0.8, "learning_rate": 2.026191938770128e-06, "loss": 1.2261, "step": 2605 }, { "epoch": 0.8, "learning_rate": 2.0201810614090343e-06, "loss": 1.367, "step": 2606 }, { "epoch": 0.8, "learning_rate": 2.0141781112573155e-06, "loss": 1.2662, "step": 2607 }, { "epoch": 0.8, "learning_rate": 2.008183094278352e-06, "loss": 1.9735, "step": 2608 }, { "epoch": 0.8, "learning_rate": 2.002196016427629e-06, "loss": 1.4474, "step": 2609 }, { "epoch": 0.8, "learning_rate": 1.996216883652756e-06, "loss": 1.3568, "step": 2610 }, { "epoch": 0.8, "learning_rate": 1.9902457018934496e-06, "loss": 1.4749, "step": 2611 }, { "epoch": 0.8, "learning_rate": 1.9842824770815194e-06, "loss": 1.5196, "step": 2612 }, { "epoch": 0.8, "learning_rate": 1.9783272151408805e-06, "loss": 1.4516, "step": 2613 }, { "epoch": 0.8, "learning_rate": 1.972379921987527e-06, "loss": 1.1927, "step": 2614 }, { "epoch": 0.8, "learning_rate": 1.9664406035295493e-06, "loss": 1.1425, "step": 2615 }, { "epoch": 0.8, "learning_rate": 1.9605092656671055e-06, "loss": 2.1226, "step": 2616 }, { "epoch": 0.81, "learning_rate": 1.954585914292426e-06, "loss": 1.4244, "step": 2617 }, { "epoch": 0.81, "learning_rate": 1.9486705552898145e-06, "loss": 1.3919, "step": 2618 }, { "epoch": 0.81, "learning_rate": 1.942763194535633e-06, "loss": 1.3921, "step": 2619 }, { "epoch": 0.81, "learning_rate": 1.9368638378982908e-06, "loss": 1.4269, "step": 2620 }, { "epoch": 0.81, "learning_rate": 1.930972491238253e-06, "loss": 1.3205, "step": 2621 }, { "epoch": 0.81, "learning_rate": 1.9250891604080302e-06, "loss": 1.7154, "step": 2622 }, { "epoch": 0.81, "learning_rate": 1.9192138512521585e-06, "loss": 1.4195, "step": 2623 }, { "epoch": 0.81, "learning_rate": 1.9133465696072174e-06, "loss": 1.091, "step": 2624 }, { "epoch": 0.81, "learning_rate": 1.907487321301802e-06, "loss": 1.1435, "step": 2625 }, { "epoch": 0.81, "learning_rate": 1.901636112156534e-06, "loss": 1.3941, "step": 2626 }, { "epoch": 0.81, "learning_rate": 1.8957929479840497e-06, "loss": 1.2989, "step": 2627 }, { "epoch": 0.81, "learning_rate": 1.8899578345889846e-06, "loss": 1.3342, "step": 2628 }, { "epoch": 0.81, "learning_rate": 1.8841307777679874e-06, "loss": 1.5118, "step": 2629 }, { "epoch": 0.81, "learning_rate": 1.8783117833096953e-06, "loss": 1.4909, "step": 2630 }, { "epoch": 0.81, "learning_rate": 1.8725008569947366e-06, "loss": 1.6813, "step": 2631 }, { "epoch": 0.81, "learning_rate": 1.866698004595733e-06, "loss": 1.1792, "step": 2632 }, { "epoch": 0.81, "learning_rate": 1.8609032318772735e-06, "loss": 2.0535, "step": 2633 }, { "epoch": 0.81, "learning_rate": 1.8551165445959308e-06, "loss": 1.5891, "step": 2634 }, { "epoch": 0.81, "learning_rate": 1.8493379485002427e-06, "loss": 1.3032, "step": 2635 }, { "epoch": 0.81, "learning_rate": 1.8435674493307043e-06, "loss": 1.4067, "step": 2636 }, { "epoch": 0.81, "learning_rate": 1.8378050528197722e-06, "loss": 1.2746, "step": 2637 }, { "epoch": 0.81, "learning_rate": 1.8320507646918552e-06, "loss": 1.4263, "step": 2638 }, { "epoch": 0.81, "learning_rate": 1.8263045906633004e-06, "loss": 1.5586, "step": 2639 }, { "epoch": 0.81, "learning_rate": 1.8205665364424019e-06, "loss": 1.4499, "step": 2640 }, { "epoch": 0.81, "learning_rate": 1.81483660772938e-06, "loss": 0.9416, "step": 2641 }, { "epoch": 0.81, "learning_rate": 1.809114810216388e-06, "loss": 1.344, "step": 2642 }, { "epoch": 0.81, "learning_rate": 1.8034011495875082e-06, "loss": 1.193, "step": 2643 }, { "epoch": 0.81, "learning_rate": 1.79769563151872e-06, "loss": 1.4571, "step": 2644 }, { "epoch": 0.81, "learning_rate": 1.7919982616779308e-06, "loss": 1.0854, "step": 2645 }, { "epoch": 0.81, "learning_rate": 1.786309045724952e-06, "loss": 1.2303, "step": 2646 }, { "epoch": 0.81, "learning_rate": 1.7806279893114874e-06, "loss": 1.1636, "step": 2647 }, { "epoch": 0.81, "learning_rate": 1.7749550980811425e-06, "loss": 1.4557, "step": 2648 }, { "epoch": 0.82, "learning_rate": 1.7692903776694037e-06, "loss": 1.1746, "step": 2649 }, { "epoch": 0.82, "learning_rate": 1.7636338337036474e-06, "loss": 1.3724, "step": 2650 }, { "epoch": 0.82, "learning_rate": 1.7579854718031285e-06, "loss": 1.5699, "step": 2651 }, { "epoch": 0.82, "learning_rate": 1.7523452975789634e-06, "loss": 1.4874, "step": 2652 }, { "epoch": 0.82, "learning_rate": 1.7467133166341454e-06, "loss": 1.7284, "step": 2653 }, { "epoch": 0.82, "learning_rate": 1.741089534563527e-06, "loss": 1.2646, "step": 2654 }, { "epoch": 0.82, "learning_rate": 1.7354739569538081e-06, "loss": 1.3907, "step": 2655 }, { "epoch": 0.82, "learning_rate": 1.7298665893835498e-06, "loss": 1.1218, "step": 2656 }, { "epoch": 0.82, "learning_rate": 1.7242674374231483e-06, "loss": 1.4082, "step": 2657 }, { "epoch": 0.82, "learning_rate": 1.7186765066348388e-06, "loss": 1.5865, "step": 2658 }, { "epoch": 0.82, "learning_rate": 1.7130938025726963e-06, "loss": 1.1326, "step": 2659 }, { "epoch": 0.82, "learning_rate": 1.7075193307826144e-06, "loss": 1.1497, "step": 2660 }, { "epoch": 0.82, "learning_rate": 1.701953096802318e-06, "loss": 1.4058, "step": 2661 }, { "epoch": 0.82, "learning_rate": 1.6963951061613426e-06, "loss": 1.1423, "step": 2662 }, { "epoch": 0.82, "learning_rate": 1.6908453643810342e-06, "loss": 1.4075, "step": 2663 }, { "epoch": 0.82, "learning_rate": 1.6853038769745466e-06, "loss": 1.1628, "step": 2664 }, { "epoch": 0.82, "learning_rate": 1.679770649446837e-06, "loss": 1.4037, "step": 2665 }, { "epoch": 0.82, "learning_rate": 1.6742456872946478e-06, "loss": 1.7946, "step": 2666 }, { "epoch": 0.82, "learning_rate": 1.6687289960065213e-06, "loss": 1.777, "step": 2667 }, { "epoch": 0.82, "learning_rate": 1.6632205810627722e-06, "loss": 1.4469, "step": 2668 }, { "epoch": 0.82, "learning_rate": 1.6577204479355026e-06, "loss": 1.5557, "step": 2669 }, { "epoch": 0.82, "learning_rate": 1.652228602088587e-06, "loss": 1.7897, "step": 2670 }, { "epoch": 0.82, "learning_rate": 1.6467450489776581e-06, "loss": 1.2053, "step": 2671 }, { "epoch": 0.82, "learning_rate": 1.6412697940501244e-06, "loss": 1.6465, "step": 2672 }, { "epoch": 0.82, "learning_rate": 1.6358028427451379e-06, "loss": 1.6095, "step": 2673 }, { "epoch": 0.82, "learning_rate": 1.6303442004936077e-06, "loss": 1.1963, "step": 2674 }, { "epoch": 0.82, "learning_rate": 1.6248938727181917e-06, "loss": 1.2632, "step": 2675 }, { "epoch": 0.82, "learning_rate": 1.619451864833279e-06, "loss": 1.463, "step": 2676 }, { "epoch": 0.82, "learning_rate": 1.614018182245003e-06, "loss": 1.2237, "step": 2677 }, { "epoch": 0.82, "learning_rate": 1.6085928303512243e-06, "loss": 1.6416, "step": 2678 }, { "epoch": 0.82, "learning_rate": 1.6031758145415222e-06, "loss": 1.464, "step": 2679 }, { "epoch": 0.82, "learning_rate": 1.5977671401972018e-06, "loss": 1.207, "step": 2680 }, { "epoch": 0.82, "learning_rate": 1.5923668126912806e-06, "loss": 1.2699, "step": 2681 }, { "epoch": 0.83, "learning_rate": 1.5869748373884786e-06, "loss": 1.5458, "step": 2682 }, { "epoch": 0.83, "learning_rate": 1.5815912196452265e-06, "loss": 1.1712, "step": 2683 }, { "epoch": 0.83, "learning_rate": 1.5762159648096453e-06, "loss": 1.4899, "step": 2684 }, { "epoch": 0.83, "learning_rate": 1.5708490782215535e-06, "loss": 1.9001, "step": 2685 }, { "epoch": 0.83, "learning_rate": 1.5654905652124596e-06, "loss": 1.3993, "step": 2686 }, { "epoch": 0.83, "learning_rate": 1.560140431105539e-06, "loss": 1.4974, "step": 2687 }, { "epoch": 0.83, "learning_rate": 1.5547986812156579e-06, "loss": 1.5852, "step": 2688 }, { "epoch": 0.83, "learning_rate": 1.5494653208493527e-06, "loss": 1.4412, "step": 2689 }, { "epoch": 0.83, "learning_rate": 1.5441403553048162e-06, "loss": 1.5521, "step": 2690 }, { "epoch": 0.83, "learning_rate": 1.5388237898719105e-06, "loss": 1.2133, "step": 2691 }, { "epoch": 0.83, "learning_rate": 1.5335156298321485e-06, "loss": 1.7156, "step": 2692 }, { "epoch": 0.83, "learning_rate": 1.5282158804586934e-06, "loss": 0.9455, "step": 2693 }, { "epoch": 0.83, "learning_rate": 1.522924547016359e-06, "loss": 1.4694, "step": 2694 }, { "epoch": 0.83, "learning_rate": 1.5176416347615886e-06, "loss": 1.1641, "step": 2695 }, { "epoch": 0.83, "learning_rate": 1.5123671489424662e-06, "loss": 1.3567, "step": 2696 }, { "epoch": 0.83, "learning_rate": 1.5071010947987085e-06, "loss": 1.4311, "step": 2697 }, { "epoch": 0.83, "learning_rate": 1.5018434775616453e-06, "loss": 1.4747, "step": 2698 }, { "epoch": 0.83, "learning_rate": 1.4965943024542361e-06, "loss": 1.6376, "step": 2699 }, { "epoch": 0.83, "learning_rate": 1.4913535746910467e-06, "loss": 1.268, "step": 2700 }, { "epoch": 0.83, "learning_rate": 1.4861212994782514e-06, "loss": 1.3031, "step": 2701 }, { "epoch": 0.83, "learning_rate": 1.4808974820136345e-06, "loss": 1.2599, "step": 2702 }, { "epoch": 0.83, "learning_rate": 1.4756821274865696e-06, "loss": 1.4308, "step": 2703 }, { "epoch": 0.83, "learning_rate": 1.4704752410780287e-06, "loss": 1.2852, "step": 2704 }, { "epoch": 0.83, "learning_rate": 1.4652768279605734e-06, "loss": 1.2491, "step": 2705 }, { "epoch": 0.83, "learning_rate": 1.4600868932983402e-06, "loss": 1.3842, "step": 2706 }, { "epoch": 0.83, "learning_rate": 1.454905442247051e-06, "loss": 1.4247, "step": 2707 }, { "epoch": 0.83, "learning_rate": 1.4497324799539914e-06, "loss": 1.3358, "step": 2708 }, { "epoch": 0.83, "learning_rate": 1.444568011558023e-06, "loss": 1.3806, "step": 2709 }, { "epoch": 0.83, "learning_rate": 1.4394120421895674e-06, "loss": 1.4426, "step": 2710 }, { "epoch": 0.83, "learning_rate": 1.4342645769705977e-06, "loss": 1.153, "step": 2711 }, { "epoch": 0.83, "learning_rate": 1.4291256210146442e-06, "loss": 1.3113, "step": 2712 }, { "epoch": 0.83, "learning_rate": 1.4239951794267848e-06, "loss": 1.3176, "step": 2713 }, { "epoch": 0.84, "learning_rate": 1.4188732573036346e-06, "loss": 1.3246, "step": 2714 }, { "epoch": 0.84, "learning_rate": 1.4137598597333458e-06, "loss": 1.4593, "step": 2715 }, { "epoch": 0.84, "learning_rate": 1.4086549917956094e-06, "loss": 1.4904, "step": 2716 }, { "epoch": 0.84, "learning_rate": 1.4035586585616323e-06, "loss": 1.842, "step": 2717 }, { "epoch": 0.84, "learning_rate": 1.3984708650941536e-06, "loss": 1.319, "step": 2718 }, { "epoch": 0.84, "learning_rate": 1.3933916164474193e-06, "loss": 1.2174, "step": 2719 }, { "epoch": 0.84, "learning_rate": 1.3883209176671942e-06, "loss": 1.2161, "step": 2720 }, { "epoch": 0.84, "learning_rate": 1.3832587737907476e-06, "loss": 1.6838, "step": 2721 }, { "epoch": 0.84, "learning_rate": 1.3782051898468463e-06, "loss": 1.3245, "step": 2722 }, { "epoch": 0.84, "learning_rate": 1.3731601708557608e-06, "loss": 1.2519, "step": 2723 }, { "epoch": 0.84, "learning_rate": 1.3681237218292442e-06, "loss": 1.7184, "step": 2724 }, { "epoch": 0.84, "learning_rate": 1.3630958477705448e-06, "loss": 1.2253, "step": 2725 }, { "epoch": 0.84, "learning_rate": 1.35807655367439e-06, "loss": 1.6702, "step": 2726 }, { "epoch": 0.84, "learning_rate": 1.3530658445269784e-06, "loss": 1.3254, "step": 2727 }, { "epoch": 0.84, "learning_rate": 1.348063725305988e-06, "loss": 1.3084, "step": 2728 }, { "epoch": 0.84, "learning_rate": 1.3430702009805586e-06, "loss": 1.4824, "step": 2729 }, { "epoch": 0.84, "learning_rate": 1.3380852765112916e-06, "loss": 1.2608, "step": 2730 }, { "epoch": 0.84, "learning_rate": 1.3331089568502465e-06, "loss": 1.3589, "step": 2731 }, { "epoch": 0.84, "learning_rate": 1.3281412469409394e-06, "loss": 1.5002, "step": 2732 }, { "epoch": 0.84, "learning_rate": 1.323182151718324e-06, "loss": 1.6301, "step": 2733 }, { "epoch": 0.84, "learning_rate": 1.3182316761088055e-06, "loss": 1.8812, "step": 2734 }, { "epoch": 0.84, "learning_rate": 1.3132898250302173e-06, "loss": 1.7355, "step": 2735 }, { "epoch": 0.84, "learning_rate": 1.3083566033918316e-06, "loss": 1.3862, "step": 2736 }, { "epoch": 0.84, "learning_rate": 1.303432016094348e-06, "loss": 1.5021, "step": 2737 }, { "epoch": 0.84, "learning_rate": 1.2985160680298835e-06, "loss": 1.3212, "step": 2738 }, { "epoch": 0.84, "learning_rate": 1.2936087640819805e-06, "loss": 1.2141, "step": 2739 }, { "epoch": 0.84, "learning_rate": 1.2887101091255849e-06, "loss": 1.3545, "step": 2740 }, { "epoch": 0.84, "learning_rate": 1.2838201080270585e-06, "loss": 1.4226, "step": 2741 }, { "epoch": 0.84, "learning_rate": 1.2789387656441676e-06, "loss": 1.6109, "step": 2742 }, { "epoch": 0.84, "learning_rate": 1.2740660868260634e-06, "loss": 1.1917, "step": 2743 }, { "epoch": 0.84, "learning_rate": 1.269202076413305e-06, "loss": 1.3703, "step": 2744 }, { "epoch": 0.84, "learning_rate": 1.2643467392378383e-06, "loss": 1.4263, "step": 2745 }, { "epoch": 0.84, "learning_rate": 1.259500080122984e-06, "loss": 1.2747, "step": 2746 }, { "epoch": 0.85, "learning_rate": 1.2546621038834506e-06, "loss": 1.2362, "step": 2747 }, { "epoch": 0.85, "learning_rate": 1.2498328153253214e-06, "loss": 1.7445, "step": 2748 }, { "epoch": 0.85, "learning_rate": 1.2450122192460412e-06, "loss": 1.3055, "step": 2749 }, { "epoch": 0.85, "learning_rate": 1.2402003204344293e-06, "loss": 1.4602, "step": 2750 }, { "epoch": 0.85, "learning_rate": 1.2353971236706564e-06, "loss": 1.593, "step": 2751 }, { "epoch": 0.85, "learning_rate": 1.2306026337262544e-06, "loss": 1.3273, "step": 2752 }, { "epoch": 0.85, "learning_rate": 1.2258168553641058e-06, "loss": 1.5382, "step": 2753 }, { "epoch": 0.85, "learning_rate": 1.2210397933384344e-06, "loss": 1.8195, "step": 2754 }, { "epoch": 0.85, "learning_rate": 1.2162714523948104e-06, "loss": 1.542, "step": 2755 }, { "epoch": 0.85, "learning_rate": 1.2115118372701373e-06, "loss": 1.5063, "step": 2756 }, { "epoch": 0.85, "learning_rate": 1.206760952692647e-06, "loss": 1.1642, "step": 2757 }, { "epoch": 0.85, "learning_rate": 1.2020188033819092e-06, "loss": 1.1647, "step": 2758 }, { "epoch": 0.85, "learning_rate": 1.1972853940488017e-06, "loss": 1.2048, "step": 2759 }, { "epoch": 0.85, "learning_rate": 1.1925607293955322e-06, "loss": 0.9597, "step": 2760 }, { "epoch": 0.85, "learning_rate": 1.1878448141156162e-06, "loss": 1.2824, "step": 2761 }, { "epoch": 0.85, "learning_rate": 1.1831376528938742e-06, "loss": 1.0194, "step": 2762 }, { "epoch": 0.85, "learning_rate": 1.1784392504064378e-06, "loss": 1.1893, "step": 2763 }, { "epoch": 0.85, "learning_rate": 1.1737496113207336e-06, "loss": 1.2623, "step": 2764 }, { "epoch": 0.85, "learning_rate": 1.1690687402954792e-06, "loss": 1.0311, "step": 2765 }, { "epoch": 0.85, "learning_rate": 1.1643966419806907e-06, "loss": 1.508, "step": 2766 }, { "epoch": 0.85, "learning_rate": 1.1597333210176587e-06, "loss": 1.1493, "step": 2767 }, { "epoch": 0.85, "learning_rate": 1.1550787820389619e-06, "loss": 1.1789, "step": 2768 }, { "epoch": 0.85, "learning_rate": 1.1504330296684552e-06, "loss": 1.6571, "step": 2769 }, { "epoch": 0.85, "learning_rate": 1.1457960685212599e-06, "loss": 1.4267, "step": 2770 }, { "epoch": 0.85, "learning_rate": 1.1411679032037636e-06, "loss": 1.589, "step": 2771 }, { "epoch": 0.85, "learning_rate": 1.1365485383136255e-06, "loss": 1.3366, "step": 2772 }, { "epoch": 0.85, "learning_rate": 1.1319379784397499e-06, "loss": 1.5478, "step": 2773 }, { "epoch": 0.85, "learning_rate": 1.1273362281623045e-06, "loss": 1.5473, "step": 2774 }, { "epoch": 0.85, "learning_rate": 1.122743292052697e-06, "loss": 1.6997, "step": 2775 }, { "epoch": 0.85, "learning_rate": 1.1181591746735842e-06, "loss": 1.6264, "step": 2776 }, { "epoch": 0.85, "learning_rate": 1.1135838805788646e-06, "loss": 1.3311, "step": 2777 }, { "epoch": 0.85, "learning_rate": 1.1090174143136635e-06, "loss": 1.2912, "step": 2778 }, { "epoch": 0.86, "learning_rate": 1.1044597804143441e-06, "loss": 1.2467, "step": 2779 }, { "epoch": 0.86, "learning_rate": 1.099910983408493e-06, "loss": 1.4647, "step": 2780 }, { "epoch": 0.86, "learning_rate": 1.0953710278149155e-06, "loss": 1.3872, "step": 2781 }, { "epoch": 0.86, "learning_rate": 1.090839918143638e-06, "loss": 1.296, "step": 2782 }, { "epoch": 0.86, "learning_rate": 1.0863176588958957e-06, "loss": 1.3604, "step": 2783 }, { "epoch": 0.86, "learning_rate": 1.0818042545641349e-06, "loss": 1.4225, "step": 2784 }, { "epoch": 0.86, "learning_rate": 1.0772997096320049e-06, "loss": 1.3426, "step": 2785 }, { "epoch": 0.86, "learning_rate": 1.072804028574348e-06, "loss": 1.4228, "step": 2786 }, { "epoch": 0.86, "learning_rate": 1.0683172158572097e-06, "loss": 1.0379, "step": 2787 }, { "epoch": 0.86, "learning_rate": 1.0638392759378236e-06, "loss": 1.1639, "step": 2788 }, { "epoch": 0.86, "learning_rate": 1.0593702132646023e-06, "loss": 1.7137, "step": 2789 }, { "epoch": 0.86, "learning_rate": 1.0549100322771498e-06, "loss": 1.393, "step": 2790 }, { "epoch": 0.86, "learning_rate": 1.0504587374062392e-06, "loss": 1.4926, "step": 2791 }, { "epoch": 0.86, "learning_rate": 1.0460163330738182e-06, "loss": 1.5008, "step": 2792 }, { "epoch": 0.86, "learning_rate": 1.0415828236930059e-06, "loss": 1.5855, "step": 2793 }, { "epoch": 0.86, "learning_rate": 1.0371582136680792e-06, "loss": 1.285, "step": 2794 }, { "epoch": 0.86, "learning_rate": 1.0327425073944785e-06, "loss": 1.6029, "step": 2795 }, { "epoch": 0.86, "learning_rate": 1.028335709258801e-06, "loss": 1.0107, "step": 2796 }, { "epoch": 0.86, "learning_rate": 1.0239378236387864e-06, "loss": 1.4162, "step": 2797 }, { "epoch": 0.86, "learning_rate": 1.0195488549033305e-06, "loss": 1.5145, "step": 2798 }, { "epoch": 0.86, "learning_rate": 1.0151688074124645e-06, "loss": 1.6029, "step": 2799 }, { "epoch": 0.86, "learning_rate": 1.0107976855173574e-06, "loss": 1.3441, "step": 2800 }, { "epoch": 0.86, "learning_rate": 1.006435493560315e-06, "loss": 1.2033, "step": 2801 }, { "epoch": 0.86, "learning_rate": 1.0020822358747683e-06, "loss": 1.494, "step": 2802 }, { "epoch": 0.86, "learning_rate": 9.977379167852763e-07, "loss": 1.2489, "step": 2803 }, { "epoch": 0.86, "learning_rate": 9.934025406075176e-07, "loss": 1.6816, "step": 2804 }, { "epoch": 0.86, "learning_rate": 9.89076111648284e-07, "loss": 1.045, "step": 2805 }, { "epoch": 0.86, "learning_rate": 9.847586342054804e-07, "loss": 1.2116, "step": 2806 }, { "epoch": 0.86, "learning_rate": 9.804501125681243e-07, "loss": 1.435, "step": 2807 }, { "epoch": 0.86, "learning_rate": 9.76150551016326e-07, "loss": 1.3362, "step": 2808 }, { "epoch": 0.86, "learning_rate": 9.71859953821307e-07, "loss": 1.6681, "step": 2809 }, { "epoch": 0.86, "learning_rate": 9.675783252453718e-07, "loss": 1.2922, "step": 2810 }, { "epoch": 0.86, "learning_rate": 9.633056695419229e-07, "loss": 1.5096, "step": 2811 }, { "epoch": 0.87, "learning_rate": 9.590419909554516e-07, "loss": 1.3404, "step": 2812 }, { "epoch": 0.87, "learning_rate": 9.547872937215187e-07, "loss": 1.2368, "step": 2813 }, { "epoch": 0.87, "learning_rate": 9.505415820667751e-07, "loss": 1.2994, "step": 2814 }, { "epoch": 0.87, "learning_rate": 9.463048602089431e-07, "loss": 1.3717, "step": 2815 }, { "epoch": 0.87, "learning_rate": 9.420771323568079e-07, "loss": 1.2866, "step": 2816 }, { "epoch": 0.87, "learning_rate": 9.3785840271023e-07, "loss": 1.2937, "step": 2817 }, { "epoch": 0.87, "learning_rate": 9.336486754601204e-07, "loss": 1.3817, "step": 2818 }, { "epoch": 0.87, "learning_rate": 9.294479547884561e-07, "loss": 1.24, "step": 2819 }, { "epoch": 0.87, "learning_rate": 9.252562448682623e-07, "loss": 1.3247, "step": 2820 }, { "epoch": 0.87, "learning_rate": 9.210735498636125e-07, "loss": 1.4985, "step": 2821 }, { "epoch": 0.87, "learning_rate": 9.168998739296264e-07, "loss": 1.5737, "step": 2822 }, { "epoch": 0.87, "learning_rate": 9.127352212124663e-07, "loss": 1.2757, "step": 2823 }, { "epoch": 0.87, "learning_rate": 9.085795958493249e-07, "loss": 1.2902, "step": 2824 }, { "epoch": 0.87, "learning_rate": 9.044330019684311e-07, "loss": 1.1353, "step": 2825 }, { "epoch": 0.87, "learning_rate": 9.002954436890421e-07, "loss": 1.3566, "step": 2826 }, { "epoch": 0.87, "learning_rate": 8.961669251214356e-07, "loss": 1.3564, "step": 2827 }, { "epoch": 0.87, "learning_rate": 8.920474503669141e-07, "loss": 1.5027, "step": 2828 }, { "epoch": 0.87, "learning_rate": 8.879370235177909e-07, "loss": 1.6656, "step": 2829 }, { "epoch": 0.87, "learning_rate": 8.838356486573941e-07, "loss": 1.2436, "step": 2830 }, { "epoch": 0.87, "learning_rate": 8.797433298600622e-07, "loss": 1.6089, "step": 2831 }, { "epoch": 0.87, "learning_rate": 8.756600711911289e-07, "loss": 1.4268, "step": 2832 }, { "epoch": 0.87, "learning_rate": 8.715858767069374e-07, "loss": 1.304, "step": 2833 }, { "epoch": 0.87, "learning_rate": 8.675207504548178e-07, "loss": 1.3828, "step": 2834 }, { "epoch": 0.87, "learning_rate": 8.634646964730975e-07, "loss": 1.2387, "step": 2835 }, { "epoch": 0.87, "learning_rate": 8.594177187910913e-07, "loss": 1.4763, "step": 2836 }, { "epoch": 0.87, "learning_rate": 8.55379821429092e-07, "loss": 1.3914, "step": 2837 }, { "epoch": 0.87, "learning_rate": 8.513510083983778e-07, "loss": 1.1502, "step": 2838 }, { "epoch": 0.87, "learning_rate": 8.473312837012027e-07, "loss": 1.6727, "step": 2839 }, { "epoch": 0.87, "learning_rate": 8.433206513307878e-07, "loss": 1.0827, "step": 2840 }, { "epoch": 0.87, "learning_rate": 8.393191152713231e-07, "loss": 1.6531, "step": 2841 }, { "epoch": 0.87, "learning_rate": 8.353266794979653e-07, "loss": 1.4707, "step": 2842 }, { "epoch": 0.87, "learning_rate": 8.313433479768262e-07, "loss": 1.4687, "step": 2843 }, { "epoch": 0.88, "learning_rate": 8.273691246649795e-07, "loss": 1.4851, "step": 2844 }, { "epoch": 0.88, "learning_rate": 8.234040135104437e-07, "loss": 1.6681, "step": 2845 }, { "epoch": 0.88, "learning_rate": 8.194480184521891e-07, "loss": 1.5478, "step": 2846 }, { "epoch": 0.88, "learning_rate": 8.155011434201332e-07, "loss": 1.416, "step": 2847 }, { "epoch": 0.88, "learning_rate": 8.115633923351252e-07, "loss": 1.4117, "step": 2848 }, { "epoch": 0.88, "learning_rate": 8.076347691089592e-07, "loss": 1.3627, "step": 2849 }, { "epoch": 0.88, "learning_rate": 8.037152776443535e-07, "loss": 1.5196, "step": 2850 }, { "epoch": 0.88, "learning_rate": 7.998049218349624e-07, "loss": 1.433, "step": 2851 }, { "epoch": 0.88, "learning_rate": 7.959037055653617e-07, "loss": 1.1253, "step": 2852 }, { "epoch": 0.88, "learning_rate": 7.920116327110449e-07, "loss": 1.2896, "step": 2853 }, { "epoch": 0.88, "learning_rate": 7.88128707138428e-07, "loss": 1.0381, "step": 2854 }, { "epoch": 0.88, "learning_rate": 7.842549327048366e-07, "loss": 1.5005, "step": 2855 }, { "epoch": 0.88, "learning_rate": 7.803903132585023e-07, "loss": 1.5039, "step": 2856 }, { "epoch": 0.88, "learning_rate": 7.765348526385685e-07, "loss": 1.0744, "step": 2857 }, { "epoch": 0.88, "learning_rate": 7.726885546750796e-07, "loss": 1.219, "step": 2858 }, { "epoch": 0.88, "learning_rate": 7.688514231889721e-07, "loss": 1.2464, "step": 2859 }, { "epoch": 0.88, "learning_rate": 7.650234619920826e-07, "loss": 1.2983, "step": 2860 }, { "epoch": 0.88, "learning_rate": 7.612046748871327e-07, "loss": 1.2922, "step": 2861 }, { "epoch": 0.88, "learning_rate": 7.573950656677342e-07, "loss": 1.6027, "step": 2862 }, { "epoch": 0.88, "learning_rate": 7.535946381183812e-07, "loss": 1.1291, "step": 2863 }, { "epoch": 0.88, "learning_rate": 7.498033960144435e-07, "loss": 1.5354, "step": 2864 }, { "epoch": 0.88, "learning_rate": 7.460213431221719e-07, "loss": 1.1072, "step": 2865 }, { "epoch": 0.88, "learning_rate": 7.422484831986798e-07, "loss": 1.3677, "step": 2866 }, { "epoch": 0.88, "learning_rate": 7.384848199919558e-07, "loss": 1.9943, "step": 2867 }, { "epoch": 0.88, "learning_rate": 7.347303572408515e-07, "loss": 1.3345, "step": 2868 }, { "epoch": 0.88, "learning_rate": 7.309850986750766e-07, "loss": 1.1227, "step": 2869 }, { "epoch": 0.88, "learning_rate": 7.272490480151939e-07, "loss": 1.3734, "step": 2870 }, { "epoch": 0.88, "learning_rate": 7.23522208972628e-07, "loss": 1.6701, "step": 2871 }, { "epoch": 0.88, "learning_rate": 7.198045852496438e-07, "loss": 2.5635, "step": 2872 }, { "epoch": 0.88, "learning_rate": 7.160961805393574e-07, "loss": 1.7899, "step": 2873 }, { "epoch": 0.88, "learning_rate": 7.123969985257262e-07, "loss": 1.2196, "step": 2874 }, { "epoch": 0.88, "learning_rate": 7.087070428835397e-07, "loss": 1.4816, "step": 2875 }, { "epoch": 0.88, "learning_rate": 7.05026317278431e-07, "loss": 1.3796, "step": 2876 }, { "epoch": 0.89, "learning_rate": 7.013548253668556e-07, "loss": 1.7681, "step": 2877 }, { "epoch": 0.89, "learning_rate": 6.976925707961002e-07, "loss": 1.2791, "step": 2878 }, { "epoch": 0.89, "learning_rate": 6.940395572042791e-07, "loss": 1.3327, "step": 2879 }, { "epoch": 0.89, "learning_rate": 6.903957882203171e-07, "loss": 1.3877, "step": 2880 }, { "epoch": 0.89, "learning_rate": 6.867612674639634e-07, "loss": 1.1897, "step": 2881 }, { "epoch": 0.89, "learning_rate": 6.831359985457753e-07, "loss": 1.3656, "step": 2882 }, { "epoch": 0.89, "learning_rate": 6.79519985067123e-07, "loss": 1.3448, "step": 2883 }, { "epoch": 0.89, "learning_rate": 6.759132306201788e-07, "loss": 1.6175, "step": 2884 }, { "epoch": 0.89, "learning_rate": 6.723157387879164e-07, "loss": 1.2321, "step": 2885 }, { "epoch": 0.89, "learning_rate": 6.687275131441129e-07, "loss": 1.6447, "step": 2886 }, { "epoch": 0.89, "learning_rate": 6.651485572533379e-07, "loss": 1.2069, "step": 2887 }, { "epoch": 0.89, "learning_rate": 6.615788746709495e-07, "loss": 1.1779, "step": 2888 }, { "epoch": 0.89, "learning_rate": 6.580184689430958e-07, "loss": 1.3162, "step": 2889 }, { "epoch": 0.89, "learning_rate": 6.54467343606714e-07, "loss": 1.3883, "step": 2890 }, { "epoch": 0.89, "learning_rate": 6.509255021895111e-07, "loss": 1.4417, "step": 2891 }, { "epoch": 0.89, "learning_rate": 6.473929482099827e-07, "loss": 1.4512, "step": 2892 }, { "epoch": 0.89, "learning_rate": 6.438696851773906e-07, "loss": 1.3571, "step": 2893 }, { "epoch": 0.89, "learning_rate": 6.403557165917695e-07, "loss": 1.4242, "step": 2894 }, { "epoch": 0.89, "learning_rate": 6.368510459439248e-07, "loss": 1.3279, "step": 2895 }, { "epoch": 0.89, "learning_rate": 6.333556767154159e-07, "loss": 1.4234, "step": 2896 }, { "epoch": 0.89, "learning_rate": 6.298696123785719e-07, "loss": 1.3089, "step": 2897 }, { "epoch": 0.89, "learning_rate": 6.263928563964727e-07, "loss": 1.344, "step": 2898 }, { "epoch": 0.89, "learning_rate": 6.2292541222295e-07, "loss": 1.043, "step": 2899 }, { "epoch": 0.89, "learning_rate": 6.194672833025927e-07, "loss": 1.3056, "step": 2900 }, { "epoch": 0.89, "learning_rate": 6.160184730707264e-07, "loss": 1.5247, "step": 2901 }, { "epoch": 0.89, "learning_rate": 6.125789849534269e-07, "loss": 2.1458, "step": 2902 }, { "epoch": 0.89, "learning_rate": 6.091488223675058e-07, "loss": 1.2948, "step": 2903 }, { "epoch": 0.89, "learning_rate": 6.057279887205092e-07, "loss": 1.2128, "step": 2904 }, { "epoch": 0.89, "learning_rate": 6.023164874107202e-07, "loss": 1.5385, "step": 2905 }, { "epoch": 0.89, "learning_rate": 5.989143218271487e-07, "loss": 1.3961, "step": 2906 }, { "epoch": 0.89, "learning_rate": 5.95521495349527e-07, "loss": 1.3153, "step": 2907 }, { "epoch": 0.89, "learning_rate": 5.921380113483188e-07, "loss": 1.5059, "step": 2908 }, { "epoch": 0.9, "learning_rate": 5.887638731846967e-07, "loss": 1.2553, "step": 2909 }, { "epoch": 0.9, "learning_rate": 5.85399084210555e-07, "loss": 1.5093, "step": 2910 }, { "epoch": 0.9, "learning_rate": 5.820436477685021e-07, "loss": 1.5094, "step": 2911 }, { "epoch": 0.9, "learning_rate": 5.786975671918482e-07, "loss": 1.4329, "step": 2912 }, { "epoch": 0.9, "learning_rate": 5.753608458046145e-07, "loss": 1.3455, "step": 2913 }, { "epoch": 0.9, "learning_rate": 5.720334869215282e-07, "loss": 1.3079, "step": 2914 }, { "epoch": 0.9, "learning_rate": 5.687154938480055e-07, "loss": 1.3543, "step": 2915 }, { "epoch": 0.9, "learning_rate": 5.654068698801696e-07, "loss": 1.3558, "step": 2916 }, { "epoch": 0.9, "learning_rate": 5.62107618304828e-07, "loss": 1.3352, "step": 2917 }, { "epoch": 0.9, "learning_rate": 5.588177423994812e-07, "loss": 1.4507, "step": 2918 }, { "epoch": 0.9, "learning_rate": 5.555372454323182e-07, "loss": 1.3969, "step": 2919 }, { "epoch": 0.9, "learning_rate": 5.522661306622046e-07, "loss": 1.4068, "step": 2920 }, { "epoch": 0.9, "learning_rate": 5.490044013386919e-07, "loss": 1.2338, "step": 2921 }, { "epoch": 0.9, "learning_rate": 5.45752060702005e-07, "loss": 1.3456, "step": 2922 }, { "epoch": 0.9, "learning_rate": 5.425091119830417e-07, "loss": 1.644, "step": 2923 }, { "epoch": 0.9, "learning_rate": 5.392755584033715e-07, "loss": 2.0859, "step": 2924 }, { "epoch": 0.9, "learning_rate": 5.360514031752284e-07, "loss": 1.3264, "step": 2925 }, { "epoch": 0.9, "learning_rate": 5.328366495015113e-07, "loss": 1.3682, "step": 2926 }, { "epoch": 0.9, "learning_rate": 5.296313005757814e-07, "loss": 1.5484, "step": 2927 }, { "epoch": 0.9, "learning_rate": 5.264353595822535e-07, "loss": 1.6126, "step": 2928 }, { "epoch": 0.9, "learning_rate": 5.232488296957983e-07, "loss": 1.3289, "step": 2929 }, { "epoch": 0.9, "learning_rate": 5.200717140819389e-07, "loss": 1.3832, "step": 2930 }, { "epoch": 0.9, "learning_rate": 5.169040158968431e-07, "loss": 1.5822, "step": 2931 }, { "epoch": 0.9, "learning_rate": 5.137457382873257e-07, "loss": 1.3771, "step": 2932 }, { "epoch": 0.9, "learning_rate": 5.105968843908449e-07, "loss": 1.2385, "step": 2933 }, { "epoch": 0.9, "learning_rate": 5.074574573354896e-07, "loss": 1.606, "step": 2934 }, { "epoch": 0.9, "learning_rate": 5.043274602399939e-07, "loss": 1.4042, "step": 2935 }, { "epoch": 0.9, "learning_rate": 5.012068962137151e-07, "loss": 1.821, "step": 2936 }, { "epoch": 0.9, "learning_rate": 4.980957683566468e-07, "loss": 1.9755, "step": 2937 }, { "epoch": 0.9, "learning_rate": 4.949940797594066e-07, "loss": 1.5351, "step": 2938 }, { "epoch": 0.9, "learning_rate": 4.919018335032322e-07, "loss": 1.6128, "step": 2939 }, { "epoch": 0.9, "learning_rate": 4.888190326599829e-07, "loss": 1.2311, "step": 2940 }, { "epoch": 0.9, "learning_rate": 4.857456802921368e-07, "loss": 1.6116, "step": 2941 }, { "epoch": 0.91, "learning_rate": 4.826817794527828e-07, "loss": 1.2975, "step": 2942 }, { "epoch": 0.91, "learning_rate": 4.796273331856227e-07, "loss": 1.1241, "step": 2943 }, { "epoch": 0.91, "learning_rate": 4.76582344524964e-07, "loss": 1.291, "step": 2944 }, { "epoch": 0.91, "learning_rate": 4.735468164957213e-07, "loss": 1.3337, "step": 2945 }, { "epoch": 0.91, "learning_rate": 4.705207521134125e-07, "loss": 1.7244, "step": 2946 }, { "epoch": 0.91, "learning_rate": 4.6750415438414745e-07, "loss": 1.3703, "step": 2947 }, { "epoch": 0.91, "learning_rate": 4.644970263046367e-07, "loss": 1.4727, "step": 2948 }, { "epoch": 0.91, "learning_rate": 4.614993708621862e-07, "loss": 1.441, "step": 2949 }, { "epoch": 0.91, "learning_rate": 4.58511191034684e-07, "loss": 1.1016, "step": 2950 }, { "epoch": 0.91, "learning_rate": 4.555324897906133e-07, "loss": 1.6036, "step": 2951 }, { "epoch": 0.91, "learning_rate": 4.525632700890348e-07, "loss": 1.3771, "step": 2952 }, { "epoch": 0.91, "learning_rate": 4.4960353487959354e-07, "loss": 1.7475, "step": 2953 }, { "epoch": 0.91, "learning_rate": 4.46653287102512e-07, "loss": 1.4229, "step": 2954 }, { "epoch": 0.91, "learning_rate": 4.437125296885847e-07, "loss": 1.431, "step": 2955 }, { "epoch": 0.91, "learning_rate": 4.407812655591848e-07, "loss": 1.3778, "step": 2956 }, { "epoch": 0.91, "learning_rate": 4.378594976262496e-07, "loss": 1.3583, "step": 2957 }, { "epoch": 0.91, "learning_rate": 4.3494722879228404e-07, "loss": 1.4101, "step": 2958 }, { "epoch": 0.91, "learning_rate": 4.320444619503583e-07, "loss": 1.84, "step": 2959 }, { "epoch": 0.91, "learning_rate": 4.2915119998409914e-07, "loss": 1.3871, "step": 2960 }, { "epoch": 0.91, "learning_rate": 4.262674457676952e-07, "loss": 1.3767, "step": 2961 }, { "epoch": 0.91, "learning_rate": 4.233932021658904e-07, "loss": 1.7263, "step": 2962 }, { "epoch": 0.91, "learning_rate": 4.205284720339764e-07, "loss": 1.4118, "step": 2963 }, { "epoch": 0.91, "learning_rate": 4.176732582177967e-07, "loss": 1.3906, "step": 2964 }, { "epoch": 0.91, "learning_rate": 4.1482756355374463e-07, "loss": 1.4991, "step": 2965 }, { "epoch": 0.91, "learning_rate": 4.1199139086874895e-07, "loss": 1.3459, "step": 2966 }, { "epoch": 0.91, "learning_rate": 4.0916474298028694e-07, "loss": 1.1746, "step": 2967 }, { "epoch": 0.91, "learning_rate": 4.063476226963703e-07, "loss": 1.2817, "step": 2968 }, { "epoch": 0.91, "learning_rate": 4.0354003281554586e-07, "loss": 1.5201, "step": 2969 }, { "epoch": 0.91, "learning_rate": 4.007419761268938e-07, "loss": 1.0179, "step": 2970 }, { "epoch": 0.91, "learning_rate": 3.9795345541002395e-07, "loss": 1.7437, "step": 2971 }, { "epoch": 0.91, "learning_rate": 3.951744734350715e-07, "loss": 1.2231, "step": 2972 }, { "epoch": 0.91, "learning_rate": 3.924050329627005e-07, "loss": 1.5422, "step": 2973 }, { "epoch": 0.92, "learning_rate": 3.896451367440901e-07, "loss": 1.4199, "step": 2974 }, { "epoch": 0.92, "learning_rate": 3.8689478752094167e-07, "loss": 1.5682, "step": 2975 }, { "epoch": 0.92, "learning_rate": 3.841539880254708e-07, "loss": 1.7482, "step": 2976 }, { "epoch": 0.92, "learning_rate": 3.814227409804094e-07, "loss": 1.2629, "step": 2977 }, { "epoch": 0.92, "learning_rate": 3.787010490989973e-07, "loss": 1.4626, "step": 2978 }, { "epoch": 0.92, "learning_rate": 3.759889150849805e-07, "loss": 1.4805, "step": 2979 }, { "epoch": 0.92, "learning_rate": 3.73286341632616e-07, "loss": 1.18, "step": 2980 }, { "epoch": 0.92, "learning_rate": 3.7059333142665854e-07, "loss": 1.2404, "step": 2981 }, { "epoch": 0.92, "learning_rate": 3.6790988714236144e-07, "loss": 1.4339, "step": 2982 }, { "epoch": 0.92, "learning_rate": 3.6523601144548003e-07, "loss": 1.4648, "step": 2983 }, { "epoch": 0.92, "learning_rate": 3.6257170699226076e-07, "loss": 1.3987, "step": 2984 }, { "epoch": 0.92, "learning_rate": 3.599169764294419e-07, "loss": 1.2372, "step": 2985 }, { "epoch": 0.92, "learning_rate": 3.5727182239425393e-07, "loss": 1.3294, "step": 2986 }, { "epoch": 0.92, "learning_rate": 3.546362475144105e-07, "loss": 1.2964, "step": 2987 }, { "epoch": 0.92, "learning_rate": 3.5201025440810943e-07, "loss": 1.4169, "step": 2988 }, { "epoch": 0.92, "learning_rate": 3.49393845684034e-07, "loss": 1.2522, "step": 2989 }, { "epoch": 0.92, "learning_rate": 3.467870239413407e-07, "loss": 1.6116, "step": 2990 }, { "epoch": 0.92, "learning_rate": 3.441897917696679e-07, "loss": 1.6494, "step": 2991 }, { "epoch": 0.92, "learning_rate": 3.416021517491208e-07, "loss": 1.2379, "step": 2992 }, { "epoch": 0.92, "learning_rate": 3.3902410645028197e-07, "loss": 1.3141, "step": 2993 }, { "epoch": 0.92, "learning_rate": 3.364556584342005e-07, "loss": 1.4326, "step": 2994 }, { "epoch": 0.92, "learning_rate": 3.3389681025239095e-07, "loss": 1.4313, "step": 2995 }, { "epoch": 0.92, "learning_rate": 3.3134756444683e-07, "loss": 1.3092, "step": 2996 }, { "epoch": 0.92, "learning_rate": 3.288079235499575e-07, "loss": 1.5026, "step": 2997 }, { "epoch": 0.92, "learning_rate": 3.262778900846708e-07, "loss": 1.538, "step": 2998 }, { "epoch": 0.92, "learning_rate": 3.2375746656432284e-07, "loss": 1.1904, "step": 2999 }, { "epoch": 0.92, "learning_rate": 3.212466554927229e-07, "loss": 1.2712, "step": 3000 }, { "epoch": 0.92, "learning_rate": 3.187454593641248e-07, "loss": 1.2696, "step": 3001 }, { "epoch": 0.92, "learning_rate": 3.162538806632387e-07, "loss": 1.2591, "step": 3002 }, { "epoch": 0.92, "learning_rate": 3.1377192186521357e-07, "loss": 1.534, "step": 3003 }, { "epoch": 0.92, "learning_rate": 3.11299585435646e-07, "loss": 1.1406, "step": 3004 }, { "epoch": 0.92, "learning_rate": 3.088368738305747e-07, "loss": 1.126, "step": 3005 }, { "epoch": 0.92, "learning_rate": 3.0638378949647275e-07, "loss": 1.251, "step": 3006 }, { "epoch": 0.93, "learning_rate": 3.03940334870253e-07, "loss": 1.4787, "step": 3007 }, { "epoch": 0.93, "learning_rate": 3.015065123792593e-07, "loss": 1.7717, "step": 3008 }, { "epoch": 0.93, "learning_rate": 2.9908232444126993e-07, "loss": 1.1727, "step": 3009 }, { "epoch": 0.93, "learning_rate": 2.966677734644896e-07, "loss": 1.3768, "step": 3010 }, { "epoch": 0.93, "learning_rate": 2.942628618475507e-07, "loss": 1.3341, "step": 3011 }, { "epoch": 0.93, "learning_rate": 2.9186759197951e-07, "loss": 1.2621, "step": 3012 }, { "epoch": 0.93, "learning_rate": 2.894819662398463e-07, "loss": 1.7663, "step": 3013 }, { "epoch": 0.93, "learning_rate": 2.8710598699845494e-07, "loss": 1.2015, "step": 3014 }, { "epoch": 0.93, "learning_rate": 2.8473965661565353e-07, "loss": 1.2168, "step": 3015 }, { "epoch": 0.93, "learning_rate": 2.823829774421716e-07, "loss": 1.2527, "step": 3016 }, { "epoch": 0.93, "learning_rate": 2.8003595181914866e-07, "loss": 1.2962, "step": 3017 }, { "epoch": 0.93, "learning_rate": 2.776985820781397e-07, "loss": 1.0615, "step": 3018 }, { "epoch": 0.93, "learning_rate": 2.753708705411018e-07, "loss": 1.2548, "step": 3019 }, { "epoch": 0.93, "learning_rate": 2.7305281952040183e-07, "loss": 1.175, "step": 3020 }, { "epoch": 0.93, "learning_rate": 2.707444313188101e-07, "loss": 1.7075, "step": 3021 }, { "epoch": 0.93, "learning_rate": 2.6844570822949224e-07, "loss": 1.7005, "step": 3022 }, { "epoch": 0.93, "learning_rate": 2.661566525360193e-07, "loss": 1.6203, "step": 3023 }, { "epoch": 0.93, "learning_rate": 2.6387726651235343e-07, "loss": 1.2617, "step": 3024 }, { "epoch": 0.93, "learning_rate": 2.616075524228545e-07, "loss": 1.3926, "step": 3025 }, { "epoch": 0.93, "learning_rate": 2.593475125222722e-07, "loss": 1.317, "step": 3026 }, { "epoch": 0.93, "learning_rate": 2.570971490557461e-07, "loss": 1.45, "step": 3027 }, { "epoch": 0.93, "learning_rate": 2.548564642588036e-07, "loss": 1.6772, "step": 3028 }, { "epoch": 0.93, "learning_rate": 2.526254603573586e-07, "loss": 1.2252, "step": 3029 }, { "epoch": 0.93, "learning_rate": 2.504041395677026e-07, "loss": 1.2445, "step": 3030 }, { "epoch": 0.93, "learning_rate": 2.4819250409651605e-07, "loss": 1.2043, "step": 3031 }, { "epoch": 0.93, "learning_rate": 2.459905561408515e-07, "loss": 1.1971, "step": 3032 }, { "epoch": 0.93, "learning_rate": 2.4379829788813815e-07, "loss": 1.547, "step": 3033 }, { "epoch": 0.93, "learning_rate": 2.416157315161849e-07, "loss": 1.1584, "step": 3034 }, { "epoch": 0.93, "learning_rate": 2.3944285919316525e-07, "loss": 1.4318, "step": 3035 }, { "epoch": 0.93, "learning_rate": 2.3727968307762806e-07, "loss": 1.9251, "step": 3036 }, { "epoch": 0.93, "learning_rate": 2.3512620531848885e-07, "loss": 1.5439, "step": 3037 }, { "epoch": 0.93, "learning_rate": 2.3298242805502414e-07, "loss": 1.4397, "step": 3038 }, { "epoch": 0.94, "learning_rate": 2.308483534168815e-07, "loss": 1.4431, "step": 3039 }, { "epoch": 0.94, "learning_rate": 2.28723983524064e-07, "loss": 1.4653, "step": 3040 }, { "epoch": 0.94, "learning_rate": 2.2660932048693686e-07, "loss": 1.4857, "step": 3041 }, { "epoch": 0.94, "learning_rate": 2.2450436640622076e-07, "loss": 1.1715, "step": 3042 }, { "epoch": 0.94, "learning_rate": 2.2240912337299304e-07, "loss": 1.0109, "step": 3043 }, { "epoch": 0.94, "learning_rate": 2.2032359346868203e-07, "loss": 1.5757, "step": 3044 }, { "epoch": 0.94, "learning_rate": 2.182477787650694e-07, "loss": 1.5002, "step": 3045 }, { "epoch": 0.94, "learning_rate": 2.161816813242823e-07, "loss": 1.2315, "step": 3046 }, { "epoch": 0.94, "learning_rate": 2.1412530319879888e-07, "loss": 1.5618, "step": 3047 }, { "epoch": 0.94, "learning_rate": 2.1207864643144061e-07, "loss": 1.6039, "step": 3048 }, { "epoch": 0.94, "learning_rate": 2.100417130553678e-07, "loss": 1.2323, "step": 3049 }, { "epoch": 0.94, "learning_rate": 2.080145050940874e-07, "loss": 1.8419, "step": 3050 }, { "epoch": 0.94, "learning_rate": 2.0599702456144178e-07, "loss": 1.2763, "step": 3051 }, { "epoch": 0.94, "learning_rate": 2.0398927346161e-07, "loss": 2.0699, "step": 3052 }, { "epoch": 0.94, "learning_rate": 2.0199125378910666e-07, "loss": 1.5458, "step": 3053 }, { "epoch": 0.94, "learning_rate": 2.0000296752877734e-07, "loss": 1.2847, "step": 3054 }, { "epoch": 0.94, "learning_rate": 1.9802441665580208e-07, "loss": 1.3646, "step": 3055 }, { "epoch": 0.94, "learning_rate": 1.960556031356875e-07, "loss": 1.6048, "step": 3056 }, { "epoch": 0.94, "learning_rate": 1.9409652892426578e-07, "loss": 1.2597, "step": 3057 }, { "epoch": 0.94, "learning_rate": 1.921471959676957e-07, "loss": 1.3251, "step": 3058 }, { "epoch": 0.94, "learning_rate": 1.9020760620245938e-07, "loss": 1.3719, "step": 3059 }, { "epoch": 0.94, "learning_rate": 1.882777615553577e-07, "loss": 1.4458, "step": 3060 }, { "epoch": 0.94, "learning_rate": 1.8635766394351384e-07, "loss": 1.2551, "step": 3061 }, { "epoch": 0.94, "learning_rate": 1.8444731527436532e-07, "loss": 1.4206, "step": 3062 }, { "epoch": 0.94, "learning_rate": 1.825467174456652e-07, "loss": 1.3383, "step": 3063 }, { "epoch": 0.94, "learning_rate": 1.8065587234548543e-07, "loss": 1.4454, "step": 3064 }, { "epoch": 0.94, "learning_rate": 1.787747818522001e-07, "loss": 1.5682, "step": 3065 }, { "epoch": 0.94, "learning_rate": 1.7690344783450108e-07, "loss": 1.4621, "step": 3066 }, { "epoch": 0.94, "learning_rate": 1.7504187215138358e-07, "loss": 1.4479, "step": 3067 }, { "epoch": 0.94, "learning_rate": 1.7319005665214938e-07, "loss": 1.0647, "step": 3068 }, { "epoch": 0.94, "learning_rate": 1.713480031764081e-07, "loss": 1.2686, "step": 3069 }, { "epoch": 0.94, "learning_rate": 1.6951571355406594e-07, "loss": 1.4626, "step": 3070 }, { "epoch": 0.94, "learning_rate": 1.6769318960533465e-07, "loss": 1.5348, "step": 3071 }, { "epoch": 0.95, "learning_rate": 1.6588043314072268e-07, "loss": 1.6317, "step": 3072 }, { "epoch": 0.95, "learning_rate": 1.640774459610328e-07, "loss": 1.2747, "step": 3073 }, { "epoch": 0.95, "learning_rate": 1.622842298573668e-07, "loss": 1.7529, "step": 3074 }, { "epoch": 0.95, "learning_rate": 1.6050078661112078e-07, "loss": 1.3554, "step": 3075 }, { "epoch": 0.95, "learning_rate": 1.5872711799397645e-07, "loss": 1.5009, "step": 3076 }, { "epoch": 0.95, "learning_rate": 1.5696322576791101e-07, "loss": 1.5501, "step": 3077 }, { "epoch": 0.95, "learning_rate": 1.5520911168518616e-07, "loss": 1.147, "step": 3078 }, { "epoch": 0.95, "learning_rate": 1.5346477748835353e-07, "loss": 1.4403, "step": 3079 }, { "epoch": 0.95, "learning_rate": 1.5173022491024703e-07, "loss": 1.4366, "step": 3080 }, { "epoch": 0.95, "learning_rate": 1.500054556739805e-07, "loss": 1.2863, "step": 3081 }, { "epoch": 0.95, "learning_rate": 1.4829047149295452e-07, "loss": 1.2435, "step": 3082 }, { "epoch": 0.95, "learning_rate": 1.4658527407084732e-07, "loss": 1.2102, "step": 3083 }, { "epoch": 0.95, "learning_rate": 1.4488986510161175e-07, "loss": 1.2077, "step": 3084 }, { "epoch": 0.95, "learning_rate": 1.432042462694827e-07, "loss": 1.1724, "step": 3085 }, { "epoch": 0.95, "learning_rate": 1.41528419248963e-07, "loss": 1.3969, "step": 3086 }, { "epoch": 0.95, "learning_rate": 1.39862385704832e-07, "loss": 1.4128, "step": 3087 }, { "epoch": 0.95, "learning_rate": 1.3820614729214033e-07, "loss": 1.5662, "step": 3088 }, { "epoch": 0.95, "learning_rate": 1.3655970565620623e-07, "loss": 1.3091, "step": 3089 }, { "epoch": 0.95, "learning_rate": 1.349230624326159e-07, "loss": 1.3974, "step": 3090 }, { "epoch": 0.95, "learning_rate": 1.3329621924722536e-07, "loss": 1.1484, "step": 3091 }, { "epoch": 0.95, "learning_rate": 1.3167917771614969e-07, "loss": 1.6047, "step": 3092 }, { "epoch": 0.95, "learning_rate": 1.3007193944577056e-07, "loss": 1.5473, "step": 3093 }, { "epoch": 0.95, "learning_rate": 1.284745060327297e-07, "loss": 1.3651, "step": 3094 }, { "epoch": 0.95, "learning_rate": 1.2688687906392772e-07, "loss": 1.2852, "step": 3095 }, { "epoch": 0.95, "learning_rate": 1.2530906011652744e-07, "loss": 1.6499, "step": 3096 }, { "epoch": 0.95, "learning_rate": 1.2374105075794175e-07, "loss": 1.359, "step": 3097 }, { "epoch": 0.95, "learning_rate": 1.221828525458457e-07, "loss": 1.4378, "step": 3098 }, { "epoch": 0.95, "learning_rate": 1.206344670281645e-07, "loss": 1.4479, "step": 3099 }, { "epoch": 0.95, "learning_rate": 1.1909589574307323e-07, "loss": 1.2726, "step": 3100 }, { "epoch": 0.95, "learning_rate": 1.175671402190004e-07, "loss": 1.533, "step": 3101 }, { "epoch": 0.95, "learning_rate": 1.1604820197462341e-07, "loss": 1.4281, "step": 3102 }, { "epoch": 0.95, "learning_rate": 1.1453908251886636e-07, "loss": 1.4369, "step": 3103 }, { "epoch": 0.96, "learning_rate": 1.1303978335089894e-07, "loss": 1.3176, "step": 3104 }, { "epoch": 0.96, "learning_rate": 1.1155030596013638e-07, "loss": 1.3957, "step": 3105 }, { "epoch": 0.96, "learning_rate": 1.10070651826234e-07, "loss": 1.2807, "step": 3106 }, { "epoch": 0.96, "learning_rate": 1.0860082241909487e-07, "loss": 1.1343, "step": 3107 }, { "epoch": 0.96, "learning_rate": 1.0714081919885432e-07, "loss": 1.4709, "step": 3108 }, { "epoch": 0.96, "learning_rate": 1.056906436158911e-07, "loss": 1.007, "step": 3109 }, { "epoch": 0.96, "learning_rate": 1.0425029711082169e-07, "loss": 1.2685, "step": 3110 }, { "epoch": 0.96, "learning_rate": 1.0281978111449375e-07, "loss": 1.6136, "step": 3111 }, { "epoch": 0.96, "learning_rate": 1.013990970479939e-07, "loss": 1.5904, "step": 3112 }, { "epoch": 0.96, "learning_rate": 9.998824632263981e-08, "loss": 1.3402, "step": 3113 }, { "epoch": 0.96, "learning_rate": 9.858723033998041e-08, "loss": 2.364, "step": 3114 }, { "epoch": 0.96, "learning_rate": 9.71960504917957e-08, "loss": 1.3387, "step": 3115 }, { "epoch": 0.96, "learning_rate": 9.581470816009242e-08, "loss": 1.3224, "step": 3116 }, { "epoch": 0.96, "learning_rate": 9.444320471710732e-08, "loss": 1.4677, "step": 3117 }, { "epoch": 0.96, "learning_rate": 9.308154152530057e-08, "loss": 1.2001, "step": 3118 }, { "epoch": 0.96, "learning_rate": 9.172971993735902e-08, "loss": 1.4528, "step": 3119 }, { "epoch": 0.96, "learning_rate": 9.038774129619177e-08, "loss": 1.4274, "step": 3120 }, { "epoch": 0.96, "learning_rate": 8.90556069349302e-08, "loss": 2.4832, "step": 3121 }, { "epoch": 0.96, "learning_rate": 8.773331817692576e-08, "loss": 1.0365, "step": 3122 }, { "epoch": 0.96, "learning_rate": 8.64208763357488e-08, "loss": 1.4516, "step": 3123 }, { "epoch": 0.96, "learning_rate": 8.511828271518974e-08, "loss": 1.1619, "step": 3124 }, { "epoch": 0.96, "learning_rate": 8.382553860925458e-08, "loss": 1.3574, "step": 3125 }, { "epoch": 0.96, "learning_rate": 8.254264530216383e-08, "loss": 1.6379, "step": 3126 }, { "epoch": 0.96, "learning_rate": 8.12696040683525e-08, "loss": 1.6137, "step": 3127 }, { "epoch": 0.96, "learning_rate": 8.000641617247006e-08, "loss": 1.4404, "step": 3128 }, { "epoch": 0.96, "learning_rate": 7.8753082869375e-08, "loss": 1.3007, "step": 3129 }, { "epoch": 0.96, "learning_rate": 7.750960540413909e-08, "loss": 1.2122, "step": 3130 }, { "epoch": 0.96, "learning_rate": 7.627598501204092e-08, "loss": 1.3673, "step": 3131 }, { "epoch": 0.96, "learning_rate": 7.505222291856795e-08, "loss": 1.5685, "step": 3132 }, { "epoch": 0.96, "learning_rate": 7.383832033941551e-08, "loss": 1.3681, "step": 3133 }, { "epoch": 0.96, "learning_rate": 7.26342784804801e-08, "loss": 1.258, "step": 3134 }, { "epoch": 0.96, "learning_rate": 7.144009853786826e-08, "loss": 1.413, "step": 3135 }, { "epoch": 0.96, "learning_rate": 7.025578169788661e-08, "loss": 1.4445, "step": 3136 }, { "epoch": 0.97, "learning_rate": 6.908132913704291e-08, "loss": 1.3849, "step": 3137 }, { "epoch": 0.97, "learning_rate": 6.791674202204724e-08, "loss": 1.3473, "step": 3138 }, { "epoch": 0.97, "learning_rate": 6.67620215098086e-08, "loss": 1.1221, "step": 3139 }, { "epoch": 0.97, "learning_rate": 6.561716874743496e-08, "loss": 1.4088, "step": 3140 }, { "epoch": 0.97, "learning_rate": 6.448218487223101e-08, "loss": 1.3902, "step": 3141 }, { "epoch": 0.97, "learning_rate": 6.33570710116982e-08, "loss": 1.7441, "step": 3142 }, { "epoch": 0.97, "learning_rate": 6.224182828353243e-08, "loss": 1.5017, "step": 3143 }, { "epoch": 0.97, "learning_rate": 6.113645779562528e-08, "loss": 1.0808, "step": 3144 }, { "epoch": 0.97, "learning_rate": 6.004096064605613e-08, "loss": 1.5707, "step": 3145 }, { "epoch": 0.97, "learning_rate": 5.8955337923103374e-08, "loss": 1.1212, "step": 3146 }, { "epoch": 0.97, "learning_rate": 5.787959070523097e-08, "loss": 1.5047, "step": 3147 }, { "epoch": 0.97, "learning_rate": 5.681372006109298e-08, "loss": 1.2626, "step": 3148 }, { "epoch": 0.97, "learning_rate": 5.575772704953464e-08, "loss": 1.435, "step": 3149 }, { "epoch": 0.97, "learning_rate": 5.4711612719586804e-08, "loss": 1.3504, "step": 3150 }, { "epoch": 0.97, "learning_rate": 5.367537811046486e-08, "loss": 1.3287, "step": 3151 }, { "epoch": 0.97, "learning_rate": 5.2649024251574256e-08, "loss": 2.6772, "step": 3152 }, { "epoch": 0.97, "learning_rate": 5.163255216250162e-08, "loss": 1.3951, "step": 3153 }, { "epoch": 0.97, "learning_rate": 5.0625962853015906e-08, "loss": 1.2649, "step": 3154 }, { "epoch": 0.97, "learning_rate": 4.962925732307278e-08, "loss": 1.4486, "step": 3155 }, { "epoch": 0.97, "learning_rate": 4.864243656280465e-08, "loss": 1.6975, "step": 3156 }, { "epoch": 0.97, "learning_rate": 4.766550155252736e-08, "loss": 1.4862, "step": 3157 }, { "epoch": 0.97, "learning_rate": 4.6698453262734587e-08, "loss": 1.2894, "step": 3158 }, { "epoch": 0.97, "learning_rate": 4.57412926541001e-08, "loss": 1.3841, "step": 3159 }, { "epoch": 0.97, "learning_rate": 4.4794020677473293e-08, "loss": 1.2212, "step": 3160 }, { "epoch": 0.97, "learning_rate": 4.385663827388031e-08, "loss": 1.1357, "step": 3161 }, { "epoch": 0.97, "learning_rate": 4.2929146374524055e-08, "loss": 1.4125, "step": 3162 }, { "epoch": 0.97, "learning_rate": 4.2011545900781936e-08, "loss": 1.7752, "step": 3163 }, { "epoch": 0.97, "learning_rate": 4.11038377642059e-08, "loss": 1.6089, "step": 3164 }, { "epoch": 0.97, "learning_rate": 4.020602286651687e-08, "loss": 1.3035, "step": 3165 }, { "epoch": 0.97, "learning_rate": 3.931810209961362e-08, "loss": 2.1204, "step": 3166 }, { "epoch": 0.97, "learning_rate": 3.8440076345561685e-08, "loss": 1.1149, "step": 3167 }, { "epoch": 0.97, "learning_rate": 3.757194647659779e-08, "loss": 1.1003, "step": 3168 }, { "epoch": 0.98, "learning_rate": 3.6713713355127634e-08, "loss": 1.5532, "step": 3169 }, { "epoch": 0.98, "learning_rate": 3.586537783372812e-08, "loss": 1.4656, "step": 3170 }, { "epoch": 0.98, "learning_rate": 3.502694075514179e-08, "loss": 1.3747, "step": 3171 }, { "epoch": 0.98, "learning_rate": 3.419840295227572e-08, "loss": 1.5061, "step": 3172 }, { "epoch": 0.98, "learning_rate": 3.3379765248207077e-08, "loss": 1.3496, "step": 3173 }, { "epoch": 0.98, "learning_rate": 3.2571028456177545e-08, "loss": 1.2043, "step": 3174 }, { "epoch": 0.98, "learning_rate": 3.177219337958892e-08, "loss": 1.4857, "step": 3175 }, { "epoch": 0.98, "learning_rate": 3.098326081201308e-08, "loss": 1.4735, "step": 3176 }, { "epoch": 0.98, "learning_rate": 3.020423153717866e-08, "loss": 1.8275, "step": 3177 }, { "epoch": 0.98, "learning_rate": 2.9435106328979946e-08, "loss": 1.228, "step": 3178 }, { "epoch": 0.98, "learning_rate": 2.8675885951472415e-08, "loss": 1.3292, "step": 3179 }, { "epoch": 0.98, "learning_rate": 2.792657115886832e-08, "loss": 1.3782, "step": 3180 }, { "epoch": 0.98, "learning_rate": 2.7187162695545554e-08, "loss": 1.6479, "step": 3181 }, { "epoch": 0.98, "learning_rate": 2.6457661296035443e-08, "loss": 1.4225, "step": 3182 }, { "epoch": 0.98, "learning_rate": 2.573806768503162e-08, "loss": 1.2893, "step": 3183 }, { "epoch": 0.98, "learning_rate": 2.5028382577382272e-08, "loss": 1.2732, "step": 3184 }, { "epoch": 0.98, "learning_rate": 2.4328606678093446e-08, "loss": 1.147, "step": 3185 }, { "epoch": 0.98, "learning_rate": 2.3638740682329075e-08, "loss": 1.1796, "step": 3186 }, { "epoch": 0.98, "learning_rate": 2.2958785275406514e-08, "loss": 1.2939, "step": 3187 }, { "epoch": 0.98, "learning_rate": 2.2288741132796554e-08, "loss": 1.6189, "step": 3188 }, { "epoch": 0.98, "learning_rate": 2.162860892012786e-08, "loss": 1.3985, "step": 3189 }, { "epoch": 0.98, "learning_rate": 2.09783892931803e-08, "loss": 1.1648, "step": 3190 }, { "epoch": 0.98, "learning_rate": 2.033808289788608e-08, "loss": 1.3692, "step": 3191 }, { "epoch": 0.98, "learning_rate": 1.9707690370330822e-08, "loss": 1.3604, "step": 3192 }, { "epoch": 0.98, "learning_rate": 1.908721233675026e-08, "loss": 1.4451, "step": 3193 }, { "epoch": 0.98, "learning_rate": 1.8476649413531333e-08, "loss": 1.3557, "step": 3194 }, { "epoch": 0.98, "learning_rate": 1.7876002207213304e-08, "loss": 1.4407, "step": 3195 }, { "epoch": 0.98, "learning_rate": 1.72852713144811e-08, "loss": 1.0455, "step": 3196 }, { "epoch": 0.98, "learning_rate": 1.6704457322173073e-08, "loss": 1.5247, "step": 3197 }, { "epoch": 0.98, "learning_rate": 1.6133560807274352e-08, "loss": 1.2941, "step": 3198 }, { "epoch": 0.98, "learning_rate": 1.557258233691572e-08, "loss": 1.1431, "step": 3199 }, { "epoch": 0.98, "learning_rate": 1.502152246837696e-08, "loss": 1.6497, "step": 3200 }, { "epoch": 0.98, "learning_rate": 1.4480381749087945e-08, "loss": 1.2585, "step": 3201 }, { "epoch": 0.99, "learning_rate": 1.394916071661867e-08, "loss": 1.4078, "step": 3202 }, { "epoch": 0.99, "learning_rate": 1.3427859898689221e-08, "loss": 1.5193, "step": 3203 }, { "epoch": 0.99, "learning_rate": 1.2916479813163129e-08, "loss": 1.568, "step": 3204 }, { "epoch": 0.99, "learning_rate": 1.2415020968049585e-08, "loss": 1.2907, "step": 3205 }, { "epoch": 0.99, "learning_rate": 1.1923483861502327e-08, "loss": 1.2634, "step": 3206 }, { "epoch": 0.99, "learning_rate": 1.1441868981815207e-08, "loss": 1.5697, "step": 3207 }, { "epoch": 0.99, "learning_rate": 1.0970176807428845e-08, "loss": 1.7897, "step": 3208 }, { "epoch": 0.99, "learning_rate": 1.0508407806927302e-08, "loss": 1.3133, "step": 3209 }, { "epoch": 0.99, "learning_rate": 1.0056562439033635e-08, "loss": 1.4996, "step": 3210 }, { "epoch": 0.99, "learning_rate": 9.614641152615457e-09, "loss": 1.4123, "step": 3211 }, { "epoch": 0.99, "learning_rate": 9.182644386678264e-09, "loss": 1.1009, "step": 3212 }, { "epoch": 0.99, "learning_rate": 8.760572570373215e-09, "loss": 1.4058, "step": 3213 }, { "epoch": 0.99, "learning_rate": 8.34842612298825e-09, "loss": 1.1655, "step": 3214 }, { "epoch": 0.99, "learning_rate": 7.946205453953638e-09, "loss": 1.6747, "step": 3215 }, { "epoch": 0.99, "learning_rate": 7.553910962836419e-09, "loss": 1.6981, "step": 3216 }, { "epoch": 0.99, "learning_rate": 7.171543039345974e-09, "loss": 1.3558, "step": 3217 }, { "epoch": 0.99, "learning_rate": 6.799102063329566e-09, "loss": 1.7951, "step": 3218 }, { "epoch": 0.99, "learning_rate": 6.4365884047723486e-09, "loss": 2.03, "step": 3219 }, { "epoch": 0.99, "learning_rate": 6.084002423798474e-09, "loss": 1.4672, "step": 3220 }, { "epoch": 0.99, "learning_rate": 5.741344470667765e-09, "loss": 1.1903, "step": 3221 }, { "epoch": 0.99, "learning_rate": 5.40861488578126e-09, "loss": 1.4321, "step": 3222 }, { "epoch": 0.99, "learning_rate": 5.08581399967345e-09, "loss": 1.5656, "step": 3223 }, { "epoch": 0.99, "learning_rate": 4.772942133017821e-09, "loss": 1.4984, "step": 3224 }, { "epoch": 0.99, "learning_rate": 4.469999596622421e-09, "loss": 1.3962, "step": 3225 }, { "epoch": 0.99, "learning_rate": 4.176986691434292e-09, "loss": 1.5291, "step": 3226 }, { "epoch": 0.99, "learning_rate": 3.8939037085339305e-09, "loss": 1.1156, "step": 3227 }, { "epoch": 0.99, "learning_rate": 3.620750929136385e-09, "loss": 1.515, "step": 3228 }, { "epoch": 0.99, "learning_rate": 3.3575286245957073e-09, "loss": 1.3884, "step": 3229 }, { "epoch": 0.99, "learning_rate": 3.1042370563971745e-09, "loss": 1.6667, "step": 3230 }, { "epoch": 0.99, "learning_rate": 2.8608764761639542e-09, "loss": 1.3063, "step": 3231 }, { "epoch": 0.99, "learning_rate": 2.62744712565155e-09, "loss": 1.2913, "step": 3232 }, { "epoch": 0.99, "learning_rate": 2.403949236750025e-09, "loss": 1.7358, "step": 3233 }, { "epoch": 1.0, "learning_rate": 2.1903830314851102e-09, "loss": 1.4452, "step": 3234 }, { "epoch": 1.0, "learning_rate": 1.9867487220126546e-09, "loss": 1.4284, "step": 3235 }, { "epoch": 1.0, "learning_rate": 1.793046510626395e-09, "loss": 1.3071, "step": 3236 }, { "epoch": 1.0, "learning_rate": 1.609276589750186e-09, "loss": 1.2222, "step": 3237 }, { "epoch": 1.0, "learning_rate": 1.4354391419435509e-09, "loss": 1.435, "step": 3238 }, { "epoch": 1.0, "learning_rate": 1.2715343398972401e-09, "loss": 1.2376, "step": 3239 }, { "epoch": 1.0, "learning_rate": 1.1175623464354524e-09, "loss": 1.078, "step": 3240 }, { "epoch": 1.0, "learning_rate": 9.73523314515834e-10, "loss": 1.6969, "step": 3241 }, { "epoch": 1.0, "learning_rate": 8.394173872261491e-10, "loss": 1.4551, "step": 3242 }, { "epoch": 1.0, "learning_rate": 7.152446977898297e-10, "loss": 1.3146, "step": 3243 }, { "epoch": 1.0, "learning_rate": 6.010053695604257e-10, "loss": 1.5375, "step": 3244 }, { "epoch": 1.0, "learning_rate": 4.966995160238242e-10, "loss": 1.6633, "step": 3245 }, { "epoch": 1.0, "learning_rate": 4.0232724079825034e-10, "loss": 1.7201, "step": 3246 }, { "epoch": 1.0, "learning_rate": 3.178886376331569e-10, "loss": 1.6338, "step": 3247 }, { "epoch": 1.0, "learning_rate": 2.4338379041255464e-10, "loss": 1.7286, "step": 3248 }, { "epoch": 1.0, "learning_rate": 1.7881277314835132e-10, "loss": 1.1633, "step": 3249 }, { "epoch": 1.0, "learning_rate": 1.2417564998590258e-10, "loss": 1.4364, "step": 3250 }, { "epoch": 1.0, "step": 3250, "total_flos": 3.998901774948762e+16, "train_loss": 1.5109170598616968, "train_runtime": 3947.9032, "train_samples_per_second": 13.172, "train_steps_per_second": 0.823 } ], "max_steps": 3250, "num_train_epochs": 1, "total_flos": 3.998901774948762e+16, "trial_name": null, "trial_params": null }