{ "best_metric": null, "best_model_checkpoint": null, "epoch": 13.089005235602095, "global_step": 10000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 5e-05, "loss": 2.493, "step": 1 }, { "epoch": 0.0, "learning_rate": 5e-05, "loss": 1.9244, "step": 2 }, { "epoch": 0.0, "learning_rate": 4.9995000000000005e-05, "loss": 1.9557, "step": 3 }, { "epoch": 0.01, "learning_rate": 4.999e-05, "loss": 1.5277, "step": 4 }, { "epoch": 0.01, "learning_rate": 4.9985e-05, "loss": 1.4193, "step": 5 }, { "epoch": 0.01, "learning_rate": 4.9980000000000006e-05, "loss": 1.3539, "step": 6 }, { "epoch": 0.01, "learning_rate": 4.9975e-05, "loss": 1.3051, "step": 7 }, { "epoch": 0.01, "learning_rate": 4.997e-05, "loss": 1.1617, "step": 8 }, { "epoch": 0.01, "learning_rate": 4.9965e-05, "loss": 1.2347, "step": 9 }, { "epoch": 0.01, "learning_rate": 4.996e-05, "loss": 1.5691, "step": 10 }, { "epoch": 0.01, "learning_rate": 4.9955e-05, "loss": 1.29, "step": 11 }, { "epoch": 0.02, "learning_rate": 4.995e-05, "loss": 1.2394, "step": 12 }, { "epoch": 0.02, "learning_rate": 4.9945000000000004e-05, "loss": 1.5466, "step": 13 }, { "epoch": 0.02, "learning_rate": 4.9940000000000006e-05, "loss": 1.1706, "step": 14 }, { "epoch": 0.02, "learning_rate": 4.9935e-05, "loss": 1.1086, "step": 15 }, { "epoch": 0.02, "learning_rate": 4.9930000000000005e-05, "loss": 1.2341, "step": 16 }, { "epoch": 0.02, "learning_rate": 4.992500000000001e-05, "loss": 1.1465, "step": 17 }, { "epoch": 0.02, "learning_rate": 4.992e-05, "loss": 1.262, "step": 18 }, { "epoch": 0.02, "learning_rate": 4.9915e-05, "loss": 1.062, "step": 19 }, { "epoch": 0.03, "learning_rate": 4.991e-05, "loss": 1.0916, "step": 20 }, { "epoch": 0.03, "learning_rate": 4.9905000000000004e-05, "loss": 1.204, "step": 21 }, { "epoch": 0.03, "learning_rate": 4.99e-05, "loss": 0.9764, "step": 22 }, { "epoch": 0.03, "learning_rate": 4.9895e-05, "loss": 1.1547, "step": 23 }, { "epoch": 0.03, "learning_rate": 4.9890000000000005e-05, "loss": 1.332, "step": 24 }, { "epoch": 0.03, "learning_rate": 4.9885e-05, "loss": 1.3518, "step": 25 }, { "epoch": 0.03, "learning_rate": 4.9880000000000004e-05, "loss": 0.981, "step": 26 }, { "epoch": 0.04, "learning_rate": 4.9875000000000006e-05, "loss": 1.1544, "step": 27 }, { "epoch": 0.04, "learning_rate": 4.987e-05, "loss": 1.0799, "step": 28 }, { "epoch": 0.04, "learning_rate": 4.9865e-05, "loss": 1.133, "step": 29 }, { "epoch": 0.04, "learning_rate": 4.986e-05, "loss": 1.2719, "step": 30 }, { "epoch": 0.04, "learning_rate": 4.9855e-05, "loss": 1.1999, "step": 31 }, { "epoch": 0.04, "learning_rate": 4.9850000000000006e-05, "loss": 1.0432, "step": 32 }, { "epoch": 0.04, "learning_rate": 4.9845e-05, "loss": 1.1838, "step": 33 }, { "epoch": 0.04, "learning_rate": 4.9840000000000004e-05, "loss": 1.1303, "step": 34 }, { "epoch": 0.05, "learning_rate": 4.9835000000000007e-05, "loss": 1.0491, "step": 35 }, { "epoch": 0.05, "learning_rate": 4.983e-05, "loss": 1.2789, "step": 36 }, { "epoch": 0.05, "learning_rate": 4.9825000000000005e-05, "loss": 0.8952, "step": 37 }, { "epoch": 0.05, "learning_rate": 4.982e-05, "loss": 1.0015, "step": 38 }, { "epoch": 0.05, "learning_rate": 4.9815e-05, "loss": 1.1024, "step": 39 }, { "epoch": 0.05, "learning_rate": 4.981e-05, "loss": 1.435, "step": 40 }, { "epoch": 0.05, "learning_rate": 4.9805e-05, "loss": 1.1639, "step": 41 }, { "epoch": 0.05, "learning_rate": 4.9800000000000004e-05, "loss": 1.1886, "step": 42 }, { "epoch": 0.06, "learning_rate": 4.9795e-05, "loss": 1.0329, "step": 43 }, { "epoch": 0.06, "learning_rate": 4.979e-05, "loss": 1.1902, "step": 44 }, { "epoch": 0.06, "learning_rate": 4.9785000000000005e-05, "loss": 1.0498, "step": 45 }, { "epoch": 0.06, "learning_rate": 4.978e-05, "loss": 1.4877, "step": 46 }, { "epoch": 0.06, "learning_rate": 4.9775000000000004e-05, "loss": 1.0838, "step": 47 }, { "epoch": 0.06, "learning_rate": 4.977e-05, "loss": 1.196, "step": 48 }, { "epoch": 0.06, "learning_rate": 4.9765e-05, "loss": 0.9751, "step": 49 }, { "epoch": 0.07, "learning_rate": 4.976e-05, "loss": 1.1618, "step": 50 }, { "epoch": 0.07, "learning_rate": 4.9755e-05, "loss": 1.0687, "step": 51 }, { "epoch": 0.07, "learning_rate": 4.975e-05, "loss": 1.0327, "step": 52 }, { "epoch": 0.07, "learning_rate": 4.9745000000000006e-05, "loss": 1.3049, "step": 53 }, { "epoch": 0.07, "learning_rate": 4.974e-05, "loss": 1.1296, "step": 54 }, { "epoch": 0.07, "learning_rate": 4.9735000000000004e-05, "loss": 1.2892, "step": 55 }, { "epoch": 0.07, "learning_rate": 4.973000000000001e-05, "loss": 0.9039, "step": 56 }, { "epoch": 0.07, "learning_rate": 4.9725e-05, "loss": 1.0778, "step": 57 }, { "epoch": 0.08, "learning_rate": 4.972e-05, "loss": 1.0511, "step": 58 }, { "epoch": 0.08, "learning_rate": 4.9715e-05, "loss": 1.3157, "step": 59 }, { "epoch": 0.08, "learning_rate": 4.9710000000000003e-05, "loss": 0.934, "step": 60 }, { "epoch": 0.08, "learning_rate": 4.9705e-05, "loss": 0.9479, "step": 61 }, { "epoch": 0.08, "learning_rate": 4.97e-05, "loss": 1.1344, "step": 62 }, { "epoch": 0.08, "learning_rate": 4.9695000000000004e-05, "loss": 1.1477, "step": 63 }, { "epoch": 0.08, "learning_rate": 4.969e-05, "loss": 1.2556, "step": 64 }, { "epoch": 0.09, "learning_rate": 4.9685e-05, "loss": 1.0411, "step": 65 }, { "epoch": 0.09, "learning_rate": 4.9680000000000005e-05, "loss": 1.0563, "step": 66 }, { "epoch": 0.09, "learning_rate": 4.967500000000001e-05, "loss": 1.1571, "step": 67 }, { "epoch": 0.09, "learning_rate": 4.967e-05, "loss": 1.0354, "step": 68 }, { "epoch": 0.09, "learning_rate": 4.9665e-05, "loss": 1.0107, "step": 69 }, { "epoch": 0.09, "learning_rate": 4.966e-05, "loss": 0.9611, "step": 70 }, { "epoch": 0.09, "learning_rate": 4.9655000000000005e-05, "loss": 1.0534, "step": 71 }, { "epoch": 0.09, "learning_rate": 4.965e-05, "loss": 1.133, "step": 72 }, { "epoch": 0.1, "learning_rate": 4.9645e-05, "loss": 1.0493, "step": 73 }, { "epoch": 0.1, "learning_rate": 4.9640000000000006e-05, "loss": 1.2378, "step": 74 }, { "epoch": 0.1, "learning_rate": 4.9635e-05, "loss": 1.0439, "step": 75 }, { "epoch": 0.1, "learning_rate": 4.9630000000000004e-05, "loss": 0.8049, "step": 76 }, { "epoch": 0.1, "learning_rate": 4.962500000000001e-05, "loss": 1.027, "step": 77 }, { "epoch": 0.1, "learning_rate": 4.962e-05, "loss": 1.0521, "step": 78 }, { "epoch": 0.1, "learning_rate": 4.9615e-05, "loss": 1.1221, "step": 79 }, { "epoch": 0.1, "learning_rate": 4.961e-05, "loss": 1.0288, "step": 80 }, { "epoch": 0.11, "learning_rate": 4.9605000000000004e-05, "loss": 1.0487, "step": 81 }, { "epoch": 0.11, "learning_rate": 4.96e-05, "loss": 1.0576, "step": 82 }, { "epoch": 0.11, "learning_rate": 4.9595e-05, "loss": 1.1094, "step": 83 }, { "epoch": 0.11, "learning_rate": 4.9590000000000005e-05, "loss": 1.0504, "step": 84 }, { "epoch": 0.11, "learning_rate": 4.9585e-05, "loss": 1.0848, "step": 85 }, { "epoch": 0.11, "learning_rate": 4.958e-05, "loss": 1.1061, "step": 86 }, { "epoch": 0.11, "learning_rate": 4.9575000000000006e-05, "loss": 0.8237, "step": 87 }, { "epoch": 0.12, "learning_rate": 4.957e-05, "loss": 0.9871, "step": 88 }, { "epoch": 0.12, "learning_rate": 4.9565e-05, "loss": 1.1254, "step": 89 }, { "epoch": 0.12, "learning_rate": 4.956e-05, "loss": 1.0392, "step": 90 }, { "epoch": 0.12, "learning_rate": 4.9555e-05, "loss": 0.9964, "step": 91 }, { "epoch": 0.12, "learning_rate": 4.9550000000000005e-05, "loss": 0.8928, "step": 92 }, { "epoch": 0.12, "learning_rate": 4.9545e-05, "loss": 1.0601, "step": 93 }, { "epoch": 0.12, "learning_rate": 4.9540000000000003e-05, "loss": 0.9606, "step": 94 }, { "epoch": 0.12, "learning_rate": 4.9535000000000006e-05, "loss": 1.0289, "step": 95 }, { "epoch": 0.13, "learning_rate": 4.953e-05, "loss": 0.9941, "step": 96 }, { "epoch": 0.13, "learning_rate": 4.9525000000000004e-05, "loss": 1.1008, "step": 97 }, { "epoch": 0.13, "learning_rate": 4.952e-05, "loss": 1.1122, "step": 98 }, { "epoch": 0.13, "learning_rate": 4.9515e-05, "loss": 0.8777, "step": 99 }, { "epoch": 0.13, "learning_rate": 4.951e-05, "loss": 1.3704, "step": 100 }, { "epoch": 0.13, "learning_rate": 4.9505e-05, "loss": 0.9593, "step": 101 }, { "epoch": 0.13, "learning_rate": 4.9500000000000004e-05, "loss": 1.3386, "step": 102 }, { "epoch": 0.13, "learning_rate": 4.9495e-05, "loss": 1.0847, "step": 103 }, { "epoch": 0.14, "learning_rate": 4.949e-05, "loss": 0.9998, "step": 104 }, { "epoch": 0.14, "learning_rate": 4.9485000000000005e-05, "loss": 1.0587, "step": 105 }, { "epoch": 0.14, "learning_rate": 4.948000000000001e-05, "loss": 0.8735, "step": 106 }, { "epoch": 0.14, "learning_rate": 4.9475e-05, "loss": 0.9792, "step": 107 }, { "epoch": 0.14, "learning_rate": 4.947e-05, "loss": 1.1384, "step": 108 }, { "epoch": 0.14, "learning_rate": 4.9465e-05, "loss": 1.162, "step": 109 }, { "epoch": 0.14, "learning_rate": 4.946e-05, "loss": 0.9491, "step": 110 }, { "epoch": 0.15, "learning_rate": 4.9455e-05, "loss": 1.1418, "step": 111 }, { "epoch": 0.15, "learning_rate": 4.945e-05, "loss": 1.0659, "step": 112 }, { "epoch": 0.15, "learning_rate": 4.9445000000000005e-05, "loss": 0.9557, "step": 113 }, { "epoch": 0.15, "learning_rate": 4.944e-05, "loss": 1.2053, "step": 114 }, { "epoch": 0.15, "learning_rate": 4.9435000000000004e-05, "loss": 1.1764, "step": 115 }, { "epoch": 0.15, "learning_rate": 4.9430000000000006e-05, "loss": 0.9042, "step": 116 }, { "epoch": 0.15, "learning_rate": 4.9425e-05, "loss": 0.7959, "step": 117 }, { "epoch": 0.15, "learning_rate": 4.942e-05, "loss": 1.17, "step": 118 }, { "epoch": 0.16, "learning_rate": 4.9415e-05, "loss": 0.9732, "step": 119 }, { "epoch": 0.16, "learning_rate": 4.941e-05, "loss": 0.9346, "step": 120 }, { "epoch": 0.16, "learning_rate": 4.9405e-05, "loss": 1.1937, "step": 121 }, { "epoch": 0.16, "learning_rate": 4.94e-05, "loss": 1.0677, "step": 122 }, { "epoch": 0.16, "learning_rate": 4.9395000000000004e-05, "loss": 1.0693, "step": 123 }, { "epoch": 0.16, "learning_rate": 4.939e-05, "loss": 1.254, "step": 124 }, { "epoch": 0.16, "learning_rate": 4.9385e-05, "loss": 1.0157, "step": 125 }, { "epoch": 0.16, "learning_rate": 4.9380000000000005e-05, "loss": 1.3604, "step": 126 }, { "epoch": 0.17, "learning_rate": 4.937500000000001e-05, "loss": 0.9622, "step": 127 }, { "epoch": 0.17, "learning_rate": 4.937e-05, "loss": 0.9459, "step": 128 }, { "epoch": 0.17, "learning_rate": 4.9365e-05, "loss": 0.9864, "step": 129 }, { "epoch": 0.17, "learning_rate": 4.936e-05, "loss": 1.05, "step": 130 }, { "epoch": 0.17, "learning_rate": 4.9355000000000004e-05, "loss": 1.1985, "step": 131 }, { "epoch": 0.17, "learning_rate": 4.935e-05, "loss": 1.0503, "step": 132 }, { "epoch": 0.17, "learning_rate": 4.9345e-05, "loss": 0.9725, "step": 133 }, { "epoch": 0.18, "learning_rate": 4.9340000000000005e-05, "loss": 1.1097, "step": 134 }, { "epoch": 0.18, "learning_rate": 4.9335e-05, "loss": 1.0361, "step": 135 }, { "epoch": 0.18, "learning_rate": 4.9330000000000004e-05, "loss": 0.8907, "step": 136 }, { "epoch": 0.18, "learning_rate": 4.9325000000000006e-05, "loss": 1.2129, "step": 137 }, { "epoch": 0.18, "learning_rate": 4.932e-05, "loss": 1.176, "step": 138 }, { "epoch": 0.18, "learning_rate": 4.9315e-05, "loss": 0.884, "step": 139 }, { "epoch": 0.18, "learning_rate": 4.931e-05, "loss": 0.952, "step": 140 }, { "epoch": 0.18, "learning_rate": 4.9305e-05, "loss": 0.9539, "step": 141 }, { "epoch": 0.19, "learning_rate": 4.93e-05, "loss": 1.0538, "step": 142 }, { "epoch": 0.19, "learning_rate": 4.9295e-05, "loss": 1.1323, "step": 143 }, { "epoch": 0.19, "learning_rate": 4.9290000000000004e-05, "loss": 0.897, "step": 144 }, { "epoch": 0.19, "learning_rate": 4.928500000000001e-05, "loss": 0.9171, "step": 145 }, { "epoch": 0.19, "learning_rate": 4.928e-05, "loss": 0.8657, "step": 146 }, { "epoch": 0.19, "learning_rate": 4.9275000000000005e-05, "loss": 1.1105, "step": 147 }, { "epoch": 0.19, "learning_rate": 4.927000000000001e-05, "loss": 1.116, "step": 148 }, { "epoch": 0.2, "learning_rate": 4.9265e-05, "loss": 1.194, "step": 149 }, { "epoch": 0.2, "learning_rate": 4.926e-05, "loss": 1.0446, "step": 150 }, { "epoch": 0.2, "learning_rate": 4.9255e-05, "loss": 0.9883, "step": 151 }, { "epoch": 0.2, "learning_rate": 4.9250000000000004e-05, "loss": 1.3495, "step": 152 }, { "epoch": 0.2, "learning_rate": 4.9245e-05, "loss": 1.1237, "step": 153 }, { "epoch": 0.2, "learning_rate": 4.924e-05, "loss": 1.0164, "step": 154 }, { "epoch": 0.2, "learning_rate": 4.9235000000000005e-05, "loss": 0.9186, "step": 155 }, { "epoch": 0.2, "learning_rate": 4.923e-05, "loss": 0.924, "step": 156 }, { "epoch": 0.21, "learning_rate": 4.9225000000000004e-05, "loss": 0.9716, "step": 157 }, { "epoch": 0.21, "learning_rate": 4.9220000000000006e-05, "loss": 1.0748, "step": 158 }, { "epoch": 0.21, "learning_rate": 4.9215e-05, "loss": 1.2057, "step": 159 }, { "epoch": 0.21, "learning_rate": 4.921e-05, "loss": 1.0141, "step": 160 }, { "epoch": 0.21, "learning_rate": 4.9205e-05, "loss": 1.3414, "step": 161 }, { "epoch": 0.21, "learning_rate": 4.92e-05, "loss": 1.1044, "step": 162 }, { "epoch": 0.21, "learning_rate": 4.9195e-05, "loss": 0.8931, "step": 163 }, { "epoch": 0.21, "learning_rate": 4.919e-05, "loss": 1.0664, "step": 164 }, { "epoch": 0.22, "learning_rate": 4.9185000000000004e-05, "loss": 1.0023, "step": 165 }, { "epoch": 0.22, "learning_rate": 4.918000000000001e-05, "loss": 0.8324, "step": 166 }, { "epoch": 0.22, "learning_rate": 4.9175e-05, "loss": 0.9214, "step": 167 }, { "epoch": 0.22, "learning_rate": 4.9170000000000005e-05, "loss": 1.0435, "step": 168 }, { "epoch": 0.22, "learning_rate": 4.9165e-05, "loss": 1.092, "step": 169 }, { "epoch": 0.22, "learning_rate": 4.9160000000000004e-05, "loss": 0.9049, "step": 170 }, { "epoch": 0.22, "learning_rate": 4.9155e-05, "loss": 0.9983, "step": 171 }, { "epoch": 0.23, "learning_rate": 4.915e-05, "loss": 1.0073, "step": 172 }, { "epoch": 0.23, "learning_rate": 4.9145000000000005e-05, "loss": 0.9658, "step": 173 }, { "epoch": 0.23, "learning_rate": 4.914e-05, "loss": 0.8826, "step": 174 }, { "epoch": 0.23, "learning_rate": 4.9135e-05, "loss": 0.8847, "step": 175 }, { "epoch": 0.23, "learning_rate": 4.9130000000000006e-05, "loss": 0.8738, "step": 176 }, { "epoch": 0.23, "learning_rate": 4.9125e-05, "loss": 0.9554, "step": 177 }, { "epoch": 0.23, "learning_rate": 4.9120000000000004e-05, "loss": 1.142, "step": 178 }, { "epoch": 0.23, "learning_rate": 4.9115e-05, "loss": 1.0584, "step": 179 }, { "epoch": 0.24, "learning_rate": 4.911e-05, "loss": 0.9727, "step": 180 }, { "epoch": 0.24, "learning_rate": 4.9105e-05, "loss": 0.859, "step": 181 }, { "epoch": 0.24, "learning_rate": 4.91e-05, "loss": 1.0289, "step": 182 }, { "epoch": 0.24, "learning_rate": 4.9095000000000003e-05, "loss": 1.002, "step": 183 }, { "epoch": 0.24, "learning_rate": 4.9090000000000006e-05, "loss": 0.7977, "step": 184 }, { "epoch": 0.24, "learning_rate": 4.9085e-05, "loss": 0.8837, "step": 185 }, { "epoch": 0.24, "learning_rate": 4.9080000000000004e-05, "loss": 0.8547, "step": 186 }, { "epoch": 0.24, "learning_rate": 4.907500000000001e-05, "loss": 0.9382, "step": 187 }, { "epoch": 0.25, "learning_rate": 4.907e-05, "loss": 0.7546, "step": 188 }, { "epoch": 0.25, "learning_rate": 4.9065e-05, "loss": 1.3638, "step": 189 }, { "epoch": 0.25, "learning_rate": 4.906e-05, "loss": 1.0635, "step": 190 }, { "epoch": 0.25, "learning_rate": 4.9055000000000004e-05, "loss": 0.8257, "step": 191 }, { "epoch": 0.25, "learning_rate": 4.905e-05, "loss": 1.0229, "step": 192 }, { "epoch": 0.25, "learning_rate": 4.9045e-05, "loss": 1.0845, "step": 193 }, { "epoch": 0.25, "learning_rate": 4.9040000000000005e-05, "loss": 1.3163, "step": 194 }, { "epoch": 0.26, "learning_rate": 4.9035e-05, "loss": 1.1227, "step": 195 }, { "epoch": 0.26, "learning_rate": 4.903e-05, "loss": 1.0605, "step": 196 }, { "epoch": 0.26, "learning_rate": 4.9025000000000006e-05, "loss": 1.1831, "step": 197 }, { "epoch": 0.26, "learning_rate": 4.902e-05, "loss": 0.9807, "step": 198 }, { "epoch": 0.26, "learning_rate": 4.9015e-05, "loss": 0.8634, "step": 199 }, { "epoch": 0.26, "learning_rate": 4.901e-05, "loss": 1.0927, "step": 200 }, { "epoch": 0.26, "learning_rate": 4.9005e-05, "loss": 1.1827, "step": 201 }, { "epoch": 0.26, "learning_rate": 4.9e-05, "loss": 0.9196, "step": 202 }, { "epoch": 0.27, "learning_rate": 4.8995e-05, "loss": 0.848, "step": 203 }, { "epoch": 0.27, "learning_rate": 4.8990000000000004e-05, "loss": 1.0508, "step": 204 }, { "epoch": 0.27, "learning_rate": 4.8985000000000006e-05, "loss": 0.8686, "step": 205 }, { "epoch": 0.27, "learning_rate": 4.898e-05, "loss": 1.1881, "step": 206 }, { "epoch": 0.27, "learning_rate": 4.8975000000000005e-05, "loss": 0.8658, "step": 207 }, { "epoch": 0.27, "learning_rate": 4.897000000000001e-05, "loss": 0.9881, "step": 208 }, { "epoch": 0.27, "learning_rate": 4.8965e-05, "loss": 1.0028, "step": 209 }, { "epoch": 0.27, "learning_rate": 4.896e-05, "loss": 1.0389, "step": 210 }, { "epoch": 0.28, "learning_rate": 4.8955e-05, "loss": 0.9998, "step": 211 }, { "epoch": 0.28, "learning_rate": 4.8950000000000004e-05, "loss": 0.8375, "step": 212 }, { "epoch": 0.28, "learning_rate": 4.8945e-05, "loss": 1.0385, "step": 213 }, { "epoch": 0.28, "learning_rate": 4.894e-05, "loss": 0.8586, "step": 214 }, { "epoch": 0.28, "learning_rate": 4.8935000000000005e-05, "loss": 1.0923, "step": 215 }, { "epoch": 0.28, "learning_rate": 4.893e-05, "loss": 0.8516, "step": 216 }, { "epoch": 0.28, "learning_rate": 4.8925e-05, "loss": 1.2719, "step": 217 }, { "epoch": 0.29, "learning_rate": 4.8920000000000006e-05, "loss": 0.8818, "step": 218 }, { "epoch": 0.29, "learning_rate": 4.8915e-05, "loss": 1.1437, "step": 219 }, { "epoch": 0.29, "learning_rate": 4.891e-05, "loss": 0.8672, "step": 220 }, { "epoch": 0.29, "learning_rate": 4.8905e-05, "loss": 1.1917, "step": 221 }, { "epoch": 0.29, "learning_rate": 4.89e-05, "loss": 1.0064, "step": 222 }, { "epoch": 0.29, "learning_rate": 4.8895e-05, "loss": 1.1262, "step": 223 }, { "epoch": 0.29, "learning_rate": 4.889e-05, "loss": 0.9609, "step": 224 }, { "epoch": 0.29, "learning_rate": 4.8885000000000004e-05, "loss": 0.9387, "step": 225 }, { "epoch": 0.3, "learning_rate": 4.8880000000000006e-05, "loss": 1.4603, "step": 226 }, { "epoch": 0.3, "learning_rate": 4.8875e-05, "loss": 1.1638, "step": 227 }, { "epoch": 0.3, "learning_rate": 4.8870000000000005e-05, "loss": 1.1136, "step": 228 }, { "epoch": 0.3, "learning_rate": 4.8865e-05, "loss": 1.0265, "step": 229 }, { "epoch": 0.3, "learning_rate": 4.886e-05, "loss": 0.8488, "step": 230 }, { "epoch": 0.3, "learning_rate": 4.8855e-05, "loss": 1.0038, "step": 231 }, { "epoch": 0.3, "learning_rate": 4.885e-05, "loss": 1.0268, "step": 232 }, { "epoch": 0.3, "learning_rate": 4.8845000000000004e-05, "loss": 1.226, "step": 233 }, { "epoch": 0.31, "learning_rate": 4.884e-05, "loss": 0.9708, "step": 234 }, { "epoch": 0.31, "learning_rate": 4.8835e-05, "loss": 0.9059, "step": 235 }, { "epoch": 0.31, "learning_rate": 4.8830000000000005e-05, "loss": 1.1241, "step": 236 }, { "epoch": 0.31, "learning_rate": 4.8825e-05, "loss": 0.9567, "step": 237 }, { "epoch": 0.31, "learning_rate": 4.8820000000000004e-05, "loss": 1.1141, "step": 238 }, { "epoch": 0.31, "learning_rate": 4.8815e-05, "loss": 1.3872, "step": 239 }, { "epoch": 0.31, "learning_rate": 4.881e-05, "loss": 0.8183, "step": 240 }, { "epoch": 0.32, "learning_rate": 4.8805e-05, "loss": 0.9641, "step": 241 }, { "epoch": 0.32, "learning_rate": 4.88e-05, "loss": 1.0737, "step": 242 }, { "epoch": 0.32, "learning_rate": 4.8795e-05, "loss": 1.071, "step": 243 }, { "epoch": 0.32, "learning_rate": 4.8790000000000006e-05, "loss": 0.9886, "step": 244 }, { "epoch": 0.32, "learning_rate": 4.8785e-05, "loss": 1.3119, "step": 245 }, { "epoch": 0.32, "learning_rate": 4.8780000000000004e-05, "loss": 0.9687, "step": 246 }, { "epoch": 0.32, "learning_rate": 4.8775000000000007e-05, "loss": 0.9879, "step": 247 }, { "epoch": 0.32, "learning_rate": 4.877e-05, "loss": 0.8308, "step": 248 }, { "epoch": 0.33, "learning_rate": 4.8765e-05, "loss": 1.1004, "step": 249 }, { "epoch": 0.33, "learning_rate": 4.876e-05, "loss": 0.951, "step": 250 }, { "epoch": 0.33, "learning_rate": 4.8755e-05, "loss": 0.9113, "step": 251 }, { "epoch": 0.33, "learning_rate": 4.875e-05, "loss": 0.8915, "step": 252 }, { "epoch": 0.33, "learning_rate": 4.8745e-05, "loss": 1.001, "step": 253 }, { "epoch": 0.33, "learning_rate": 4.8740000000000004e-05, "loss": 1.1171, "step": 254 }, { "epoch": 0.33, "learning_rate": 4.8735e-05, "loss": 0.9136, "step": 255 }, { "epoch": 0.34, "learning_rate": 4.873e-05, "loss": 1.165, "step": 256 }, { "epoch": 0.34, "learning_rate": 4.8725000000000005e-05, "loss": 1.0931, "step": 257 }, { "epoch": 0.34, "learning_rate": 4.872000000000001e-05, "loss": 0.8351, "step": 258 }, { "epoch": 0.34, "learning_rate": 4.8715000000000004e-05, "loss": 0.8134, "step": 259 }, { "epoch": 0.34, "learning_rate": 4.871e-05, "loss": 1.2317, "step": 260 }, { "epoch": 0.34, "learning_rate": 4.8705e-05, "loss": 0.9612, "step": 261 }, { "epoch": 0.34, "learning_rate": 4.87e-05, "loss": 0.845, "step": 262 }, { "epoch": 0.34, "learning_rate": 4.8695e-05, "loss": 0.9568, "step": 263 }, { "epoch": 0.35, "learning_rate": 4.869e-05, "loss": 1.0104, "step": 264 }, { "epoch": 0.35, "learning_rate": 4.8685000000000006e-05, "loss": 0.8468, "step": 265 }, { "epoch": 0.35, "learning_rate": 4.868e-05, "loss": 0.9904, "step": 266 }, { "epoch": 0.35, "learning_rate": 4.8675000000000004e-05, "loss": 0.9057, "step": 267 }, { "epoch": 0.35, "learning_rate": 4.867000000000001e-05, "loss": 1.0353, "step": 268 }, { "epoch": 0.35, "learning_rate": 4.8665e-05, "loss": 1.161, "step": 269 }, { "epoch": 0.35, "learning_rate": 4.866e-05, "loss": 1.0005, "step": 270 }, { "epoch": 0.35, "learning_rate": 4.8655e-05, "loss": 1.1971, "step": 271 }, { "epoch": 0.36, "learning_rate": 4.8650000000000003e-05, "loss": 0.9908, "step": 272 }, { "epoch": 0.36, "learning_rate": 4.8645e-05, "loss": 1.1434, "step": 273 }, { "epoch": 0.36, "learning_rate": 4.864e-05, "loss": 0.9481, "step": 274 }, { "epoch": 0.36, "learning_rate": 4.8635000000000004e-05, "loss": 0.9522, "step": 275 }, { "epoch": 0.36, "learning_rate": 4.863e-05, "loss": 0.9144, "step": 276 }, { "epoch": 0.36, "learning_rate": 4.8625e-05, "loss": 0.9243, "step": 277 }, { "epoch": 0.36, "learning_rate": 4.8620000000000005e-05, "loss": 0.8601, "step": 278 }, { "epoch": 0.37, "learning_rate": 4.861500000000001e-05, "loss": 1.0374, "step": 279 }, { "epoch": 0.37, "learning_rate": 4.861e-05, "loss": 0.8276, "step": 280 }, { "epoch": 0.37, "learning_rate": 4.8605e-05, "loss": 0.8931, "step": 281 }, { "epoch": 0.37, "learning_rate": 4.86e-05, "loss": 1.0224, "step": 282 }, { "epoch": 0.37, "learning_rate": 4.8595000000000005e-05, "loss": 0.7966, "step": 283 }, { "epoch": 0.37, "learning_rate": 4.859e-05, "loss": 0.9948, "step": 284 }, { "epoch": 0.37, "learning_rate": 4.8585e-05, "loss": 0.9496, "step": 285 }, { "epoch": 0.37, "learning_rate": 4.8580000000000006e-05, "loss": 0.9637, "step": 286 }, { "epoch": 0.38, "learning_rate": 4.8575e-05, "loss": 1.0753, "step": 287 }, { "epoch": 0.38, "learning_rate": 4.8570000000000004e-05, "loss": 1.136, "step": 288 }, { "epoch": 0.38, "learning_rate": 4.856500000000001e-05, "loss": 0.9623, "step": 289 }, { "epoch": 0.38, "learning_rate": 4.856e-05, "loss": 0.9395, "step": 290 }, { "epoch": 0.38, "learning_rate": 4.8555e-05, "loss": 1.0259, "step": 291 }, { "epoch": 0.38, "learning_rate": 4.855e-05, "loss": 1.0602, "step": 292 }, { "epoch": 0.38, "learning_rate": 4.8545000000000004e-05, "loss": 0.8122, "step": 293 }, { "epoch": 0.38, "learning_rate": 4.854e-05, "loss": 0.9963, "step": 294 }, { "epoch": 0.39, "learning_rate": 4.8535e-05, "loss": 1.0787, "step": 295 }, { "epoch": 0.39, "learning_rate": 4.8530000000000005e-05, "loss": 1.0264, "step": 296 }, { "epoch": 0.39, "learning_rate": 4.8525e-05, "loss": 0.8854, "step": 297 }, { "epoch": 0.39, "learning_rate": 4.852e-05, "loss": 1.0472, "step": 298 }, { "epoch": 0.39, "learning_rate": 4.8515000000000006e-05, "loss": 0.8604, "step": 299 }, { "epoch": 0.39, "learning_rate": 4.851e-05, "loss": 1.2423, "step": 300 }, { "epoch": 0.39, "learning_rate": 4.8505e-05, "loss": 0.8124, "step": 301 }, { "epoch": 0.4, "learning_rate": 4.85e-05, "loss": 0.8976, "step": 302 }, { "epoch": 0.4, "learning_rate": 4.8495e-05, "loss": 0.9854, "step": 303 }, { "epoch": 0.4, "learning_rate": 4.8490000000000005e-05, "loss": 1.1059, "step": 304 }, { "epoch": 0.4, "learning_rate": 4.8485e-05, "loss": 0.9352, "step": 305 }, { "epoch": 0.4, "learning_rate": 4.8480000000000003e-05, "loss": 0.8203, "step": 306 }, { "epoch": 0.4, "learning_rate": 4.8475000000000006e-05, "loss": 0.8752, "step": 307 }, { "epoch": 0.4, "learning_rate": 4.847e-05, "loss": 0.8176, "step": 308 }, { "epoch": 0.4, "learning_rate": 4.8465000000000004e-05, "loss": 1.2748, "step": 309 }, { "epoch": 0.41, "learning_rate": 4.846e-05, "loss": 0.8861, "step": 310 }, { "epoch": 0.41, "learning_rate": 4.8455e-05, "loss": 0.8223, "step": 311 }, { "epoch": 0.41, "learning_rate": 4.845e-05, "loss": 1.0336, "step": 312 }, { "epoch": 0.41, "learning_rate": 4.8445e-05, "loss": 1.379, "step": 313 }, { "epoch": 0.41, "learning_rate": 4.8440000000000004e-05, "loss": 1.0829, "step": 314 }, { "epoch": 0.41, "learning_rate": 4.8435e-05, "loss": 1.0375, "step": 315 }, { "epoch": 0.41, "learning_rate": 4.843e-05, "loss": 1.2387, "step": 316 }, { "epoch": 0.41, "learning_rate": 4.8425000000000005e-05, "loss": 1.1049, "step": 317 }, { "epoch": 0.42, "learning_rate": 4.842000000000001e-05, "loss": 1.0294, "step": 318 }, { "epoch": 0.42, "learning_rate": 4.8415e-05, "loss": 1.0098, "step": 319 }, { "epoch": 0.42, "learning_rate": 4.841e-05, "loss": 1.3153, "step": 320 }, { "epoch": 0.42, "learning_rate": 4.8405e-05, "loss": 1.1627, "step": 321 }, { "epoch": 0.42, "learning_rate": 4.8400000000000004e-05, "loss": 0.9937, "step": 322 }, { "epoch": 0.42, "learning_rate": 4.8395e-05, "loss": 0.9938, "step": 323 }, { "epoch": 0.42, "learning_rate": 4.839e-05, "loss": 0.8659, "step": 324 }, { "epoch": 0.43, "learning_rate": 4.8385000000000005e-05, "loss": 0.8741, "step": 325 }, { "epoch": 0.43, "learning_rate": 4.838e-05, "loss": 0.9767, "step": 326 }, { "epoch": 0.43, "learning_rate": 4.8375000000000004e-05, "loss": 0.9759, "step": 327 }, { "epoch": 0.43, "learning_rate": 4.8370000000000006e-05, "loss": 0.8588, "step": 328 }, { "epoch": 0.43, "learning_rate": 4.8365e-05, "loss": 0.9538, "step": 329 }, { "epoch": 0.43, "learning_rate": 4.836e-05, "loss": 0.841, "step": 330 }, { "epoch": 0.43, "learning_rate": 4.8355e-05, "loss": 1.1929, "step": 331 }, { "epoch": 0.43, "learning_rate": 4.835e-05, "loss": 0.897, "step": 332 }, { "epoch": 0.44, "learning_rate": 4.8345e-05, "loss": 0.9758, "step": 333 }, { "epoch": 0.44, "learning_rate": 4.834e-05, "loss": 0.9275, "step": 334 }, { "epoch": 0.44, "learning_rate": 4.8335000000000004e-05, "loss": 1.2234, "step": 335 }, { "epoch": 0.44, "learning_rate": 4.833e-05, "loss": 0.8679, "step": 336 }, { "epoch": 0.44, "learning_rate": 4.8325e-05, "loss": 1.0121, "step": 337 }, { "epoch": 0.44, "learning_rate": 4.8320000000000005e-05, "loss": 0.8055, "step": 338 }, { "epoch": 0.44, "learning_rate": 4.831500000000001e-05, "loss": 0.9001, "step": 339 }, { "epoch": 0.45, "learning_rate": 4.8309999999999997e-05, "loss": 1.0467, "step": 340 }, { "epoch": 0.45, "learning_rate": 4.8305e-05, "loss": 1.0771, "step": 341 }, { "epoch": 0.45, "learning_rate": 4.83e-05, "loss": 1.0733, "step": 342 }, { "epoch": 0.45, "learning_rate": 4.8295000000000004e-05, "loss": 0.8405, "step": 343 }, { "epoch": 0.45, "learning_rate": 4.829e-05, "loss": 0.9773, "step": 344 }, { "epoch": 0.45, "learning_rate": 4.8285e-05, "loss": 1.2372, "step": 345 }, { "epoch": 0.45, "learning_rate": 4.8280000000000005e-05, "loss": 0.9112, "step": 346 }, { "epoch": 0.45, "learning_rate": 4.8275e-05, "loss": 0.8111, "step": 347 }, { "epoch": 0.46, "learning_rate": 4.8270000000000004e-05, "loss": 1.0308, "step": 348 }, { "epoch": 0.46, "learning_rate": 4.8265000000000006e-05, "loss": 0.8704, "step": 349 }, { "epoch": 0.46, "learning_rate": 4.826e-05, "loss": 1.2091, "step": 350 }, { "epoch": 0.46, "learning_rate": 4.8255e-05, "loss": 1.1954, "step": 351 }, { "epoch": 0.46, "learning_rate": 4.825e-05, "loss": 0.91, "step": 352 }, { "epoch": 0.46, "learning_rate": 4.8245e-05, "loss": 0.9495, "step": 353 }, { "epoch": 0.46, "learning_rate": 4.824e-05, "loss": 0.912, "step": 354 }, { "epoch": 0.46, "learning_rate": 4.8235e-05, "loss": 1.0716, "step": 355 }, { "epoch": 0.47, "learning_rate": 4.8230000000000004e-05, "loss": 0.9116, "step": 356 }, { "epoch": 0.47, "learning_rate": 4.822500000000001e-05, "loss": 0.9479, "step": 357 }, { "epoch": 0.47, "learning_rate": 4.822e-05, "loss": 1.0415, "step": 358 }, { "epoch": 0.47, "learning_rate": 4.8215000000000005e-05, "loss": 0.9703, "step": 359 }, { "epoch": 0.47, "learning_rate": 4.821e-05, "loss": 0.8931, "step": 360 }, { "epoch": 0.47, "learning_rate": 4.8205000000000003e-05, "loss": 1.025, "step": 361 }, { "epoch": 0.47, "learning_rate": 4.82e-05, "loss": 0.8893, "step": 362 }, { "epoch": 0.48, "learning_rate": 4.8195e-05, "loss": 0.9929, "step": 363 }, { "epoch": 0.48, "learning_rate": 4.8190000000000004e-05, "loss": 1.2372, "step": 364 }, { "epoch": 0.48, "learning_rate": 4.8185e-05, "loss": 0.876, "step": 365 }, { "epoch": 0.48, "learning_rate": 4.818e-05, "loss": 0.9533, "step": 366 }, { "epoch": 0.48, "learning_rate": 4.8175000000000005e-05, "loss": 0.7874, "step": 367 }, { "epoch": 0.48, "learning_rate": 4.817e-05, "loss": 0.9386, "step": 368 }, { "epoch": 0.48, "learning_rate": 4.8165000000000004e-05, "loss": 0.975, "step": 369 }, { "epoch": 0.48, "learning_rate": 4.816e-05, "loss": 1.004, "step": 370 }, { "epoch": 0.49, "learning_rate": 4.8155e-05, "loss": 0.9017, "step": 371 }, { "epoch": 0.49, "learning_rate": 4.815e-05, "loss": 0.9697, "step": 372 }, { "epoch": 0.49, "learning_rate": 4.8145e-05, "loss": 1.0979, "step": 373 }, { "epoch": 0.49, "learning_rate": 4.814e-05, "loss": 1.0298, "step": 374 }, { "epoch": 0.49, "learning_rate": 4.8135e-05, "loss": 0.9384, "step": 375 }, { "epoch": 0.49, "learning_rate": 4.813e-05, "loss": 0.9761, "step": 376 }, { "epoch": 0.49, "learning_rate": 4.8125000000000004e-05, "loss": 1.0137, "step": 377 }, { "epoch": 0.49, "learning_rate": 4.812000000000001e-05, "loss": 0.9922, "step": 378 }, { "epoch": 0.5, "learning_rate": 4.8115e-05, "loss": 0.9694, "step": 379 }, { "epoch": 0.5, "learning_rate": 4.8110000000000005e-05, "loss": 1.3526, "step": 380 }, { "epoch": 0.5, "learning_rate": 4.8105e-05, "loss": 0.8072, "step": 381 }, { "epoch": 0.5, "learning_rate": 4.8100000000000004e-05, "loss": 1.2042, "step": 382 }, { "epoch": 0.5, "learning_rate": 4.8095e-05, "loss": 0.9266, "step": 383 }, { "epoch": 0.5, "learning_rate": 4.809e-05, "loss": 0.8513, "step": 384 }, { "epoch": 0.5, "learning_rate": 4.8085000000000005e-05, "loss": 1.0164, "step": 385 }, { "epoch": 0.51, "learning_rate": 4.808e-05, "loss": 1.1594, "step": 386 }, { "epoch": 0.51, "learning_rate": 4.8075e-05, "loss": 0.9869, "step": 387 }, { "epoch": 0.51, "learning_rate": 4.8070000000000006e-05, "loss": 0.9873, "step": 388 }, { "epoch": 0.51, "learning_rate": 4.8065e-05, "loss": 1.0241, "step": 389 }, { "epoch": 0.51, "learning_rate": 4.8060000000000004e-05, "loss": 0.861, "step": 390 }, { "epoch": 0.51, "learning_rate": 4.8055e-05, "loss": 0.8593, "step": 391 }, { "epoch": 0.51, "learning_rate": 4.805e-05, "loss": 1.1614, "step": 392 }, { "epoch": 0.51, "learning_rate": 4.8045e-05, "loss": 0.7522, "step": 393 }, { "epoch": 0.52, "learning_rate": 4.804e-05, "loss": 0.9687, "step": 394 }, { "epoch": 0.52, "learning_rate": 4.8035000000000003e-05, "loss": 1.0756, "step": 395 }, { "epoch": 0.52, "learning_rate": 4.8030000000000006e-05, "loss": 1.0197, "step": 396 }, { "epoch": 0.52, "learning_rate": 4.8025e-05, "loss": 0.9562, "step": 397 }, { "epoch": 0.52, "learning_rate": 4.8020000000000004e-05, "loss": 0.8951, "step": 398 }, { "epoch": 0.52, "learning_rate": 4.801500000000001e-05, "loss": 0.9743, "step": 399 }, { "epoch": 0.52, "learning_rate": 4.801e-05, "loss": 0.866, "step": 400 }, { "epoch": 0.52, "learning_rate": 4.8005e-05, "loss": 1.1208, "step": 401 }, { "epoch": 0.53, "learning_rate": 4.8e-05, "loss": 0.9494, "step": 402 }, { "epoch": 0.53, "learning_rate": 4.7995000000000004e-05, "loss": 0.9399, "step": 403 }, { "epoch": 0.53, "learning_rate": 4.799e-05, "loss": 1.022, "step": 404 }, { "epoch": 0.53, "learning_rate": 4.7985e-05, "loss": 1.1854, "step": 405 }, { "epoch": 0.53, "learning_rate": 4.7980000000000005e-05, "loss": 0.8429, "step": 406 }, { "epoch": 0.53, "learning_rate": 4.7975e-05, "loss": 0.9398, "step": 407 }, { "epoch": 0.53, "learning_rate": 4.797e-05, "loss": 0.7363, "step": 408 }, { "epoch": 0.54, "learning_rate": 4.7965000000000006e-05, "loss": 0.84, "step": 409 }, { "epoch": 0.54, "learning_rate": 4.796e-05, "loss": 1.1485, "step": 410 }, { "epoch": 0.54, "learning_rate": 4.7955e-05, "loss": 1.1136, "step": 411 }, { "epoch": 0.54, "learning_rate": 4.795e-05, "loss": 0.9627, "step": 412 }, { "epoch": 0.54, "learning_rate": 4.7945e-05, "loss": 1.0065, "step": 413 }, { "epoch": 0.54, "learning_rate": 4.794e-05, "loss": 1.0542, "step": 414 }, { "epoch": 0.54, "learning_rate": 4.7935e-05, "loss": 1.0974, "step": 415 }, { "epoch": 0.54, "learning_rate": 4.7930000000000004e-05, "loss": 0.9376, "step": 416 }, { "epoch": 0.55, "learning_rate": 4.7925000000000006e-05, "loss": 0.8115, "step": 417 }, { "epoch": 0.55, "learning_rate": 4.792e-05, "loss": 0.9539, "step": 418 }, { "epoch": 0.55, "learning_rate": 4.7915000000000005e-05, "loss": 0.727, "step": 419 }, { "epoch": 0.55, "learning_rate": 4.791000000000001e-05, "loss": 0.877, "step": 420 }, { "epoch": 0.55, "learning_rate": 4.7905e-05, "loss": 0.9069, "step": 421 }, { "epoch": 0.55, "learning_rate": 4.7905e-05, "loss": 0.911, "step": 422 }, { "epoch": 0.55, "learning_rate": 4.79e-05, "loss": 0.9617, "step": 423 }, { "epoch": 0.55, "learning_rate": 4.7895e-05, "loss": 0.8047, "step": 424 }, { "epoch": 0.56, "learning_rate": 4.7890000000000004e-05, "loss": 0.8611, "step": 425 }, { "epoch": 0.56, "learning_rate": 4.7885e-05, "loss": 0.7866, "step": 426 }, { "epoch": 0.56, "learning_rate": 4.788e-05, "loss": 1.046, "step": 427 }, { "epoch": 0.56, "learning_rate": 4.7875000000000005e-05, "loss": 0.9482, "step": 428 }, { "epoch": 0.56, "learning_rate": 4.787e-05, "loss": 0.9005, "step": 429 }, { "epoch": 0.56, "learning_rate": 4.7865e-05, "loss": 1.0614, "step": 430 }, { "epoch": 0.56, "learning_rate": 4.7860000000000006e-05, "loss": 1.1554, "step": 431 }, { "epoch": 0.57, "learning_rate": 4.7855e-05, "loss": 0.8815, "step": 432 }, { "epoch": 0.57, "learning_rate": 4.785e-05, "loss": 1.0509, "step": 433 }, { "epoch": 0.57, "learning_rate": 4.7845e-05, "loss": 0.8848, "step": 434 }, { "epoch": 0.57, "learning_rate": 4.784e-05, "loss": 0.961, "step": 435 }, { "epoch": 0.57, "learning_rate": 4.7835000000000005e-05, "loss": 0.8655, "step": 436 }, { "epoch": 0.57, "learning_rate": 4.783e-05, "loss": 0.9245, "step": 437 }, { "epoch": 0.57, "learning_rate": 4.7825000000000004e-05, "loss": 0.956, "step": 438 }, { "epoch": 0.57, "learning_rate": 4.7820000000000006e-05, "loss": 1.072, "step": 439 }, { "epoch": 0.58, "learning_rate": 4.7815e-05, "loss": 1.0318, "step": 440 }, { "epoch": 0.58, "learning_rate": 4.7810000000000005e-05, "loss": 0.9706, "step": 441 }, { "epoch": 0.58, "learning_rate": 4.7805e-05, "loss": 1.1428, "step": 442 }, { "epoch": 0.58, "learning_rate": 4.78e-05, "loss": 0.9171, "step": 443 }, { "epoch": 0.58, "learning_rate": 4.7795e-05, "loss": 1.0811, "step": 444 }, { "epoch": 0.58, "learning_rate": 4.779e-05, "loss": 0.8371, "step": 445 }, { "epoch": 0.58, "learning_rate": 4.7785000000000004e-05, "loss": 0.9252, "step": 446 }, { "epoch": 0.59, "learning_rate": 4.778e-05, "loss": 1.1784, "step": 447 }, { "epoch": 0.59, "learning_rate": 4.7775e-05, "loss": 0.9094, "step": 448 }, { "epoch": 0.59, "learning_rate": 4.7770000000000005e-05, "loss": 1.0783, "step": 449 }, { "epoch": 0.59, "learning_rate": 4.7765e-05, "loss": 1.0859, "step": 450 }, { "epoch": 0.59, "learning_rate": 4.7760000000000004e-05, "loss": 0.8835, "step": 451 }, { "epoch": 0.59, "learning_rate": 4.7755e-05, "loss": 1.1458, "step": 452 }, { "epoch": 0.59, "learning_rate": 4.775e-05, "loss": 0.8774, "step": 453 }, { "epoch": 0.59, "learning_rate": 4.7745e-05, "loss": 1.1058, "step": 454 }, { "epoch": 0.6, "learning_rate": 4.774e-05, "loss": 0.8463, "step": 455 }, { "epoch": 0.6, "learning_rate": 4.7735e-05, "loss": 1.0673, "step": 456 }, { "epoch": 0.6, "learning_rate": 4.7730000000000005e-05, "loss": 0.8472, "step": 457 }, { "epoch": 0.6, "learning_rate": 4.7725e-05, "loss": 0.9483, "step": 458 }, { "epoch": 0.6, "learning_rate": 4.7720000000000004e-05, "loss": 1.1638, "step": 459 }, { "epoch": 0.6, "learning_rate": 4.7715000000000006e-05, "loss": 0.7696, "step": 460 }, { "epoch": 0.6, "learning_rate": 4.771e-05, "loss": 0.8679, "step": 461 }, { "epoch": 0.6, "learning_rate": 4.7705e-05, "loss": 0.838, "step": 462 }, { "epoch": 0.61, "learning_rate": 4.77e-05, "loss": 0.8732, "step": 463 }, { "epoch": 0.61, "learning_rate": 4.7695e-05, "loss": 0.85, "step": 464 }, { "epoch": 0.61, "learning_rate": 4.769e-05, "loss": 0.916, "step": 465 }, { "epoch": 0.61, "learning_rate": 4.7685e-05, "loss": 1.1771, "step": 466 }, { "epoch": 0.61, "learning_rate": 4.7680000000000004e-05, "loss": 0.8578, "step": 467 }, { "epoch": 0.61, "learning_rate": 4.7675e-05, "loss": 0.7909, "step": 468 }, { "epoch": 0.61, "learning_rate": 4.767e-05, "loss": 1.102, "step": 469 }, { "epoch": 0.62, "learning_rate": 4.7665000000000005e-05, "loss": 0.9071, "step": 470 }, { "epoch": 0.62, "learning_rate": 4.766000000000001e-05, "loss": 1.4196, "step": 471 }, { "epoch": 0.62, "learning_rate": 4.7655e-05, "loss": 1.0134, "step": 472 }, { "epoch": 0.62, "learning_rate": 4.765e-05, "loss": 1.0439, "step": 473 }, { "epoch": 0.62, "learning_rate": 4.7645e-05, "loss": 1.1512, "step": 474 }, { "epoch": 0.62, "learning_rate": 4.7640000000000005e-05, "loss": 1.1807, "step": 475 }, { "epoch": 0.62, "learning_rate": 4.7635e-05, "loss": 0.84, "step": 476 }, { "epoch": 0.62, "learning_rate": 4.763e-05, "loss": 0.803, "step": 477 }, { "epoch": 0.63, "learning_rate": 4.7625000000000006e-05, "loss": 0.7105, "step": 478 }, { "epoch": 0.63, "learning_rate": 4.762e-05, "loss": 1.1844, "step": 479 }, { "epoch": 0.63, "learning_rate": 4.7615000000000004e-05, "loss": 0.8011, "step": 480 }, { "epoch": 0.63, "learning_rate": 4.761000000000001e-05, "loss": 0.9138, "step": 481 }, { "epoch": 0.63, "learning_rate": 4.7605e-05, "loss": 1.0131, "step": 482 }, { "epoch": 0.63, "learning_rate": 4.76e-05, "loss": 0.8559, "step": 483 }, { "epoch": 0.63, "learning_rate": 4.7595e-05, "loss": 1.0717, "step": 484 }, { "epoch": 0.63, "learning_rate": 4.7590000000000003e-05, "loss": 1.0077, "step": 485 }, { "epoch": 0.64, "learning_rate": 4.7585e-05, "loss": 1.2938, "step": 486 }, { "epoch": 0.64, "learning_rate": 4.758e-05, "loss": 0.8785, "step": 487 }, { "epoch": 0.64, "learning_rate": 4.7575000000000004e-05, "loss": 0.9354, "step": 488 }, { "epoch": 0.64, "learning_rate": 4.757e-05, "loss": 1.1447, "step": 489 }, { "epoch": 0.64, "learning_rate": 4.7565e-05, "loss": 0.9996, "step": 490 }, { "epoch": 0.64, "learning_rate": 4.7560000000000005e-05, "loss": 1.0106, "step": 491 }, { "epoch": 0.64, "learning_rate": 4.7555e-05, "loss": 0.9033, "step": 492 }, { "epoch": 0.65, "learning_rate": 4.755e-05, "loss": 1.1715, "step": 493 }, { "epoch": 0.65, "learning_rate": 4.7545e-05, "loss": 0.971, "step": 494 }, { "epoch": 0.65, "learning_rate": 4.754e-05, "loss": 0.955, "step": 495 }, { "epoch": 0.65, "learning_rate": 4.7535000000000005e-05, "loss": 0.8633, "step": 496 }, { "epoch": 0.65, "learning_rate": 4.753e-05, "loss": 0.8306, "step": 497 }, { "epoch": 0.65, "learning_rate": 4.7525e-05, "loss": 0.8759, "step": 498 }, { "epoch": 0.65, "learning_rate": 4.7520000000000006e-05, "loss": 0.955, "step": 499 }, { "epoch": 0.65, "learning_rate": 4.7515e-05, "loss": 0.7848, "step": 500 }, { "epoch": 0.66, "learning_rate": 4.7510000000000004e-05, "loss": 1.1597, "step": 501 }, { "epoch": 0.66, "learning_rate": 4.7505e-05, "loss": 0.7209, "step": 502 }, { "epoch": 0.66, "learning_rate": 4.75e-05, "loss": 0.8751, "step": 503 }, { "epoch": 0.66, "learning_rate": 4.7495e-05, "loss": 1.0763, "step": 504 }, { "epoch": 0.66, "learning_rate": 4.749e-05, "loss": 1.0124, "step": 505 }, { "epoch": 0.66, "learning_rate": 4.7485000000000004e-05, "loss": 0.8246, "step": 506 }, { "epoch": 0.66, "learning_rate": 4.748e-05, "loss": 0.9802, "step": 507 }, { "epoch": 0.66, "learning_rate": 4.7475e-05, "loss": 0.8757, "step": 508 }, { "epoch": 0.67, "learning_rate": 4.7470000000000005e-05, "loss": 1.1055, "step": 509 }, { "epoch": 0.67, "learning_rate": 4.746500000000001e-05, "loss": 0.9797, "step": 510 }, { "epoch": 0.67, "learning_rate": 4.746e-05, "loss": 0.8071, "step": 511 }, { "epoch": 0.67, "learning_rate": 4.7455000000000006e-05, "loss": 0.8686, "step": 512 }, { "epoch": 0.67, "learning_rate": 4.745e-05, "loss": 1.0384, "step": 513 }, { "epoch": 0.67, "learning_rate": 4.7445e-05, "loss": 0.7847, "step": 514 }, { "epoch": 0.67, "learning_rate": 4.744e-05, "loss": 0.9966, "step": 515 }, { "epoch": 0.68, "learning_rate": 4.7435e-05, "loss": 0.9612, "step": 516 }, { "epoch": 0.68, "learning_rate": 4.7430000000000005e-05, "loss": 0.9641, "step": 517 }, { "epoch": 0.68, "learning_rate": 4.7425e-05, "loss": 0.9827, "step": 518 }, { "epoch": 0.68, "learning_rate": 4.742e-05, "loss": 0.879, "step": 519 }, { "epoch": 0.68, "learning_rate": 4.7415000000000006e-05, "loss": 0.9515, "step": 520 }, { "epoch": 0.68, "learning_rate": 4.741e-05, "loss": 0.8694, "step": 521 }, { "epoch": 0.68, "learning_rate": 4.7405000000000004e-05, "loss": 0.8906, "step": 522 }, { "epoch": 0.68, "learning_rate": 4.74e-05, "loss": 0.9185, "step": 523 }, { "epoch": 0.69, "learning_rate": 4.7395e-05, "loss": 1.0126, "step": 524 }, { "epoch": 0.69, "learning_rate": 4.739e-05, "loss": 0.9462, "step": 525 }, { "epoch": 0.69, "learning_rate": 4.7385e-05, "loss": 1.103, "step": 526 }, { "epoch": 0.69, "learning_rate": 4.7380000000000004e-05, "loss": 0.9831, "step": 527 }, { "epoch": 0.69, "learning_rate": 4.7375e-05, "loss": 0.8049, "step": 528 }, { "epoch": 0.69, "learning_rate": 4.737e-05, "loss": 0.8046, "step": 529 }, { "epoch": 0.69, "learning_rate": 4.7365000000000005e-05, "loss": 0.9381, "step": 530 }, { "epoch": 0.7, "learning_rate": 4.736000000000001e-05, "loss": 0.9118, "step": 531 }, { "epoch": 0.7, "learning_rate": 4.7355e-05, "loss": 1.0331, "step": 532 }, { "epoch": 0.7, "learning_rate": 4.735e-05, "loss": 1.1238, "step": 533 }, { "epoch": 0.7, "learning_rate": 4.7345e-05, "loss": 0.7914, "step": 534 }, { "epoch": 0.7, "learning_rate": 4.7340000000000004e-05, "loss": 0.8408, "step": 535 }, { "epoch": 0.7, "learning_rate": 4.7335e-05, "loss": 1.1953, "step": 536 }, { "epoch": 0.7, "learning_rate": 4.733e-05, "loss": 1.2695, "step": 537 }, { "epoch": 0.7, "learning_rate": 4.7325000000000005e-05, "loss": 0.8392, "step": 538 }, { "epoch": 0.71, "learning_rate": 4.732e-05, "loss": 0.9125, "step": 539 }, { "epoch": 0.71, "learning_rate": 4.7315000000000004e-05, "loss": 1.0938, "step": 540 }, { "epoch": 0.71, "learning_rate": 4.7310000000000006e-05, "loss": 1.3212, "step": 541 }, { "epoch": 0.71, "learning_rate": 4.7305e-05, "loss": 0.8802, "step": 542 }, { "epoch": 0.71, "learning_rate": 4.73e-05, "loss": 0.8341, "step": 543 }, { "epoch": 0.71, "learning_rate": 4.7295e-05, "loss": 0.999, "step": 544 }, { "epoch": 0.71, "learning_rate": 4.729e-05, "loss": 1.2435, "step": 545 }, { "epoch": 0.71, "learning_rate": 4.7285e-05, "loss": 1.0853, "step": 546 }, { "epoch": 0.72, "learning_rate": 4.728e-05, "loss": 0.8639, "step": 547 }, { "epoch": 0.72, "learning_rate": 4.7275000000000004e-05, "loss": 0.9827, "step": 548 }, { "epoch": 0.72, "learning_rate": 4.7270000000000007e-05, "loss": 1.1498, "step": 549 }, { "epoch": 0.72, "learning_rate": 4.7265e-05, "loss": 0.7856, "step": 550 }, { "epoch": 0.72, "learning_rate": 4.7260000000000005e-05, "loss": 0.8733, "step": 551 }, { "epoch": 0.72, "learning_rate": 4.725500000000001e-05, "loss": 1.0324, "step": 552 }, { "epoch": 0.72, "learning_rate": 4.7249999999999997e-05, "loss": 1.1517, "step": 553 }, { "epoch": 0.73, "learning_rate": 4.7245e-05, "loss": 1.0397, "step": 554 }, { "epoch": 0.73, "learning_rate": 4.724e-05, "loss": 0.9391, "step": 555 }, { "epoch": 0.73, "learning_rate": 4.7235000000000004e-05, "loss": 1.0823, "step": 556 }, { "epoch": 0.73, "learning_rate": 4.723e-05, "loss": 0.9634, "step": 557 }, { "epoch": 0.73, "learning_rate": 4.7225e-05, "loss": 1.1104, "step": 558 }, { "epoch": 0.73, "learning_rate": 4.7220000000000005e-05, "loss": 0.9149, "step": 559 }, { "epoch": 0.73, "learning_rate": 4.7215e-05, "loss": 0.8582, "step": 560 }, { "epoch": 0.73, "learning_rate": 4.7210000000000004e-05, "loss": 0.7157, "step": 561 }, { "epoch": 0.74, "learning_rate": 4.7205000000000006e-05, "loss": 0.999, "step": 562 }, { "epoch": 0.74, "learning_rate": 4.72e-05, "loss": 1.0583, "step": 563 }, { "epoch": 0.74, "learning_rate": 4.7195e-05, "loss": 0.9158, "step": 564 }, { "epoch": 0.74, "learning_rate": 4.719e-05, "loss": 1.0309, "step": 565 }, { "epoch": 0.74, "learning_rate": 4.7185e-05, "loss": 0.7801, "step": 566 }, { "epoch": 0.74, "learning_rate": 4.718e-05, "loss": 0.9798, "step": 567 }, { "epoch": 0.74, "learning_rate": 4.7175e-05, "loss": 0.8442, "step": 568 }, { "epoch": 0.74, "learning_rate": 4.7170000000000004e-05, "loss": 0.9968, "step": 569 }, { "epoch": 0.75, "learning_rate": 4.716500000000001e-05, "loss": 0.9027, "step": 570 }, { "epoch": 0.75, "learning_rate": 4.716e-05, "loss": 1.0192, "step": 571 }, { "epoch": 0.75, "learning_rate": 4.7155000000000005e-05, "loss": 0.7408, "step": 572 }, { "epoch": 0.75, "learning_rate": 4.715e-05, "loss": 0.9749, "step": 573 }, { "epoch": 0.75, "learning_rate": 4.7145000000000003e-05, "loss": 0.9011, "step": 574 }, { "epoch": 0.75, "learning_rate": 4.714e-05, "loss": 0.6923, "step": 575 }, { "epoch": 0.75, "learning_rate": 4.7135e-05, "loss": 0.8668, "step": 576 }, { "epoch": 0.76, "learning_rate": 4.7130000000000004e-05, "loss": 0.9996, "step": 577 }, { "epoch": 0.76, "learning_rate": 4.7125e-05, "loss": 0.8956, "step": 578 }, { "epoch": 0.76, "learning_rate": 4.712e-05, "loss": 0.8906, "step": 579 }, { "epoch": 0.76, "learning_rate": 4.7115000000000005e-05, "loss": 0.951, "step": 580 }, { "epoch": 0.76, "learning_rate": 4.711e-05, "loss": 0.9211, "step": 581 }, { "epoch": 0.76, "learning_rate": 4.7105000000000004e-05, "loss": 0.8627, "step": 582 }, { "epoch": 0.76, "learning_rate": 4.71e-05, "loss": 1.2368, "step": 583 }, { "epoch": 0.76, "learning_rate": 4.7095e-05, "loss": 0.988, "step": 584 }, { "epoch": 0.77, "learning_rate": 4.709e-05, "loss": 0.8478, "step": 585 }, { "epoch": 0.77, "learning_rate": 4.7085e-05, "loss": 0.8419, "step": 586 }, { "epoch": 0.77, "learning_rate": 4.708e-05, "loss": 0.9499, "step": 587 }, { "epoch": 0.77, "learning_rate": 4.7075e-05, "loss": 1.0634, "step": 588 }, { "epoch": 0.77, "learning_rate": 4.707e-05, "loss": 0.7289, "step": 589 }, { "epoch": 0.77, "learning_rate": 4.7065000000000004e-05, "loss": 1.0728, "step": 590 }, { "epoch": 0.77, "learning_rate": 4.706000000000001e-05, "loss": 0.9505, "step": 591 }, { "epoch": 0.77, "learning_rate": 4.7055e-05, "loss": 1.1179, "step": 592 }, { "epoch": 0.78, "learning_rate": 4.705e-05, "loss": 1.1087, "step": 593 }, { "epoch": 0.78, "learning_rate": 4.7045e-05, "loss": 0.9962, "step": 594 }, { "epoch": 0.78, "learning_rate": 4.7040000000000004e-05, "loss": 0.8274, "step": 595 }, { "epoch": 0.78, "learning_rate": 4.7035e-05, "loss": 1.1123, "step": 596 }, { "epoch": 0.78, "learning_rate": 4.703e-05, "loss": 0.9789, "step": 597 }, { "epoch": 0.78, "learning_rate": 4.7025000000000005e-05, "loss": 0.8778, "step": 598 }, { "epoch": 0.78, "learning_rate": 4.702e-05, "loss": 1.1416, "step": 599 }, { "epoch": 0.79, "learning_rate": 4.7015e-05, "loss": 0.9456, "step": 600 }, { "epoch": 0.79, "learning_rate": 4.7010000000000006e-05, "loss": 0.9099, "step": 601 }, { "epoch": 0.79, "learning_rate": 4.7005e-05, "loss": 0.9017, "step": 602 }, { "epoch": 0.79, "learning_rate": 4.7e-05, "loss": 0.9054, "step": 603 }, { "epoch": 0.79, "learning_rate": 4.6995e-05, "loss": 0.9617, "step": 604 }, { "epoch": 0.79, "learning_rate": 4.699e-05, "loss": 1.0609, "step": 605 }, { "epoch": 0.79, "learning_rate": 4.6985e-05, "loss": 1.064, "step": 606 }, { "epoch": 0.79, "learning_rate": 4.698e-05, "loss": 1.0648, "step": 607 }, { "epoch": 0.8, "learning_rate": 4.6975000000000003e-05, "loss": 0.9504, "step": 608 }, { "epoch": 0.8, "learning_rate": 4.6970000000000006e-05, "loss": 0.9659, "step": 609 }, { "epoch": 0.8, "learning_rate": 4.6965e-05, "loss": 1.0206, "step": 610 }, { "epoch": 0.8, "learning_rate": 4.6960000000000004e-05, "loss": 0.8751, "step": 611 }, { "epoch": 0.8, "learning_rate": 4.695500000000001e-05, "loss": 0.9523, "step": 612 }, { "epoch": 0.8, "learning_rate": 4.695e-05, "loss": 0.7328, "step": 613 }, { "epoch": 0.8, "learning_rate": 4.6945e-05, "loss": 1.194, "step": 614 }, { "epoch": 0.8, "learning_rate": 4.694e-05, "loss": 0.8258, "step": 615 }, { "epoch": 0.81, "learning_rate": 4.6935000000000004e-05, "loss": 0.862, "step": 616 }, { "epoch": 0.81, "learning_rate": 4.693e-05, "loss": 1.1786, "step": 617 }, { "epoch": 0.81, "learning_rate": 4.6925e-05, "loss": 1.0133, "step": 618 }, { "epoch": 0.81, "learning_rate": 4.6920000000000005e-05, "loss": 1.0225, "step": 619 }, { "epoch": 0.81, "learning_rate": 4.6915e-05, "loss": 0.9821, "step": 620 }, { "epoch": 0.81, "learning_rate": 4.691e-05, "loss": 0.9788, "step": 621 }, { "epoch": 0.81, "learning_rate": 4.6905000000000006e-05, "loss": 0.821, "step": 622 }, { "epoch": 0.82, "learning_rate": 4.69e-05, "loss": 0.988, "step": 623 }, { "epoch": 0.82, "learning_rate": 4.6895e-05, "loss": 0.9283, "step": 624 }, { "epoch": 0.82, "learning_rate": 4.689e-05, "loss": 1.0326, "step": 625 }, { "epoch": 0.82, "learning_rate": 4.6885e-05, "loss": 0.7573, "step": 626 }, { "epoch": 0.82, "learning_rate": 4.688e-05, "loss": 0.9193, "step": 627 }, { "epoch": 0.82, "learning_rate": 4.6875e-05, "loss": 0.9022, "step": 628 }, { "epoch": 0.82, "learning_rate": 4.6870000000000004e-05, "loss": 1.0624, "step": 629 }, { "epoch": 0.82, "learning_rate": 4.6865000000000006e-05, "loss": 0.8442, "step": 630 }, { "epoch": 0.83, "learning_rate": 4.686e-05, "loss": 0.9662, "step": 631 }, { "epoch": 0.83, "learning_rate": 4.6855000000000005e-05, "loss": 1.0819, "step": 632 }, { "epoch": 0.83, "learning_rate": 4.685000000000001e-05, "loss": 0.949, "step": 633 }, { "epoch": 0.83, "learning_rate": 4.6845e-05, "loss": 0.7761, "step": 634 }, { "epoch": 0.83, "learning_rate": 4.684e-05, "loss": 0.9946, "step": 635 }, { "epoch": 0.83, "learning_rate": 4.6835e-05, "loss": 1.016, "step": 636 }, { "epoch": 0.83, "learning_rate": 4.6830000000000004e-05, "loss": 0.9721, "step": 637 }, { "epoch": 0.84, "learning_rate": 4.6825e-05, "loss": 0.8492, "step": 638 }, { "epoch": 0.84, "learning_rate": 4.682e-05, "loss": 0.9068, "step": 639 }, { "epoch": 0.84, "learning_rate": 4.6815000000000005e-05, "loss": 0.9359, "step": 640 }, { "epoch": 0.84, "learning_rate": 4.681e-05, "loss": 0.784, "step": 641 }, { "epoch": 0.84, "learning_rate": 4.6805e-05, "loss": 0.9368, "step": 642 }, { "epoch": 0.84, "learning_rate": 4.6800000000000006e-05, "loss": 0.8724, "step": 643 }, { "epoch": 0.84, "learning_rate": 4.6795e-05, "loss": 0.8871, "step": 644 }, { "epoch": 0.84, "learning_rate": 4.679e-05, "loss": 0.8973, "step": 645 }, { "epoch": 0.85, "learning_rate": 4.6785e-05, "loss": 0.9305, "step": 646 }, { "epoch": 0.85, "learning_rate": 4.678e-05, "loss": 1.0288, "step": 647 }, { "epoch": 0.85, "learning_rate": 4.6775000000000005e-05, "loss": 0.8545, "step": 648 }, { "epoch": 0.85, "learning_rate": 4.677e-05, "loss": 0.8731, "step": 649 }, { "epoch": 0.85, "learning_rate": 4.6765000000000004e-05, "loss": 0.7454, "step": 650 }, { "epoch": 0.85, "learning_rate": 4.6760000000000006e-05, "loss": 0.9136, "step": 651 }, { "epoch": 0.85, "learning_rate": 4.6755e-05, "loss": 0.9764, "step": 652 }, { "epoch": 0.85, "learning_rate": 4.6750000000000005e-05, "loss": 0.7126, "step": 653 }, { "epoch": 0.86, "learning_rate": 4.6745e-05, "loss": 0.9466, "step": 654 }, { "epoch": 0.86, "learning_rate": 4.674e-05, "loss": 0.9468, "step": 655 }, { "epoch": 0.86, "learning_rate": 4.6735e-05, "loss": 0.8549, "step": 656 }, { "epoch": 0.86, "learning_rate": 4.673e-05, "loss": 0.8315, "step": 657 }, { "epoch": 0.86, "learning_rate": 4.6725000000000004e-05, "loss": 0.9993, "step": 658 }, { "epoch": 0.86, "learning_rate": 4.672e-05, "loss": 1.0018, "step": 659 }, { "epoch": 0.86, "learning_rate": 4.6715e-05, "loss": 0.9999, "step": 660 }, { "epoch": 0.87, "learning_rate": 4.6710000000000005e-05, "loss": 0.7642, "step": 661 }, { "epoch": 0.87, "learning_rate": 4.670500000000001e-05, "loss": 1.0583, "step": 662 }, { "epoch": 0.87, "learning_rate": 4.6700000000000003e-05, "loss": 1.4718, "step": 663 }, { "epoch": 0.87, "learning_rate": 4.6695e-05, "loss": 1.0648, "step": 664 }, { "epoch": 0.87, "learning_rate": 4.669e-05, "loss": 0.8954, "step": 665 }, { "epoch": 0.87, "learning_rate": 4.6685e-05, "loss": 1.1181, "step": 666 }, { "epoch": 0.87, "learning_rate": 4.668e-05, "loss": 0.9017, "step": 667 }, { "epoch": 0.87, "learning_rate": 4.6675e-05, "loss": 0.9929, "step": 668 }, { "epoch": 0.88, "learning_rate": 4.6670000000000005e-05, "loss": 0.9011, "step": 669 }, { "epoch": 0.88, "learning_rate": 4.6665e-05, "loss": 0.8968, "step": 670 }, { "epoch": 0.88, "learning_rate": 4.6660000000000004e-05, "loss": 0.9369, "step": 671 }, { "epoch": 0.88, "learning_rate": 4.6655000000000006e-05, "loss": 0.7612, "step": 672 }, { "epoch": 0.88, "learning_rate": 4.665e-05, "loss": 0.9257, "step": 673 }, { "epoch": 0.88, "learning_rate": 4.6645e-05, "loss": 1.1022, "step": 674 }, { "epoch": 0.88, "learning_rate": 4.664e-05, "loss": 1.0178, "step": 675 }, { "epoch": 0.88, "learning_rate": 4.6635e-05, "loss": 0.9943, "step": 676 }, { "epoch": 0.89, "learning_rate": 4.663e-05, "loss": 0.8352, "step": 677 }, { "epoch": 0.89, "learning_rate": 4.6625e-05, "loss": 0.9079, "step": 678 }, { "epoch": 0.89, "learning_rate": 4.6620000000000004e-05, "loss": 0.9146, "step": 679 }, { "epoch": 0.89, "learning_rate": 4.6615e-05, "loss": 0.8866, "step": 680 }, { "epoch": 0.89, "learning_rate": 4.661e-05, "loss": 0.9406, "step": 681 }, { "epoch": 0.89, "learning_rate": 4.6605000000000005e-05, "loss": 1.2457, "step": 682 }, { "epoch": 0.89, "learning_rate": 4.660000000000001e-05, "loss": 0.9355, "step": 683 }, { "epoch": 0.9, "learning_rate": 4.6595e-05, "loss": 1.1029, "step": 684 }, { "epoch": 0.9, "learning_rate": 4.659e-05, "loss": 1.0058, "step": 685 }, { "epoch": 0.9, "learning_rate": 4.6585e-05, "loss": 0.889, "step": 686 }, { "epoch": 0.9, "learning_rate": 4.6580000000000005e-05, "loss": 0.881, "step": 687 }, { "epoch": 0.9, "learning_rate": 4.6575e-05, "loss": 1.0272, "step": 688 }, { "epoch": 0.9, "learning_rate": 4.657e-05, "loss": 1.0553, "step": 689 }, { "epoch": 0.9, "learning_rate": 4.6565000000000006e-05, "loss": 0.8764, "step": 690 }, { "epoch": 0.9, "learning_rate": 4.656e-05, "loss": 1.2845, "step": 691 }, { "epoch": 0.91, "learning_rate": 4.6555000000000004e-05, "loss": 1.0675, "step": 692 }, { "epoch": 0.91, "learning_rate": 4.655000000000001e-05, "loss": 0.8433, "step": 693 }, { "epoch": 0.91, "learning_rate": 4.6545e-05, "loss": 0.8774, "step": 694 }, { "epoch": 0.91, "learning_rate": 4.654e-05, "loss": 1.0627, "step": 695 }, { "epoch": 0.91, "learning_rate": 4.6535e-05, "loss": 0.8818, "step": 696 }, { "epoch": 0.91, "learning_rate": 4.6530000000000003e-05, "loss": 0.877, "step": 697 }, { "epoch": 0.91, "learning_rate": 4.6525e-05, "loss": 0.956, "step": 698 }, { "epoch": 0.91, "learning_rate": 4.652e-05, "loss": 1.1255, "step": 699 }, { "epoch": 0.92, "learning_rate": 4.6515000000000004e-05, "loss": 0.8945, "step": 700 }, { "epoch": 0.92, "learning_rate": 4.651e-05, "loss": 1.1845, "step": 701 }, { "epoch": 0.92, "learning_rate": 4.6505e-05, "loss": 0.972, "step": 702 }, { "epoch": 0.92, "learning_rate": 4.6500000000000005e-05, "loss": 1.1082, "step": 703 }, { "epoch": 0.92, "learning_rate": 4.6495e-05, "loss": 0.8104, "step": 704 }, { "epoch": 0.92, "learning_rate": 4.649e-05, "loss": 0.7955, "step": 705 }, { "epoch": 0.92, "learning_rate": 4.6485e-05, "loss": 0.9673, "step": 706 }, { "epoch": 0.93, "learning_rate": 4.648e-05, "loss": 1.0446, "step": 707 }, { "epoch": 0.93, "learning_rate": 4.6475000000000005e-05, "loss": 0.9902, "step": 708 }, { "epoch": 0.93, "learning_rate": 4.647e-05, "loss": 0.8445, "step": 709 }, { "epoch": 0.93, "learning_rate": 4.6465e-05, "loss": 0.8785, "step": 710 }, { "epoch": 0.93, "learning_rate": 4.6460000000000006e-05, "loss": 0.9222, "step": 711 }, { "epoch": 0.93, "learning_rate": 4.6455e-05, "loss": 0.9375, "step": 712 }, { "epoch": 0.93, "learning_rate": 4.6450000000000004e-05, "loss": 0.9171, "step": 713 }, { "epoch": 0.93, "learning_rate": 4.6445e-05, "loss": 0.8032, "step": 714 }, { "epoch": 0.94, "learning_rate": 4.644e-05, "loss": 0.9399, "step": 715 }, { "epoch": 0.94, "learning_rate": 4.6435e-05, "loss": 0.9097, "step": 716 }, { "epoch": 0.94, "learning_rate": 4.643e-05, "loss": 0.8224, "step": 717 }, { "epoch": 0.94, "learning_rate": 4.6425000000000004e-05, "loss": 0.9174, "step": 718 }, { "epoch": 0.94, "learning_rate": 4.642e-05, "loss": 0.9025, "step": 719 }, { "epoch": 0.94, "learning_rate": 4.6415e-05, "loss": 0.8342, "step": 720 }, { "epoch": 0.94, "learning_rate": 4.6410000000000005e-05, "loss": 1.1155, "step": 721 }, { "epoch": 0.95, "learning_rate": 4.640500000000001e-05, "loss": 1.1207, "step": 722 }, { "epoch": 0.95, "learning_rate": 4.64e-05, "loss": 0.7767, "step": 723 }, { "epoch": 0.95, "learning_rate": 4.6395e-05, "loss": 1.1853, "step": 724 }, { "epoch": 0.95, "learning_rate": 4.639e-05, "loss": 0.9474, "step": 725 }, { "epoch": 0.95, "learning_rate": 4.6385000000000004e-05, "loss": 0.9361, "step": 726 }, { "epoch": 0.95, "learning_rate": 4.638e-05, "loss": 1.2294, "step": 727 }, { "epoch": 0.95, "learning_rate": 4.6375e-05, "loss": 0.7878, "step": 728 }, { "epoch": 0.95, "learning_rate": 4.6370000000000005e-05, "loss": 0.8989, "step": 729 }, { "epoch": 0.96, "learning_rate": 4.6365e-05, "loss": 0.9554, "step": 730 }, { "epoch": 0.96, "learning_rate": 4.636e-05, "loss": 0.8646, "step": 731 }, { "epoch": 0.96, "learning_rate": 4.6355000000000006e-05, "loss": 1.0831, "step": 732 }, { "epoch": 0.96, "learning_rate": 4.635e-05, "loss": 0.9, "step": 733 }, { "epoch": 0.96, "learning_rate": 4.6345e-05, "loss": 1.0377, "step": 734 }, { "epoch": 0.96, "learning_rate": 4.634e-05, "loss": 0.9044, "step": 735 }, { "epoch": 0.96, "learning_rate": 4.6335e-05, "loss": 1.0234, "step": 736 }, { "epoch": 0.96, "learning_rate": 4.633e-05, "loss": 0.915, "step": 737 }, { "epoch": 0.97, "learning_rate": 4.6325e-05, "loss": 0.8516, "step": 738 }, { "epoch": 0.97, "learning_rate": 4.6320000000000004e-05, "loss": 0.9338, "step": 739 }, { "epoch": 0.97, "learning_rate": 4.6315e-05, "loss": 0.9885, "step": 740 }, { "epoch": 0.97, "learning_rate": 4.631e-05, "loss": 1.0036, "step": 741 }, { "epoch": 0.97, "learning_rate": 4.6305000000000005e-05, "loss": 0.8386, "step": 742 }, { "epoch": 0.97, "learning_rate": 4.630000000000001e-05, "loss": 0.7953, "step": 743 }, { "epoch": 0.97, "learning_rate": 4.6294999999999996e-05, "loss": 0.989, "step": 744 }, { "epoch": 0.98, "learning_rate": 4.629e-05, "loss": 1.0504, "step": 745 }, { "epoch": 0.98, "learning_rate": 4.6285e-05, "loss": 0.8721, "step": 746 }, { "epoch": 0.98, "learning_rate": 4.6280000000000004e-05, "loss": 0.8353, "step": 747 }, { "epoch": 0.98, "learning_rate": 4.6275e-05, "loss": 0.7908, "step": 748 }, { "epoch": 0.98, "learning_rate": 4.627e-05, "loss": 0.9932, "step": 749 }, { "epoch": 0.98, "learning_rate": 4.6265000000000005e-05, "loss": 0.8553, "step": 750 }, { "epoch": 0.98, "learning_rate": 4.626e-05, "loss": 0.7297, "step": 751 }, { "epoch": 0.98, "learning_rate": 4.6255000000000004e-05, "loss": 0.8229, "step": 752 }, { "epoch": 0.99, "learning_rate": 4.6250000000000006e-05, "loss": 0.9847, "step": 753 }, { "epoch": 0.99, "learning_rate": 4.6245e-05, "loss": 1.0427, "step": 754 }, { "epoch": 0.99, "learning_rate": 4.624e-05, "loss": 0.9766, "step": 755 }, { "epoch": 0.99, "learning_rate": 4.6235e-05, "loss": 0.8629, "step": 756 }, { "epoch": 0.99, "learning_rate": 4.623e-05, "loss": 1.141, "step": 757 }, { "epoch": 0.99, "learning_rate": 4.6225e-05, "loss": 1.0004, "step": 758 }, { "epoch": 0.99, "learning_rate": 4.622e-05, "loss": 1.0183, "step": 759 }, { "epoch": 0.99, "learning_rate": 4.6215000000000004e-05, "loss": 1.2385, "step": 760 }, { "epoch": 1.0, "learning_rate": 4.6210000000000006e-05, "loss": 0.8413, "step": 761 }, { "epoch": 1.0, "learning_rate": 4.6205e-05, "loss": 0.8755, "step": 762 }, { "epoch": 1.0, "learning_rate": 4.6200000000000005e-05, "loss": 0.7853, "step": 763 }, { "epoch": 1.0, "learning_rate": 4.619500000000001e-05, "loss": 0.7659, "step": 764 }, { "epoch": 1.0, "learning_rate": 4.619e-05, "loss": 0.6121, "step": 765 }, { "epoch": 1.0, "learning_rate": 4.6185e-05, "loss": 0.8022, "step": 766 }, { "epoch": 1.0, "learning_rate": 4.618e-05, "loss": 0.6172, "step": 767 }, { "epoch": 1.01, "learning_rate": 4.6175000000000004e-05, "loss": 0.6093, "step": 768 }, { "epoch": 1.01, "learning_rate": 4.617e-05, "loss": 0.5324, "step": 769 }, { "epoch": 1.01, "learning_rate": 4.6165e-05, "loss": 0.6579, "step": 770 }, { "epoch": 1.01, "learning_rate": 4.6160000000000005e-05, "loss": 0.5984, "step": 771 }, { "epoch": 1.01, "learning_rate": 4.6155e-05, "loss": 0.5234, "step": 772 }, { "epoch": 1.01, "learning_rate": 4.6150000000000004e-05, "loss": 0.7531, "step": 773 }, { "epoch": 1.01, "learning_rate": 4.6145000000000006e-05, "loss": 0.5738, "step": 774 }, { "epoch": 1.01, "learning_rate": 4.614e-05, "loss": 0.6364, "step": 775 }, { "epoch": 1.02, "learning_rate": 4.6135e-05, "loss": 0.7311, "step": 776 }, { "epoch": 1.02, "learning_rate": 4.613e-05, "loss": 0.5477, "step": 777 }, { "epoch": 1.02, "learning_rate": 4.6125e-05, "loss": 0.5899, "step": 778 }, { "epoch": 1.02, "learning_rate": 4.612e-05, "loss": 0.5699, "step": 779 }, { "epoch": 1.02, "learning_rate": 4.6115e-05, "loss": 0.593, "step": 780 }, { "epoch": 1.02, "learning_rate": 4.6110000000000004e-05, "loss": 0.5065, "step": 781 }, { "epoch": 1.02, "learning_rate": 4.610500000000001e-05, "loss": 0.5892, "step": 782 }, { "epoch": 1.02, "learning_rate": 4.61e-05, "loss": 0.5957, "step": 783 }, { "epoch": 1.03, "learning_rate": 4.6095000000000005e-05, "loss": 0.6689, "step": 784 }, { "epoch": 1.03, "learning_rate": 4.609e-05, "loss": 0.5482, "step": 785 }, { "epoch": 1.03, "learning_rate": 4.6085000000000003e-05, "loss": 0.6407, "step": 786 }, { "epoch": 1.03, "learning_rate": 4.608e-05, "loss": 0.5532, "step": 787 }, { "epoch": 1.03, "learning_rate": 4.6075e-05, "loss": 0.5516, "step": 788 }, { "epoch": 1.03, "learning_rate": 4.6070000000000004e-05, "loss": 0.5754, "step": 789 }, { "epoch": 1.03, "learning_rate": 4.6065e-05, "loss": 0.6982, "step": 790 }, { "epoch": 1.04, "learning_rate": 4.606e-05, "loss": 0.5117, "step": 791 }, { "epoch": 1.04, "learning_rate": 4.6055000000000005e-05, "loss": 0.5348, "step": 792 }, { "epoch": 1.04, "learning_rate": 4.605e-05, "loss": 0.5244, "step": 793 }, { "epoch": 1.04, "learning_rate": 4.6045000000000004e-05, "loss": 0.5124, "step": 794 }, { "epoch": 1.04, "learning_rate": 4.604e-05, "loss": 0.4848, "step": 795 }, { "epoch": 1.04, "learning_rate": 4.6035e-05, "loss": 0.5509, "step": 796 }, { "epoch": 1.04, "learning_rate": 4.603e-05, "loss": 0.5407, "step": 797 }, { "epoch": 1.04, "learning_rate": 4.6025e-05, "loss": 0.6558, "step": 798 }, { "epoch": 1.05, "learning_rate": 4.602e-05, "loss": 0.6191, "step": 799 }, { "epoch": 1.05, "learning_rate": 4.6015000000000006e-05, "loss": 0.4214, "step": 800 }, { "epoch": 1.05, "learning_rate": 4.601e-05, "loss": 0.6567, "step": 801 }, { "epoch": 1.05, "learning_rate": 4.6005000000000004e-05, "loss": 0.7937, "step": 802 }, { "epoch": 1.05, "learning_rate": 4.600000000000001e-05, "loss": 0.5406, "step": 803 }, { "epoch": 1.05, "learning_rate": 4.5995e-05, "loss": 0.6405, "step": 804 }, { "epoch": 1.05, "learning_rate": 4.599e-05, "loss": 0.6404, "step": 805 }, { "epoch": 1.05, "learning_rate": 4.5985e-05, "loss": 0.522, "step": 806 }, { "epoch": 1.06, "learning_rate": 4.5980000000000004e-05, "loss": 0.4715, "step": 807 }, { "epoch": 1.06, "learning_rate": 4.5975e-05, "loss": 0.6197, "step": 808 }, { "epoch": 1.06, "learning_rate": 4.597e-05, "loss": 0.4675, "step": 809 }, { "epoch": 1.06, "learning_rate": 4.5965000000000005e-05, "loss": 0.6953, "step": 810 }, { "epoch": 1.06, "learning_rate": 4.596e-05, "loss": 0.4702, "step": 811 }, { "epoch": 1.06, "learning_rate": 4.5955e-05, "loss": 0.64, "step": 812 }, { "epoch": 1.06, "learning_rate": 4.5950000000000006e-05, "loss": 0.7632, "step": 813 }, { "epoch": 1.07, "learning_rate": 4.5945e-05, "loss": 0.6144, "step": 814 }, { "epoch": 1.07, "learning_rate": 4.594e-05, "loss": 0.5461, "step": 815 }, { "epoch": 1.07, "learning_rate": 4.5935e-05, "loss": 0.7641, "step": 816 }, { "epoch": 1.07, "learning_rate": 4.593e-05, "loss": 0.6639, "step": 817 }, { "epoch": 1.07, "learning_rate": 4.5925e-05, "loss": 0.9299, "step": 818 }, { "epoch": 1.07, "learning_rate": 4.592e-05, "loss": 0.5727, "step": 819 }, { "epoch": 1.07, "learning_rate": 4.5915000000000003e-05, "loss": 0.6164, "step": 820 }, { "epoch": 1.07, "learning_rate": 4.5910000000000006e-05, "loss": 0.6412, "step": 821 }, { "epoch": 1.08, "learning_rate": 4.5905e-05, "loss": 0.6242, "step": 822 }, { "epoch": 1.08, "learning_rate": 4.5900000000000004e-05, "loss": 0.7899, "step": 823 }, { "epoch": 1.08, "learning_rate": 4.589500000000001e-05, "loss": 0.5628, "step": 824 }, { "epoch": 1.08, "learning_rate": 4.589e-05, "loss": 0.6942, "step": 825 }, { "epoch": 1.08, "learning_rate": 4.5885e-05, "loss": 0.6749, "step": 826 }, { "epoch": 1.08, "learning_rate": 4.588e-05, "loss": 0.5475, "step": 827 }, { "epoch": 1.08, "learning_rate": 4.5875000000000004e-05, "loss": 0.5528, "step": 828 }, { "epoch": 1.09, "learning_rate": 4.587e-05, "loss": 0.7676, "step": 829 }, { "epoch": 1.09, "learning_rate": 4.5865e-05, "loss": 0.6449, "step": 830 }, { "epoch": 1.09, "learning_rate": 4.5860000000000005e-05, "loss": 0.5352, "step": 831 }, { "epoch": 1.09, "learning_rate": 4.5855e-05, "loss": 0.6775, "step": 832 }, { "epoch": 1.09, "learning_rate": 4.585e-05, "loss": 0.5045, "step": 833 }, { "epoch": 1.09, "learning_rate": 4.5845000000000006e-05, "loss": 0.6284, "step": 834 }, { "epoch": 1.09, "learning_rate": 4.584e-05, "loss": 0.5957, "step": 835 }, { "epoch": 1.09, "learning_rate": 4.5835e-05, "loss": 0.5215, "step": 836 }, { "epoch": 1.1, "learning_rate": 4.583e-05, "loss": 0.6045, "step": 837 }, { "epoch": 1.1, "learning_rate": 4.5825e-05, "loss": 0.5606, "step": 838 }, { "epoch": 1.1, "learning_rate": 4.5820000000000005e-05, "loss": 0.647, "step": 839 }, { "epoch": 1.1, "learning_rate": 4.5815e-05, "loss": 0.6046, "step": 840 }, { "epoch": 1.1, "learning_rate": 4.5810000000000004e-05, "loss": 0.73, "step": 841 }, { "epoch": 1.1, "learning_rate": 4.5805000000000006e-05, "loss": 0.7749, "step": 842 }, { "epoch": 1.1, "learning_rate": 4.58e-05, "loss": 0.6999, "step": 843 }, { "epoch": 1.1, "learning_rate": 4.5795000000000005e-05, "loss": 0.6506, "step": 844 }, { "epoch": 1.11, "learning_rate": 4.579e-05, "loss": 0.6221, "step": 845 }, { "epoch": 1.11, "learning_rate": 4.5785e-05, "loss": 0.5783, "step": 846 }, { "epoch": 1.11, "learning_rate": 4.578e-05, "loss": 0.5429, "step": 847 }, { "epoch": 1.11, "learning_rate": 4.5775e-05, "loss": 0.5391, "step": 848 }, { "epoch": 1.11, "learning_rate": 4.5770000000000004e-05, "loss": 0.5789, "step": 849 }, { "epoch": 1.11, "learning_rate": 4.5765e-05, "loss": 0.6742, "step": 850 }, { "epoch": 1.11, "learning_rate": 4.576e-05, "loss": 0.5768, "step": 851 }, { "epoch": 1.12, "learning_rate": 4.5755000000000005e-05, "loss": 0.6935, "step": 852 }, { "epoch": 1.12, "learning_rate": 4.575e-05, "loss": 0.7472, "step": 853 }, { "epoch": 1.12, "learning_rate": 4.5745e-05, "loss": 0.5289, "step": 854 }, { "epoch": 1.12, "learning_rate": 4.574e-05, "loss": 0.636, "step": 855 }, { "epoch": 1.12, "learning_rate": 4.5735e-05, "loss": 0.6183, "step": 856 }, { "epoch": 1.12, "learning_rate": 4.573e-05, "loss": 0.6353, "step": 857 }, { "epoch": 1.12, "learning_rate": 4.5725e-05, "loss": 0.5963, "step": 858 }, { "epoch": 1.12, "learning_rate": 4.572e-05, "loss": 0.5858, "step": 859 }, { "epoch": 1.13, "learning_rate": 4.5715000000000005e-05, "loss": 0.6064, "step": 860 }, { "epoch": 1.13, "learning_rate": 4.571e-05, "loss": 0.6058, "step": 861 }, { "epoch": 1.13, "learning_rate": 4.5705000000000004e-05, "loss": 0.5529, "step": 862 }, { "epoch": 1.13, "learning_rate": 4.5700000000000006e-05, "loss": 0.5465, "step": 863 }, { "epoch": 1.13, "learning_rate": 4.5695e-05, "loss": 0.5659, "step": 864 }, { "epoch": 1.13, "learning_rate": 4.569e-05, "loss": 0.5788, "step": 865 }, { "epoch": 1.13, "learning_rate": 4.5685e-05, "loss": 0.7236, "step": 866 }, { "epoch": 1.13, "learning_rate": 4.568e-05, "loss": 0.4999, "step": 867 }, { "epoch": 1.14, "learning_rate": 4.5675e-05, "loss": 0.5608, "step": 868 }, { "epoch": 1.14, "learning_rate": 4.567e-05, "loss": 0.5639, "step": 869 }, { "epoch": 1.14, "learning_rate": 4.5665000000000004e-05, "loss": 0.7227, "step": 870 }, { "epoch": 1.14, "learning_rate": 4.566e-05, "loss": 0.6908, "step": 871 }, { "epoch": 1.14, "learning_rate": 4.5655e-05, "loss": 0.6038, "step": 872 }, { "epoch": 1.14, "learning_rate": 4.5650000000000005e-05, "loss": 0.6251, "step": 873 }, { "epoch": 1.14, "learning_rate": 4.564500000000001e-05, "loss": 0.7227, "step": 874 }, { "epoch": 1.15, "learning_rate": 4.564e-05, "loss": 0.6878, "step": 875 }, { "epoch": 1.15, "learning_rate": 4.5635e-05, "loss": 0.6429, "step": 876 }, { "epoch": 1.15, "learning_rate": 4.563e-05, "loss": 0.5482, "step": 877 }, { "epoch": 1.15, "learning_rate": 4.5625e-05, "loss": 0.5715, "step": 878 }, { "epoch": 1.15, "learning_rate": 4.562e-05, "loss": 0.6107, "step": 879 }, { "epoch": 1.15, "learning_rate": 4.5615e-05, "loss": 0.6337, "step": 880 }, { "epoch": 1.15, "learning_rate": 4.5610000000000005e-05, "loss": 0.5822, "step": 881 }, { "epoch": 1.15, "learning_rate": 4.5605e-05, "loss": 0.6877, "step": 882 }, { "epoch": 1.16, "learning_rate": 4.5600000000000004e-05, "loss": 0.5343, "step": 883 }, { "epoch": 1.16, "learning_rate": 4.5595000000000006e-05, "loss": 0.6461, "step": 884 }, { "epoch": 1.16, "learning_rate": 4.559e-05, "loss": 0.5533, "step": 885 }, { "epoch": 1.16, "learning_rate": 4.5585e-05, "loss": 0.623, "step": 886 }, { "epoch": 1.16, "learning_rate": 4.558e-05, "loss": 0.6016, "step": 887 }, { "epoch": 1.16, "learning_rate": 4.5575e-05, "loss": 0.4784, "step": 888 }, { "epoch": 1.16, "learning_rate": 4.557e-05, "loss": 0.5117, "step": 889 }, { "epoch": 1.16, "learning_rate": 4.5565e-05, "loss": 0.5142, "step": 890 }, { "epoch": 1.17, "learning_rate": 4.5560000000000004e-05, "loss": 0.5544, "step": 891 }, { "epoch": 1.17, "learning_rate": 4.5555e-05, "loss": 0.5979, "step": 892 }, { "epoch": 1.17, "learning_rate": 4.555e-05, "loss": 0.6488, "step": 893 }, { "epoch": 1.17, "learning_rate": 4.5545000000000005e-05, "loss": 0.589, "step": 894 }, { "epoch": 1.17, "learning_rate": 4.554000000000001e-05, "loss": 0.5574, "step": 895 }, { "epoch": 1.17, "learning_rate": 4.5535e-05, "loss": 0.5755, "step": 896 }, { "epoch": 1.17, "learning_rate": 4.553e-05, "loss": 0.578, "step": 897 }, { "epoch": 1.18, "learning_rate": 4.5525e-05, "loss": 0.6892, "step": 898 }, { "epoch": 1.18, "learning_rate": 4.5520000000000005e-05, "loss": 0.6012, "step": 899 }, { "epoch": 1.18, "learning_rate": 4.5515e-05, "loss": 0.5878, "step": 900 }, { "epoch": 1.18, "learning_rate": 4.551e-05, "loss": 0.6805, "step": 901 }, { "epoch": 1.18, "learning_rate": 4.5505000000000006e-05, "loss": 0.5334, "step": 902 }, { "epoch": 1.18, "learning_rate": 4.55e-05, "loss": 0.5013, "step": 903 }, { "epoch": 1.18, "learning_rate": 4.5495000000000004e-05, "loss": 0.8237, "step": 904 }, { "epoch": 1.18, "learning_rate": 4.549000000000001e-05, "loss": 0.6894, "step": 905 }, { "epoch": 1.19, "learning_rate": 4.5485e-05, "loss": 0.5776, "step": 906 }, { "epoch": 1.19, "learning_rate": 4.548e-05, "loss": 0.7534, "step": 907 }, { "epoch": 1.19, "learning_rate": 4.5475e-05, "loss": 0.8477, "step": 908 }, { "epoch": 1.19, "learning_rate": 4.5470000000000003e-05, "loss": 0.5592, "step": 909 }, { "epoch": 1.19, "learning_rate": 4.5465e-05, "loss": 0.5704, "step": 910 }, { "epoch": 1.19, "learning_rate": 4.546e-05, "loss": 0.5608, "step": 911 }, { "epoch": 1.19, "learning_rate": 4.5455000000000004e-05, "loss": 0.7286, "step": 912 }, { "epoch": 1.2, "learning_rate": 4.545000000000001e-05, "loss": 0.5547, "step": 913 }, { "epoch": 1.2, "learning_rate": 4.5445e-05, "loss": 0.6314, "step": 914 }, { "epoch": 1.2, "learning_rate": 4.5440000000000005e-05, "loss": 0.5182, "step": 915 }, { "epoch": 1.2, "learning_rate": 4.5435e-05, "loss": 0.6636, "step": 916 }, { "epoch": 1.2, "learning_rate": 4.543e-05, "loss": 0.564, "step": 917 }, { "epoch": 1.2, "learning_rate": 4.5425e-05, "loss": 0.5926, "step": 918 }, { "epoch": 1.2, "learning_rate": 4.542e-05, "loss": 0.5602, "step": 919 }, { "epoch": 1.2, "learning_rate": 4.5415000000000005e-05, "loss": 0.7353, "step": 920 }, { "epoch": 1.21, "learning_rate": 4.541e-05, "loss": 0.6411, "step": 921 }, { "epoch": 1.21, "learning_rate": 4.5405e-05, "loss": 0.6486, "step": 922 }, { "epoch": 1.21, "learning_rate": 4.5400000000000006e-05, "loss": 0.7203, "step": 923 }, { "epoch": 1.21, "learning_rate": 4.5395e-05, "loss": 0.6559, "step": 924 }, { "epoch": 1.21, "learning_rate": 4.5390000000000004e-05, "loss": 0.6701, "step": 925 }, { "epoch": 1.21, "learning_rate": 4.5385e-05, "loss": 0.5261, "step": 926 }, { "epoch": 1.21, "learning_rate": 4.538e-05, "loss": 0.6567, "step": 927 }, { "epoch": 1.21, "learning_rate": 4.5375e-05, "loss": 0.7458, "step": 928 }, { "epoch": 1.22, "learning_rate": 4.537e-05, "loss": 0.4503, "step": 929 }, { "epoch": 1.22, "learning_rate": 4.5365000000000004e-05, "loss": 0.6208, "step": 930 }, { "epoch": 1.22, "learning_rate": 4.536e-05, "loss": 0.6111, "step": 931 }, { "epoch": 1.22, "learning_rate": 4.5355e-05, "loss": 0.8205, "step": 932 }, { "epoch": 1.22, "learning_rate": 4.5350000000000005e-05, "loss": 0.466, "step": 933 }, { "epoch": 1.22, "learning_rate": 4.534500000000001e-05, "loss": 0.5165, "step": 934 }, { "epoch": 1.22, "learning_rate": 4.534e-05, "loss": 0.6131, "step": 935 }, { "epoch": 1.23, "learning_rate": 4.5335e-05, "loss": 0.5616, "step": 936 }, { "epoch": 1.23, "learning_rate": 4.533e-05, "loss": 0.687, "step": 937 }, { "epoch": 1.23, "learning_rate": 4.5325000000000004e-05, "loss": 0.5673, "step": 938 }, { "epoch": 1.23, "learning_rate": 4.532e-05, "loss": 0.6733, "step": 939 }, { "epoch": 1.23, "learning_rate": 4.5315e-05, "loss": 0.6051, "step": 940 }, { "epoch": 1.23, "learning_rate": 4.5310000000000005e-05, "loss": 0.6232, "step": 941 }, { "epoch": 1.23, "learning_rate": 4.5305e-05, "loss": 0.6182, "step": 942 }, { "epoch": 1.23, "learning_rate": 4.53e-05, "loss": 0.6927, "step": 943 }, { "epoch": 1.24, "learning_rate": 4.5295000000000006e-05, "loss": 0.4954, "step": 944 }, { "epoch": 1.24, "learning_rate": 4.529e-05, "loss": 0.5703, "step": 945 }, { "epoch": 1.24, "learning_rate": 4.5285e-05, "loss": 0.6132, "step": 946 }, { "epoch": 1.24, "learning_rate": 4.528e-05, "loss": 0.4792, "step": 947 }, { "epoch": 1.24, "learning_rate": 4.5275e-05, "loss": 0.6449, "step": 948 }, { "epoch": 1.24, "learning_rate": 4.527e-05, "loss": 0.817, "step": 949 }, { "epoch": 1.24, "learning_rate": 4.5265e-05, "loss": 0.4237, "step": 950 }, { "epoch": 1.24, "learning_rate": 4.5260000000000004e-05, "loss": 0.4814, "step": 951 }, { "epoch": 1.25, "learning_rate": 4.5255000000000006e-05, "loss": 0.6678, "step": 952 }, { "epoch": 1.25, "learning_rate": 4.525e-05, "loss": 0.5536, "step": 953 }, { "epoch": 1.25, "learning_rate": 4.5245000000000005e-05, "loss": 0.5041, "step": 954 }, { "epoch": 1.25, "learning_rate": 4.524000000000001e-05, "loss": 0.6548, "step": 955 }, { "epoch": 1.25, "learning_rate": 4.5234999999999996e-05, "loss": 0.8377, "step": 956 }, { "epoch": 1.25, "learning_rate": 4.523e-05, "loss": 0.5742, "step": 957 }, { "epoch": 1.25, "learning_rate": 4.5225e-05, "loss": 0.6514, "step": 958 }, { "epoch": 1.26, "learning_rate": 4.5220000000000004e-05, "loss": 0.5995, "step": 959 }, { "epoch": 1.26, "learning_rate": 4.5215e-05, "loss": 0.5399, "step": 960 }, { "epoch": 1.26, "learning_rate": 4.521e-05, "loss": 0.5508, "step": 961 }, { "epoch": 1.26, "learning_rate": 4.5205000000000005e-05, "loss": 0.7065, "step": 962 }, { "epoch": 1.26, "learning_rate": 4.52e-05, "loss": 0.5331, "step": 963 }, { "epoch": 1.26, "learning_rate": 4.5195000000000004e-05, "loss": 0.5958, "step": 964 }, { "epoch": 1.26, "learning_rate": 4.5190000000000006e-05, "loss": 0.7665, "step": 965 }, { "epoch": 1.26, "learning_rate": 4.5185e-05, "loss": 0.6617, "step": 966 }, { "epoch": 1.27, "learning_rate": 4.518e-05, "loss": 0.6051, "step": 967 }, { "epoch": 1.27, "learning_rate": 4.5175e-05, "loss": 0.637, "step": 968 }, { "epoch": 1.27, "learning_rate": 4.517e-05, "loss": 0.7131, "step": 969 }, { "epoch": 1.27, "learning_rate": 4.5165e-05, "loss": 0.5297, "step": 970 }, { "epoch": 1.27, "learning_rate": 4.516e-05, "loss": 0.6778, "step": 971 }, { "epoch": 1.27, "learning_rate": 4.5155000000000004e-05, "loss": 0.6273, "step": 972 }, { "epoch": 1.27, "learning_rate": 4.5150000000000006e-05, "loss": 0.6467, "step": 973 }, { "epoch": 1.27, "learning_rate": 4.5145e-05, "loss": 0.5666, "step": 974 }, { "epoch": 1.28, "learning_rate": 4.5140000000000005e-05, "loss": 0.9065, "step": 975 }, { "epoch": 1.28, "learning_rate": 4.5135e-05, "loss": 0.5052, "step": 976 }, { "epoch": 1.28, "learning_rate": 4.513e-05, "loss": 0.6579, "step": 977 }, { "epoch": 1.28, "learning_rate": 4.5125e-05, "loss": 0.5921, "step": 978 }, { "epoch": 1.28, "learning_rate": 4.512e-05, "loss": 0.6317, "step": 979 }, { "epoch": 1.28, "learning_rate": 4.5115000000000004e-05, "loss": 0.6058, "step": 980 }, { "epoch": 1.28, "learning_rate": 4.511e-05, "loss": 0.5437, "step": 981 }, { "epoch": 1.29, "learning_rate": 4.5105e-05, "loss": 0.5201, "step": 982 }, { "epoch": 1.29, "learning_rate": 4.5100000000000005e-05, "loss": 0.4862, "step": 983 }, { "epoch": 1.29, "learning_rate": 4.5095e-05, "loss": 0.5613, "step": 984 }, { "epoch": 1.29, "learning_rate": 4.5090000000000004e-05, "loss": 0.637, "step": 985 }, { "epoch": 1.29, "learning_rate": 4.5085e-05, "loss": 0.6602, "step": 986 }, { "epoch": 1.29, "learning_rate": 4.508e-05, "loss": 0.5879, "step": 987 }, { "epoch": 1.29, "learning_rate": 4.5075e-05, "loss": 0.5699, "step": 988 }, { "epoch": 1.29, "learning_rate": 4.507e-05, "loss": 0.6436, "step": 989 }, { "epoch": 1.3, "learning_rate": 4.5065e-05, "loss": 0.516, "step": 990 }, { "epoch": 1.3, "learning_rate": 4.506e-05, "loss": 0.6631, "step": 991 }, { "epoch": 1.3, "learning_rate": 4.5055e-05, "loss": 0.5491, "step": 992 }, { "epoch": 1.3, "learning_rate": 4.5050000000000004e-05, "loss": 0.6844, "step": 993 }, { "epoch": 1.3, "learning_rate": 4.504500000000001e-05, "loss": 0.6052, "step": 994 }, { "epoch": 1.3, "learning_rate": 4.504e-05, "loss": 0.6889, "step": 995 }, { "epoch": 1.3, "learning_rate": 4.5035e-05, "loss": 0.5212, "step": 996 }, { "epoch": 1.3, "learning_rate": 4.503e-05, "loss": 0.5889, "step": 997 }, { "epoch": 1.31, "learning_rate": 4.5025000000000003e-05, "loss": 0.5359, "step": 998 }, { "epoch": 1.31, "learning_rate": 4.502e-05, "loss": 0.7551, "step": 999 }, { "epoch": 1.31, "learning_rate": 4.5015e-05, "loss": 0.5294, "step": 1000 }, { "epoch": 1.31, "learning_rate": 4.5010000000000004e-05, "loss": 0.572, "step": 1001 }, { "epoch": 1.31, "learning_rate": 4.5005e-05, "loss": 0.6195, "step": 1002 }, { "epoch": 1.31, "learning_rate": 4.5e-05, "loss": 0.5311, "step": 1003 }, { "epoch": 1.31, "learning_rate": 4.4995000000000005e-05, "loss": 0.613, "step": 1004 }, { "epoch": 1.32, "learning_rate": 4.499e-05, "loss": 0.6714, "step": 1005 }, { "epoch": 1.32, "learning_rate": 4.4985000000000004e-05, "loss": 0.7039, "step": 1006 }, { "epoch": 1.32, "learning_rate": 4.498e-05, "loss": 0.5115, "step": 1007 }, { "epoch": 1.32, "learning_rate": 4.4975e-05, "loss": 0.6861, "step": 1008 }, { "epoch": 1.32, "learning_rate": 4.497e-05, "loss": 0.5338, "step": 1009 }, { "epoch": 1.32, "learning_rate": 4.4965e-05, "loss": 0.6671, "step": 1010 }, { "epoch": 1.32, "learning_rate": 4.496e-05, "loss": 0.7405, "step": 1011 }, { "epoch": 1.32, "learning_rate": 4.4955000000000006e-05, "loss": 0.8011, "step": 1012 }, { "epoch": 1.33, "learning_rate": 4.495e-05, "loss": 0.5703, "step": 1013 }, { "epoch": 1.33, "learning_rate": 4.4945000000000004e-05, "loss": 0.4499, "step": 1014 }, { "epoch": 1.33, "learning_rate": 4.494000000000001e-05, "loss": 0.6003, "step": 1015 }, { "epoch": 1.33, "learning_rate": 4.4935e-05, "loss": 0.656, "step": 1016 }, { "epoch": 1.33, "learning_rate": 4.493e-05, "loss": 0.7386, "step": 1017 }, { "epoch": 1.33, "learning_rate": 4.4925e-05, "loss": 0.6519, "step": 1018 }, { "epoch": 1.33, "learning_rate": 4.4920000000000004e-05, "loss": 0.603, "step": 1019 }, { "epoch": 1.34, "learning_rate": 4.4915e-05, "loss": 0.5307, "step": 1020 }, { "epoch": 1.34, "learning_rate": 4.491e-05, "loss": 0.6138, "step": 1021 }, { "epoch": 1.34, "learning_rate": 4.4905000000000005e-05, "loss": 0.6346, "step": 1022 }, { "epoch": 1.34, "learning_rate": 4.49e-05, "loss": 0.6764, "step": 1023 }, { "epoch": 1.34, "learning_rate": 4.4895e-05, "loss": 0.6808, "step": 1024 }, { "epoch": 1.34, "learning_rate": 4.4890000000000006e-05, "loss": 0.5487, "step": 1025 }, { "epoch": 1.34, "learning_rate": 4.488500000000001e-05, "loss": 0.6566, "step": 1026 }, { "epoch": 1.34, "learning_rate": 4.488e-05, "loss": 0.5869, "step": 1027 }, { "epoch": 1.35, "learning_rate": 4.4875e-05, "loss": 0.5351, "step": 1028 }, { "epoch": 1.35, "learning_rate": 4.487e-05, "loss": 0.6305, "step": 1029 }, { "epoch": 1.35, "learning_rate": 4.4865e-05, "loss": 0.6481, "step": 1030 }, { "epoch": 1.35, "learning_rate": 4.486e-05, "loss": 0.5292, "step": 1031 }, { "epoch": 1.35, "learning_rate": 4.4855e-05, "loss": 0.5262, "step": 1032 }, { "epoch": 1.35, "learning_rate": 4.4850000000000006e-05, "loss": 0.5387, "step": 1033 }, { "epoch": 1.35, "learning_rate": 4.4845e-05, "loss": 0.5864, "step": 1034 }, { "epoch": 1.35, "learning_rate": 4.4840000000000004e-05, "loss": 0.6016, "step": 1035 }, { "epoch": 1.36, "learning_rate": 4.483500000000001e-05, "loss": 0.5851, "step": 1036 }, { "epoch": 1.36, "learning_rate": 4.483e-05, "loss": 0.5576, "step": 1037 }, { "epoch": 1.36, "learning_rate": 4.4825e-05, "loss": 0.7418, "step": 1038 }, { "epoch": 1.36, "learning_rate": 4.482e-05, "loss": 0.617, "step": 1039 }, { "epoch": 1.36, "learning_rate": 4.4815000000000004e-05, "loss": 0.6897, "step": 1040 }, { "epoch": 1.36, "learning_rate": 4.481e-05, "loss": 0.6209, "step": 1041 }, { "epoch": 1.36, "learning_rate": 4.4805e-05, "loss": 0.5322, "step": 1042 }, { "epoch": 1.37, "learning_rate": 4.4800000000000005e-05, "loss": 0.8139, "step": 1043 }, { "epoch": 1.37, "learning_rate": 4.4795e-05, "loss": 0.5749, "step": 1044 }, { "epoch": 1.37, "learning_rate": 4.479e-05, "loss": 0.6793, "step": 1045 }, { "epoch": 1.37, "learning_rate": 4.4785000000000006e-05, "loss": 0.6486, "step": 1046 }, { "epoch": 1.37, "learning_rate": 4.478e-05, "loss": 0.6467, "step": 1047 }, { "epoch": 1.37, "learning_rate": 4.4775e-05, "loss": 0.5946, "step": 1048 }, { "epoch": 1.37, "learning_rate": 4.477e-05, "loss": 0.5921, "step": 1049 }, { "epoch": 1.37, "learning_rate": 4.4765e-05, "loss": 0.653, "step": 1050 }, { "epoch": 1.38, "learning_rate": 4.4760000000000005e-05, "loss": 0.6626, "step": 1051 }, { "epoch": 1.38, "learning_rate": 4.4755e-05, "loss": 0.4525, "step": 1052 }, { "epoch": 1.38, "learning_rate": 4.4750000000000004e-05, "loss": 0.619, "step": 1053 }, { "epoch": 1.38, "learning_rate": 4.4745000000000006e-05, "loss": 0.7734, "step": 1054 }, { "epoch": 1.38, "learning_rate": 4.474e-05, "loss": 0.6881, "step": 1055 }, { "epoch": 1.38, "learning_rate": 4.4735000000000005e-05, "loss": 0.7872, "step": 1056 }, { "epoch": 1.38, "learning_rate": 4.473e-05, "loss": 0.6717, "step": 1057 }, { "epoch": 1.38, "learning_rate": 4.4725e-05, "loss": 0.4917, "step": 1058 }, { "epoch": 1.39, "learning_rate": 4.472e-05, "loss": 0.827, "step": 1059 }, { "epoch": 1.39, "learning_rate": 4.4715e-05, "loss": 0.6613, "step": 1060 }, { "epoch": 1.39, "learning_rate": 4.4710000000000004e-05, "loss": 0.5802, "step": 1061 }, { "epoch": 1.39, "learning_rate": 4.4705e-05, "loss": 0.6985, "step": 1062 }, { "epoch": 1.39, "learning_rate": 4.47e-05, "loss": 0.54, "step": 1063 }, { "epoch": 1.39, "learning_rate": 4.4695000000000005e-05, "loss": 0.753, "step": 1064 }, { "epoch": 1.39, "learning_rate": 4.469e-05, "loss": 0.511, "step": 1065 }, { "epoch": 1.4, "learning_rate": 4.4685e-05, "loss": 0.7011, "step": 1066 }, { "epoch": 1.4, "learning_rate": 4.468e-05, "loss": 0.5172, "step": 1067 }, { "epoch": 1.4, "learning_rate": 4.4675e-05, "loss": 0.7054, "step": 1068 }, { "epoch": 1.4, "learning_rate": 4.467e-05, "loss": 0.6131, "step": 1069 }, { "epoch": 1.4, "learning_rate": 4.4665e-05, "loss": 0.7326, "step": 1070 }, { "epoch": 1.4, "learning_rate": 4.466e-05, "loss": 0.8679, "step": 1071 }, { "epoch": 1.4, "learning_rate": 4.4655000000000005e-05, "loss": 0.5562, "step": 1072 }, { "epoch": 1.4, "learning_rate": 4.465e-05, "loss": 0.6029, "step": 1073 }, { "epoch": 1.41, "learning_rate": 4.4645000000000004e-05, "loss": 0.6284, "step": 1074 }, { "epoch": 1.41, "learning_rate": 4.4640000000000006e-05, "loss": 0.4807, "step": 1075 }, { "epoch": 1.41, "learning_rate": 4.4635e-05, "loss": 0.7166, "step": 1076 }, { "epoch": 1.41, "learning_rate": 4.463e-05, "loss": 0.5568, "step": 1077 }, { "epoch": 1.41, "learning_rate": 4.4625e-05, "loss": 0.5538, "step": 1078 }, { "epoch": 1.41, "learning_rate": 4.462e-05, "loss": 0.6075, "step": 1079 }, { "epoch": 1.41, "learning_rate": 4.4615e-05, "loss": 0.5625, "step": 1080 }, { "epoch": 1.41, "learning_rate": 4.461e-05, "loss": 0.596, "step": 1081 }, { "epoch": 1.42, "learning_rate": 4.4605000000000004e-05, "loss": 0.6121, "step": 1082 }, { "epoch": 1.42, "learning_rate": 4.46e-05, "loss": 0.5124, "step": 1083 }, { "epoch": 1.42, "learning_rate": 4.4595e-05, "loss": 0.5353, "step": 1084 }, { "epoch": 1.42, "learning_rate": 4.4590000000000005e-05, "loss": 0.7318, "step": 1085 }, { "epoch": 1.42, "learning_rate": 4.458500000000001e-05, "loss": 0.5736, "step": 1086 }, { "epoch": 1.42, "learning_rate": 4.458e-05, "loss": 0.7325, "step": 1087 }, { "epoch": 1.42, "learning_rate": 4.4575e-05, "loss": 0.6344, "step": 1088 }, { "epoch": 1.43, "learning_rate": 4.457e-05, "loss": 0.5642, "step": 1089 }, { "epoch": 1.43, "learning_rate": 4.4565000000000004e-05, "loss": 0.4136, "step": 1090 }, { "epoch": 1.43, "learning_rate": 4.456e-05, "loss": 0.9219, "step": 1091 }, { "epoch": 1.43, "learning_rate": 4.4555e-05, "loss": 0.5998, "step": 1092 }, { "epoch": 1.43, "learning_rate": 4.4550000000000005e-05, "loss": 0.5999, "step": 1093 }, { "epoch": 1.43, "learning_rate": 4.4545e-05, "loss": 0.7621, "step": 1094 }, { "epoch": 1.43, "learning_rate": 4.4540000000000004e-05, "loss": 0.6213, "step": 1095 }, { "epoch": 1.43, "learning_rate": 4.4535000000000006e-05, "loss": 0.6006, "step": 1096 }, { "epoch": 1.44, "learning_rate": 4.453e-05, "loss": 0.4961, "step": 1097 }, { "epoch": 1.44, "learning_rate": 4.4525e-05, "loss": 0.5485, "step": 1098 }, { "epoch": 1.44, "learning_rate": 4.452e-05, "loss": 0.5146, "step": 1099 }, { "epoch": 1.44, "learning_rate": 4.4515e-05, "loss": 0.6867, "step": 1100 }, { "epoch": 1.44, "learning_rate": 4.451e-05, "loss": 0.5842, "step": 1101 }, { "epoch": 1.44, "learning_rate": 4.4505e-05, "loss": 0.5015, "step": 1102 }, { "epoch": 1.44, "learning_rate": 4.4500000000000004e-05, "loss": 0.7086, "step": 1103 }, { "epoch": 1.45, "learning_rate": 4.4495e-05, "loss": 0.6456, "step": 1104 }, { "epoch": 1.45, "learning_rate": 4.449e-05, "loss": 0.7506, "step": 1105 }, { "epoch": 1.45, "learning_rate": 4.4485000000000005e-05, "loss": 0.6325, "step": 1106 }, { "epoch": 1.45, "learning_rate": 4.448e-05, "loss": 0.5829, "step": 1107 }, { "epoch": 1.45, "learning_rate": 4.4475e-05, "loss": 0.9496, "step": 1108 }, { "epoch": 1.45, "learning_rate": 4.447e-05, "loss": 0.6981, "step": 1109 }, { "epoch": 1.45, "learning_rate": 4.4465e-05, "loss": 0.6571, "step": 1110 }, { "epoch": 1.45, "learning_rate": 4.4460000000000005e-05, "loss": 0.7308, "step": 1111 }, { "epoch": 1.46, "learning_rate": 4.4455e-05, "loss": 0.6724, "step": 1112 }, { "epoch": 1.46, "learning_rate": 4.445e-05, "loss": 0.6601, "step": 1113 }, { "epoch": 1.46, "learning_rate": 4.4445000000000006e-05, "loss": 0.5423, "step": 1114 }, { "epoch": 1.46, "learning_rate": 4.444e-05, "loss": 0.6329, "step": 1115 }, { "epoch": 1.46, "learning_rate": 4.4435000000000004e-05, "loss": 0.6298, "step": 1116 }, { "epoch": 1.46, "learning_rate": 4.443e-05, "loss": 0.6714, "step": 1117 }, { "epoch": 1.46, "learning_rate": 4.4425e-05, "loss": 0.5593, "step": 1118 }, { "epoch": 1.46, "learning_rate": 4.442e-05, "loss": 0.5731, "step": 1119 }, { "epoch": 1.47, "learning_rate": 4.4415e-05, "loss": 0.6683, "step": 1120 }, { "epoch": 1.47, "learning_rate": 4.4410000000000003e-05, "loss": 0.6129, "step": 1121 }, { "epoch": 1.47, "learning_rate": 4.4405e-05, "loss": 0.574, "step": 1122 }, { "epoch": 1.47, "learning_rate": 4.44e-05, "loss": 0.7732, "step": 1123 }, { "epoch": 1.47, "learning_rate": 4.4395000000000004e-05, "loss": 0.5939, "step": 1124 }, { "epoch": 1.47, "learning_rate": 4.439000000000001e-05, "loss": 0.5883, "step": 1125 }, { "epoch": 1.47, "learning_rate": 4.4385e-05, "loss": 0.6137, "step": 1126 }, { "epoch": 1.48, "learning_rate": 4.438e-05, "loss": 0.6114, "step": 1127 }, { "epoch": 1.48, "learning_rate": 4.4375e-05, "loss": 0.7731, "step": 1128 }, { "epoch": 1.48, "learning_rate": 4.4370000000000004e-05, "loss": 0.7335, "step": 1129 }, { "epoch": 1.48, "learning_rate": 4.4365e-05, "loss": 0.594, "step": 1130 }, { "epoch": 1.48, "learning_rate": 4.436e-05, "loss": 0.606, "step": 1131 }, { "epoch": 1.48, "learning_rate": 4.4355000000000005e-05, "loss": 0.5592, "step": 1132 }, { "epoch": 1.48, "learning_rate": 4.435e-05, "loss": 0.5432, "step": 1133 }, { "epoch": 1.48, "learning_rate": 4.4345e-05, "loss": 0.6055, "step": 1134 }, { "epoch": 1.49, "learning_rate": 4.4340000000000006e-05, "loss": 0.5841, "step": 1135 }, { "epoch": 1.49, "learning_rate": 4.4335e-05, "loss": 0.61, "step": 1136 }, { "epoch": 1.49, "learning_rate": 4.4330000000000004e-05, "loss": 0.5798, "step": 1137 }, { "epoch": 1.49, "learning_rate": 4.4325e-05, "loss": 0.6405, "step": 1138 }, { "epoch": 1.49, "learning_rate": 4.432e-05, "loss": 0.5578, "step": 1139 }, { "epoch": 1.49, "learning_rate": 4.4315e-05, "loss": 0.61, "step": 1140 }, { "epoch": 1.49, "learning_rate": 4.431e-05, "loss": 0.6324, "step": 1141 }, { "epoch": 1.49, "learning_rate": 4.4305000000000004e-05, "loss": 0.8244, "step": 1142 }, { "epoch": 1.5, "learning_rate": 4.43e-05, "loss": 0.5885, "step": 1143 }, { "epoch": 1.5, "learning_rate": 4.4295e-05, "loss": 0.8334, "step": 1144 }, { "epoch": 1.5, "learning_rate": 4.4290000000000005e-05, "loss": 0.6452, "step": 1145 }, { "epoch": 1.5, "learning_rate": 4.428500000000001e-05, "loss": 0.653, "step": 1146 }, { "epoch": 1.5, "learning_rate": 4.428e-05, "loss": 0.6787, "step": 1147 }, { "epoch": 1.5, "learning_rate": 4.4275e-05, "loss": 0.6612, "step": 1148 }, { "epoch": 1.5, "learning_rate": 4.427e-05, "loss": 0.6126, "step": 1149 }, { "epoch": 1.51, "learning_rate": 4.4265000000000004e-05, "loss": 0.5891, "step": 1150 }, { "epoch": 1.51, "learning_rate": 4.426e-05, "loss": 0.6445, "step": 1151 }, { "epoch": 1.51, "learning_rate": 4.4255e-05, "loss": 0.7164, "step": 1152 }, { "epoch": 1.51, "learning_rate": 4.4250000000000005e-05, "loss": 0.7443, "step": 1153 }, { "epoch": 1.51, "learning_rate": 4.4245e-05, "loss": 0.5169, "step": 1154 }, { "epoch": 1.51, "learning_rate": 4.424e-05, "loss": 0.6542, "step": 1155 }, { "epoch": 1.51, "learning_rate": 4.4235000000000006e-05, "loss": 0.437, "step": 1156 }, { "epoch": 1.51, "learning_rate": 4.423e-05, "loss": 0.5236, "step": 1157 }, { "epoch": 1.52, "learning_rate": 4.4225e-05, "loss": 0.5756, "step": 1158 }, { "epoch": 1.52, "learning_rate": 4.422e-05, "loss": 0.6014, "step": 1159 }, { "epoch": 1.52, "learning_rate": 4.4215e-05, "loss": 0.6563, "step": 1160 }, { "epoch": 1.52, "learning_rate": 4.421e-05, "loss": 0.7307, "step": 1161 }, { "epoch": 1.52, "learning_rate": 4.4205e-05, "loss": 0.5545, "step": 1162 }, { "epoch": 1.52, "learning_rate": 4.4200000000000004e-05, "loss": 0.4994, "step": 1163 }, { "epoch": 1.52, "learning_rate": 4.4195000000000006e-05, "loss": 0.787, "step": 1164 }, { "epoch": 1.52, "learning_rate": 4.419e-05, "loss": 0.7274, "step": 1165 }, { "epoch": 1.53, "learning_rate": 4.4185000000000005e-05, "loss": 0.6336, "step": 1166 }, { "epoch": 1.53, "learning_rate": 4.418000000000001e-05, "loss": 0.5211, "step": 1167 }, { "epoch": 1.53, "learning_rate": 4.4174999999999996e-05, "loss": 0.7553, "step": 1168 }, { "epoch": 1.53, "learning_rate": 4.417e-05, "loss": 0.5703, "step": 1169 }, { "epoch": 1.53, "learning_rate": 4.4165e-05, "loss": 0.5732, "step": 1170 }, { "epoch": 1.53, "learning_rate": 4.4160000000000004e-05, "loss": 0.5925, "step": 1171 }, { "epoch": 1.53, "learning_rate": 4.4155e-05, "loss": 0.6006, "step": 1172 }, { "epoch": 1.54, "learning_rate": 4.415e-05, "loss": 0.8236, "step": 1173 }, { "epoch": 1.54, "learning_rate": 4.4145000000000005e-05, "loss": 0.5282, "step": 1174 }, { "epoch": 1.54, "learning_rate": 4.414e-05, "loss": 0.7734, "step": 1175 }, { "epoch": 1.54, "learning_rate": 4.4135000000000003e-05, "loss": 0.6717, "step": 1176 }, { "epoch": 1.54, "learning_rate": 4.4130000000000006e-05, "loss": 0.7127, "step": 1177 }, { "epoch": 1.54, "learning_rate": 4.4125e-05, "loss": 0.6315, "step": 1178 }, { "epoch": 1.54, "learning_rate": 4.412e-05, "loss": 0.6975, "step": 1179 }, { "epoch": 1.54, "learning_rate": 4.4115e-05, "loss": 0.6541, "step": 1180 }, { "epoch": 1.55, "learning_rate": 4.411e-05, "loss": 0.5614, "step": 1181 }, { "epoch": 1.55, "learning_rate": 4.4105e-05, "loss": 0.6569, "step": 1182 }, { "epoch": 1.55, "learning_rate": 4.41e-05, "loss": 0.7225, "step": 1183 }, { "epoch": 1.55, "learning_rate": 4.4095000000000004e-05, "loss": 0.5399, "step": 1184 }, { "epoch": 1.55, "learning_rate": 4.4090000000000006e-05, "loss": 0.559, "step": 1185 }, { "epoch": 1.55, "learning_rate": 4.4085e-05, "loss": 0.5803, "step": 1186 }, { "epoch": 1.55, "learning_rate": 4.4080000000000005e-05, "loss": 0.6493, "step": 1187 }, { "epoch": 1.55, "learning_rate": 4.4075e-05, "loss": 0.5793, "step": 1188 }, { "epoch": 1.56, "learning_rate": 4.407e-05, "loss": 0.63, "step": 1189 }, { "epoch": 1.56, "learning_rate": 4.4065e-05, "loss": 0.5882, "step": 1190 }, { "epoch": 1.56, "learning_rate": 4.406e-05, "loss": 0.6876, "step": 1191 }, { "epoch": 1.56, "learning_rate": 4.4055000000000004e-05, "loss": 0.5922, "step": 1192 }, { "epoch": 1.56, "learning_rate": 4.405e-05, "loss": 0.5438, "step": 1193 }, { "epoch": 1.56, "learning_rate": 4.4045e-05, "loss": 0.7443, "step": 1194 }, { "epoch": 1.56, "learning_rate": 4.4040000000000005e-05, "loss": 0.6469, "step": 1195 }, { "epoch": 1.57, "learning_rate": 4.4035e-05, "loss": 0.666, "step": 1196 }, { "epoch": 1.57, "learning_rate": 4.4030000000000004e-05, "loss": 0.6194, "step": 1197 }, { "epoch": 1.57, "learning_rate": 4.4025e-05, "loss": 0.6526, "step": 1198 }, { "epoch": 1.57, "learning_rate": 4.402e-05, "loss": 0.7105, "step": 1199 }, { "epoch": 1.57, "learning_rate": 4.4015e-05, "loss": 0.5476, "step": 1200 }, { "epoch": 1.57, "learning_rate": 4.401e-05, "loss": 0.5121, "step": 1201 }, { "epoch": 1.57, "learning_rate": 4.4005e-05, "loss": 0.6477, "step": 1202 }, { "epoch": 1.57, "learning_rate": 4.4000000000000006e-05, "loss": 0.7619, "step": 1203 }, { "epoch": 1.58, "learning_rate": 4.3995e-05, "loss": 0.7018, "step": 1204 }, { "epoch": 1.58, "learning_rate": 4.3990000000000004e-05, "loss": 0.6238, "step": 1205 }, { "epoch": 1.58, "learning_rate": 4.398500000000001e-05, "loss": 0.5166, "step": 1206 }, { "epoch": 1.58, "learning_rate": 4.398e-05, "loss": 0.6903, "step": 1207 }, { "epoch": 1.58, "learning_rate": 4.3975e-05, "loss": 0.7041, "step": 1208 }, { "epoch": 1.58, "learning_rate": 4.397e-05, "loss": 0.5905, "step": 1209 }, { "epoch": 1.58, "learning_rate": 4.3965000000000003e-05, "loss": 0.6458, "step": 1210 }, { "epoch": 1.59, "learning_rate": 4.396e-05, "loss": 0.6422, "step": 1211 }, { "epoch": 1.59, "learning_rate": 4.3955e-05, "loss": 0.5981, "step": 1212 }, { "epoch": 1.59, "learning_rate": 4.3950000000000004e-05, "loss": 0.7856, "step": 1213 }, { "epoch": 1.59, "learning_rate": 4.3945e-05, "loss": 0.6765, "step": 1214 }, { "epoch": 1.59, "learning_rate": 4.394e-05, "loss": 0.6675, "step": 1215 }, { "epoch": 1.59, "learning_rate": 4.3935000000000005e-05, "loss": 0.531, "step": 1216 }, { "epoch": 1.59, "learning_rate": 4.393e-05, "loss": 0.6272, "step": 1217 }, { "epoch": 1.59, "learning_rate": 4.3925e-05, "loss": 0.7354, "step": 1218 }, { "epoch": 1.6, "learning_rate": 4.392e-05, "loss": 0.7163, "step": 1219 }, { "epoch": 1.6, "learning_rate": 4.3915e-05, "loss": 0.5091, "step": 1220 }, { "epoch": 1.6, "learning_rate": 4.391e-05, "loss": 0.5118, "step": 1221 }, { "epoch": 1.6, "learning_rate": 4.3905e-05, "loss": 0.6357, "step": 1222 }, { "epoch": 1.6, "learning_rate": 4.39e-05, "loss": 0.6122, "step": 1223 }, { "epoch": 1.6, "learning_rate": 4.3895000000000006e-05, "loss": 0.5995, "step": 1224 }, { "epoch": 1.6, "learning_rate": 4.389e-05, "loss": 0.7743, "step": 1225 }, { "epoch": 1.6, "learning_rate": 4.3885000000000004e-05, "loss": 0.6441, "step": 1226 }, { "epoch": 1.61, "learning_rate": 4.388000000000001e-05, "loss": 0.6969, "step": 1227 }, { "epoch": 1.61, "learning_rate": 4.3875e-05, "loss": 0.8982, "step": 1228 }, { "epoch": 1.61, "learning_rate": 4.387e-05, "loss": 0.7527, "step": 1229 }, { "epoch": 1.61, "learning_rate": 4.3865e-05, "loss": 0.6091, "step": 1230 }, { "epoch": 1.61, "learning_rate": 4.3860000000000004e-05, "loss": 0.6958, "step": 1231 }, { "epoch": 1.61, "learning_rate": 4.3855e-05, "loss": 0.6484, "step": 1232 }, { "epoch": 1.61, "learning_rate": 4.385e-05, "loss": 0.6215, "step": 1233 }, { "epoch": 1.62, "learning_rate": 4.3845000000000005e-05, "loss": 0.6599, "step": 1234 }, { "epoch": 1.62, "learning_rate": 4.384e-05, "loss": 0.605, "step": 1235 }, { "epoch": 1.62, "learning_rate": 4.3835e-05, "loss": 0.4562, "step": 1236 }, { "epoch": 1.62, "learning_rate": 4.3830000000000006e-05, "loss": 0.7233, "step": 1237 }, { "epoch": 1.62, "learning_rate": 4.3825e-05, "loss": 0.6468, "step": 1238 }, { "epoch": 1.62, "learning_rate": 4.382e-05, "loss": 0.7029, "step": 1239 }, { "epoch": 1.62, "learning_rate": 4.3815e-05, "loss": 0.7883, "step": 1240 }, { "epoch": 1.62, "learning_rate": 4.381e-05, "loss": 0.5886, "step": 1241 }, { "epoch": 1.63, "learning_rate": 4.3805000000000005e-05, "loss": 0.6915, "step": 1242 }, { "epoch": 1.63, "learning_rate": 4.38e-05, "loss": 0.515, "step": 1243 }, { "epoch": 1.63, "learning_rate": 4.3795e-05, "loss": 0.5367, "step": 1244 }, { "epoch": 1.63, "learning_rate": 4.3790000000000006e-05, "loss": 0.5799, "step": 1245 }, { "epoch": 1.63, "learning_rate": 4.3785e-05, "loss": 0.7674, "step": 1246 }, { "epoch": 1.63, "learning_rate": 4.3780000000000004e-05, "loss": 0.7472, "step": 1247 }, { "epoch": 1.63, "learning_rate": 4.3775e-05, "loss": 0.624, "step": 1248 }, { "epoch": 1.63, "learning_rate": 4.377e-05, "loss": 0.5452, "step": 1249 }, { "epoch": 1.64, "learning_rate": 4.3765e-05, "loss": 0.5385, "step": 1250 }, { "epoch": 1.64, "learning_rate": 4.376e-05, "loss": 0.5707, "step": 1251 }, { "epoch": 1.64, "learning_rate": 4.3755000000000004e-05, "loss": 0.7019, "step": 1252 }, { "epoch": 1.64, "learning_rate": 4.375e-05, "loss": 0.6106, "step": 1253 }, { "epoch": 1.64, "learning_rate": 4.3745e-05, "loss": 0.7644, "step": 1254 }, { "epoch": 1.64, "learning_rate": 4.3740000000000005e-05, "loss": 0.7192, "step": 1255 }, { "epoch": 1.64, "learning_rate": 4.3735e-05, "loss": 0.7011, "step": 1256 }, { "epoch": 1.65, "learning_rate": 4.373e-05, "loss": 0.4966, "step": 1257 }, { "epoch": 1.65, "learning_rate": 4.3725000000000006e-05, "loss": 0.7213, "step": 1258 }, { "epoch": 1.65, "learning_rate": 4.372e-05, "loss": 0.6475, "step": 1259 }, { "epoch": 1.65, "learning_rate": 4.3715e-05, "loss": 0.6182, "step": 1260 }, { "epoch": 1.65, "learning_rate": 4.371e-05, "loss": 0.7578, "step": 1261 }, { "epoch": 1.65, "learning_rate": 4.3705e-05, "loss": 0.5094, "step": 1262 }, { "epoch": 1.65, "learning_rate": 4.3700000000000005e-05, "loss": 0.4847, "step": 1263 }, { "epoch": 1.65, "learning_rate": 4.3695e-05, "loss": 0.6145, "step": 1264 }, { "epoch": 1.66, "learning_rate": 4.3690000000000004e-05, "loss": 0.6436, "step": 1265 }, { "epoch": 1.66, "learning_rate": 4.3685000000000006e-05, "loss": 0.6649, "step": 1266 }, { "epoch": 1.66, "learning_rate": 4.368e-05, "loss": 0.6523, "step": 1267 }, { "epoch": 1.66, "learning_rate": 4.3675000000000005e-05, "loss": 0.5292, "step": 1268 }, { "epoch": 1.66, "learning_rate": 4.367e-05, "loss": 0.5953, "step": 1269 }, { "epoch": 1.66, "learning_rate": 4.3665e-05, "loss": 0.6079, "step": 1270 }, { "epoch": 1.66, "learning_rate": 4.366e-05, "loss": 0.7093, "step": 1271 }, { "epoch": 1.66, "learning_rate": 4.3655e-05, "loss": 0.6169, "step": 1272 }, { "epoch": 1.67, "learning_rate": 4.3650000000000004e-05, "loss": 0.6944, "step": 1273 }, { "epoch": 1.67, "learning_rate": 4.3645e-05, "loss": 0.5184, "step": 1274 }, { "epoch": 1.67, "learning_rate": 4.364e-05, "loss": 0.7321, "step": 1275 }, { "epoch": 1.67, "learning_rate": 4.3635000000000005e-05, "loss": 0.7252, "step": 1276 }, { "epoch": 1.67, "learning_rate": 4.363000000000001e-05, "loss": 0.6179, "step": 1277 }, { "epoch": 1.67, "learning_rate": 4.3625e-05, "loss": 0.8616, "step": 1278 }, { "epoch": 1.67, "learning_rate": 4.362e-05, "loss": 0.5379, "step": 1279 }, { "epoch": 1.68, "learning_rate": 4.3615e-05, "loss": 0.6418, "step": 1280 }, { "epoch": 1.68, "learning_rate": 4.361e-05, "loss": 0.5853, "step": 1281 }, { "epoch": 1.68, "learning_rate": 4.3605e-05, "loss": 0.5811, "step": 1282 }, { "epoch": 1.68, "learning_rate": 4.36e-05, "loss": 0.7042, "step": 1283 }, { "epoch": 1.68, "learning_rate": 4.3595000000000005e-05, "loss": 0.5282, "step": 1284 }, { "epoch": 1.68, "learning_rate": 4.359e-05, "loss": 0.5164, "step": 1285 }, { "epoch": 1.68, "learning_rate": 4.3585000000000004e-05, "loss": 0.5158, "step": 1286 }, { "epoch": 1.68, "learning_rate": 4.3580000000000006e-05, "loss": 0.7134, "step": 1287 }, { "epoch": 1.69, "learning_rate": 4.3575e-05, "loss": 0.8047, "step": 1288 }, { "epoch": 1.69, "learning_rate": 4.357e-05, "loss": 0.6878, "step": 1289 }, { "epoch": 1.69, "learning_rate": 4.3565e-05, "loss": 0.5307, "step": 1290 }, { "epoch": 1.69, "learning_rate": 4.356e-05, "loss": 0.6064, "step": 1291 }, { "epoch": 1.69, "learning_rate": 4.3555e-05, "loss": 0.7875, "step": 1292 }, { "epoch": 1.69, "learning_rate": 4.355e-05, "loss": 0.7766, "step": 1293 }, { "epoch": 1.69, "learning_rate": 4.3545000000000004e-05, "loss": 0.6484, "step": 1294 }, { "epoch": 1.7, "learning_rate": 4.3545000000000004e-05, "loss": 0.6046, "step": 1295 }, { "epoch": 1.7, "learning_rate": 4.354e-05, "loss": 0.6554, "step": 1296 }, { "epoch": 1.7, "learning_rate": 4.3535e-05, "loss": 0.6629, "step": 1297 }, { "epoch": 1.7, "learning_rate": 4.3530000000000005e-05, "loss": 0.6729, "step": 1298 }, { "epoch": 1.7, "learning_rate": 4.352500000000001e-05, "loss": 0.6907, "step": 1299 }, { "epoch": 1.7, "learning_rate": 4.352e-05, "loss": 0.605, "step": 1300 }, { "epoch": 1.7, "learning_rate": 4.3515e-05, "loss": 0.8584, "step": 1301 }, { "epoch": 1.7, "learning_rate": 4.351e-05, "loss": 0.6486, "step": 1302 }, { "epoch": 1.71, "learning_rate": 4.3505000000000004e-05, "loss": 0.5695, "step": 1303 }, { "epoch": 1.71, "learning_rate": 4.35e-05, "loss": 0.6643, "step": 1304 }, { "epoch": 1.71, "learning_rate": 4.3495e-05, "loss": 0.6483, "step": 1305 }, { "epoch": 1.71, "learning_rate": 4.3490000000000005e-05, "loss": 0.5908, "step": 1306 }, { "epoch": 1.71, "learning_rate": 4.3485e-05, "loss": 0.4829, "step": 1307 }, { "epoch": 1.71, "learning_rate": 4.3480000000000004e-05, "loss": 0.6889, "step": 1308 }, { "epoch": 1.71, "learning_rate": 4.3475000000000006e-05, "loss": 0.5604, "step": 1309 }, { "epoch": 1.71, "learning_rate": 4.347e-05, "loss": 0.5515, "step": 1310 }, { "epoch": 1.72, "learning_rate": 4.3465e-05, "loss": 0.6175, "step": 1311 }, { "epoch": 1.72, "learning_rate": 4.346e-05, "loss": 0.5928, "step": 1312 }, { "epoch": 1.72, "learning_rate": 4.3455e-05, "loss": 0.6545, "step": 1313 }, { "epoch": 1.72, "learning_rate": 4.345e-05, "loss": 0.72, "step": 1314 }, { "epoch": 1.72, "learning_rate": 4.3445e-05, "loss": 0.5217, "step": 1315 }, { "epoch": 1.72, "learning_rate": 4.3440000000000004e-05, "loss": 0.6959, "step": 1316 }, { "epoch": 1.72, "learning_rate": 4.343500000000001e-05, "loss": 0.6716, "step": 1317 }, { "epoch": 1.73, "learning_rate": 4.343e-05, "loss": 0.5625, "step": 1318 }, { "epoch": 1.73, "learning_rate": 4.3425000000000005e-05, "loss": 0.7862, "step": 1319 }, { "epoch": 1.73, "learning_rate": 4.342e-05, "loss": 0.7901, "step": 1320 }, { "epoch": 1.73, "learning_rate": 4.3415e-05, "loss": 0.806, "step": 1321 }, { "epoch": 1.73, "learning_rate": 4.341e-05, "loss": 0.4893, "step": 1322 }, { "epoch": 1.73, "learning_rate": 4.3405e-05, "loss": 0.7184, "step": 1323 }, { "epoch": 1.73, "learning_rate": 4.3400000000000005e-05, "loss": 0.5296, "step": 1324 }, { "epoch": 1.73, "learning_rate": 4.3395e-05, "loss": 0.5851, "step": 1325 }, { "epoch": 1.74, "learning_rate": 4.339e-05, "loss": 0.5971, "step": 1326 }, { "epoch": 1.74, "learning_rate": 4.3385000000000006e-05, "loss": 0.6414, "step": 1327 }, { "epoch": 1.74, "learning_rate": 4.338e-05, "loss": 0.7059, "step": 1328 }, { "epoch": 1.74, "learning_rate": 4.3375000000000004e-05, "loss": 0.617, "step": 1329 }, { "epoch": 1.74, "learning_rate": 4.337e-05, "loss": 0.5895, "step": 1330 }, { "epoch": 1.74, "learning_rate": 4.3365e-05, "loss": 0.7116, "step": 1331 }, { "epoch": 1.74, "learning_rate": 4.336e-05, "loss": 0.6014, "step": 1332 }, { "epoch": 1.74, "learning_rate": 4.3355e-05, "loss": 0.599, "step": 1333 }, { "epoch": 1.75, "learning_rate": 4.335e-05, "loss": 0.6602, "step": 1334 }, { "epoch": 1.75, "learning_rate": 4.3345e-05, "loss": 0.5847, "step": 1335 }, { "epoch": 1.75, "learning_rate": 4.334e-05, "loss": 0.5747, "step": 1336 }, { "epoch": 1.75, "learning_rate": 4.3335000000000004e-05, "loss": 0.5997, "step": 1337 }, { "epoch": 1.75, "learning_rate": 4.333000000000001e-05, "loss": 0.6738, "step": 1338 }, { "epoch": 1.75, "learning_rate": 4.3325e-05, "loss": 0.8249, "step": 1339 }, { "epoch": 1.75, "learning_rate": 4.332e-05, "loss": 0.5425, "step": 1340 }, { "epoch": 1.76, "learning_rate": 4.3315e-05, "loss": 0.6793, "step": 1341 }, { "epoch": 1.76, "learning_rate": 4.3310000000000004e-05, "loss": 0.5728, "step": 1342 }, { "epoch": 1.76, "learning_rate": 4.3305e-05, "loss": 0.639, "step": 1343 }, { "epoch": 1.76, "learning_rate": 4.33e-05, "loss": 0.5369, "step": 1344 }, { "epoch": 1.76, "learning_rate": 4.3295000000000005e-05, "loss": 0.4427, "step": 1345 }, { "epoch": 1.76, "learning_rate": 4.329e-05, "loss": 0.5626, "step": 1346 }, { "epoch": 1.76, "learning_rate": 4.3285e-05, "loss": 0.6742, "step": 1347 }, { "epoch": 1.76, "learning_rate": 4.3280000000000006e-05, "loss": 0.6324, "step": 1348 }, { "epoch": 1.77, "learning_rate": 4.3275e-05, "loss": 0.5496, "step": 1349 }, { "epoch": 1.77, "learning_rate": 4.327e-05, "loss": 0.591, "step": 1350 }, { "epoch": 1.77, "learning_rate": 4.3265e-05, "loss": 0.7729, "step": 1351 }, { "epoch": 1.77, "learning_rate": 4.326e-05, "loss": 0.6834, "step": 1352 }, { "epoch": 1.77, "learning_rate": 4.3255e-05, "loss": 0.5762, "step": 1353 }, { "epoch": 1.77, "learning_rate": 4.325e-05, "loss": 0.5916, "step": 1354 }, { "epoch": 1.77, "learning_rate": 4.3245000000000004e-05, "loss": 0.5055, "step": 1355 }, { "epoch": 1.77, "learning_rate": 4.324e-05, "loss": 0.7113, "step": 1356 }, { "epoch": 1.78, "learning_rate": 4.3235e-05, "loss": 0.6245, "step": 1357 }, { "epoch": 1.78, "learning_rate": 4.3230000000000005e-05, "loss": 0.5859, "step": 1358 }, { "epoch": 1.78, "learning_rate": 4.322500000000001e-05, "loss": 0.4327, "step": 1359 }, { "epoch": 1.78, "learning_rate": 4.3219999999999996e-05, "loss": 0.5536, "step": 1360 }, { "epoch": 1.78, "learning_rate": 4.3215e-05, "loss": 0.5287, "step": 1361 }, { "epoch": 1.78, "learning_rate": 4.321e-05, "loss": 0.6322, "step": 1362 }, { "epoch": 1.78, "learning_rate": 4.3205000000000004e-05, "loss": 0.6076, "step": 1363 }, { "epoch": 1.79, "learning_rate": 4.32e-05, "loss": 0.5162, "step": 1364 }, { "epoch": 1.79, "learning_rate": 4.3195e-05, "loss": 0.4868, "step": 1365 }, { "epoch": 1.79, "learning_rate": 4.3190000000000005e-05, "loss": 0.6183, "step": 1366 }, { "epoch": 1.79, "learning_rate": 4.3185e-05, "loss": 0.606, "step": 1367 }, { "epoch": 1.79, "learning_rate": 4.318e-05, "loss": 0.6122, "step": 1368 }, { "epoch": 1.79, "learning_rate": 4.3175000000000006e-05, "loss": 0.5956, "step": 1369 }, { "epoch": 1.79, "learning_rate": 4.317e-05, "loss": 0.7241, "step": 1370 }, { "epoch": 1.79, "learning_rate": 4.3165e-05, "loss": 0.6037, "step": 1371 }, { "epoch": 1.8, "learning_rate": 4.316e-05, "loss": 0.5904, "step": 1372 }, { "epoch": 1.8, "learning_rate": 4.3155e-05, "loss": 0.6897, "step": 1373 }, { "epoch": 1.8, "learning_rate": 4.315e-05, "loss": 0.5639, "step": 1374 }, { "epoch": 1.8, "learning_rate": 4.3145e-05, "loss": 0.5964, "step": 1375 }, { "epoch": 1.8, "learning_rate": 4.3140000000000004e-05, "loss": 0.6373, "step": 1376 }, { "epoch": 1.8, "learning_rate": 4.3135000000000006e-05, "loss": 0.6552, "step": 1377 }, { "epoch": 1.8, "learning_rate": 4.313e-05, "loss": 0.6531, "step": 1378 }, { "epoch": 1.8, "learning_rate": 4.3125000000000005e-05, "loss": 0.6091, "step": 1379 }, { "epoch": 1.81, "learning_rate": 4.312000000000001e-05, "loss": 0.6423, "step": 1380 }, { "epoch": 1.81, "learning_rate": 4.3115e-05, "loss": 0.6043, "step": 1381 }, { "epoch": 1.81, "learning_rate": 4.311e-05, "loss": 0.5102, "step": 1382 }, { "epoch": 1.81, "learning_rate": 4.3105e-05, "loss": 0.6609, "step": 1383 }, { "epoch": 1.81, "learning_rate": 4.3100000000000004e-05, "loss": 0.581, "step": 1384 }, { "epoch": 1.81, "learning_rate": 4.3095e-05, "loss": 0.6147, "step": 1385 }, { "epoch": 1.81, "learning_rate": 4.309e-05, "loss": 0.7118, "step": 1386 }, { "epoch": 1.82, "learning_rate": 4.3085000000000005e-05, "loss": 0.7869, "step": 1387 }, { "epoch": 1.82, "learning_rate": 4.308e-05, "loss": 0.6782, "step": 1388 }, { "epoch": 1.82, "learning_rate": 4.3075000000000003e-05, "loss": 0.6579, "step": 1389 }, { "epoch": 1.82, "learning_rate": 4.3070000000000006e-05, "loss": 0.6512, "step": 1390 }, { "epoch": 1.82, "learning_rate": 4.3065e-05, "loss": 0.8332, "step": 1391 }, { "epoch": 1.82, "learning_rate": 4.306e-05, "loss": 0.589, "step": 1392 }, { "epoch": 1.82, "learning_rate": 4.3055e-05, "loss": 0.6721, "step": 1393 }, { "epoch": 1.82, "learning_rate": 4.305e-05, "loss": 0.6522, "step": 1394 }, { "epoch": 1.83, "learning_rate": 4.3045e-05, "loss": 0.7928, "step": 1395 }, { "epoch": 1.83, "learning_rate": 4.304e-05, "loss": 0.6789, "step": 1396 }, { "epoch": 1.83, "learning_rate": 4.3035000000000004e-05, "loss": 0.4857, "step": 1397 }, { "epoch": 1.83, "learning_rate": 4.3030000000000006e-05, "loss": 0.6034, "step": 1398 }, { "epoch": 1.83, "learning_rate": 4.3025e-05, "loss": 0.3824, "step": 1399 }, { "epoch": 1.83, "learning_rate": 4.3020000000000005e-05, "loss": 0.483, "step": 1400 }, { "epoch": 1.83, "learning_rate": 4.3015e-05, "loss": 0.657, "step": 1401 }, { "epoch": 1.84, "learning_rate": 4.301e-05, "loss": 0.6302, "step": 1402 }, { "epoch": 1.84, "learning_rate": 4.3005e-05, "loss": 0.6237, "step": 1403 }, { "epoch": 1.84, "learning_rate": 4.3e-05, "loss": 0.6418, "step": 1404 }, { "epoch": 1.84, "learning_rate": 4.2995000000000004e-05, "loss": 0.5741, "step": 1405 }, { "epoch": 1.84, "learning_rate": 4.299e-05, "loss": 0.5755, "step": 1406 }, { "epoch": 1.84, "learning_rate": 4.2985e-05, "loss": 0.5613, "step": 1407 }, { "epoch": 1.84, "learning_rate": 4.2980000000000005e-05, "loss": 0.5921, "step": 1408 }, { "epoch": 1.84, "learning_rate": 4.2975e-05, "loss": 0.5783, "step": 1409 }, { "epoch": 1.85, "learning_rate": 4.2970000000000004e-05, "loss": 0.5934, "step": 1410 }, { "epoch": 1.85, "learning_rate": 4.2965e-05, "loss": 0.635, "step": 1411 }, { "epoch": 1.85, "learning_rate": 4.296e-05, "loss": 0.6309, "step": 1412 }, { "epoch": 1.85, "learning_rate": 4.2955e-05, "loss": 0.568, "step": 1413 }, { "epoch": 1.85, "learning_rate": 4.295e-05, "loss": 0.5965, "step": 1414 }, { "epoch": 1.85, "learning_rate": 4.2945e-05, "loss": 0.8903, "step": 1415 }, { "epoch": 1.85, "learning_rate": 4.2940000000000006e-05, "loss": 0.7283, "step": 1416 }, { "epoch": 1.85, "learning_rate": 4.2935e-05, "loss": 0.5194, "step": 1417 }, { "epoch": 1.86, "learning_rate": 4.2930000000000004e-05, "loss": 0.5518, "step": 1418 }, { "epoch": 1.86, "learning_rate": 4.2925000000000007e-05, "loss": 0.5839, "step": 1419 }, { "epoch": 1.86, "learning_rate": 4.292e-05, "loss": 0.728, "step": 1420 }, { "epoch": 1.86, "learning_rate": 4.2915e-05, "loss": 0.8006, "step": 1421 }, { "epoch": 1.86, "learning_rate": 4.291e-05, "loss": 0.6852, "step": 1422 }, { "epoch": 1.86, "learning_rate": 4.2905000000000003e-05, "loss": 0.6508, "step": 1423 }, { "epoch": 1.86, "learning_rate": 4.29e-05, "loss": 0.6889, "step": 1424 }, { "epoch": 1.87, "learning_rate": 4.2895e-05, "loss": 0.5331, "step": 1425 }, { "epoch": 1.87, "learning_rate": 4.2890000000000004e-05, "loss": 0.6851, "step": 1426 }, { "epoch": 1.87, "learning_rate": 4.2885e-05, "loss": 0.5964, "step": 1427 }, { "epoch": 1.87, "learning_rate": 4.288e-05, "loss": 0.6599, "step": 1428 }, { "epoch": 1.87, "learning_rate": 4.2875000000000005e-05, "loss": 0.6395, "step": 1429 }, { "epoch": 1.87, "learning_rate": 4.287000000000001e-05, "loss": 0.7758, "step": 1430 }, { "epoch": 1.87, "learning_rate": 4.2865e-05, "loss": 0.6046, "step": 1431 }, { "epoch": 1.87, "learning_rate": 4.286e-05, "loss": 0.6941, "step": 1432 }, { "epoch": 1.88, "learning_rate": 4.2855e-05, "loss": 0.5298, "step": 1433 }, { "epoch": 1.88, "learning_rate": 4.285e-05, "loss": 0.6617, "step": 1434 }, { "epoch": 1.88, "learning_rate": 4.2845e-05, "loss": 0.6122, "step": 1435 }, { "epoch": 1.88, "learning_rate": 4.284e-05, "loss": 0.5886, "step": 1436 }, { "epoch": 1.88, "learning_rate": 4.2835000000000006e-05, "loss": 0.551, "step": 1437 }, { "epoch": 1.88, "learning_rate": 4.283e-05, "loss": 0.7972, "step": 1438 }, { "epoch": 1.88, "learning_rate": 4.2825000000000004e-05, "loss": 0.6932, "step": 1439 }, { "epoch": 1.88, "learning_rate": 4.282000000000001e-05, "loss": 0.6305, "step": 1440 }, { "epoch": 1.89, "learning_rate": 4.2815e-05, "loss": 0.6887, "step": 1441 }, { "epoch": 1.89, "learning_rate": 4.281e-05, "loss": 0.9205, "step": 1442 }, { "epoch": 1.89, "learning_rate": 4.2805e-05, "loss": 0.7403, "step": 1443 }, { "epoch": 1.89, "learning_rate": 4.2800000000000004e-05, "loss": 0.7084, "step": 1444 }, { "epoch": 1.89, "learning_rate": 4.2795e-05, "loss": 0.5992, "step": 1445 }, { "epoch": 1.89, "learning_rate": 4.279e-05, "loss": 0.5849, "step": 1446 }, { "epoch": 1.89, "learning_rate": 4.2785000000000005e-05, "loss": 0.6968, "step": 1447 }, { "epoch": 1.9, "learning_rate": 4.278e-05, "loss": 0.5597, "step": 1448 }, { "epoch": 1.9, "learning_rate": 4.2775e-05, "loss": 0.6324, "step": 1449 }, { "epoch": 1.9, "learning_rate": 4.2770000000000006e-05, "loss": 0.7743, "step": 1450 }, { "epoch": 1.9, "learning_rate": 4.2765e-05, "loss": 0.6024, "step": 1451 }, { "epoch": 1.9, "learning_rate": 4.276e-05, "loss": 0.5417, "step": 1452 }, { "epoch": 1.9, "learning_rate": 4.2755e-05, "loss": 0.6219, "step": 1453 }, { "epoch": 1.9, "learning_rate": 4.275e-05, "loss": 0.6311, "step": 1454 }, { "epoch": 1.9, "learning_rate": 4.2745000000000005e-05, "loss": 0.7234, "step": 1455 }, { "epoch": 1.91, "learning_rate": 4.274e-05, "loss": 0.6164, "step": 1456 }, { "epoch": 1.91, "learning_rate": 4.2735e-05, "loss": 0.7138, "step": 1457 }, { "epoch": 1.91, "learning_rate": 4.2730000000000006e-05, "loss": 0.5615, "step": 1458 }, { "epoch": 1.91, "learning_rate": 4.2725e-05, "loss": 0.4963, "step": 1459 }, { "epoch": 1.91, "learning_rate": 4.2720000000000004e-05, "loss": 0.5612, "step": 1460 }, { "epoch": 1.91, "learning_rate": 4.2715e-05, "loss": 0.5955, "step": 1461 }, { "epoch": 1.91, "learning_rate": 4.271e-05, "loss": 0.6282, "step": 1462 }, { "epoch": 1.91, "learning_rate": 4.2705e-05, "loss": 0.5416, "step": 1463 }, { "epoch": 1.92, "learning_rate": 4.27e-05, "loss": 0.588, "step": 1464 }, { "epoch": 1.92, "learning_rate": 4.2695000000000004e-05, "loss": 0.6563, "step": 1465 }, { "epoch": 1.92, "learning_rate": 4.269e-05, "loss": 0.6123, "step": 1466 }, { "epoch": 1.92, "learning_rate": 4.2685e-05, "loss": 0.5904, "step": 1467 }, { "epoch": 1.92, "learning_rate": 4.2680000000000005e-05, "loss": 0.647, "step": 1468 }, { "epoch": 1.92, "learning_rate": 4.2675e-05, "loss": 0.7227, "step": 1469 }, { "epoch": 1.92, "learning_rate": 4.267e-05, "loss": 0.6995, "step": 1470 }, { "epoch": 1.93, "learning_rate": 4.2665e-05, "loss": 0.5114, "step": 1471 }, { "epoch": 1.93, "learning_rate": 4.266e-05, "loss": 0.5814, "step": 1472 }, { "epoch": 1.93, "learning_rate": 4.2655e-05, "loss": 0.693, "step": 1473 }, { "epoch": 1.93, "learning_rate": 4.265e-05, "loss": 0.777, "step": 1474 }, { "epoch": 1.93, "learning_rate": 4.2645e-05, "loss": 0.6855, "step": 1475 }, { "epoch": 1.93, "learning_rate": 4.2640000000000005e-05, "loss": 0.7619, "step": 1476 }, { "epoch": 1.93, "learning_rate": 4.2635e-05, "loss": 0.9015, "step": 1477 }, { "epoch": 1.93, "learning_rate": 4.2630000000000004e-05, "loss": 0.6491, "step": 1478 }, { "epoch": 1.94, "learning_rate": 4.2625000000000006e-05, "loss": 0.5643, "step": 1479 }, { "epoch": 1.94, "learning_rate": 4.262e-05, "loss": 0.6608, "step": 1480 }, { "epoch": 1.94, "learning_rate": 4.2615e-05, "loss": 0.506, "step": 1481 }, { "epoch": 1.94, "learning_rate": 4.261e-05, "loss": 0.5918, "step": 1482 }, { "epoch": 1.94, "learning_rate": 4.2605e-05, "loss": 0.7319, "step": 1483 }, { "epoch": 1.94, "learning_rate": 4.26e-05, "loss": 0.5844, "step": 1484 }, { "epoch": 1.94, "learning_rate": 4.2595e-05, "loss": 0.7239, "step": 1485 }, { "epoch": 1.95, "learning_rate": 4.2590000000000004e-05, "loss": 0.7651, "step": 1486 }, { "epoch": 1.95, "learning_rate": 4.2585e-05, "loss": 0.5031, "step": 1487 }, { "epoch": 1.95, "learning_rate": 4.258e-05, "loss": 0.6131, "step": 1488 }, { "epoch": 1.95, "learning_rate": 4.2575000000000005e-05, "loss": 0.6242, "step": 1489 }, { "epoch": 1.95, "learning_rate": 4.257000000000001e-05, "loss": 0.5771, "step": 1490 }, { "epoch": 1.95, "learning_rate": 4.2564999999999997e-05, "loss": 0.5546, "step": 1491 }, { "epoch": 1.95, "learning_rate": 4.256e-05, "loss": 0.6122, "step": 1492 }, { "epoch": 1.95, "learning_rate": 4.2555e-05, "loss": 0.5872, "step": 1493 }, { "epoch": 1.96, "learning_rate": 4.2550000000000004e-05, "loss": 0.5594, "step": 1494 }, { "epoch": 1.96, "learning_rate": 4.2545e-05, "loss": 0.5677, "step": 1495 }, { "epoch": 1.96, "learning_rate": 4.254e-05, "loss": 0.898, "step": 1496 }, { "epoch": 1.96, "learning_rate": 4.2535000000000005e-05, "loss": 0.652, "step": 1497 }, { "epoch": 1.96, "learning_rate": 4.253e-05, "loss": 0.5977, "step": 1498 }, { "epoch": 1.96, "learning_rate": 4.2525000000000004e-05, "loss": 0.819, "step": 1499 }, { "epoch": 1.96, "learning_rate": 4.2520000000000006e-05, "loss": 0.6304, "step": 1500 }, { "epoch": 1.96, "learning_rate": 4.2515e-05, "loss": 0.7215, "step": 1501 }, { "epoch": 1.97, "learning_rate": 4.251e-05, "loss": 0.6221, "step": 1502 }, { "epoch": 1.97, "learning_rate": 4.2505e-05, "loss": 0.4996, "step": 1503 }, { "epoch": 1.97, "learning_rate": 4.25e-05, "loss": 0.5926, "step": 1504 }, { "epoch": 1.97, "learning_rate": 4.2495e-05, "loss": 0.5357, "step": 1505 }, { "epoch": 1.97, "learning_rate": 4.249e-05, "loss": 0.7315, "step": 1506 }, { "epoch": 1.97, "learning_rate": 4.2485000000000004e-05, "loss": 0.5113, "step": 1507 }, { "epoch": 1.97, "learning_rate": 4.248e-05, "loss": 0.6161, "step": 1508 }, { "epoch": 1.98, "learning_rate": 4.2475e-05, "loss": 0.5406, "step": 1509 }, { "epoch": 1.98, "learning_rate": 4.2470000000000005e-05, "loss": 0.8068, "step": 1510 }, { "epoch": 1.98, "learning_rate": 4.246500000000001e-05, "loss": 0.5028, "step": 1511 }, { "epoch": 1.98, "learning_rate": 4.246e-05, "loss": 0.6376, "step": 1512 }, { "epoch": 1.98, "learning_rate": 4.2455e-05, "loss": 0.6484, "step": 1513 }, { "epoch": 1.98, "learning_rate": 4.245e-05, "loss": 0.6228, "step": 1514 }, { "epoch": 1.98, "learning_rate": 4.2445000000000004e-05, "loss": 0.6398, "step": 1515 }, { "epoch": 1.98, "learning_rate": 4.244e-05, "loss": 0.5078, "step": 1516 }, { "epoch": 1.99, "learning_rate": 4.2435e-05, "loss": 0.622, "step": 1517 }, { "epoch": 1.99, "learning_rate": 4.2430000000000005e-05, "loss": 0.5719, "step": 1518 }, { "epoch": 1.99, "learning_rate": 4.2425e-05, "loss": 0.6554, "step": 1519 }, { "epoch": 1.99, "learning_rate": 4.2420000000000004e-05, "loss": 0.6234, "step": 1520 }, { "epoch": 1.99, "learning_rate": 4.2415000000000006e-05, "loss": 0.5471, "step": 1521 }, { "epoch": 1.99, "learning_rate": 4.241e-05, "loss": 0.7987, "step": 1522 }, { "epoch": 1.99, "learning_rate": 4.2405e-05, "loss": 0.7057, "step": 1523 }, { "epoch": 1.99, "learning_rate": 4.24e-05, "loss": 0.6362, "step": 1524 }, { "epoch": 2.0, "learning_rate": 4.2395e-05, "loss": 0.7612, "step": 1525 }, { "epoch": 2.0, "learning_rate": 4.239e-05, "loss": 0.8176, "step": 1526 }, { "epoch": 2.0, "learning_rate": 4.2385e-05, "loss": 0.6831, "step": 1527 }, { "epoch": 2.0, "learning_rate": 4.2380000000000004e-05, "loss": 0.6111, "step": 1528 }, { "epoch": 2.0, "learning_rate": 4.237500000000001e-05, "loss": 0.4211, "step": 1529 }, { "epoch": 2.0, "learning_rate": 4.237e-05, "loss": 0.2717, "step": 1530 }, { "epoch": 2.0, "learning_rate": 4.2365000000000005e-05, "loss": 0.3564, "step": 1531 }, { "epoch": 2.01, "learning_rate": 4.236e-05, "loss": 0.3298, "step": 1532 }, { "epoch": 2.01, "learning_rate": 4.2355000000000004e-05, "loss": 0.3159, "step": 1533 }, { "epoch": 2.01, "learning_rate": 4.235e-05, "loss": 0.3569, "step": 1534 }, { "epoch": 2.01, "learning_rate": 4.2345e-05, "loss": 0.3231, "step": 1535 }, { "epoch": 2.01, "learning_rate": 4.2340000000000005e-05, "loss": 0.3051, "step": 1536 }, { "epoch": 2.01, "learning_rate": 4.2335e-05, "loss": 0.3269, "step": 1537 }, { "epoch": 2.01, "learning_rate": 4.233e-05, "loss": 0.3232, "step": 1538 }, { "epoch": 2.01, "learning_rate": 4.2325000000000006e-05, "loss": 0.3587, "step": 1539 }, { "epoch": 2.02, "learning_rate": 4.232e-05, "loss": 0.2753, "step": 1540 }, { "epoch": 2.02, "learning_rate": 4.2315000000000004e-05, "loss": 0.3948, "step": 1541 }, { "epoch": 2.02, "learning_rate": 4.231e-05, "loss": 0.3288, "step": 1542 }, { "epoch": 2.02, "learning_rate": 4.2305e-05, "loss": 0.2817, "step": 1543 }, { "epoch": 2.02, "learning_rate": 4.23e-05, "loss": 0.2588, "step": 1544 }, { "epoch": 2.02, "learning_rate": 4.2295e-05, "loss": 0.3694, "step": 1545 }, { "epoch": 2.02, "learning_rate": 4.229e-05, "loss": 0.312, "step": 1546 }, { "epoch": 2.02, "learning_rate": 4.2285e-05, "loss": 0.3899, "step": 1547 }, { "epoch": 2.03, "learning_rate": 4.228e-05, "loss": 0.3143, "step": 1548 }, { "epoch": 2.03, "learning_rate": 4.2275000000000004e-05, "loss": 0.2959, "step": 1549 }, { "epoch": 2.03, "learning_rate": 4.227000000000001e-05, "loss": 0.2729, "step": 1550 }, { "epoch": 2.03, "learning_rate": 4.2265e-05, "loss": 0.2988, "step": 1551 }, { "epoch": 2.03, "learning_rate": 4.226e-05, "loss": 0.382, "step": 1552 }, { "epoch": 2.03, "learning_rate": 4.2255e-05, "loss": 0.3827, "step": 1553 }, { "epoch": 2.03, "learning_rate": 4.2250000000000004e-05, "loss": 0.3008, "step": 1554 }, { "epoch": 2.04, "learning_rate": 4.2245e-05, "loss": 0.2906, "step": 1555 }, { "epoch": 2.04, "learning_rate": 4.224e-05, "loss": 0.272, "step": 1556 }, { "epoch": 2.04, "learning_rate": 4.2235000000000005e-05, "loss": 0.3354, "step": 1557 }, { "epoch": 2.04, "learning_rate": 4.223e-05, "loss": 0.3328, "step": 1558 }, { "epoch": 2.04, "learning_rate": 4.2225e-05, "loss": 0.3303, "step": 1559 }, { "epoch": 2.04, "learning_rate": 4.2220000000000006e-05, "loss": 0.2802, "step": 1560 }, { "epoch": 2.04, "learning_rate": 4.2215e-05, "loss": 0.3158, "step": 1561 }, { "epoch": 2.04, "learning_rate": 4.221e-05, "loss": 0.292, "step": 1562 }, { "epoch": 2.05, "learning_rate": 4.2205e-05, "loss": 0.2472, "step": 1563 }, { "epoch": 2.05, "learning_rate": 4.22e-05, "loss": 0.3769, "step": 1564 }, { "epoch": 2.05, "learning_rate": 4.2195e-05, "loss": 0.3355, "step": 1565 }, { "epoch": 2.05, "learning_rate": 4.219e-05, "loss": 0.3446, "step": 1566 }, { "epoch": 2.05, "learning_rate": 4.2185000000000004e-05, "loss": 0.2497, "step": 1567 }, { "epoch": 2.05, "learning_rate": 4.2180000000000006e-05, "loss": 0.3456, "step": 1568 }, { "epoch": 2.05, "learning_rate": 4.2175e-05, "loss": 0.4579, "step": 1569 }, { "epoch": 2.05, "learning_rate": 4.2170000000000005e-05, "loss": 0.2384, "step": 1570 }, { "epoch": 2.06, "learning_rate": 4.216500000000001e-05, "loss": 0.3328, "step": 1571 }, { "epoch": 2.06, "learning_rate": 4.2159999999999996e-05, "loss": 0.28, "step": 1572 }, { "epoch": 2.06, "learning_rate": 4.2155e-05, "loss": 0.3203, "step": 1573 }, { "epoch": 2.06, "learning_rate": 4.215e-05, "loss": 0.4162, "step": 1574 }, { "epoch": 2.06, "learning_rate": 4.2145000000000004e-05, "loss": 0.4055, "step": 1575 }, { "epoch": 2.06, "learning_rate": 4.214e-05, "loss": 0.324, "step": 1576 }, { "epoch": 2.06, "learning_rate": 4.2135e-05, "loss": 0.2356, "step": 1577 }, { "epoch": 2.07, "learning_rate": 4.2130000000000005e-05, "loss": 0.3229, "step": 1578 }, { "epoch": 2.07, "learning_rate": 4.2125e-05, "loss": 0.3569, "step": 1579 }, { "epoch": 2.07, "learning_rate": 4.212e-05, "loss": 0.325, "step": 1580 }, { "epoch": 2.07, "learning_rate": 4.2115000000000006e-05, "loss": 0.2554, "step": 1581 }, { "epoch": 2.07, "learning_rate": 4.211e-05, "loss": 0.4428, "step": 1582 }, { "epoch": 2.07, "learning_rate": 4.2105e-05, "loss": 0.251, "step": 1583 }, { "epoch": 2.07, "learning_rate": 4.21e-05, "loss": 0.4141, "step": 1584 }, { "epoch": 2.07, "learning_rate": 4.2095e-05, "loss": 0.275, "step": 1585 }, { "epoch": 2.08, "learning_rate": 4.209e-05, "loss": 0.3538, "step": 1586 }, { "epoch": 2.08, "learning_rate": 4.2085e-05, "loss": 0.2654, "step": 1587 }, { "epoch": 2.08, "learning_rate": 4.2080000000000004e-05, "loss": 0.3514, "step": 1588 }, { "epoch": 2.08, "learning_rate": 4.2075000000000006e-05, "loss": 0.3404, "step": 1589 }, { "epoch": 2.08, "learning_rate": 4.207e-05, "loss": 0.3179, "step": 1590 }, { "epoch": 2.08, "learning_rate": 4.2065000000000005e-05, "loss": 0.3524, "step": 1591 }, { "epoch": 2.08, "learning_rate": 4.206e-05, "loss": 0.2821, "step": 1592 }, { "epoch": 2.09, "learning_rate": 4.2055e-05, "loss": 0.2664, "step": 1593 }, { "epoch": 2.09, "learning_rate": 4.205e-05, "loss": 0.344, "step": 1594 }, { "epoch": 2.09, "learning_rate": 4.2045e-05, "loss": 0.2593, "step": 1595 }, { "epoch": 2.09, "learning_rate": 4.2040000000000004e-05, "loss": 0.4171, "step": 1596 }, { "epoch": 2.09, "learning_rate": 4.2035e-05, "loss": 0.2749, "step": 1597 }, { "epoch": 2.09, "learning_rate": 4.203e-05, "loss": 0.4619, "step": 1598 }, { "epoch": 2.09, "learning_rate": 4.2025000000000005e-05, "loss": 0.3171, "step": 1599 }, { "epoch": 2.09, "learning_rate": 4.202e-05, "loss": 0.348, "step": 1600 }, { "epoch": 2.1, "learning_rate": 4.2015000000000003e-05, "loss": 0.2944, "step": 1601 }, { "epoch": 2.1, "learning_rate": 4.201e-05, "loss": 0.2174, "step": 1602 }, { "epoch": 2.1, "learning_rate": 4.2005e-05, "loss": 0.324, "step": 1603 }, { "epoch": 2.1, "learning_rate": 4.2e-05, "loss": 0.3175, "step": 1604 }, { "epoch": 2.1, "learning_rate": 4.1995e-05, "loss": 0.2976, "step": 1605 }, { "epoch": 2.1, "learning_rate": 4.199e-05, "loss": 0.3703, "step": 1606 }, { "epoch": 2.1, "learning_rate": 4.1985000000000005e-05, "loss": 0.2903, "step": 1607 }, { "epoch": 2.1, "learning_rate": 4.198e-05, "loss": 0.4046, "step": 1608 }, { "epoch": 2.11, "learning_rate": 4.1975000000000004e-05, "loss": 0.3156, "step": 1609 }, { "epoch": 2.11, "learning_rate": 4.1970000000000006e-05, "loss": 0.3706, "step": 1610 }, { "epoch": 2.11, "learning_rate": 4.1965e-05, "loss": 0.2953, "step": 1611 }, { "epoch": 2.11, "learning_rate": 4.196e-05, "loss": 0.2764, "step": 1612 }, { "epoch": 2.11, "learning_rate": 4.1955e-05, "loss": 0.3728, "step": 1613 }, { "epoch": 2.11, "learning_rate": 4.195e-05, "loss": 0.2934, "step": 1614 }, { "epoch": 2.11, "learning_rate": 4.1945e-05, "loss": 0.3308, "step": 1615 }, { "epoch": 2.12, "learning_rate": 4.194e-05, "loss": 0.315, "step": 1616 }, { "epoch": 2.12, "learning_rate": 4.1935000000000004e-05, "loss": 0.3041, "step": 1617 }, { "epoch": 2.12, "learning_rate": 4.193e-05, "loss": 0.2571, "step": 1618 }, { "epoch": 2.12, "learning_rate": 4.1925e-05, "loss": 0.3105, "step": 1619 }, { "epoch": 2.12, "learning_rate": 4.1920000000000005e-05, "loss": 0.2692, "step": 1620 }, { "epoch": 2.12, "learning_rate": 4.1915e-05, "loss": 0.3936, "step": 1621 }, { "epoch": 2.12, "learning_rate": 4.191e-05, "loss": 0.2815, "step": 1622 }, { "epoch": 2.12, "learning_rate": 4.1905e-05, "loss": 0.3448, "step": 1623 }, { "epoch": 2.13, "learning_rate": 4.19e-05, "loss": 0.3261, "step": 1624 }, { "epoch": 2.13, "learning_rate": 4.1895e-05, "loss": 0.2742, "step": 1625 }, { "epoch": 2.13, "learning_rate": 4.189e-05, "loss": 0.3567, "step": 1626 }, { "epoch": 2.13, "learning_rate": 4.1885e-05, "loss": 0.3429, "step": 1627 }, { "epoch": 2.13, "learning_rate": 4.1880000000000006e-05, "loss": 0.3237, "step": 1628 }, { "epoch": 2.13, "learning_rate": 4.1875e-05, "loss": 0.3011, "step": 1629 }, { "epoch": 2.13, "learning_rate": 4.1870000000000004e-05, "loss": 0.2864, "step": 1630 }, { "epoch": 2.13, "learning_rate": 4.1865000000000007e-05, "loss": 0.2656, "step": 1631 }, { "epoch": 2.14, "learning_rate": 4.186e-05, "loss": 0.2928, "step": 1632 }, { "epoch": 2.14, "learning_rate": 4.1855e-05, "loss": 0.3293, "step": 1633 }, { "epoch": 2.14, "learning_rate": 4.185e-05, "loss": 0.3051, "step": 1634 }, { "epoch": 2.14, "learning_rate": 4.1845000000000003e-05, "loss": 0.3039, "step": 1635 }, { "epoch": 2.14, "learning_rate": 4.184e-05, "loss": 0.2867, "step": 1636 }, { "epoch": 2.14, "learning_rate": 4.1835e-05, "loss": 0.3641, "step": 1637 }, { "epoch": 2.14, "learning_rate": 4.1830000000000004e-05, "loss": 0.349, "step": 1638 }, { "epoch": 2.15, "learning_rate": 4.1825e-05, "loss": 0.3757, "step": 1639 }, { "epoch": 2.15, "learning_rate": 4.182e-05, "loss": 0.3391, "step": 1640 }, { "epoch": 2.15, "learning_rate": 4.1815000000000005e-05, "loss": 0.2923, "step": 1641 }, { "epoch": 2.15, "learning_rate": 4.181000000000001e-05, "loss": 0.3192, "step": 1642 }, { "epoch": 2.15, "learning_rate": 4.1805e-05, "loss": 0.3486, "step": 1643 }, { "epoch": 2.15, "learning_rate": 4.18e-05, "loss": 0.3567, "step": 1644 }, { "epoch": 2.15, "learning_rate": 4.1795e-05, "loss": 0.2219, "step": 1645 }, { "epoch": 2.15, "learning_rate": 4.179e-05, "loss": 0.2944, "step": 1646 }, { "epoch": 2.16, "learning_rate": 4.1785e-05, "loss": 0.2494, "step": 1647 }, { "epoch": 2.16, "learning_rate": 4.178e-05, "loss": 0.2891, "step": 1648 }, { "epoch": 2.16, "learning_rate": 4.1775000000000006e-05, "loss": 0.2877, "step": 1649 }, { "epoch": 2.16, "learning_rate": 4.177e-05, "loss": 0.2797, "step": 1650 }, { "epoch": 2.16, "learning_rate": 4.1765000000000004e-05, "loss": 0.3754, "step": 1651 }, { "epoch": 2.16, "learning_rate": 4.176000000000001e-05, "loss": 0.4105, "step": 1652 }, { "epoch": 2.16, "learning_rate": 4.1755e-05, "loss": 0.3697, "step": 1653 }, { "epoch": 2.16, "learning_rate": 4.175e-05, "loss": 0.3338, "step": 1654 }, { "epoch": 2.17, "learning_rate": 4.1745e-05, "loss": 0.3007, "step": 1655 }, { "epoch": 2.17, "learning_rate": 4.1740000000000004e-05, "loss": 0.3441, "step": 1656 }, { "epoch": 2.17, "learning_rate": 4.1735e-05, "loss": 0.3035, "step": 1657 }, { "epoch": 2.17, "learning_rate": 4.173e-05, "loss": 0.32, "step": 1658 }, { "epoch": 2.17, "learning_rate": 4.1725000000000005e-05, "loss": 0.442, "step": 1659 }, { "epoch": 2.17, "learning_rate": 4.172e-05, "loss": 0.3129, "step": 1660 }, { "epoch": 2.17, "learning_rate": 4.1715e-05, "loss": 0.2412, "step": 1661 }, { "epoch": 2.18, "learning_rate": 4.1710000000000006e-05, "loss": 0.3548, "step": 1662 }, { "epoch": 2.18, "learning_rate": 4.1705e-05, "loss": 0.2905, "step": 1663 }, { "epoch": 2.18, "learning_rate": 4.17e-05, "loss": 0.3689, "step": 1664 }, { "epoch": 2.18, "learning_rate": 4.1695e-05, "loss": 0.299, "step": 1665 }, { "epoch": 2.18, "learning_rate": 4.169e-05, "loss": 0.359, "step": 1666 }, { "epoch": 2.18, "learning_rate": 4.1685000000000005e-05, "loss": 0.4102, "step": 1667 }, { "epoch": 2.18, "learning_rate": 4.168e-05, "loss": 0.4641, "step": 1668 }, { "epoch": 2.18, "learning_rate": 4.1675e-05, "loss": 0.3554, "step": 1669 }, { "epoch": 2.19, "learning_rate": 4.1670000000000006e-05, "loss": 0.2876, "step": 1670 }, { "epoch": 2.19, "learning_rate": 4.1665e-05, "loss": 0.3166, "step": 1671 }, { "epoch": 2.19, "learning_rate": 4.1660000000000004e-05, "loss": 0.3159, "step": 1672 }, { "epoch": 2.19, "learning_rate": 4.1655e-05, "loss": 0.3192, "step": 1673 }, { "epoch": 2.19, "learning_rate": 4.165e-05, "loss": 0.2619, "step": 1674 }, { "epoch": 2.19, "learning_rate": 4.1645e-05, "loss": 0.304, "step": 1675 }, { "epoch": 2.19, "learning_rate": 4.164e-05, "loss": 0.2741, "step": 1676 }, { "epoch": 2.2, "learning_rate": 4.1635000000000004e-05, "loss": 0.2538, "step": 1677 }, { "epoch": 2.2, "learning_rate": 4.163e-05, "loss": 0.3124, "step": 1678 }, { "epoch": 2.2, "learning_rate": 4.1625e-05, "loss": 0.2812, "step": 1679 }, { "epoch": 2.2, "learning_rate": 4.1620000000000005e-05, "loss": 0.2628, "step": 1680 }, { "epoch": 2.2, "learning_rate": 4.161500000000001e-05, "loss": 0.3342, "step": 1681 }, { "epoch": 2.2, "learning_rate": 4.161e-05, "loss": 0.3308, "step": 1682 }, { "epoch": 2.2, "learning_rate": 4.1605e-05, "loss": 0.3962, "step": 1683 }, { "epoch": 2.2, "learning_rate": 4.16e-05, "loss": 0.3415, "step": 1684 }, { "epoch": 2.21, "learning_rate": 4.1595e-05, "loss": 0.2747, "step": 1685 }, { "epoch": 2.21, "learning_rate": 4.159e-05, "loss": 0.2318, "step": 1686 }, { "epoch": 2.21, "learning_rate": 4.1585e-05, "loss": 0.3392, "step": 1687 }, { "epoch": 2.21, "learning_rate": 4.1580000000000005e-05, "loss": 0.2673, "step": 1688 }, { "epoch": 2.21, "learning_rate": 4.1575e-05, "loss": 0.3236, "step": 1689 }, { "epoch": 2.21, "learning_rate": 4.1570000000000003e-05, "loss": 0.3188, "step": 1690 }, { "epoch": 2.21, "learning_rate": 4.1565000000000006e-05, "loss": 0.3549, "step": 1691 }, { "epoch": 2.21, "learning_rate": 4.156e-05, "loss": 0.3499, "step": 1692 }, { "epoch": 2.22, "learning_rate": 4.1555e-05, "loss": 0.2864, "step": 1693 }, { "epoch": 2.22, "learning_rate": 4.155e-05, "loss": 0.2763, "step": 1694 }, { "epoch": 2.22, "learning_rate": 4.1545e-05, "loss": 0.2873, "step": 1695 }, { "epoch": 2.22, "learning_rate": 4.154e-05, "loss": 0.323, "step": 1696 }, { "epoch": 2.22, "learning_rate": 4.1535e-05, "loss": 0.348, "step": 1697 }, { "epoch": 2.22, "learning_rate": 4.1530000000000004e-05, "loss": 0.3022, "step": 1698 }, { "epoch": 2.22, "learning_rate": 4.1525e-05, "loss": 0.2634, "step": 1699 }, { "epoch": 2.23, "learning_rate": 4.152e-05, "loss": 0.3013, "step": 1700 }, { "epoch": 2.23, "learning_rate": 4.1515000000000005e-05, "loss": 0.2834, "step": 1701 }, { "epoch": 2.23, "learning_rate": 4.151000000000001e-05, "loss": 0.3295, "step": 1702 }, { "epoch": 2.23, "learning_rate": 4.1504999999999996e-05, "loss": 0.2767, "step": 1703 }, { "epoch": 2.23, "learning_rate": 4.15e-05, "loss": 0.2919, "step": 1704 }, { "epoch": 2.23, "learning_rate": 4.1495e-05, "loss": 0.3793, "step": 1705 }, { "epoch": 2.23, "learning_rate": 4.1490000000000004e-05, "loss": 0.3633, "step": 1706 }, { "epoch": 2.23, "learning_rate": 4.1485e-05, "loss": 0.3292, "step": 1707 }, { "epoch": 2.24, "learning_rate": 4.148e-05, "loss": 0.2708, "step": 1708 }, { "epoch": 2.24, "learning_rate": 4.1475000000000005e-05, "loss": 0.3505, "step": 1709 }, { "epoch": 2.24, "learning_rate": 4.147e-05, "loss": 0.3505, "step": 1710 }, { "epoch": 2.24, "learning_rate": 4.1465000000000004e-05, "loss": 0.4256, "step": 1711 }, { "epoch": 2.24, "learning_rate": 4.1460000000000006e-05, "loss": 0.2889, "step": 1712 }, { "epoch": 2.24, "learning_rate": 4.1455e-05, "loss": 0.2609, "step": 1713 }, { "epoch": 2.24, "learning_rate": 4.145e-05, "loss": 0.3167, "step": 1714 }, { "epoch": 2.24, "learning_rate": 4.1445e-05, "loss": 0.357, "step": 1715 }, { "epoch": 2.25, "learning_rate": 4.144e-05, "loss": 0.2885, "step": 1716 }, { "epoch": 2.25, "learning_rate": 4.1435e-05, "loss": 0.2972, "step": 1717 }, { "epoch": 2.25, "learning_rate": 4.143e-05, "loss": 0.3058, "step": 1718 }, { "epoch": 2.25, "learning_rate": 4.1425000000000004e-05, "loss": 0.3495, "step": 1719 }, { "epoch": 2.25, "learning_rate": 4.142000000000001e-05, "loss": 0.3858, "step": 1720 }, { "epoch": 2.25, "learning_rate": 4.1415e-05, "loss": 0.3398, "step": 1721 }, { "epoch": 2.25, "learning_rate": 4.1410000000000005e-05, "loss": 0.3383, "step": 1722 }, { "epoch": 2.26, "learning_rate": 4.1405e-05, "loss": 0.3423, "step": 1723 }, { "epoch": 2.26, "learning_rate": 4.14e-05, "loss": 0.3133, "step": 1724 }, { "epoch": 2.26, "learning_rate": 4.1395e-05, "loss": 0.3498, "step": 1725 }, { "epoch": 2.26, "learning_rate": 4.139e-05, "loss": 0.3363, "step": 1726 }, { "epoch": 2.26, "learning_rate": 4.1385000000000004e-05, "loss": 0.36, "step": 1727 }, { "epoch": 2.26, "learning_rate": 4.138e-05, "loss": 0.258, "step": 1728 }, { "epoch": 2.26, "learning_rate": 4.1375e-05, "loss": 0.342, "step": 1729 }, { "epoch": 2.26, "learning_rate": 4.1370000000000005e-05, "loss": 0.347, "step": 1730 }, { "epoch": 2.27, "learning_rate": 4.1365e-05, "loss": 0.2887, "step": 1731 }, { "epoch": 2.27, "learning_rate": 4.1360000000000004e-05, "loss": 0.2634, "step": 1732 }, { "epoch": 2.27, "learning_rate": 4.1355e-05, "loss": 0.3289, "step": 1733 }, { "epoch": 2.27, "learning_rate": 4.135e-05, "loss": 0.2948, "step": 1734 }, { "epoch": 2.27, "learning_rate": 4.1345e-05, "loss": 0.3676, "step": 1735 }, { "epoch": 2.27, "learning_rate": 4.134e-05, "loss": 0.2393, "step": 1736 }, { "epoch": 2.27, "learning_rate": 4.1335e-05, "loss": 0.3351, "step": 1737 }, { "epoch": 2.27, "learning_rate": 4.133e-05, "loss": 0.3519, "step": 1738 }, { "epoch": 2.28, "learning_rate": 4.1325e-05, "loss": 0.3845, "step": 1739 }, { "epoch": 2.28, "learning_rate": 4.1320000000000004e-05, "loss": 0.2989, "step": 1740 }, { "epoch": 2.28, "learning_rate": 4.131500000000001e-05, "loss": 0.3242, "step": 1741 }, { "epoch": 2.28, "learning_rate": 4.131e-05, "loss": 0.3054, "step": 1742 }, { "epoch": 2.28, "learning_rate": 4.1305e-05, "loss": 0.2942, "step": 1743 }, { "epoch": 2.28, "learning_rate": 4.13e-05, "loss": 0.2807, "step": 1744 }, { "epoch": 2.28, "learning_rate": 4.1295000000000004e-05, "loss": 0.2782, "step": 1745 }, { "epoch": 2.29, "learning_rate": 4.129e-05, "loss": 0.3007, "step": 1746 }, { "epoch": 2.29, "learning_rate": 4.1285e-05, "loss": 0.3307, "step": 1747 }, { "epoch": 2.29, "learning_rate": 4.1280000000000005e-05, "loss": 0.3663, "step": 1748 }, { "epoch": 2.29, "learning_rate": 4.1275e-05, "loss": 0.4097, "step": 1749 }, { "epoch": 2.29, "learning_rate": 4.127e-05, "loss": 0.2834, "step": 1750 }, { "epoch": 2.29, "learning_rate": 4.1265000000000006e-05, "loss": 0.2654, "step": 1751 }, { "epoch": 2.29, "learning_rate": 4.126e-05, "loss": 0.3431, "step": 1752 }, { "epoch": 2.29, "learning_rate": 4.1255e-05, "loss": 0.3615, "step": 1753 }, { "epoch": 2.3, "learning_rate": 4.125e-05, "loss": 0.3973, "step": 1754 }, { "epoch": 2.3, "learning_rate": 4.1245e-05, "loss": 0.2994, "step": 1755 }, { "epoch": 2.3, "learning_rate": 4.124e-05, "loss": 0.2908, "step": 1756 }, { "epoch": 2.3, "learning_rate": 4.1235e-05, "loss": 0.3081, "step": 1757 }, { "epoch": 2.3, "learning_rate": 4.123e-05, "loss": 0.4095, "step": 1758 }, { "epoch": 2.3, "learning_rate": 4.1225e-05, "loss": 0.3492, "step": 1759 }, { "epoch": 2.3, "learning_rate": 4.122e-05, "loss": 0.3958, "step": 1760 }, { "epoch": 2.3, "learning_rate": 4.1215000000000004e-05, "loss": 0.2831, "step": 1761 }, { "epoch": 2.31, "learning_rate": 4.121000000000001e-05, "loss": 0.304, "step": 1762 }, { "epoch": 2.31, "learning_rate": 4.1205e-05, "loss": 0.3452, "step": 1763 }, { "epoch": 2.31, "learning_rate": 4.12e-05, "loss": 0.3454, "step": 1764 }, { "epoch": 2.31, "learning_rate": 4.1195e-05, "loss": 0.3856, "step": 1765 }, { "epoch": 2.31, "learning_rate": 4.1190000000000004e-05, "loss": 0.3027, "step": 1766 }, { "epoch": 2.31, "learning_rate": 4.1185e-05, "loss": 0.3281, "step": 1767 }, { "epoch": 2.31, "learning_rate": 4.118e-05, "loss": 0.243, "step": 1768 }, { "epoch": 2.32, "learning_rate": 4.1175000000000005e-05, "loss": 0.3376, "step": 1769 }, { "epoch": 2.32, "learning_rate": 4.117e-05, "loss": 0.283, "step": 1770 }, { "epoch": 2.32, "learning_rate": 4.1165e-05, "loss": 0.2676, "step": 1771 }, { "epoch": 2.32, "learning_rate": 4.1160000000000006e-05, "loss": 0.2968, "step": 1772 }, { "epoch": 2.32, "learning_rate": 4.1155e-05, "loss": 0.3453, "step": 1773 }, { "epoch": 2.32, "learning_rate": 4.115e-05, "loss": 0.3079, "step": 1774 }, { "epoch": 2.32, "learning_rate": 4.1145e-05, "loss": 0.3592, "step": 1775 }, { "epoch": 2.32, "learning_rate": 4.114e-05, "loss": 0.3574, "step": 1776 }, { "epoch": 2.33, "learning_rate": 4.1135e-05, "loss": 0.313, "step": 1777 }, { "epoch": 2.33, "learning_rate": 4.113e-05, "loss": 0.3518, "step": 1778 }, { "epoch": 2.33, "learning_rate": 4.1125000000000004e-05, "loss": 0.3197, "step": 1779 }, { "epoch": 2.33, "learning_rate": 4.1120000000000006e-05, "loss": 0.3873, "step": 1780 }, { "epoch": 2.33, "learning_rate": 4.1115e-05, "loss": 0.2871, "step": 1781 }, { "epoch": 2.33, "learning_rate": 4.1110000000000005e-05, "loss": 0.3961, "step": 1782 }, { "epoch": 2.33, "learning_rate": 4.110500000000001e-05, "loss": 0.3354, "step": 1783 }, { "epoch": 2.34, "learning_rate": 4.11e-05, "loss": 0.2912, "step": 1784 }, { "epoch": 2.34, "learning_rate": 4.1095e-05, "loss": 0.3255, "step": 1785 }, { "epoch": 2.34, "learning_rate": 4.109e-05, "loss": 0.3752, "step": 1786 }, { "epoch": 2.34, "learning_rate": 4.1085000000000004e-05, "loss": 0.3857, "step": 1787 }, { "epoch": 2.34, "learning_rate": 4.108e-05, "loss": 0.3509, "step": 1788 }, { "epoch": 2.34, "learning_rate": 4.1075e-05, "loss": 0.288, "step": 1789 }, { "epoch": 2.34, "learning_rate": 4.1070000000000005e-05, "loss": 0.3108, "step": 1790 }, { "epoch": 2.34, "learning_rate": 4.1065e-05, "loss": 0.3539, "step": 1791 }, { "epoch": 2.35, "learning_rate": 4.106e-05, "loss": 0.3774, "step": 1792 }, { "epoch": 2.35, "learning_rate": 4.1055000000000006e-05, "loss": 0.3917, "step": 1793 }, { "epoch": 2.35, "learning_rate": 4.105e-05, "loss": 0.4233, "step": 1794 }, { "epoch": 2.35, "learning_rate": 4.1045e-05, "loss": 0.3654, "step": 1795 }, { "epoch": 2.35, "learning_rate": 4.104e-05, "loss": 0.291, "step": 1796 }, { "epoch": 2.35, "learning_rate": 4.1035e-05, "loss": 0.3032, "step": 1797 }, { "epoch": 2.35, "learning_rate": 4.103e-05, "loss": 0.3674, "step": 1798 }, { "epoch": 2.35, "learning_rate": 4.1025e-05, "loss": 0.3337, "step": 1799 }, { "epoch": 2.36, "learning_rate": 4.1020000000000004e-05, "loss": 0.3046, "step": 1800 }, { "epoch": 2.36, "learning_rate": 4.1015000000000006e-05, "loss": 0.3291, "step": 1801 }, { "epoch": 2.36, "learning_rate": 4.101e-05, "loss": 0.3165, "step": 1802 }, { "epoch": 2.36, "learning_rate": 4.1005000000000005e-05, "loss": 0.3604, "step": 1803 }, { "epoch": 2.36, "learning_rate": 4.1e-05, "loss": 0.2907, "step": 1804 }, { "epoch": 2.36, "learning_rate": 4.0995e-05, "loss": 0.2677, "step": 1805 }, { "epoch": 2.36, "learning_rate": 4.099e-05, "loss": 0.289, "step": 1806 }, { "epoch": 2.37, "learning_rate": 4.0985e-05, "loss": 0.4074, "step": 1807 }, { "epoch": 2.37, "learning_rate": 4.0980000000000004e-05, "loss": 0.3201, "step": 1808 }, { "epoch": 2.37, "learning_rate": 4.0975e-05, "loss": 0.3037, "step": 1809 }, { "epoch": 2.37, "learning_rate": 4.097e-05, "loss": 0.3757, "step": 1810 }, { "epoch": 2.37, "learning_rate": 4.0965000000000005e-05, "loss": 0.3026, "step": 1811 }, { "epoch": 2.37, "learning_rate": 4.096e-05, "loss": 0.3501, "step": 1812 }, { "epoch": 2.37, "learning_rate": 4.0955000000000003e-05, "loss": 0.4073, "step": 1813 }, { "epoch": 2.37, "learning_rate": 4.095e-05, "loss": 0.3419, "step": 1814 }, { "epoch": 2.38, "learning_rate": 4.0945e-05, "loss": 0.3219, "step": 1815 }, { "epoch": 2.38, "learning_rate": 4.094e-05, "loss": 0.3447, "step": 1816 }, { "epoch": 2.38, "learning_rate": 4.0935e-05, "loss": 0.241, "step": 1817 }, { "epoch": 2.38, "learning_rate": 4.093e-05, "loss": 0.3074, "step": 1818 }, { "epoch": 2.38, "learning_rate": 4.0925000000000005e-05, "loss": 0.3184, "step": 1819 }, { "epoch": 2.38, "learning_rate": 4.092e-05, "loss": 0.4582, "step": 1820 }, { "epoch": 2.38, "learning_rate": 4.0915000000000004e-05, "loss": 0.3076, "step": 1821 }, { "epoch": 2.38, "learning_rate": 4.0910000000000006e-05, "loss": 0.2425, "step": 1822 }, { "epoch": 2.39, "learning_rate": 4.0905e-05, "loss": 0.2814, "step": 1823 }, { "epoch": 2.39, "learning_rate": 4.09e-05, "loss": 0.3428, "step": 1824 }, { "epoch": 2.39, "learning_rate": 4.0895e-05, "loss": 0.3085, "step": 1825 }, { "epoch": 2.39, "learning_rate": 4.089e-05, "loss": 0.3901, "step": 1826 }, { "epoch": 2.39, "learning_rate": 4.0885e-05, "loss": 0.3415, "step": 1827 }, { "epoch": 2.39, "learning_rate": 4.088e-05, "loss": 0.4483, "step": 1828 }, { "epoch": 2.39, "learning_rate": 4.0875000000000004e-05, "loss": 0.2692, "step": 1829 }, { "epoch": 2.4, "learning_rate": 4.087e-05, "loss": 0.3915, "step": 1830 }, { "epoch": 2.4, "learning_rate": 4.0865e-05, "loss": 0.3217, "step": 1831 }, { "epoch": 2.4, "learning_rate": 4.0860000000000005e-05, "loss": 0.3205, "step": 1832 }, { "epoch": 2.4, "learning_rate": 4.0855e-05, "loss": 0.3109, "step": 1833 }, { "epoch": 2.4, "learning_rate": 4.085e-05, "loss": 0.3544, "step": 1834 }, { "epoch": 2.4, "learning_rate": 4.0845e-05, "loss": 0.3056, "step": 1835 }, { "epoch": 2.4, "learning_rate": 4.084e-05, "loss": 0.3749, "step": 1836 }, { "epoch": 2.4, "learning_rate": 4.0835e-05, "loss": 0.3414, "step": 1837 }, { "epoch": 2.41, "learning_rate": 4.083e-05, "loss": 0.3458, "step": 1838 }, { "epoch": 2.41, "learning_rate": 4.0825e-05, "loss": 0.3, "step": 1839 }, { "epoch": 2.41, "learning_rate": 4.0820000000000006e-05, "loss": 0.3048, "step": 1840 }, { "epoch": 2.41, "learning_rate": 4.0815e-05, "loss": 0.3354, "step": 1841 }, { "epoch": 2.41, "learning_rate": 4.0810000000000004e-05, "loss": 0.2943, "step": 1842 }, { "epoch": 2.41, "learning_rate": 4.0805000000000007e-05, "loss": 0.3474, "step": 1843 }, { "epoch": 2.41, "learning_rate": 4.08e-05, "loss": 0.3252, "step": 1844 }, { "epoch": 2.41, "learning_rate": 4.0795e-05, "loss": 0.431, "step": 1845 }, { "epoch": 2.42, "learning_rate": 4.079e-05, "loss": 0.328, "step": 1846 }, { "epoch": 2.42, "learning_rate": 4.0785e-05, "loss": 0.2933, "step": 1847 }, { "epoch": 2.42, "learning_rate": 4.078e-05, "loss": 0.3405, "step": 1848 }, { "epoch": 2.42, "learning_rate": 4.0775e-05, "loss": 0.3414, "step": 1849 }, { "epoch": 2.42, "learning_rate": 4.0770000000000004e-05, "loss": 0.3559, "step": 1850 }, { "epoch": 2.42, "learning_rate": 4.0765e-05, "loss": 0.4188, "step": 1851 }, { "epoch": 2.42, "learning_rate": 4.076e-05, "loss": 0.2406, "step": 1852 }, { "epoch": 2.43, "learning_rate": 4.0755000000000005e-05, "loss": 0.3752, "step": 1853 }, { "epoch": 2.43, "learning_rate": 4.075e-05, "loss": 0.3736, "step": 1854 }, { "epoch": 2.43, "learning_rate": 4.0745e-05, "loss": 0.3293, "step": 1855 }, { "epoch": 2.43, "learning_rate": 4.074e-05, "loss": 0.3737, "step": 1856 }, { "epoch": 2.43, "learning_rate": 4.0735e-05, "loss": 0.2964, "step": 1857 }, { "epoch": 2.43, "learning_rate": 4.0730000000000005e-05, "loss": 0.322, "step": 1858 }, { "epoch": 2.43, "learning_rate": 4.0725e-05, "loss": 0.3356, "step": 1859 }, { "epoch": 2.43, "learning_rate": 4.072e-05, "loss": 0.3987, "step": 1860 }, { "epoch": 2.44, "learning_rate": 4.0715000000000006e-05, "loss": 0.2965, "step": 1861 }, { "epoch": 2.44, "learning_rate": 4.071e-05, "loss": 0.3215, "step": 1862 }, { "epoch": 2.44, "learning_rate": 4.0705000000000004e-05, "loss": 0.3685, "step": 1863 }, { "epoch": 2.44, "learning_rate": 4.07e-05, "loss": 0.3369, "step": 1864 }, { "epoch": 2.44, "learning_rate": 4.0695e-05, "loss": 0.3531, "step": 1865 }, { "epoch": 2.44, "learning_rate": 4.069e-05, "loss": 0.3957, "step": 1866 }, { "epoch": 2.44, "learning_rate": 4.0685e-05, "loss": 0.3223, "step": 1867 }, { "epoch": 2.45, "learning_rate": 4.0680000000000004e-05, "loss": 0.3611, "step": 1868 }, { "epoch": 2.45, "learning_rate": 4.0675e-05, "loss": 0.3873, "step": 1869 }, { "epoch": 2.45, "learning_rate": 4.067e-05, "loss": 0.3736, "step": 1870 }, { "epoch": 2.45, "learning_rate": 4.0665000000000005e-05, "loss": 0.3745, "step": 1871 }, { "epoch": 2.45, "learning_rate": 4.066e-05, "loss": 0.3636, "step": 1872 }, { "epoch": 2.45, "learning_rate": 4.0655e-05, "loss": 0.2981, "step": 1873 }, { "epoch": 2.45, "learning_rate": 4.065e-05, "loss": 0.3844, "step": 1874 }, { "epoch": 2.45, "learning_rate": 4.0645e-05, "loss": 0.4024, "step": 1875 }, { "epoch": 2.46, "learning_rate": 4.064e-05, "loss": 0.3305, "step": 1876 }, { "epoch": 2.46, "learning_rate": 4.0635e-05, "loss": 0.3903, "step": 1877 }, { "epoch": 2.46, "learning_rate": 4.063e-05, "loss": 0.3736, "step": 1878 }, { "epoch": 2.46, "learning_rate": 4.0625000000000005e-05, "loss": 0.3656, "step": 1879 }, { "epoch": 2.46, "learning_rate": 4.062e-05, "loss": 0.3847, "step": 1880 }, { "epoch": 2.46, "learning_rate": 4.0615e-05, "loss": 0.3859, "step": 1881 }, { "epoch": 2.46, "learning_rate": 4.0610000000000006e-05, "loss": 0.3001, "step": 1882 }, { "epoch": 2.46, "learning_rate": 4.0605e-05, "loss": 0.3194, "step": 1883 }, { "epoch": 2.47, "learning_rate": 4.0600000000000004e-05, "loss": 0.279, "step": 1884 }, { "epoch": 2.47, "learning_rate": 4.0595e-05, "loss": 0.3395, "step": 1885 }, { "epoch": 2.47, "learning_rate": 4.059e-05, "loss": 0.343, "step": 1886 }, { "epoch": 2.47, "learning_rate": 4.0585e-05, "loss": 0.3112, "step": 1887 }, { "epoch": 2.47, "learning_rate": 4.058e-05, "loss": 0.4298, "step": 1888 }, { "epoch": 2.47, "learning_rate": 4.0575000000000004e-05, "loss": 0.2715, "step": 1889 }, { "epoch": 2.47, "learning_rate": 4.057e-05, "loss": 0.2858, "step": 1890 }, { "epoch": 2.48, "learning_rate": 4.0565e-05, "loss": 0.3608, "step": 1891 }, { "epoch": 2.48, "learning_rate": 4.0560000000000005e-05, "loss": 0.2997, "step": 1892 }, { "epoch": 2.48, "learning_rate": 4.055500000000001e-05, "loss": 0.4019, "step": 1893 }, { "epoch": 2.48, "learning_rate": 4.055e-05, "loss": 0.3876, "step": 1894 }, { "epoch": 2.48, "learning_rate": 4.0545e-05, "loss": 0.3923, "step": 1895 }, { "epoch": 2.48, "learning_rate": 4.054e-05, "loss": 0.3376, "step": 1896 }, { "epoch": 2.48, "learning_rate": 4.0535000000000004e-05, "loss": 0.3032, "step": 1897 }, { "epoch": 2.48, "learning_rate": 4.053e-05, "loss": 0.3847, "step": 1898 }, { "epoch": 2.49, "learning_rate": 4.0525e-05, "loss": 0.3973, "step": 1899 }, { "epoch": 2.49, "learning_rate": 4.0520000000000005e-05, "loss": 0.2824, "step": 1900 }, { "epoch": 2.49, "learning_rate": 4.0515e-05, "loss": 0.457, "step": 1901 }, { "epoch": 2.49, "learning_rate": 4.0510000000000003e-05, "loss": 0.3124, "step": 1902 }, { "epoch": 2.49, "learning_rate": 4.0505000000000006e-05, "loss": 0.2899, "step": 1903 }, { "epoch": 2.49, "learning_rate": 4.05e-05, "loss": 0.2965, "step": 1904 }, { "epoch": 2.49, "learning_rate": 4.0495e-05, "loss": 0.3596, "step": 1905 }, { "epoch": 2.49, "learning_rate": 4.049e-05, "loss": 0.3727, "step": 1906 }, { "epoch": 2.5, "learning_rate": 4.0485e-05, "loss": 0.463, "step": 1907 }, { "epoch": 2.5, "learning_rate": 4.048e-05, "loss": 0.3569, "step": 1908 }, { "epoch": 2.5, "learning_rate": 4.0475e-05, "loss": 0.3118, "step": 1909 }, { "epoch": 2.5, "learning_rate": 4.0470000000000004e-05, "loss": 0.3549, "step": 1910 }, { "epoch": 2.5, "learning_rate": 4.0465e-05, "loss": 0.3339, "step": 1911 }, { "epoch": 2.5, "learning_rate": 4.046e-05, "loss": 0.3528, "step": 1912 }, { "epoch": 2.5, "learning_rate": 4.0455000000000005e-05, "loss": 0.3114, "step": 1913 }, { "epoch": 2.51, "learning_rate": 4.045000000000001e-05, "loss": 0.3621, "step": 1914 }, { "epoch": 2.51, "learning_rate": 4.0444999999999996e-05, "loss": 0.4449, "step": 1915 }, { "epoch": 2.51, "learning_rate": 4.044e-05, "loss": 0.353, "step": 1916 }, { "epoch": 2.51, "learning_rate": 4.0435e-05, "loss": 0.3421, "step": 1917 }, { "epoch": 2.51, "learning_rate": 4.0430000000000004e-05, "loss": 0.3354, "step": 1918 }, { "epoch": 2.51, "learning_rate": 4.0425e-05, "loss": 0.2878, "step": 1919 }, { "epoch": 2.51, "learning_rate": 4.042e-05, "loss": 0.4211, "step": 1920 }, { "epoch": 2.51, "learning_rate": 4.0415000000000005e-05, "loss": 0.3111, "step": 1921 }, { "epoch": 2.52, "learning_rate": 4.041e-05, "loss": 0.4217, "step": 1922 }, { "epoch": 2.52, "learning_rate": 4.0405000000000004e-05, "loss": 0.3267, "step": 1923 }, { "epoch": 2.52, "learning_rate": 4.0400000000000006e-05, "loss": 0.3275, "step": 1924 }, { "epoch": 2.52, "learning_rate": 4.0395e-05, "loss": 0.3608, "step": 1925 }, { "epoch": 2.52, "learning_rate": 4.039e-05, "loss": 0.3314, "step": 1926 }, { "epoch": 2.52, "learning_rate": 4.0385e-05, "loss": 0.4196, "step": 1927 }, { "epoch": 2.52, "learning_rate": 4.038e-05, "loss": 0.3102, "step": 1928 }, { "epoch": 2.52, "learning_rate": 4.0375e-05, "loss": 0.3281, "step": 1929 }, { "epoch": 2.53, "learning_rate": 4.037e-05, "loss": 0.364, "step": 1930 }, { "epoch": 2.53, "learning_rate": 4.0365000000000004e-05, "loss": 0.3542, "step": 1931 }, { "epoch": 2.53, "learning_rate": 4.0360000000000007e-05, "loss": 0.3358, "step": 1932 }, { "epoch": 2.53, "learning_rate": 4.0355e-05, "loss": 0.3069, "step": 1933 }, { "epoch": 2.53, "learning_rate": 4.0350000000000005e-05, "loss": 0.3884, "step": 1934 }, { "epoch": 2.53, "learning_rate": 4.0345e-05, "loss": 0.3582, "step": 1935 }, { "epoch": 2.53, "learning_rate": 4.034e-05, "loss": 0.355, "step": 1936 }, { "epoch": 2.54, "learning_rate": 4.0335e-05, "loss": 0.3981, "step": 1937 }, { "epoch": 2.54, "learning_rate": 4.033e-05, "loss": 0.3513, "step": 1938 }, { "epoch": 2.54, "learning_rate": 4.0325000000000004e-05, "loss": 0.3902, "step": 1939 }, { "epoch": 2.54, "learning_rate": 4.032e-05, "loss": 0.3943, "step": 1940 }, { "epoch": 2.54, "learning_rate": 4.0315e-05, "loss": 0.354, "step": 1941 }, { "epoch": 2.54, "learning_rate": 4.0310000000000005e-05, "loss": 0.3658, "step": 1942 }, { "epoch": 2.54, "learning_rate": 4.0305e-05, "loss": 0.3507, "step": 1943 }, { "epoch": 2.54, "learning_rate": 4.0300000000000004e-05, "loss": 0.3172, "step": 1944 }, { "epoch": 2.55, "learning_rate": 4.0295e-05, "loss": 0.3145, "step": 1945 }, { "epoch": 2.55, "learning_rate": 4.029e-05, "loss": 0.3091, "step": 1946 }, { "epoch": 2.55, "learning_rate": 4.0285e-05, "loss": 0.3479, "step": 1947 }, { "epoch": 2.55, "learning_rate": 4.028e-05, "loss": 0.3559, "step": 1948 }, { "epoch": 2.55, "learning_rate": 4.0275e-05, "loss": 0.3871, "step": 1949 }, { "epoch": 2.55, "learning_rate": 4.027e-05, "loss": 0.3534, "step": 1950 }, { "epoch": 2.55, "learning_rate": 4.0265e-05, "loss": 0.3505, "step": 1951 }, { "epoch": 2.55, "learning_rate": 4.0260000000000004e-05, "loss": 0.3875, "step": 1952 }, { "epoch": 2.56, "learning_rate": 4.025500000000001e-05, "loss": 0.3907, "step": 1953 }, { "epoch": 2.56, "learning_rate": 4.025e-05, "loss": 0.2574, "step": 1954 }, { "epoch": 2.56, "learning_rate": 4.0245e-05, "loss": 0.3622, "step": 1955 }, { "epoch": 2.56, "learning_rate": 4.024e-05, "loss": 0.3146, "step": 1956 }, { "epoch": 2.56, "learning_rate": 4.0235000000000004e-05, "loss": 0.3393, "step": 1957 }, { "epoch": 2.56, "learning_rate": 4.023e-05, "loss": 0.3141, "step": 1958 }, { "epoch": 2.56, "learning_rate": 4.0225e-05, "loss": 0.3921, "step": 1959 }, { "epoch": 2.57, "learning_rate": 4.0220000000000005e-05, "loss": 0.3441, "step": 1960 }, { "epoch": 2.57, "learning_rate": 4.0215e-05, "loss": 0.3335, "step": 1961 }, { "epoch": 2.57, "learning_rate": 4.021e-05, "loss": 0.4799, "step": 1962 }, { "epoch": 2.57, "learning_rate": 4.0205000000000006e-05, "loss": 0.294, "step": 1963 }, { "epoch": 2.57, "learning_rate": 4.02e-05, "loss": 0.3542, "step": 1964 }, { "epoch": 2.57, "learning_rate": 4.0195e-05, "loss": 0.3136, "step": 1965 }, { "epoch": 2.57, "learning_rate": 4.019e-05, "loss": 0.277, "step": 1966 }, { "epoch": 2.57, "learning_rate": 4.0185e-05, "loss": 0.3058, "step": 1967 }, { "epoch": 2.58, "learning_rate": 4.018e-05, "loss": 0.2766, "step": 1968 }, { "epoch": 2.58, "learning_rate": 4.0175e-05, "loss": 0.3284, "step": 1969 }, { "epoch": 2.58, "learning_rate": 4.017e-05, "loss": 0.3695, "step": 1970 }, { "epoch": 2.58, "learning_rate": 4.0165000000000006e-05, "loss": 0.4057, "step": 1971 }, { "epoch": 2.58, "learning_rate": 4.016e-05, "loss": 0.3651, "step": 1972 }, { "epoch": 2.58, "learning_rate": 4.0155000000000004e-05, "loss": 0.3805, "step": 1973 }, { "epoch": 2.58, "learning_rate": 4.015000000000001e-05, "loss": 0.3536, "step": 1974 }, { "epoch": 2.59, "learning_rate": 4.0144999999999996e-05, "loss": 0.3362, "step": 1975 }, { "epoch": 2.59, "learning_rate": 4.014e-05, "loss": 0.3047, "step": 1976 }, { "epoch": 2.59, "learning_rate": 4.0135e-05, "loss": 0.3169, "step": 1977 }, { "epoch": 2.59, "learning_rate": 4.0130000000000004e-05, "loss": 0.352, "step": 1978 }, { "epoch": 2.59, "learning_rate": 4.0125e-05, "loss": 0.3912, "step": 1979 }, { "epoch": 2.59, "learning_rate": 4.012e-05, "loss": 0.4038, "step": 1980 }, { "epoch": 2.59, "learning_rate": 4.0115000000000005e-05, "loss": 0.4145, "step": 1981 }, { "epoch": 2.59, "learning_rate": 4.011e-05, "loss": 0.4171, "step": 1982 }, { "epoch": 2.6, "learning_rate": 4.0105e-05, "loss": 0.3067, "step": 1983 }, { "epoch": 2.6, "learning_rate": 4.0100000000000006e-05, "loss": 0.405, "step": 1984 }, { "epoch": 2.6, "learning_rate": 4.0095e-05, "loss": 0.3285, "step": 1985 }, { "epoch": 2.6, "learning_rate": 4.009e-05, "loss": 0.3333, "step": 1986 }, { "epoch": 2.6, "learning_rate": 4.0085e-05, "loss": 0.2913, "step": 1987 }, { "epoch": 2.6, "learning_rate": 4.008e-05, "loss": 0.3593, "step": 1988 }, { "epoch": 2.6, "learning_rate": 4.0075e-05, "loss": 0.3486, "step": 1989 }, { "epoch": 2.6, "learning_rate": 4.007e-05, "loss": 0.3708, "step": 1990 }, { "epoch": 2.61, "learning_rate": 4.0065000000000003e-05, "loss": 0.3751, "step": 1991 }, { "epoch": 2.61, "learning_rate": 4.0060000000000006e-05, "loss": 0.3515, "step": 1992 }, { "epoch": 2.61, "learning_rate": 4.0055e-05, "loss": 0.4082, "step": 1993 }, { "epoch": 2.61, "learning_rate": 4.0050000000000004e-05, "loss": 0.3594, "step": 1994 }, { "epoch": 2.61, "learning_rate": 4.0045e-05, "loss": 0.3905, "step": 1995 }, { "epoch": 2.61, "learning_rate": 4.004e-05, "loss": 0.3814, "step": 1996 }, { "epoch": 2.61, "learning_rate": 4.0035e-05, "loss": 0.3407, "step": 1997 }, { "epoch": 2.62, "learning_rate": 4.003e-05, "loss": 0.2639, "step": 1998 }, { "epoch": 2.62, "learning_rate": 4.0025000000000004e-05, "loss": 0.392, "step": 1999 }, { "epoch": 2.62, "learning_rate": 4.002e-05, "loss": 0.3274, "step": 2000 }, { "epoch": 2.62, "learning_rate": 4.0015e-05, "loss": 0.3518, "step": 2001 }, { "epoch": 2.62, "learning_rate": 4.0010000000000005e-05, "loss": 0.3017, "step": 2002 }, { "epoch": 2.62, "learning_rate": 4.0005e-05, "loss": 0.3142, "step": 2003 }, { "epoch": 2.62, "learning_rate": 4e-05, "loss": 0.3865, "step": 2004 }, { "epoch": 2.62, "learning_rate": 3.9995000000000006e-05, "loss": 0.3674, "step": 2005 }, { "epoch": 2.63, "learning_rate": 3.999e-05, "loss": 0.3394, "step": 2006 }, { "epoch": 2.63, "learning_rate": 3.9985e-05, "loss": 0.3544, "step": 2007 }, { "epoch": 2.63, "learning_rate": 3.998e-05, "loss": 0.3147, "step": 2008 }, { "epoch": 2.63, "learning_rate": 3.9975e-05, "loss": 0.3867, "step": 2009 }, { "epoch": 2.63, "learning_rate": 3.9970000000000005e-05, "loss": 0.3062, "step": 2010 }, { "epoch": 2.63, "learning_rate": 3.9965e-05, "loss": 0.3969, "step": 2011 }, { "epoch": 2.63, "learning_rate": 3.9960000000000004e-05, "loss": 0.3453, "step": 2012 }, { "epoch": 2.63, "learning_rate": 3.9955000000000006e-05, "loss": 0.3114, "step": 2013 }, { "epoch": 2.64, "learning_rate": 3.995e-05, "loss": 0.3963, "step": 2014 }, { "epoch": 2.64, "learning_rate": 3.9945000000000005e-05, "loss": 0.3387, "step": 2015 }, { "epoch": 2.64, "learning_rate": 3.994e-05, "loss": 0.4589, "step": 2016 }, { "epoch": 2.64, "learning_rate": 3.9935e-05, "loss": 0.4004, "step": 2017 }, { "epoch": 2.64, "learning_rate": 3.993e-05, "loss": 0.3234, "step": 2018 }, { "epoch": 2.64, "learning_rate": 3.9925e-05, "loss": 0.3351, "step": 2019 }, { "epoch": 2.64, "learning_rate": 3.9920000000000004e-05, "loss": 0.3222, "step": 2020 }, { "epoch": 2.65, "learning_rate": 3.9915e-05, "loss": 0.3768, "step": 2021 }, { "epoch": 2.65, "learning_rate": 3.991e-05, "loss": 0.4757, "step": 2022 }, { "epoch": 2.65, "learning_rate": 3.9905000000000005e-05, "loss": 0.3501, "step": 2023 }, { "epoch": 2.65, "learning_rate": 3.99e-05, "loss": 0.3616, "step": 2024 }, { "epoch": 2.65, "learning_rate": 3.9895000000000003e-05, "loss": 0.3546, "step": 2025 }, { "epoch": 2.65, "learning_rate": 3.989e-05, "loss": 0.3603, "step": 2026 }, { "epoch": 2.65, "learning_rate": 3.9885e-05, "loss": 0.5043, "step": 2027 }, { "epoch": 2.65, "learning_rate": 3.988e-05, "loss": 0.3673, "step": 2028 }, { "epoch": 2.66, "learning_rate": 3.9875e-05, "loss": 0.3671, "step": 2029 }, { "epoch": 2.66, "learning_rate": 3.987e-05, "loss": 0.3281, "step": 2030 }, { "epoch": 2.66, "learning_rate": 3.9865000000000005e-05, "loss": 0.3363, "step": 2031 }, { "epoch": 2.66, "learning_rate": 3.986e-05, "loss": 0.3211, "step": 2032 }, { "epoch": 2.66, "learning_rate": 3.9855000000000004e-05, "loss": 0.3283, "step": 2033 }, { "epoch": 2.66, "learning_rate": 3.9850000000000006e-05, "loss": 0.3677, "step": 2034 }, { "epoch": 2.66, "learning_rate": 3.9845e-05, "loss": 0.3823, "step": 2035 }, { "epoch": 2.66, "learning_rate": 3.984e-05, "loss": 0.3694, "step": 2036 }, { "epoch": 2.67, "learning_rate": 3.9835e-05, "loss": 0.3303, "step": 2037 }, { "epoch": 2.67, "learning_rate": 3.983e-05, "loss": 0.3785, "step": 2038 }, { "epoch": 2.67, "learning_rate": 3.9825e-05, "loss": 0.3727, "step": 2039 }, { "epoch": 2.67, "learning_rate": 3.982e-05, "loss": 0.3473, "step": 2040 }, { "epoch": 2.67, "learning_rate": 3.9815000000000004e-05, "loss": 0.3221, "step": 2041 }, { "epoch": 2.67, "learning_rate": 3.981e-05, "loss": 0.3174, "step": 2042 }, { "epoch": 2.67, "learning_rate": 3.9805e-05, "loss": 0.2973, "step": 2043 }, { "epoch": 2.68, "learning_rate": 3.9800000000000005e-05, "loss": 0.36, "step": 2044 }, { "epoch": 2.68, "learning_rate": 3.979500000000001e-05, "loss": 0.3062, "step": 2045 }, { "epoch": 2.68, "learning_rate": 3.979e-05, "loss": 0.3585, "step": 2046 }, { "epoch": 2.68, "learning_rate": 3.9785e-05, "loss": 0.3475, "step": 2047 }, { "epoch": 2.68, "learning_rate": 3.978e-05, "loss": 0.3851, "step": 2048 }, { "epoch": 2.68, "learning_rate": 3.9775e-05, "loss": 0.4498, "step": 2049 }, { "epoch": 2.68, "learning_rate": 3.977e-05, "loss": 0.4059, "step": 2050 }, { "epoch": 2.68, "learning_rate": 3.9765e-05, "loss": 0.2794, "step": 2051 }, { "epoch": 2.69, "learning_rate": 3.9760000000000006e-05, "loss": 0.3834, "step": 2052 }, { "epoch": 2.69, "learning_rate": 3.9755e-05, "loss": 0.3558, "step": 2053 }, { "epoch": 2.69, "learning_rate": 3.9750000000000004e-05, "loss": 0.4561, "step": 2054 }, { "epoch": 2.69, "learning_rate": 3.9745000000000007e-05, "loss": 0.3138, "step": 2055 }, { "epoch": 2.69, "learning_rate": 3.974e-05, "loss": 0.3181, "step": 2056 }, { "epoch": 2.69, "learning_rate": 3.9735e-05, "loss": 0.3657, "step": 2057 }, { "epoch": 2.69, "learning_rate": 3.973e-05, "loss": 0.3915, "step": 2058 }, { "epoch": 2.7, "learning_rate": 3.9725e-05, "loss": 0.2536, "step": 2059 }, { "epoch": 2.7, "learning_rate": 3.972e-05, "loss": 0.3201, "step": 2060 }, { "epoch": 2.7, "learning_rate": 3.9715e-05, "loss": 0.2973, "step": 2061 }, { "epoch": 2.7, "learning_rate": 3.9710000000000004e-05, "loss": 0.2983, "step": 2062 }, { "epoch": 2.7, "learning_rate": 3.9705e-05, "loss": 0.4328, "step": 2063 }, { "epoch": 2.7, "learning_rate": 3.97e-05, "loss": 0.3735, "step": 2064 }, { "epoch": 2.7, "learning_rate": 3.9695000000000005e-05, "loss": 0.3353, "step": 2065 }, { "epoch": 2.7, "learning_rate": 3.969e-05, "loss": 0.3085, "step": 2066 }, { "epoch": 2.71, "learning_rate": 3.9685e-05, "loss": 0.3053, "step": 2067 }, { "epoch": 2.71, "learning_rate": 3.968e-05, "loss": 0.3517, "step": 2068 }, { "epoch": 2.71, "learning_rate": 3.9675e-05, "loss": 0.3695, "step": 2069 }, { "epoch": 2.71, "learning_rate": 3.9670000000000005e-05, "loss": 0.3424, "step": 2070 }, { "epoch": 2.71, "learning_rate": 3.9665e-05, "loss": 0.3276, "step": 2071 }, { "epoch": 2.71, "learning_rate": 3.966e-05, "loss": 0.3246, "step": 2072 }, { "epoch": 2.71, "learning_rate": 3.9655000000000006e-05, "loss": 0.3121, "step": 2073 }, { "epoch": 2.71, "learning_rate": 3.965e-05, "loss": 0.2926, "step": 2074 }, { "epoch": 2.72, "learning_rate": 3.9645000000000004e-05, "loss": 0.4189, "step": 2075 }, { "epoch": 2.72, "learning_rate": 3.964e-05, "loss": 0.3517, "step": 2076 }, { "epoch": 2.72, "learning_rate": 3.9635e-05, "loss": 0.3133, "step": 2077 }, { "epoch": 2.72, "learning_rate": 3.963e-05, "loss": 0.3965, "step": 2078 }, { "epoch": 2.72, "learning_rate": 3.9625e-05, "loss": 0.4023, "step": 2079 }, { "epoch": 2.72, "learning_rate": 3.9620000000000004e-05, "loss": 0.3506, "step": 2080 }, { "epoch": 2.72, "learning_rate": 3.9615e-05, "loss": 0.3768, "step": 2081 }, { "epoch": 2.73, "learning_rate": 3.961e-05, "loss": 0.3514, "step": 2082 }, { "epoch": 2.73, "learning_rate": 3.9605000000000005e-05, "loss": 0.3087, "step": 2083 }, { "epoch": 2.73, "learning_rate": 3.960000000000001e-05, "loss": 0.3844, "step": 2084 }, { "epoch": 2.73, "learning_rate": 3.9595e-05, "loss": 0.4455, "step": 2085 }, { "epoch": 2.73, "learning_rate": 3.959e-05, "loss": 0.39, "step": 2086 }, { "epoch": 2.73, "learning_rate": 3.9585e-05, "loss": 0.3586, "step": 2087 }, { "epoch": 2.73, "learning_rate": 3.958e-05, "loss": 0.3862, "step": 2088 }, { "epoch": 2.73, "learning_rate": 3.9575e-05, "loss": 0.3552, "step": 2089 }, { "epoch": 2.74, "learning_rate": 3.957e-05, "loss": 0.3988, "step": 2090 }, { "epoch": 2.74, "learning_rate": 3.9565000000000005e-05, "loss": 0.4096, "step": 2091 }, { "epoch": 2.74, "learning_rate": 3.956e-05, "loss": 0.3731, "step": 2092 }, { "epoch": 2.74, "learning_rate": 3.9555e-05, "loss": 0.3867, "step": 2093 }, { "epoch": 2.74, "learning_rate": 3.9550000000000006e-05, "loss": 0.3715, "step": 2094 }, { "epoch": 2.74, "learning_rate": 3.9545e-05, "loss": 0.315, "step": 2095 }, { "epoch": 2.74, "learning_rate": 3.954e-05, "loss": 0.3273, "step": 2096 }, { "epoch": 2.74, "learning_rate": 3.9535e-05, "loss": 0.3192, "step": 2097 }, { "epoch": 2.75, "learning_rate": 3.953e-05, "loss": 0.2966, "step": 2098 }, { "epoch": 2.75, "learning_rate": 3.9525e-05, "loss": 0.3566, "step": 2099 }, { "epoch": 2.75, "learning_rate": 3.952e-05, "loss": 0.2955, "step": 2100 }, { "epoch": 2.75, "learning_rate": 3.9515000000000004e-05, "loss": 0.406, "step": 2101 }, { "epoch": 2.75, "learning_rate": 3.951e-05, "loss": 0.3417, "step": 2102 }, { "epoch": 2.75, "learning_rate": 3.9505e-05, "loss": 0.3727, "step": 2103 }, { "epoch": 2.75, "learning_rate": 3.9500000000000005e-05, "loss": 0.3218, "step": 2104 }, { "epoch": 2.76, "learning_rate": 3.949500000000001e-05, "loss": 0.3099, "step": 2105 }, { "epoch": 2.76, "learning_rate": 3.9489999999999996e-05, "loss": 0.3741, "step": 2106 }, { "epoch": 2.76, "learning_rate": 3.9485e-05, "loss": 0.3754, "step": 2107 }, { "epoch": 2.76, "learning_rate": 3.948e-05, "loss": 0.4158, "step": 2108 }, { "epoch": 2.76, "learning_rate": 3.9475000000000004e-05, "loss": 0.3372, "step": 2109 }, { "epoch": 2.76, "learning_rate": 3.947e-05, "loss": 0.3395, "step": 2110 }, { "epoch": 2.76, "learning_rate": 3.9465e-05, "loss": 0.4418, "step": 2111 }, { "epoch": 2.76, "learning_rate": 3.9460000000000005e-05, "loss": 0.3166, "step": 2112 }, { "epoch": 2.77, "learning_rate": 3.9455e-05, "loss": 0.3565, "step": 2113 }, { "epoch": 2.77, "learning_rate": 3.9450000000000003e-05, "loss": 0.3647, "step": 2114 }, { "epoch": 2.77, "learning_rate": 3.9445000000000006e-05, "loss": 0.3598, "step": 2115 }, { "epoch": 2.77, "learning_rate": 3.944e-05, "loss": 0.3991, "step": 2116 }, { "epoch": 2.77, "learning_rate": 3.9435e-05, "loss": 0.3914, "step": 2117 }, { "epoch": 2.77, "learning_rate": 3.943e-05, "loss": 0.2433, "step": 2118 }, { "epoch": 2.77, "learning_rate": 3.9425e-05, "loss": 0.3602, "step": 2119 }, { "epoch": 2.77, "learning_rate": 3.942e-05, "loss": 0.3086, "step": 2120 }, { "epoch": 2.78, "learning_rate": 3.9415e-05, "loss": 0.3214, "step": 2121 }, { "epoch": 2.78, "learning_rate": 3.9410000000000004e-05, "loss": 0.4753, "step": 2122 }, { "epoch": 2.78, "learning_rate": 3.9405e-05, "loss": 0.3987, "step": 2123 }, { "epoch": 2.78, "learning_rate": 3.94e-05, "loss": 0.3298, "step": 2124 }, { "epoch": 2.78, "learning_rate": 3.9395000000000005e-05, "loss": 0.3705, "step": 2125 }, { "epoch": 2.78, "learning_rate": 3.939e-05, "loss": 0.3801, "step": 2126 }, { "epoch": 2.78, "learning_rate": 3.9384999999999996e-05, "loss": 0.3036, "step": 2127 }, { "epoch": 2.79, "learning_rate": 3.938e-05, "loss": 0.2943, "step": 2128 }, { "epoch": 2.79, "learning_rate": 3.9375e-05, "loss": 0.3567, "step": 2129 }, { "epoch": 2.79, "learning_rate": 3.9370000000000004e-05, "loss": 0.3797, "step": 2130 }, { "epoch": 2.79, "learning_rate": 3.9365e-05, "loss": 0.3006, "step": 2131 }, { "epoch": 2.79, "learning_rate": 3.936e-05, "loss": 0.2769, "step": 2132 }, { "epoch": 2.79, "learning_rate": 3.9355000000000005e-05, "loss": 0.3399, "step": 2133 }, { "epoch": 2.79, "learning_rate": 3.935e-05, "loss": 0.3453, "step": 2134 }, { "epoch": 2.79, "learning_rate": 3.9345000000000004e-05, "loss": 0.3499, "step": 2135 }, { "epoch": 2.8, "learning_rate": 3.9340000000000006e-05, "loss": 0.3166, "step": 2136 }, { "epoch": 2.8, "learning_rate": 3.9335e-05, "loss": 0.3433, "step": 2137 }, { "epoch": 2.8, "learning_rate": 3.933e-05, "loss": 0.379, "step": 2138 }, { "epoch": 2.8, "learning_rate": 3.9325e-05, "loss": 0.3577, "step": 2139 }, { "epoch": 2.8, "learning_rate": 3.932e-05, "loss": 0.322, "step": 2140 }, { "epoch": 2.8, "learning_rate": 3.9315e-05, "loss": 0.3339, "step": 2141 }, { "epoch": 2.8, "learning_rate": 3.931e-05, "loss": 0.3165, "step": 2142 }, { "epoch": 2.8, "learning_rate": 3.9305000000000004e-05, "loss": 0.4252, "step": 2143 }, { "epoch": 2.81, "learning_rate": 3.9300000000000007e-05, "loss": 0.3811, "step": 2144 }, { "epoch": 2.81, "learning_rate": 3.9295e-05, "loss": 0.3691, "step": 2145 }, { "epoch": 2.81, "learning_rate": 3.9290000000000005e-05, "loss": 0.4013, "step": 2146 }, { "epoch": 2.81, "learning_rate": 3.9285e-05, "loss": 0.3839, "step": 2147 }, { "epoch": 2.81, "learning_rate": 3.9280000000000003e-05, "loss": 0.3021, "step": 2148 }, { "epoch": 2.81, "learning_rate": 3.9275e-05, "loss": 0.3311, "step": 2149 }, { "epoch": 2.81, "learning_rate": 3.927e-05, "loss": 0.3191, "step": 2150 }, { "epoch": 2.82, "learning_rate": 3.9265000000000004e-05, "loss": 0.3228, "step": 2151 }, { "epoch": 2.82, "learning_rate": 3.926e-05, "loss": 0.2992, "step": 2152 }, { "epoch": 2.82, "learning_rate": 3.9255e-05, "loss": 0.3077, "step": 2153 }, { "epoch": 2.82, "learning_rate": 3.9250000000000005e-05, "loss": 0.3489, "step": 2154 }, { "epoch": 2.82, "learning_rate": 3.9245e-05, "loss": 0.3957, "step": 2155 }, { "epoch": 2.82, "learning_rate": 3.9240000000000004e-05, "loss": 0.3527, "step": 2156 }, { "epoch": 2.82, "learning_rate": 3.9235e-05, "loss": 0.4673, "step": 2157 }, { "epoch": 2.82, "learning_rate": 3.923e-05, "loss": 0.3519, "step": 2158 }, { "epoch": 2.83, "learning_rate": 3.9225e-05, "loss": 0.3471, "step": 2159 }, { "epoch": 2.83, "learning_rate": 3.922e-05, "loss": 0.3161, "step": 2160 }, { "epoch": 2.83, "learning_rate": 3.9215e-05, "loss": 0.3125, "step": 2161 }, { "epoch": 2.83, "learning_rate": 3.921e-05, "loss": 0.3479, "step": 2162 }, { "epoch": 2.83, "learning_rate": 3.9205e-05, "loss": 0.35, "step": 2163 }, { "epoch": 2.83, "learning_rate": 3.9200000000000004e-05, "loss": 0.3094, "step": 2164 }, { "epoch": 2.83, "learning_rate": 3.919500000000001e-05, "loss": 0.4238, "step": 2165 }, { "epoch": 2.84, "learning_rate": 3.919e-05, "loss": 0.3569, "step": 2166 }, { "epoch": 2.84, "learning_rate": 3.9185e-05, "loss": 0.2969, "step": 2167 }, { "epoch": 2.84, "learning_rate": 3.918e-05, "loss": 0.4248, "step": 2168 }, { "epoch": 2.84, "learning_rate": 3.9175000000000004e-05, "loss": 0.3741, "step": 2169 }, { "epoch": 2.84, "learning_rate": 3.917e-05, "loss": 0.3347, "step": 2170 }, { "epoch": 2.84, "learning_rate": 3.9165e-05, "loss": 0.3477, "step": 2171 }, { "epoch": 2.84, "learning_rate": 3.9160000000000005e-05, "loss": 0.4127, "step": 2172 }, { "epoch": 2.84, "learning_rate": 3.9155e-05, "loss": 0.3885, "step": 2173 }, { "epoch": 2.85, "learning_rate": 3.915e-05, "loss": 0.34, "step": 2174 }, { "epoch": 2.85, "learning_rate": 3.9145000000000006e-05, "loss": 0.3091, "step": 2175 }, { "epoch": 2.85, "learning_rate": 3.914e-05, "loss": 0.3324, "step": 2176 }, { "epoch": 2.85, "learning_rate": 3.9135e-05, "loss": 0.3018, "step": 2177 }, { "epoch": 2.85, "learning_rate": 3.913e-05, "loss": 0.3195, "step": 2178 }, { "epoch": 2.85, "learning_rate": 3.9125e-05, "loss": 0.3756, "step": 2179 }, { "epoch": 2.85, "learning_rate": 3.912e-05, "loss": 0.3127, "step": 2180 }, { "epoch": 2.85, "learning_rate": 3.9115e-05, "loss": 0.3331, "step": 2181 }, { "epoch": 2.86, "learning_rate": 3.911e-05, "loss": 0.3245, "step": 2182 }, { "epoch": 2.86, "learning_rate": 3.9105000000000006e-05, "loss": 0.3203, "step": 2183 }, { "epoch": 2.86, "learning_rate": 3.91e-05, "loss": 0.3262, "step": 2184 }, { "epoch": 2.86, "learning_rate": 3.9095000000000004e-05, "loss": 0.4796, "step": 2185 }, { "epoch": 2.86, "learning_rate": 3.909000000000001e-05, "loss": 0.3845, "step": 2186 }, { "epoch": 2.86, "learning_rate": 3.9085e-05, "loss": 0.2725, "step": 2187 }, { "epoch": 2.86, "learning_rate": 3.908e-05, "loss": 0.3443, "step": 2188 }, { "epoch": 2.87, "learning_rate": 3.9075e-05, "loss": 0.3239, "step": 2189 }, { "epoch": 2.87, "learning_rate": 3.9070000000000004e-05, "loss": 0.3233, "step": 2190 }, { "epoch": 2.87, "learning_rate": 3.9065e-05, "loss": 0.2848, "step": 2191 }, { "epoch": 2.87, "learning_rate": 3.906e-05, "loss": 0.4323, "step": 2192 }, { "epoch": 2.87, "learning_rate": 3.9055000000000005e-05, "loss": 0.3577, "step": 2193 }, { "epoch": 2.87, "learning_rate": 3.905e-05, "loss": 0.4078, "step": 2194 }, { "epoch": 2.87, "learning_rate": 3.9045e-05, "loss": 0.4691, "step": 2195 }, { "epoch": 2.87, "learning_rate": 3.9040000000000006e-05, "loss": 0.4545, "step": 2196 }, { "epoch": 2.88, "learning_rate": 3.9035e-05, "loss": 0.4028, "step": 2197 }, { "epoch": 2.88, "learning_rate": 3.903e-05, "loss": 0.3825, "step": 2198 }, { "epoch": 2.88, "learning_rate": 3.9025e-05, "loss": 0.3502, "step": 2199 }, { "epoch": 2.88, "learning_rate": 3.902e-05, "loss": 0.3968, "step": 2200 }, { "epoch": 2.88, "learning_rate": 3.9015e-05, "loss": 0.3001, "step": 2201 }, { "epoch": 2.88, "learning_rate": 3.901e-05, "loss": 0.2937, "step": 2202 }, { "epoch": 2.88, "learning_rate": 3.9005000000000003e-05, "loss": 0.3334, "step": 2203 }, { "epoch": 2.88, "learning_rate": 3.9000000000000006e-05, "loss": 0.478, "step": 2204 }, { "epoch": 2.89, "learning_rate": 3.8995e-05, "loss": 0.2333, "step": 2205 }, { "epoch": 2.89, "learning_rate": 3.8990000000000004e-05, "loss": 0.3829, "step": 2206 }, { "epoch": 2.89, "learning_rate": 3.8985e-05, "loss": 0.3906, "step": 2207 }, { "epoch": 2.89, "learning_rate": 3.898e-05, "loss": 0.3699, "step": 2208 }, { "epoch": 2.89, "learning_rate": 3.8975e-05, "loss": 0.3615, "step": 2209 }, { "epoch": 2.89, "learning_rate": 3.897e-05, "loss": 0.3205, "step": 2210 }, { "epoch": 2.89, "learning_rate": 3.8965000000000004e-05, "loss": 0.3541, "step": 2211 }, { "epoch": 2.9, "learning_rate": 3.896e-05, "loss": 0.3946, "step": 2212 }, { "epoch": 2.9, "learning_rate": 3.8955e-05, "loss": 0.4003, "step": 2213 }, { "epoch": 2.9, "learning_rate": 3.8950000000000005e-05, "loss": 0.2914, "step": 2214 }, { "epoch": 2.9, "learning_rate": 3.8945e-05, "loss": 0.4146, "step": 2215 }, { "epoch": 2.9, "learning_rate": 3.894e-05, "loss": 0.3312, "step": 2216 }, { "epoch": 2.9, "learning_rate": 3.8935e-05, "loss": 0.3657, "step": 2217 }, { "epoch": 2.9, "learning_rate": 3.893e-05, "loss": 0.3677, "step": 2218 }, { "epoch": 2.9, "learning_rate": 3.8925e-05, "loss": 0.3457, "step": 2219 }, { "epoch": 2.91, "learning_rate": 3.892e-05, "loss": 0.3544, "step": 2220 }, { "epoch": 2.91, "learning_rate": 3.8915e-05, "loss": 0.3513, "step": 2221 }, { "epoch": 2.91, "learning_rate": 3.8910000000000005e-05, "loss": 0.3333, "step": 2222 }, { "epoch": 2.91, "learning_rate": 3.8905e-05, "loss": 0.3839, "step": 2223 }, { "epoch": 2.91, "learning_rate": 3.8900000000000004e-05, "loss": 0.3348, "step": 2224 }, { "epoch": 2.91, "learning_rate": 3.8895000000000006e-05, "loss": 0.4732, "step": 2225 }, { "epoch": 2.91, "learning_rate": 3.889e-05, "loss": 0.2938, "step": 2226 }, { "epoch": 2.91, "learning_rate": 3.8885e-05, "loss": 0.3608, "step": 2227 }, { "epoch": 2.92, "learning_rate": 3.888e-05, "loss": 0.3225, "step": 2228 }, { "epoch": 2.92, "learning_rate": 3.8875e-05, "loss": 0.3497, "step": 2229 }, { "epoch": 2.92, "learning_rate": 3.887e-05, "loss": 0.3939, "step": 2230 }, { "epoch": 2.92, "learning_rate": 3.8865e-05, "loss": 0.3885, "step": 2231 }, { "epoch": 2.92, "learning_rate": 3.8860000000000004e-05, "loss": 0.5581, "step": 2232 }, { "epoch": 2.92, "learning_rate": 3.8855e-05, "loss": 0.3681, "step": 2233 }, { "epoch": 2.92, "learning_rate": 3.885e-05, "loss": 0.3383, "step": 2234 }, { "epoch": 2.93, "learning_rate": 3.8845000000000005e-05, "loss": 0.3911, "step": 2235 }, { "epoch": 2.93, "learning_rate": 3.884e-05, "loss": 0.3524, "step": 2236 }, { "epoch": 2.93, "learning_rate": 3.8835e-05, "loss": 0.3077, "step": 2237 }, { "epoch": 2.93, "learning_rate": 3.883e-05, "loss": 0.3071, "step": 2238 }, { "epoch": 2.93, "learning_rate": 3.8825e-05, "loss": 0.3446, "step": 2239 }, { "epoch": 2.93, "learning_rate": 3.882e-05, "loss": 0.551, "step": 2240 }, { "epoch": 2.93, "learning_rate": 3.8815e-05, "loss": 0.3325, "step": 2241 }, { "epoch": 2.93, "learning_rate": 3.881e-05, "loss": 0.375, "step": 2242 }, { "epoch": 2.94, "learning_rate": 3.8805000000000005e-05, "loss": 0.3296, "step": 2243 }, { "epoch": 2.94, "learning_rate": 3.88e-05, "loss": 0.3733, "step": 2244 }, { "epoch": 2.94, "learning_rate": 3.8795000000000004e-05, "loss": 0.3502, "step": 2245 }, { "epoch": 2.94, "learning_rate": 3.8790000000000006e-05, "loss": 0.3342, "step": 2246 }, { "epoch": 2.94, "learning_rate": 3.8785e-05, "loss": 0.3906, "step": 2247 }, { "epoch": 2.94, "learning_rate": 3.878e-05, "loss": 0.3529, "step": 2248 }, { "epoch": 2.94, "learning_rate": 3.8775e-05, "loss": 0.3154, "step": 2249 }, { "epoch": 2.95, "learning_rate": 3.877e-05, "loss": 0.3415, "step": 2250 }, { "epoch": 2.95, "learning_rate": 3.8765e-05, "loss": 0.3821, "step": 2251 }, { "epoch": 2.95, "learning_rate": 3.876e-05, "loss": 0.3084, "step": 2252 }, { "epoch": 2.95, "learning_rate": 3.8755000000000004e-05, "loss": 0.3834, "step": 2253 }, { "epoch": 2.95, "learning_rate": 3.875e-05, "loss": 0.2939, "step": 2254 }, { "epoch": 2.95, "learning_rate": 3.8745e-05, "loss": 0.4187, "step": 2255 }, { "epoch": 2.95, "learning_rate": 3.8740000000000005e-05, "loss": 0.3591, "step": 2256 }, { "epoch": 2.95, "learning_rate": 3.873500000000001e-05, "loss": 0.33, "step": 2257 }, { "epoch": 2.96, "learning_rate": 3.873e-05, "loss": 0.2997, "step": 2258 }, { "epoch": 2.96, "learning_rate": 3.8725e-05, "loss": 0.4219, "step": 2259 }, { "epoch": 2.96, "learning_rate": 3.872e-05, "loss": 0.3514, "step": 2260 }, { "epoch": 2.96, "learning_rate": 3.8715000000000005e-05, "loss": 0.3604, "step": 2261 }, { "epoch": 2.96, "learning_rate": 3.871e-05, "loss": 0.3819, "step": 2262 }, { "epoch": 2.96, "learning_rate": 3.8705e-05, "loss": 0.3438, "step": 2263 }, { "epoch": 2.96, "learning_rate": 3.8700000000000006e-05, "loss": 0.3181, "step": 2264 }, { "epoch": 2.96, "learning_rate": 3.8695e-05, "loss": 0.4213, "step": 2265 }, { "epoch": 2.97, "learning_rate": 3.8690000000000004e-05, "loss": 0.4253, "step": 2266 }, { "epoch": 2.97, "learning_rate": 3.8685000000000007e-05, "loss": 0.3352, "step": 2267 }, { "epoch": 2.97, "learning_rate": 3.868e-05, "loss": 0.3154, "step": 2268 }, { "epoch": 2.97, "learning_rate": 3.8675e-05, "loss": 0.3981, "step": 2269 }, { "epoch": 2.97, "learning_rate": 3.867e-05, "loss": 0.3172, "step": 2270 }, { "epoch": 2.97, "learning_rate": 3.8665e-05, "loss": 0.3335, "step": 2271 }, { "epoch": 2.97, "learning_rate": 3.866e-05, "loss": 0.3302, "step": 2272 }, { "epoch": 2.98, "learning_rate": 3.8655e-05, "loss": 0.3963, "step": 2273 }, { "epoch": 2.98, "learning_rate": 3.8650000000000004e-05, "loss": 0.3598, "step": 2274 }, { "epoch": 2.98, "learning_rate": 3.8645e-05, "loss": 0.3352, "step": 2275 }, { "epoch": 2.98, "learning_rate": 3.864e-05, "loss": 0.4012, "step": 2276 }, { "epoch": 2.98, "learning_rate": 3.8635000000000005e-05, "loss": 0.3249, "step": 2277 }, { "epoch": 2.98, "learning_rate": 3.863e-05, "loss": 0.3611, "step": 2278 }, { "epoch": 2.98, "learning_rate": 3.8625e-05, "loss": 0.4381, "step": 2279 }, { "epoch": 2.98, "learning_rate": 3.862e-05, "loss": 0.3648, "step": 2280 }, { "epoch": 2.99, "learning_rate": 3.8615e-05, "loss": 0.3679, "step": 2281 }, { "epoch": 2.99, "learning_rate": 3.8610000000000005e-05, "loss": 0.3302, "step": 2282 }, { "epoch": 2.99, "learning_rate": 3.8605e-05, "loss": 0.3491, "step": 2283 }, { "epoch": 2.99, "learning_rate": 3.86e-05, "loss": 0.2951, "step": 2284 }, { "epoch": 2.99, "learning_rate": 3.8595000000000006e-05, "loss": 0.3828, "step": 2285 }, { "epoch": 2.99, "learning_rate": 3.859e-05, "loss": 0.3982, "step": 2286 }, { "epoch": 2.99, "learning_rate": 3.8585000000000004e-05, "loss": 0.3322, "step": 2287 }, { "epoch": 2.99, "learning_rate": 3.858e-05, "loss": 0.4381, "step": 2288 }, { "epoch": 3.0, "learning_rate": 3.8575e-05, "loss": 0.3841, "step": 2289 }, { "epoch": 3.0, "learning_rate": 3.857e-05, "loss": 0.3365, "step": 2290 }, { "epoch": 3.0, "learning_rate": 3.8565e-05, "loss": 0.4769, "step": 2291 }, { "epoch": 3.0, "learning_rate": 3.8560000000000004e-05, "loss": 0.332, "step": 2292 }, { "epoch": 3.0, "learning_rate": 3.8555e-05, "loss": 0.1694, "step": 2293 }, { "epoch": 3.0, "learning_rate": 3.855e-05, "loss": 0.1913, "step": 2294 }, { "epoch": 3.0, "learning_rate": 3.8545000000000004e-05, "loss": 0.1581, "step": 2295 }, { "epoch": 3.01, "learning_rate": 3.854000000000001e-05, "loss": 0.1788, "step": 2296 }, { "epoch": 3.01, "learning_rate": 3.8535e-05, "loss": 0.1341, "step": 2297 }, { "epoch": 3.01, "learning_rate": 3.853e-05, "loss": 0.126, "step": 2298 }, { "epoch": 3.01, "learning_rate": 3.8525e-05, "loss": 0.1357, "step": 2299 }, { "epoch": 3.01, "learning_rate": 3.8520000000000004e-05, "loss": 0.1671, "step": 2300 }, { "epoch": 3.01, "learning_rate": 3.8515e-05, "loss": 0.1385, "step": 2301 }, { "epoch": 3.01, "learning_rate": 3.851e-05, "loss": 0.1836, "step": 2302 }, { "epoch": 3.01, "learning_rate": 3.8505000000000005e-05, "loss": 0.1718, "step": 2303 }, { "epoch": 3.02, "learning_rate": 3.85e-05, "loss": 0.148, "step": 2304 }, { "epoch": 3.02, "learning_rate": 3.8495e-05, "loss": 0.1634, "step": 2305 }, { "epoch": 3.02, "learning_rate": 3.8490000000000006e-05, "loss": 0.149, "step": 2306 }, { "epoch": 3.02, "learning_rate": 3.8485e-05, "loss": 0.1442, "step": 2307 }, { "epoch": 3.02, "learning_rate": 3.848e-05, "loss": 0.1526, "step": 2308 }, { "epoch": 3.02, "learning_rate": 3.8475e-05, "loss": 0.1401, "step": 2309 }, { "epoch": 3.02, "learning_rate": 3.847e-05, "loss": 0.1137, "step": 2310 }, { "epoch": 3.02, "learning_rate": 3.8465e-05, "loss": 0.1519, "step": 2311 }, { "epoch": 3.03, "learning_rate": 3.846e-05, "loss": 0.1247, "step": 2312 }, { "epoch": 3.03, "learning_rate": 3.8455000000000004e-05, "loss": 0.1412, "step": 2313 }, { "epoch": 3.03, "learning_rate": 3.845e-05, "loss": 0.1283, "step": 2314 }, { "epoch": 3.03, "learning_rate": 3.8445e-05, "loss": 0.1489, "step": 2315 }, { "epoch": 3.03, "learning_rate": 3.8440000000000005e-05, "loss": 0.1191, "step": 2316 }, { "epoch": 3.03, "learning_rate": 3.843500000000001e-05, "loss": 0.1726, "step": 2317 }, { "epoch": 3.03, "learning_rate": 3.8429999999999996e-05, "loss": 0.1837, "step": 2318 }, { "epoch": 3.04, "learning_rate": 3.8425e-05, "loss": 0.1449, "step": 2319 }, { "epoch": 3.04, "learning_rate": 3.842e-05, "loss": 0.1189, "step": 2320 }, { "epoch": 3.04, "learning_rate": 3.8415000000000004e-05, "loss": 0.1732, "step": 2321 }, { "epoch": 3.04, "learning_rate": 3.841e-05, "loss": 0.154, "step": 2322 }, { "epoch": 3.04, "learning_rate": 3.8405e-05, "loss": 0.1506, "step": 2323 }, { "epoch": 3.04, "learning_rate": 3.8400000000000005e-05, "loss": 0.1587, "step": 2324 }, { "epoch": 3.04, "learning_rate": 3.8395e-05, "loss": 0.1511, "step": 2325 }, { "epoch": 3.04, "learning_rate": 3.8390000000000003e-05, "loss": 0.139, "step": 2326 }, { "epoch": 3.05, "learning_rate": 3.8385000000000006e-05, "loss": 0.1291, "step": 2327 }, { "epoch": 3.05, "learning_rate": 3.838e-05, "loss": 0.157, "step": 2328 }, { "epoch": 3.05, "learning_rate": 3.8375e-05, "loss": 0.1223, "step": 2329 }, { "epoch": 3.05, "learning_rate": 3.837e-05, "loss": 0.1402, "step": 2330 }, { "epoch": 3.05, "learning_rate": 3.8365e-05, "loss": 0.1695, "step": 2331 }, { "epoch": 3.05, "learning_rate": 3.836e-05, "loss": 0.1395, "step": 2332 }, { "epoch": 3.05, "learning_rate": 3.8355e-05, "loss": 0.1465, "step": 2333 }, { "epoch": 3.05, "learning_rate": 3.8350000000000004e-05, "loss": 0.1703, "step": 2334 }, { "epoch": 3.06, "learning_rate": 3.8345000000000006e-05, "loss": 0.1352, "step": 2335 }, { "epoch": 3.06, "learning_rate": 3.834e-05, "loss": 0.1411, "step": 2336 }, { "epoch": 3.06, "learning_rate": 3.8335000000000005e-05, "loss": 0.2126, "step": 2337 }, { "epoch": 3.06, "learning_rate": 3.833e-05, "loss": 0.1434, "step": 2338 }, { "epoch": 3.06, "learning_rate": 3.8324999999999996e-05, "loss": 0.1473, "step": 2339 }, { "epoch": 3.06, "learning_rate": 3.832e-05, "loss": 0.1447, "step": 2340 }, { "epoch": 3.06, "learning_rate": 3.8315e-05, "loss": 0.1425, "step": 2341 }, { "epoch": 3.07, "learning_rate": 3.8310000000000004e-05, "loss": 0.1535, "step": 2342 }, { "epoch": 3.07, "learning_rate": 3.8305e-05, "loss": 0.1271, "step": 2343 }, { "epoch": 3.07, "learning_rate": 3.83e-05, "loss": 0.1623, "step": 2344 }, { "epoch": 3.07, "learning_rate": 3.8295000000000005e-05, "loss": 0.1424, "step": 2345 }, { "epoch": 3.07, "learning_rate": 3.829e-05, "loss": 0.1472, "step": 2346 }, { "epoch": 3.07, "learning_rate": 3.8285000000000004e-05, "loss": 0.1603, "step": 2347 }, { "epoch": 3.07, "learning_rate": 3.828e-05, "loss": 0.1878, "step": 2348 }, { "epoch": 3.07, "learning_rate": 3.8275e-05, "loss": 0.1335, "step": 2349 }, { "epoch": 3.08, "learning_rate": 3.827e-05, "loss": 0.1569, "step": 2350 }, { "epoch": 3.08, "learning_rate": 3.8265e-05, "loss": 0.1588, "step": 2351 }, { "epoch": 3.08, "learning_rate": 3.826e-05, "loss": 0.1511, "step": 2352 }, { "epoch": 3.08, "learning_rate": 3.8255e-05, "loss": 0.1501, "step": 2353 }, { "epoch": 3.08, "learning_rate": 3.825e-05, "loss": 0.1561, "step": 2354 }, { "epoch": 3.08, "learning_rate": 3.8245000000000004e-05, "loss": 0.1588, "step": 2355 }, { "epoch": 3.08, "learning_rate": 3.8240000000000007e-05, "loss": 0.1494, "step": 2356 }, { "epoch": 3.09, "learning_rate": 3.8235e-05, "loss": 0.1404, "step": 2357 }, { "epoch": 3.09, "learning_rate": 3.823e-05, "loss": 0.1528, "step": 2358 }, { "epoch": 3.09, "learning_rate": 3.8225e-05, "loss": 0.1407, "step": 2359 }, { "epoch": 3.09, "learning_rate": 3.822e-05, "loss": 0.1688, "step": 2360 }, { "epoch": 3.09, "learning_rate": 3.8215e-05, "loss": 0.1267, "step": 2361 }, { "epoch": 3.09, "learning_rate": 3.821e-05, "loss": 0.1055, "step": 2362 }, { "epoch": 3.09, "learning_rate": 3.8205000000000004e-05, "loss": 0.1278, "step": 2363 }, { "epoch": 3.09, "learning_rate": 3.82e-05, "loss": 0.1304, "step": 2364 }, { "epoch": 3.1, "learning_rate": 3.8195e-05, "loss": 0.1568, "step": 2365 }, { "epoch": 3.1, "learning_rate": 3.8190000000000005e-05, "loss": 0.1596, "step": 2366 }, { "epoch": 3.1, "learning_rate": 3.8185e-05, "loss": 0.1394, "step": 2367 }, { "epoch": 3.1, "learning_rate": 3.818e-05, "loss": 0.1414, "step": 2368 }, { "epoch": 3.1, "learning_rate": 3.8175e-05, "loss": 0.1654, "step": 2369 }, { "epoch": 3.1, "learning_rate": 3.817e-05, "loss": 0.1472, "step": 2370 }, { "epoch": 3.1, "learning_rate": 3.8165e-05, "loss": 0.1317, "step": 2371 }, { "epoch": 3.1, "learning_rate": 3.816e-05, "loss": 0.1674, "step": 2372 }, { "epoch": 3.11, "learning_rate": 3.8155e-05, "loss": 0.1583, "step": 2373 }, { "epoch": 3.11, "learning_rate": 3.8150000000000006e-05, "loss": 0.1404, "step": 2374 }, { "epoch": 3.11, "learning_rate": 3.8145e-05, "loss": 0.1477, "step": 2375 }, { "epoch": 3.11, "learning_rate": 3.8140000000000004e-05, "loss": 0.1557, "step": 2376 }, { "epoch": 3.11, "learning_rate": 3.813500000000001e-05, "loss": 0.1376, "step": 2377 }, { "epoch": 3.11, "learning_rate": 3.8129999999999996e-05, "loss": 0.1654, "step": 2378 }, { "epoch": 3.11, "learning_rate": 3.8125e-05, "loss": 0.1243, "step": 2379 }, { "epoch": 3.12, "learning_rate": 3.812e-05, "loss": 0.1358, "step": 2380 }, { "epoch": 3.12, "learning_rate": 3.8115000000000004e-05, "loss": 0.146, "step": 2381 }, { "epoch": 3.12, "learning_rate": 3.811e-05, "loss": 0.1281, "step": 2382 }, { "epoch": 3.12, "learning_rate": 3.8105e-05, "loss": 0.1626, "step": 2383 }, { "epoch": 3.12, "learning_rate": 3.8100000000000005e-05, "loss": 0.1384, "step": 2384 }, { "epoch": 3.12, "learning_rate": 3.8095e-05, "loss": 0.1351, "step": 2385 }, { "epoch": 3.12, "learning_rate": 3.809e-05, "loss": 0.1316, "step": 2386 }, { "epoch": 3.12, "learning_rate": 3.8085000000000006e-05, "loss": 0.1387, "step": 2387 }, { "epoch": 3.13, "learning_rate": 3.808e-05, "loss": 0.1109, "step": 2388 }, { "epoch": 3.13, "learning_rate": 3.8075e-05, "loss": 0.1527, "step": 2389 }, { "epoch": 3.13, "learning_rate": 3.807e-05, "loss": 0.1627, "step": 2390 }, { "epoch": 3.13, "learning_rate": 3.8065e-05, "loss": 0.1576, "step": 2391 }, { "epoch": 3.13, "learning_rate": 3.806e-05, "loss": 0.1465, "step": 2392 }, { "epoch": 3.13, "learning_rate": 3.8055e-05, "loss": 0.1494, "step": 2393 }, { "epoch": 3.13, "learning_rate": 3.805e-05, "loss": 0.1439, "step": 2394 }, { "epoch": 3.13, "learning_rate": 3.8045000000000006e-05, "loss": 0.1323, "step": 2395 }, { "epoch": 3.14, "learning_rate": 3.804e-05, "loss": 0.1658, "step": 2396 }, { "epoch": 3.14, "learning_rate": 3.8035000000000004e-05, "loss": 0.1742, "step": 2397 }, { "epoch": 3.14, "learning_rate": 3.803000000000001e-05, "loss": 0.1499, "step": 2398 }, { "epoch": 3.14, "learning_rate": 3.8025e-05, "loss": 0.1335, "step": 2399 }, { "epoch": 3.14, "learning_rate": 3.802e-05, "loss": 0.1364, "step": 2400 }, { "epoch": 3.14, "learning_rate": 3.8015e-05, "loss": 0.1485, "step": 2401 }, { "epoch": 3.14, "learning_rate": 3.8010000000000004e-05, "loss": 0.141, "step": 2402 }, { "epoch": 3.15, "learning_rate": 3.8005e-05, "loss": 0.1368, "step": 2403 }, { "epoch": 3.15, "learning_rate": 3.8e-05, "loss": 0.1388, "step": 2404 }, { "epoch": 3.15, "learning_rate": 3.7995000000000005e-05, "loss": 0.1455, "step": 2405 }, { "epoch": 3.15, "learning_rate": 3.799e-05, "loss": 0.1191, "step": 2406 }, { "epoch": 3.15, "learning_rate": 3.7985e-05, "loss": 0.1473, "step": 2407 }, { "epoch": 3.15, "learning_rate": 3.7980000000000006e-05, "loss": 0.1323, "step": 2408 }, { "epoch": 3.15, "learning_rate": 3.7975e-05, "loss": 0.1428, "step": 2409 }, { "epoch": 3.15, "learning_rate": 3.797e-05, "loss": 0.1342, "step": 2410 }, { "epoch": 3.16, "learning_rate": 3.7965e-05, "loss": 0.1613, "step": 2411 }, { "epoch": 3.16, "learning_rate": 3.796e-05, "loss": 0.1419, "step": 2412 }, { "epoch": 3.16, "learning_rate": 3.7955e-05, "loss": 0.1327, "step": 2413 }, { "epoch": 3.16, "learning_rate": 3.795e-05, "loss": 0.1249, "step": 2414 }, { "epoch": 3.16, "learning_rate": 3.7945000000000003e-05, "loss": 0.1483, "step": 2415 }, { "epoch": 3.16, "learning_rate": 3.7940000000000006e-05, "loss": 0.1301, "step": 2416 }, { "epoch": 3.16, "learning_rate": 3.7935e-05, "loss": 0.1348, "step": 2417 }, { "epoch": 3.16, "learning_rate": 3.7930000000000004e-05, "loss": 0.1317, "step": 2418 }, { "epoch": 3.17, "learning_rate": 3.7925e-05, "loss": 0.1568, "step": 2419 }, { "epoch": 3.17, "learning_rate": 3.792e-05, "loss": 0.1527, "step": 2420 }, { "epoch": 3.17, "learning_rate": 3.7915e-05, "loss": 0.1487, "step": 2421 }, { "epoch": 3.17, "learning_rate": 3.791e-05, "loss": 0.1077, "step": 2422 }, { "epoch": 3.17, "learning_rate": 3.7905000000000004e-05, "loss": 0.1348, "step": 2423 }, { "epoch": 3.17, "learning_rate": 3.79e-05, "loss": 0.1489, "step": 2424 }, { "epoch": 3.17, "learning_rate": 3.7895e-05, "loss": 0.1819, "step": 2425 }, { "epoch": 3.18, "learning_rate": 3.7890000000000005e-05, "loss": 0.1449, "step": 2426 }, { "epoch": 3.18, "learning_rate": 3.7885e-05, "loss": 0.134, "step": 2427 }, { "epoch": 3.18, "learning_rate": 3.788e-05, "loss": 0.1581, "step": 2428 }, { "epoch": 3.18, "learning_rate": 3.7875e-05, "loss": 0.1493, "step": 2429 }, { "epoch": 3.18, "learning_rate": 3.787e-05, "loss": 0.1378, "step": 2430 }, { "epoch": 3.18, "learning_rate": 3.7865e-05, "loss": 0.1387, "step": 2431 }, { "epoch": 3.18, "learning_rate": 3.786e-05, "loss": 0.1591, "step": 2432 }, { "epoch": 3.18, "learning_rate": 3.7855e-05, "loss": 0.1638, "step": 2433 }, { "epoch": 3.19, "learning_rate": 3.7850000000000005e-05, "loss": 0.1333, "step": 2434 }, { "epoch": 3.19, "learning_rate": 3.7845e-05, "loss": 0.1273, "step": 2435 }, { "epoch": 3.19, "learning_rate": 3.7840000000000004e-05, "loss": 0.16, "step": 2436 }, { "epoch": 3.19, "learning_rate": 3.7835000000000006e-05, "loss": 0.1505, "step": 2437 }, { "epoch": 3.19, "learning_rate": 3.783e-05, "loss": 0.1252, "step": 2438 }, { "epoch": 3.19, "learning_rate": 3.7825e-05, "loss": 0.1233, "step": 2439 }, { "epoch": 3.19, "learning_rate": 3.782e-05, "loss": 0.1302, "step": 2440 }, { "epoch": 3.2, "learning_rate": 3.7815e-05, "loss": 0.1622, "step": 2441 }, { "epoch": 3.2, "learning_rate": 3.781e-05, "loss": 0.1274, "step": 2442 }, { "epoch": 3.2, "learning_rate": 3.7805e-05, "loss": 0.1397, "step": 2443 }, { "epoch": 3.2, "learning_rate": 3.7800000000000004e-05, "loss": 0.1677, "step": 2444 }, { "epoch": 3.2, "learning_rate": 3.7795e-05, "loss": 0.0998, "step": 2445 }, { "epoch": 3.2, "learning_rate": 3.779e-05, "loss": 0.163, "step": 2446 }, { "epoch": 3.2, "learning_rate": 3.7785000000000005e-05, "loss": 0.1507, "step": 2447 }, { "epoch": 3.2, "learning_rate": 3.778000000000001e-05, "loss": 0.1299, "step": 2448 }, { "epoch": 3.21, "learning_rate": 3.7775e-05, "loss": 0.1445, "step": 2449 }, { "epoch": 3.21, "learning_rate": 3.777e-05, "loss": 0.1395, "step": 2450 }, { "epoch": 3.21, "learning_rate": 3.7765e-05, "loss": 0.1444, "step": 2451 }, { "epoch": 3.21, "learning_rate": 3.776e-05, "loss": 0.1548, "step": 2452 }, { "epoch": 3.21, "learning_rate": 3.7755e-05, "loss": 0.1301, "step": 2453 }, { "epoch": 3.21, "learning_rate": 3.775e-05, "loss": 0.1324, "step": 2454 }, { "epoch": 3.21, "learning_rate": 3.7745000000000005e-05, "loss": 0.1202, "step": 2455 }, { "epoch": 3.21, "learning_rate": 3.774e-05, "loss": 0.1044, "step": 2456 }, { "epoch": 3.22, "learning_rate": 3.7735000000000004e-05, "loss": 0.1556, "step": 2457 }, { "epoch": 3.22, "learning_rate": 3.7730000000000006e-05, "loss": 0.1416, "step": 2458 }, { "epoch": 3.22, "learning_rate": 3.7725e-05, "loss": 0.1887, "step": 2459 }, { "epoch": 3.22, "learning_rate": 3.772e-05, "loss": 0.1401, "step": 2460 }, { "epoch": 3.22, "learning_rate": 3.7715e-05, "loss": 0.137, "step": 2461 }, { "epoch": 3.22, "learning_rate": 3.771e-05, "loss": 0.1415, "step": 2462 }, { "epoch": 3.22, "learning_rate": 3.7705e-05, "loss": 0.1645, "step": 2463 }, { "epoch": 3.23, "learning_rate": 3.77e-05, "loss": 0.143, "step": 2464 }, { "epoch": 3.23, "learning_rate": 3.7695000000000004e-05, "loss": 0.1216, "step": 2465 }, { "epoch": 3.23, "learning_rate": 3.769e-05, "loss": 0.1648, "step": 2466 }, { "epoch": 3.23, "learning_rate": 3.7685e-05, "loss": 0.1555, "step": 2467 }, { "epoch": 3.23, "learning_rate": 3.7680000000000005e-05, "loss": 0.14, "step": 2468 }, { "epoch": 3.23, "learning_rate": 3.7675e-05, "loss": 0.14, "step": 2469 }, { "epoch": 3.23, "learning_rate": 3.767e-05, "loss": 0.1656, "step": 2470 }, { "epoch": 3.23, "learning_rate": 3.7665e-05, "loss": 0.1578, "step": 2471 }, { "epoch": 3.24, "learning_rate": 3.766e-05, "loss": 0.1588, "step": 2472 }, { "epoch": 3.24, "learning_rate": 3.7655000000000005e-05, "loss": 0.1284, "step": 2473 }, { "epoch": 3.24, "learning_rate": 3.765e-05, "loss": 0.1687, "step": 2474 }, { "epoch": 3.24, "learning_rate": 3.7645e-05, "loss": 0.1596, "step": 2475 }, { "epoch": 3.24, "learning_rate": 3.7640000000000006e-05, "loss": 0.1386, "step": 2476 }, { "epoch": 3.24, "learning_rate": 3.7635e-05, "loss": 0.1361, "step": 2477 }, { "epoch": 3.24, "learning_rate": 3.7630000000000004e-05, "loss": 0.1318, "step": 2478 }, { "epoch": 3.24, "learning_rate": 3.7625e-05, "loss": 0.1496, "step": 2479 }, { "epoch": 3.25, "learning_rate": 3.762e-05, "loss": 0.1271, "step": 2480 }, { "epoch": 3.25, "learning_rate": 3.7615e-05, "loss": 0.1415, "step": 2481 }, { "epoch": 3.25, "learning_rate": 3.761e-05, "loss": 0.1324, "step": 2482 }, { "epoch": 3.25, "learning_rate": 3.7605e-05, "loss": 0.2164, "step": 2483 }, { "epoch": 3.25, "learning_rate": 3.76e-05, "loss": 0.1723, "step": 2484 }, { "epoch": 3.25, "learning_rate": 3.7595e-05, "loss": 0.1245, "step": 2485 }, { "epoch": 3.25, "learning_rate": 3.7590000000000004e-05, "loss": 0.1492, "step": 2486 }, { "epoch": 3.26, "learning_rate": 3.758500000000001e-05, "loss": 0.1865, "step": 2487 }, { "epoch": 3.26, "learning_rate": 3.758e-05, "loss": 0.1473, "step": 2488 }, { "epoch": 3.26, "learning_rate": 3.7575e-05, "loss": 0.1462, "step": 2489 }, { "epoch": 3.26, "learning_rate": 3.757e-05, "loss": 0.117, "step": 2490 }, { "epoch": 3.26, "learning_rate": 3.7565e-05, "loss": 0.1509, "step": 2491 }, { "epoch": 3.26, "learning_rate": 3.756e-05, "loss": 0.142, "step": 2492 }, { "epoch": 3.26, "learning_rate": 3.7555e-05, "loss": 0.1387, "step": 2493 }, { "epoch": 3.26, "learning_rate": 3.7550000000000005e-05, "loss": 0.1447, "step": 2494 }, { "epoch": 3.27, "learning_rate": 3.7545e-05, "loss": 0.1412, "step": 2495 }, { "epoch": 3.27, "learning_rate": 3.754e-05, "loss": 0.1493, "step": 2496 }, { "epoch": 3.27, "learning_rate": 3.7535000000000006e-05, "loss": 0.1414, "step": 2497 }, { "epoch": 3.27, "learning_rate": 3.753e-05, "loss": 0.1492, "step": 2498 }, { "epoch": 3.27, "learning_rate": 3.7525e-05, "loss": 0.1126, "step": 2499 }, { "epoch": 3.27, "learning_rate": 3.752e-05, "loss": 0.1258, "step": 2500 }, { "epoch": 3.27, "learning_rate": 3.7515e-05, "loss": 0.1456, "step": 2501 }, { "epoch": 3.27, "learning_rate": 3.751e-05, "loss": 0.1337, "step": 2502 }, { "epoch": 3.28, "learning_rate": 3.7505e-05, "loss": 0.1582, "step": 2503 }, { "epoch": 3.28, "learning_rate": 3.7500000000000003e-05, "loss": 0.1425, "step": 2504 }, { "epoch": 3.28, "learning_rate": 3.7495e-05, "loss": 0.138, "step": 2505 }, { "epoch": 3.28, "learning_rate": 3.749e-05, "loss": 0.1112, "step": 2506 }, { "epoch": 3.28, "learning_rate": 3.7485000000000004e-05, "loss": 0.1776, "step": 2507 }, { "epoch": 3.28, "learning_rate": 3.748000000000001e-05, "loss": 0.1573, "step": 2508 }, { "epoch": 3.28, "learning_rate": 3.7475e-05, "loss": 0.1806, "step": 2509 }, { "epoch": 3.29, "learning_rate": 3.747e-05, "loss": 0.1243, "step": 2510 }, { "epoch": 3.29, "learning_rate": 3.7465e-05, "loss": 0.1912, "step": 2511 }, { "epoch": 3.29, "learning_rate": 3.7460000000000004e-05, "loss": 0.1115, "step": 2512 }, { "epoch": 3.29, "learning_rate": 3.7455e-05, "loss": 0.1435, "step": 2513 }, { "epoch": 3.29, "learning_rate": 3.745e-05, "loss": 0.1208, "step": 2514 }, { "epoch": 3.29, "learning_rate": 3.7445000000000005e-05, "loss": 0.1336, "step": 2515 }, { "epoch": 3.29, "learning_rate": 3.744e-05, "loss": 0.1469, "step": 2516 }, { "epoch": 3.29, "learning_rate": 3.7435e-05, "loss": 0.156, "step": 2517 }, { "epoch": 3.3, "learning_rate": 3.7430000000000006e-05, "loss": 0.1678, "step": 2518 }, { "epoch": 3.3, "learning_rate": 3.7425e-05, "loss": 0.1256, "step": 2519 }, { "epoch": 3.3, "learning_rate": 3.742e-05, "loss": 0.1492, "step": 2520 }, { "epoch": 3.3, "learning_rate": 3.7415e-05, "loss": 0.1379, "step": 2521 }, { "epoch": 3.3, "learning_rate": 3.741e-05, "loss": 0.1329, "step": 2522 }, { "epoch": 3.3, "learning_rate": 3.7405e-05, "loss": 0.1532, "step": 2523 }, { "epoch": 3.3, "learning_rate": 3.74e-05, "loss": 0.138, "step": 2524 }, { "epoch": 3.3, "learning_rate": 3.7395000000000004e-05, "loss": 0.1645, "step": 2525 }, { "epoch": 3.31, "learning_rate": 3.739e-05, "loss": 0.1615, "step": 2526 }, { "epoch": 3.31, "learning_rate": 3.7385e-05, "loss": 0.1488, "step": 2527 }, { "epoch": 3.31, "learning_rate": 3.7380000000000005e-05, "loss": 0.1497, "step": 2528 }, { "epoch": 3.31, "learning_rate": 3.737500000000001e-05, "loss": 0.1632, "step": 2529 }, { "epoch": 3.31, "learning_rate": 3.7369999999999996e-05, "loss": 0.1799, "step": 2530 }, { "epoch": 3.31, "learning_rate": 3.7365e-05, "loss": 0.1476, "step": 2531 }, { "epoch": 3.31, "learning_rate": 3.736e-05, "loss": 0.1399, "step": 2532 }, { "epoch": 3.32, "learning_rate": 3.7355000000000004e-05, "loss": 0.1645, "step": 2533 }, { "epoch": 3.32, "learning_rate": 3.735e-05, "loss": 0.1295, "step": 2534 }, { "epoch": 3.32, "learning_rate": 3.7345e-05, "loss": 0.1571, "step": 2535 }, { "epoch": 3.32, "learning_rate": 3.7340000000000005e-05, "loss": 0.1488, "step": 2536 }, { "epoch": 3.32, "learning_rate": 3.7335e-05, "loss": 0.144, "step": 2537 }, { "epoch": 3.32, "learning_rate": 3.7330000000000003e-05, "loss": 0.1636, "step": 2538 }, { "epoch": 3.32, "learning_rate": 3.7325000000000006e-05, "loss": 0.147, "step": 2539 }, { "epoch": 3.32, "learning_rate": 3.732e-05, "loss": 0.1452, "step": 2540 }, { "epoch": 3.33, "learning_rate": 3.7315e-05, "loss": 0.1564, "step": 2541 }, { "epoch": 3.33, "learning_rate": 3.731e-05, "loss": 0.15, "step": 2542 }, { "epoch": 3.33, "learning_rate": 3.7305e-05, "loss": 0.1463, "step": 2543 }, { "epoch": 3.33, "learning_rate": 3.73e-05, "loss": 0.1357, "step": 2544 }, { "epoch": 3.33, "learning_rate": 3.7295e-05, "loss": 0.1525, "step": 2545 }, { "epoch": 3.33, "learning_rate": 3.7290000000000004e-05, "loss": 0.1679, "step": 2546 }, { "epoch": 3.33, "learning_rate": 3.7285000000000006e-05, "loss": 0.1273, "step": 2547 }, { "epoch": 3.34, "learning_rate": 3.728e-05, "loss": 0.1364, "step": 2548 }, { "epoch": 3.34, "learning_rate": 3.7275000000000005e-05, "loss": 0.1465, "step": 2549 }, { "epoch": 3.34, "learning_rate": 3.727e-05, "loss": 0.1481, "step": 2550 }, { "epoch": 3.34, "learning_rate": 3.7265e-05, "loss": 0.1596, "step": 2551 }, { "epoch": 3.34, "learning_rate": 3.726e-05, "loss": 0.1375, "step": 2552 }, { "epoch": 3.34, "learning_rate": 3.7255e-05, "loss": 0.1525, "step": 2553 }, { "epoch": 3.34, "learning_rate": 3.7250000000000004e-05, "loss": 0.1748, "step": 2554 }, { "epoch": 3.34, "learning_rate": 3.7245e-05, "loss": 0.1645, "step": 2555 }, { "epoch": 3.35, "learning_rate": 3.724e-05, "loss": 0.1376, "step": 2556 }, { "epoch": 3.35, "learning_rate": 3.7235000000000005e-05, "loss": 0.2017, "step": 2557 }, { "epoch": 3.35, "learning_rate": 3.723e-05, "loss": 0.1169, "step": 2558 }, { "epoch": 3.35, "learning_rate": 3.7225000000000004e-05, "loss": 0.161, "step": 2559 }, { "epoch": 3.35, "learning_rate": 3.722e-05, "loss": 0.1593, "step": 2560 }, { "epoch": 3.35, "learning_rate": 3.7215e-05, "loss": 0.1598, "step": 2561 }, { "epoch": 3.35, "learning_rate": 3.721e-05, "loss": 0.1813, "step": 2562 }, { "epoch": 3.35, "learning_rate": 3.7205e-05, "loss": 0.1368, "step": 2563 }, { "epoch": 3.36, "learning_rate": 3.72e-05, "loss": 0.1134, "step": 2564 }, { "epoch": 3.36, "learning_rate": 3.7195e-05, "loss": 0.1619, "step": 2565 }, { "epoch": 3.36, "learning_rate": 3.719e-05, "loss": 0.1264, "step": 2566 }, { "epoch": 3.36, "learning_rate": 3.7185000000000004e-05, "loss": 0.136, "step": 2567 }, { "epoch": 3.36, "learning_rate": 3.7180000000000007e-05, "loss": 0.1278, "step": 2568 }, { "epoch": 3.36, "learning_rate": 3.7175e-05, "loss": 0.1562, "step": 2569 }, { "epoch": 3.36, "learning_rate": 3.717e-05, "loss": 0.1306, "step": 2570 }, { "epoch": 3.37, "learning_rate": 3.7165e-05, "loss": 0.1789, "step": 2571 }, { "epoch": 3.37, "learning_rate": 3.716e-05, "loss": 0.1246, "step": 2572 }, { "epoch": 3.37, "learning_rate": 3.7155e-05, "loss": 0.1732, "step": 2573 }, { "epoch": 3.37, "learning_rate": 3.715e-05, "loss": 0.1746, "step": 2574 }, { "epoch": 3.37, "learning_rate": 3.7145000000000004e-05, "loss": 0.1664, "step": 2575 }, { "epoch": 3.37, "learning_rate": 3.714e-05, "loss": 0.1502, "step": 2576 }, { "epoch": 3.37, "learning_rate": 3.7135e-05, "loss": 0.1803, "step": 2577 }, { "epoch": 3.37, "learning_rate": 3.7130000000000005e-05, "loss": 0.1489, "step": 2578 }, { "epoch": 3.38, "learning_rate": 3.7125e-05, "loss": 0.1435, "step": 2579 }, { "epoch": 3.38, "learning_rate": 3.712e-05, "loss": 0.1678, "step": 2580 }, { "epoch": 3.38, "learning_rate": 3.7115e-05, "loss": 0.1678, "step": 2581 }, { "epoch": 3.38, "learning_rate": 3.711e-05, "loss": 0.1668, "step": 2582 }, { "epoch": 3.38, "learning_rate": 3.7105e-05, "loss": 0.1381, "step": 2583 }, { "epoch": 3.38, "learning_rate": 3.71e-05, "loss": 0.1448, "step": 2584 }, { "epoch": 3.38, "learning_rate": 3.7095e-05, "loss": 0.1684, "step": 2585 }, { "epoch": 3.38, "learning_rate": 3.7090000000000006e-05, "loss": 0.1424, "step": 2586 }, { "epoch": 3.39, "learning_rate": 3.7085e-05, "loss": 0.1395, "step": 2587 }, { "epoch": 3.39, "learning_rate": 3.7080000000000004e-05, "loss": 0.1175, "step": 2588 }, { "epoch": 3.39, "learning_rate": 3.707500000000001e-05, "loss": 0.1622, "step": 2589 }, { "epoch": 3.39, "learning_rate": 3.707e-05, "loss": 0.125, "step": 2590 }, { "epoch": 3.39, "learning_rate": 3.7065e-05, "loss": 0.1387, "step": 2591 }, { "epoch": 3.39, "learning_rate": 3.706e-05, "loss": 0.1575, "step": 2592 }, { "epoch": 3.39, "learning_rate": 3.7055000000000004e-05, "loss": 0.1526, "step": 2593 }, { "epoch": 3.4, "learning_rate": 3.705e-05, "loss": 0.1623, "step": 2594 }, { "epoch": 3.4, "learning_rate": 3.7045e-05, "loss": 0.1756, "step": 2595 }, { "epoch": 3.4, "learning_rate": 3.7040000000000005e-05, "loss": 0.1571, "step": 2596 }, { "epoch": 3.4, "learning_rate": 3.7035e-05, "loss": 0.1504, "step": 2597 }, { "epoch": 3.4, "learning_rate": 3.703e-05, "loss": 0.1881, "step": 2598 }, { "epoch": 3.4, "learning_rate": 3.7025000000000005e-05, "loss": 0.1826, "step": 2599 }, { "epoch": 3.4, "learning_rate": 3.702e-05, "loss": 0.1749, "step": 2600 }, { "epoch": 3.4, "learning_rate": 3.7015e-05, "loss": 0.1832, "step": 2601 }, { "epoch": 3.41, "learning_rate": 3.701e-05, "loss": 0.1462, "step": 2602 }, { "epoch": 3.41, "learning_rate": 3.7005e-05, "loss": 0.1481, "step": 2603 }, { "epoch": 3.41, "learning_rate": 3.7e-05, "loss": 0.1265, "step": 2604 }, { "epoch": 3.41, "learning_rate": 3.6995e-05, "loss": 0.1819, "step": 2605 }, { "epoch": 3.41, "learning_rate": 3.699e-05, "loss": 0.1567, "step": 2606 }, { "epoch": 3.41, "learning_rate": 3.6985000000000006e-05, "loss": 0.1559, "step": 2607 }, { "epoch": 3.41, "learning_rate": 3.698e-05, "loss": 0.16, "step": 2608 }, { "epoch": 3.41, "learning_rate": 3.6975000000000004e-05, "loss": 0.1566, "step": 2609 }, { "epoch": 3.42, "learning_rate": 3.697e-05, "loss": 0.1733, "step": 2610 }, { "epoch": 3.42, "learning_rate": 3.6965e-05, "loss": 0.1434, "step": 2611 }, { "epoch": 3.42, "learning_rate": 3.696e-05, "loss": 0.1215, "step": 2612 }, { "epoch": 3.42, "learning_rate": 3.6955e-05, "loss": 0.1722, "step": 2613 }, { "epoch": 3.42, "learning_rate": 3.6950000000000004e-05, "loss": 0.1987, "step": 2614 }, { "epoch": 3.42, "learning_rate": 3.6945e-05, "loss": 0.1804, "step": 2615 }, { "epoch": 3.42, "learning_rate": 3.694e-05, "loss": 0.1468, "step": 2616 }, { "epoch": 3.43, "learning_rate": 3.6935000000000005e-05, "loss": 0.1565, "step": 2617 }, { "epoch": 3.43, "learning_rate": 3.693e-05, "loss": 0.1245, "step": 2618 }, { "epoch": 3.43, "learning_rate": 3.6925e-05, "loss": 0.1508, "step": 2619 }, { "epoch": 3.43, "learning_rate": 3.692e-05, "loss": 0.1286, "step": 2620 }, { "epoch": 3.43, "learning_rate": 3.6915e-05, "loss": 0.1744, "step": 2621 }, { "epoch": 3.43, "learning_rate": 3.691e-05, "loss": 0.1662, "step": 2622 }, { "epoch": 3.43, "learning_rate": 3.6905e-05, "loss": 0.1711, "step": 2623 }, { "epoch": 3.43, "learning_rate": 3.69e-05, "loss": 0.1519, "step": 2624 }, { "epoch": 3.44, "learning_rate": 3.6895000000000005e-05, "loss": 0.1229, "step": 2625 }, { "epoch": 3.44, "learning_rate": 3.689e-05, "loss": 0.1551, "step": 2626 }, { "epoch": 3.44, "learning_rate": 3.6885000000000003e-05, "loss": 0.1245, "step": 2627 }, { "epoch": 3.44, "learning_rate": 3.6880000000000006e-05, "loss": 0.1661, "step": 2628 }, { "epoch": 3.44, "learning_rate": 3.6875e-05, "loss": 0.137, "step": 2629 }, { "epoch": 3.44, "learning_rate": 3.6870000000000004e-05, "loss": 0.1414, "step": 2630 }, { "epoch": 3.44, "learning_rate": 3.6865e-05, "loss": 0.1596, "step": 2631 }, { "epoch": 3.45, "learning_rate": 3.686e-05, "loss": 0.1414, "step": 2632 }, { "epoch": 3.45, "learning_rate": 3.6855e-05, "loss": 0.1408, "step": 2633 }, { "epoch": 3.45, "learning_rate": 3.685e-05, "loss": 0.1744, "step": 2634 }, { "epoch": 3.45, "learning_rate": 3.6845000000000004e-05, "loss": 0.1367, "step": 2635 }, { "epoch": 3.45, "learning_rate": 3.684e-05, "loss": 0.1503, "step": 2636 }, { "epoch": 3.45, "learning_rate": 3.6835e-05, "loss": 0.2008, "step": 2637 }, { "epoch": 3.45, "learning_rate": 3.6830000000000005e-05, "loss": 0.1714, "step": 2638 }, { "epoch": 3.45, "learning_rate": 3.6825e-05, "loss": 0.1357, "step": 2639 }, { "epoch": 3.46, "learning_rate": 3.682e-05, "loss": 0.145, "step": 2640 }, { "epoch": 3.46, "learning_rate": 3.6815e-05, "loss": 0.1504, "step": 2641 }, { "epoch": 3.46, "learning_rate": 3.681e-05, "loss": 0.1455, "step": 2642 }, { "epoch": 3.46, "learning_rate": 3.6805e-05, "loss": 0.1922, "step": 2643 }, { "epoch": 3.46, "learning_rate": 3.68e-05, "loss": 0.1357, "step": 2644 }, { "epoch": 3.46, "learning_rate": 3.6795e-05, "loss": 0.1582, "step": 2645 }, { "epoch": 3.46, "learning_rate": 3.6790000000000005e-05, "loss": 0.1341, "step": 2646 }, { "epoch": 3.46, "learning_rate": 3.6785e-05, "loss": 0.1702, "step": 2647 }, { "epoch": 3.47, "learning_rate": 3.6780000000000004e-05, "loss": 0.1274, "step": 2648 }, { "epoch": 3.47, "learning_rate": 3.6775000000000006e-05, "loss": 0.147, "step": 2649 }, { "epoch": 3.47, "learning_rate": 3.677e-05, "loss": 0.1679, "step": 2650 }, { "epoch": 3.47, "learning_rate": 3.6765e-05, "loss": 0.1719, "step": 2651 }, { "epoch": 3.47, "learning_rate": 3.676e-05, "loss": 0.1922, "step": 2652 }, { "epoch": 3.47, "learning_rate": 3.6755e-05, "loss": 0.1356, "step": 2653 }, { "epoch": 3.47, "learning_rate": 3.675e-05, "loss": 0.1534, "step": 2654 }, { "epoch": 3.48, "learning_rate": 3.6745e-05, "loss": 0.1729, "step": 2655 }, { "epoch": 3.48, "learning_rate": 3.6740000000000004e-05, "loss": 0.1561, "step": 2656 }, { "epoch": 3.48, "learning_rate": 3.6735e-05, "loss": 0.1503, "step": 2657 }, { "epoch": 3.48, "learning_rate": 3.673e-05, "loss": 0.1587, "step": 2658 }, { "epoch": 3.48, "learning_rate": 3.6725000000000005e-05, "loss": 0.1376, "step": 2659 }, { "epoch": 3.48, "learning_rate": 3.672000000000001e-05, "loss": 0.113, "step": 2660 }, { "epoch": 3.48, "learning_rate": 3.6714999999999997e-05, "loss": 0.1244, "step": 2661 }, { "epoch": 3.48, "learning_rate": 3.671e-05, "loss": 0.1494, "step": 2662 }, { "epoch": 3.49, "learning_rate": 3.6705e-05, "loss": 0.1488, "step": 2663 }, { "epoch": 3.49, "learning_rate": 3.6700000000000004e-05, "loss": 0.1562, "step": 2664 }, { "epoch": 3.49, "learning_rate": 3.6695e-05, "loss": 0.155, "step": 2665 }, { "epoch": 3.49, "learning_rate": 3.669e-05, "loss": 0.1576, "step": 2666 }, { "epoch": 3.49, "learning_rate": 3.6685000000000005e-05, "loss": 0.1748, "step": 2667 }, { "epoch": 3.49, "learning_rate": 3.668e-05, "loss": 0.1692, "step": 2668 }, { "epoch": 3.49, "learning_rate": 3.6675000000000004e-05, "loss": 0.1512, "step": 2669 }, { "epoch": 3.49, "learning_rate": 3.6670000000000006e-05, "loss": 0.1341, "step": 2670 }, { "epoch": 3.5, "learning_rate": 3.6665e-05, "loss": 0.1294, "step": 2671 }, { "epoch": 3.5, "learning_rate": 3.666e-05, "loss": 0.1392, "step": 2672 }, { "epoch": 3.5, "learning_rate": 3.6655e-05, "loss": 0.1448, "step": 2673 }, { "epoch": 3.5, "learning_rate": 3.665e-05, "loss": 0.1517, "step": 2674 }, { "epoch": 3.5, "learning_rate": 3.6645e-05, "loss": 0.183, "step": 2675 }, { "epoch": 3.5, "learning_rate": 3.664e-05, "loss": 0.1574, "step": 2676 }, { "epoch": 3.5, "learning_rate": 3.6635000000000004e-05, "loss": 0.1415, "step": 2677 }, { "epoch": 3.51, "learning_rate": 3.663e-05, "loss": 0.1669, "step": 2678 }, { "epoch": 3.51, "learning_rate": 3.6625e-05, "loss": 0.1644, "step": 2679 }, { "epoch": 3.51, "learning_rate": 3.6620000000000005e-05, "loss": 0.194, "step": 2680 }, { "epoch": 3.51, "learning_rate": 3.6615e-05, "loss": 0.1651, "step": 2681 }, { "epoch": 3.51, "learning_rate": 3.661e-05, "loss": 0.1412, "step": 2682 }, { "epoch": 3.51, "learning_rate": 3.6605e-05, "loss": 0.1766, "step": 2683 }, { "epoch": 3.51, "learning_rate": 3.66e-05, "loss": 0.1516, "step": 2684 }, { "epoch": 3.51, "learning_rate": 3.6595000000000005e-05, "loss": 0.1508, "step": 2685 }, { "epoch": 3.52, "learning_rate": 3.659e-05, "loss": 0.1295, "step": 2686 }, { "epoch": 3.52, "learning_rate": 3.6585e-05, "loss": 0.1504, "step": 2687 }, { "epoch": 3.52, "learning_rate": 3.6580000000000006e-05, "loss": 0.1629, "step": 2688 }, { "epoch": 3.52, "learning_rate": 3.6575e-05, "loss": 0.1831, "step": 2689 }, { "epoch": 3.52, "learning_rate": 3.6570000000000004e-05, "loss": 0.1694, "step": 2690 }, { "epoch": 3.52, "learning_rate": 3.6565e-05, "loss": 0.1578, "step": 2691 }, { "epoch": 3.52, "learning_rate": 3.656e-05, "loss": 0.1351, "step": 2692 }, { "epoch": 3.52, "learning_rate": 3.6555e-05, "loss": 0.1378, "step": 2693 }, { "epoch": 3.53, "learning_rate": 3.655e-05, "loss": 0.1432, "step": 2694 }, { "epoch": 3.53, "learning_rate": 3.6545e-05, "loss": 0.1431, "step": 2695 }, { "epoch": 3.53, "learning_rate": 3.654e-05, "loss": 0.1689, "step": 2696 }, { "epoch": 3.53, "learning_rate": 3.6535e-05, "loss": 0.154, "step": 2697 }, { "epoch": 3.53, "learning_rate": 3.6530000000000004e-05, "loss": 0.1639, "step": 2698 }, { "epoch": 3.53, "learning_rate": 3.652500000000001e-05, "loss": 0.1652, "step": 2699 }, { "epoch": 3.53, "learning_rate": 3.652e-05, "loss": 0.1783, "step": 2700 }, { "epoch": 3.54, "learning_rate": 3.6515e-05, "loss": 0.1639, "step": 2701 }, { "epoch": 3.54, "learning_rate": 3.651e-05, "loss": 0.1711, "step": 2702 }, { "epoch": 3.54, "learning_rate": 3.6505e-05, "loss": 0.16, "step": 2703 }, { "epoch": 3.54, "learning_rate": 3.65e-05, "loss": 0.1556, "step": 2704 }, { "epoch": 3.54, "learning_rate": 3.6495e-05, "loss": 0.1434, "step": 2705 }, { "epoch": 3.54, "learning_rate": 3.6490000000000005e-05, "loss": 0.1437, "step": 2706 }, { "epoch": 3.54, "learning_rate": 3.6485e-05, "loss": 0.1764, "step": 2707 }, { "epoch": 3.54, "learning_rate": 3.648e-05, "loss": 0.1677, "step": 2708 }, { "epoch": 3.55, "learning_rate": 3.6475000000000006e-05, "loss": 0.1737, "step": 2709 }, { "epoch": 3.55, "learning_rate": 3.647e-05, "loss": 0.1666, "step": 2710 }, { "epoch": 3.55, "learning_rate": 3.6465e-05, "loss": 0.1523, "step": 2711 }, { "epoch": 3.55, "learning_rate": 3.646e-05, "loss": 0.1594, "step": 2712 }, { "epoch": 3.55, "learning_rate": 3.6455e-05, "loss": 0.1692, "step": 2713 }, { "epoch": 3.55, "learning_rate": 3.645e-05, "loss": 0.1672, "step": 2714 }, { "epoch": 3.55, "learning_rate": 3.6445e-05, "loss": 0.1548, "step": 2715 }, { "epoch": 3.55, "learning_rate": 3.6440000000000003e-05, "loss": 0.1773, "step": 2716 }, { "epoch": 3.56, "learning_rate": 3.6435e-05, "loss": 0.1423, "step": 2717 }, { "epoch": 3.56, "learning_rate": 3.643e-05, "loss": 0.1358, "step": 2718 }, { "epoch": 3.56, "learning_rate": 3.6425000000000004e-05, "loss": 0.1536, "step": 2719 }, { "epoch": 3.56, "learning_rate": 3.642000000000001e-05, "loss": 0.173, "step": 2720 }, { "epoch": 3.56, "learning_rate": 3.6414999999999996e-05, "loss": 0.1279, "step": 2721 }, { "epoch": 3.56, "learning_rate": 3.641e-05, "loss": 0.1263, "step": 2722 }, { "epoch": 3.56, "learning_rate": 3.6405e-05, "loss": 0.1531, "step": 2723 }, { "epoch": 3.57, "learning_rate": 3.6400000000000004e-05, "loss": 0.1518, "step": 2724 }, { "epoch": 3.57, "learning_rate": 3.6395e-05, "loss": 0.1705, "step": 2725 }, { "epoch": 3.57, "learning_rate": 3.639e-05, "loss": 0.1383, "step": 2726 }, { "epoch": 3.57, "learning_rate": 3.6385000000000005e-05, "loss": 0.1673, "step": 2727 }, { "epoch": 3.57, "learning_rate": 3.638e-05, "loss": 0.1588, "step": 2728 }, { "epoch": 3.57, "learning_rate": 3.6375e-05, "loss": 0.1721, "step": 2729 }, { "epoch": 3.57, "learning_rate": 3.6370000000000006e-05, "loss": 0.1505, "step": 2730 }, { "epoch": 3.57, "learning_rate": 3.6365e-05, "loss": 0.1809, "step": 2731 }, { "epoch": 3.58, "learning_rate": 3.636e-05, "loss": 0.1494, "step": 2732 }, { "epoch": 3.58, "learning_rate": 3.6355e-05, "loss": 0.1536, "step": 2733 }, { "epoch": 3.58, "learning_rate": 3.635e-05, "loss": 0.1632, "step": 2734 }, { "epoch": 3.58, "learning_rate": 3.6345e-05, "loss": 0.1899, "step": 2735 }, { "epoch": 3.58, "learning_rate": 3.634e-05, "loss": 0.1528, "step": 2736 }, { "epoch": 3.58, "learning_rate": 3.6335000000000004e-05, "loss": 0.1637, "step": 2737 }, { "epoch": 3.58, "learning_rate": 3.6330000000000006e-05, "loss": 0.1817, "step": 2738 }, { "epoch": 3.59, "learning_rate": 3.6325e-05, "loss": 0.1984, "step": 2739 }, { "epoch": 3.59, "learning_rate": 3.6320000000000005e-05, "loss": 0.1697, "step": 2740 }, { "epoch": 3.59, "learning_rate": 3.6315e-05, "loss": 0.1661, "step": 2741 }, { "epoch": 3.59, "learning_rate": 3.6309999999999996e-05, "loss": 0.151, "step": 2742 }, { "epoch": 3.59, "learning_rate": 3.6305e-05, "loss": 0.1525, "step": 2743 }, { "epoch": 3.59, "learning_rate": 3.63e-05, "loss": 0.1356, "step": 2744 }, { "epoch": 3.59, "learning_rate": 3.6295000000000004e-05, "loss": 0.1725, "step": 2745 }, { "epoch": 3.59, "learning_rate": 3.629e-05, "loss": 0.1541, "step": 2746 }, { "epoch": 3.6, "learning_rate": 3.6285e-05, "loss": 0.1395, "step": 2747 }, { "epoch": 3.6, "learning_rate": 3.6280000000000005e-05, "loss": 0.1696, "step": 2748 }, { "epoch": 3.6, "learning_rate": 3.6275e-05, "loss": 0.1731, "step": 2749 }, { "epoch": 3.6, "learning_rate": 3.6270000000000003e-05, "loss": 0.1759, "step": 2750 }, { "epoch": 3.6, "learning_rate": 3.6265e-05, "loss": 0.1426, "step": 2751 }, { "epoch": 3.6, "learning_rate": 3.626e-05, "loss": 0.1652, "step": 2752 }, { "epoch": 3.6, "learning_rate": 3.6255e-05, "loss": 0.1455, "step": 2753 }, { "epoch": 3.6, "learning_rate": 3.625e-05, "loss": 0.1495, "step": 2754 }, { "epoch": 3.61, "learning_rate": 3.6245e-05, "loss": 0.1568, "step": 2755 }, { "epoch": 3.61, "learning_rate": 3.624e-05, "loss": 0.1499, "step": 2756 }, { "epoch": 3.61, "learning_rate": 3.6235e-05, "loss": 0.1458, "step": 2757 }, { "epoch": 3.61, "learning_rate": 3.6230000000000004e-05, "loss": 0.182, "step": 2758 }, { "epoch": 3.61, "learning_rate": 3.6225000000000006e-05, "loss": 0.1533, "step": 2759 }, { "epoch": 3.61, "learning_rate": 3.622e-05, "loss": 0.1412, "step": 2760 }, { "epoch": 3.61, "learning_rate": 3.6215000000000005e-05, "loss": 0.1684, "step": 2761 }, { "epoch": 3.62, "learning_rate": 3.621e-05, "loss": 0.1298, "step": 2762 }, { "epoch": 3.62, "learning_rate": 3.6205e-05, "loss": 0.1909, "step": 2763 }, { "epoch": 3.62, "learning_rate": 3.62e-05, "loss": 0.1893, "step": 2764 }, { "epoch": 3.62, "learning_rate": 3.6195e-05, "loss": 0.155, "step": 2765 }, { "epoch": 3.62, "learning_rate": 3.6190000000000004e-05, "loss": 0.1579, "step": 2766 }, { "epoch": 3.62, "learning_rate": 3.6185e-05, "loss": 0.1725, "step": 2767 }, { "epoch": 3.62, "learning_rate": 3.618e-05, "loss": 0.1741, "step": 2768 }, { "epoch": 3.62, "learning_rate": 3.6175000000000005e-05, "loss": 0.1675, "step": 2769 }, { "epoch": 3.63, "learning_rate": 3.617e-05, "loss": 0.1573, "step": 2770 }, { "epoch": 3.63, "learning_rate": 3.6165000000000004e-05, "loss": 0.1391, "step": 2771 }, { "epoch": 3.63, "learning_rate": 3.616e-05, "loss": 0.1918, "step": 2772 }, { "epoch": 3.63, "learning_rate": 3.6155e-05, "loss": 0.1613, "step": 2773 }, { "epoch": 3.63, "learning_rate": 3.615e-05, "loss": 0.1531, "step": 2774 }, { "epoch": 3.63, "learning_rate": 3.6145e-05, "loss": 0.1809, "step": 2775 }, { "epoch": 3.63, "learning_rate": 3.614e-05, "loss": 0.1678, "step": 2776 }, { "epoch": 3.63, "learning_rate": 3.6135000000000006e-05, "loss": 0.1515, "step": 2777 }, { "epoch": 3.64, "learning_rate": 3.613e-05, "loss": 0.157, "step": 2778 }, { "epoch": 3.64, "learning_rate": 3.6125000000000004e-05, "loss": 0.1513, "step": 2779 }, { "epoch": 3.64, "learning_rate": 3.6120000000000007e-05, "loss": 0.2127, "step": 2780 }, { "epoch": 3.64, "learning_rate": 3.6115e-05, "loss": 0.1424, "step": 2781 }, { "epoch": 3.64, "learning_rate": 3.611e-05, "loss": 0.1602, "step": 2782 }, { "epoch": 3.64, "learning_rate": 3.6105e-05, "loss": 0.1779, "step": 2783 }, { "epoch": 3.64, "learning_rate": 3.61e-05, "loss": 0.1549, "step": 2784 }, { "epoch": 3.65, "learning_rate": 3.6095e-05, "loss": 0.1627, "step": 2785 }, { "epoch": 3.65, "learning_rate": 3.609e-05, "loss": 0.181, "step": 2786 }, { "epoch": 3.65, "learning_rate": 3.6085000000000004e-05, "loss": 0.1744, "step": 2787 }, { "epoch": 3.65, "learning_rate": 3.608e-05, "loss": 0.1652, "step": 2788 }, { "epoch": 3.65, "learning_rate": 3.6075e-05, "loss": 0.1758, "step": 2789 }, { "epoch": 3.65, "learning_rate": 3.6070000000000005e-05, "loss": 0.1647, "step": 2790 }, { "epoch": 3.65, "learning_rate": 3.6065e-05, "loss": 0.1297, "step": 2791 }, { "epoch": 3.65, "learning_rate": 3.606e-05, "loss": 0.1536, "step": 2792 }, { "epoch": 3.66, "learning_rate": 3.6055e-05, "loss": 0.1872, "step": 2793 }, { "epoch": 3.66, "learning_rate": 3.605e-05, "loss": 0.1623, "step": 2794 }, { "epoch": 3.66, "learning_rate": 3.6045e-05, "loss": 0.1667, "step": 2795 }, { "epoch": 3.66, "learning_rate": 3.604e-05, "loss": 0.1716, "step": 2796 }, { "epoch": 3.66, "learning_rate": 3.6035e-05, "loss": 0.1479, "step": 2797 }, { "epoch": 3.66, "learning_rate": 3.6030000000000006e-05, "loss": 0.1522, "step": 2798 }, { "epoch": 3.66, "learning_rate": 3.6025e-05, "loss": 0.2088, "step": 2799 }, { "epoch": 3.66, "learning_rate": 3.6020000000000004e-05, "loss": 0.1484, "step": 2800 }, { "epoch": 3.67, "learning_rate": 3.601500000000001e-05, "loss": 0.159, "step": 2801 }, { "epoch": 3.67, "learning_rate": 3.601e-05, "loss": 0.1157, "step": 2802 }, { "epoch": 3.67, "learning_rate": 3.6005e-05, "loss": 0.1801, "step": 2803 }, { "epoch": 3.67, "learning_rate": 3.6e-05, "loss": 0.1871, "step": 2804 }, { "epoch": 3.67, "learning_rate": 3.5995000000000004e-05, "loss": 0.1419, "step": 2805 }, { "epoch": 3.67, "learning_rate": 3.599e-05, "loss": 0.1341, "step": 2806 }, { "epoch": 3.67, "learning_rate": 3.5985e-05, "loss": 0.1656, "step": 2807 }, { "epoch": 3.68, "learning_rate": 3.5980000000000004e-05, "loss": 0.1722, "step": 2808 }, { "epoch": 3.68, "learning_rate": 3.5975e-05, "loss": 0.1548, "step": 2809 }, { "epoch": 3.68, "learning_rate": 3.597e-05, "loss": 0.1638, "step": 2810 }, { "epoch": 3.68, "learning_rate": 3.5965000000000005e-05, "loss": 0.1704, "step": 2811 }, { "epoch": 3.68, "learning_rate": 3.596e-05, "loss": 0.1538, "step": 2812 }, { "epoch": 3.68, "learning_rate": 3.5955e-05, "loss": 0.1431, "step": 2813 }, { "epoch": 3.68, "learning_rate": 3.595e-05, "loss": 0.1435, "step": 2814 }, { "epoch": 3.68, "learning_rate": 3.5945e-05, "loss": 0.1781, "step": 2815 }, { "epoch": 3.69, "learning_rate": 3.594e-05, "loss": 0.1915, "step": 2816 }, { "epoch": 3.69, "learning_rate": 3.5935e-05, "loss": 0.1633, "step": 2817 }, { "epoch": 3.69, "learning_rate": 3.593e-05, "loss": 0.2085, "step": 2818 }, { "epoch": 3.69, "learning_rate": 3.5925000000000006e-05, "loss": 0.1368, "step": 2819 }, { "epoch": 3.69, "learning_rate": 3.592e-05, "loss": 0.1281, "step": 2820 }, { "epoch": 3.69, "learning_rate": 3.5915000000000004e-05, "loss": 0.1705, "step": 2821 }, { "epoch": 3.69, "learning_rate": 3.591e-05, "loss": 0.1673, "step": 2822 }, { "epoch": 3.7, "learning_rate": 3.5905e-05, "loss": 0.1735, "step": 2823 }, { "epoch": 3.7, "learning_rate": 3.59e-05, "loss": 0.1456, "step": 2824 }, { "epoch": 3.7, "learning_rate": 3.5895e-05, "loss": 0.1482, "step": 2825 }, { "epoch": 3.7, "learning_rate": 3.5890000000000004e-05, "loss": 0.1875, "step": 2826 }, { "epoch": 3.7, "learning_rate": 3.5885e-05, "loss": 0.1894, "step": 2827 }, { "epoch": 3.7, "learning_rate": 3.588e-05, "loss": 0.1788, "step": 2828 }, { "epoch": 3.7, "learning_rate": 3.5875000000000005e-05, "loss": 0.1648, "step": 2829 }, { "epoch": 3.7, "learning_rate": 3.587e-05, "loss": 0.1495, "step": 2830 }, { "epoch": 3.71, "learning_rate": 3.5865e-05, "loss": 0.1619, "step": 2831 }, { "epoch": 3.71, "learning_rate": 3.586e-05, "loss": 0.1317, "step": 2832 }, { "epoch": 3.71, "learning_rate": 3.5855e-05, "loss": 0.1331, "step": 2833 }, { "epoch": 3.71, "learning_rate": 3.585e-05, "loss": 0.1572, "step": 2834 }, { "epoch": 3.71, "learning_rate": 3.5845e-05, "loss": 0.1805, "step": 2835 }, { "epoch": 3.71, "learning_rate": 3.584e-05, "loss": 0.139, "step": 2836 }, { "epoch": 3.71, "learning_rate": 3.5835000000000005e-05, "loss": 0.2029, "step": 2837 }, { "epoch": 3.71, "learning_rate": 3.583e-05, "loss": 0.1246, "step": 2838 }, { "epoch": 3.72, "learning_rate": 3.5825000000000003e-05, "loss": 0.1788, "step": 2839 }, { "epoch": 3.72, "learning_rate": 3.5820000000000006e-05, "loss": 0.1509, "step": 2840 }, { "epoch": 3.72, "learning_rate": 3.5815e-05, "loss": 0.1648, "step": 2841 }, { "epoch": 3.72, "learning_rate": 3.581e-05, "loss": 0.1717, "step": 2842 }, { "epoch": 3.72, "learning_rate": 3.5805e-05, "loss": 0.1819, "step": 2843 }, { "epoch": 3.72, "learning_rate": 3.58e-05, "loss": 0.1464, "step": 2844 }, { "epoch": 3.72, "learning_rate": 3.5795e-05, "loss": 0.1706, "step": 2845 }, { "epoch": 3.73, "learning_rate": 3.579e-05, "loss": 0.1839, "step": 2846 }, { "epoch": 3.73, "learning_rate": 3.5785000000000004e-05, "loss": 0.1731, "step": 2847 }, { "epoch": 3.73, "learning_rate": 3.578e-05, "loss": 0.1665, "step": 2848 }, { "epoch": 3.73, "learning_rate": 3.5775e-05, "loss": 0.1571, "step": 2849 }, { "epoch": 3.73, "learning_rate": 3.5770000000000005e-05, "loss": 0.1501, "step": 2850 }, { "epoch": 3.73, "learning_rate": 3.576500000000001e-05, "loss": 0.1519, "step": 2851 }, { "epoch": 3.73, "learning_rate": 3.5759999999999996e-05, "loss": 0.1304, "step": 2852 }, { "epoch": 3.73, "learning_rate": 3.5755e-05, "loss": 0.1578, "step": 2853 }, { "epoch": 3.74, "learning_rate": 3.575e-05, "loss": 0.1598, "step": 2854 }, { "epoch": 3.74, "learning_rate": 3.5745e-05, "loss": 0.1903, "step": 2855 }, { "epoch": 3.74, "learning_rate": 3.574e-05, "loss": 0.1639, "step": 2856 }, { "epoch": 3.74, "learning_rate": 3.5735e-05, "loss": 0.1678, "step": 2857 }, { "epoch": 3.74, "learning_rate": 3.5730000000000005e-05, "loss": 0.17, "step": 2858 }, { "epoch": 3.74, "learning_rate": 3.5725e-05, "loss": 0.1642, "step": 2859 }, { "epoch": 3.74, "learning_rate": 3.5720000000000004e-05, "loss": 0.1587, "step": 2860 }, { "epoch": 3.74, "learning_rate": 3.5715000000000006e-05, "loss": 0.1618, "step": 2861 }, { "epoch": 3.75, "learning_rate": 3.571e-05, "loss": 0.1471, "step": 2862 }, { "epoch": 3.75, "learning_rate": 3.5705e-05, "loss": 0.1666, "step": 2863 }, { "epoch": 3.75, "learning_rate": 3.57e-05, "loss": 0.1861, "step": 2864 }, { "epoch": 3.75, "learning_rate": 3.5695e-05, "loss": 0.1436, "step": 2865 }, { "epoch": 3.75, "learning_rate": 3.569e-05, "loss": 0.1442, "step": 2866 }, { "epoch": 3.75, "learning_rate": 3.5685e-05, "loss": 0.1702, "step": 2867 }, { "epoch": 3.75, "learning_rate": 3.5680000000000004e-05, "loss": 0.1446, "step": 2868 }, { "epoch": 3.76, "learning_rate": 3.5675e-05, "loss": 0.117, "step": 2869 }, { "epoch": 3.76, "learning_rate": 3.567e-05, "loss": 0.1779, "step": 2870 }, { "epoch": 3.76, "learning_rate": 3.5665000000000005e-05, "loss": 0.1636, "step": 2871 }, { "epoch": 3.76, "learning_rate": 3.566e-05, "loss": 0.1988, "step": 2872 }, { "epoch": 3.76, "learning_rate": 3.5654999999999997e-05, "loss": 0.1624, "step": 2873 }, { "epoch": 3.76, "learning_rate": 3.565e-05, "loss": 0.1691, "step": 2874 }, { "epoch": 3.76, "learning_rate": 3.5645e-05, "loss": 0.1532, "step": 2875 }, { "epoch": 3.76, "learning_rate": 3.5640000000000004e-05, "loss": 0.1125, "step": 2876 }, { "epoch": 3.77, "learning_rate": 3.5635e-05, "loss": 0.1912, "step": 2877 }, { "epoch": 3.77, "learning_rate": 3.563e-05, "loss": 0.1909, "step": 2878 }, { "epoch": 3.77, "learning_rate": 3.5625000000000005e-05, "loss": 0.1674, "step": 2879 }, { "epoch": 3.77, "learning_rate": 3.562e-05, "loss": 0.166, "step": 2880 }, { "epoch": 3.77, "learning_rate": 3.5615000000000004e-05, "loss": 0.1524, "step": 2881 }, { "epoch": 3.77, "learning_rate": 3.5610000000000006e-05, "loss": 0.1856, "step": 2882 }, { "epoch": 3.77, "learning_rate": 3.5605e-05, "loss": 0.1528, "step": 2883 }, { "epoch": 3.77, "learning_rate": 3.56e-05, "loss": 0.1532, "step": 2884 }, { "epoch": 3.78, "learning_rate": 3.5595e-05, "loss": 0.2325, "step": 2885 }, { "epoch": 3.78, "learning_rate": 3.559e-05, "loss": 0.1946, "step": 2886 }, { "epoch": 3.78, "learning_rate": 3.5585e-05, "loss": 0.1651, "step": 2887 }, { "epoch": 3.78, "learning_rate": 3.558e-05, "loss": 0.1491, "step": 2888 }, { "epoch": 3.78, "learning_rate": 3.5575000000000004e-05, "loss": 0.176, "step": 2889 }, { "epoch": 3.78, "learning_rate": 3.557e-05, "loss": 0.1637, "step": 2890 }, { "epoch": 3.78, "learning_rate": 3.5565e-05, "loss": 0.1482, "step": 2891 }, { "epoch": 3.79, "learning_rate": 3.5560000000000005e-05, "loss": 0.1764, "step": 2892 }, { "epoch": 3.79, "learning_rate": 3.5555e-05, "loss": 0.1737, "step": 2893 }, { "epoch": 3.79, "learning_rate": 3.555e-05, "loss": 0.1646, "step": 2894 }, { "epoch": 3.79, "learning_rate": 3.5545e-05, "loss": 0.1527, "step": 2895 }, { "epoch": 3.79, "learning_rate": 3.554e-05, "loss": 0.1851, "step": 2896 }, { "epoch": 3.79, "learning_rate": 3.5535000000000005e-05, "loss": 0.1675, "step": 2897 }, { "epoch": 3.79, "learning_rate": 3.553e-05, "loss": 0.1487, "step": 2898 }, { "epoch": 3.79, "learning_rate": 3.5525e-05, "loss": 0.1454, "step": 2899 }, { "epoch": 3.8, "learning_rate": 3.5520000000000006e-05, "loss": 0.1683, "step": 2900 }, { "epoch": 3.8, "learning_rate": 3.5515e-05, "loss": 0.2025, "step": 2901 }, { "epoch": 3.8, "learning_rate": 3.5510000000000004e-05, "loss": 0.2055, "step": 2902 }, { "epoch": 3.8, "learning_rate": 3.5505e-05, "loss": 0.1901, "step": 2903 }, { "epoch": 3.8, "learning_rate": 3.55e-05, "loss": 0.1677, "step": 2904 }, { "epoch": 3.8, "learning_rate": 3.5495e-05, "loss": 0.1581, "step": 2905 }, { "epoch": 3.8, "learning_rate": 3.549e-05, "loss": 0.161, "step": 2906 }, { "epoch": 3.8, "learning_rate": 3.5485e-05, "loss": 0.1588, "step": 2907 }, { "epoch": 3.81, "learning_rate": 3.548e-05, "loss": 0.158, "step": 2908 }, { "epoch": 3.81, "learning_rate": 3.5475e-05, "loss": 0.1577, "step": 2909 }, { "epoch": 3.81, "learning_rate": 3.5470000000000004e-05, "loss": 0.1261, "step": 2910 }, { "epoch": 3.81, "learning_rate": 3.546500000000001e-05, "loss": 0.1702, "step": 2911 }, { "epoch": 3.81, "learning_rate": 3.546e-05, "loss": 0.1426, "step": 2912 }, { "epoch": 3.81, "learning_rate": 3.5455e-05, "loss": 0.1669, "step": 2913 }, { "epoch": 3.81, "learning_rate": 3.545e-05, "loss": 0.1628, "step": 2914 }, { "epoch": 3.82, "learning_rate": 3.5445000000000004e-05, "loss": 0.1518, "step": 2915 }, { "epoch": 3.82, "learning_rate": 3.544e-05, "loss": 0.1443, "step": 2916 }, { "epoch": 3.82, "learning_rate": 3.5435e-05, "loss": 0.2022, "step": 2917 }, { "epoch": 3.82, "learning_rate": 3.5430000000000005e-05, "loss": 0.1911, "step": 2918 }, { "epoch": 3.82, "learning_rate": 3.5425e-05, "loss": 0.1315, "step": 2919 }, { "epoch": 3.82, "learning_rate": 3.542e-05, "loss": 0.1824, "step": 2920 }, { "epoch": 3.82, "learning_rate": 3.5415000000000006e-05, "loss": 0.1709, "step": 2921 }, { "epoch": 3.82, "learning_rate": 3.541e-05, "loss": 0.1349, "step": 2922 }, { "epoch": 3.83, "learning_rate": 3.5405e-05, "loss": 0.1482, "step": 2923 }, { "epoch": 3.83, "learning_rate": 3.54e-05, "loss": 0.1922, "step": 2924 }, { "epoch": 3.83, "learning_rate": 3.5395e-05, "loss": 0.1759, "step": 2925 }, { "epoch": 3.83, "learning_rate": 3.539e-05, "loss": 0.1544, "step": 2926 }, { "epoch": 3.83, "learning_rate": 3.5385e-05, "loss": 0.1568, "step": 2927 }, { "epoch": 3.83, "learning_rate": 3.5380000000000003e-05, "loss": 0.1331, "step": 2928 }, { "epoch": 3.83, "learning_rate": 3.5375e-05, "loss": 0.1808, "step": 2929 }, { "epoch": 3.84, "learning_rate": 3.537e-05, "loss": 0.1718, "step": 2930 }, { "epoch": 3.84, "learning_rate": 3.5365000000000004e-05, "loss": 0.1264, "step": 2931 }, { "epoch": 3.84, "learning_rate": 3.536000000000001e-05, "loss": 0.1915, "step": 2932 }, { "epoch": 3.84, "learning_rate": 3.5354999999999996e-05, "loss": 0.2023, "step": 2933 }, { "epoch": 3.84, "learning_rate": 3.535e-05, "loss": 0.1623, "step": 2934 }, { "epoch": 3.84, "learning_rate": 3.5345e-05, "loss": 0.1637, "step": 2935 }, { "epoch": 3.84, "learning_rate": 3.5340000000000004e-05, "loss": 0.1789, "step": 2936 }, { "epoch": 3.84, "learning_rate": 3.5335e-05, "loss": 0.1769, "step": 2937 }, { "epoch": 3.85, "learning_rate": 3.533e-05, "loss": 0.172, "step": 2938 }, { "epoch": 3.85, "learning_rate": 3.5325000000000005e-05, "loss": 0.1736, "step": 2939 }, { "epoch": 3.85, "learning_rate": 3.532e-05, "loss": 0.2014, "step": 2940 }, { "epoch": 3.85, "learning_rate": 3.5315e-05, "loss": 0.1583, "step": 2941 }, { "epoch": 3.85, "learning_rate": 3.5310000000000006e-05, "loss": 0.1368, "step": 2942 }, { "epoch": 3.85, "learning_rate": 3.5305e-05, "loss": 0.1829, "step": 2943 }, { "epoch": 3.85, "learning_rate": 3.53e-05, "loss": 0.1694, "step": 2944 }, { "epoch": 3.85, "learning_rate": 3.5295e-05, "loss": 0.1611, "step": 2945 }, { "epoch": 3.86, "learning_rate": 3.529e-05, "loss": 0.1673, "step": 2946 }, { "epoch": 3.86, "learning_rate": 3.5285e-05, "loss": 0.1791, "step": 2947 }, { "epoch": 3.86, "learning_rate": 3.528e-05, "loss": 0.1727, "step": 2948 }, { "epoch": 3.86, "learning_rate": 3.5275000000000004e-05, "loss": 0.1568, "step": 2949 }, { "epoch": 3.86, "learning_rate": 3.5270000000000006e-05, "loss": 0.1721, "step": 2950 }, { "epoch": 3.86, "learning_rate": 3.5265e-05, "loss": 0.1319, "step": 2951 }, { "epoch": 3.86, "learning_rate": 3.5260000000000005e-05, "loss": 0.1673, "step": 2952 }, { "epoch": 3.87, "learning_rate": 3.5255e-05, "loss": 0.1355, "step": 2953 }, { "epoch": 3.87, "learning_rate": 3.525e-05, "loss": 0.1756, "step": 2954 }, { "epoch": 3.87, "learning_rate": 3.5245e-05, "loss": 0.1438, "step": 2955 }, { "epoch": 3.87, "learning_rate": 3.524e-05, "loss": 0.1358, "step": 2956 }, { "epoch": 3.87, "learning_rate": 3.5235000000000004e-05, "loss": 0.1811, "step": 2957 }, { "epoch": 3.87, "learning_rate": 3.523e-05, "loss": 0.1541, "step": 2958 }, { "epoch": 3.87, "learning_rate": 3.5225e-05, "loss": 0.19, "step": 2959 }, { "epoch": 3.87, "learning_rate": 3.5220000000000005e-05, "loss": 0.1535, "step": 2960 }, { "epoch": 3.88, "learning_rate": 3.5215e-05, "loss": 0.1808, "step": 2961 }, { "epoch": 3.88, "learning_rate": 3.5210000000000003e-05, "loss": 0.1665, "step": 2962 }, { "epoch": 3.88, "learning_rate": 3.5205e-05, "loss": 0.1518, "step": 2963 }, { "epoch": 3.88, "learning_rate": 3.52e-05, "loss": 0.1647, "step": 2964 }, { "epoch": 3.88, "learning_rate": 3.5195e-05, "loss": 0.1842, "step": 2965 }, { "epoch": 3.88, "learning_rate": 3.519e-05, "loss": 0.1377, "step": 2966 }, { "epoch": 3.88, "learning_rate": 3.5185e-05, "loss": 0.1456, "step": 2967 }, { "epoch": 3.88, "learning_rate": 3.518e-05, "loss": 0.1777, "step": 2968 }, { "epoch": 3.89, "learning_rate": 3.5175e-05, "loss": 0.1739, "step": 2969 }, { "epoch": 3.89, "learning_rate": 3.5170000000000004e-05, "loss": 0.1904, "step": 2970 }, { "epoch": 3.89, "learning_rate": 3.5165000000000006e-05, "loss": 0.1565, "step": 2971 }, { "epoch": 3.89, "learning_rate": 3.516e-05, "loss": 0.1705, "step": 2972 }, { "epoch": 3.89, "learning_rate": 3.5155e-05, "loss": 0.1392, "step": 2973 }, { "epoch": 3.89, "learning_rate": 3.515e-05, "loss": 0.1498, "step": 2974 }, { "epoch": 3.89, "learning_rate": 3.5145e-05, "loss": 0.1731, "step": 2975 }, { "epoch": 3.9, "learning_rate": 3.514e-05, "loss": 0.1664, "step": 2976 }, { "epoch": 3.9, "learning_rate": 3.5135e-05, "loss": 0.1604, "step": 2977 }, { "epoch": 3.9, "learning_rate": 3.5130000000000004e-05, "loss": 0.1615, "step": 2978 }, { "epoch": 3.9, "learning_rate": 3.5125e-05, "loss": 0.1501, "step": 2979 }, { "epoch": 3.9, "learning_rate": 3.512e-05, "loss": 0.1852, "step": 2980 }, { "epoch": 3.9, "learning_rate": 3.5115000000000005e-05, "loss": 0.1478, "step": 2981 }, { "epoch": 3.9, "learning_rate": 3.511e-05, "loss": 0.174, "step": 2982 }, { "epoch": 3.9, "learning_rate": 3.5105e-05, "loss": 0.1498, "step": 2983 }, { "epoch": 3.91, "learning_rate": 3.51e-05, "loss": 0.1664, "step": 2984 }, { "epoch": 3.91, "learning_rate": 3.5095e-05, "loss": 0.1923, "step": 2985 }, { "epoch": 3.91, "learning_rate": 3.509e-05, "loss": 0.1812, "step": 2986 }, { "epoch": 3.91, "learning_rate": 3.5085e-05, "loss": 0.1495, "step": 2987 }, { "epoch": 3.91, "learning_rate": 3.508e-05, "loss": 0.1531, "step": 2988 }, { "epoch": 3.91, "learning_rate": 3.5075000000000006e-05, "loss": 0.1884, "step": 2989 }, { "epoch": 3.91, "learning_rate": 3.507e-05, "loss": 0.1633, "step": 2990 }, { "epoch": 3.91, "learning_rate": 3.5065000000000004e-05, "loss": 0.161, "step": 2991 }, { "epoch": 3.92, "learning_rate": 3.5060000000000007e-05, "loss": 0.1333, "step": 2992 }, { "epoch": 3.92, "learning_rate": 3.5055e-05, "loss": 0.1845, "step": 2993 }, { "epoch": 3.92, "learning_rate": 3.505e-05, "loss": 0.1776, "step": 2994 }, { "epoch": 3.92, "learning_rate": 3.5045e-05, "loss": 0.1757, "step": 2995 }, { "epoch": 3.92, "learning_rate": 3.504e-05, "loss": 0.188, "step": 2996 }, { "epoch": 3.92, "learning_rate": 3.5035e-05, "loss": 0.1647, "step": 2997 }, { "epoch": 3.92, "learning_rate": 3.503e-05, "loss": 0.1804, "step": 2998 }, { "epoch": 3.93, "learning_rate": 3.5025000000000004e-05, "loss": 0.1851, "step": 2999 }, { "epoch": 3.93, "learning_rate": 3.502e-05, "loss": 0.1656, "step": 3000 }, { "epoch": 3.93, "learning_rate": 3.5015e-05, "loss": 0.1415, "step": 3001 }, { "epoch": 3.93, "learning_rate": 3.5010000000000005e-05, "loss": 0.1469, "step": 3002 }, { "epoch": 3.93, "learning_rate": 3.5005e-05, "loss": 0.1665, "step": 3003 }, { "epoch": 3.93, "learning_rate": 3.5e-05, "loss": 0.1273, "step": 3004 }, { "epoch": 3.93, "learning_rate": 3.4995e-05, "loss": 0.1593, "step": 3005 }, { "epoch": 3.93, "learning_rate": 3.499e-05, "loss": 0.1519, "step": 3006 }, { "epoch": 3.94, "learning_rate": 3.4985e-05, "loss": 0.1719, "step": 3007 }, { "epoch": 3.94, "learning_rate": 3.498e-05, "loss": 0.1685, "step": 3008 }, { "epoch": 3.94, "learning_rate": 3.4975e-05, "loss": 0.1522, "step": 3009 }, { "epoch": 3.94, "learning_rate": 3.4970000000000006e-05, "loss": 0.1634, "step": 3010 }, { "epoch": 3.94, "learning_rate": 3.4965e-05, "loss": 0.1465, "step": 3011 }, { "epoch": 3.94, "learning_rate": 3.4960000000000004e-05, "loss": 0.2149, "step": 3012 }, { "epoch": 3.94, "learning_rate": 3.495500000000001e-05, "loss": 0.1657, "step": 3013 }, { "epoch": 3.95, "learning_rate": 3.495e-05, "loss": 0.2161, "step": 3014 }, { "epoch": 3.95, "learning_rate": 3.4945e-05, "loss": 0.1357, "step": 3015 }, { "epoch": 3.95, "learning_rate": 3.494e-05, "loss": 0.1735, "step": 3016 }, { "epoch": 3.95, "learning_rate": 3.4935000000000003e-05, "loss": 0.1407, "step": 3017 }, { "epoch": 3.95, "learning_rate": 3.493e-05, "loss": 0.1693, "step": 3018 }, { "epoch": 3.95, "learning_rate": 3.4925e-05, "loss": 0.1494, "step": 3019 }, { "epoch": 3.95, "learning_rate": 3.4920000000000004e-05, "loss": 0.1528, "step": 3020 }, { "epoch": 3.95, "learning_rate": 3.4915e-05, "loss": 0.1664, "step": 3021 }, { "epoch": 3.96, "learning_rate": 3.491e-05, "loss": 0.1871, "step": 3022 }, { "epoch": 3.96, "learning_rate": 3.4905000000000005e-05, "loss": 0.1834, "step": 3023 }, { "epoch": 3.96, "learning_rate": 3.49e-05, "loss": 0.1962, "step": 3024 }, { "epoch": 3.96, "learning_rate": 3.4895e-05, "loss": 0.1791, "step": 3025 }, { "epoch": 3.96, "learning_rate": 3.489e-05, "loss": 0.1625, "step": 3026 }, { "epoch": 3.96, "learning_rate": 3.4885e-05, "loss": 0.1495, "step": 3027 }, { "epoch": 3.96, "learning_rate": 3.4880000000000005e-05, "loss": 0.1725, "step": 3028 }, { "epoch": 3.96, "learning_rate": 3.4875e-05, "loss": 0.1603, "step": 3029 }, { "epoch": 3.97, "learning_rate": 3.487e-05, "loss": 0.1696, "step": 3030 }, { "epoch": 3.97, "learning_rate": 3.4865000000000006e-05, "loss": 0.1568, "step": 3031 }, { "epoch": 3.97, "learning_rate": 3.486e-05, "loss": 0.1897, "step": 3032 }, { "epoch": 3.97, "learning_rate": 3.4855000000000004e-05, "loss": 0.1545, "step": 3033 }, { "epoch": 3.97, "learning_rate": 3.485e-05, "loss": 0.1372, "step": 3034 }, { "epoch": 3.97, "learning_rate": 3.4845e-05, "loss": 0.1899, "step": 3035 }, { "epoch": 3.97, "learning_rate": 3.484e-05, "loss": 0.1642, "step": 3036 }, { "epoch": 3.98, "learning_rate": 3.4835e-05, "loss": 0.1736, "step": 3037 }, { "epoch": 3.98, "learning_rate": 3.4830000000000004e-05, "loss": 0.1559, "step": 3038 }, { "epoch": 3.98, "learning_rate": 3.4825e-05, "loss": 0.1736, "step": 3039 }, { "epoch": 3.98, "learning_rate": 3.482e-05, "loss": 0.1598, "step": 3040 }, { "epoch": 3.98, "learning_rate": 3.4815000000000005e-05, "loss": 0.1881, "step": 3041 }, { "epoch": 3.98, "learning_rate": 3.481e-05, "loss": 0.1523, "step": 3042 }, { "epoch": 3.98, "learning_rate": 3.4805e-05, "loss": 0.1524, "step": 3043 }, { "epoch": 3.98, "learning_rate": 3.48e-05, "loss": 0.1946, "step": 3044 }, { "epoch": 3.99, "learning_rate": 3.4795e-05, "loss": 0.2023, "step": 3045 }, { "epoch": 3.99, "learning_rate": 3.479e-05, "loss": 0.1609, "step": 3046 }, { "epoch": 3.99, "learning_rate": 3.4785e-05, "loss": 0.1568, "step": 3047 }, { "epoch": 3.99, "learning_rate": 3.478e-05, "loss": 0.1505, "step": 3048 }, { "epoch": 3.99, "learning_rate": 3.4775000000000005e-05, "loss": 0.1895, "step": 3049 }, { "epoch": 3.99, "learning_rate": 3.477e-05, "loss": 0.1535, "step": 3050 }, { "epoch": 3.99, "learning_rate": 3.4765000000000003e-05, "loss": 0.1766, "step": 3051 }, { "epoch": 3.99, "learning_rate": 3.4760000000000006e-05, "loss": 0.185, "step": 3052 }, { "epoch": 4.0, "learning_rate": 3.4755e-05, "loss": 0.1922, "step": 3053 }, { "epoch": 4.0, "learning_rate": 3.475e-05, "loss": 0.1722, "step": 3054 }, { "epoch": 4.0, "learning_rate": 3.4745e-05, "loss": 0.1659, "step": 3055 }, { "epoch": 4.0, "learning_rate": 3.474e-05, "loss": 0.1645, "step": 3056 }, { "epoch": 4.0, "learning_rate": 3.4735e-05, "loss": 0.0775, "step": 3057 }, { "epoch": 4.0, "learning_rate": 3.473e-05, "loss": 0.0726, "step": 3058 }, { "epoch": 4.0, "learning_rate": 3.4725000000000004e-05, "loss": 0.0734, "step": 3059 }, { "epoch": 4.01, "learning_rate": 3.472e-05, "loss": 0.0812, "step": 3060 }, { "epoch": 4.01, "learning_rate": 3.4715e-05, "loss": 0.0678, "step": 3061 }, { "epoch": 4.01, "learning_rate": 3.4710000000000005e-05, "loss": 0.0583, "step": 3062 }, { "epoch": 4.01, "learning_rate": 3.470500000000001e-05, "loss": 0.0584, "step": 3063 }, { "epoch": 4.01, "learning_rate": 3.4699999999999996e-05, "loss": 0.0675, "step": 3064 }, { "epoch": 4.01, "learning_rate": 3.4695e-05, "loss": 0.0739, "step": 3065 }, { "epoch": 4.01, "learning_rate": 3.469e-05, "loss": 0.0741, "step": 3066 }, { "epoch": 4.01, "learning_rate": 3.4685000000000004e-05, "loss": 0.0841, "step": 3067 }, { "epoch": 4.02, "learning_rate": 3.468e-05, "loss": 0.0783, "step": 3068 }, { "epoch": 4.02, "learning_rate": 3.4675e-05, "loss": 0.0887, "step": 3069 }, { "epoch": 4.02, "learning_rate": 3.4670000000000005e-05, "loss": 0.0635, "step": 3070 }, { "epoch": 4.02, "learning_rate": 3.4665e-05, "loss": 0.0784, "step": 3071 }, { "epoch": 4.02, "learning_rate": 3.4660000000000004e-05, "loss": 0.074, "step": 3072 }, { "epoch": 4.02, "learning_rate": 3.4655000000000006e-05, "loss": 0.0593, "step": 3073 }, { "epoch": 4.02, "learning_rate": 3.465e-05, "loss": 0.0669, "step": 3074 }, { "epoch": 4.02, "learning_rate": 3.4645e-05, "loss": 0.0785, "step": 3075 }, { "epoch": 4.03, "learning_rate": 3.464e-05, "loss": 0.0786, "step": 3076 }, { "epoch": 4.03, "learning_rate": 3.4635e-05, "loss": 0.0785, "step": 3077 }, { "epoch": 4.03, "learning_rate": 3.463e-05, "loss": 0.0547, "step": 3078 }, { "epoch": 4.03, "learning_rate": 3.4625e-05, "loss": 0.0771, "step": 3079 }, { "epoch": 4.03, "learning_rate": 3.4620000000000004e-05, "loss": 0.0667, "step": 3080 }, { "epoch": 4.03, "learning_rate": 3.4615e-05, "loss": 0.0583, "step": 3081 }, { "epoch": 4.03, "learning_rate": 3.461e-05, "loss": 0.0688, "step": 3082 }, { "epoch": 4.04, "learning_rate": 3.4605000000000005e-05, "loss": 0.0645, "step": 3083 }, { "epoch": 4.04, "learning_rate": 3.46e-05, "loss": 0.0651, "step": 3084 }, { "epoch": 4.04, "learning_rate": 3.4594999999999997e-05, "loss": 0.0684, "step": 3085 }, { "epoch": 4.04, "learning_rate": 3.459e-05, "loss": 0.0663, "step": 3086 }, { "epoch": 4.04, "learning_rate": 3.4585e-05, "loss": 0.0488, "step": 3087 }, { "epoch": 4.04, "learning_rate": 3.4580000000000004e-05, "loss": 0.0643, "step": 3088 }, { "epoch": 4.04, "learning_rate": 3.4575e-05, "loss": 0.0636, "step": 3089 }, { "epoch": 4.04, "learning_rate": 3.457e-05, "loss": 0.0623, "step": 3090 }, { "epoch": 4.05, "learning_rate": 3.4565000000000005e-05, "loss": 0.0858, "step": 3091 }, { "epoch": 4.05, "learning_rate": 3.456e-05, "loss": 0.0679, "step": 3092 }, { "epoch": 4.05, "learning_rate": 3.4555000000000004e-05, "loss": 0.0628, "step": 3093 }, { "epoch": 4.05, "learning_rate": 3.455e-05, "loss": 0.0679, "step": 3094 }, { "epoch": 4.05, "learning_rate": 3.4545e-05, "loss": 0.0543, "step": 3095 }, { "epoch": 4.05, "learning_rate": 3.454e-05, "loss": 0.0667, "step": 3096 }, { "epoch": 4.05, "learning_rate": 3.4535e-05, "loss": 0.0521, "step": 3097 }, { "epoch": 4.05, "learning_rate": 3.453e-05, "loss": 0.0664, "step": 3098 }, { "epoch": 4.06, "learning_rate": 3.4525e-05, "loss": 0.0513, "step": 3099 }, { "epoch": 4.06, "learning_rate": 3.452e-05, "loss": 0.0706, "step": 3100 }, { "epoch": 4.06, "learning_rate": 3.4515000000000004e-05, "loss": 0.0524, "step": 3101 }, { "epoch": 4.06, "learning_rate": 3.451000000000001e-05, "loss": 0.0566, "step": 3102 }, { "epoch": 4.06, "learning_rate": 3.4505e-05, "loss": 0.0711, "step": 3103 }, { "epoch": 4.06, "learning_rate": 3.45e-05, "loss": 0.0733, "step": 3104 }, { "epoch": 4.06, "learning_rate": 3.4495e-05, "loss": 0.0663, "step": 3105 }, { "epoch": 4.07, "learning_rate": 3.449e-05, "loss": 0.0533, "step": 3106 }, { "epoch": 4.07, "learning_rate": 3.4485e-05, "loss": 0.0506, "step": 3107 }, { "epoch": 4.07, "learning_rate": 3.448e-05, "loss": 0.0709, "step": 3108 }, { "epoch": 4.07, "learning_rate": 3.4475000000000005e-05, "loss": 0.0726, "step": 3109 }, { "epoch": 4.07, "learning_rate": 3.447e-05, "loss": 0.0646, "step": 3110 }, { "epoch": 4.07, "learning_rate": 3.4465e-05, "loss": 0.0585, "step": 3111 }, { "epoch": 4.07, "learning_rate": 3.4460000000000005e-05, "loss": 0.0661, "step": 3112 }, { "epoch": 4.07, "learning_rate": 3.4455e-05, "loss": 0.0539, "step": 3113 }, { "epoch": 4.08, "learning_rate": 3.445e-05, "loss": 0.0573, "step": 3114 }, { "epoch": 4.08, "learning_rate": 3.4445e-05, "loss": 0.0767, "step": 3115 }, { "epoch": 4.08, "learning_rate": 3.444e-05, "loss": 0.0558, "step": 3116 }, { "epoch": 4.08, "learning_rate": 3.4435e-05, "loss": 0.0558, "step": 3117 }, { "epoch": 4.08, "learning_rate": 3.443e-05, "loss": 0.0652, "step": 3118 }, { "epoch": 4.08, "learning_rate": 3.4425e-05, "loss": 0.0586, "step": 3119 }, { "epoch": 4.08, "learning_rate": 3.442e-05, "loss": 0.0649, "step": 3120 }, { "epoch": 4.09, "learning_rate": 3.4415e-05, "loss": 0.0554, "step": 3121 }, { "epoch": 4.09, "learning_rate": 3.4410000000000004e-05, "loss": 0.0767, "step": 3122 }, { "epoch": 4.09, "learning_rate": 3.440500000000001e-05, "loss": 0.0704, "step": 3123 }, { "epoch": 4.09, "learning_rate": 3.4399999999999996e-05, "loss": 0.0773, "step": 3124 }, { "epoch": 4.09, "learning_rate": 3.4395e-05, "loss": 0.0845, "step": 3125 }, { "epoch": 4.09, "learning_rate": 3.439e-05, "loss": 0.0513, "step": 3126 }, { "epoch": 4.09, "learning_rate": 3.4385000000000004e-05, "loss": 0.0545, "step": 3127 }, { "epoch": 4.09, "learning_rate": 3.438e-05, "loss": 0.0615, "step": 3128 }, { "epoch": 4.1, "learning_rate": 3.4375e-05, "loss": 0.0584, "step": 3129 }, { "epoch": 4.1, "learning_rate": 3.4370000000000005e-05, "loss": 0.0565, "step": 3130 }, { "epoch": 4.1, "learning_rate": 3.4365e-05, "loss": 0.0634, "step": 3131 }, { "epoch": 4.1, "learning_rate": 3.436e-05, "loss": 0.0718, "step": 3132 }, { "epoch": 4.1, "learning_rate": 3.4355000000000006e-05, "loss": 0.062, "step": 3133 }, { "epoch": 4.1, "learning_rate": 3.435e-05, "loss": 0.0759, "step": 3134 }, { "epoch": 4.1, "learning_rate": 3.4345e-05, "loss": 0.0616, "step": 3135 }, { "epoch": 4.1, "learning_rate": 3.434e-05, "loss": 0.0587, "step": 3136 }, { "epoch": 4.11, "learning_rate": 3.4335e-05, "loss": 0.0639, "step": 3137 }, { "epoch": 4.11, "learning_rate": 3.433e-05, "loss": 0.0708, "step": 3138 }, { "epoch": 4.11, "learning_rate": 3.4325e-05, "loss": 0.0912, "step": 3139 }, { "epoch": 4.11, "learning_rate": 3.4320000000000003e-05, "loss": 0.0571, "step": 3140 }, { "epoch": 4.11, "learning_rate": 3.4315000000000006e-05, "loss": 0.0643, "step": 3141 }, { "epoch": 4.11, "learning_rate": 3.431e-05, "loss": 0.0564, "step": 3142 }, { "epoch": 4.11, "learning_rate": 3.4305000000000004e-05, "loss": 0.076, "step": 3143 }, { "epoch": 4.12, "learning_rate": 3.430000000000001e-05, "loss": 0.0652, "step": 3144 }, { "epoch": 4.12, "learning_rate": 3.4294999999999996e-05, "loss": 0.0692, "step": 3145 }, { "epoch": 4.12, "learning_rate": 3.429e-05, "loss": 0.0599, "step": 3146 }, { "epoch": 4.12, "learning_rate": 3.4285e-05, "loss": 0.0776, "step": 3147 }, { "epoch": 4.12, "learning_rate": 3.4280000000000004e-05, "loss": 0.0618, "step": 3148 }, { "epoch": 4.12, "learning_rate": 3.4275e-05, "loss": 0.0598, "step": 3149 }, { "epoch": 4.12, "learning_rate": 3.427e-05, "loss": 0.0606, "step": 3150 }, { "epoch": 4.12, "learning_rate": 3.4265000000000005e-05, "loss": 0.0627, "step": 3151 }, { "epoch": 4.13, "learning_rate": 3.426e-05, "loss": 0.0636, "step": 3152 }, { "epoch": 4.13, "learning_rate": 3.4255e-05, "loss": 0.0623, "step": 3153 }, { "epoch": 4.13, "learning_rate": 3.4250000000000006e-05, "loss": 0.0498, "step": 3154 }, { "epoch": 4.13, "learning_rate": 3.4245e-05, "loss": 0.0603, "step": 3155 }, { "epoch": 4.13, "learning_rate": 3.424e-05, "loss": 0.0851, "step": 3156 }, { "epoch": 4.13, "learning_rate": 3.4235e-05, "loss": 0.0575, "step": 3157 }, { "epoch": 4.13, "learning_rate": 3.423e-05, "loss": 0.0544, "step": 3158 }, { "epoch": 4.13, "learning_rate": 3.4225e-05, "loss": 0.0658, "step": 3159 }, { "epoch": 4.14, "learning_rate": 3.422e-05, "loss": 0.0551, "step": 3160 }, { "epoch": 4.14, "learning_rate": 3.4215000000000004e-05, "loss": 0.0628, "step": 3161 }, { "epoch": 4.14, "learning_rate": 3.4210000000000006e-05, "loss": 0.0805, "step": 3162 }, { "epoch": 4.14, "learning_rate": 3.4205e-05, "loss": 0.0698, "step": 3163 }, { "epoch": 4.14, "learning_rate": 3.4200000000000005e-05, "loss": 0.0388, "step": 3164 }, { "epoch": 4.14, "learning_rate": 3.4195e-05, "loss": 0.071, "step": 3165 }, { "epoch": 4.14, "learning_rate": 3.419e-05, "loss": 0.0588, "step": 3166 }, { "epoch": 4.15, "learning_rate": 3.4185e-05, "loss": 0.0578, "step": 3167 }, { "epoch": 4.15, "learning_rate": 3.418e-05, "loss": 0.0749, "step": 3168 }, { "epoch": 4.15, "learning_rate": 3.4175000000000004e-05, "loss": 0.0649, "step": 3169 }, { "epoch": 4.15, "learning_rate": 3.417e-05, "loss": 0.0759, "step": 3170 }, { "epoch": 4.15, "learning_rate": 3.4165e-05, "loss": 0.0627, "step": 3171 }, { "epoch": 4.15, "learning_rate": 3.4160000000000005e-05, "loss": 0.0572, "step": 3172 }, { "epoch": 4.15, "learning_rate": 3.4155e-05, "loss": 0.0743, "step": 3173 }, { "epoch": 4.15, "learning_rate": 3.415e-05, "loss": 0.0668, "step": 3174 }, { "epoch": 4.16, "learning_rate": 3.4145e-05, "loss": 0.0596, "step": 3175 }, { "epoch": 4.16, "learning_rate": 3.414e-05, "loss": 0.0774, "step": 3176 }, { "epoch": 4.16, "learning_rate": 3.4135e-05, "loss": 0.0683, "step": 3177 }, { "epoch": 4.16, "learning_rate": 3.413e-05, "loss": 0.0602, "step": 3178 }, { "epoch": 4.16, "learning_rate": 3.4125e-05, "loss": 0.0524, "step": 3179 }, { "epoch": 4.16, "learning_rate": 3.412e-05, "loss": 0.0667, "step": 3180 }, { "epoch": 4.16, "learning_rate": 3.4115e-05, "loss": 0.0539, "step": 3181 }, { "epoch": 4.16, "learning_rate": 3.4110000000000004e-05, "loss": 0.068, "step": 3182 }, { "epoch": 4.17, "learning_rate": 3.4105000000000006e-05, "loss": 0.0614, "step": 3183 }, { "epoch": 4.17, "learning_rate": 3.41e-05, "loss": 0.0737, "step": 3184 }, { "epoch": 4.17, "learning_rate": 3.4095e-05, "loss": 0.0668, "step": 3185 }, { "epoch": 4.17, "learning_rate": 3.409e-05, "loss": 0.0651, "step": 3186 }, { "epoch": 4.17, "learning_rate": 3.4085e-05, "loss": 0.0557, "step": 3187 }, { "epoch": 4.17, "learning_rate": 3.408e-05, "loss": 0.0641, "step": 3188 }, { "epoch": 4.17, "learning_rate": 3.4075e-05, "loss": 0.0429, "step": 3189 }, { "epoch": 4.18, "learning_rate": 3.4070000000000004e-05, "loss": 0.0534, "step": 3190 }, { "epoch": 4.18, "learning_rate": 3.4065e-05, "loss": 0.0726, "step": 3191 }, { "epoch": 4.18, "learning_rate": 3.406e-05, "loss": 0.0645, "step": 3192 }, { "epoch": 4.18, "learning_rate": 3.4055000000000005e-05, "loss": 0.0604, "step": 3193 }, { "epoch": 4.18, "learning_rate": 3.405e-05, "loss": 0.0744, "step": 3194 }, { "epoch": 4.18, "learning_rate": 3.4045e-05, "loss": 0.0628, "step": 3195 }, { "epoch": 4.18, "learning_rate": 3.404e-05, "loss": 0.0592, "step": 3196 }, { "epoch": 4.18, "learning_rate": 3.4035e-05, "loss": 0.0728, "step": 3197 }, { "epoch": 4.19, "learning_rate": 3.403e-05, "loss": 0.0555, "step": 3198 }, { "epoch": 4.19, "learning_rate": 3.4025e-05, "loss": 0.0645, "step": 3199 }, { "epoch": 4.19, "learning_rate": 3.402e-05, "loss": 0.0696, "step": 3200 }, { "epoch": 4.19, "learning_rate": 3.4015000000000006e-05, "loss": 0.0636, "step": 3201 }, { "epoch": 4.19, "learning_rate": 3.401e-05, "loss": 0.0602, "step": 3202 }, { "epoch": 4.19, "learning_rate": 3.4005000000000004e-05, "loss": 0.0634, "step": 3203 }, { "epoch": 4.19, "learning_rate": 3.4000000000000007e-05, "loss": 0.0777, "step": 3204 }, { "epoch": 4.2, "learning_rate": 3.3995e-05, "loss": 0.0619, "step": 3205 }, { "epoch": 4.2, "learning_rate": 3.399e-05, "loss": 0.0689, "step": 3206 }, { "epoch": 4.2, "learning_rate": 3.3985e-05, "loss": 0.0668, "step": 3207 }, { "epoch": 4.2, "learning_rate": 3.398e-05, "loss": 0.054, "step": 3208 }, { "epoch": 4.2, "learning_rate": 3.3975e-05, "loss": 0.0553, "step": 3209 }, { "epoch": 4.2, "learning_rate": 3.397e-05, "loss": 0.0771, "step": 3210 }, { "epoch": 4.2, "learning_rate": 3.3965000000000004e-05, "loss": 0.0689, "step": 3211 }, { "epoch": 4.2, "learning_rate": 3.396e-05, "loss": 0.0637, "step": 3212 }, { "epoch": 4.21, "learning_rate": 3.3955e-05, "loss": 0.075, "step": 3213 }, { "epoch": 4.21, "learning_rate": 3.3950000000000005e-05, "loss": 0.0644, "step": 3214 }, { "epoch": 4.21, "learning_rate": 3.3945e-05, "loss": 0.059, "step": 3215 }, { "epoch": 4.21, "learning_rate": 3.394e-05, "loss": 0.0621, "step": 3216 }, { "epoch": 4.21, "learning_rate": 3.3935e-05, "loss": 0.0646, "step": 3217 }, { "epoch": 4.21, "learning_rate": 3.393e-05, "loss": 0.0639, "step": 3218 }, { "epoch": 4.21, "learning_rate": 3.3925e-05, "loss": 0.0668, "step": 3219 }, { "epoch": 4.21, "learning_rate": 3.392e-05, "loss": 0.0652, "step": 3220 }, { "epoch": 4.22, "learning_rate": 3.3915e-05, "loss": 0.0707, "step": 3221 }, { "epoch": 4.22, "learning_rate": 3.3910000000000006e-05, "loss": 0.0563, "step": 3222 }, { "epoch": 4.22, "learning_rate": 3.3905e-05, "loss": 0.0704, "step": 3223 }, { "epoch": 4.22, "learning_rate": 3.3900000000000004e-05, "loss": 0.0543, "step": 3224 }, { "epoch": 4.22, "learning_rate": 3.3895e-05, "loss": 0.0689, "step": 3225 }, { "epoch": 4.22, "learning_rate": 3.389e-05, "loss": 0.059, "step": 3226 }, { "epoch": 4.22, "learning_rate": 3.3885e-05, "loss": 0.0578, "step": 3227 }, { "epoch": 4.23, "learning_rate": 3.388e-05, "loss": 0.0607, "step": 3228 }, { "epoch": 4.23, "learning_rate": 3.3875000000000003e-05, "loss": 0.056, "step": 3229 }, { "epoch": 4.23, "learning_rate": 3.387e-05, "loss": 0.0763, "step": 3230 }, { "epoch": 4.23, "learning_rate": 3.3865e-05, "loss": 0.0537, "step": 3231 }, { "epoch": 4.23, "learning_rate": 3.3860000000000004e-05, "loss": 0.058, "step": 3232 }, { "epoch": 4.23, "learning_rate": 3.3855e-05, "loss": 0.0592, "step": 3233 }, { "epoch": 4.23, "learning_rate": 3.385e-05, "loss": 0.0523, "step": 3234 }, { "epoch": 4.23, "learning_rate": 3.3845e-05, "loss": 0.0657, "step": 3235 }, { "epoch": 4.24, "learning_rate": 3.384e-05, "loss": 0.0533, "step": 3236 }, { "epoch": 4.24, "learning_rate": 3.3835e-05, "loss": 0.0709, "step": 3237 }, { "epoch": 4.24, "learning_rate": 3.383e-05, "loss": 0.051, "step": 3238 }, { "epoch": 4.24, "learning_rate": 3.3825e-05, "loss": 0.0668, "step": 3239 }, { "epoch": 4.24, "learning_rate": 3.3820000000000005e-05, "loss": 0.0657, "step": 3240 }, { "epoch": 4.24, "learning_rate": 3.3815e-05, "loss": 0.0509, "step": 3241 }, { "epoch": 4.24, "learning_rate": 3.381e-05, "loss": 0.0755, "step": 3242 }, { "epoch": 4.24, "learning_rate": 3.3805000000000006e-05, "loss": 0.053, "step": 3243 }, { "epoch": 4.25, "learning_rate": 3.38e-05, "loss": 0.0674, "step": 3244 }, { "epoch": 4.25, "learning_rate": 3.3795e-05, "loss": 0.0561, "step": 3245 }, { "epoch": 4.25, "learning_rate": 3.379e-05, "loss": 0.068, "step": 3246 }, { "epoch": 4.25, "learning_rate": 3.3785e-05, "loss": 0.0566, "step": 3247 }, { "epoch": 4.25, "learning_rate": 3.378e-05, "loss": 0.0598, "step": 3248 }, { "epoch": 4.25, "learning_rate": 3.3775e-05, "loss": 0.0648, "step": 3249 }, { "epoch": 4.25, "learning_rate": 3.3770000000000004e-05, "loss": 0.0666, "step": 3250 }, { "epoch": 4.26, "learning_rate": 3.3765e-05, "loss": 0.0639, "step": 3251 }, { "epoch": 4.26, "learning_rate": 3.376e-05, "loss": 0.0617, "step": 3252 }, { "epoch": 4.26, "learning_rate": 3.3755000000000005e-05, "loss": 0.0646, "step": 3253 }, { "epoch": 4.26, "learning_rate": 3.375000000000001e-05, "loss": 0.0649, "step": 3254 }, { "epoch": 4.26, "learning_rate": 3.3745e-05, "loss": 0.0617, "step": 3255 }, { "epoch": 4.26, "learning_rate": 3.374e-05, "loss": 0.0468, "step": 3256 }, { "epoch": 4.26, "learning_rate": 3.3735e-05, "loss": 0.0653, "step": 3257 }, { "epoch": 4.26, "learning_rate": 3.373e-05, "loss": 0.0642, "step": 3258 }, { "epoch": 4.27, "learning_rate": 3.3725e-05, "loss": 0.0569, "step": 3259 }, { "epoch": 4.27, "learning_rate": 3.372e-05, "loss": 0.0673, "step": 3260 }, { "epoch": 4.27, "learning_rate": 3.3715000000000005e-05, "loss": 0.0666, "step": 3261 }, { "epoch": 4.27, "learning_rate": 3.371e-05, "loss": 0.0547, "step": 3262 }, { "epoch": 4.27, "learning_rate": 3.3705000000000003e-05, "loss": 0.0782, "step": 3263 }, { "epoch": 4.27, "learning_rate": 3.3700000000000006e-05, "loss": 0.0637, "step": 3264 }, { "epoch": 4.27, "learning_rate": 3.3695e-05, "loss": 0.067, "step": 3265 }, { "epoch": 4.27, "learning_rate": 3.369e-05, "loss": 0.0668, "step": 3266 }, { "epoch": 4.28, "learning_rate": 3.3685e-05, "loss": 0.0585, "step": 3267 }, { "epoch": 4.28, "learning_rate": 3.368e-05, "loss": 0.075, "step": 3268 }, { "epoch": 4.28, "learning_rate": 3.3675e-05, "loss": 0.0538, "step": 3269 }, { "epoch": 4.28, "learning_rate": 3.367e-05, "loss": 0.0591, "step": 3270 }, { "epoch": 4.28, "learning_rate": 3.3665000000000004e-05, "loss": 0.0645, "step": 3271 }, { "epoch": 4.28, "learning_rate": 3.366e-05, "loss": 0.069, "step": 3272 }, { "epoch": 4.28, "learning_rate": 3.3655e-05, "loss": 0.0701, "step": 3273 }, { "epoch": 4.29, "learning_rate": 3.3650000000000005e-05, "loss": 0.0682, "step": 3274 }, { "epoch": 4.29, "learning_rate": 3.364500000000001e-05, "loss": 0.0669, "step": 3275 }, { "epoch": 4.29, "learning_rate": 3.3639999999999996e-05, "loss": 0.068, "step": 3276 }, { "epoch": 4.29, "learning_rate": 3.3635e-05, "loss": 0.0699, "step": 3277 }, { "epoch": 4.29, "learning_rate": 3.363e-05, "loss": 0.0557, "step": 3278 }, { "epoch": 4.29, "learning_rate": 3.3625000000000004e-05, "loss": 0.0508, "step": 3279 }, { "epoch": 4.29, "learning_rate": 3.362e-05, "loss": 0.0569, "step": 3280 }, { "epoch": 4.29, "learning_rate": 3.3615e-05, "loss": 0.0668, "step": 3281 }, { "epoch": 4.3, "learning_rate": 3.3610000000000005e-05, "loss": 0.0778, "step": 3282 }, { "epoch": 4.3, "learning_rate": 3.3605e-05, "loss": 0.0653, "step": 3283 }, { "epoch": 4.3, "learning_rate": 3.3600000000000004e-05, "loss": 0.0664, "step": 3284 }, { "epoch": 4.3, "learning_rate": 3.3595000000000006e-05, "loss": 0.059, "step": 3285 }, { "epoch": 4.3, "learning_rate": 3.359e-05, "loss": 0.0717, "step": 3286 }, { "epoch": 4.3, "learning_rate": 3.3585e-05, "loss": 0.0806, "step": 3287 }, { "epoch": 4.3, "learning_rate": 3.358e-05, "loss": 0.0548, "step": 3288 }, { "epoch": 4.3, "learning_rate": 3.3575e-05, "loss": 0.0671, "step": 3289 }, { "epoch": 4.31, "learning_rate": 3.357e-05, "loss": 0.0674, "step": 3290 }, { "epoch": 4.31, "learning_rate": 3.3565e-05, "loss": 0.0628, "step": 3291 }, { "epoch": 4.31, "learning_rate": 3.3560000000000004e-05, "loss": 0.0729, "step": 3292 }, { "epoch": 4.31, "learning_rate": 3.3555e-05, "loss": 0.0737, "step": 3293 }, { "epoch": 4.31, "learning_rate": 3.355e-05, "loss": 0.0492, "step": 3294 }, { "epoch": 4.31, "learning_rate": 3.3545000000000005e-05, "loss": 0.0589, "step": 3295 }, { "epoch": 4.31, "learning_rate": 3.354e-05, "loss": 0.0548, "step": 3296 }, { "epoch": 4.32, "learning_rate": 3.3534999999999997e-05, "loss": 0.0596, "step": 3297 }, { "epoch": 4.32, "learning_rate": 3.353e-05, "loss": 0.0609, "step": 3298 }, { "epoch": 4.32, "learning_rate": 3.3525e-05, "loss": 0.0706, "step": 3299 }, { "epoch": 4.32, "learning_rate": 3.3520000000000004e-05, "loss": 0.0653, "step": 3300 }, { "epoch": 4.32, "learning_rate": 3.3515e-05, "loss": 0.0699, "step": 3301 }, { "epoch": 4.32, "learning_rate": 3.351e-05, "loss": 0.0632, "step": 3302 }, { "epoch": 4.32, "learning_rate": 3.3505000000000005e-05, "loss": 0.0832, "step": 3303 }, { "epoch": 4.32, "learning_rate": 3.35e-05, "loss": 0.0864, "step": 3304 }, { "epoch": 4.33, "learning_rate": 3.3495000000000004e-05, "loss": 0.0545, "step": 3305 }, { "epoch": 4.33, "learning_rate": 3.349e-05, "loss": 0.0685, "step": 3306 }, { "epoch": 4.33, "learning_rate": 3.3485e-05, "loss": 0.055, "step": 3307 }, { "epoch": 4.33, "learning_rate": 3.348e-05, "loss": 0.0665, "step": 3308 }, { "epoch": 4.33, "learning_rate": 3.3475e-05, "loss": 0.0626, "step": 3309 }, { "epoch": 4.33, "learning_rate": 3.347e-05, "loss": 0.0612, "step": 3310 }, { "epoch": 4.33, "learning_rate": 3.3465e-05, "loss": 0.0624, "step": 3311 }, { "epoch": 4.34, "learning_rate": 3.346e-05, "loss": 0.0581, "step": 3312 }, { "epoch": 4.34, "learning_rate": 3.3455000000000004e-05, "loss": 0.0601, "step": 3313 }, { "epoch": 4.34, "learning_rate": 3.345000000000001e-05, "loss": 0.0721, "step": 3314 }, { "epoch": 4.34, "learning_rate": 3.3445e-05, "loss": 0.0678, "step": 3315 }, { "epoch": 4.34, "learning_rate": 3.344e-05, "loss": 0.0646, "step": 3316 }, { "epoch": 4.34, "learning_rate": 3.3435e-05, "loss": 0.0372, "step": 3317 }, { "epoch": 4.34, "learning_rate": 3.3430000000000003e-05, "loss": 0.0702, "step": 3318 }, { "epoch": 4.34, "learning_rate": 3.3425e-05, "loss": 0.055, "step": 3319 }, { "epoch": 4.35, "learning_rate": 3.342e-05, "loss": 0.0675, "step": 3320 }, { "epoch": 4.35, "learning_rate": 3.3415000000000004e-05, "loss": 0.0672, "step": 3321 }, { "epoch": 4.35, "learning_rate": 3.341e-05, "loss": 0.0841, "step": 3322 }, { "epoch": 4.35, "learning_rate": 3.3405e-05, "loss": 0.0591, "step": 3323 }, { "epoch": 4.35, "learning_rate": 3.3400000000000005e-05, "loss": 0.0862, "step": 3324 }, { "epoch": 4.35, "learning_rate": 3.3395e-05, "loss": 0.065, "step": 3325 }, { "epoch": 4.35, "learning_rate": 3.339e-05, "loss": 0.0694, "step": 3326 }, { "epoch": 4.35, "learning_rate": 3.3385e-05, "loss": 0.0624, "step": 3327 }, { "epoch": 4.36, "learning_rate": 3.338e-05, "loss": 0.0729, "step": 3328 }, { "epoch": 4.36, "learning_rate": 3.3375e-05, "loss": 0.0625, "step": 3329 }, { "epoch": 4.36, "learning_rate": 3.337e-05, "loss": 0.0534, "step": 3330 }, { "epoch": 4.36, "learning_rate": 3.3365e-05, "loss": 0.0677, "step": 3331 }, { "epoch": 4.36, "learning_rate": 3.336e-05, "loss": 0.0693, "step": 3332 }, { "epoch": 4.36, "learning_rate": 3.3355e-05, "loss": 0.0663, "step": 3333 }, { "epoch": 4.36, "learning_rate": 3.3350000000000004e-05, "loss": 0.0628, "step": 3334 }, { "epoch": 4.37, "learning_rate": 3.334500000000001e-05, "loss": 0.0585, "step": 3335 }, { "epoch": 4.37, "learning_rate": 3.3339999999999996e-05, "loss": 0.0536, "step": 3336 }, { "epoch": 4.37, "learning_rate": 3.3335e-05, "loss": 0.0671, "step": 3337 }, { "epoch": 4.37, "learning_rate": 3.333e-05, "loss": 0.0757, "step": 3338 }, { "epoch": 4.37, "learning_rate": 3.3325000000000004e-05, "loss": 0.0663, "step": 3339 }, { "epoch": 4.37, "learning_rate": 3.332e-05, "loss": 0.0626, "step": 3340 }, { "epoch": 4.37, "learning_rate": 3.3315e-05, "loss": 0.0654, "step": 3341 }, { "epoch": 4.37, "learning_rate": 3.3310000000000005e-05, "loss": 0.0743, "step": 3342 }, { "epoch": 4.38, "learning_rate": 3.3305e-05, "loss": 0.0601, "step": 3343 }, { "epoch": 4.38, "learning_rate": 3.33e-05, "loss": 0.0803, "step": 3344 }, { "epoch": 4.38, "learning_rate": 3.3295000000000006e-05, "loss": 0.0759, "step": 3345 }, { "epoch": 4.38, "learning_rate": 3.329e-05, "loss": 0.0771, "step": 3346 }, { "epoch": 4.38, "learning_rate": 3.3285e-05, "loss": 0.0707, "step": 3347 }, { "epoch": 4.38, "learning_rate": 3.328e-05, "loss": 0.0674, "step": 3348 }, { "epoch": 4.38, "learning_rate": 3.3275e-05, "loss": 0.0623, "step": 3349 }, { "epoch": 4.38, "learning_rate": 3.327e-05, "loss": 0.0571, "step": 3350 }, { "epoch": 4.39, "learning_rate": 3.3265e-05, "loss": 0.066, "step": 3351 }, { "epoch": 4.39, "learning_rate": 3.3260000000000003e-05, "loss": 0.0603, "step": 3352 }, { "epoch": 4.39, "learning_rate": 3.3255000000000006e-05, "loss": 0.076, "step": 3353 }, { "epoch": 4.39, "learning_rate": 3.325e-05, "loss": 0.0594, "step": 3354 }, { "epoch": 4.39, "learning_rate": 3.3245000000000004e-05, "loss": 0.0739, "step": 3355 }, { "epoch": 4.39, "learning_rate": 3.324e-05, "loss": 0.0814, "step": 3356 }, { "epoch": 4.39, "learning_rate": 3.3235e-05, "loss": 0.0657, "step": 3357 }, { "epoch": 4.4, "learning_rate": 3.323e-05, "loss": 0.0596, "step": 3358 }, { "epoch": 4.4, "learning_rate": 3.3225e-05, "loss": 0.0591, "step": 3359 }, { "epoch": 4.4, "learning_rate": 3.3220000000000004e-05, "loss": 0.0801, "step": 3360 }, { "epoch": 4.4, "learning_rate": 3.3215e-05, "loss": 0.0506, "step": 3361 }, { "epoch": 4.4, "learning_rate": 3.321e-05, "loss": 0.061, "step": 3362 }, { "epoch": 4.4, "learning_rate": 3.3205000000000005e-05, "loss": 0.0852, "step": 3363 }, { "epoch": 4.4, "learning_rate": 3.32e-05, "loss": 0.0805, "step": 3364 }, { "epoch": 4.4, "learning_rate": 3.3195e-05, "loss": 0.0731, "step": 3365 }, { "epoch": 4.41, "learning_rate": 3.319e-05, "loss": 0.0581, "step": 3366 }, { "epoch": 4.41, "learning_rate": 3.3185e-05, "loss": 0.0775, "step": 3367 }, { "epoch": 4.41, "learning_rate": 3.318e-05, "loss": 0.071, "step": 3368 }, { "epoch": 4.41, "learning_rate": 3.3175e-05, "loss": 0.0805, "step": 3369 }, { "epoch": 4.41, "learning_rate": 3.317e-05, "loss": 0.0603, "step": 3370 }, { "epoch": 4.41, "learning_rate": 3.3165e-05, "loss": 0.0724, "step": 3371 }, { "epoch": 4.41, "learning_rate": 3.316e-05, "loss": 0.0737, "step": 3372 }, { "epoch": 4.41, "learning_rate": 3.3155000000000004e-05, "loss": 0.0792, "step": 3373 }, { "epoch": 4.42, "learning_rate": 3.3150000000000006e-05, "loss": 0.0558, "step": 3374 }, { "epoch": 4.42, "learning_rate": 3.3145e-05, "loss": 0.0729, "step": 3375 }, { "epoch": 4.42, "learning_rate": 3.314e-05, "loss": 0.0775, "step": 3376 }, { "epoch": 4.42, "learning_rate": 3.3135e-05, "loss": 0.0627, "step": 3377 }, { "epoch": 4.42, "learning_rate": 3.313e-05, "loss": 0.065, "step": 3378 }, { "epoch": 4.42, "learning_rate": 3.3125e-05, "loss": 0.0624, "step": 3379 }, { "epoch": 4.42, "learning_rate": 3.312e-05, "loss": 0.0518, "step": 3380 }, { "epoch": 4.43, "learning_rate": 3.3115000000000004e-05, "loss": 0.0617, "step": 3381 }, { "epoch": 4.43, "learning_rate": 3.311e-05, "loss": 0.0603, "step": 3382 }, { "epoch": 4.43, "learning_rate": 3.3105e-05, "loss": 0.0585, "step": 3383 }, { "epoch": 4.43, "learning_rate": 3.3100000000000005e-05, "loss": 0.0776, "step": 3384 }, { "epoch": 4.43, "learning_rate": 3.3095e-05, "loss": 0.0822, "step": 3385 }, { "epoch": 4.43, "learning_rate": 3.309e-05, "loss": 0.0693, "step": 3386 }, { "epoch": 4.43, "learning_rate": 3.3085e-05, "loss": 0.0568, "step": 3387 }, { "epoch": 4.43, "learning_rate": 3.308e-05, "loss": 0.057, "step": 3388 }, { "epoch": 4.44, "learning_rate": 3.3075e-05, "loss": 0.0812, "step": 3389 }, { "epoch": 4.44, "learning_rate": 3.307e-05, "loss": 0.0615, "step": 3390 }, { "epoch": 4.44, "learning_rate": 3.3065e-05, "loss": 0.0659, "step": 3391 }, { "epoch": 4.44, "learning_rate": 3.3060000000000005e-05, "loss": 0.0735, "step": 3392 }, { "epoch": 4.44, "learning_rate": 3.3055e-05, "loss": 0.0718, "step": 3393 }, { "epoch": 4.44, "learning_rate": 3.3050000000000004e-05, "loss": 0.06, "step": 3394 }, { "epoch": 4.44, "learning_rate": 3.3045000000000006e-05, "loss": 0.0744, "step": 3395 }, { "epoch": 4.45, "learning_rate": 3.304e-05, "loss": 0.0739, "step": 3396 }, { "epoch": 4.45, "learning_rate": 3.3035e-05, "loss": 0.0785, "step": 3397 }, { "epoch": 4.45, "learning_rate": 3.303e-05, "loss": 0.0738, "step": 3398 }, { "epoch": 4.45, "learning_rate": 3.3025e-05, "loss": 0.0589, "step": 3399 }, { "epoch": 4.45, "learning_rate": 3.302e-05, "loss": 0.0718, "step": 3400 }, { "epoch": 4.45, "learning_rate": 3.3015e-05, "loss": 0.0781, "step": 3401 }, { "epoch": 4.45, "learning_rate": 3.3010000000000004e-05, "loss": 0.0702, "step": 3402 }, { "epoch": 4.45, "learning_rate": 3.3005e-05, "loss": 0.0695, "step": 3403 }, { "epoch": 4.46, "learning_rate": 3.3e-05, "loss": 0.0645, "step": 3404 }, { "epoch": 4.46, "learning_rate": 3.2995000000000005e-05, "loss": 0.0631, "step": 3405 }, { "epoch": 4.46, "learning_rate": 3.299e-05, "loss": 0.0757, "step": 3406 }, { "epoch": 4.46, "learning_rate": 3.2985e-05, "loss": 0.0553, "step": 3407 }, { "epoch": 4.46, "learning_rate": 3.298e-05, "loss": 0.0866, "step": 3408 }, { "epoch": 4.46, "learning_rate": 3.2975e-05, "loss": 0.0816, "step": 3409 }, { "epoch": 4.46, "learning_rate": 3.297e-05, "loss": 0.0747, "step": 3410 }, { "epoch": 4.46, "learning_rate": 3.2965e-05, "loss": 0.0743, "step": 3411 }, { "epoch": 4.47, "learning_rate": 3.296e-05, "loss": 0.0797, "step": 3412 }, { "epoch": 4.47, "learning_rate": 3.2955000000000006e-05, "loss": 0.0665, "step": 3413 }, { "epoch": 4.47, "learning_rate": 3.295e-05, "loss": 0.06, "step": 3414 }, { "epoch": 4.47, "learning_rate": 3.2945000000000004e-05, "loss": 0.0608, "step": 3415 }, { "epoch": 4.47, "learning_rate": 3.2940000000000006e-05, "loss": 0.0639, "step": 3416 }, { "epoch": 4.47, "learning_rate": 3.2935e-05, "loss": 0.0646, "step": 3417 }, { "epoch": 4.47, "learning_rate": 3.293e-05, "loss": 0.0573, "step": 3418 }, { "epoch": 4.48, "learning_rate": 3.2925e-05, "loss": 0.0846, "step": 3419 }, { "epoch": 4.48, "learning_rate": 3.292e-05, "loss": 0.0682, "step": 3420 }, { "epoch": 4.48, "learning_rate": 3.2915e-05, "loss": 0.0733, "step": 3421 }, { "epoch": 4.48, "learning_rate": 3.291e-05, "loss": 0.0665, "step": 3422 }, { "epoch": 4.48, "learning_rate": 3.2905000000000004e-05, "loss": 0.0815, "step": 3423 }, { "epoch": 4.48, "learning_rate": 3.29e-05, "loss": 0.0649, "step": 3424 }, { "epoch": 4.48, "learning_rate": 3.2895e-05, "loss": 0.066, "step": 3425 }, { "epoch": 4.48, "learning_rate": 3.2890000000000005e-05, "loss": 0.0686, "step": 3426 }, { "epoch": 4.49, "learning_rate": 3.2885e-05, "loss": 0.0714, "step": 3427 }, { "epoch": 4.49, "learning_rate": 3.288e-05, "loss": 0.073, "step": 3428 }, { "epoch": 4.49, "learning_rate": 3.2875e-05, "loss": 0.064, "step": 3429 }, { "epoch": 4.49, "learning_rate": 3.287e-05, "loss": 0.0688, "step": 3430 }, { "epoch": 4.49, "learning_rate": 3.2865000000000005e-05, "loss": 0.0703, "step": 3431 }, { "epoch": 4.49, "learning_rate": 3.286e-05, "loss": 0.0835, "step": 3432 }, { "epoch": 4.49, "learning_rate": 3.2855e-05, "loss": 0.0587, "step": 3433 }, { "epoch": 4.49, "learning_rate": 3.2850000000000006e-05, "loss": 0.065, "step": 3434 }, { "epoch": 4.5, "learning_rate": 3.2845e-05, "loss": 0.0637, "step": 3435 }, { "epoch": 4.5, "learning_rate": 3.2840000000000004e-05, "loss": 0.0589, "step": 3436 }, { "epoch": 4.5, "learning_rate": 3.2835e-05, "loss": 0.0702, "step": 3437 }, { "epoch": 4.5, "learning_rate": 3.283e-05, "loss": 0.082, "step": 3438 }, { "epoch": 4.5, "learning_rate": 3.2825e-05, "loss": 0.0742, "step": 3439 }, { "epoch": 4.5, "learning_rate": 3.282e-05, "loss": 0.0573, "step": 3440 }, { "epoch": 4.5, "learning_rate": 3.2815000000000003e-05, "loss": 0.0705, "step": 3441 }, { "epoch": 4.51, "learning_rate": 3.281e-05, "loss": 0.0642, "step": 3442 }, { "epoch": 4.51, "learning_rate": 3.2805e-05, "loss": 0.0738, "step": 3443 }, { "epoch": 4.51, "learning_rate": 3.2800000000000004e-05, "loss": 0.0847, "step": 3444 }, { "epoch": 4.51, "learning_rate": 3.2795e-05, "loss": 0.0824, "step": 3445 }, { "epoch": 4.51, "learning_rate": 3.279e-05, "loss": 0.0605, "step": 3446 }, { "epoch": 4.51, "learning_rate": 3.2785e-05, "loss": 0.0603, "step": 3447 }, { "epoch": 4.51, "learning_rate": 3.278e-05, "loss": 0.0626, "step": 3448 }, { "epoch": 4.51, "learning_rate": 3.2775e-05, "loss": 0.0745, "step": 3449 }, { "epoch": 4.52, "learning_rate": 3.277e-05, "loss": 0.0735, "step": 3450 }, { "epoch": 4.52, "learning_rate": 3.2765e-05, "loss": 0.0587, "step": 3451 }, { "epoch": 4.52, "learning_rate": 3.2760000000000005e-05, "loss": 0.0673, "step": 3452 }, { "epoch": 4.52, "learning_rate": 3.2755e-05, "loss": 0.068, "step": 3453 }, { "epoch": 4.52, "learning_rate": 3.275e-05, "loss": 0.0719, "step": 3454 }, { "epoch": 4.52, "learning_rate": 3.2745000000000006e-05, "loss": 0.0536, "step": 3455 }, { "epoch": 4.52, "learning_rate": 3.274e-05, "loss": 0.0727, "step": 3456 }, { "epoch": 4.52, "learning_rate": 3.2735e-05, "loss": 0.0634, "step": 3457 }, { "epoch": 4.53, "learning_rate": 3.273e-05, "loss": 0.0786, "step": 3458 }, { "epoch": 4.53, "learning_rate": 3.2725e-05, "loss": 0.0787, "step": 3459 }, { "epoch": 4.53, "learning_rate": 3.272e-05, "loss": 0.057, "step": 3460 }, { "epoch": 4.53, "learning_rate": 3.2715e-05, "loss": 0.0716, "step": 3461 }, { "epoch": 4.53, "learning_rate": 3.2710000000000004e-05, "loss": 0.0718, "step": 3462 }, { "epoch": 4.53, "learning_rate": 3.2705e-05, "loss": 0.0778, "step": 3463 }, { "epoch": 4.53, "learning_rate": 3.27e-05, "loss": 0.0745, "step": 3464 }, { "epoch": 4.54, "learning_rate": 3.2695000000000005e-05, "loss": 0.0644, "step": 3465 }, { "epoch": 4.54, "learning_rate": 3.269000000000001e-05, "loss": 0.0684, "step": 3466 }, { "epoch": 4.54, "learning_rate": 3.2684999999999996e-05, "loss": 0.0761, "step": 3467 }, { "epoch": 4.54, "learning_rate": 3.268e-05, "loss": 0.0727, "step": 3468 }, { "epoch": 4.54, "learning_rate": 3.2675e-05, "loss": 0.0669, "step": 3469 }, { "epoch": 4.54, "learning_rate": 3.267e-05, "loss": 0.0817, "step": 3470 }, { "epoch": 4.54, "learning_rate": 3.2665e-05, "loss": 0.0743, "step": 3471 }, { "epoch": 4.54, "learning_rate": 3.266e-05, "loss": 0.071, "step": 3472 }, { "epoch": 4.55, "learning_rate": 3.2655000000000005e-05, "loss": 0.0711, "step": 3473 }, { "epoch": 4.55, "learning_rate": 3.265e-05, "loss": 0.0674, "step": 3474 }, { "epoch": 4.55, "learning_rate": 3.2645e-05, "loss": 0.0516, "step": 3475 }, { "epoch": 4.55, "learning_rate": 3.2640000000000006e-05, "loss": 0.0708, "step": 3476 }, { "epoch": 4.55, "learning_rate": 3.2635e-05, "loss": 0.0651, "step": 3477 }, { "epoch": 4.55, "learning_rate": 3.263e-05, "loss": 0.0713, "step": 3478 }, { "epoch": 4.55, "learning_rate": 3.2625e-05, "loss": 0.0658, "step": 3479 }, { "epoch": 4.55, "learning_rate": 3.262e-05, "loss": 0.072, "step": 3480 }, { "epoch": 4.56, "learning_rate": 3.2615e-05, "loss": 0.0722, "step": 3481 }, { "epoch": 4.56, "learning_rate": 3.261e-05, "loss": 0.0754, "step": 3482 }, { "epoch": 4.56, "learning_rate": 3.2605000000000004e-05, "loss": 0.0785, "step": 3483 }, { "epoch": 4.56, "learning_rate": 3.26e-05, "loss": 0.0625, "step": 3484 }, { "epoch": 4.56, "learning_rate": 3.2595e-05, "loss": 0.0635, "step": 3485 }, { "epoch": 4.56, "learning_rate": 3.2590000000000005e-05, "loss": 0.0586, "step": 3486 }, { "epoch": 4.56, "learning_rate": 3.2585e-05, "loss": 0.0664, "step": 3487 }, { "epoch": 4.57, "learning_rate": 3.2579999999999996e-05, "loss": 0.0786, "step": 3488 }, { "epoch": 4.57, "learning_rate": 3.2575e-05, "loss": 0.0606, "step": 3489 }, { "epoch": 4.57, "learning_rate": 3.257e-05, "loss": 0.0786, "step": 3490 }, { "epoch": 4.57, "learning_rate": 3.2565000000000004e-05, "loss": 0.0637, "step": 3491 }, { "epoch": 4.57, "learning_rate": 3.256e-05, "loss": 0.0656, "step": 3492 }, { "epoch": 4.57, "learning_rate": 3.2555e-05, "loss": 0.0767, "step": 3493 }, { "epoch": 4.57, "learning_rate": 3.2550000000000005e-05, "loss": 0.0612, "step": 3494 }, { "epoch": 4.57, "learning_rate": 3.2545e-05, "loss": 0.071, "step": 3495 }, { "epoch": 4.58, "learning_rate": 3.2540000000000004e-05, "loss": 0.0756, "step": 3496 }, { "epoch": 4.58, "learning_rate": 3.2535e-05, "loss": 0.0701, "step": 3497 }, { "epoch": 4.58, "learning_rate": 3.253e-05, "loss": 0.0616, "step": 3498 }, { "epoch": 4.58, "learning_rate": 3.2525e-05, "loss": 0.0612, "step": 3499 }, { "epoch": 4.58, "learning_rate": 3.252e-05, "loss": 0.0854, "step": 3500 }, { "epoch": 4.58, "learning_rate": 3.2515e-05, "loss": 0.0814, "step": 3501 }, { "epoch": 4.58, "learning_rate": 3.251e-05, "loss": 0.0843, "step": 3502 }, { "epoch": 4.59, "learning_rate": 3.2505e-05, "loss": 0.0656, "step": 3503 }, { "epoch": 4.59, "learning_rate": 3.2500000000000004e-05, "loss": 0.0755, "step": 3504 }, { "epoch": 4.59, "learning_rate": 3.2495000000000007e-05, "loss": 0.0698, "step": 3505 }, { "epoch": 4.59, "learning_rate": 3.249e-05, "loss": 0.076, "step": 3506 }, { "epoch": 4.59, "learning_rate": 3.2485000000000005e-05, "loss": 0.0686, "step": 3507 }, { "epoch": 4.59, "learning_rate": 3.248e-05, "loss": 0.0701, "step": 3508 }, { "epoch": 4.59, "learning_rate": 3.2474999999999997e-05, "loss": 0.0719, "step": 3509 }, { "epoch": 4.59, "learning_rate": 3.247e-05, "loss": 0.0645, "step": 3510 }, { "epoch": 4.6, "learning_rate": 3.2465e-05, "loss": 0.0598, "step": 3511 }, { "epoch": 4.6, "learning_rate": 3.2460000000000004e-05, "loss": 0.0613, "step": 3512 }, { "epoch": 4.6, "learning_rate": 3.2455e-05, "loss": 0.0813, "step": 3513 }, { "epoch": 4.6, "learning_rate": 3.245e-05, "loss": 0.0718, "step": 3514 }, { "epoch": 4.6, "learning_rate": 3.2445000000000005e-05, "loss": 0.0737, "step": 3515 }, { "epoch": 4.6, "learning_rate": 3.244e-05, "loss": 0.0551, "step": 3516 }, { "epoch": 4.6, "learning_rate": 3.2435000000000004e-05, "loss": 0.0743, "step": 3517 }, { "epoch": 4.6, "learning_rate": 3.243e-05, "loss": 0.0585, "step": 3518 }, { "epoch": 4.61, "learning_rate": 3.2425e-05, "loss": 0.0638, "step": 3519 }, { "epoch": 4.61, "learning_rate": 3.242e-05, "loss": 0.0678, "step": 3520 }, { "epoch": 4.61, "learning_rate": 3.2415e-05, "loss": 0.0742, "step": 3521 }, { "epoch": 4.61, "learning_rate": 3.241e-05, "loss": 0.0658, "step": 3522 }, { "epoch": 4.61, "learning_rate": 3.2405e-05, "loss": 0.0577, "step": 3523 }, { "epoch": 4.61, "learning_rate": 3.24e-05, "loss": 0.0752, "step": 3524 }, { "epoch": 4.61, "learning_rate": 3.2395000000000004e-05, "loss": 0.0714, "step": 3525 }, { "epoch": 4.62, "learning_rate": 3.239000000000001e-05, "loss": 0.0662, "step": 3526 }, { "epoch": 4.62, "learning_rate": 3.2385e-05, "loss": 0.0624, "step": 3527 }, { "epoch": 4.62, "learning_rate": 3.238e-05, "loss": 0.0677, "step": 3528 }, { "epoch": 4.62, "learning_rate": 3.2375e-05, "loss": 0.0856, "step": 3529 }, { "epoch": 4.62, "learning_rate": 3.2370000000000003e-05, "loss": 0.0672, "step": 3530 }, { "epoch": 4.62, "learning_rate": 3.2365e-05, "loss": 0.0671, "step": 3531 }, { "epoch": 4.62, "learning_rate": 3.236e-05, "loss": 0.0612, "step": 3532 }, { "epoch": 4.62, "learning_rate": 3.2355000000000004e-05, "loss": 0.0718, "step": 3533 }, { "epoch": 4.63, "learning_rate": 3.235e-05, "loss": 0.0785, "step": 3534 }, { "epoch": 4.63, "learning_rate": 3.2345e-05, "loss": 0.0727, "step": 3535 }, { "epoch": 4.63, "learning_rate": 3.2340000000000005e-05, "loss": 0.0624, "step": 3536 }, { "epoch": 4.63, "learning_rate": 3.2335e-05, "loss": 0.0769, "step": 3537 }, { "epoch": 4.63, "learning_rate": 3.233e-05, "loss": 0.0691, "step": 3538 }, { "epoch": 4.63, "learning_rate": 3.2325e-05, "loss": 0.0858, "step": 3539 }, { "epoch": 4.63, "learning_rate": 3.232e-05, "loss": 0.0674, "step": 3540 }, { "epoch": 4.63, "learning_rate": 3.2315e-05, "loss": 0.0785, "step": 3541 }, { "epoch": 4.64, "learning_rate": 3.231e-05, "loss": 0.0757, "step": 3542 }, { "epoch": 4.64, "learning_rate": 3.2305e-05, "loss": 0.065, "step": 3543 }, { "epoch": 4.64, "learning_rate": 3.2300000000000006e-05, "loss": 0.0745, "step": 3544 }, { "epoch": 4.64, "learning_rate": 3.2295e-05, "loss": 0.0723, "step": 3545 }, { "epoch": 4.64, "learning_rate": 3.2290000000000004e-05, "loss": 0.0701, "step": 3546 }, { "epoch": 4.64, "learning_rate": 3.228500000000001e-05, "loss": 0.0725, "step": 3547 }, { "epoch": 4.64, "learning_rate": 3.2279999999999996e-05, "loss": 0.0621, "step": 3548 }, { "epoch": 4.65, "learning_rate": 3.2275e-05, "loss": 0.0792, "step": 3549 }, { "epoch": 4.65, "learning_rate": 3.227e-05, "loss": 0.0743, "step": 3550 }, { "epoch": 4.65, "learning_rate": 3.2265000000000004e-05, "loss": 0.0743, "step": 3551 }, { "epoch": 4.65, "learning_rate": 3.226e-05, "loss": 0.0667, "step": 3552 }, { "epoch": 4.65, "learning_rate": 3.2255e-05, "loss": 0.0606, "step": 3553 }, { "epoch": 4.65, "learning_rate": 3.2250000000000005e-05, "loss": 0.0605, "step": 3554 }, { "epoch": 4.65, "learning_rate": 3.2245e-05, "loss": 0.073, "step": 3555 }, { "epoch": 4.65, "learning_rate": 3.224e-05, "loss": 0.0761, "step": 3556 }, { "epoch": 4.66, "learning_rate": 3.2235000000000006e-05, "loss": 0.0855, "step": 3557 }, { "epoch": 4.66, "learning_rate": 3.223e-05, "loss": 0.064, "step": 3558 }, { "epoch": 4.66, "learning_rate": 3.2225e-05, "loss": 0.0721, "step": 3559 }, { "epoch": 4.66, "learning_rate": 3.222e-05, "loss": 0.0684, "step": 3560 }, { "epoch": 4.66, "learning_rate": 3.2215e-05, "loss": 0.0808, "step": 3561 }, { "epoch": 4.66, "learning_rate": 3.221e-05, "loss": 0.0785, "step": 3562 }, { "epoch": 4.66, "learning_rate": 3.2205e-05, "loss": 0.0738, "step": 3563 }, { "epoch": 4.66, "learning_rate": 3.2200000000000003e-05, "loss": 0.0704, "step": 3564 }, { "epoch": 4.67, "learning_rate": 3.2195000000000006e-05, "loss": 0.074, "step": 3565 }, { "epoch": 4.67, "learning_rate": 3.219e-05, "loss": 0.059, "step": 3566 }, { "epoch": 4.67, "learning_rate": 3.2185000000000004e-05, "loss": 0.0914, "step": 3567 }, { "epoch": 4.67, "learning_rate": 3.218e-05, "loss": 0.0643, "step": 3568 }, { "epoch": 4.67, "learning_rate": 3.2175e-05, "loss": 0.0628, "step": 3569 }, { "epoch": 4.67, "learning_rate": 3.217e-05, "loss": 0.0763, "step": 3570 }, { "epoch": 4.67, "learning_rate": 3.2165e-05, "loss": 0.0793, "step": 3571 }, { "epoch": 4.68, "learning_rate": 3.2160000000000004e-05, "loss": 0.08, "step": 3572 }, { "epoch": 4.68, "learning_rate": 3.2155e-05, "loss": 0.0795, "step": 3573 }, { "epoch": 4.68, "learning_rate": 3.215e-05, "loss": 0.0748, "step": 3574 }, { "epoch": 4.68, "learning_rate": 3.2145000000000005e-05, "loss": 0.0603, "step": 3575 }, { "epoch": 4.68, "learning_rate": 3.214e-05, "loss": 0.0838, "step": 3576 }, { "epoch": 4.68, "learning_rate": 3.2135e-05, "loss": 0.0741, "step": 3577 }, { "epoch": 4.68, "learning_rate": 3.213e-05, "loss": 0.0788, "step": 3578 }, { "epoch": 4.68, "learning_rate": 3.2125e-05, "loss": 0.0623, "step": 3579 }, { "epoch": 4.69, "learning_rate": 3.212e-05, "loss": 0.0831, "step": 3580 }, { "epoch": 4.69, "learning_rate": 3.2115e-05, "loss": 0.0517, "step": 3581 }, { "epoch": 4.69, "learning_rate": 3.211e-05, "loss": 0.0693, "step": 3582 }, { "epoch": 4.69, "learning_rate": 3.2105e-05, "loss": 0.0675, "step": 3583 }, { "epoch": 4.69, "learning_rate": 3.21e-05, "loss": 0.0743, "step": 3584 }, { "epoch": 4.69, "learning_rate": 3.2095000000000004e-05, "loss": 0.0685, "step": 3585 }, { "epoch": 4.69, "learning_rate": 3.2090000000000006e-05, "loss": 0.0818, "step": 3586 }, { "epoch": 4.7, "learning_rate": 3.2085e-05, "loss": 0.066, "step": 3587 }, { "epoch": 4.7, "learning_rate": 3.208e-05, "loss": 0.0686, "step": 3588 }, { "epoch": 4.7, "learning_rate": 3.2075e-05, "loss": 0.0632, "step": 3589 }, { "epoch": 4.7, "learning_rate": 3.207e-05, "loss": 0.0619, "step": 3590 }, { "epoch": 4.7, "learning_rate": 3.2065e-05, "loss": 0.0774, "step": 3591 }, { "epoch": 4.7, "learning_rate": 3.206e-05, "loss": 0.06, "step": 3592 }, { "epoch": 4.7, "learning_rate": 3.2055000000000004e-05, "loss": 0.0636, "step": 3593 }, { "epoch": 4.7, "learning_rate": 3.205e-05, "loss": 0.0691, "step": 3594 }, { "epoch": 4.71, "learning_rate": 3.2045e-05, "loss": 0.0695, "step": 3595 }, { "epoch": 4.71, "learning_rate": 3.2040000000000005e-05, "loss": 0.08, "step": 3596 }, { "epoch": 4.71, "learning_rate": 3.2035e-05, "loss": 0.0711, "step": 3597 }, { "epoch": 4.71, "learning_rate": 3.2029999999999997e-05, "loss": 0.088, "step": 3598 }, { "epoch": 4.71, "learning_rate": 3.2025e-05, "loss": 0.0747, "step": 3599 }, { "epoch": 4.71, "learning_rate": 3.202e-05, "loss": 0.1029, "step": 3600 }, { "epoch": 4.71, "learning_rate": 3.2015e-05, "loss": 0.0847, "step": 3601 }, { "epoch": 4.71, "learning_rate": 3.201e-05, "loss": 0.0761, "step": 3602 }, { "epoch": 4.72, "learning_rate": 3.2005e-05, "loss": 0.0703, "step": 3603 }, { "epoch": 4.72, "learning_rate": 3.2000000000000005e-05, "loss": 0.0845, "step": 3604 }, { "epoch": 4.72, "learning_rate": 3.1995e-05, "loss": 0.0832, "step": 3605 }, { "epoch": 4.72, "learning_rate": 3.1990000000000004e-05, "loss": 0.0668, "step": 3606 }, { "epoch": 4.72, "learning_rate": 3.1985000000000006e-05, "loss": 0.0809, "step": 3607 }, { "epoch": 4.72, "learning_rate": 3.198e-05, "loss": 0.0883, "step": 3608 }, { "epoch": 4.72, "learning_rate": 3.1975e-05, "loss": 0.0592, "step": 3609 }, { "epoch": 4.73, "learning_rate": 3.197e-05, "loss": 0.0601, "step": 3610 }, { "epoch": 4.73, "learning_rate": 3.1965e-05, "loss": 0.058, "step": 3611 }, { "epoch": 4.73, "learning_rate": 3.196e-05, "loss": 0.063, "step": 3612 }, { "epoch": 4.73, "learning_rate": 3.1955e-05, "loss": 0.0718, "step": 3613 }, { "epoch": 4.73, "learning_rate": 3.1950000000000004e-05, "loss": 0.0703, "step": 3614 }, { "epoch": 4.73, "learning_rate": 3.1945e-05, "loss": 0.087, "step": 3615 }, { "epoch": 4.73, "learning_rate": 3.194e-05, "loss": 0.0815, "step": 3616 }, { "epoch": 4.73, "learning_rate": 3.1935000000000005e-05, "loss": 0.0814, "step": 3617 }, { "epoch": 4.74, "learning_rate": 3.193e-05, "loss": 0.0732, "step": 3618 }, { "epoch": 4.74, "learning_rate": 3.1925e-05, "loss": 0.0662, "step": 3619 }, { "epoch": 4.74, "learning_rate": 3.192e-05, "loss": 0.0786, "step": 3620 }, { "epoch": 4.74, "learning_rate": 3.1915e-05, "loss": 0.0841, "step": 3621 }, { "epoch": 4.74, "learning_rate": 3.191e-05, "loss": 0.077, "step": 3622 }, { "epoch": 4.74, "learning_rate": 3.1905e-05, "loss": 0.0831, "step": 3623 }, { "epoch": 4.74, "learning_rate": 3.19e-05, "loss": 0.0802, "step": 3624 }, { "epoch": 4.74, "learning_rate": 3.1895000000000005e-05, "loss": 0.0701, "step": 3625 }, { "epoch": 4.75, "learning_rate": 3.189e-05, "loss": 0.0783, "step": 3626 }, { "epoch": 4.75, "learning_rate": 3.1885000000000004e-05, "loss": 0.0719, "step": 3627 }, { "epoch": 4.75, "learning_rate": 3.188e-05, "loss": 0.0741, "step": 3628 }, { "epoch": 4.75, "learning_rate": 3.1875e-05, "loss": 0.0798, "step": 3629 }, { "epoch": 4.75, "learning_rate": 3.187e-05, "loss": 0.083, "step": 3630 }, { "epoch": 4.75, "learning_rate": 3.1865e-05, "loss": 0.0691, "step": 3631 }, { "epoch": 4.75, "learning_rate": 3.186e-05, "loss": 0.0769, "step": 3632 }, { "epoch": 4.76, "learning_rate": 3.1855e-05, "loss": 0.0803, "step": 3633 }, { "epoch": 4.76, "learning_rate": 3.185e-05, "loss": 0.0808, "step": 3634 }, { "epoch": 4.76, "learning_rate": 3.1845000000000004e-05, "loss": 0.0678, "step": 3635 }, { "epoch": 4.76, "learning_rate": 3.184e-05, "loss": 0.0844, "step": 3636 }, { "epoch": 4.76, "learning_rate": 3.1835e-05, "loss": 0.0896, "step": 3637 }, { "epoch": 4.76, "learning_rate": 3.1830000000000005e-05, "loss": 0.0679, "step": 3638 }, { "epoch": 4.76, "learning_rate": 3.1825e-05, "loss": 0.0674, "step": 3639 }, { "epoch": 4.76, "learning_rate": 3.182e-05, "loss": 0.0707, "step": 3640 }, { "epoch": 4.77, "learning_rate": 3.1815e-05, "loss": 0.076, "step": 3641 }, { "epoch": 4.77, "learning_rate": 3.181e-05, "loss": 0.0868, "step": 3642 }, { "epoch": 4.77, "learning_rate": 3.1805000000000005e-05, "loss": 0.0755, "step": 3643 }, { "epoch": 4.77, "learning_rate": 3.18e-05, "loss": 0.0607, "step": 3644 }, { "epoch": 4.77, "learning_rate": 3.1795e-05, "loss": 0.0737, "step": 3645 }, { "epoch": 4.77, "learning_rate": 3.1790000000000006e-05, "loss": 0.0741, "step": 3646 }, { "epoch": 4.77, "learning_rate": 3.1785e-05, "loss": 0.0819, "step": 3647 }, { "epoch": 4.77, "learning_rate": 3.1780000000000004e-05, "loss": 0.0732, "step": 3648 }, { "epoch": 4.78, "learning_rate": 3.1775e-05, "loss": 0.0803, "step": 3649 }, { "epoch": 4.78, "learning_rate": 3.177e-05, "loss": 0.0633, "step": 3650 }, { "epoch": 4.78, "learning_rate": 3.1765e-05, "loss": 0.063, "step": 3651 }, { "epoch": 4.78, "learning_rate": 3.176e-05, "loss": 0.0895, "step": 3652 }, { "epoch": 4.78, "learning_rate": 3.1755000000000003e-05, "loss": 0.0633, "step": 3653 }, { "epoch": 4.78, "learning_rate": 3.175e-05, "loss": 0.0783, "step": 3654 }, { "epoch": 4.78, "learning_rate": 3.1745e-05, "loss": 0.0662, "step": 3655 }, { "epoch": 4.79, "learning_rate": 3.1740000000000004e-05, "loss": 0.0575, "step": 3656 }, { "epoch": 4.79, "learning_rate": 3.1735e-05, "loss": 0.0901, "step": 3657 }, { "epoch": 4.79, "learning_rate": 3.173e-05, "loss": 0.0766, "step": 3658 }, { "epoch": 4.79, "learning_rate": 3.1725e-05, "loss": 0.0662, "step": 3659 }, { "epoch": 4.79, "learning_rate": 3.172e-05, "loss": 0.077, "step": 3660 }, { "epoch": 4.79, "learning_rate": 3.1715e-05, "loss": 0.0771, "step": 3661 }, { "epoch": 4.79, "learning_rate": 3.171e-05, "loss": 0.0704, "step": 3662 }, { "epoch": 4.79, "learning_rate": 3.1705e-05, "loss": 0.0622, "step": 3663 }, { "epoch": 4.8, "learning_rate": 3.1700000000000005e-05, "loss": 0.0698, "step": 3664 }, { "epoch": 4.8, "learning_rate": 3.1695e-05, "loss": 0.0741, "step": 3665 }, { "epoch": 4.8, "learning_rate": 3.169e-05, "loss": 0.0657, "step": 3666 }, { "epoch": 4.8, "learning_rate": 3.1685000000000006e-05, "loss": 0.0586, "step": 3667 }, { "epoch": 4.8, "learning_rate": 3.168e-05, "loss": 0.0642, "step": 3668 }, { "epoch": 4.8, "learning_rate": 3.1675e-05, "loss": 0.0719, "step": 3669 }, { "epoch": 4.8, "learning_rate": 3.167e-05, "loss": 0.0842, "step": 3670 }, { "epoch": 4.8, "learning_rate": 3.1665e-05, "loss": 0.0692, "step": 3671 }, { "epoch": 4.81, "learning_rate": 3.166e-05, "loss": 0.0873, "step": 3672 }, { "epoch": 4.81, "learning_rate": 3.1655e-05, "loss": 0.0863, "step": 3673 }, { "epoch": 4.81, "learning_rate": 3.1650000000000004e-05, "loss": 0.0926, "step": 3674 }, { "epoch": 4.81, "learning_rate": 3.1645e-05, "loss": 0.0952, "step": 3675 }, { "epoch": 4.81, "learning_rate": 3.164e-05, "loss": 0.0627, "step": 3676 }, { "epoch": 4.81, "learning_rate": 3.1635000000000005e-05, "loss": 0.0812, "step": 3677 }, { "epoch": 4.81, "learning_rate": 3.163000000000001e-05, "loss": 0.0678, "step": 3678 }, { "epoch": 4.82, "learning_rate": 3.1624999999999996e-05, "loss": 0.0723, "step": 3679 }, { "epoch": 4.82, "learning_rate": 3.162e-05, "loss": 0.0653, "step": 3680 }, { "epoch": 4.82, "learning_rate": 3.1615e-05, "loss": 0.0732, "step": 3681 }, { "epoch": 4.82, "learning_rate": 3.1610000000000004e-05, "loss": 0.0605, "step": 3682 }, { "epoch": 4.82, "learning_rate": 3.1605e-05, "loss": 0.0714, "step": 3683 }, { "epoch": 4.82, "learning_rate": 3.16e-05, "loss": 0.0714, "step": 3684 }, { "epoch": 4.82, "learning_rate": 3.1595000000000005e-05, "loss": 0.0724, "step": 3685 }, { "epoch": 4.82, "learning_rate": 3.159e-05, "loss": 0.0784, "step": 3686 }, { "epoch": 4.83, "learning_rate": 3.1585e-05, "loss": 0.0713, "step": 3687 }, { "epoch": 4.83, "learning_rate": 3.1580000000000006e-05, "loss": 0.0785, "step": 3688 }, { "epoch": 4.83, "learning_rate": 3.1575e-05, "loss": 0.0783, "step": 3689 }, { "epoch": 4.83, "learning_rate": 3.157e-05, "loss": 0.0812, "step": 3690 }, { "epoch": 4.83, "learning_rate": 3.1565e-05, "loss": 0.0788, "step": 3691 }, { "epoch": 4.83, "learning_rate": 3.156e-05, "loss": 0.0697, "step": 3692 }, { "epoch": 4.83, "learning_rate": 3.1555e-05, "loss": 0.0779, "step": 3693 }, { "epoch": 4.84, "learning_rate": 3.155e-05, "loss": 0.0746, "step": 3694 }, { "epoch": 4.84, "learning_rate": 3.1545000000000004e-05, "loss": 0.0878, "step": 3695 }, { "epoch": 4.84, "learning_rate": 3.154e-05, "loss": 0.0661, "step": 3696 }, { "epoch": 4.84, "learning_rate": 3.1535e-05, "loss": 0.0825, "step": 3697 }, { "epoch": 4.84, "learning_rate": 3.1530000000000005e-05, "loss": 0.0612, "step": 3698 }, { "epoch": 4.84, "learning_rate": 3.1525e-05, "loss": 0.0667, "step": 3699 }, { "epoch": 4.84, "learning_rate": 3.1519999999999996e-05, "loss": 0.0767, "step": 3700 }, { "epoch": 4.84, "learning_rate": 3.1515e-05, "loss": 0.0866, "step": 3701 }, { "epoch": 4.85, "learning_rate": 3.151e-05, "loss": 0.0836, "step": 3702 }, { "epoch": 4.85, "learning_rate": 3.1505000000000004e-05, "loss": 0.0721, "step": 3703 }, { "epoch": 4.85, "learning_rate": 3.15e-05, "loss": 0.0735, "step": 3704 }, { "epoch": 4.85, "learning_rate": 3.1495e-05, "loss": 0.0801, "step": 3705 }, { "epoch": 4.85, "learning_rate": 3.1490000000000005e-05, "loss": 0.0711, "step": 3706 }, { "epoch": 4.85, "learning_rate": 3.1485e-05, "loss": 0.0729, "step": 3707 }, { "epoch": 4.85, "learning_rate": 3.1480000000000004e-05, "loss": 0.0774, "step": 3708 }, { "epoch": 4.85, "learning_rate": 3.1475e-05, "loss": 0.0841, "step": 3709 }, { "epoch": 4.86, "learning_rate": 3.147e-05, "loss": 0.0672, "step": 3710 }, { "epoch": 4.86, "learning_rate": 3.1465e-05, "loss": 0.0794, "step": 3711 }, { "epoch": 4.86, "learning_rate": 3.146e-05, "loss": 0.0717, "step": 3712 }, { "epoch": 4.86, "learning_rate": 3.1455e-05, "loss": 0.0855, "step": 3713 }, { "epoch": 4.86, "learning_rate": 3.145e-05, "loss": 0.0797, "step": 3714 }, { "epoch": 4.86, "learning_rate": 3.1445e-05, "loss": 0.0695, "step": 3715 }, { "epoch": 4.86, "learning_rate": 3.1440000000000004e-05, "loss": 0.0772, "step": 3716 }, { "epoch": 4.87, "learning_rate": 3.1435000000000007e-05, "loss": 0.0865, "step": 3717 }, { "epoch": 4.87, "learning_rate": 3.143e-05, "loss": 0.0812, "step": 3718 }, { "epoch": 4.87, "learning_rate": 3.1425e-05, "loss": 0.0738, "step": 3719 }, { "epoch": 4.87, "learning_rate": 3.142e-05, "loss": 0.0856, "step": 3720 }, { "epoch": 4.87, "learning_rate": 3.1415e-05, "loss": 0.0794, "step": 3721 }, { "epoch": 4.87, "learning_rate": 3.141e-05, "loss": 0.0756, "step": 3722 }, { "epoch": 4.87, "learning_rate": 3.1405e-05, "loss": 0.0729, "step": 3723 }, { "epoch": 4.87, "learning_rate": 3.1400000000000004e-05, "loss": 0.0946, "step": 3724 }, { "epoch": 4.88, "learning_rate": 3.1395e-05, "loss": 0.0778, "step": 3725 }, { "epoch": 4.88, "learning_rate": 3.139e-05, "loss": 0.0861, "step": 3726 }, { "epoch": 4.88, "learning_rate": 3.1385000000000005e-05, "loss": 0.0719, "step": 3727 }, { "epoch": 4.88, "learning_rate": 3.138e-05, "loss": 0.0701, "step": 3728 }, { "epoch": 4.88, "learning_rate": 3.1375e-05, "loss": 0.0701, "step": 3729 }, { "epoch": 4.88, "learning_rate": 3.137e-05, "loss": 0.0628, "step": 3730 }, { "epoch": 4.88, "learning_rate": 3.1365e-05, "loss": 0.093, "step": 3731 }, { "epoch": 4.88, "learning_rate": 3.136e-05, "loss": 0.0629, "step": 3732 }, { "epoch": 4.89, "learning_rate": 3.1355e-05, "loss": 0.0759, "step": 3733 }, { "epoch": 4.89, "learning_rate": 3.135e-05, "loss": 0.081, "step": 3734 }, { "epoch": 4.89, "learning_rate": 3.1345e-05, "loss": 0.0767, "step": 3735 }, { "epoch": 4.89, "learning_rate": 3.134e-05, "loss": 0.07, "step": 3736 }, { "epoch": 4.89, "learning_rate": 3.1335000000000004e-05, "loss": 0.0799, "step": 3737 }, { "epoch": 4.89, "learning_rate": 3.133000000000001e-05, "loss": 0.0683, "step": 3738 }, { "epoch": 4.89, "learning_rate": 3.1324999999999996e-05, "loss": 0.0806, "step": 3739 }, { "epoch": 4.9, "learning_rate": 3.132e-05, "loss": 0.0763, "step": 3740 }, { "epoch": 4.9, "learning_rate": 3.1315e-05, "loss": 0.0716, "step": 3741 }, { "epoch": 4.9, "learning_rate": 3.1310000000000003e-05, "loss": 0.0835, "step": 3742 }, { "epoch": 4.9, "learning_rate": 3.1305e-05, "loss": 0.0842, "step": 3743 }, { "epoch": 4.9, "learning_rate": 3.13e-05, "loss": 0.0857, "step": 3744 }, { "epoch": 4.9, "learning_rate": 3.1295000000000004e-05, "loss": 0.08, "step": 3745 }, { "epoch": 4.9, "learning_rate": 3.129e-05, "loss": 0.0841, "step": 3746 }, { "epoch": 4.9, "learning_rate": 3.1285e-05, "loss": 0.0712, "step": 3747 }, { "epoch": 4.91, "learning_rate": 3.1280000000000005e-05, "loss": 0.0917, "step": 3748 }, { "epoch": 4.91, "learning_rate": 3.1275e-05, "loss": 0.0746, "step": 3749 }, { "epoch": 4.91, "learning_rate": 3.127e-05, "loss": 0.0773, "step": 3750 }, { "epoch": 4.91, "learning_rate": 3.1265e-05, "loss": 0.0742, "step": 3751 }, { "epoch": 4.91, "learning_rate": 3.126e-05, "loss": 0.0687, "step": 3752 }, { "epoch": 4.91, "learning_rate": 3.1255e-05, "loss": 0.072, "step": 3753 }, { "epoch": 4.91, "learning_rate": 3.125e-05, "loss": 0.0765, "step": 3754 }, { "epoch": 4.91, "learning_rate": 3.1245e-05, "loss": 0.0879, "step": 3755 }, { "epoch": 4.92, "learning_rate": 3.1240000000000006e-05, "loss": 0.0622, "step": 3756 }, { "epoch": 4.92, "learning_rate": 3.1235e-05, "loss": 0.0659, "step": 3757 }, { "epoch": 4.92, "learning_rate": 3.1230000000000004e-05, "loss": 0.0885, "step": 3758 }, { "epoch": 4.92, "learning_rate": 3.122500000000001e-05, "loss": 0.0872, "step": 3759 }, { "epoch": 4.92, "learning_rate": 3.122e-05, "loss": 0.0782, "step": 3760 }, { "epoch": 4.92, "learning_rate": 3.1215e-05, "loss": 0.0643, "step": 3761 }, { "epoch": 4.92, "learning_rate": 3.121e-05, "loss": 0.0635, "step": 3762 }, { "epoch": 4.93, "learning_rate": 3.1205000000000004e-05, "loss": 0.0785, "step": 3763 }, { "epoch": 4.93, "learning_rate": 3.12e-05, "loss": 0.0635, "step": 3764 }, { "epoch": 4.93, "learning_rate": 3.1195e-05, "loss": 0.0863, "step": 3765 }, { "epoch": 4.93, "learning_rate": 3.1190000000000005e-05, "loss": 0.0879, "step": 3766 }, { "epoch": 4.93, "learning_rate": 3.1185e-05, "loss": 0.0771, "step": 3767 }, { "epoch": 4.93, "learning_rate": 3.118e-05, "loss": 0.078, "step": 3768 }, { "epoch": 4.93, "learning_rate": 3.1175000000000006e-05, "loss": 0.0756, "step": 3769 }, { "epoch": 4.93, "learning_rate": 3.117e-05, "loss": 0.0703, "step": 3770 }, { "epoch": 4.94, "learning_rate": 3.1165e-05, "loss": 0.0786, "step": 3771 }, { "epoch": 4.94, "learning_rate": 3.116e-05, "loss": 0.0778, "step": 3772 }, { "epoch": 4.94, "learning_rate": 3.1155e-05, "loss": 0.0759, "step": 3773 }, { "epoch": 4.94, "learning_rate": 3.115e-05, "loss": 0.0656, "step": 3774 }, { "epoch": 4.94, "learning_rate": 3.1145e-05, "loss": 0.0871, "step": 3775 }, { "epoch": 4.94, "learning_rate": 3.1140000000000003e-05, "loss": 0.0653, "step": 3776 }, { "epoch": 4.94, "learning_rate": 3.1135000000000006e-05, "loss": 0.0818, "step": 3777 }, { "epoch": 4.95, "learning_rate": 3.113e-05, "loss": 0.0711, "step": 3778 }, { "epoch": 4.95, "learning_rate": 3.1125000000000004e-05, "loss": 0.0705, "step": 3779 }, { "epoch": 4.95, "learning_rate": 3.112e-05, "loss": 0.0734, "step": 3780 }, { "epoch": 4.95, "learning_rate": 3.1115e-05, "loss": 0.0811, "step": 3781 }, { "epoch": 4.95, "learning_rate": 3.111e-05, "loss": 0.0735, "step": 3782 }, { "epoch": 4.95, "learning_rate": 3.1105e-05, "loss": 0.0704, "step": 3783 }, { "epoch": 4.95, "learning_rate": 3.1100000000000004e-05, "loss": 0.0664, "step": 3784 }, { "epoch": 4.95, "learning_rate": 3.1095e-05, "loss": 0.0775, "step": 3785 }, { "epoch": 4.96, "learning_rate": 3.109e-05, "loss": 0.0657, "step": 3786 }, { "epoch": 4.96, "learning_rate": 3.1085000000000005e-05, "loss": 0.0791, "step": 3787 }, { "epoch": 4.96, "learning_rate": 3.108e-05, "loss": 0.0774, "step": 3788 }, { "epoch": 4.96, "learning_rate": 3.1075e-05, "loss": 0.0688, "step": 3789 }, { "epoch": 4.96, "learning_rate": 3.107e-05, "loss": 0.0842, "step": 3790 }, { "epoch": 4.96, "learning_rate": 3.1065e-05, "loss": 0.0836, "step": 3791 }, { "epoch": 4.96, "learning_rate": 3.106e-05, "loss": 0.0846, "step": 3792 }, { "epoch": 4.96, "learning_rate": 3.1055e-05, "loss": 0.0676, "step": 3793 }, { "epoch": 4.97, "learning_rate": 3.105e-05, "loss": 0.0741, "step": 3794 }, { "epoch": 4.97, "learning_rate": 3.1045000000000005e-05, "loss": 0.0843, "step": 3795 }, { "epoch": 4.97, "learning_rate": 3.104e-05, "loss": 0.0622, "step": 3796 }, { "epoch": 4.97, "learning_rate": 3.1035000000000004e-05, "loss": 0.0583, "step": 3797 }, { "epoch": 4.97, "learning_rate": 3.1030000000000006e-05, "loss": 0.079, "step": 3798 }, { "epoch": 4.97, "learning_rate": 3.1025e-05, "loss": 0.0685, "step": 3799 }, { "epoch": 4.97, "learning_rate": 3.102e-05, "loss": 0.0772, "step": 3800 }, { "epoch": 4.98, "learning_rate": 3.1015e-05, "loss": 0.074, "step": 3801 }, { "epoch": 4.98, "learning_rate": 3.101e-05, "loss": 0.0661, "step": 3802 }, { "epoch": 4.98, "learning_rate": 3.1005e-05, "loss": 0.0817, "step": 3803 }, { "epoch": 4.98, "learning_rate": 3.1e-05, "loss": 0.0665, "step": 3804 }, { "epoch": 4.98, "learning_rate": 3.0995000000000004e-05, "loss": 0.0614, "step": 3805 }, { "epoch": 4.98, "learning_rate": 3.099e-05, "loss": 0.0659, "step": 3806 }, { "epoch": 4.98, "learning_rate": 3.0985e-05, "loss": 0.0672, "step": 3807 }, { "epoch": 4.98, "learning_rate": 3.0980000000000005e-05, "loss": 0.0933, "step": 3808 }, { "epoch": 4.99, "learning_rate": 3.0975e-05, "loss": 0.0803, "step": 3809 }, { "epoch": 4.99, "learning_rate": 3.0969999999999997e-05, "loss": 0.0597, "step": 3810 }, { "epoch": 4.99, "learning_rate": 3.0965e-05, "loss": 0.0753, "step": 3811 }, { "epoch": 4.99, "learning_rate": 3.096e-05, "loss": 0.0729, "step": 3812 }, { "epoch": 4.99, "learning_rate": 3.0955e-05, "loss": 0.0702, "step": 3813 }, { "epoch": 4.99, "learning_rate": 3.095e-05, "loss": 0.0802, "step": 3814 }, { "epoch": 4.99, "learning_rate": 3.0945e-05, "loss": 0.0854, "step": 3815 }, { "epoch": 4.99, "learning_rate": 3.0940000000000005e-05, "loss": 0.0809, "step": 3816 }, { "epoch": 5.0, "learning_rate": 3.0935e-05, "loss": 0.0737, "step": 3817 }, { "epoch": 5.0, "learning_rate": 3.0930000000000004e-05, "loss": 0.077, "step": 3818 }, { "epoch": 5.0, "learning_rate": 3.0925000000000006e-05, "loss": 0.0919, "step": 3819 }, { "epoch": 5.0, "learning_rate": 3.092e-05, "loss": 0.0639, "step": 3820 }, { "epoch": 5.0, "learning_rate": 3.0915e-05, "loss": 0.0326, "step": 3821 }, { "epoch": 5.0, "learning_rate": 3.091e-05, "loss": 0.0304, "step": 3822 }, { "epoch": 5.0, "learning_rate": 3.0905e-05, "loss": 0.0392, "step": 3823 }, { "epoch": 5.01, "learning_rate": 3.09e-05, "loss": 0.0344, "step": 3824 }, { "epoch": 5.01, "learning_rate": 3.0895e-05, "loss": 0.0332, "step": 3825 }, { "epoch": 5.01, "learning_rate": 3.0890000000000004e-05, "loss": 0.0266, "step": 3826 }, { "epoch": 5.01, "learning_rate": 3.0885e-05, "loss": 0.0378, "step": 3827 }, { "epoch": 5.01, "learning_rate": 3.088e-05, "loss": 0.0355, "step": 3828 }, { "epoch": 5.01, "learning_rate": 3.0875000000000005e-05, "loss": 0.0418, "step": 3829 }, { "epoch": 5.01, "learning_rate": 3.087e-05, "loss": 0.0298, "step": 3830 }, { "epoch": 5.01, "learning_rate": 3.0865e-05, "loss": 0.0303, "step": 3831 }, { "epoch": 5.02, "learning_rate": 3.086e-05, "loss": 0.0371, "step": 3832 }, { "epoch": 5.02, "learning_rate": 3.0855e-05, "loss": 0.0372, "step": 3833 }, { "epoch": 5.02, "learning_rate": 3.0850000000000004e-05, "loss": 0.0316, "step": 3834 }, { "epoch": 5.02, "learning_rate": 3.0845e-05, "loss": 0.0277, "step": 3835 }, { "epoch": 5.02, "learning_rate": 3.084e-05, "loss": 0.0336, "step": 3836 }, { "epoch": 5.02, "learning_rate": 3.0835000000000005e-05, "loss": 0.0462, "step": 3837 }, { "epoch": 5.02, "learning_rate": 3.083e-05, "loss": 0.0298, "step": 3838 }, { "epoch": 5.02, "learning_rate": 3.0825000000000004e-05, "loss": 0.0313, "step": 3839 }, { "epoch": 5.03, "learning_rate": 3.082e-05, "loss": 0.035, "step": 3840 }, { "epoch": 5.03, "learning_rate": 3.0815e-05, "loss": 0.0363, "step": 3841 }, { "epoch": 5.03, "learning_rate": 3.081e-05, "loss": 0.0387, "step": 3842 }, { "epoch": 5.03, "learning_rate": 3.0805e-05, "loss": 0.0361, "step": 3843 }, { "epoch": 5.03, "learning_rate": 3.08e-05, "loss": 0.032, "step": 3844 }, { "epoch": 5.03, "learning_rate": 3.0795e-05, "loss": 0.0333, "step": 3845 }, { "epoch": 5.03, "learning_rate": 3.079e-05, "loss": 0.0387, "step": 3846 }, { "epoch": 5.04, "learning_rate": 3.0785000000000004e-05, "loss": 0.0374, "step": 3847 }, { "epoch": 5.04, "learning_rate": 3.078e-05, "loss": 0.033, "step": 3848 }, { "epoch": 5.04, "learning_rate": 3.0775e-05, "loss": 0.0322, "step": 3849 }, { "epoch": 5.04, "learning_rate": 3.077e-05, "loss": 0.0294, "step": 3850 }, { "epoch": 5.04, "learning_rate": 3.0765e-05, "loss": 0.0299, "step": 3851 }, { "epoch": 5.04, "learning_rate": 3.076e-05, "loss": 0.0326, "step": 3852 }, { "epoch": 5.04, "learning_rate": 3.0755e-05, "loss": 0.0322, "step": 3853 }, { "epoch": 5.04, "learning_rate": 3.075e-05, "loss": 0.0324, "step": 3854 }, { "epoch": 5.05, "learning_rate": 3.0745000000000005e-05, "loss": 0.0365, "step": 3855 }, { "epoch": 5.05, "learning_rate": 3.074e-05, "loss": 0.041, "step": 3856 }, { "epoch": 5.05, "learning_rate": 3.0735e-05, "loss": 0.026, "step": 3857 }, { "epoch": 5.05, "learning_rate": 3.0730000000000006e-05, "loss": 0.0261, "step": 3858 }, { "epoch": 5.05, "learning_rate": 3.0725e-05, "loss": 0.038, "step": 3859 }, { "epoch": 5.05, "learning_rate": 3.072e-05, "loss": 0.0311, "step": 3860 }, { "epoch": 5.05, "learning_rate": 3.0715e-05, "loss": 0.0304, "step": 3861 }, { "epoch": 5.05, "learning_rate": 3.071e-05, "loss": 0.0332, "step": 3862 }, { "epoch": 5.06, "learning_rate": 3.0705e-05, "loss": 0.0333, "step": 3863 }, { "epoch": 5.06, "learning_rate": 3.07e-05, "loss": 0.0393, "step": 3864 }, { "epoch": 5.06, "learning_rate": 3.0695000000000003e-05, "loss": 0.0302, "step": 3865 }, { "epoch": 5.06, "learning_rate": 3.069e-05, "loss": 0.0341, "step": 3866 }, { "epoch": 5.06, "learning_rate": 3.0685e-05, "loss": 0.0245, "step": 3867 }, { "epoch": 5.06, "learning_rate": 3.0680000000000004e-05, "loss": 0.0345, "step": 3868 }, { "epoch": 5.06, "learning_rate": 3.067500000000001e-05, "loss": 0.0373, "step": 3869 }, { "epoch": 5.07, "learning_rate": 3.0669999999999996e-05, "loss": 0.0303, "step": 3870 }, { "epoch": 5.07, "learning_rate": 3.0665e-05, "loss": 0.0363, "step": 3871 }, { "epoch": 5.07, "learning_rate": 3.066e-05, "loss": 0.0324, "step": 3872 }, { "epoch": 5.07, "learning_rate": 3.0655e-05, "loss": 0.0305, "step": 3873 }, { "epoch": 5.07, "learning_rate": 3.065e-05, "loss": 0.0285, "step": 3874 }, { "epoch": 5.07, "learning_rate": 3.0645e-05, "loss": 0.0285, "step": 3875 }, { "epoch": 5.07, "learning_rate": 3.0640000000000005e-05, "loss": 0.029, "step": 3876 }, { "epoch": 5.07, "learning_rate": 3.0635e-05, "loss": 0.0382, "step": 3877 }, { "epoch": 5.08, "learning_rate": 3.063e-05, "loss": 0.0283, "step": 3878 }, { "epoch": 5.08, "learning_rate": 3.0625000000000006e-05, "loss": 0.0307, "step": 3879 }, { "epoch": 5.08, "learning_rate": 3.062e-05, "loss": 0.0372, "step": 3880 }, { "epoch": 5.08, "learning_rate": 3.0615e-05, "loss": 0.0301, "step": 3881 }, { "epoch": 5.08, "learning_rate": 3.061e-05, "loss": 0.0302, "step": 3882 }, { "epoch": 5.08, "learning_rate": 3.0605e-05, "loss": 0.0421, "step": 3883 }, { "epoch": 5.08, "learning_rate": 3.06e-05, "loss": 0.0294, "step": 3884 }, { "epoch": 5.09, "learning_rate": 3.0595e-05, "loss": 0.0335, "step": 3885 }, { "epoch": 5.09, "learning_rate": 3.0590000000000004e-05, "loss": 0.0364, "step": 3886 }, { "epoch": 5.09, "learning_rate": 3.0585e-05, "loss": 0.0309, "step": 3887 }, { "epoch": 5.09, "learning_rate": 3.058e-05, "loss": 0.0338, "step": 3888 }, { "epoch": 5.09, "learning_rate": 3.0575000000000005e-05, "loss": 0.028, "step": 3889 }, { "epoch": 5.09, "learning_rate": 3.057000000000001e-05, "loss": 0.034, "step": 3890 }, { "epoch": 5.09, "learning_rate": 3.0564999999999996e-05, "loss": 0.0346, "step": 3891 }, { "epoch": 5.09, "learning_rate": 3.056e-05, "loss": 0.0309, "step": 3892 }, { "epoch": 5.1, "learning_rate": 3.0555e-05, "loss": 0.0288, "step": 3893 }, { "epoch": 5.1, "learning_rate": 3.0550000000000004e-05, "loss": 0.0366, "step": 3894 }, { "epoch": 5.1, "learning_rate": 3.0545e-05, "loss": 0.0391, "step": 3895 }, { "epoch": 5.1, "learning_rate": 3.054e-05, "loss": 0.0387, "step": 3896 }, { "epoch": 5.1, "learning_rate": 3.0535000000000005e-05, "loss": 0.0366, "step": 3897 }, { "epoch": 5.1, "learning_rate": 3.053e-05, "loss": 0.0308, "step": 3898 }, { "epoch": 5.1, "learning_rate": 3.0525e-05, "loss": 0.0367, "step": 3899 }, { "epoch": 5.1, "learning_rate": 3.0520000000000006e-05, "loss": 0.0335, "step": 3900 }, { "epoch": 5.11, "learning_rate": 3.0515e-05, "loss": 0.0369, "step": 3901 }, { "epoch": 5.11, "learning_rate": 3.051e-05, "loss": 0.0297, "step": 3902 }, { "epoch": 5.11, "learning_rate": 3.0505e-05, "loss": 0.0285, "step": 3903 }, { "epoch": 5.11, "learning_rate": 3.05e-05, "loss": 0.0275, "step": 3904 }, { "epoch": 5.11, "learning_rate": 3.0495000000000002e-05, "loss": 0.0405, "step": 3905 }, { "epoch": 5.11, "learning_rate": 3.049e-05, "loss": 0.0299, "step": 3906 }, { "epoch": 5.11, "learning_rate": 3.0485000000000004e-05, "loss": 0.0282, "step": 3907 }, { "epoch": 5.12, "learning_rate": 3.0480000000000003e-05, "loss": 0.035, "step": 3908 }, { "epoch": 5.12, "learning_rate": 3.0475000000000002e-05, "loss": 0.0302, "step": 3909 }, { "epoch": 5.12, "learning_rate": 3.0470000000000005e-05, "loss": 0.0322, "step": 3910 }, { "epoch": 5.12, "learning_rate": 3.0465e-05, "loss": 0.0473, "step": 3911 }, { "epoch": 5.12, "learning_rate": 3.046e-05, "loss": 0.028, "step": 3912 }, { "epoch": 5.12, "learning_rate": 3.0455e-05, "loss": 0.0325, "step": 3913 }, { "epoch": 5.12, "learning_rate": 3.045e-05, "loss": 0.0305, "step": 3914 }, { "epoch": 5.12, "learning_rate": 3.0445e-05, "loss": 0.0313, "step": 3915 }, { "epoch": 5.13, "learning_rate": 3.0440000000000003e-05, "loss": 0.0324, "step": 3916 }, { "epoch": 5.13, "learning_rate": 3.0435000000000003e-05, "loss": 0.0383, "step": 3917 }, { "epoch": 5.13, "learning_rate": 3.0430000000000002e-05, "loss": 0.0317, "step": 3918 }, { "epoch": 5.13, "learning_rate": 3.0425000000000004e-05, "loss": 0.0341, "step": 3919 }, { "epoch": 5.13, "learning_rate": 3.0420000000000004e-05, "loss": 0.0336, "step": 3920 }, { "epoch": 5.13, "learning_rate": 3.0415e-05, "loss": 0.034, "step": 3921 }, { "epoch": 5.13, "learning_rate": 3.041e-05, "loss": 0.0247, "step": 3922 }, { "epoch": 5.13, "learning_rate": 3.041e-05, "loss": 0.1777, "step": 3923 }, { "epoch": 5.14, "learning_rate": 3.0405e-05, "loss": 0.0256, "step": 3924 }, { "epoch": 5.14, "learning_rate": 3.04e-05, "loss": 0.0359, "step": 3925 }, { "epoch": 5.14, "learning_rate": 3.0395000000000003e-05, "loss": 0.0384, "step": 3926 }, { "epoch": 5.14, "learning_rate": 3.0390000000000002e-05, "loss": 0.0313, "step": 3927 }, { "epoch": 5.14, "learning_rate": 3.0385e-05, "loss": 0.0339, "step": 3928 }, { "epoch": 5.14, "learning_rate": 3.0380000000000004e-05, "loss": 0.0405, "step": 3929 }, { "epoch": 5.14, "learning_rate": 3.0375000000000003e-05, "loss": 0.0303, "step": 3930 }, { "epoch": 5.15, "learning_rate": 3.0370000000000006e-05, "loss": 0.0386, "step": 3931 }, { "epoch": 5.15, "learning_rate": 3.0364999999999998e-05, "loss": 0.0363, "step": 3932 }, { "epoch": 5.15, "learning_rate": 3.036e-05, "loss": 0.0279, "step": 3933 }, { "epoch": 5.15, "learning_rate": 3.0355e-05, "loss": 0.0328, "step": 3934 }, { "epoch": 5.15, "learning_rate": 3.035e-05, "loss": 0.0374, "step": 3935 }, { "epoch": 5.15, "learning_rate": 3.0345e-05, "loss": 0.0372, "step": 3936 }, { "epoch": 5.15, "learning_rate": 3.034e-05, "loss": 0.0269, "step": 3937 }, { "epoch": 5.15, "learning_rate": 3.0335000000000003e-05, "loss": 0.0276, "step": 3938 }, { "epoch": 5.16, "learning_rate": 3.0330000000000003e-05, "loss": 0.0312, "step": 3939 }, { "epoch": 5.16, "learning_rate": 3.0325000000000002e-05, "loss": 0.035, "step": 3940 }, { "epoch": 5.16, "learning_rate": 3.0320000000000004e-05, "loss": 0.0426, "step": 3941 }, { "epoch": 5.16, "learning_rate": 3.0315e-05, "loss": 0.025, "step": 3942 }, { "epoch": 5.16, "learning_rate": 3.031e-05, "loss": 0.0238, "step": 3943 }, { "epoch": 5.16, "learning_rate": 3.0305e-05, "loss": 0.0355, "step": 3944 }, { "epoch": 5.16, "learning_rate": 3.03e-05, "loss": 0.0329, "step": 3945 }, { "epoch": 5.16, "learning_rate": 3.0295e-05, "loss": 0.0291, "step": 3946 }, { "epoch": 5.17, "learning_rate": 3.0290000000000003e-05, "loss": 0.0328, "step": 3947 }, { "epoch": 5.17, "learning_rate": 3.0285000000000002e-05, "loss": 0.0343, "step": 3948 }, { "epoch": 5.17, "learning_rate": 3.028e-05, "loss": 0.0285, "step": 3949 }, { "epoch": 5.17, "learning_rate": 3.0275000000000004e-05, "loss": 0.0331, "step": 3950 }, { "epoch": 5.17, "learning_rate": 3.0270000000000003e-05, "loss": 0.0279, "step": 3951 }, { "epoch": 5.17, "learning_rate": 3.0265e-05, "loss": 0.0309, "step": 3952 }, { "epoch": 5.17, "learning_rate": 3.0259999999999998e-05, "loss": 0.0344, "step": 3953 }, { "epoch": 5.18, "learning_rate": 3.0255e-05, "loss": 0.0361, "step": 3954 }, { "epoch": 5.18, "learning_rate": 3.025e-05, "loss": 0.0311, "step": 3955 }, { "epoch": 5.18, "learning_rate": 3.0245000000000003e-05, "loss": 0.0329, "step": 3956 }, { "epoch": 5.18, "learning_rate": 3.0240000000000002e-05, "loss": 0.031, "step": 3957 }, { "epoch": 5.18, "learning_rate": 3.0235e-05, "loss": 0.0273, "step": 3958 }, { "epoch": 5.18, "learning_rate": 3.0230000000000004e-05, "loss": 0.0367, "step": 3959 }, { "epoch": 5.18, "learning_rate": 3.0225000000000003e-05, "loss": 0.0333, "step": 3960 }, { "epoch": 5.18, "learning_rate": 3.0220000000000005e-05, "loss": 0.0316, "step": 3961 }, { "epoch": 5.19, "learning_rate": 3.0214999999999998e-05, "loss": 0.0313, "step": 3962 }, { "epoch": 5.19, "learning_rate": 3.021e-05, "loss": 0.0378, "step": 3963 }, { "epoch": 5.19, "learning_rate": 3.0205e-05, "loss": 0.0385, "step": 3964 }, { "epoch": 5.19, "learning_rate": 3.02e-05, "loss": 0.0333, "step": 3965 }, { "epoch": 5.19, "learning_rate": 3.0195e-05, "loss": 0.028, "step": 3966 }, { "epoch": 5.19, "learning_rate": 3.019e-05, "loss": 0.0388, "step": 3967 }, { "epoch": 5.19, "learning_rate": 3.0185000000000003e-05, "loss": 0.0428, "step": 3968 }, { "epoch": 5.2, "learning_rate": 3.0180000000000002e-05, "loss": 0.0228, "step": 3969 }, { "epoch": 5.2, "learning_rate": 3.0175e-05, "loss": 0.0365, "step": 3970 }, { "epoch": 5.2, "learning_rate": 3.0170000000000004e-05, "loss": 0.0356, "step": 3971 }, { "epoch": 5.2, "learning_rate": 3.0165e-05, "loss": 0.0359, "step": 3972 }, { "epoch": 5.2, "learning_rate": 3.016e-05, "loss": 0.0334, "step": 3973 }, { "epoch": 5.2, "learning_rate": 3.0155e-05, "loss": 0.0297, "step": 3974 }, { "epoch": 5.2, "learning_rate": 3.015e-05, "loss": 0.034, "step": 3975 }, { "epoch": 5.2, "learning_rate": 3.0145e-05, "loss": 0.0313, "step": 3976 }, { "epoch": 5.21, "learning_rate": 3.0140000000000003e-05, "loss": 0.0211, "step": 3977 }, { "epoch": 5.21, "learning_rate": 3.0135000000000002e-05, "loss": 0.0285, "step": 3978 }, { "epoch": 5.21, "learning_rate": 3.013e-05, "loss": 0.0323, "step": 3979 }, { "epoch": 5.21, "learning_rate": 3.0125000000000004e-05, "loss": 0.032, "step": 3980 }, { "epoch": 5.21, "learning_rate": 3.0120000000000003e-05, "loss": 0.0337, "step": 3981 }, { "epoch": 5.21, "learning_rate": 3.0115e-05, "loss": 0.0319, "step": 3982 }, { "epoch": 5.21, "learning_rate": 3.0109999999999998e-05, "loss": 0.0305, "step": 3983 }, { "epoch": 5.21, "learning_rate": 3.0105e-05, "loss": 0.0259, "step": 3984 }, { "epoch": 5.22, "learning_rate": 3.01e-05, "loss": 0.0389, "step": 3985 }, { "epoch": 5.22, "learning_rate": 3.0095000000000002e-05, "loss": 0.0304, "step": 3986 }, { "epoch": 5.22, "learning_rate": 3.009e-05, "loss": 0.0304, "step": 3987 }, { "epoch": 5.22, "learning_rate": 3.0085e-05, "loss": 0.0288, "step": 3988 }, { "epoch": 5.22, "learning_rate": 3.0080000000000003e-05, "loss": 0.0287, "step": 3989 }, { "epoch": 5.22, "learning_rate": 3.0075000000000003e-05, "loss": 0.0362, "step": 3990 }, { "epoch": 5.22, "learning_rate": 3.0070000000000005e-05, "loss": 0.0304, "step": 3991 }, { "epoch": 5.23, "learning_rate": 3.0064999999999998e-05, "loss": 0.0293, "step": 3992 }, { "epoch": 5.23, "learning_rate": 3.006e-05, "loss": 0.0271, "step": 3993 }, { "epoch": 5.23, "learning_rate": 3.0055e-05, "loss": 0.0277, "step": 3994 }, { "epoch": 5.23, "learning_rate": 3.0050000000000002e-05, "loss": 0.0395, "step": 3995 }, { "epoch": 5.23, "learning_rate": 3.0045e-05, "loss": 0.0311, "step": 3996 }, { "epoch": 5.23, "learning_rate": 3.004e-05, "loss": 0.0306, "step": 3997 }, { "epoch": 5.23, "learning_rate": 3.0035000000000003e-05, "loss": 0.0348, "step": 3998 }, { "epoch": 5.23, "learning_rate": 3.0030000000000002e-05, "loss": 0.0312, "step": 3999 }, { "epoch": 5.24, "learning_rate": 3.0025000000000005e-05, "loss": 0.0378, "step": 4000 }, { "epoch": 5.24, "learning_rate": 3.0020000000000004e-05, "loss": 0.0359, "step": 4001 }, { "epoch": 5.24, "learning_rate": 3.0015e-05, "loss": 0.029, "step": 4002 }, { "epoch": 5.24, "learning_rate": 3.001e-05, "loss": 0.0344, "step": 4003 }, { "epoch": 5.24, "learning_rate": 3.0004999999999998e-05, "loss": 0.0363, "step": 4004 }, { "epoch": 5.24, "learning_rate": 3e-05, "loss": 0.0331, "step": 4005 }, { "epoch": 5.24, "learning_rate": 2.9995e-05, "loss": 0.0425, "step": 4006 }, { "epoch": 5.24, "learning_rate": 2.9990000000000003e-05, "loss": 0.0308, "step": 4007 }, { "epoch": 5.25, "learning_rate": 2.9985000000000002e-05, "loss": 0.031, "step": 4008 }, { "epoch": 5.25, "learning_rate": 2.998e-05, "loss": 0.034, "step": 4009 }, { "epoch": 5.25, "learning_rate": 2.9975000000000004e-05, "loss": 0.0233, "step": 4010 }, { "epoch": 5.25, "learning_rate": 2.9970000000000003e-05, "loss": 0.0351, "step": 4011 }, { "epoch": 5.25, "learning_rate": 2.9965000000000005e-05, "loss": 0.033, "step": 4012 }, { "epoch": 5.25, "learning_rate": 2.9959999999999998e-05, "loss": 0.0357, "step": 4013 }, { "epoch": 5.25, "learning_rate": 2.9955e-05, "loss": 0.033, "step": 4014 }, { "epoch": 5.26, "learning_rate": 2.995e-05, "loss": 0.0342, "step": 4015 }, { "epoch": 5.26, "learning_rate": 2.9945000000000002e-05, "loss": 0.0272, "step": 4016 }, { "epoch": 5.26, "learning_rate": 2.994e-05, "loss": 0.032, "step": 4017 }, { "epoch": 5.26, "learning_rate": 2.9935e-05, "loss": 0.0333, "step": 4018 }, { "epoch": 5.26, "learning_rate": 2.9930000000000003e-05, "loss": 0.0347, "step": 4019 }, { "epoch": 5.26, "learning_rate": 2.9925000000000002e-05, "loss": 0.0291, "step": 4020 }, { "epoch": 5.26, "learning_rate": 2.9920000000000005e-05, "loss": 0.0356, "step": 4021 }, { "epoch": 5.26, "learning_rate": 2.9915000000000004e-05, "loss": 0.0361, "step": 4022 }, { "epoch": 5.27, "learning_rate": 2.991e-05, "loss": 0.0321, "step": 4023 }, { "epoch": 5.27, "learning_rate": 2.9905e-05, "loss": 0.0344, "step": 4024 }, { "epoch": 5.27, "learning_rate": 2.9900000000000002e-05, "loss": 0.0275, "step": 4025 }, { "epoch": 5.27, "learning_rate": 2.9895e-05, "loss": 0.0313, "step": 4026 }, { "epoch": 5.27, "learning_rate": 2.989e-05, "loss": 0.0324, "step": 4027 }, { "epoch": 5.27, "learning_rate": 2.9885000000000003e-05, "loss": 0.0306, "step": 4028 }, { "epoch": 5.27, "learning_rate": 2.9880000000000002e-05, "loss": 0.0294, "step": 4029 }, { "epoch": 5.27, "learning_rate": 2.9875000000000004e-05, "loss": 0.0317, "step": 4030 }, { "epoch": 5.28, "learning_rate": 2.9870000000000004e-05, "loss": 0.0315, "step": 4031 }, { "epoch": 5.28, "learning_rate": 2.9865000000000003e-05, "loss": 0.0276, "step": 4032 }, { "epoch": 5.28, "learning_rate": 2.986e-05, "loss": 0.034, "step": 4033 }, { "epoch": 5.28, "learning_rate": 2.9855e-05, "loss": 0.0399, "step": 4034 }, { "epoch": 5.28, "learning_rate": 2.985e-05, "loss": 0.0364, "step": 4035 }, { "epoch": 5.28, "learning_rate": 2.9845e-05, "loss": 0.0344, "step": 4036 }, { "epoch": 5.28, "learning_rate": 2.9840000000000002e-05, "loss": 0.0289, "step": 4037 }, { "epoch": 5.29, "learning_rate": 2.9835e-05, "loss": 0.0345, "step": 4038 }, { "epoch": 5.29, "learning_rate": 2.9830000000000004e-05, "loss": 0.029, "step": 4039 }, { "epoch": 5.29, "learning_rate": 2.9825000000000003e-05, "loss": 0.031, "step": 4040 }, { "epoch": 5.29, "learning_rate": 2.9820000000000002e-05, "loss": 0.0323, "step": 4041 }, { "epoch": 5.29, "learning_rate": 2.9815000000000005e-05, "loss": 0.0371, "step": 4042 }, { "epoch": 5.29, "learning_rate": 2.9809999999999997e-05, "loss": 0.0293, "step": 4043 }, { "epoch": 5.29, "learning_rate": 2.9805e-05, "loss": 0.026, "step": 4044 }, { "epoch": 5.29, "learning_rate": 2.98e-05, "loss": 0.0396, "step": 4045 }, { "epoch": 5.3, "learning_rate": 2.9795000000000002e-05, "loss": 0.0381, "step": 4046 }, { "epoch": 5.3, "learning_rate": 2.979e-05, "loss": 0.0383, "step": 4047 }, { "epoch": 5.3, "learning_rate": 2.9785e-05, "loss": 0.0324, "step": 4048 }, { "epoch": 5.3, "learning_rate": 2.9780000000000003e-05, "loss": 0.038, "step": 4049 }, { "epoch": 5.3, "learning_rate": 2.9775000000000002e-05, "loss": 0.0336, "step": 4050 }, { "epoch": 5.3, "learning_rate": 2.9770000000000005e-05, "loss": 0.0335, "step": 4051 }, { "epoch": 5.3, "learning_rate": 2.9765000000000004e-05, "loss": 0.0355, "step": 4052 }, { "epoch": 5.3, "learning_rate": 2.976e-05, "loss": 0.033, "step": 4053 }, { "epoch": 5.31, "learning_rate": 2.9755e-05, "loss": 0.0289, "step": 4054 }, { "epoch": 5.31, "learning_rate": 2.975e-05, "loss": 0.0403, "step": 4055 }, { "epoch": 5.31, "learning_rate": 2.9745e-05, "loss": 0.0289, "step": 4056 }, { "epoch": 5.31, "learning_rate": 2.974e-05, "loss": 0.0364, "step": 4057 }, { "epoch": 5.31, "learning_rate": 2.9735000000000002e-05, "loss": 0.0334, "step": 4058 }, { "epoch": 5.31, "learning_rate": 2.973e-05, "loss": 0.0317, "step": 4059 }, { "epoch": 5.31, "learning_rate": 2.9725000000000004e-05, "loss": 0.0369, "step": 4060 }, { "epoch": 5.32, "learning_rate": 2.9720000000000003e-05, "loss": 0.0332, "step": 4061 }, { "epoch": 5.32, "learning_rate": 2.9715000000000003e-05, "loss": 0.0259, "step": 4062 }, { "epoch": 5.32, "learning_rate": 2.971e-05, "loss": 0.0311, "step": 4063 }, { "epoch": 5.32, "learning_rate": 2.9705e-05, "loss": 0.0323, "step": 4064 }, { "epoch": 5.32, "learning_rate": 2.97e-05, "loss": 0.0263, "step": 4065 }, { "epoch": 5.32, "learning_rate": 2.9695e-05, "loss": 0.0356, "step": 4066 }, { "epoch": 5.32, "learning_rate": 2.9690000000000002e-05, "loss": 0.0354, "step": 4067 }, { "epoch": 5.32, "learning_rate": 2.9685e-05, "loss": 0.0364, "step": 4068 }, { "epoch": 5.33, "learning_rate": 2.9680000000000004e-05, "loss": 0.0362, "step": 4069 }, { "epoch": 5.33, "learning_rate": 2.9675000000000003e-05, "loss": 0.0379, "step": 4070 }, { "epoch": 5.33, "learning_rate": 2.9670000000000002e-05, "loss": 0.0278, "step": 4071 }, { "epoch": 5.33, "learning_rate": 2.9665000000000005e-05, "loss": 0.0261, "step": 4072 }, { "epoch": 5.33, "learning_rate": 2.9659999999999997e-05, "loss": 0.0299, "step": 4073 }, { "epoch": 5.33, "learning_rate": 2.9655e-05, "loss": 0.0275, "step": 4074 }, { "epoch": 5.33, "learning_rate": 2.965e-05, "loss": 0.0327, "step": 4075 }, { "epoch": 5.34, "learning_rate": 2.9645e-05, "loss": 0.0323, "step": 4076 }, { "epoch": 5.34, "learning_rate": 2.964e-05, "loss": 0.032, "step": 4077 }, { "epoch": 5.34, "learning_rate": 2.9635e-05, "loss": 0.0288, "step": 4078 }, { "epoch": 5.34, "learning_rate": 2.9630000000000003e-05, "loss": 0.033, "step": 4079 }, { "epoch": 5.34, "learning_rate": 2.9625000000000002e-05, "loss": 0.0349, "step": 4080 }, { "epoch": 5.34, "learning_rate": 2.9620000000000004e-05, "loss": 0.0324, "step": 4081 }, { "epoch": 5.34, "learning_rate": 2.9615000000000004e-05, "loss": 0.0344, "step": 4082 }, { "epoch": 5.34, "learning_rate": 2.961e-05, "loss": 0.0424, "step": 4083 }, { "epoch": 5.35, "learning_rate": 2.9605e-05, "loss": 0.0359, "step": 4084 }, { "epoch": 5.35, "learning_rate": 2.96e-05, "loss": 0.0304, "step": 4085 }, { "epoch": 5.35, "learning_rate": 2.9595e-05, "loss": 0.0322, "step": 4086 }, { "epoch": 5.35, "learning_rate": 2.959e-05, "loss": 0.0331, "step": 4087 }, { "epoch": 5.35, "learning_rate": 2.9585000000000002e-05, "loss": 0.0292, "step": 4088 }, { "epoch": 5.35, "learning_rate": 2.958e-05, "loss": 0.037, "step": 4089 }, { "epoch": 5.35, "learning_rate": 2.9575000000000004e-05, "loss": 0.0308, "step": 4090 }, { "epoch": 5.35, "learning_rate": 2.9570000000000003e-05, "loss": 0.0301, "step": 4091 }, { "epoch": 5.36, "learning_rate": 2.9565000000000002e-05, "loss": 0.0325, "step": 4092 }, { "epoch": 5.36, "learning_rate": 2.9559999999999998e-05, "loss": 0.032, "step": 4093 }, { "epoch": 5.36, "learning_rate": 2.9555e-05, "loss": 0.0302, "step": 4094 }, { "epoch": 5.36, "learning_rate": 2.955e-05, "loss": 0.0354, "step": 4095 }, { "epoch": 5.36, "learning_rate": 2.9545e-05, "loss": 0.0293, "step": 4096 }, { "epoch": 5.36, "learning_rate": 2.9540000000000002e-05, "loss": 0.0288, "step": 4097 }, { "epoch": 5.36, "learning_rate": 2.9535e-05, "loss": 0.0249, "step": 4098 }, { "epoch": 5.37, "learning_rate": 2.9530000000000004e-05, "loss": 0.0398, "step": 4099 }, { "epoch": 5.37, "learning_rate": 2.9525000000000003e-05, "loss": 0.032, "step": 4100 }, { "epoch": 5.37, "learning_rate": 2.9520000000000002e-05, "loss": 0.034, "step": 4101 }, { "epoch": 5.37, "learning_rate": 2.9515000000000005e-05, "loss": 0.0335, "step": 4102 }, { "epoch": 5.37, "learning_rate": 2.951e-05, "loss": 0.0407, "step": 4103 }, { "epoch": 5.37, "learning_rate": 2.9505e-05, "loss": 0.032, "step": 4104 }, { "epoch": 5.37, "learning_rate": 2.95e-05, "loss": 0.037, "step": 4105 }, { "epoch": 5.37, "learning_rate": 2.9495e-05, "loss": 0.0284, "step": 4106 }, { "epoch": 5.38, "learning_rate": 2.949e-05, "loss": 0.0305, "step": 4107 }, { "epoch": 5.38, "learning_rate": 2.9485000000000003e-05, "loss": 0.0311, "step": 4108 }, { "epoch": 5.38, "learning_rate": 2.9480000000000002e-05, "loss": 0.0386, "step": 4109 }, { "epoch": 5.38, "learning_rate": 2.9475e-05, "loss": 0.036, "step": 4110 }, { "epoch": 5.38, "learning_rate": 2.9470000000000004e-05, "loss": 0.0356, "step": 4111 }, { "epoch": 5.38, "learning_rate": 2.9465000000000003e-05, "loss": 0.038, "step": 4112 }, { "epoch": 5.38, "learning_rate": 2.946e-05, "loss": 0.029, "step": 4113 }, { "epoch": 5.38, "learning_rate": 2.9455e-05, "loss": 0.0296, "step": 4114 }, { "epoch": 5.39, "learning_rate": 2.945e-05, "loss": 0.0324, "step": 4115 }, { "epoch": 5.39, "learning_rate": 2.9445e-05, "loss": 0.0347, "step": 4116 }, { "epoch": 5.39, "learning_rate": 2.944e-05, "loss": 0.0378, "step": 4117 }, { "epoch": 5.39, "learning_rate": 2.9435000000000002e-05, "loss": 0.0354, "step": 4118 }, { "epoch": 5.39, "learning_rate": 2.943e-05, "loss": 0.0338, "step": 4119 }, { "epoch": 5.39, "learning_rate": 2.9425000000000004e-05, "loss": 0.0352, "step": 4120 }, { "epoch": 5.39, "learning_rate": 2.9420000000000003e-05, "loss": 0.0351, "step": 4121 }, { "epoch": 5.4, "learning_rate": 2.9415000000000002e-05, "loss": 0.0313, "step": 4122 }, { "epoch": 5.4, "learning_rate": 2.9409999999999998e-05, "loss": 0.0261, "step": 4123 }, { "epoch": 5.4, "learning_rate": 2.9405e-05, "loss": 0.0332, "step": 4124 }, { "epoch": 5.4, "learning_rate": 2.94e-05, "loss": 0.0332, "step": 4125 }, { "epoch": 5.4, "learning_rate": 2.9395e-05, "loss": 0.0406, "step": 4126 }, { "epoch": 5.4, "learning_rate": 2.939e-05, "loss": 0.0393, "step": 4127 }, { "epoch": 5.4, "learning_rate": 2.9385e-05, "loss": 0.036, "step": 4128 }, { "epoch": 5.4, "learning_rate": 2.9380000000000003e-05, "loss": 0.0286, "step": 4129 }, { "epoch": 5.41, "learning_rate": 2.9375000000000003e-05, "loss": 0.0339, "step": 4130 }, { "epoch": 5.41, "learning_rate": 2.9370000000000002e-05, "loss": 0.038, "step": 4131 }, { "epoch": 5.41, "learning_rate": 2.9365000000000004e-05, "loss": 0.0343, "step": 4132 }, { "epoch": 5.41, "learning_rate": 2.9360000000000003e-05, "loss": 0.0299, "step": 4133 }, { "epoch": 5.41, "learning_rate": 2.9355e-05, "loss": 0.0367, "step": 4134 }, { "epoch": 5.41, "learning_rate": 2.935e-05, "loss": 0.0405, "step": 4135 }, { "epoch": 5.41, "learning_rate": 2.9345e-05, "loss": 0.031, "step": 4136 }, { "epoch": 5.41, "learning_rate": 2.934e-05, "loss": 0.0284, "step": 4137 }, { "epoch": 5.42, "learning_rate": 2.9335000000000003e-05, "loss": 0.0345, "step": 4138 }, { "epoch": 5.42, "learning_rate": 2.9330000000000002e-05, "loss": 0.0343, "step": 4139 }, { "epoch": 5.42, "learning_rate": 2.9325e-05, "loss": 0.0343, "step": 4140 }, { "epoch": 5.42, "learning_rate": 2.9320000000000004e-05, "loss": 0.0292, "step": 4141 }, { "epoch": 5.42, "learning_rate": 2.9315000000000003e-05, "loss": 0.032, "step": 4142 }, { "epoch": 5.42, "learning_rate": 2.9310000000000006e-05, "loss": 0.0372, "step": 4143 }, { "epoch": 5.42, "learning_rate": 2.9304999999999998e-05, "loss": 0.033, "step": 4144 }, { "epoch": 5.43, "learning_rate": 2.93e-05, "loss": 0.0331, "step": 4145 }, { "epoch": 5.43, "learning_rate": 2.9295e-05, "loss": 0.0334, "step": 4146 }, { "epoch": 5.43, "learning_rate": 2.929e-05, "loss": 0.0388, "step": 4147 }, { "epoch": 5.43, "learning_rate": 2.9285e-05, "loss": 0.0282, "step": 4148 }, { "epoch": 5.43, "learning_rate": 2.928e-05, "loss": 0.0364, "step": 4149 }, { "epoch": 5.43, "learning_rate": 2.9275000000000003e-05, "loss": 0.0309, "step": 4150 }, { "epoch": 5.43, "learning_rate": 2.9270000000000003e-05, "loss": 0.0321, "step": 4151 }, { "epoch": 5.43, "learning_rate": 2.9265000000000002e-05, "loss": 0.0391, "step": 4152 }, { "epoch": 5.44, "learning_rate": 2.9260000000000004e-05, "loss": 0.0404, "step": 4153 }, { "epoch": 5.44, "learning_rate": 2.9255e-05, "loss": 0.0426, "step": 4154 }, { "epoch": 5.44, "learning_rate": 2.925e-05, "loss": 0.0299, "step": 4155 }, { "epoch": 5.44, "learning_rate": 2.9245e-05, "loss": 0.025, "step": 4156 }, { "epoch": 5.44, "learning_rate": 2.924e-05, "loss": 0.0367, "step": 4157 }, { "epoch": 5.44, "learning_rate": 2.9235e-05, "loss": 0.0342, "step": 4158 }, { "epoch": 5.44, "learning_rate": 2.9230000000000003e-05, "loss": 0.0302, "step": 4159 }, { "epoch": 5.45, "learning_rate": 2.9225000000000002e-05, "loss": 0.0392, "step": 4160 }, { "epoch": 5.45, "learning_rate": 2.922e-05, "loss": 0.0365, "step": 4161 }, { "epoch": 5.45, "learning_rate": 2.9215000000000004e-05, "loss": 0.0337, "step": 4162 }, { "epoch": 5.45, "learning_rate": 2.9210000000000003e-05, "loss": 0.0335, "step": 4163 }, { "epoch": 5.45, "learning_rate": 2.9205e-05, "loss": 0.0205, "step": 4164 }, { "epoch": 5.45, "learning_rate": 2.9199999999999998e-05, "loss": 0.0417, "step": 4165 }, { "epoch": 5.45, "learning_rate": 2.9195e-05, "loss": 0.0354, "step": 4166 }, { "epoch": 5.45, "learning_rate": 2.919e-05, "loss": 0.0378, "step": 4167 }, { "epoch": 5.46, "learning_rate": 2.9185000000000003e-05, "loss": 0.0341, "step": 4168 }, { "epoch": 5.46, "learning_rate": 2.9180000000000002e-05, "loss": 0.0397, "step": 4169 }, { "epoch": 5.46, "learning_rate": 2.9175e-05, "loss": 0.0315, "step": 4170 }, { "epoch": 5.46, "learning_rate": 2.9170000000000004e-05, "loss": 0.0335, "step": 4171 }, { "epoch": 5.46, "learning_rate": 2.9165000000000003e-05, "loss": 0.0279, "step": 4172 }, { "epoch": 5.46, "learning_rate": 2.9160000000000005e-05, "loss": 0.0332, "step": 4173 }, { "epoch": 5.46, "learning_rate": 2.9154999999999998e-05, "loss": 0.027, "step": 4174 }, { "epoch": 5.46, "learning_rate": 2.915e-05, "loss": 0.0234, "step": 4175 }, { "epoch": 5.47, "learning_rate": 2.9145e-05, "loss": 0.0349, "step": 4176 }, { "epoch": 5.47, "learning_rate": 2.9140000000000002e-05, "loss": 0.0342, "step": 4177 }, { "epoch": 5.47, "learning_rate": 2.9135e-05, "loss": 0.0284, "step": 4178 }, { "epoch": 5.47, "learning_rate": 2.913e-05, "loss": 0.033, "step": 4179 }, { "epoch": 5.47, "learning_rate": 2.9125000000000003e-05, "loss": 0.029, "step": 4180 }, { "epoch": 5.47, "learning_rate": 2.9120000000000002e-05, "loss": 0.0326, "step": 4181 }, { "epoch": 5.47, "learning_rate": 2.9115000000000005e-05, "loss": 0.034, "step": 4182 }, { "epoch": 5.48, "learning_rate": 2.9110000000000004e-05, "loss": 0.036, "step": 4183 }, { "epoch": 5.48, "learning_rate": 2.9105e-05, "loss": 0.0321, "step": 4184 }, { "epoch": 5.48, "learning_rate": 2.91e-05, "loss": 0.0353, "step": 4185 }, { "epoch": 5.48, "learning_rate": 2.9095e-05, "loss": 0.0292, "step": 4186 }, { "epoch": 5.48, "learning_rate": 2.909e-05, "loss": 0.0317, "step": 4187 }, { "epoch": 5.48, "learning_rate": 2.9085e-05, "loss": 0.0342, "step": 4188 }, { "epoch": 5.48, "learning_rate": 2.9080000000000003e-05, "loss": 0.0277, "step": 4189 }, { "epoch": 5.48, "learning_rate": 2.9075000000000002e-05, "loss": 0.0332, "step": 4190 }, { "epoch": 5.49, "learning_rate": 2.907e-05, "loss": 0.0387, "step": 4191 }, { "epoch": 5.49, "learning_rate": 2.9065000000000004e-05, "loss": 0.0285, "step": 4192 }, { "epoch": 5.49, "learning_rate": 2.9060000000000003e-05, "loss": 0.0348, "step": 4193 }, { "epoch": 5.49, "learning_rate": 2.9055e-05, "loss": 0.0294, "step": 4194 }, { "epoch": 5.49, "learning_rate": 2.9049999999999998e-05, "loss": 0.0413, "step": 4195 }, { "epoch": 5.49, "learning_rate": 2.9045e-05, "loss": 0.0341, "step": 4196 }, { "epoch": 5.49, "learning_rate": 2.904e-05, "loss": 0.0399, "step": 4197 }, { "epoch": 5.49, "learning_rate": 2.9035000000000002e-05, "loss": 0.0373, "step": 4198 }, { "epoch": 5.5, "learning_rate": 2.903e-05, "loss": 0.0307, "step": 4199 }, { "epoch": 5.5, "learning_rate": 2.9025e-05, "loss": 0.038, "step": 4200 }, { "epoch": 5.5, "learning_rate": 2.9020000000000003e-05, "loss": 0.0322, "step": 4201 }, { "epoch": 5.5, "learning_rate": 2.9015000000000003e-05, "loss": 0.0334, "step": 4202 }, { "epoch": 5.5, "learning_rate": 2.9010000000000005e-05, "loss": 0.0406, "step": 4203 }, { "epoch": 5.5, "learning_rate": 2.9004999999999998e-05, "loss": 0.0344, "step": 4204 }, { "epoch": 5.5, "learning_rate": 2.9e-05, "loss": 0.0359, "step": 4205 }, { "epoch": 5.51, "learning_rate": 2.8995e-05, "loss": 0.0307, "step": 4206 }, { "epoch": 5.51, "learning_rate": 2.8990000000000002e-05, "loss": 0.0357, "step": 4207 }, { "epoch": 5.51, "learning_rate": 2.8985e-05, "loss": 0.0345, "step": 4208 }, { "epoch": 5.51, "learning_rate": 2.898e-05, "loss": 0.0321, "step": 4209 }, { "epoch": 5.51, "learning_rate": 2.8975000000000003e-05, "loss": 0.033, "step": 4210 }, { "epoch": 5.51, "learning_rate": 2.8970000000000002e-05, "loss": 0.0428, "step": 4211 }, { "epoch": 5.51, "learning_rate": 2.8965000000000005e-05, "loss": 0.0353, "step": 4212 }, { "epoch": 5.51, "learning_rate": 2.8960000000000004e-05, "loss": 0.0305, "step": 4213 }, { "epoch": 5.52, "learning_rate": 2.8955e-05, "loss": 0.0307, "step": 4214 }, { "epoch": 5.52, "learning_rate": 2.895e-05, "loss": 0.0343, "step": 4215 }, { "epoch": 5.52, "learning_rate": 2.8945e-05, "loss": 0.0344, "step": 4216 }, { "epoch": 5.52, "learning_rate": 2.894e-05, "loss": 0.0312, "step": 4217 }, { "epoch": 5.52, "learning_rate": 2.8935e-05, "loss": 0.0355, "step": 4218 }, { "epoch": 5.52, "learning_rate": 2.8930000000000003e-05, "loss": 0.0362, "step": 4219 }, { "epoch": 5.52, "learning_rate": 2.8925000000000002e-05, "loss": 0.0346, "step": 4220 }, { "epoch": 5.52, "learning_rate": 2.8920000000000004e-05, "loss": 0.0334, "step": 4221 }, { "epoch": 5.53, "learning_rate": 2.8915000000000004e-05, "loss": 0.0346, "step": 4222 }, { "epoch": 5.53, "learning_rate": 2.8910000000000003e-05, "loss": 0.0375, "step": 4223 }, { "epoch": 5.53, "learning_rate": 2.8905e-05, "loss": 0.0358, "step": 4224 }, { "epoch": 5.53, "learning_rate": 2.8899999999999998e-05, "loss": 0.035, "step": 4225 }, { "epoch": 5.53, "learning_rate": 2.8895e-05, "loss": 0.0407, "step": 4226 }, { "epoch": 5.53, "learning_rate": 2.889e-05, "loss": 0.0304, "step": 4227 }, { "epoch": 5.53, "learning_rate": 2.8885000000000002e-05, "loss": 0.034, "step": 4228 }, { "epoch": 5.54, "learning_rate": 2.888e-05, "loss": 0.0342, "step": 4229 }, { "epoch": 5.54, "learning_rate": 2.8875e-05, "loss": 0.0364, "step": 4230 }, { "epoch": 5.54, "learning_rate": 2.8870000000000003e-05, "loss": 0.0393, "step": 4231 }, { "epoch": 5.54, "learning_rate": 2.8865000000000002e-05, "loss": 0.0336, "step": 4232 }, { "epoch": 5.54, "learning_rate": 2.8860000000000005e-05, "loss": 0.0329, "step": 4233 }, { "epoch": 5.54, "learning_rate": 2.8854999999999997e-05, "loss": 0.0316, "step": 4234 }, { "epoch": 5.54, "learning_rate": 2.885e-05, "loss": 0.0326, "step": 4235 }, { "epoch": 5.54, "learning_rate": 2.8845e-05, "loss": 0.0387, "step": 4236 }, { "epoch": 5.55, "learning_rate": 2.8840000000000002e-05, "loss": 0.0353, "step": 4237 }, { "epoch": 5.55, "learning_rate": 2.8835e-05, "loss": 0.036, "step": 4238 }, { "epoch": 5.55, "learning_rate": 2.883e-05, "loss": 0.039, "step": 4239 }, { "epoch": 5.55, "learning_rate": 2.8825000000000003e-05, "loss": 0.0299, "step": 4240 }, { "epoch": 5.55, "learning_rate": 2.8820000000000002e-05, "loss": 0.0413, "step": 4241 }, { "epoch": 5.55, "learning_rate": 2.8815000000000004e-05, "loss": 0.0401, "step": 4242 }, { "epoch": 5.55, "learning_rate": 2.8810000000000004e-05, "loss": 0.0356, "step": 4243 }, { "epoch": 5.55, "learning_rate": 2.8805e-05, "loss": 0.0277, "step": 4244 }, { "epoch": 5.56, "learning_rate": 2.88e-05, "loss": 0.0328, "step": 4245 }, { "epoch": 5.56, "learning_rate": 2.8795e-05, "loss": 0.0359, "step": 4246 }, { "epoch": 5.56, "learning_rate": 2.879e-05, "loss": 0.0288, "step": 4247 }, { "epoch": 5.56, "learning_rate": 2.8785e-05, "loss": 0.0394, "step": 4248 }, { "epoch": 5.56, "learning_rate": 2.8780000000000002e-05, "loss": 0.0385, "step": 4249 }, { "epoch": 5.56, "learning_rate": 2.8775e-05, "loss": 0.0327, "step": 4250 }, { "epoch": 5.56, "learning_rate": 2.8770000000000004e-05, "loss": 0.0317, "step": 4251 }, { "epoch": 5.57, "learning_rate": 2.8765000000000003e-05, "loss": 0.0352, "step": 4252 }, { "epoch": 5.57, "learning_rate": 2.8760000000000002e-05, "loss": 0.0358, "step": 4253 }, { "epoch": 5.57, "learning_rate": 2.8754999999999998e-05, "loss": 0.031, "step": 4254 }, { "epoch": 5.57, "learning_rate": 2.8749999999999997e-05, "loss": 0.037, "step": 4255 }, { "epoch": 5.57, "learning_rate": 2.8745e-05, "loss": 0.0286, "step": 4256 }, { "epoch": 5.57, "learning_rate": 2.874e-05, "loss": 0.0281, "step": 4257 }, { "epoch": 5.57, "learning_rate": 2.8735000000000002e-05, "loss": 0.0336, "step": 4258 }, { "epoch": 5.57, "learning_rate": 2.873e-05, "loss": 0.043, "step": 4259 }, { "epoch": 5.58, "learning_rate": 2.8725e-05, "loss": 0.0465, "step": 4260 }, { "epoch": 5.58, "learning_rate": 2.8720000000000003e-05, "loss": 0.0329, "step": 4261 }, { "epoch": 5.58, "learning_rate": 2.8715000000000002e-05, "loss": 0.0303, "step": 4262 }, { "epoch": 5.58, "learning_rate": 2.8710000000000005e-05, "loss": 0.0317, "step": 4263 }, { "epoch": 5.58, "learning_rate": 2.8705000000000004e-05, "loss": 0.0367, "step": 4264 }, { "epoch": 5.58, "learning_rate": 2.87e-05, "loss": 0.0319, "step": 4265 }, { "epoch": 5.58, "learning_rate": 2.8695e-05, "loss": 0.0357, "step": 4266 }, { "epoch": 5.59, "learning_rate": 2.869e-05, "loss": 0.032, "step": 4267 }, { "epoch": 5.59, "learning_rate": 2.8685e-05, "loss": 0.0362, "step": 4268 }, { "epoch": 5.59, "learning_rate": 2.868e-05, "loss": 0.0404, "step": 4269 }, { "epoch": 5.59, "learning_rate": 2.8675000000000002e-05, "loss": 0.0378, "step": 4270 }, { "epoch": 5.59, "learning_rate": 2.867e-05, "loss": 0.0372, "step": 4271 }, { "epoch": 5.59, "learning_rate": 2.8665000000000004e-05, "loss": 0.0409, "step": 4272 }, { "epoch": 5.59, "learning_rate": 2.8660000000000003e-05, "loss": 0.0411, "step": 4273 }, { "epoch": 5.59, "learning_rate": 2.8655000000000003e-05, "loss": 0.0369, "step": 4274 }, { "epoch": 5.6, "learning_rate": 2.865e-05, "loss": 0.0318, "step": 4275 }, { "epoch": 5.6, "learning_rate": 2.8645e-05, "loss": 0.0324, "step": 4276 }, { "epoch": 5.6, "learning_rate": 2.864e-05, "loss": 0.0362, "step": 4277 }, { "epoch": 5.6, "learning_rate": 2.8635e-05, "loss": 0.0354, "step": 4278 }, { "epoch": 5.6, "learning_rate": 2.8630000000000002e-05, "loss": 0.0293, "step": 4279 }, { "epoch": 5.6, "learning_rate": 2.8625e-05, "loss": 0.0363, "step": 4280 }, { "epoch": 5.6, "learning_rate": 2.8620000000000004e-05, "loss": 0.0366, "step": 4281 }, { "epoch": 5.6, "learning_rate": 2.8615000000000003e-05, "loss": 0.036, "step": 4282 }, { "epoch": 5.61, "learning_rate": 2.8610000000000002e-05, "loss": 0.0347, "step": 4283 }, { "epoch": 5.61, "learning_rate": 2.8605000000000005e-05, "loss": 0.0381, "step": 4284 }, { "epoch": 5.61, "learning_rate": 2.86e-05, "loss": 0.0327, "step": 4285 }, { "epoch": 5.61, "learning_rate": 2.8595e-05, "loss": 0.037, "step": 4286 }, { "epoch": 5.61, "learning_rate": 2.859e-05, "loss": 0.0253, "step": 4287 }, { "epoch": 5.61, "learning_rate": 2.8585e-05, "loss": 0.0393, "step": 4288 }, { "epoch": 5.61, "learning_rate": 2.858e-05, "loss": 0.0318, "step": 4289 }, { "epoch": 5.62, "learning_rate": 2.8575000000000003e-05, "loss": 0.0411, "step": 4290 }, { "epoch": 5.62, "learning_rate": 2.8570000000000003e-05, "loss": 0.0357, "step": 4291 }, { "epoch": 5.62, "learning_rate": 2.8565000000000002e-05, "loss": 0.029, "step": 4292 }, { "epoch": 5.62, "learning_rate": 2.8560000000000004e-05, "loss": 0.0413, "step": 4293 }, { "epoch": 5.62, "learning_rate": 2.8555000000000004e-05, "loss": 0.034, "step": 4294 }, { "epoch": 5.62, "learning_rate": 2.855e-05, "loss": 0.0344, "step": 4295 }, { "epoch": 5.62, "learning_rate": 2.8545e-05, "loss": 0.0287, "step": 4296 }, { "epoch": 5.62, "learning_rate": 2.854e-05, "loss": 0.034, "step": 4297 }, { "epoch": 5.63, "learning_rate": 2.8535e-05, "loss": 0.0391, "step": 4298 }, { "epoch": 5.63, "learning_rate": 2.853e-05, "loss": 0.0314, "step": 4299 }, { "epoch": 5.63, "learning_rate": 2.8525000000000002e-05, "loss": 0.0391, "step": 4300 }, { "epoch": 5.63, "learning_rate": 2.852e-05, "loss": 0.0301, "step": 4301 }, { "epoch": 5.63, "learning_rate": 2.8515000000000004e-05, "loss": 0.0369, "step": 4302 }, { "epoch": 5.63, "learning_rate": 2.8510000000000003e-05, "loss": 0.03, "step": 4303 }, { "epoch": 5.63, "learning_rate": 2.8505000000000002e-05, "loss": 0.0359, "step": 4304 }, { "epoch": 5.63, "learning_rate": 2.8499999999999998e-05, "loss": 0.0374, "step": 4305 }, { "epoch": 5.64, "learning_rate": 2.8495e-05, "loss": 0.0454, "step": 4306 }, { "epoch": 5.64, "learning_rate": 2.849e-05, "loss": 0.03, "step": 4307 }, { "epoch": 5.64, "learning_rate": 2.8485e-05, "loss": 0.0321, "step": 4308 }, { "epoch": 5.64, "learning_rate": 2.8480000000000002e-05, "loss": 0.0303, "step": 4309 }, { "epoch": 5.64, "learning_rate": 2.8475e-05, "loss": 0.0333, "step": 4310 }, { "epoch": 5.64, "learning_rate": 2.8470000000000004e-05, "loss": 0.031, "step": 4311 }, { "epoch": 5.64, "learning_rate": 2.8465000000000003e-05, "loss": 0.037, "step": 4312 }, { "epoch": 5.65, "learning_rate": 2.8460000000000002e-05, "loss": 0.0321, "step": 4313 }, { "epoch": 5.65, "learning_rate": 2.8455000000000005e-05, "loss": 0.0368, "step": 4314 }, { "epoch": 5.65, "learning_rate": 2.845e-05, "loss": 0.0347, "step": 4315 }, { "epoch": 5.65, "learning_rate": 2.8445e-05, "loss": 0.0386, "step": 4316 }, { "epoch": 5.65, "learning_rate": 2.844e-05, "loss": 0.0474, "step": 4317 }, { "epoch": 5.65, "learning_rate": 2.8435e-05, "loss": 0.035, "step": 4318 }, { "epoch": 5.65, "learning_rate": 2.843e-05, "loss": 0.0376, "step": 4319 }, { "epoch": 5.65, "learning_rate": 2.8425000000000003e-05, "loss": 0.0295, "step": 4320 }, { "epoch": 5.66, "learning_rate": 2.8420000000000002e-05, "loss": 0.0356, "step": 4321 }, { "epoch": 5.66, "learning_rate": 2.8415e-05, "loss": 0.0456, "step": 4322 }, { "epoch": 5.66, "learning_rate": 2.8410000000000004e-05, "loss": 0.0373, "step": 4323 }, { "epoch": 5.66, "learning_rate": 2.8405000000000003e-05, "loss": 0.0303, "step": 4324 }, { "epoch": 5.66, "learning_rate": 2.84e-05, "loss": 0.0296, "step": 4325 }, { "epoch": 5.66, "learning_rate": 2.8395e-05, "loss": 0.0305, "step": 4326 }, { "epoch": 5.66, "learning_rate": 2.839e-05, "loss": 0.0294, "step": 4327 }, { "epoch": 5.66, "learning_rate": 2.8385e-05, "loss": 0.0328, "step": 4328 }, { "epoch": 5.67, "learning_rate": 2.8380000000000003e-05, "loss": 0.0418, "step": 4329 }, { "epoch": 5.67, "learning_rate": 2.8375000000000002e-05, "loss": 0.035, "step": 4330 }, { "epoch": 5.67, "learning_rate": 2.837e-05, "loss": 0.029, "step": 4331 }, { "epoch": 5.67, "learning_rate": 2.8365000000000004e-05, "loss": 0.0401, "step": 4332 }, { "epoch": 5.67, "learning_rate": 2.8360000000000003e-05, "loss": 0.0372, "step": 4333 }, { "epoch": 5.67, "learning_rate": 2.8355000000000002e-05, "loss": 0.042, "step": 4334 }, { "epoch": 5.67, "learning_rate": 2.8349999999999998e-05, "loss": 0.0319, "step": 4335 }, { "epoch": 5.68, "learning_rate": 2.8345e-05, "loss": 0.0316, "step": 4336 }, { "epoch": 5.68, "learning_rate": 2.834e-05, "loss": 0.0373, "step": 4337 }, { "epoch": 5.68, "learning_rate": 2.8335e-05, "loss": 0.0374, "step": 4338 }, { "epoch": 5.68, "learning_rate": 2.833e-05, "loss": 0.0363, "step": 4339 }, { "epoch": 5.68, "learning_rate": 2.8325e-05, "loss": 0.0365, "step": 4340 }, { "epoch": 5.68, "learning_rate": 2.8320000000000003e-05, "loss": 0.0406, "step": 4341 }, { "epoch": 5.68, "learning_rate": 2.8315000000000002e-05, "loss": 0.0433, "step": 4342 }, { "epoch": 5.68, "learning_rate": 2.8310000000000002e-05, "loss": 0.038, "step": 4343 }, { "epoch": 5.69, "learning_rate": 2.8305000000000004e-05, "loss": 0.0368, "step": 4344 }, { "epoch": 5.69, "learning_rate": 2.83e-05, "loss": 0.0322, "step": 4345 }, { "epoch": 5.69, "learning_rate": 2.8295e-05, "loss": 0.0394, "step": 4346 }, { "epoch": 5.69, "learning_rate": 2.829e-05, "loss": 0.0364, "step": 4347 }, { "epoch": 5.69, "learning_rate": 2.8285e-05, "loss": 0.0385, "step": 4348 }, { "epoch": 5.69, "learning_rate": 2.828e-05, "loss": 0.0403, "step": 4349 }, { "epoch": 5.69, "learning_rate": 2.8275000000000003e-05, "loss": 0.0353, "step": 4350 }, { "epoch": 5.7, "learning_rate": 2.8270000000000002e-05, "loss": 0.0378, "step": 4351 }, { "epoch": 5.7, "learning_rate": 2.8265e-05, "loss": 0.0355, "step": 4352 }, { "epoch": 5.7, "learning_rate": 2.8260000000000004e-05, "loss": 0.0275, "step": 4353 }, { "epoch": 5.7, "learning_rate": 2.8255000000000003e-05, "loss": 0.0382, "step": 4354 }, { "epoch": 5.7, "learning_rate": 2.825e-05, "loss": 0.0342, "step": 4355 }, { "epoch": 5.7, "learning_rate": 2.8244999999999998e-05, "loss": 0.0414, "step": 4356 }, { "epoch": 5.7, "learning_rate": 2.824e-05, "loss": 0.0412, "step": 4357 }, { "epoch": 5.7, "learning_rate": 2.8235e-05, "loss": 0.0374, "step": 4358 }, { "epoch": 5.71, "learning_rate": 2.8230000000000002e-05, "loss": 0.0424, "step": 4359 }, { "epoch": 5.71, "learning_rate": 2.8225e-05, "loss": 0.0361, "step": 4360 }, { "epoch": 5.71, "learning_rate": 2.822e-05, "loss": 0.0306, "step": 4361 }, { "epoch": 5.71, "learning_rate": 2.8215000000000003e-05, "loss": 0.0332, "step": 4362 }, { "epoch": 5.71, "learning_rate": 2.8210000000000003e-05, "loss": 0.0421, "step": 4363 }, { "epoch": 5.71, "learning_rate": 2.8205000000000005e-05, "loss": 0.0373, "step": 4364 }, { "epoch": 5.71, "learning_rate": 2.8199999999999998e-05, "loss": 0.0372, "step": 4365 }, { "epoch": 5.71, "learning_rate": 2.8195e-05, "loss": 0.0363, "step": 4366 }, { "epoch": 5.72, "learning_rate": 2.819e-05, "loss": 0.041, "step": 4367 }, { "epoch": 5.72, "learning_rate": 2.8185e-05, "loss": 0.0359, "step": 4368 }, { "epoch": 5.72, "learning_rate": 2.818e-05, "loss": 0.0364, "step": 4369 }, { "epoch": 5.72, "learning_rate": 2.8175e-05, "loss": 0.0384, "step": 4370 }, { "epoch": 5.72, "learning_rate": 2.8170000000000003e-05, "loss": 0.0322, "step": 4371 }, { "epoch": 5.72, "learning_rate": 2.8165000000000002e-05, "loss": 0.0407, "step": 4372 }, { "epoch": 5.72, "learning_rate": 2.816e-05, "loss": 0.04, "step": 4373 }, { "epoch": 5.73, "learning_rate": 2.8155000000000004e-05, "loss": 0.0272, "step": 4374 }, { "epoch": 5.73, "learning_rate": 2.815e-05, "loss": 0.0344, "step": 4375 }, { "epoch": 5.73, "learning_rate": 2.8145e-05, "loss": 0.0335, "step": 4376 }, { "epoch": 5.73, "learning_rate": 2.8139999999999998e-05, "loss": 0.0347, "step": 4377 }, { "epoch": 5.73, "learning_rate": 2.8135e-05, "loss": 0.0395, "step": 4378 }, { "epoch": 5.73, "learning_rate": 2.813e-05, "loss": 0.0398, "step": 4379 }, { "epoch": 5.73, "learning_rate": 2.8125000000000003e-05, "loss": 0.0339, "step": 4380 }, { "epoch": 5.73, "learning_rate": 2.8120000000000002e-05, "loss": 0.0387, "step": 4381 }, { "epoch": 5.74, "learning_rate": 2.8115e-05, "loss": 0.0327, "step": 4382 }, { "epoch": 5.74, "learning_rate": 2.8110000000000004e-05, "loss": 0.0355, "step": 4383 }, { "epoch": 5.74, "learning_rate": 2.8105000000000003e-05, "loss": 0.0327, "step": 4384 }, { "epoch": 5.74, "learning_rate": 2.8100000000000005e-05, "loss": 0.0339, "step": 4385 }, { "epoch": 5.74, "learning_rate": 2.8094999999999998e-05, "loss": 0.039, "step": 4386 }, { "epoch": 5.74, "learning_rate": 2.809e-05, "loss": 0.0341, "step": 4387 }, { "epoch": 5.74, "learning_rate": 2.8085e-05, "loss": 0.0375, "step": 4388 }, { "epoch": 5.74, "learning_rate": 2.8080000000000002e-05, "loss": 0.0391, "step": 4389 }, { "epoch": 5.75, "learning_rate": 2.8075e-05, "loss": 0.0409, "step": 4390 }, { "epoch": 5.75, "learning_rate": 2.807e-05, "loss": 0.0342, "step": 4391 }, { "epoch": 5.75, "learning_rate": 2.8065000000000003e-05, "loss": 0.0411, "step": 4392 }, { "epoch": 5.75, "learning_rate": 2.8060000000000002e-05, "loss": 0.0338, "step": 4393 }, { "epoch": 5.75, "learning_rate": 2.8055000000000005e-05, "loss": 0.0371, "step": 4394 }, { "epoch": 5.75, "learning_rate": 2.8050000000000004e-05, "loss": 0.0297, "step": 4395 }, { "epoch": 5.75, "learning_rate": 2.8045e-05, "loss": 0.0398, "step": 4396 }, { "epoch": 5.76, "learning_rate": 2.804e-05, "loss": 0.0404, "step": 4397 }, { "epoch": 5.76, "learning_rate": 2.8035000000000002e-05, "loss": 0.038, "step": 4398 }, { "epoch": 5.76, "learning_rate": 2.803e-05, "loss": 0.0365, "step": 4399 }, { "epoch": 5.76, "learning_rate": 2.8025e-05, "loss": 0.0366, "step": 4400 }, { "epoch": 5.76, "learning_rate": 2.8020000000000003e-05, "loss": 0.0373, "step": 4401 }, { "epoch": 5.76, "learning_rate": 2.8015000000000002e-05, "loss": 0.0337, "step": 4402 }, { "epoch": 5.76, "learning_rate": 2.8010000000000005e-05, "loss": 0.0344, "step": 4403 }, { "epoch": 5.76, "learning_rate": 2.8005000000000004e-05, "loss": 0.036, "step": 4404 }, { "epoch": 5.77, "learning_rate": 2.8000000000000003e-05, "loss": 0.0302, "step": 4405 }, { "epoch": 5.77, "learning_rate": 2.7995e-05, "loss": 0.0401, "step": 4406 }, { "epoch": 5.77, "learning_rate": 2.7989999999999998e-05, "loss": 0.0326, "step": 4407 }, { "epoch": 5.77, "learning_rate": 2.7985e-05, "loss": 0.034, "step": 4408 }, { "epoch": 5.77, "learning_rate": 2.798e-05, "loss": 0.0355, "step": 4409 }, { "epoch": 5.77, "learning_rate": 2.7975000000000002e-05, "loss": 0.0339, "step": 4410 }, { "epoch": 5.77, "learning_rate": 2.797e-05, "loss": 0.0394, "step": 4411 }, { "epoch": 5.77, "learning_rate": 2.7965e-05, "loss": 0.0332, "step": 4412 }, { "epoch": 5.78, "learning_rate": 2.7960000000000003e-05, "loss": 0.0385, "step": 4413 }, { "epoch": 5.78, "learning_rate": 2.7955000000000003e-05, "loss": 0.0306, "step": 4414 }, { "epoch": 5.78, "learning_rate": 2.7950000000000005e-05, "loss": 0.04, "step": 4415 }, { "epoch": 5.78, "learning_rate": 2.7944999999999998e-05, "loss": 0.0411, "step": 4416 }, { "epoch": 5.78, "learning_rate": 2.794e-05, "loss": 0.0376, "step": 4417 }, { "epoch": 5.78, "learning_rate": 2.7935e-05, "loss": 0.0371, "step": 4418 }, { "epoch": 5.78, "learning_rate": 2.7930000000000002e-05, "loss": 0.0438, "step": 4419 }, { "epoch": 5.79, "learning_rate": 2.7925e-05, "loss": 0.0352, "step": 4420 }, { "epoch": 5.79, "learning_rate": 2.792e-05, "loss": 0.0289, "step": 4421 }, { "epoch": 5.79, "learning_rate": 2.7915000000000003e-05, "loss": 0.0315, "step": 4422 }, { "epoch": 5.79, "learning_rate": 2.7910000000000002e-05, "loss": 0.0415, "step": 4423 }, { "epoch": 5.79, "learning_rate": 2.7905000000000005e-05, "loss": 0.0374, "step": 4424 }, { "epoch": 5.79, "learning_rate": 2.7900000000000004e-05, "loss": 0.0391, "step": 4425 }, { "epoch": 5.79, "learning_rate": 2.7895e-05, "loss": 0.0405, "step": 4426 }, { "epoch": 5.79, "learning_rate": 2.789e-05, "loss": 0.034, "step": 4427 }, { "epoch": 5.8, "learning_rate": 2.7885e-05, "loss": 0.0327, "step": 4428 }, { "epoch": 5.8, "learning_rate": 2.788e-05, "loss": 0.0379, "step": 4429 }, { "epoch": 5.8, "learning_rate": 2.7875e-05, "loss": 0.0413, "step": 4430 }, { "epoch": 5.8, "learning_rate": 2.7870000000000003e-05, "loss": 0.0264, "step": 4431 }, { "epoch": 5.8, "learning_rate": 2.7865000000000002e-05, "loss": 0.0368, "step": 4432 }, { "epoch": 5.8, "learning_rate": 2.7860000000000004e-05, "loss": 0.0342, "step": 4433 }, { "epoch": 5.8, "learning_rate": 2.7855000000000004e-05, "loss": 0.0297, "step": 4434 }, { "epoch": 5.8, "learning_rate": 2.7850000000000003e-05, "loss": 0.0374, "step": 4435 }, { "epoch": 5.81, "learning_rate": 2.7845e-05, "loss": 0.0416, "step": 4436 }, { "epoch": 5.81, "learning_rate": 2.7839999999999998e-05, "loss": 0.0517, "step": 4437 }, { "epoch": 5.81, "learning_rate": 2.7835e-05, "loss": 0.0358, "step": 4438 }, { "epoch": 5.81, "learning_rate": 2.783e-05, "loss": 0.04, "step": 4439 }, { "epoch": 5.81, "learning_rate": 2.7825000000000002e-05, "loss": 0.0341, "step": 4440 }, { "epoch": 5.81, "learning_rate": 2.782e-05, "loss": 0.0334, "step": 4441 }, { "epoch": 5.81, "learning_rate": 2.7815e-05, "loss": 0.0441, "step": 4442 }, { "epoch": 5.82, "learning_rate": 2.7810000000000003e-05, "loss": 0.045, "step": 4443 }, { "epoch": 5.82, "learning_rate": 2.7805000000000002e-05, "loss": 0.0428, "step": 4444 }, { "epoch": 5.82, "learning_rate": 2.7800000000000005e-05, "loss": 0.035, "step": 4445 }, { "epoch": 5.82, "learning_rate": 2.7794999999999997e-05, "loss": 0.0445, "step": 4446 }, { "epoch": 5.82, "learning_rate": 2.779e-05, "loss": 0.035, "step": 4447 }, { "epoch": 5.82, "learning_rate": 2.7785e-05, "loss": 0.0395, "step": 4448 }, { "epoch": 5.82, "learning_rate": 2.778e-05, "loss": 0.0352, "step": 4449 }, { "epoch": 5.82, "learning_rate": 2.7775e-05, "loss": 0.036, "step": 4450 }, { "epoch": 5.83, "learning_rate": 2.777e-05, "loss": 0.0406, "step": 4451 }, { "epoch": 5.83, "learning_rate": 2.7765000000000003e-05, "loss": 0.0385, "step": 4452 }, { "epoch": 5.83, "learning_rate": 2.7760000000000002e-05, "loss": 0.0369, "step": 4453 }, { "epoch": 5.83, "learning_rate": 2.7755000000000004e-05, "loss": 0.0333, "step": 4454 }, { "epoch": 5.83, "learning_rate": 2.7750000000000004e-05, "loss": 0.0433, "step": 4455 }, { "epoch": 5.83, "learning_rate": 2.7745e-05, "loss": 0.0365, "step": 4456 }, { "epoch": 5.83, "learning_rate": 2.774e-05, "loss": 0.028, "step": 4457 }, { "epoch": 5.84, "learning_rate": 2.7735e-05, "loss": 0.0383, "step": 4458 }, { "epoch": 5.84, "learning_rate": 2.773e-05, "loss": 0.0399, "step": 4459 }, { "epoch": 5.84, "learning_rate": 2.7725e-05, "loss": 0.0279, "step": 4460 }, { "epoch": 5.84, "learning_rate": 2.7720000000000002e-05, "loss": 0.0369, "step": 4461 }, { "epoch": 5.84, "learning_rate": 2.7715e-05, "loss": 0.0313, "step": 4462 }, { "epoch": 5.84, "learning_rate": 2.7710000000000004e-05, "loss": 0.0393, "step": 4463 }, { "epoch": 5.84, "learning_rate": 2.7705000000000003e-05, "loss": 0.0412, "step": 4464 }, { "epoch": 5.84, "learning_rate": 2.7700000000000002e-05, "loss": 0.0406, "step": 4465 }, { "epoch": 5.85, "learning_rate": 2.7694999999999998e-05, "loss": 0.0343, "step": 4466 }, { "epoch": 5.85, "learning_rate": 2.769e-05, "loss": 0.0375, "step": 4467 }, { "epoch": 5.85, "learning_rate": 2.7685e-05, "loss": 0.0423, "step": 4468 }, { "epoch": 5.85, "learning_rate": 2.768e-05, "loss": 0.039, "step": 4469 }, { "epoch": 5.85, "learning_rate": 2.7675000000000002e-05, "loss": 0.0284, "step": 4470 }, { "epoch": 5.85, "learning_rate": 2.767e-05, "loss": 0.0379, "step": 4471 }, { "epoch": 5.85, "learning_rate": 2.7665000000000004e-05, "loss": 0.0381, "step": 4472 }, { "epoch": 5.85, "learning_rate": 2.7660000000000003e-05, "loss": 0.0322, "step": 4473 }, { "epoch": 5.86, "learning_rate": 2.7655000000000002e-05, "loss": 0.0364, "step": 4474 }, { "epoch": 5.86, "learning_rate": 2.7650000000000005e-05, "loss": 0.0423, "step": 4475 }, { "epoch": 5.86, "learning_rate": 2.7644999999999997e-05, "loss": 0.0393, "step": 4476 }, { "epoch": 5.86, "learning_rate": 2.764e-05, "loss": 0.0394, "step": 4477 }, { "epoch": 5.86, "learning_rate": 2.7635e-05, "loss": 0.0349, "step": 4478 }, { "epoch": 5.86, "learning_rate": 2.763e-05, "loss": 0.0374, "step": 4479 }, { "epoch": 5.86, "learning_rate": 2.7625e-05, "loss": 0.0347, "step": 4480 }, { "epoch": 5.87, "learning_rate": 2.762e-05, "loss": 0.0494, "step": 4481 }, { "epoch": 5.87, "learning_rate": 2.7615000000000002e-05, "loss": 0.038, "step": 4482 }, { "epoch": 5.87, "learning_rate": 2.761e-05, "loss": 0.044, "step": 4483 }, { "epoch": 5.87, "learning_rate": 2.7605000000000004e-05, "loss": 0.0334, "step": 4484 }, { "epoch": 5.87, "learning_rate": 2.7600000000000003e-05, "loss": 0.0382, "step": 4485 }, { "epoch": 5.87, "learning_rate": 2.7595e-05, "loss": 0.0364, "step": 4486 }, { "epoch": 5.87, "learning_rate": 2.759e-05, "loss": 0.0405, "step": 4487 }, { "epoch": 5.87, "learning_rate": 2.7585e-05, "loss": 0.0398, "step": 4488 }, { "epoch": 5.88, "learning_rate": 2.758e-05, "loss": 0.0378, "step": 4489 }, { "epoch": 5.88, "learning_rate": 2.7575e-05, "loss": 0.0418, "step": 4490 }, { "epoch": 5.88, "learning_rate": 2.7570000000000002e-05, "loss": 0.041, "step": 4491 }, { "epoch": 5.88, "learning_rate": 2.7565e-05, "loss": 0.0447, "step": 4492 }, { "epoch": 5.88, "learning_rate": 2.7560000000000004e-05, "loss": 0.0393, "step": 4493 }, { "epoch": 5.88, "learning_rate": 2.7555000000000003e-05, "loss": 0.0407, "step": 4494 }, { "epoch": 5.88, "learning_rate": 2.7550000000000002e-05, "loss": 0.0437, "step": 4495 }, { "epoch": 5.88, "learning_rate": 2.7544999999999998e-05, "loss": 0.0343, "step": 4496 }, { "epoch": 5.89, "learning_rate": 2.754e-05, "loss": 0.0413, "step": 4497 }, { "epoch": 5.89, "learning_rate": 2.7535e-05, "loss": 0.0353, "step": 4498 }, { "epoch": 5.89, "learning_rate": 2.753e-05, "loss": 0.0444, "step": 4499 }, { "epoch": 5.89, "learning_rate": 2.7525e-05, "loss": 0.0326, "step": 4500 }, { "epoch": 5.89, "learning_rate": 2.752e-05, "loss": 0.0365, "step": 4501 }, { "epoch": 5.89, "learning_rate": 2.7515000000000003e-05, "loss": 0.0312, "step": 4502 }, { "epoch": 5.89, "learning_rate": 2.7510000000000003e-05, "loss": 0.0346, "step": 4503 }, { "epoch": 5.9, "learning_rate": 2.7505000000000002e-05, "loss": 0.0325, "step": 4504 }, { "epoch": 5.9, "learning_rate": 2.7500000000000004e-05, "loss": 0.0413, "step": 4505 }, { "epoch": 5.9, "learning_rate": 2.7495000000000004e-05, "loss": 0.0427, "step": 4506 }, { "epoch": 5.9, "learning_rate": 2.749e-05, "loss": 0.0478, "step": 4507 }, { "epoch": 5.9, "learning_rate": 2.7485e-05, "loss": 0.0357, "step": 4508 }, { "epoch": 5.9, "learning_rate": 2.748e-05, "loss": 0.0415, "step": 4509 }, { "epoch": 5.9, "learning_rate": 2.7475e-05, "loss": 0.0376, "step": 4510 }, { "epoch": 5.9, "learning_rate": 2.7470000000000003e-05, "loss": 0.034, "step": 4511 }, { "epoch": 5.91, "learning_rate": 2.7465000000000002e-05, "loss": 0.0384, "step": 4512 }, { "epoch": 5.91, "learning_rate": 2.746e-05, "loss": 0.0328, "step": 4513 }, { "epoch": 5.91, "learning_rate": 2.7455000000000004e-05, "loss": 0.0388, "step": 4514 }, { "epoch": 5.91, "learning_rate": 2.7450000000000003e-05, "loss": 0.0397, "step": 4515 }, { "epoch": 5.91, "learning_rate": 2.7445000000000002e-05, "loss": 0.0414, "step": 4516 }, { "epoch": 5.91, "learning_rate": 2.7439999999999998e-05, "loss": 0.035, "step": 4517 }, { "epoch": 5.91, "learning_rate": 2.7435e-05, "loss": 0.0331, "step": 4518 }, { "epoch": 5.91, "learning_rate": 2.743e-05, "loss": 0.0357, "step": 4519 }, { "epoch": 5.92, "learning_rate": 2.7425e-05, "loss": 0.0377, "step": 4520 }, { "epoch": 5.92, "learning_rate": 2.7420000000000002e-05, "loss": 0.0269, "step": 4521 }, { "epoch": 5.92, "learning_rate": 2.7415e-05, "loss": 0.0439, "step": 4522 }, { "epoch": 5.92, "learning_rate": 2.7410000000000004e-05, "loss": 0.0313, "step": 4523 }, { "epoch": 5.92, "learning_rate": 2.7405000000000003e-05, "loss": 0.0389, "step": 4524 }, { "epoch": 5.92, "learning_rate": 2.7400000000000002e-05, "loss": 0.043, "step": 4525 }, { "epoch": 5.92, "learning_rate": 2.7395000000000005e-05, "loss": 0.0433, "step": 4526 }, { "epoch": 5.93, "learning_rate": 2.739e-05, "loss": 0.0307, "step": 4527 }, { "epoch": 5.93, "learning_rate": 2.7385e-05, "loss": 0.033, "step": 4528 }, { "epoch": 5.93, "learning_rate": 2.738e-05, "loss": 0.0446, "step": 4529 }, { "epoch": 5.93, "learning_rate": 2.7375e-05, "loss": 0.034, "step": 4530 }, { "epoch": 5.93, "learning_rate": 2.737e-05, "loss": 0.0425, "step": 4531 }, { "epoch": 5.93, "learning_rate": 2.7365000000000003e-05, "loss": 0.0373, "step": 4532 }, { "epoch": 5.93, "learning_rate": 2.7360000000000002e-05, "loss": 0.0393, "step": 4533 }, { "epoch": 5.93, "learning_rate": 2.7355e-05, "loss": 0.0433, "step": 4534 }, { "epoch": 5.94, "learning_rate": 2.7350000000000004e-05, "loss": 0.0424, "step": 4535 }, { "epoch": 5.94, "learning_rate": 2.7345000000000003e-05, "loss": 0.0339, "step": 4536 }, { "epoch": 5.94, "learning_rate": 2.734e-05, "loss": 0.0445, "step": 4537 }, { "epoch": 5.94, "learning_rate": 2.7335e-05, "loss": 0.0356, "step": 4538 }, { "epoch": 5.94, "learning_rate": 2.733e-05, "loss": 0.0385, "step": 4539 }, { "epoch": 5.94, "learning_rate": 2.7325e-05, "loss": 0.0388, "step": 4540 }, { "epoch": 5.94, "learning_rate": 2.7320000000000003e-05, "loss": 0.0448, "step": 4541 }, { "epoch": 5.95, "learning_rate": 2.7315000000000002e-05, "loss": 0.0354, "step": 4542 }, { "epoch": 5.95, "learning_rate": 2.731e-05, "loss": 0.0389, "step": 4543 }, { "epoch": 5.95, "learning_rate": 2.7305000000000004e-05, "loss": 0.0342, "step": 4544 }, { "epoch": 5.95, "learning_rate": 2.7300000000000003e-05, "loss": 0.0403, "step": 4545 }, { "epoch": 5.95, "learning_rate": 2.7295000000000005e-05, "loss": 0.0378, "step": 4546 }, { "epoch": 5.95, "learning_rate": 2.7289999999999998e-05, "loss": 0.0347, "step": 4547 }, { "epoch": 5.95, "learning_rate": 2.7285e-05, "loss": 0.0395, "step": 4548 }, { "epoch": 5.95, "learning_rate": 2.728e-05, "loss": 0.0381, "step": 4549 }, { "epoch": 5.96, "learning_rate": 2.7275e-05, "loss": 0.0371, "step": 4550 }, { "epoch": 5.96, "learning_rate": 2.727e-05, "loss": 0.041, "step": 4551 }, { "epoch": 5.96, "learning_rate": 2.7265e-05, "loss": 0.0488, "step": 4552 }, { "epoch": 5.96, "learning_rate": 2.7260000000000003e-05, "loss": 0.0416, "step": 4553 }, { "epoch": 5.96, "learning_rate": 2.7255000000000002e-05, "loss": 0.0399, "step": 4554 }, { "epoch": 5.96, "learning_rate": 2.725e-05, "loss": 0.0442, "step": 4555 }, { "epoch": 5.96, "learning_rate": 2.7245000000000004e-05, "loss": 0.0355, "step": 4556 }, { "epoch": 5.96, "learning_rate": 2.724e-05, "loss": 0.0367, "step": 4557 }, { "epoch": 5.97, "learning_rate": 2.7235e-05, "loss": 0.0377, "step": 4558 }, { "epoch": 5.97, "learning_rate": 2.723e-05, "loss": 0.0385, "step": 4559 }, { "epoch": 5.97, "learning_rate": 2.7225e-05, "loss": 0.0447, "step": 4560 }, { "epoch": 5.97, "learning_rate": 2.722e-05, "loss": 0.0341, "step": 4561 }, { "epoch": 5.97, "learning_rate": 2.7215000000000003e-05, "loss": 0.0389, "step": 4562 }, { "epoch": 5.97, "learning_rate": 2.7210000000000002e-05, "loss": 0.0357, "step": 4563 }, { "epoch": 5.97, "learning_rate": 2.7205e-05, "loss": 0.0434, "step": 4564 }, { "epoch": 5.98, "learning_rate": 2.7200000000000004e-05, "loss": 0.0355, "step": 4565 }, { "epoch": 5.98, "learning_rate": 2.7195000000000003e-05, "loss": 0.0402, "step": 4566 }, { "epoch": 5.98, "learning_rate": 2.719e-05, "loss": 0.0351, "step": 4567 }, { "epoch": 5.98, "learning_rate": 2.7184999999999998e-05, "loss": 0.0354, "step": 4568 }, { "epoch": 5.98, "learning_rate": 2.718e-05, "loss": 0.0312, "step": 4569 }, { "epoch": 5.98, "learning_rate": 2.7175e-05, "loss": 0.0379, "step": 4570 }, { "epoch": 5.98, "learning_rate": 2.7170000000000002e-05, "loss": 0.0366, "step": 4571 }, { "epoch": 5.98, "learning_rate": 2.7165e-05, "loss": 0.038, "step": 4572 }, { "epoch": 5.99, "learning_rate": 2.716e-05, "loss": 0.0436, "step": 4573 }, { "epoch": 5.99, "learning_rate": 2.7155000000000003e-05, "loss": 0.0393, "step": 4574 }, { "epoch": 5.99, "learning_rate": 2.7150000000000003e-05, "loss": 0.0341, "step": 4575 }, { "epoch": 5.99, "learning_rate": 2.7145000000000005e-05, "loss": 0.0316, "step": 4576 }, { "epoch": 5.99, "learning_rate": 2.7139999999999998e-05, "loss": 0.0294, "step": 4577 }, { "epoch": 5.99, "learning_rate": 2.7135e-05, "loss": 0.0814, "step": 4578 }, { "epoch": 5.99, "learning_rate": 2.713e-05, "loss": 0.047, "step": 4579 }, { "epoch": 5.99, "learning_rate": 2.7125000000000002e-05, "loss": 0.0384, "step": 4580 }, { "epoch": 6.0, "learning_rate": 2.712e-05, "loss": 0.0269, "step": 4581 }, { "epoch": 6.0, "learning_rate": 2.7115e-05, "loss": 0.0371, "step": 4582 }, { "epoch": 6.0, "learning_rate": 2.7110000000000003e-05, "loss": 0.0374, "step": 4583 }, { "epoch": 6.0, "learning_rate": 2.7105000000000002e-05, "loss": 0.0397, "step": 4584 }, { "epoch": 6.0, "learning_rate": 2.7100000000000005e-05, "loss": 0.0238, "step": 4585 }, { "epoch": 6.0, "learning_rate": 2.7095000000000004e-05, "loss": 0.0227, "step": 4586 }, { "epoch": 6.0, "learning_rate": 2.709e-05, "loss": 0.0193, "step": 4587 }, { "epoch": 6.01, "learning_rate": 2.7085e-05, "loss": 0.0243, "step": 4588 }, { "epoch": 6.01, "learning_rate": 2.7079999999999998e-05, "loss": 0.0221, "step": 4589 }, { "epoch": 6.01, "learning_rate": 2.7075e-05, "loss": 0.0243, "step": 4590 }, { "epoch": 6.01, "learning_rate": 2.707e-05, "loss": 0.0216, "step": 4591 }, { "epoch": 6.01, "learning_rate": 2.7065000000000003e-05, "loss": 0.0234, "step": 4592 }, { "epoch": 6.01, "learning_rate": 2.7060000000000002e-05, "loss": 0.0243, "step": 4593 }, { "epoch": 6.01, "learning_rate": 2.7055e-05, "loss": 0.0193, "step": 4594 }, { "epoch": 6.01, "learning_rate": 2.7050000000000004e-05, "loss": 0.0178, "step": 4595 }, { "epoch": 6.02, "learning_rate": 2.7045000000000003e-05, "loss": 0.0215, "step": 4596 }, { "epoch": 6.02, "learning_rate": 2.704e-05, "loss": 0.0283, "step": 4597 }, { "epoch": 6.02, "learning_rate": 2.7034999999999998e-05, "loss": 0.0218, "step": 4598 }, { "epoch": 6.02, "learning_rate": 2.703e-05, "loss": 0.0211, "step": 4599 }, { "epoch": 6.02, "learning_rate": 2.7025e-05, "loss": 0.0211, "step": 4600 }, { "epoch": 6.02, "learning_rate": 2.7020000000000002e-05, "loss": 0.0191, "step": 4601 }, { "epoch": 6.02, "learning_rate": 2.7015e-05, "loss": 0.0222, "step": 4602 }, { "epoch": 6.02, "learning_rate": 2.701e-05, "loss": 0.0236, "step": 4603 }, { "epoch": 6.03, "learning_rate": 2.7005000000000003e-05, "loss": 0.0179, "step": 4604 }, { "epoch": 6.03, "learning_rate": 2.7000000000000002e-05, "loss": 0.023, "step": 4605 }, { "epoch": 6.03, "learning_rate": 2.6995000000000005e-05, "loss": 0.0197, "step": 4606 }, { "epoch": 6.03, "learning_rate": 2.6989999999999997e-05, "loss": 0.0185, "step": 4607 }, { "epoch": 6.03, "learning_rate": 2.6985e-05, "loss": 0.0175, "step": 4608 }, { "epoch": 6.03, "learning_rate": 2.698e-05, "loss": 0.0266, "step": 4609 }, { "epoch": 6.03, "learning_rate": 2.6975000000000002e-05, "loss": 0.0173, "step": 4610 }, { "epoch": 6.04, "learning_rate": 2.697e-05, "loss": 0.0224, "step": 4611 }, { "epoch": 6.04, "learning_rate": 2.6965e-05, "loss": 0.0198, "step": 4612 }, { "epoch": 6.04, "learning_rate": 2.6960000000000003e-05, "loss": 0.0215, "step": 4613 }, { "epoch": 6.04, "learning_rate": 2.6955000000000002e-05, "loss": 0.0229, "step": 4614 }, { "epoch": 6.04, "learning_rate": 2.6950000000000005e-05, "loss": 0.0225, "step": 4615 }, { "epoch": 6.04, "learning_rate": 2.6945000000000004e-05, "loss": 0.0207, "step": 4616 }, { "epoch": 6.04, "learning_rate": 2.694e-05, "loss": 0.0178, "step": 4617 }, { "epoch": 6.04, "learning_rate": 2.6935e-05, "loss": 0.0189, "step": 4618 }, { "epoch": 6.05, "learning_rate": 2.693e-05, "loss": 0.0221, "step": 4619 }, { "epoch": 6.05, "learning_rate": 2.6925e-05, "loss": 0.0215, "step": 4620 }, { "epoch": 6.05, "learning_rate": 2.692e-05, "loss": 0.0201, "step": 4621 }, { "epoch": 6.05, "learning_rate": 2.6915000000000002e-05, "loss": 0.0181, "step": 4622 }, { "epoch": 6.05, "learning_rate": 2.691e-05, "loss": 0.019, "step": 4623 }, { "epoch": 6.05, "learning_rate": 2.6905e-05, "loss": 0.0186, "step": 4624 }, { "epoch": 6.05, "learning_rate": 2.6900000000000003e-05, "loss": 0.0194, "step": 4625 }, { "epoch": 6.05, "learning_rate": 2.6895000000000003e-05, "loss": 0.019, "step": 4626 }, { "epoch": 6.06, "learning_rate": 2.689e-05, "loss": 0.023, "step": 4627 }, { "epoch": 6.06, "learning_rate": 2.6884999999999998e-05, "loss": 0.0253, "step": 4628 }, { "epoch": 6.06, "learning_rate": 2.688e-05, "loss": 0.0166, "step": 4629 }, { "epoch": 6.06, "learning_rate": 2.6875e-05, "loss": 0.0191, "step": 4630 }, { "epoch": 6.06, "learning_rate": 2.6870000000000002e-05, "loss": 0.0173, "step": 4631 }, { "epoch": 6.06, "learning_rate": 2.6865e-05, "loss": 0.0198, "step": 4632 }, { "epoch": 6.06, "learning_rate": 2.686e-05, "loss": 0.0133, "step": 4633 }, { "epoch": 6.07, "learning_rate": 2.6855000000000003e-05, "loss": 0.0166, "step": 4634 }, { "epoch": 6.07, "learning_rate": 2.6850000000000002e-05, "loss": 0.0166, "step": 4635 }, { "epoch": 6.07, "learning_rate": 2.6845000000000005e-05, "loss": 0.0177, "step": 4636 }, { "epoch": 6.07, "learning_rate": 2.6840000000000004e-05, "loss": 0.0212, "step": 4637 }, { "epoch": 6.07, "learning_rate": 2.6835e-05, "loss": 0.017, "step": 4638 }, { "epoch": 6.07, "learning_rate": 2.683e-05, "loss": 0.0179, "step": 4639 }, { "epoch": 6.07, "learning_rate": 2.6825e-05, "loss": 0.0142, "step": 4640 }, { "epoch": 6.07, "learning_rate": 2.682e-05, "loss": 0.0249, "step": 4641 }, { "epoch": 6.08, "learning_rate": 2.6815e-05, "loss": 0.0191, "step": 4642 }, { "epoch": 6.08, "learning_rate": 2.6810000000000003e-05, "loss": 0.0178, "step": 4643 }, { "epoch": 6.08, "learning_rate": 2.6805000000000002e-05, "loss": 0.0203, "step": 4644 }, { "epoch": 6.08, "learning_rate": 2.6800000000000004e-05, "loss": 0.0181, "step": 4645 }, { "epoch": 6.08, "learning_rate": 2.6795000000000003e-05, "loss": 0.0185, "step": 4646 }, { "epoch": 6.08, "learning_rate": 2.6790000000000003e-05, "loss": 0.0223, "step": 4647 }, { "epoch": 6.08, "learning_rate": 2.6785e-05, "loss": 0.0177, "step": 4648 }, { "epoch": 6.09, "learning_rate": 2.678e-05, "loss": 0.0237, "step": 4649 }, { "epoch": 6.09, "learning_rate": 2.6775e-05, "loss": 0.0192, "step": 4650 }, { "epoch": 6.09, "learning_rate": 2.677e-05, "loss": 0.0219, "step": 4651 }, { "epoch": 6.09, "learning_rate": 2.6765000000000002e-05, "loss": 0.0202, "step": 4652 }, { "epoch": 6.09, "learning_rate": 2.676e-05, "loss": 0.0191, "step": 4653 }, { "epoch": 6.09, "learning_rate": 2.6755000000000004e-05, "loss": 0.0264, "step": 4654 }, { "epoch": 6.09, "learning_rate": 2.6750000000000003e-05, "loss": 0.0177, "step": 4655 }, { "epoch": 6.09, "learning_rate": 2.6745000000000002e-05, "loss": 0.0171, "step": 4656 }, { "epoch": 6.1, "learning_rate": 2.6740000000000005e-05, "loss": 0.0157, "step": 4657 }, { "epoch": 6.1, "learning_rate": 2.6734999999999997e-05, "loss": 0.0203, "step": 4658 }, { "epoch": 6.1, "learning_rate": 2.673e-05, "loss": 0.0205, "step": 4659 }, { "epoch": 6.1, "learning_rate": 2.6725e-05, "loss": 0.0207, "step": 4660 }, { "epoch": 6.1, "learning_rate": 2.672e-05, "loss": 0.021, "step": 4661 }, { "epoch": 6.1, "learning_rate": 2.6715e-05, "loss": 0.0209, "step": 4662 }, { "epoch": 6.1, "learning_rate": 2.671e-05, "loss": 0.0204, "step": 4663 }, { "epoch": 6.1, "learning_rate": 2.6705000000000003e-05, "loss": 0.0183, "step": 4664 }, { "epoch": 6.11, "learning_rate": 2.6700000000000002e-05, "loss": 0.0221, "step": 4665 }, { "epoch": 6.11, "learning_rate": 2.6695000000000004e-05, "loss": 0.0178, "step": 4666 }, { "epoch": 6.11, "learning_rate": 2.6690000000000004e-05, "loss": 0.019, "step": 4667 }, { "epoch": 6.11, "learning_rate": 2.6685e-05, "loss": 0.0209, "step": 4668 }, { "epoch": 6.11, "learning_rate": 2.668e-05, "loss": 0.0199, "step": 4669 }, { "epoch": 6.11, "learning_rate": 2.6675e-05, "loss": 0.0232, "step": 4670 }, { "epoch": 6.11, "learning_rate": 2.667e-05, "loss": 0.0204, "step": 4671 }, { "epoch": 6.12, "learning_rate": 2.6665e-05, "loss": 0.0181, "step": 4672 }, { "epoch": 6.12, "learning_rate": 2.6660000000000002e-05, "loss": 0.0211, "step": 4673 }, { "epoch": 6.12, "learning_rate": 2.6655e-05, "loss": 0.0191, "step": 4674 }, { "epoch": 6.12, "learning_rate": 2.6650000000000004e-05, "loss": 0.0225, "step": 4675 }, { "epoch": 6.12, "learning_rate": 2.6645000000000003e-05, "loss": 0.0224, "step": 4676 }, { "epoch": 6.12, "learning_rate": 2.6640000000000002e-05, "loss": 0.0199, "step": 4677 }, { "epoch": 6.12, "learning_rate": 2.6634999999999998e-05, "loss": 0.015, "step": 4678 }, { "epoch": 6.12, "learning_rate": 2.663e-05, "loss": 0.0192, "step": 4679 }, { "epoch": 6.13, "learning_rate": 2.6625e-05, "loss": 0.0227, "step": 4680 }, { "epoch": 6.13, "learning_rate": 2.662e-05, "loss": 0.0175, "step": 4681 }, { "epoch": 6.13, "learning_rate": 2.6615000000000002e-05, "loss": 0.0225, "step": 4682 }, { "epoch": 6.13, "learning_rate": 2.661e-05, "loss": 0.0232, "step": 4683 }, { "epoch": 6.13, "learning_rate": 2.6605000000000004e-05, "loss": 0.0222, "step": 4684 }, { "epoch": 6.13, "learning_rate": 2.6600000000000003e-05, "loss": 0.0214, "step": 4685 }, { "epoch": 6.13, "learning_rate": 2.6595000000000002e-05, "loss": 0.0253, "step": 4686 }, { "epoch": 6.13, "learning_rate": 2.6590000000000005e-05, "loss": 0.0189, "step": 4687 }, { "epoch": 6.14, "learning_rate": 2.6585e-05, "loss": 0.0166, "step": 4688 }, { "epoch": 6.14, "learning_rate": 2.658e-05, "loss": 0.0186, "step": 4689 }, { "epoch": 6.14, "learning_rate": 2.6575e-05, "loss": 0.0177, "step": 4690 }, { "epoch": 6.14, "learning_rate": 2.657e-05, "loss": 0.0253, "step": 4691 }, { "epoch": 6.14, "learning_rate": 2.6565e-05, "loss": 0.019, "step": 4692 }, { "epoch": 6.14, "learning_rate": 2.6560000000000003e-05, "loss": 0.0176, "step": 4693 }, { "epoch": 6.14, "learning_rate": 2.6555000000000002e-05, "loss": 0.0206, "step": 4694 }, { "epoch": 6.15, "learning_rate": 2.655e-05, "loss": 0.0186, "step": 4695 }, { "epoch": 6.15, "learning_rate": 2.6545000000000004e-05, "loss": 0.0227, "step": 4696 }, { "epoch": 6.15, "learning_rate": 2.6540000000000003e-05, "loss": 0.0176, "step": 4697 }, { "epoch": 6.15, "learning_rate": 2.6535e-05, "loss": 0.0235, "step": 4698 }, { "epoch": 6.15, "learning_rate": 2.653e-05, "loss": 0.0268, "step": 4699 }, { "epoch": 6.15, "learning_rate": 2.6525e-05, "loss": 0.0184, "step": 4700 }, { "epoch": 6.15, "learning_rate": 2.652e-05, "loss": 0.0206, "step": 4701 }, { "epoch": 6.15, "learning_rate": 2.6515e-05, "loss": 0.0213, "step": 4702 }, { "epoch": 6.16, "learning_rate": 2.6510000000000002e-05, "loss": 0.0238, "step": 4703 }, { "epoch": 6.16, "learning_rate": 2.6505e-05, "loss": 0.0216, "step": 4704 }, { "epoch": 6.16, "learning_rate": 2.6500000000000004e-05, "loss": 0.0194, "step": 4705 }, { "epoch": 6.16, "learning_rate": 2.6495000000000003e-05, "loss": 0.0293, "step": 4706 }, { "epoch": 6.16, "learning_rate": 2.6490000000000002e-05, "loss": 0.0169, "step": 4707 }, { "epoch": 6.16, "learning_rate": 2.6484999999999998e-05, "loss": 0.0203, "step": 4708 }, { "epoch": 6.16, "learning_rate": 2.648e-05, "loss": 0.0195, "step": 4709 }, { "epoch": 6.16, "learning_rate": 2.6475e-05, "loss": 0.0193, "step": 4710 }, { "epoch": 6.17, "learning_rate": 2.647e-05, "loss": 0.0205, "step": 4711 }, { "epoch": 6.17, "learning_rate": 2.6465e-05, "loss": 0.0219, "step": 4712 }, { "epoch": 6.17, "learning_rate": 2.646e-05, "loss": 0.0183, "step": 4713 }, { "epoch": 6.17, "learning_rate": 2.6455000000000003e-05, "loss": 0.0214, "step": 4714 }, { "epoch": 6.17, "learning_rate": 2.6450000000000003e-05, "loss": 0.0252, "step": 4715 }, { "epoch": 6.17, "learning_rate": 2.6445000000000002e-05, "loss": 0.0167, "step": 4716 }, { "epoch": 6.17, "learning_rate": 2.6440000000000004e-05, "loss": 0.0188, "step": 4717 }, { "epoch": 6.18, "learning_rate": 2.6435e-05, "loss": 0.0236, "step": 4718 }, { "epoch": 6.18, "learning_rate": 2.643e-05, "loss": 0.0211, "step": 4719 }, { "epoch": 6.18, "learning_rate": 2.6425e-05, "loss": 0.0177, "step": 4720 }, { "epoch": 6.18, "learning_rate": 2.642e-05, "loss": 0.0266, "step": 4721 }, { "epoch": 6.18, "learning_rate": 2.6415e-05, "loss": 0.0217, "step": 4722 }, { "epoch": 6.18, "learning_rate": 2.6410000000000003e-05, "loss": 0.0171, "step": 4723 }, { "epoch": 6.18, "learning_rate": 2.6405000000000002e-05, "loss": 0.0153, "step": 4724 }, { "epoch": 6.18, "learning_rate": 2.64e-05, "loss": 0.0199, "step": 4725 }, { "epoch": 6.19, "learning_rate": 2.6395000000000004e-05, "loss": 0.0156, "step": 4726 }, { "epoch": 6.19, "learning_rate": 2.6390000000000003e-05, "loss": 0.0199, "step": 4727 }, { "epoch": 6.19, "learning_rate": 2.6385e-05, "loss": 0.0271, "step": 4728 }, { "epoch": 6.19, "learning_rate": 2.6379999999999998e-05, "loss": 0.0195, "step": 4729 }, { "epoch": 6.19, "learning_rate": 2.6375e-05, "loss": 0.0202, "step": 4730 }, { "epoch": 6.19, "learning_rate": 2.637e-05, "loss": 0.021, "step": 4731 }, { "epoch": 6.19, "learning_rate": 2.6365e-05, "loss": 0.0279, "step": 4732 }, { "epoch": 6.2, "learning_rate": 2.6360000000000002e-05, "loss": 0.0187, "step": 4733 }, { "epoch": 6.2, "learning_rate": 2.6355e-05, "loss": 0.0174, "step": 4734 }, { "epoch": 6.2, "learning_rate": 2.6350000000000004e-05, "loss": 0.0203, "step": 4735 }, { "epoch": 6.2, "learning_rate": 2.6345000000000003e-05, "loss": 0.0209, "step": 4736 }, { "epoch": 6.2, "learning_rate": 2.6340000000000002e-05, "loss": 0.0207, "step": 4737 }, { "epoch": 6.2, "learning_rate": 2.6334999999999998e-05, "loss": 0.0216, "step": 4738 }, { "epoch": 6.2, "learning_rate": 2.633e-05, "loss": 0.017, "step": 4739 }, { "epoch": 6.2, "learning_rate": 2.6325e-05, "loss": 0.0175, "step": 4740 }, { "epoch": 6.21, "learning_rate": 2.632e-05, "loss": 0.0183, "step": 4741 }, { "epoch": 6.21, "learning_rate": 2.6315e-05, "loss": 0.0208, "step": 4742 }, { "epoch": 6.21, "learning_rate": 2.631e-05, "loss": 0.0217, "step": 4743 }, { "epoch": 6.21, "learning_rate": 2.6305000000000003e-05, "loss": 0.0139, "step": 4744 }, { "epoch": 6.21, "learning_rate": 2.6300000000000002e-05, "loss": 0.0217, "step": 4745 }, { "epoch": 6.21, "learning_rate": 2.6295e-05, "loss": 0.0191, "step": 4746 }, { "epoch": 6.21, "learning_rate": 2.6290000000000004e-05, "loss": 0.0215, "step": 4747 }, { "epoch": 6.21, "learning_rate": 2.6285e-05, "loss": 0.0266, "step": 4748 }, { "epoch": 6.22, "learning_rate": 2.628e-05, "loss": 0.0188, "step": 4749 }, { "epoch": 6.22, "learning_rate": 2.6275e-05, "loss": 0.0235, "step": 4750 }, { "epoch": 6.22, "learning_rate": 2.627e-05, "loss": 0.0165, "step": 4751 }, { "epoch": 6.22, "learning_rate": 2.6265e-05, "loss": 0.0195, "step": 4752 }, { "epoch": 6.22, "learning_rate": 2.6260000000000003e-05, "loss": 0.0158, "step": 4753 }, { "epoch": 6.22, "learning_rate": 2.6255000000000002e-05, "loss": 0.0213, "step": 4754 }, { "epoch": 6.22, "learning_rate": 2.625e-05, "loss": 0.0183, "step": 4755 }, { "epoch": 6.23, "learning_rate": 2.6245000000000004e-05, "loss": 0.0197, "step": 4756 }, { "epoch": 6.23, "learning_rate": 2.6240000000000003e-05, "loss": 0.0158, "step": 4757 }, { "epoch": 6.23, "learning_rate": 2.6235000000000005e-05, "loss": 0.0163, "step": 4758 }, { "epoch": 6.23, "learning_rate": 2.6229999999999998e-05, "loss": 0.0194, "step": 4759 }, { "epoch": 6.23, "learning_rate": 2.6225e-05, "loss": 0.021, "step": 4760 }, { "epoch": 6.23, "learning_rate": 2.622e-05, "loss": 0.022, "step": 4761 }, { "epoch": 6.23, "learning_rate": 2.6215000000000002e-05, "loss": 0.0173, "step": 4762 }, { "epoch": 6.23, "learning_rate": 2.621e-05, "loss": 0.0213, "step": 4763 }, { "epoch": 6.24, "learning_rate": 2.6205e-05, "loss": 0.0177, "step": 4764 }, { "epoch": 6.24, "learning_rate": 2.6200000000000003e-05, "loss": 0.0211, "step": 4765 }, { "epoch": 6.24, "learning_rate": 2.6195000000000002e-05, "loss": 0.0289, "step": 4766 }, { "epoch": 6.24, "learning_rate": 2.6190000000000005e-05, "loss": 0.0222, "step": 4767 }, { "epoch": 6.24, "learning_rate": 2.6185000000000004e-05, "loss": 0.0165, "step": 4768 }, { "epoch": 6.24, "learning_rate": 2.618e-05, "loss": 0.0207, "step": 4769 }, { "epoch": 6.24, "learning_rate": 2.6175e-05, "loss": 0.018, "step": 4770 }, { "epoch": 6.24, "learning_rate": 2.617e-05, "loss": 0.0172, "step": 4771 }, { "epoch": 6.25, "learning_rate": 2.6165e-05, "loss": 0.0135, "step": 4772 }, { "epoch": 6.25, "learning_rate": 2.616e-05, "loss": 0.0193, "step": 4773 }, { "epoch": 6.25, "learning_rate": 2.6155000000000003e-05, "loss": 0.0172, "step": 4774 }, { "epoch": 6.25, "learning_rate": 2.6150000000000002e-05, "loss": 0.0243, "step": 4775 }, { "epoch": 6.25, "learning_rate": 2.6145e-05, "loss": 0.0184, "step": 4776 }, { "epoch": 6.25, "learning_rate": 2.6140000000000004e-05, "loss": 0.0199, "step": 4777 }, { "epoch": 6.25, "learning_rate": 2.6135000000000003e-05, "loss": 0.023, "step": 4778 }, { "epoch": 6.26, "learning_rate": 2.613e-05, "loss": 0.0201, "step": 4779 }, { "epoch": 6.26, "learning_rate": 2.6124999999999998e-05, "loss": 0.0198, "step": 4780 }, { "epoch": 6.26, "learning_rate": 2.612e-05, "loss": 0.017, "step": 4781 }, { "epoch": 6.26, "learning_rate": 2.6115e-05, "loss": 0.0184, "step": 4782 }, { "epoch": 6.26, "learning_rate": 2.6110000000000002e-05, "loss": 0.0256, "step": 4783 }, { "epoch": 6.26, "learning_rate": 2.6105e-05, "loss": 0.0211, "step": 4784 }, { "epoch": 6.26, "learning_rate": 2.61e-05, "loss": 0.0218, "step": 4785 }, { "epoch": 6.26, "learning_rate": 2.6095000000000003e-05, "loss": 0.0191, "step": 4786 }, { "epoch": 6.27, "learning_rate": 2.6090000000000003e-05, "loss": 0.0211, "step": 4787 }, { "epoch": 6.27, "learning_rate": 2.6085000000000005e-05, "loss": 0.0221, "step": 4788 }, { "epoch": 6.27, "learning_rate": 2.6079999999999998e-05, "loss": 0.0185, "step": 4789 }, { "epoch": 6.27, "learning_rate": 2.6075e-05, "loss": 0.0155, "step": 4790 }, { "epoch": 6.27, "learning_rate": 2.607e-05, "loss": 0.0173, "step": 4791 }, { "epoch": 6.27, "learning_rate": 2.6065000000000002e-05, "loss": 0.0186, "step": 4792 }, { "epoch": 6.27, "learning_rate": 2.606e-05, "loss": 0.0187, "step": 4793 }, { "epoch": 6.27, "learning_rate": 2.6055e-05, "loss": 0.0192, "step": 4794 }, { "epoch": 6.28, "learning_rate": 2.6050000000000003e-05, "loss": 0.017, "step": 4795 }, { "epoch": 6.28, "learning_rate": 2.6045000000000002e-05, "loss": 0.0192, "step": 4796 }, { "epoch": 6.28, "learning_rate": 2.6040000000000005e-05, "loss": 0.0173, "step": 4797 }, { "epoch": 6.28, "learning_rate": 2.6035000000000004e-05, "loss": 0.0206, "step": 4798 }, { "epoch": 6.28, "learning_rate": 2.603e-05, "loss": 0.0204, "step": 4799 }, { "epoch": 6.28, "learning_rate": 2.6025e-05, "loss": 0.0145, "step": 4800 }, { "epoch": 6.28, "learning_rate": 2.602e-05, "loss": 0.0177, "step": 4801 }, { "epoch": 6.29, "learning_rate": 2.6015e-05, "loss": 0.0226, "step": 4802 }, { "epoch": 6.29, "learning_rate": 2.601e-05, "loss": 0.0198, "step": 4803 }, { "epoch": 6.29, "learning_rate": 2.6005000000000003e-05, "loss": 0.0171, "step": 4804 }, { "epoch": 6.29, "learning_rate": 2.6000000000000002e-05, "loss": 0.0275, "step": 4805 }, { "epoch": 6.29, "learning_rate": 2.5995000000000004e-05, "loss": 0.0223, "step": 4806 }, { "epoch": 6.29, "learning_rate": 2.5990000000000004e-05, "loss": 0.0212, "step": 4807 }, { "epoch": 6.29, "learning_rate": 2.5985000000000003e-05, "loss": 0.018, "step": 4808 }, { "epoch": 6.29, "learning_rate": 2.598e-05, "loss": 0.0168, "step": 4809 }, { "epoch": 6.3, "learning_rate": 2.5974999999999998e-05, "loss": 0.0169, "step": 4810 }, { "epoch": 6.3, "learning_rate": 2.597e-05, "loss": 0.0214, "step": 4811 }, { "epoch": 6.3, "learning_rate": 2.5965e-05, "loss": 0.0163, "step": 4812 }, { "epoch": 6.3, "learning_rate": 2.5960000000000002e-05, "loss": 0.0222, "step": 4813 }, { "epoch": 6.3, "learning_rate": 2.5955e-05, "loss": 0.0208, "step": 4814 }, { "epoch": 6.3, "learning_rate": 2.595e-05, "loss": 0.0184, "step": 4815 }, { "epoch": 6.3, "learning_rate": 2.5945000000000003e-05, "loss": 0.0168, "step": 4816 }, { "epoch": 6.3, "learning_rate": 2.5940000000000002e-05, "loss": 0.0249, "step": 4817 }, { "epoch": 6.31, "learning_rate": 2.5935000000000005e-05, "loss": 0.0241, "step": 4818 }, { "epoch": 6.31, "learning_rate": 2.5929999999999997e-05, "loss": 0.0263, "step": 4819 }, { "epoch": 6.31, "learning_rate": 2.5925e-05, "loss": 0.0173, "step": 4820 }, { "epoch": 6.31, "learning_rate": 2.592e-05, "loss": 0.0218, "step": 4821 }, { "epoch": 6.31, "learning_rate": 2.5915000000000002e-05, "loss": 0.0213, "step": 4822 }, { "epoch": 6.31, "learning_rate": 2.591e-05, "loss": 0.023, "step": 4823 }, { "epoch": 6.31, "learning_rate": 2.5905e-05, "loss": 0.0166, "step": 4824 }, { "epoch": 6.32, "learning_rate": 2.5900000000000003e-05, "loss": 0.022, "step": 4825 }, { "epoch": 6.32, "learning_rate": 2.5895000000000002e-05, "loss": 0.0208, "step": 4826 }, { "epoch": 6.32, "learning_rate": 2.5890000000000005e-05, "loss": 0.0201, "step": 4827 }, { "epoch": 6.32, "learning_rate": 2.5885000000000004e-05, "loss": 0.02, "step": 4828 }, { "epoch": 6.32, "learning_rate": 2.588e-05, "loss": 0.0205, "step": 4829 }, { "epoch": 6.32, "learning_rate": 2.5875e-05, "loss": 0.0198, "step": 4830 }, { "epoch": 6.32, "learning_rate": 2.587e-05, "loss": 0.0171, "step": 4831 }, { "epoch": 6.32, "learning_rate": 2.5865e-05, "loss": 0.0223, "step": 4832 }, { "epoch": 6.33, "learning_rate": 2.586e-05, "loss": 0.0189, "step": 4833 }, { "epoch": 6.33, "learning_rate": 2.5855000000000002e-05, "loss": 0.0195, "step": 4834 }, { "epoch": 6.33, "learning_rate": 2.585e-05, "loss": 0.0245, "step": 4835 }, { "epoch": 6.33, "learning_rate": 2.5845000000000004e-05, "loss": 0.0205, "step": 4836 }, { "epoch": 6.33, "learning_rate": 2.5840000000000003e-05, "loss": 0.0165, "step": 4837 }, { "epoch": 6.33, "learning_rate": 2.5835000000000003e-05, "loss": 0.0175, "step": 4838 }, { "epoch": 6.33, "learning_rate": 2.583e-05, "loss": 0.0199, "step": 4839 }, { "epoch": 6.34, "learning_rate": 2.5824999999999998e-05, "loss": 0.0206, "step": 4840 }, { "epoch": 6.34, "learning_rate": 2.582e-05, "loss": 0.0216, "step": 4841 }, { "epoch": 6.34, "learning_rate": 2.5815e-05, "loss": 0.0224, "step": 4842 }, { "epoch": 6.34, "learning_rate": 2.5810000000000002e-05, "loss": 0.024, "step": 4843 }, { "epoch": 6.34, "learning_rate": 2.5805e-05, "loss": 0.0202, "step": 4844 }, { "epoch": 6.34, "learning_rate": 2.58e-05, "loss": 0.0186, "step": 4845 }, { "epoch": 6.34, "learning_rate": 2.5795000000000003e-05, "loss": 0.0241, "step": 4846 }, { "epoch": 6.34, "learning_rate": 2.5790000000000002e-05, "loss": 0.0208, "step": 4847 }, { "epoch": 6.35, "learning_rate": 2.5785000000000005e-05, "loss": 0.0214, "step": 4848 }, { "epoch": 6.35, "learning_rate": 2.5779999999999997e-05, "loss": 0.0161, "step": 4849 }, { "epoch": 6.35, "learning_rate": 2.5775e-05, "loss": 0.0206, "step": 4850 }, { "epoch": 6.35, "learning_rate": 2.577e-05, "loss": 0.0218, "step": 4851 }, { "epoch": 6.35, "learning_rate": 2.5765e-05, "loss": 0.0195, "step": 4852 }, { "epoch": 6.35, "learning_rate": 2.576e-05, "loss": 0.0215, "step": 4853 }, { "epoch": 6.35, "learning_rate": 2.5755e-05, "loss": 0.0187, "step": 4854 }, { "epoch": 6.35, "learning_rate": 2.5750000000000002e-05, "loss": 0.0184, "step": 4855 }, { "epoch": 6.36, "learning_rate": 2.5745e-05, "loss": 0.0193, "step": 4856 }, { "epoch": 6.36, "learning_rate": 2.5740000000000004e-05, "loss": 0.0207, "step": 4857 }, { "epoch": 6.36, "learning_rate": 2.5735000000000003e-05, "loss": 0.0208, "step": 4858 }, { "epoch": 6.36, "learning_rate": 2.573e-05, "loss": 0.021, "step": 4859 }, { "epoch": 6.36, "learning_rate": 2.5725e-05, "loss": 0.0173, "step": 4860 }, { "epoch": 6.36, "learning_rate": 2.572e-05, "loss": 0.0165, "step": 4861 }, { "epoch": 6.36, "learning_rate": 2.5715e-05, "loss": 0.0203, "step": 4862 }, { "epoch": 6.37, "learning_rate": 2.571e-05, "loss": 0.0215, "step": 4863 }, { "epoch": 6.37, "learning_rate": 2.5705000000000002e-05, "loss": 0.0211, "step": 4864 }, { "epoch": 6.37, "learning_rate": 2.57e-05, "loss": 0.0191, "step": 4865 }, { "epoch": 6.37, "learning_rate": 2.5695000000000004e-05, "loss": 0.0202, "step": 4866 }, { "epoch": 6.37, "learning_rate": 2.5690000000000003e-05, "loss": 0.0201, "step": 4867 }, { "epoch": 6.37, "learning_rate": 2.5685000000000002e-05, "loss": 0.0192, "step": 4868 }, { "epoch": 6.37, "learning_rate": 2.5679999999999998e-05, "loss": 0.0182, "step": 4869 }, { "epoch": 6.37, "learning_rate": 2.5675e-05, "loss": 0.0183, "step": 4870 }, { "epoch": 6.38, "learning_rate": 2.567e-05, "loss": 0.0227, "step": 4871 }, { "epoch": 6.38, "learning_rate": 2.5665e-05, "loss": 0.0209, "step": 4872 }, { "epoch": 6.38, "learning_rate": 2.566e-05, "loss": 0.0411, "step": 4873 }, { "epoch": 6.38, "learning_rate": 2.5655e-05, "loss": 0.0186, "step": 4874 }, { "epoch": 6.38, "learning_rate": 2.5650000000000003e-05, "loss": 0.0193, "step": 4875 }, { "epoch": 6.38, "learning_rate": 2.5645000000000003e-05, "loss": 0.0205, "step": 4876 }, { "epoch": 6.38, "learning_rate": 2.5640000000000002e-05, "loss": 0.0188, "step": 4877 }, { "epoch": 6.38, "learning_rate": 2.5635000000000004e-05, "loss": 0.019, "step": 4878 }, { "epoch": 6.39, "learning_rate": 2.5629999999999997e-05, "loss": 0.0185, "step": 4879 }, { "epoch": 6.39, "learning_rate": 2.5625e-05, "loss": 0.0189, "step": 4880 }, { "epoch": 6.39, "learning_rate": 2.562e-05, "loss": 0.0201, "step": 4881 }, { "epoch": 6.39, "learning_rate": 2.5615e-05, "loss": 0.0165, "step": 4882 }, { "epoch": 6.39, "learning_rate": 2.561e-05, "loss": 0.0182, "step": 4883 }, { "epoch": 6.39, "learning_rate": 2.5605e-05, "loss": 0.0227, "step": 4884 }, { "epoch": 6.39, "learning_rate": 2.5600000000000002e-05, "loss": 0.0204, "step": 4885 }, { "epoch": 6.4, "learning_rate": 2.5595e-05, "loss": 0.0183, "step": 4886 }, { "epoch": 6.4, "learning_rate": 2.5590000000000004e-05, "loss": 0.0189, "step": 4887 }, { "epoch": 6.4, "learning_rate": 2.5585000000000003e-05, "loss": 0.0152, "step": 4888 }, { "epoch": 6.4, "learning_rate": 2.5580000000000002e-05, "loss": 0.0198, "step": 4889 }, { "epoch": 6.4, "learning_rate": 2.5574999999999998e-05, "loss": 0.0207, "step": 4890 }, { "epoch": 6.4, "learning_rate": 2.557e-05, "loss": 0.0196, "step": 4891 }, { "epoch": 6.4, "learning_rate": 2.5565e-05, "loss": 0.0188, "step": 4892 }, { "epoch": 6.4, "learning_rate": 2.556e-05, "loss": 0.0171, "step": 4893 }, { "epoch": 6.41, "learning_rate": 2.5555000000000002e-05, "loss": 0.0233, "step": 4894 }, { "epoch": 6.41, "learning_rate": 2.555e-05, "loss": 0.0215, "step": 4895 }, { "epoch": 6.41, "learning_rate": 2.5545000000000004e-05, "loss": 0.0246, "step": 4896 }, { "epoch": 6.41, "learning_rate": 2.5540000000000003e-05, "loss": 0.0176, "step": 4897 }, { "epoch": 6.41, "learning_rate": 2.5535000000000002e-05, "loss": 0.0221, "step": 4898 }, { "epoch": 6.41, "learning_rate": 2.5530000000000005e-05, "loss": 0.0265, "step": 4899 }, { "epoch": 6.41, "learning_rate": 2.5525e-05, "loss": 0.0179, "step": 4900 }, { "epoch": 6.41, "learning_rate": 2.552e-05, "loss": 0.0221, "step": 4901 }, { "epoch": 6.42, "learning_rate": 2.5515e-05, "loss": 0.0218, "step": 4902 }, { "epoch": 6.42, "learning_rate": 2.551e-05, "loss": 0.0169, "step": 4903 }, { "epoch": 6.42, "learning_rate": 2.5505e-05, "loss": 0.0231, "step": 4904 }, { "epoch": 6.42, "learning_rate": 2.5500000000000003e-05, "loss": 0.0199, "step": 4905 }, { "epoch": 6.42, "learning_rate": 2.5495000000000002e-05, "loss": 0.0233, "step": 4906 }, { "epoch": 6.42, "learning_rate": 2.549e-05, "loss": 0.0236, "step": 4907 }, { "epoch": 6.42, "learning_rate": 2.5485000000000004e-05, "loss": 0.0191, "step": 4908 }, { "epoch": 6.43, "learning_rate": 2.5480000000000003e-05, "loss": 0.0193, "step": 4909 }, { "epoch": 6.43, "learning_rate": 2.5475e-05, "loss": 0.0247, "step": 4910 }, { "epoch": 6.43, "learning_rate": 2.547e-05, "loss": 0.0215, "step": 4911 }, { "epoch": 6.43, "learning_rate": 2.5465e-05, "loss": 0.0198, "step": 4912 }, { "epoch": 6.43, "learning_rate": 2.546e-05, "loss": 0.0203, "step": 4913 }, { "epoch": 6.43, "learning_rate": 2.5455e-05, "loss": 0.0178, "step": 4914 }, { "epoch": 6.43, "learning_rate": 2.5450000000000002e-05, "loss": 0.0179, "step": 4915 }, { "epoch": 6.43, "learning_rate": 2.5445e-05, "loss": 0.0225, "step": 4916 }, { "epoch": 6.44, "learning_rate": 2.5440000000000004e-05, "loss": 0.0231, "step": 4917 }, { "epoch": 6.44, "learning_rate": 2.5435000000000003e-05, "loss": 0.0273, "step": 4918 }, { "epoch": 6.44, "learning_rate": 2.5430000000000002e-05, "loss": 0.0197, "step": 4919 }, { "epoch": 6.44, "learning_rate": 2.5424999999999998e-05, "loss": 0.0217, "step": 4920 }, { "epoch": 6.44, "learning_rate": 2.542e-05, "loss": 0.0184, "step": 4921 }, { "epoch": 6.44, "learning_rate": 2.5415e-05, "loss": 0.0252, "step": 4922 }, { "epoch": 6.44, "learning_rate": 2.541e-05, "loss": 0.0178, "step": 4923 }, { "epoch": 6.45, "learning_rate": 2.541e-05, "loss": 0.0253, "step": 4924 }, { "epoch": 6.45, "learning_rate": 2.541e-05, "loss": 0.0183, "step": 4925 }, { "epoch": 6.45, "learning_rate": 2.5405e-05, "loss": 0.0275, "step": 4926 }, { "epoch": 6.45, "learning_rate": 2.54e-05, "loss": 0.017, "step": 4927 }, { "epoch": 6.45, "learning_rate": 2.5395000000000003e-05, "loss": 0.0196, "step": 4928 }, { "epoch": 6.45, "learning_rate": 2.5390000000000003e-05, "loss": 0.0282, "step": 4929 }, { "epoch": 6.45, "learning_rate": 2.5385000000000002e-05, "loss": 0.0217, "step": 4930 }, { "epoch": 6.45, "learning_rate": 2.5380000000000004e-05, "loss": 0.0234, "step": 4931 }, { "epoch": 6.46, "learning_rate": 2.5375e-05, "loss": 0.0182, "step": 4932 }, { "epoch": 6.46, "learning_rate": 2.537e-05, "loss": 0.0186, "step": 4933 }, { "epoch": 6.46, "learning_rate": 2.5365e-05, "loss": 0.0218, "step": 4934 }, { "epoch": 6.46, "learning_rate": 2.536e-05, "loss": 0.022, "step": 4935 }, { "epoch": 6.46, "learning_rate": 2.5355e-05, "loss": 0.0217, "step": 4936 }, { "epoch": 6.46, "learning_rate": 2.5350000000000003e-05, "loss": 0.0202, "step": 4937 }, { "epoch": 6.46, "learning_rate": 2.5345000000000002e-05, "loss": 0.0187, "step": 4938 }, { "epoch": 6.46, "learning_rate": 2.534e-05, "loss": 0.0207, "step": 4939 }, { "epoch": 6.47, "learning_rate": 2.5335000000000004e-05, "loss": 0.0248, "step": 4940 }, { "epoch": 6.47, "learning_rate": 2.5330000000000003e-05, "loss": 0.0217, "step": 4941 }, { "epoch": 6.47, "learning_rate": 2.5325e-05, "loss": 0.0218, "step": 4942 }, { "epoch": 6.47, "learning_rate": 2.5319999999999998e-05, "loss": 0.0177, "step": 4943 }, { "epoch": 6.47, "learning_rate": 2.5315e-05, "loss": 0.0236, "step": 4944 }, { "epoch": 6.47, "learning_rate": 2.531e-05, "loss": 0.0211, "step": 4945 }, { "epoch": 6.47, "learning_rate": 2.5305000000000003e-05, "loss": 0.0167, "step": 4946 }, { "epoch": 6.48, "learning_rate": 2.5300000000000002e-05, "loss": 0.0205, "step": 4947 }, { "epoch": 6.48, "learning_rate": 2.5295e-05, "loss": 0.0183, "step": 4948 }, { "epoch": 6.48, "learning_rate": 2.5290000000000004e-05, "loss": 0.0196, "step": 4949 }, { "epoch": 6.48, "learning_rate": 2.5285000000000003e-05, "loss": 0.0231, "step": 4950 }, { "epoch": 6.48, "learning_rate": 2.5280000000000005e-05, "loss": 0.0227, "step": 4951 }, { "epoch": 6.48, "learning_rate": 2.5274999999999998e-05, "loss": 0.0163, "step": 4952 }, { "epoch": 6.48, "learning_rate": 2.527e-05, "loss": 0.0181, "step": 4953 }, { "epoch": 6.48, "learning_rate": 2.5265e-05, "loss": 0.0181, "step": 4954 }, { "epoch": 6.49, "learning_rate": 2.526e-05, "loss": 0.0227, "step": 4955 }, { "epoch": 6.49, "learning_rate": 2.5255e-05, "loss": 0.0231, "step": 4956 }, { "epoch": 6.49, "learning_rate": 2.525e-05, "loss": 0.0204, "step": 4957 }, { "epoch": 6.49, "learning_rate": 2.5245000000000003e-05, "loss": 0.0196, "step": 4958 }, { "epoch": 6.49, "learning_rate": 2.5240000000000002e-05, "loss": 0.0223, "step": 4959 }, { "epoch": 6.49, "learning_rate": 2.5235e-05, "loss": 0.0281, "step": 4960 }, { "epoch": 6.49, "learning_rate": 2.5230000000000004e-05, "loss": 0.0186, "step": 4961 }, { "epoch": 6.49, "learning_rate": 2.5225e-05, "loss": 0.017, "step": 4962 }, { "epoch": 6.5, "learning_rate": 2.522e-05, "loss": 0.0194, "step": 4963 }, { "epoch": 6.5, "learning_rate": 2.5214999999999998e-05, "loss": 0.017, "step": 4964 }, { "epoch": 6.5, "learning_rate": 2.521e-05, "loss": 0.0215, "step": 4965 }, { "epoch": 6.5, "learning_rate": 2.5205e-05, "loss": 0.0178, "step": 4966 }, { "epoch": 6.5, "learning_rate": 2.5200000000000003e-05, "loss": 0.0182, "step": 4967 }, { "epoch": 6.5, "learning_rate": 2.5195000000000002e-05, "loss": 0.0189, "step": 4968 }, { "epoch": 6.5, "learning_rate": 2.519e-05, "loss": 0.023, "step": 4969 }, { "epoch": 6.51, "learning_rate": 2.5185000000000004e-05, "loss": 0.0163, "step": 4970 }, { "epoch": 6.51, "learning_rate": 2.5180000000000003e-05, "loss": 0.0228, "step": 4971 }, { "epoch": 6.51, "learning_rate": 2.5175e-05, "loss": 0.0207, "step": 4972 }, { "epoch": 6.51, "learning_rate": 2.5169999999999998e-05, "loss": 0.0199, "step": 4973 }, { "epoch": 6.51, "learning_rate": 2.5165e-05, "loss": 0.0221, "step": 4974 }, { "epoch": 6.51, "learning_rate": 2.516e-05, "loss": 0.0223, "step": 4975 }, { "epoch": 6.51, "learning_rate": 2.5155000000000002e-05, "loss": 0.0212, "step": 4976 }, { "epoch": 6.51, "learning_rate": 2.515e-05, "loss": 0.0223, "step": 4977 }, { "epoch": 6.52, "learning_rate": 2.5145e-05, "loss": 0.0196, "step": 4978 }, { "epoch": 6.52, "learning_rate": 2.5140000000000003e-05, "loss": 0.0181, "step": 4979 }, { "epoch": 6.52, "learning_rate": 2.5135000000000002e-05, "loss": 0.0189, "step": 4980 }, { "epoch": 6.52, "learning_rate": 2.5130000000000005e-05, "loss": 0.0219, "step": 4981 }, { "epoch": 6.52, "learning_rate": 2.5124999999999997e-05, "loss": 0.0187, "step": 4982 }, { "epoch": 6.52, "learning_rate": 2.512e-05, "loss": 0.0222, "step": 4983 }, { "epoch": 6.52, "learning_rate": 2.5115e-05, "loss": 0.0176, "step": 4984 }, { "epoch": 6.52, "learning_rate": 2.5110000000000002e-05, "loss": 0.0224, "step": 4985 }, { "epoch": 6.53, "learning_rate": 2.5105e-05, "loss": 0.0246, "step": 4986 }, { "epoch": 6.53, "learning_rate": 2.51e-05, "loss": 0.0241, "step": 4987 }, { "epoch": 6.53, "learning_rate": 2.5095000000000003e-05, "loss": 0.0191, "step": 4988 }, { "epoch": 6.53, "learning_rate": 2.5090000000000002e-05, "loss": 0.0178, "step": 4989 }, { "epoch": 6.53, "learning_rate": 2.5085000000000005e-05, "loss": 0.0173, "step": 4990 }, { "epoch": 6.53, "learning_rate": 2.5080000000000004e-05, "loss": 0.0187, "step": 4991 }, { "epoch": 6.53, "learning_rate": 2.5075e-05, "loss": 0.0222, "step": 4992 }, { "epoch": 6.54, "learning_rate": 2.507e-05, "loss": 0.0219, "step": 4993 }, { "epoch": 6.54, "learning_rate": 2.5064999999999998e-05, "loss": 0.0174, "step": 4994 }, { "epoch": 6.54, "learning_rate": 2.506e-05, "loss": 0.029, "step": 4995 }, { "epoch": 6.54, "learning_rate": 2.5055e-05, "loss": 0.021, "step": 4996 }, { "epoch": 6.54, "learning_rate": 2.5050000000000002e-05, "loss": 0.017, "step": 4997 }, { "epoch": 6.54, "learning_rate": 2.5045e-05, "loss": 0.0204, "step": 4998 }, { "epoch": 6.54, "learning_rate": 2.504e-05, "loss": 0.0205, "step": 4999 }, { "epoch": 6.54, "learning_rate": 2.5035000000000003e-05, "loss": 0.0223, "step": 5000 }, { "epoch": 6.55, "learning_rate": 2.5030000000000003e-05, "loss": 0.0199, "step": 5001 }, { "epoch": 6.55, "learning_rate": 2.5025e-05, "loss": 0.0194, "step": 5002 }, { "epoch": 6.55, "learning_rate": 2.5019999999999998e-05, "loss": 0.0197, "step": 5003 }, { "epoch": 6.55, "learning_rate": 2.5015e-05, "loss": 0.025, "step": 5004 }, { "epoch": 6.55, "learning_rate": 2.501e-05, "loss": 0.0217, "step": 5005 }, { "epoch": 6.55, "learning_rate": 2.5005000000000002e-05, "loss": 0.0165, "step": 5006 }, { "epoch": 6.55, "learning_rate": 2.5e-05, "loss": 0.0244, "step": 5007 }, { "epoch": 6.55, "learning_rate": 2.4995e-05, "loss": 0.0367, "step": 5008 }, { "epoch": 6.56, "learning_rate": 2.4990000000000003e-05, "loss": 0.0207, "step": 5009 }, { "epoch": 6.56, "learning_rate": 2.4985e-05, "loss": 0.0193, "step": 5010 }, { "epoch": 6.56, "learning_rate": 2.498e-05, "loss": 0.0197, "step": 5011 }, { "epoch": 6.56, "learning_rate": 2.4975e-05, "loss": 0.0233, "step": 5012 }, { "epoch": 6.56, "learning_rate": 2.4970000000000003e-05, "loss": 0.0227, "step": 5013 }, { "epoch": 6.56, "learning_rate": 2.4965000000000002e-05, "loss": 0.024, "step": 5014 }, { "epoch": 6.56, "learning_rate": 2.496e-05, "loss": 0.0222, "step": 5015 }, { "epoch": 6.57, "learning_rate": 2.4955e-05, "loss": 0.0221, "step": 5016 }, { "epoch": 6.57, "learning_rate": 2.495e-05, "loss": 0.0215, "step": 5017 }, { "epoch": 6.57, "learning_rate": 2.4945000000000003e-05, "loss": 0.0284, "step": 5018 }, { "epoch": 6.57, "learning_rate": 2.4940000000000002e-05, "loss": 0.0249, "step": 5019 }, { "epoch": 6.57, "learning_rate": 2.4935e-05, "loss": 0.0187, "step": 5020 }, { "epoch": 6.57, "learning_rate": 2.493e-05, "loss": 0.0281, "step": 5021 }, { "epoch": 6.57, "learning_rate": 2.4925000000000003e-05, "loss": 0.0226, "step": 5022 }, { "epoch": 6.57, "learning_rate": 2.4920000000000002e-05, "loss": 0.0234, "step": 5023 }, { "epoch": 6.58, "learning_rate": 2.4915e-05, "loss": 0.0222, "step": 5024 }, { "epoch": 6.58, "learning_rate": 2.491e-05, "loss": 0.022, "step": 5025 }, { "epoch": 6.58, "learning_rate": 2.4905e-05, "loss": 0.0338, "step": 5026 }, { "epoch": 6.58, "learning_rate": 2.4900000000000002e-05, "loss": 0.0203, "step": 5027 }, { "epoch": 6.58, "learning_rate": 2.4895e-05, "loss": 0.0231, "step": 5028 }, { "epoch": 6.58, "learning_rate": 2.489e-05, "loss": 0.0176, "step": 5029 }, { "epoch": 6.58, "learning_rate": 2.4885e-05, "loss": 0.0204, "step": 5030 }, { "epoch": 6.59, "learning_rate": 2.488e-05, "loss": 0.0233, "step": 5031 }, { "epoch": 6.59, "learning_rate": 2.4875e-05, "loss": 0.0199, "step": 5032 }, { "epoch": 6.59, "learning_rate": 2.487e-05, "loss": 0.0234, "step": 5033 }, { "epoch": 6.59, "learning_rate": 2.4865000000000003e-05, "loss": 0.0267, "step": 5034 }, { "epoch": 6.59, "learning_rate": 2.486e-05, "loss": 0.0194, "step": 5035 }, { "epoch": 6.59, "learning_rate": 2.4855000000000002e-05, "loss": 0.0211, "step": 5036 }, { "epoch": 6.59, "learning_rate": 2.485e-05, "loss": 0.0217, "step": 5037 }, { "epoch": 6.59, "learning_rate": 2.4845e-05, "loss": 0.0241, "step": 5038 }, { "epoch": 6.6, "learning_rate": 2.4840000000000003e-05, "loss": 0.025, "step": 5039 }, { "epoch": 6.6, "learning_rate": 2.4835e-05, "loss": 0.0234, "step": 5040 }, { "epoch": 6.6, "learning_rate": 2.483e-05, "loss": 0.0199, "step": 5041 }, { "epoch": 6.6, "learning_rate": 2.4825e-05, "loss": 0.0205, "step": 5042 }, { "epoch": 6.6, "learning_rate": 2.4820000000000003e-05, "loss": 0.0247, "step": 5043 }, { "epoch": 6.6, "learning_rate": 2.4815000000000002e-05, "loss": 0.0262, "step": 5044 }, { "epoch": 6.6, "learning_rate": 2.481e-05, "loss": 0.0252, "step": 5045 }, { "epoch": 6.6, "learning_rate": 2.4805e-05, "loss": 0.0229, "step": 5046 }, { "epoch": 6.61, "learning_rate": 2.48e-05, "loss": 0.02, "step": 5047 }, { "epoch": 6.61, "learning_rate": 2.4795000000000002e-05, "loss": 0.022, "step": 5048 }, { "epoch": 6.61, "learning_rate": 2.479e-05, "loss": 0.0201, "step": 5049 }, { "epoch": 6.61, "learning_rate": 2.4785e-05, "loss": 0.0215, "step": 5050 }, { "epoch": 6.61, "learning_rate": 2.478e-05, "loss": 0.0216, "step": 5051 }, { "epoch": 6.61, "learning_rate": 2.4775000000000003e-05, "loss": 0.0151, "step": 5052 }, { "epoch": 6.61, "learning_rate": 2.4770000000000002e-05, "loss": 0.0219, "step": 5053 }, { "epoch": 6.62, "learning_rate": 2.4765e-05, "loss": 0.0252, "step": 5054 }, { "epoch": 6.62, "learning_rate": 2.476e-05, "loss": 0.0219, "step": 5055 }, { "epoch": 6.62, "learning_rate": 2.4755e-05, "loss": 0.0194, "step": 5056 }, { "epoch": 6.62, "learning_rate": 2.4750000000000002e-05, "loss": 0.0233, "step": 5057 }, { "epoch": 6.62, "learning_rate": 2.4745e-05, "loss": 0.0253, "step": 5058 }, { "epoch": 6.62, "learning_rate": 2.4740000000000004e-05, "loss": 0.0214, "step": 5059 }, { "epoch": 6.62, "learning_rate": 2.4735e-05, "loss": 0.0246, "step": 5060 }, { "epoch": 6.62, "learning_rate": 2.473e-05, "loss": 0.024, "step": 5061 }, { "epoch": 6.63, "learning_rate": 2.4725e-05, "loss": 0.0235, "step": 5062 }, { "epoch": 6.63, "learning_rate": 2.472e-05, "loss": 0.0223, "step": 5063 }, { "epoch": 6.63, "learning_rate": 2.4715000000000003e-05, "loss": 0.0199, "step": 5064 }, { "epoch": 6.63, "learning_rate": 2.471e-05, "loss": 0.0265, "step": 5065 }, { "epoch": 6.63, "learning_rate": 2.4705e-05, "loss": 0.0209, "step": 5066 }, { "epoch": 6.63, "learning_rate": 2.47e-05, "loss": 0.0279, "step": 5067 }, { "epoch": 6.63, "learning_rate": 2.4695e-05, "loss": 0.0233, "step": 5068 }, { "epoch": 6.63, "learning_rate": 2.4690000000000002e-05, "loss": 0.0173, "step": 5069 }, { "epoch": 6.64, "learning_rate": 2.4685e-05, "loss": 0.0189, "step": 5070 }, { "epoch": 6.64, "learning_rate": 2.468e-05, "loss": 0.0234, "step": 5071 }, { "epoch": 6.64, "learning_rate": 2.4675e-05, "loss": 0.0212, "step": 5072 }, { "epoch": 6.64, "learning_rate": 2.4670000000000003e-05, "loss": 0.0276, "step": 5073 }, { "epoch": 6.64, "learning_rate": 2.4665000000000002e-05, "loss": 0.0172, "step": 5074 }, { "epoch": 6.64, "learning_rate": 2.466e-05, "loss": 0.0231, "step": 5075 }, { "epoch": 6.64, "learning_rate": 2.4655e-05, "loss": 0.0179, "step": 5076 }, { "epoch": 6.65, "learning_rate": 2.465e-05, "loss": 0.0185, "step": 5077 }, { "epoch": 6.65, "learning_rate": 2.4645000000000002e-05, "loss": 0.0218, "step": 5078 }, { "epoch": 6.65, "learning_rate": 2.464e-05, "loss": 0.0209, "step": 5079 }, { "epoch": 6.65, "learning_rate": 2.4635000000000004e-05, "loss": 0.0212, "step": 5080 }, { "epoch": 6.65, "learning_rate": 2.463e-05, "loss": 0.0195, "step": 5081 }, { "epoch": 6.65, "learning_rate": 2.4625000000000002e-05, "loss": 0.0269, "step": 5082 }, { "epoch": 6.65, "learning_rate": 2.462e-05, "loss": 0.0201, "step": 5083 }, { "epoch": 6.65, "learning_rate": 2.4615e-05, "loss": 0.0203, "step": 5084 }, { "epoch": 6.66, "learning_rate": 2.4610000000000003e-05, "loss": 0.0211, "step": 5085 }, { "epoch": 6.66, "learning_rate": 2.4605e-05, "loss": 0.0225, "step": 5086 }, { "epoch": 6.66, "learning_rate": 2.46e-05, "loss": 0.0203, "step": 5087 }, { "epoch": 6.66, "learning_rate": 2.4595e-05, "loss": 0.0224, "step": 5088 }, { "epoch": 6.66, "learning_rate": 2.4590000000000003e-05, "loss": 0.0231, "step": 5089 }, { "epoch": 6.66, "learning_rate": 2.4585000000000003e-05, "loss": 0.0212, "step": 5090 }, { "epoch": 6.66, "learning_rate": 2.4580000000000002e-05, "loss": 0.0204, "step": 5091 }, { "epoch": 6.66, "learning_rate": 2.4575e-05, "loss": 0.0236, "step": 5092 }, { "epoch": 6.67, "learning_rate": 2.457e-05, "loss": 0.0211, "step": 5093 }, { "epoch": 6.67, "learning_rate": 2.4565000000000003e-05, "loss": 0.0243, "step": 5094 }, { "epoch": 6.67, "learning_rate": 2.4560000000000002e-05, "loss": 0.0244, "step": 5095 }, { "epoch": 6.67, "learning_rate": 2.4555e-05, "loss": 0.0172, "step": 5096 }, { "epoch": 6.67, "learning_rate": 2.455e-05, "loss": 0.0196, "step": 5097 }, { "epoch": 6.67, "learning_rate": 2.4545000000000003e-05, "loss": 0.0233, "step": 5098 }, { "epoch": 6.67, "learning_rate": 2.4540000000000002e-05, "loss": 0.0177, "step": 5099 }, { "epoch": 6.68, "learning_rate": 2.4535e-05, "loss": 0.0198, "step": 5100 }, { "epoch": 6.68, "learning_rate": 2.453e-05, "loss": 0.021, "step": 5101 }, { "epoch": 6.68, "learning_rate": 2.4525e-05, "loss": 0.0205, "step": 5102 }, { "epoch": 6.68, "learning_rate": 2.4520000000000002e-05, "loss": 0.0171, "step": 5103 }, { "epoch": 6.68, "learning_rate": 2.4515e-05, "loss": 0.024, "step": 5104 }, { "epoch": 6.68, "learning_rate": 2.451e-05, "loss": 0.0236, "step": 5105 }, { "epoch": 6.68, "learning_rate": 2.4505e-05, "loss": 0.022, "step": 5106 }, { "epoch": 6.68, "learning_rate": 2.45e-05, "loss": 0.0217, "step": 5107 }, { "epoch": 6.69, "learning_rate": 2.4495000000000002e-05, "loss": 0.0233, "step": 5108 }, { "epoch": 6.69, "learning_rate": 2.449e-05, "loss": 0.0225, "step": 5109 }, { "epoch": 6.69, "learning_rate": 2.4485000000000004e-05, "loss": 0.0213, "step": 5110 }, { "epoch": 6.69, "learning_rate": 2.448e-05, "loss": 0.0206, "step": 5111 }, { "epoch": 6.69, "learning_rate": 2.4475000000000002e-05, "loss": 0.023, "step": 5112 }, { "epoch": 6.69, "learning_rate": 2.447e-05, "loss": 0.0225, "step": 5113 }, { "epoch": 6.69, "learning_rate": 2.4465e-05, "loss": 0.0231, "step": 5114 }, { "epoch": 6.7, "learning_rate": 2.4460000000000003e-05, "loss": 0.0239, "step": 5115 }, { "epoch": 6.7, "learning_rate": 2.4455e-05, "loss": 0.0219, "step": 5116 }, { "epoch": 6.7, "learning_rate": 2.445e-05, "loss": 0.0219, "step": 5117 }, { "epoch": 6.7, "learning_rate": 2.4445e-05, "loss": 0.0222, "step": 5118 }, { "epoch": 6.7, "learning_rate": 2.4440000000000003e-05, "loss": 0.0251, "step": 5119 }, { "epoch": 6.7, "learning_rate": 2.4435000000000002e-05, "loss": 0.0216, "step": 5120 }, { "epoch": 6.7, "learning_rate": 2.443e-05, "loss": 0.0194, "step": 5121 }, { "epoch": 6.7, "learning_rate": 2.4425e-05, "loss": 0.0219, "step": 5122 }, { "epoch": 6.71, "learning_rate": 2.442e-05, "loss": 0.0253, "step": 5123 }, { "epoch": 6.71, "learning_rate": 2.4415000000000003e-05, "loss": 0.0202, "step": 5124 }, { "epoch": 6.71, "learning_rate": 2.4410000000000002e-05, "loss": 0.0182, "step": 5125 }, { "epoch": 6.71, "learning_rate": 2.4405e-05, "loss": 0.0195, "step": 5126 }, { "epoch": 6.71, "learning_rate": 2.44e-05, "loss": 0.0258, "step": 5127 }, { "epoch": 6.71, "learning_rate": 2.4395000000000003e-05, "loss": 0.0231, "step": 5128 }, { "epoch": 6.71, "learning_rate": 2.4390000000000002e-05, "loss": 0.0178, "step": 5129 }, { "epoch": 6.71, "learning_rate": 2.4385e-05, "loss": 0.0238, "step": 5130 }, { "epoch": 6.72, "learning_rate": 2.438e-05, "loss": 0.0244, "step": 5131 }, { "epoch": 6.72, "learning_rate": 2.4375e-05, "loss": 0.0232, "step": 5132 }, { "epoch": 6.72, "learning_rate": 2.4370000000000002e-05, "loss": 0.0222, "step": 5133 }, { "epoch": 6.72, "learning_rate": 2.4365e-05, "loss": 0.0228, "step": 5134 }, { "epoch": 6.72, "learning_rate": 2.4360000000000004e-05, "loss": 0.0231, "step": 5135 }, { "epoch": 6.72, "learning_rate": 2.4355e-05, "loss": 0.0169, "step": 5136 }, { "epoch": 6.72, "learning_rate": 2.435e-05, "loss": 0.0228, "step": 5137 }, { "epoch": 6.73, "learning_rate": 2.4345e-05, "loss": 0.0177, "step": 5138 }, { "epoch": 6.73, "learning_rate": 2.434e-05, "loss": 0.022, "step": 5139 }, { "epoch": 6.73, "learning_rate": 2.4335000000000003e-05, "loss": 0.0209, "step": 5140 }, { "epoch": 6.73, "learning_rate": 2.433e-05, "loss": 0.0262, "step": 5141 }, { "epoch": 6.73, "learning_rate": 2.4325000000000002e-05, "loss": 0.0198, "step": 5142 }, { "epoch": 6.73, "learning_rate": 2.432e-05, "loss": 0.0233, "step": 5143 }, { "epoch": 6.73, "learning_rate": 2.4315e-05, "loss": 0.0209, "step": 5144 }, { "epoch": 6.73, "learning_rate": 2.4310000000000003e-05, "loss": 0.018, "step": 5145 }, { "epoch": 6.74, "learning_rate": 2.4305e-05, "loss": 0.0233, "step": 5146 }, { "epoch": 6.74, "learning_rate": 2.43e-05, "loss": 0.0281, "step": 5147 }, { "epoch": 6.74, "learning_rate": 2.4295e-05, "loss": 0.0246, "step": 5148 }, { "epoch": 6.74, "learning_rate": 2.4290000000000003e-05, "loss": 0.0253, "step": 5149 }, { "epoch": 6.74, "learning_rate": 2.4285000000000002e-05, "loss": 0.0198, "step": 5150 }, { "epoch": 6.74, "learning_rate": 2.428e-05, "loss": 0.0244, "step": 5151 }, { "epoch": 6.74, "learning_rate": 2.4275e-05, "loss": 0.0228, "step": 5152 }, { "epoch": 6.74, "learning_rate": 2.427e-05, "loss": 0.0202, "step": 5153 }, { "epoch": 6.75, "learning_rate": 2.4265000000000002e-05, "loss": 0.0196, "step": 5154 }, { "epoch": 6.75, "learning_rate": 2.426e-05, "loss": 0.021, "step": 5155 }, { "epoch": 6.75, "learning_rate": 2.4255e-05, "loss": 0.0254, "step": 5156 }, { "epoch": 6.75, "learning_rate": 2.425e-05, "loss": 0.0251, "step": 5157 }, { "epoch": 6.75, "learning_rate": 2.4245000000000002e-05, "loss": 0.0224, "step": 5158 }, { "epoch": 6.75, "learning_rate": 2.4240000000000002e-05, "loss": 0.0247, "step": 5159 }, { "epoch": 6.75, "learning_rate": 2.4235e-05, "loss": 0.0216, "step": 5160 }, { "epoch": 6.76, "learning_rate": 2.423e-05, "loss": 0.0202, "step": 5161 }, { "epoch": 6.76, "learning_rate": 2.4225e-05, "loss": 0.0291, "step": 5162 }, { "epoch": 6.76, "learning_rate": 2.4220000000000002e-05, "loss": 0.0246, "step": 5163 }, { "epoch": 6.76, "learning_rate": 2.4215e-05, "loss": 0.0213, "step": 5164 }, { "epoch": 6.76, "learning_rate": 2.4210000000000004e-05, "loss": 0.0216, "step": 5165 }, { "epoch": 6.76, "learning_rate": 2.4205e-05, "loss": 0.019, "step": 5166 }, { "epoch": 6.76, "learning_rate": 2.4200000000000002e-05, "loss": 0.023, "step": 5167 }, { "epoch": 6.76, "learning_rate": 2.4195e-05, "loss": 0.0237, "step": 5168 }, { "epoch": 6.77, "learning_rate": 2.419e-05, "loss": 0.0248, "step": 5169 }, { "epoch": 6.77, "learning_rate": 2.4185000000000003e-05, "loss": 0.023, "step": 5170 }, { "epoch": 6.77, "learning_rate": 2.418e-05, "loss": 0.021, "step": 5171 }, { "epoch": 6.77, "learning_rate": 2.4175e-05, "loss": 0.0166, "step": 5172 }, { "epoch": 6.77, "learning_rate": 2.417e-05, "loss": 0.0206, "step": 5173 }, { "epoch": 6.77, "learning_rate": 2.4165e-05, "loss": 0.0179, "step": 5174 }, { "epoch": 6.77, "learning_rate": 2.4160000000000002e-05, "loss": 0.0184, "step": 5175 }, { "epoch": 6.77, "learning_rate": 2.4154999999999998e-05, "loss": 0.0192, "step": 5176 }, { "epoch": 6.78, "learning_rate": 2.415e-05, "loss": 0.1087, "step": 5177 }, { "epoch": 6.78, "learning_rate": 2.4145e-05, "loss": 0.0215, "step": 5178 }, { "epoch": 6.78, "learning_rate": 2.4140000000000003e-05, "loss": 0.0249, "step": 5179 }, { "epoch": 6.78, "learning_rate": 2.4135000000000002e-05, "loss": 0.0188, "step": 5180 }, { "epoch": 6.78, "learning_rate": 2.413e-05, "loss": 0.023, "step": 5181 }, { "epoch": 6.78, "learning_rate": 2.413e-05, "loss": 0.0987, "step": 5182 }, { "epoch": 6.78, "learning_rate": 2.4125e-05, "loss": 0.0207, "step": 5183 }, { "epoch": 6.79, "learning_rate": 2.412e-05, "loss": 0.0231, "step": 5184 }, { "epoch": 6.79, "learning_rate": 2.4115000000000002e-05, "loss": 0.026, "step": 5185 }, { "epoch": 6.79, "learning_rate": 2.411e-05, "loss": 0.0209, "step": 5186 }, { "epoch": 6.79, "learning_rate": 2.4105e-05, "loss": 0.0235, "step": 5187 }, { "epoch": 6.79, "learning_rate": 2.41e-05, "loss": 0.0205, "step": 5188 }, { "epoch": 6.79, "learning_rate": 2.4095000000000002e-05, "loss": 0.0233, "step": 5189 }, { "epoch": 6.79, "learning_rate": 2.409e-05, "loss": 0.0219, "step": 5190 }, { "epoch": 6.79, "learning_rate": 2.4085e-05, "loss": 0.0249, "step": 5191 }, { "epoch": 6.8, "learning_rate": 2.408e-05, "loss": 0.0224, "step": 5192 }, { "epoch": 6.8, "learning_rate": 2.4075e-05, "loss": 0.0227, "step": 5193 }, { "epoch": 6.8, "learning_rate": 2.407e-05, "loss": 0.022, "step": 5194 }, { "epoch": 6.8, "learning_rate": 2.4065e-05, "loss": 0.0201, "step": 5195 }, { "epoch": 6.8, "learning_rate": 2.4060000000000003e-05, "loss": 0.0194, "step": 5196 }, { "epoch": 6.8, "learning_rate": 2.4055000000000003e-05, "loss": 0.0195, "step": 5197 }, { "epoch": 6.8, "learning_rate": 2.4050000000000002e-05, "loss": 0.0221, "step": 5198 }, { "epoch": 6.8, "learning_rate": 2.4045e-05, "loss": 0.0208, "step": 5199 }, { "epoch": 6.81, "learning_rate": 2.404e-05, "loss": 0.0208, "step": 5200 }, { "epoch": 6.81, "learning_rate": 2.4035000000000003e-05, "loss": 0.0225, "step": 5201 }, { "epoch": 6.81, "learning_rate": 2.4030000000000002e-05, "loss": 0.0223, "step": 5202 }, { "epoch": 6.81, "learning_rate": 2.4025e-05, "loss": 0.0218, "step": 5203 }, { "epoch": 6.81, "learning_rate": 2.402e-05, "loss": 0.0293, "step": 5204 }, { "epoch": 6.81, "learning_rate": 2.4015000000000003e-05, "loss": 0.0208, "step": 5205 }, { "epoch": 6.81, "learning_rate": 2.4010000000000002e-05, "loss": 0.0235, "step": 5206 }, { "epoch": 6.82, "learning_rate": 2.4005e-05, "loss": 0.0244, "step": 5207 }, { "epoch": 6.82, "learning_rate": 2.4e-05, "loss": 0.0262, "step": 5208 }, { "epoch": 6.82, "learning_rate": 2.3995e-05, "loss": 0.0222, "step": 5209 }, { "epoch": 6.82, "learning_rate": 2.3990000000000002e-05, "loss": 0.0224, "step": 5210 }, { "epoch": 6.82, "learning_rate": 2.3985e-05, "loss": 0.0226, "step": 5211 }, { "epoch": 6.82, "learning_rate": 2.398e-05, "loss": 0.0239, "step": 5212 }, { "epoch": 6.82, "learning_rate": 2.3975e-05, "loss": 0.0204, "step": 5213 }, { "epoch": 6.82, "learning_rate": 2.397e-05, "loss": 0.0269, "step": 5214 }, { "epoch": 6.83, "learning_rate": 2.3965000000000002e-05, "loss": 0.0208, "step": 5215 }, { "epoch": 6.83, "learning_rate": 2.396e-05, "loss": 0.0244, "step": 5216 }, { "epoch": 6.83, "learning_rate": 2.3955000000000004e-05, "loss": 0.0217, "step": 5217 }, { "epoch": 6.83, "learning_rate": 2.395e-05, "loss": 0.0204, "step": 5218 }, { "epoch": 6.83, "learning_rate": 2.3945000000000002e-05, "loss": 0.0226, "step": 5219 }, { "epoch": 6.83, "learning_rate": 2.394e-05, "loss": 0.0208, "step": 5220 }, { "epoch": 6.83, "learning_rate": 2.3935e-05, "loss": 0.0205, "step": 5221 }, { "epoch": 6.84, "learning_rate": 2.3930000000000003e-05, "loss": 0.0195, "step": 5222 }, { "epoch": 6.84, "learning_rate": 2.3925e-05, "loss": 0.024, "step": 5223 }, { "epoch": 6.84, "learning_rate": 2.392e-05, "loss": 0.024, "step": 5224 }, { "epoch": 6.84, "learning_rate": 2.3915e-05, "loss": 0.0196, "step": 5225 }, { "epoch": 6.84, "learning_rate": 2.3910000000000003e-05, "loss": 0.0208, "step": 5226 }, { "epoch": 6.84, "learning_rate": 2.3905000000000002e-05, "loss": 0.0204, "step": 5227 }, { "epoch": 6.84, "learning_rate": 2.39e-05, "loss": 0.0207, "step": 5228 }, { "epoch": 6.84, "learning_rate": 2.3895e-05, "loss": 0.0218, "step": 5229 }, { "epoch": 6.85, "learning_rate": 2.389e-05, "loss": 0.0209, "step": 5230 }, { "epoch": 6.85, "learning_rate": 2.3885000000000003e-05, "loss": 0.0192, "step": 5231 }, { "epoch": 6.85, "learning_rate": 2.3880000000000002e-05, "loss": 0.0173, "step": 5232 }, { "epoch": 6.85, "learning_rate": 2.3875e-05, "loss": 0.0278, "step": 5233 }, { "epoch": 6.85, "learning_rate": 2.387e-05, "loss": 0.0192, "step": 5234 }, { "epoch": 6.85, "learning_rate": 2.3865000000000003e-05, "loss": 0.0215, "step": 5235 }, { "epoch": 6.85, "learning_rate": 2.3860000000000002e-05, "loss": 0.0206, "step": 5236 }, { "epoch": 6.85, "learning_rate": 2.3855e-05, "loss": 0.0195, "step": 5237 }, { "epoch": 6.86, "learning_rate": 2.385e-05, "loss": 0.0178, "step": 5238 }, { "epoch": 6.86, "learning_rate": 2.3845e-05, "loss": 0.0177, "step": 5239 }, { "epoch": 6.86, "learning_rate": 2.3840000000000002e-05, "loss": 0.0255, "step": 5240 }, { "epoch": 6.86, "learning_rate": 2.3835e-05, "loss": 0.0238, "step": 5241 }, { "epoch": 6.86, "learning_rate": 2.3830000000000004e-05, "loss": 0.0205, "step": 5242 }, { "epoch": 6.86, "learning_rate": 2.3825e-05, "loss": 0.0244, "step": 5243 }, { "epoch": 6.86, "learning_rate": 2.3820000000000002e-05, "loss": 0.0219, "step": 5244 }, { "epoch": 6.87, "learning_rate": 2.3815e-05, "loss": 0.0243, "step": 5245 }, { "epoch": 6.87, "learning_rate": 2.381e-05, "loss": 0.0217, "step": 5246 }, { "epoch": 6.87, "learning_rate": 2.3805000000000003e-05, "loss": 0.0215, "step": 5247 }, { "epoch": 6.87, "learning_rate": 2.38e-05, "loss": 0.0268, "step": 5248 }, { "epoch": 6.87, "learning_rate": 2.3795000000000002e-05, "loss": 0.0214, "step": 5249 }, { "epoch": 6.87, "learning_rate": 2.379e-05, "loss": 0.0224, "step": 5250 }, { "epoch": 6.87, "learning_rate": 2.3785e-05, "loss": 0.0206, "step": 5251 }, { "epoch": 6.87, "learning_rate": 2.3780000000000003e-05, "loss": 0.0233, "step": 5252 }, { "epoch": 6.88, "learning_rate": 2.3775e-05, "loss": 0.0188, "step": 5253 }, { "epoch": 6.88, "learning_rate": 2.377e-05, "loss": 0.0223, "step": 5254 }, { "epoch": 6.88, "learning_rate": 2.3765e-05, "loss": 0.0227, "step": 5255 }, { "epoch": 6.88, "learning_rate": 2.3760000000000003e-05, "loss": 0.0247, "step": 5256 }, { "epoch": 6.88, "learning_rate": 2.3755000000000002e-05, "loss": 0.0208, "step": 5257 }, { "epoch": 6.88, "learning_rate": 2.375e-05, "loss": 0.0239, "step": 5258 }, { "epoch": 6.88, "learning_rate": 2.3745e-05, "loss": 0.0255, "step": 5259 }, { "epoch": 6.88, "learning_rate": 2.374e-05, "loss": 0.0265, "step": 5260 }, { "epoch": 6.89, "learning_rate": 2.3735000000000002e-05, "loss": 0.0213, "step": 5261 }, { "epoch": 6.89, "learning_rate": 2.373e-05, "loss": 0.023, "step": 5262 }, { "epoch": 6.89, "learning_rate": 2.3725e-05, "loss": 0.0214, "step": 5263 }, { "epoch": 6.89, "learning_rate": 2.372e-05, "loss": 0.0231, "step": 5264 }, { "epoch": 6.89, "learning_rate": 2.3715000000000002e-05, "loss": 0.0249, "step": 5265 }, { "epoch": 6.89, "learning_rate": 2.371e-05, "loss": 0.0183, "step": 5266 }, { "epoch": 6.89, "learning_rate": 2.3705e-05, "loss": 0.0232, "step": 5267 }, { "epoch": 6.9, "learning_rate": 2.37e-05, "loss": 0.0267, "step": 5268 }, { "epoch": 6.9, "learning_rate": 2.3695e-05, "loss": 0.0251, "step": 5269 }, { "epoch": 6.9, "learning_rate": 2.3690000000000002e-05, "loss": 0.0222, "step": 5270 }, { "epoch": 6.9, "learning_rate": 2.3685e-05, "loss": 0.0191, "step": 5271 }, { "epoch": 6.9, "learning_rate": 2.3680000000000004e-05, "loss": 0.0208, "step": 5272 }, { "epoch": 6.9, "learning_rate": 2.3675e-05, "loss": 0.0266, "step": 5273 }, { "epoch": 6.9, "learning_rate": 2.3670000000000002e-05, "loss": 0.0202, "step": 5274 }, { "epoch": 6.9, "learning_rate": 2.3665e-05, "loss": 0.0222, "step": 5275 }, { "epoch": 6.91, "learning_rate": 2.366e-05, "loss": 0.0207, "step": 5276 }, { "epoch": 6.91, "learning_rate": 2.3655000000000003e-05, "loss": 0.022, "step": 5277 }, { "epoch": 6.91, "learning_rate": 2.365e-05, "loss": 0.0224, "step": 5278 }, { "epoch": 6.91, "learning_rate": 2.3645e-05, "loss": 0.0269, "step": 5279 }, { "epoch": 6.91, "learning_rate": 2.364e-05, "loss": 0.0222, "step": 5280 }, { "epoch": 6.91, "learning_rate": 2.3635000000000003e-05, "loss": 0.0213, "step": 5281 }, { "epoch": 6.91, "learning_rate": 2.3630000000000002e-05, "loss": 0.0219, "step": 5282 }, { "epoch": 6.91, "learning_rate": 2.3624999999999998e-05, "loss": 0.0266, "step": 5283 }, { "epoch": 6.92, "learning_rate": 2.362e-05, "loss": 0.0235, "step": 5284 }, { "epoch": 6.92, "learning_rate": 2.3615e-05, "loss": 0.0232, "step": 5285 }, { "epoch": 6.92, "learning_rate": 2.3610000000000003e-05, "loss": 0.0222, "step": 5286 }, { "epoch": 6.92, "learning_rate": 2.3605000000000002e-05, "loss": 0.0216, "step": 5287 }, { "epoch": 6.92, "learning_rate": 2.36e-05, "loss": 0.0232, "step": 5288 }, { "epoch": 6.92, "learning_rate": 2.3595e-05, "loss": 0.0249, "step": 5289 }, { "epoch": 6.92, "learning_rate": 2.359e-05, "loss": 0.0193, "step": 5290 }, { "epoch": 6.93, "learning_rate": 2.3585000000000002e-05, "loss": 0.0184, "step": 5291 }, { "epoch": 6.93, "learning_rate": 2.358e-05, "loss": 0.0264, "step": 5292 }, { "epoch": 6.93, "learning_rate": 2.3575e-05, "loss": 0.0207, "step": 5293 }, { "epoch": 6.93, "learning_rate": 2.357e-05, "loss": 0.02, "step": 5294 }, { "epoch": 6.93, "learning_rate": 2.3565000000000002e-05, "loss": 0.0222, "step": 5295 }, { "epoch": 6.93, "learning_rate": 2.356e-05, "loss": 0.0212, "step": 5296 }, { "epoch": 6.93, "learning_rate": 2.3555e-05, "loss": 0.0204, "step": 5297 }, { "epoch": 6.93, "learning_rate": 2.355e-05, "loss": 0.0228, "step": 5298 }, { "epoch": 6.94, "learning_rate": 2.3545e-05, "loss": 0.0222, "step": 5299 }, { "epoch": 6.94, "learning_rate": 2.354e-05, "loss": 0.0232, "step": 5300 }, { "epoch": 6.94, "learning_rate": 2.3535e-05, "loss": 0.0255, "step": 5301 }, { "epoch": 6.94, "learning_rate": 2.3530000000000003e-05, "loss": 0.0187, "step": 5302 }, { "epoch": 6.94, "learning_rate": 2.3525e-05, "loss": 0.026, "step": 5303 }, { "epoch": 6.94, "learning_rate": 2.3520000000000002e-05, "loss": 0.0215, "step": 5304 }, { "epoch": 6.94, "learning_rate": 2.3515e-05, "loss": 0.024, "step": 5305 }, { "epoch": 6.95, "learning_rate": 2.351e-05, "loss": 0.0188, "step": 5306 }, { "epoch": 6.95, "learning_rate": 2.3505000000000003e-05, "loss": 0.0233, "step": 5307 }, { "epoch": 6.95, "learning_rate": 2.35e-05, "loss": 0.0239, "step": 5308 }, { "epoch": 6.95, "learning_rate": 2.3495e-05, "loss": 0.0245, "step": 5309 }, { "epoch": 6.95, "learning_rate": 2.349e-05, "loss": 0.0283, "step": 5310 }, { "epoch": 6.95, "learning_rate": 2.3485000000000003e-05, "loss": 0.0179, "step": 5311 }, { "epoch": 6.95, "learning_rate": 2.3480000000000002e-05, "loss": 0.0194, "step": 5312 }, { "epoch": 6.95, "learning_rate": 2.3475e-05, "loss": 0.0214, "step": 5313 }, { "epoch": 6.96, "learning_rate": 2.347e-05, "loss": 0.0176, "step": 5314 }, { "epoch": 6.96, "learning_rate": 2.3465e-05, "loss": 0.0192, "step": 5315 }, { "epoch": 6.96, "learning_rate": 2.3460000000000002e-05, "loss": 0.0197, "step": 5316 }, { "epoch": 6.96, "learning_rate": 2.3455e-05, "loss": 0.024, "step": 5317 }, { "epoch": 6.96, "learning_rate": 2.345e-05, "loss": 0.0185, "step": 5318 }, { "epoch": 6.96, "learning_rate": 2.3445e-05, "loss": 0.0209, "step": 5319 }, { "epoch": 6.96, "learning_rate": 2.344e-05, "loss": 0.0202, "step": 5320 }, { "epoch": 6.96, "learning_rate": 2.3435000000000002e-05, "loss": 0.0223, "step": 5321 }, { "epoch": 6.97, "learning_rate": 2.343e-05, "loss": 0.0236, "step": 5322 }, { "epoch": 6.97, "learning_rate": 2.3425000000000004e-05, "loss": 0.0262, "step": 5323 }, { "epoch": 6.97, "learning_rate": 2.342e-05, "loss": 0.0278, "step": 5324 }, { "epoch": 6.97, "learning_rate": 2.3415000000000002e-05, "loss": 0.0221, "step": 5325 }, { "epoch": 6.97, "learning_rate": 2.341e-05, "loss": 0.0261, "step": 5326 }, { "epoch": 6.97, "learning_rate": 2.3405e-05, "loss": 0.0243, "step": 5327 }, { "epoch": 6.97, "learning_rate": 2.3400000000000003e-05, "loss": 0.0249, "step": 5328 }, { "epoch": 6.98, "learning_rate": 2.3395e-05, "loss": 0.022, "step": 5329 }, { "epoch": 6.98, "learning_rate": 2.339e-05, "loss": 0.0208, "step": 5330 }, { "epoch": 6.98, "learning_rate": 2.3385e-05, "loss": 0.0193, "step": 5331 }, { "epoch": 6.98, "learning_rate": 2.3380000000000003e-05, "loss": 0.0249, "step": 5332 }, { "epoch": 6.98, "learning_rate": 2.3375000000000002e-05, "loss": 0.0163, "step": 5333 }, { "epoch": 6.98, "learning_rate": 2.337e-05, "loss": 0.0262, "step": 5334 }, { "epoch": 6.98, "learning_rate": 2.3365e-05, "loss": 0.0278, "step": 5335 }, { "epoch": 6.98, "learning_rate": 2.336e-05, "loss": 0.0204, "step": 5336 }, { "epoch": 6.99, "learning_rate": 2.3355000000000003e-05, "loss": 0.0262, "step": 5337 }, { "epoch": 6.99, "learning_rate": 2.3350000000000002e-05, "loss": 0.0283, "step": 5338 }, { "epoch": 6.99, "learning_rate": 2.3345e-05, "loss": 0.0177, "step": 5339 }, { "epoch": 6.99, "learning_rate": 2.334e-05, "loss": 0.0237, "step": 5340 }, { "epoch": 6.99, "learning_rate": 2.3335000000000003e-05, "loss": 0.018, "step": 5341 }, { "epoch": 6.99, "learning_rate": 2.3330000000000002e-05, "loss": 0.0208, "step": 5342 }, { "epoch": 6.99, "learning_rate": 2.3325e-05, "loss": 0.0189, "step": 5343 }, { "epoch": 6.99, "learning_rate": 2.332e-05, "loss": 0.0216, "step": 5344 }, { "epoch": 7.0, "learning_rate": 2.3315e-05, "loss": 0.0238, "step": 5345 }, { "epoch": 7.0, "learning_rate": 2.3310000000000002e-05, "loss": 0.0206, "step": 5346 }, { "epoch": 7.0, "learning_rate": 2.3305e-05, "loss": 0.0232, "step": 5347 }, { "epoch": 7.0, "learning_rate": 2.3300000000000004e-05, "loss": 0.0262, "step": 5348 }, { "epoch": 7.0, "learning_rate": 2.3295e-05, "loss": 0.0137, "step": 5349 }, { "epoch": 7.0, "learning_rate": 2.3290000000000002e-05, "loss": 0.0131, "step": 5350 }, { "epoch": 7.0, "learning_rate": 2.3285e-05, "loss": 0.0137, "step": 5351 }, { "epoch": 7.01, "learning_rate": 2.328e-05, "loss": 0.0132, "step": 5352 }, { "epoch": 7.01, "learning_rate": 2.3275000000000003e-05, "loss": 0.0141, "step": 5353 }, { "epoch": 7.01, "learning_rate": 2.327e-05, "loss": 0.0107, "step": 5354 }, { "epoch": 7.01, "learning_rate": 2.3265000000000002e-05, "loss": 0.0102, "step": 5355 }, { "epoch": 7.01, "learning_rate": 2.326e-05, "loss": 0.0114, "step": 5356 }, { "epoch": 7.01, "learning_rate": 2.3255e-05, "loss": 0.0115, "step": 5357 }, { "epoch": 7.01, "learning_rate": 2.3250000000000003e-05, "loss": 0.0104, "step": 5358 }, { "epoch": 7.01, "learning_rate": 2.3245e-05, "loss": 0.0104, "step": 5359 }, { "epoch": 7.02, "learning_rate": 2.324e-05, "loss": 0.0144, "step": 5360 }, { "epoch": 7.02, "learning_rate": 2.3235e-05, "loss": 0.0151, "step": 5361 }, { "epoch": 7.02, "learning_rate": 2.3230000000000003e-05, "loss": 0.0117, "step": 5362 }, { "epoch": 7.02, "learning_rate": 2.3225000000000002e-05, "loss": 0.0166, "step": 5363 }, { "epoch": 7.02, "learning_rate": 2.322e-05, "loss": 0.0121, "step": 5364 }, { "epoch": 7.02, "learning_rate": 2.3215e-05, "loss": 0.0132, "step": 5365 }, { "epoch": 7.02, "learning_rate": 2.321e-05, "loss": 0.0125, "step": 5366 }, { "epoch": 7.02, "learning_rate": 2.3205000000000002e-05, "loss": 0.0141, "step": 5367 }, { "epoch": 7.03, "learning_rate": 2.32e-05, "loss": 0.0103, "step": 5368 }, { "epoch": 7.03, "learning_rate": 2.3195e-05, "loss": 0.0121, "step": 5369 }, { "epoch": 7.03, "learning_rate": 2.319e-05, "loss": 0.0133, "step": 5370 }, { "epoch": 7.03, "learning_rate": 2.3185000000000002e-05, "loss": 0.0142, "step": 5371 }, { "epoch": 7.03, "learning_rate": 2.318e-05, "loss": 0.0134, "step": 5372 }, { "epoch": 7.03, "learning_rate": 2.3175e-05, "loss": 0.0109, "step": 5373 }, { "epoch": 7.03, "learning_rate": 2.317e-05, "loss": 0.0132, "step": 5374 }, { "epoch": 7.04, "learning_rate": 2.3165e-05, "loss": 0.0107, "step": 5375 }, { "epoch": 7.04, "learning_rate": 2.3160000000000002e-05, "loss": 0.0117, "step": 5376 }, { "epoch": 7.04, "learning_rate": 2.3155e-05, "loss": 0.0133, "step": 5377 }, { "epoch": 7.04, "learning_rate": 2.3150000000000004e-05, "loss": 0.012, "step": 5378 }, { "epoch": 7.04, "learning_rate": 2.3145e-05, "loss": 0.0092, "step": 5379 }, { "epoch": 7.04, "learning_rate": 2.3140000000000002e-05, "loss": 0.0134, "step": 5380 }, { "epoch": 7.04, "learning_rate": 2.3135e-05, "loss": 0.0111, "step": 5381 }, { "epoch": 7.04, "learning_rate": 2.313e-05, "loss": 0.0126, "step": 5382 }, { "epoch": 7.05, "learning_rate": 2.3125000000000003e-05, "loss": 0.0105, "step": 5383 }, { "epoch": 7.05, "learning_rate": 2.312e-05, "loss": 0.0136, "step": 5384 }, { "epoch": 7.05, "learning_rate": 2.3115e-05, "loss": 0.0125, "step": 5385 }, { "epoch": 7.05, "learning_rate": 2.311e-05, "loss": 0.0112, "step": 5386 }, { "epoch": 7.05, "learning_rate": 2.3105000000000003e-05, "loss": 0.0084, "step": 5387 }, { "epoch": 7.05, "learning_rate": 2.3100000000000002e-05, "loss": 0.0132, "step": 5388 }, { "epoch": 7.05, "learning_rate": 2.3095e-05, "loss": 0.0146, "step": 5389 }, { "epoch": 7.05, "learning_rate": 2.309e-05, "loss": 0.0138, "step": 5390 }, { "epoch": 7.06, "learning_rate": 2.3085e-05, "loss": 0.0118, "step": 5391 }, { "epoch": 7.06, "learning_rate": 2.3080000000000003e-05, "loss": 0.0113, "step": 5392 }, { "epoch": 7.06, "learning_rate": 2.3075000000000002e-05, "loss": 0.0148, "step": 5393 }, { "epoch": 7.06, "learning_rate": 2.307e-05, "loss": 0.0139, "step": 5394 }, { "epoch": 7.06, "learning_rate": 2.3065e-05, "loss": 0.0122, "step": 5395 }, { "epoch": 7.06, "learning_rate": 2.306e-05, "loss": 0.014, "step": 5396 }, { "epoch": 7.06, "learning_rate": 2.3055000000000002e-05, "loss": 0.0149, "step": 5397 }, { "epoch": 7.07, "learning_rate": 2.305e-05, "loss": 0.0127, "step": 5398 }, { "epoch": 7.07, "learning_rate": 2.3045e-05, "loss": 0.0151, "step": 5399 }, { "epoch": 7.07, "learning_rate": 2.304e-05, "loss": 0.0119, "step": 5400 }, { "epoch": 7.07, "learning_rate": 2.3035000000000002e-05, "loss": 0.0108, "step": 5401 }, { "epoch": 7.07, "learning_rate": 2.303e-05, "loss": 0.0126, "step": 5402 }, { "epoch": 7.07, "learning_rate": 2.3025e-05, "loss": 0.0125, "step": 5403 }, { "epoch": 7.07, "learning_rate": 2.302e-05, "loss": 0.0114, "step": 5404 }, { "epoch": 7.07, "learning_rate": 2.3015e-05, "loss": 0.0158, "step": 5405 }, { "epoch": 7.08, "learning_rate": 2.301e-05, "loss": 0.0168, "step": 5406 }, { "epoch": 7.08, "learning_rate": 2.3005e-05, "loss": 0.0148, "step": 5407 }, { "epoch": 7.08, "learning_rate": 2.3000000000000003e-05, "loss": 0.0113, "step": 5408 }, { "epoch": 7.08, "learning_rate": 2.2995e-05, "loss": 0.0107, "step": 5409 }, { "epoch": 7.08, "learning_rate": 2.2990000000000002e-05, "loss": 0.0123, "step": 5410 }, { "epoch": 7.08, "learning_rate": 2.2985e-05, "loss": 0.0112, "step": 5411 }, { "epoch": 7.08, "learning_rate": 2.298e-05, "loss": 0.0109, "step": 5412 }, { "epoch": 7.09, "learning_rate": 2.2975000000000003e-05, "loss": 0.0122, "step": 5413 }, { "epoch": 7.09, "learning_rate": 2.297e-05, "loss": 0.0085, "step": 5414 }, { "epoch": 7.09, "learning_rate": 2.2965e-05, "loss": 0.0128, "step": 5415 }, { "epoch": 7.09, "learning_rate": 2.296e-05, "loss": 0.0141, "step": 5416 }, { "epoch": 7.09, "learning_rate": 2.2955000000000003e-05, "loss": 0.0122, "step": 5417 }, { "epoch": 7.09, "learning_rate": 2.2950000000000002e-05, "loss": 0.011, "step": 5418 }, { "epoch": 7.09, "learning_rate": 2.2945e-05, "loss": 0.0109, "step": 5419 }, { "epoch": 7.09, "learning_rate": 2.294e-05, "loss": 0.0121, "step": 5420 }, { "epoch": 7.1, "learning_rate": 2.2935e-05, "loss": 0.0109, "step": 5421 }, { "epoch": 7.1, "learning_rate": 2.2930000000000002e-05, "loss": 0.012, "step": 5422 }, { "epoch": 7.1, "learning_rate": 2.2925e-05, "loss": 0.0148, "step": 5423 }, { "epoch": 7.1, "learning_rate": 2.292e-05, "loss": 0.0136, "step": 5424 }, { "epoch": 7.1, "learning_rate": 2.2915e-05, "loss": 0.0122, "step": 5425 }, { "epoch": 7.1, "learning_rate": 2.2910000000000003e-05, "loss": 0.0158, "step": 5426 }, { "epoch": 7.1, "learning_rate": 2.2905000000000002e-05, "loss": 0.0142, "step": 5427 }, { "epoch": 7.1, "learning_rate": 2.29e-05, "loss": 0.0137, "step": 5428 }, { "epoch": 7.11, "learning_rate": 2.2895e-05, "loss": 0.0119, "step": 5429 }, { "epoch": 7.11, "learning_rate": 2.289e-05, "loss": 0.0122, "step": 5430 }, { "epoch": 7.11, "learning_rate": 2.2885000000000002e-05, "loss": 0.0109, "step": 5431 }, { "epoch": 7.11, "learning_rate": 2.288e-05, "loss": 0.0134, "step": 5432 }, { "epoch": 7.11, "learning_rate": 2.2875e-05, "loss": 0.0129, "step": 5433 }, { "epoch": 7.11, "learning_rate": 2.287e-05, "loss": 0.0161, "step": 5434 }, { "epoch": 7.11, "learning_rate": 2.2865e-05, "loss": 0.0116, "step": 5435 }, { "epoch": 7.12, "learning_rate": 2.286e-05, "loss": 0.011, "step": 5436 }, { "epoch": 7.12, "learning_rate": 2.2855e-05, "loss": 0.0152, "step": 5437 }, { "epoch": 7.12, "learning_rate": 2.2850000000000003e-05, "loss": 0.0096, "step": 5438 }, { "epoch": 7.12, "learning_rate": 2.2845e-05, "loss": 0.0123, "step": 5439 }, { "epoch": 7.12, "learning_rate": 2.284e-05, "loss": 0.0113, "step": 5440 }, { "epoch": 7.12, "learning_rate": 2.2835e-05, "loss": 0.0115, "step": 5441 }, { "epoch": 7.12, "learning_rate": 2.283e-05, "loss": 0.0154, "step": 5442 }, { "epoch": 7.12, "learning_rate": 2.2825000000000003e-05, "loss": 0.0114, "step": 5443 }, { "epoch": 7.13, "learning_rate": 2.282e-05, "loss": 0.0121, "step": 5444 }, { "epoch": 7.13, "learning_rate": 2.2815e-05, "loss": 0.0125, "step": 5445 }, { "epoch": 7.13, "learning_rate": 2.281e-05, "loss": 0.0139, "step": 5446 }, { "epoch": 7.13, "learning_rate": 2.2805000000000003e-05, "loss": 0.0129, "step": 5447 }, { "epoch": 7.13, "learning_rate": 2.2800000000000002e-05, "loss": 0.0135, "step": 5448 }, { "epoch": 7.13, "learning_rate": 2.2795e-05, "loss": 0.0119, "step": 5449 }, { "epoch": 7.13, "learning_rate": 2.279e-05, "loss": 0.0116, "step": 5450 }, { "epoch": 7.13, "learning_rate": 2.2785e-05, "loss": 0.0138, "step": 5451 }, { "epoch": 7.14, "learning_rate": 2.2780000000000002e-05, "loss": 0.0118, "step": 5452 }, { "epoch": 7.14, "learning_rate": 2.2775e-05, "loss": 0.0134, "step": 5453 }, { "epoch": 7.14, "learning_rate": 2.2770000000000004e-05, "loss": 0.0148, "step": 5454 }, { "epoch": 7.14, "learning_rate": 2.2765e-05, "loss": 0.0094, "step": 5455 }, { "epoch": 7.14, "learning_rate": 2.2760000000000002e-05, "loss": 0.0118, "step": 5456 }, { "epoch": 7.14, "learning_rate": 2.2755e-05, "loss": 0.0175, "step": 5457 }, { "epoch": 7.14, "learning_rate": 2.275e-05, "loss": 0.01, "step": 5458 }, { "epoch": 7.15, "learning_rate": 2.2745000000000003e-05, "loss": 0.0116, "step": 5459 }, { "epoch": 7.15, "learning_rate": 2.274e-05, "loss": 0.0144, "step": 5460 }, { "epoch": 7.15, "learning_rate": 2.2735000000000002e-05, "loss": 0.0126, "step": 5461 }, { "epoch": 7.15, "learning_rate": 2.273e-05, "loss": 0.0118, "step": 5462 }, { "epoch": 7.15, "learning_rate": 2.2725000000000003e-05, "loss": 0.0165, "step": 5463 }, { "epoch": 7.15, "learning_rate": 2.2720000000000003e-05, "loss": 0.0127, "step": 5464 }, { "epoch": 7.15, "learning_rate": 2.2715e-05, "loss": 0.0088, "step": 5465 }, { "epoch": 7.15, "learning_rate": 2.271e-05, "loss": 0.0134, "step": 5466 }, { "epoch": 7.16, "learning_rate": 2.2705e-05, "loss": 0.0126, "step": 5467 }, { "epoch": 7.16, "learning_rate": 2.2700000000000003e-05, "loss": 0.0106, "step": 5468 }, { "epoch": 7.16, "learning_rate": 2.2695000000000002e-05, "loss": 0.0101, "step": 5469 }, { "epoch": 7.16, "learning_rate": 2.269e-05, "loss": 0.0113, "step": 5470 }, { "epoch": 7.16, "learning_rate": 2.2685e-05, "loss": 0.0114, "step": 5471 }, { "epoch": 7.16, "learning_rate": 2.268e-05, "loss": 0.0118, "step": 5472 }, { "epoch": 7.16, "learning_rate": 2.2675000000000002e-05, "loss": 0.0093, "step": 5473 }, { "epoch": 7.16, "learning_rate": 2.267e-05, "loss": 0.0197, "step": 5474 }, { "epoch": 7.17, "learning_rate": 2.2665e-05, "loss": 0.0111, "step": 5475 }, { "epoch": 7.17, "learning_rate": 2.266e-05, "loss": 0.0116, "step": 5476 }, { "epoch": 7.17, "learning_rate": 2.2655000000000002e-05, "loss": 0.012, "step": 5477 }, { "epoch": 7.17, "learning_rate": 2.265e-05, "loss": 0.0111, "step": 5478 }, { "epoch": 7.17, "learning_rate": 2.2645e-05, "loss": 0.0119, "step": 5479 }, { "epoch": 7.17, "learning_rate": 2.264e-05, "loss": 0.0085, "step": 5480 }, { "epoch": 7.17, "learning_rate": 2.2635e-05, "loss": 0.0146, "step": 5481 }, { "epoch": 7.18, "learning_rate": 2.2630000000000002e-05, "loss": 0.0103, "step": 5482 }, { "epoch": 7.18, "learning_rate": 2.2625e-05, "loss": 0.013, "step": 5483 }, { "epoch": 7.18, "learning_rate": 2.2620000000000004e-05, "loss": 0.0129, "step": 5484 }, { "epoch": 7.18, "learning_rate": 2.2615e-05, "loss": 0.0137, "step": 5485 }, { "epoch": 7.18, "learning_rate": 2.2610000000000002e-05, "loss": 0.0098, "step": 5486 }, { "epoch": 7.18, "learning_rate": 2.2605e-05, "loss": 0.0131, "step": 5487 }, { "epoch": 7.18, "learning_rate": 2.26e-05, "loss": 0.0117, "step": 5488 }, { "epoch": 7.18, "learning_rate": 2.2595000000000003e-05, "loss": 0.0131, "step": 5489 }, { "epoch": 7.19, "learning_rate": 2.259e-05, "loss": 0.0127, "step": 5490 }, { "epoch": 7.19, "learning_rate": 2.2585e-05, "loss": 0.0087, "step": 5491 }, { "epoch": 7.19, "learning_rate": 2.258e-05, "loss": 0.0093, "step": 5492 }, { "epoch": 7.19, "learning_rate": 2.2575000000000003e-05, "loss": 0.0111, "step": 5493 }, { "epoch": 7.19, "learning_rate": 2.2570000000000002e-05, "loss": 0.0127, "step": 5494 }, { "epoch": 7.19, "learning_rate": 2.2565e-05, "loss": 0.0135, "step": 5495 }, { "epoch": 7.19, "learning_rate": 2.256e-05, "loss": 0.0111, "step": 5496 }, { "epoch": 7.2, "learning_rate": 2.2555e-05, "loss": 0.0144, "step": 5497 }, { "epoch": 7.2, "learning_rate": 2.2550000000000003e-05, "loss": 0.0131, "step": 5498 }, { "epoch": 7.2, "learning_rate": 2.2545000000000002e-05, "loss": 0.014, "step": 5499 }, { "epoch": 7.2, "learning_rate": 2.254e-05, "loss": 0.0131, "step": 5500 }, { "epoch": 7.2, "learning_rate": 2.2535e-05, "loss": 0.0144, "step": 5501 }, { "epoch": 7.2, "learning_rate": 2.253e-05, "loss": 0.0128, "step": 5502 }, { "epoch": 7.2, "learning_rate": 2.2525000000000002e-05, "loss": 0.0113, "step": 5503 }, { "epoch": 7.2, "learning_rate": 2.252e-05, "loss": 0.0099, "step": 5504 }, { "epoch": 7.21, "learning_rate": 2.2515e-05, "loss": 0.0122, "step": 5505 }, { "epoch": 7.21, "learning_rate": 2.251e-05, "loss": 0.0126, "step": 5506 }, { "epoch": 7.21, "learning_rate": 2.2505000000000002e-05, "loss": 0.0133, "step": 5507 }, { "epoch": 7.21, "learning_rate": 2.25e-05, "loss": 0.0121, "step": 5508 }, { "epoch": 7.21, "learning_rate": 2.2495e-05, "loss": 0.0106, "step": 5509 }, { "epoch": 7.21, "learning_rate": 2.249e-05, "loss": 0.0178, "step": 5510 }, { "epoch": 7.21, "learning_rate": 2.2485e-05, "loss": 0.0131, "step": 5511 }, { "epoch": 7.21, "learning_rate": 2.248e-05, "loss": 0.0144, "step": 5512 }, { "epoch": 7.22, "learning_rate": 2.2475e-05, "loss": 0.0137, "step": 5513 }, { "epoch": 7.22, "learning_rate": 2.2470000000000003e-05, "loss": 0.0102, "step": 5514 }, { "epoch": 7.22, "learning_rate": 2.2465e-05, "loss": 0.0131, "step": 5515 }, { "epoch": 7.22, "learning_rate": 2.2460000000000002e-05, "loss": 0.0134, "step": 5516 }, { "epoch": 7.22, "learning_rate": 2.2455e-05, "loss": 0.016, "step": 5517 }, { "epoch": 7.22, "learning_rate": 2.245e-05, "loss": 0.012, "step": 5518 }, { "epoch": 7.22, "learning_rate": 2.2445000000000003e-05, "loss": 0.0143, "step": 5519 }, { "epoch": 7.23, "learning_rate": 2.244e-05, "loss": 0.0144, "step": 5520 }, { "epoch": 7.23, "learning_rate": 2.2435e-05, "loss": 0.0149, "step": 5521 }, { "epoch": 7.23, "learning_rate": 2.243e-05, "loss": 0.0122, "step": 5522 }, { "epoch": 7.23, "learning_rate": 2.2425000000000003e-05, "loss": 0.0122, "step": 5523 }, { "epoch": 7.23, "learning_rate": 2.2420000000000002e-05, "loss": 0.0125, "step": 5524 }, { "epoch": 7.23, "learning_rate": 2.2415e-05, "loss": 0.0118, "step": 5525 }, { "epoch": 7.23, "learning_rate": 2.241e-05, "loss": 0.0108, "step": 5526 }, { "epoch": 7.23, "learning_rate": 2.2405e-05, "loss": 0.0107, "step": 5527 }, { "epoch": 7.24, "learning_rate": 2.2400000000000002e-05, "loss": 0.0165, "step": 5528 }, { "epoch": 7.24, "learning_rate": 2.2395e-05, "loss": 0.0133, "step": 5529 }, { "epoch": 7.24, "learning_rate": 2.239e-05, "loss": 0.0112, "step": 5530 }, { "epoch": 7.24, "learning_rate": 2.2385e-05, "loss": 0.014, "step": 5531 }, { "epoch": 7.24, "learning_rate": 2.2380000000000003e-05, "loss": 0.0126, "step": 5532 }, { "epoch": 7.24, "learning_rate": 2.2375000000000002e-05, "loss": 0.0103, "step": 5533 }, { "epoch": 7.24, "learning_rate": 2.237e-05, "loss": 0.0138, "step": 5534 }, { "epoch": 7.24, "learning_rate": 2.2365e-05, "loss": 0.009, "step": 5535 }, { "epoch": 7.25, "learning_rate": 2.236e-05, "loss": 0.0098, "step": 5536 }, { "epoch": 7.25, "learning_rate": 2.2355000000000002e-05, "loss": 0.0132, "step": 5537 }, { "epoch": 7.25, "learning_rate": 2.235e-05, "loss": 0.0105, "step": 5538 }, { "epoch": 7.25, "learning_rate": 2.2345e-05, "loss": 0.0124, "step": 5539 }, { "epoch": 7.25, "learning_rate": 2.234e-05, "loss": 0.0108, "step": 5540 }, { "epoch": 7.25, "learning_rate": 2.2335e-05, "loss": 0.013, "step": 5541 }, { "epoch": 7.25, "learning_rate": 2.233e-05, "loss": 0.0136, "step": 5542 }, { "epoch": 7.26, "learning_rate": 2.2325e-05, "loss": 0.016, "step": 5543 }, { "epoch": 7.26, "learning_rate": 2.2320000000000003e-05, "loss": 0.0102, "step": 5544 }, { "epoch": 7.26, "learning_rate": 2.2315e-05, "loss": 0.0243, "step": 5545 }, { "epoch": 7.26, "learning_rate": 2.231e-05, "loss": 0.0127, "step": 5546 }, { "epoch": 7.26, "learning_rate": 2.2305e-05, "loss": 0.0174, "step": 5547 }, { "epoch": 7.26, "learning_rate": 2.23e-05, "loss": 0.014, "step": 5548 }, { "epoch": 7.26, "learning_rate": 2.2295000000000003e-05, "loss": 0.0124, "step": 5549 }, { "epoch": 7.26, "learning_rate": 2.229e-05, "loss": 0.0133, "step": 5550 }, { "epoch": 7.27, "learning_rate": 2.2285e-05, "loss": 0.0131, "step": 5551 }, { "epoch": 7.27, "learning_rate": 2.228e-05, "loss": 0.0133, "step": 5552 }, { "epoch": 7.27, "learning_rate": 2.2275000000000003e-05, "loss": 0.0182, "step": 5553 }, { "epoch": 7.27, "learning_rate": 2.2270000000000002e-05, "loss": 0.013, "step": 5554 }, { "epoch": 7.27, "learning_rate": 2.2265e-05, "loss": 0.0129, "step": 5555 }, { "epoch": 7.27, "learning_rate": 2.226e-05, "loss": 0.0113, "step": 5556 }, { "epoch": 7.27, "learning_rate": 2.2255e-05, "loss": 0.0133, "step": 5557 }, { "epoch": 7.27, "learning_rate": 2.2250000000000002e-05, "loss": 0.0134, "step": 5558 }, { "epoch": 7.28, "learning_rate": 2.2245e-05, "loss": 0.0111, "step": 5559 }, { "epoch": 7.28, "learning_rate": 2.224e-05, "loss": 0.0128, "step": 5560 }, { "epoch": 7.28, "learning_rate": 2.2235e-05, "loss": 0.012, "step": 5561 }, { "epoch": 7.28, "learning_rate": 2.2230000000000002e-05, "loss": 0.0145, "step": 5562 }, { "epoch": 7.28, "learning_rate": 2.2225e-05, "loss": 0.0119, "step": 5563 }, { "epoch": 7.28, "learning_rate": 2.222e-05, "loss": 0.0156, "step": 5564 }, { "epoch": 7.28, "learning_rate": 2.2215e-05, "loss": 0.0119, "step": 5565 }, { "epoch": 7.29, "learning_rate": 2.221e-05, "loss": 0.0134, "step": 5566 }, { "epoch": 7.29, "learning_rate": 2.2205000000000002e-05, "loss": 0.014, "step": 5567 }, { "epoch": 7.29, "learning_rate": 2.22e-05, "loss": 0.011, "step": 5568 }, { "epoch": 7.29, "learning_rate": 2.2195000000000003e-05, "loss": 0.0137, "step": 5569 }, { "epoch": 7.29, "learning_rate": 2.219e-05, "loss": 0.0105, "step": 5570 }, { "epoch": 7.29, "learning_rate": 2.2185000000000002e-05, "loss": 0.0118, "step": 5571 }, { "epoch": 7.29, "learning_rate": 2.218e-05, "loss": 0.011, "step": 5572 }, { "epoch": 7.29, "learning_rate": 2.2175e-05, "loss": 0.0108, "step": 5573 }, { "epoch": 7.3, "learning_rate": 2.2170000000000003e-05, "loss": 0.0106, "step": 5574 }, { "epoch": 7.3, "learning_rate": 2.2165000000000002e-05, "loss": 0.0157, "step": 5575 }, { "epoch": 7.3, "learning_rate": 2.216e-05, "loss": 0.0092, "step": 5576 }, { "epoch": 7.3, "learning_rate": 2.2155e-05, "loss": 0.0141, "step": 5577 }, { "epoch": 7.3, "learning_rate": 2.215e-05, "loss": 0.0135, "step": 5578 }, { "epoch": 7.3, "learning_rate": 2.2145000000000002e-05, "loss": 0.0117, "step": 5579 }, { "epoch": 7.3, "learning_rate": 2.214e-05, "loss": 0.0118, "step": 5580 }, { "epoch": 7.3, "learning_rate": 2.2135e-05, "loss": 0.0103, "step": 5581 }, { "epoch": 7.31, "learning_rate": 2.213e-05, "loss": 0.0117, "step": 5582 }, { "epoch": 7.31, "learning_rate": 2.2125000000000002e-05, "loss": 0.011, "step": 5583 }, { "epoch": 7.31, "learning_rate": 2.212e-05, "loss": 0.0111, "step": 5584 }, { "epoch": 7.31, "learning_rate": 2.2115e-05, "loss": 0.0133, "step": 5585 }, { "epoch": 7.31, "learning_rate": 2.211e-05, "loss": 0.0118, "step": 5586 }, { "epoch": 7.31, "learning_rate": 2.2105e-05, "loss": 0.0104, "step": 5587 }, { "epoch": 7.31, "learning_rate": 2.2100000000000002e-05, "loss": 0.0105, "step": 5588 }, { "epoch": 7.32, "learning_rate": 2.2095e-05, "loss": 0.0089, "step": 5589 }, { "epoch": 7.32, "learning_rate": 2.2090000000000004e-05, "loss": 0.0114, "step": 5590 }, { "epoch": 7.32, "learning_rate": 2.2085e-05, "loss": 0.013, "step": 5591 }, { "epoch": 7.32, "learning_rate": 2.2080000000000002e-05, "loss": 0.0132, "step": 5592 }, { "epoch": 7.32, "learning_rate": 2.2075e-05, "loss": 0.0144, "step": 5593 }, { "epoch": 7.32, "learning_rate": 2.207e-05, "loss": 0.0119, "step": 5594 }, { "epoch": 7.32, "learning_rate": 2.2065000000000003e-05, "loss": 0.0134, "step": 5595 }, { "epoch": 7.32, "learning_rate": 2.206e-05, "loss": 0.0134, "step": 5596 }, { "epoch": 7.33, "learning_rate": 2.2055e-05, "loss": 0.0091, "step": 5597 }, { "epoch": 7.33, "learning_rate": 2.205e-05, "loss": 0.0144, "step": 5598 }, { "epoch": 7.33, "learning_rate": 2.2045000000000003e-05, "loss": 0.0132, "step": 5599 }, { "epoch": 7.33, "learning_rate": 2.2040000000000002e-05, "loss": 0.0144, "step": 5600 }, { "epoch": 7.33, "learning_rate": 2.2035e-05, "loss": 0.0104, "step": 5601 }, { "epoch": 7.33, "learning_rate": 2.203e-05, "loss": 0.0127, "step": 5602 }, { "epoch": 7.33, "learning_rate": 2.2025e-05, "loss": 0.0093, "step": 5603 }, { "epoch": 7.34, "learning_rate": 2.2020000000000003e-05, "loss": 0.0136, "step": 5604 }, { "epoch": 7.34, "learning_rate": 2.2015000000000002e-05, "loss": 0.0131, "step": 5605 }, { "epoch": 7.34, "learning_rate": 2.201e-05, "loss": 0.0122, "step": 5606 }, { "epoch": 7.34, "learning_rate": 2.2005e-05, "loss": 0.0149, "step": 5607 }, { "epoch": 7.34, "learning_rate": 2.2000000000000003e-05, "loss": 0.0126, "step": 5608 }, { "epoch": 7.34, "learning_rate": 2.1995000000000002e-05, "loss": 0.0127, "step": 5609 }, { "epoch": 7.34, "learning_rate": 2.199e-05, "loss": 0.017, "step": 5610 }, { "epoch": 7.34, "learning_rate": 2.1985e-05, "loss": 0.0141, "step": 5611 }, { "epoch": 7.35, "learning_rate": 2.198e-05, "loss": 0.0104, "step": 5612 }, { "epoch": 7.35, "learning_rate": 2.1975000000000002e-05, "loss": 0.014, "step": 5613 }, { "epoch": 7.35, "learning_rate": 2.197e-05, "loss": 0.0133, "step": 5614 }, { "epoch": 7.35, "learning_rate": 2.1965e-05, "loss": 0.012, "step": 5615 }, { "epoch": 7.35, "learning_rate": 2.196e-05, "loss": 0.0102, "step": 5616 }, { "epoch": 7.35, "learning_rate": 2.1955e-05, "loss": 0.0108, "step": 5617 }, { "epoch": 7.35, "learning_rate": 2.195e-05, "loss": 0.0118, "step": 5618 }, { "epoch": 7.35, "learning_rate": 2.1945e-05, "loss": 0.0127, "step": 5619 }, { "epoch": 7.36, "learning_rate": 2.1940000000000003e-05, "loss": 0.0183, "step": 5620 }, { "epoch": 7.36, "learning_rate": 2.1935e-05, "loss": 0.0137, "step": 5621 }, { "epoch": 7.36, "learning_rate": 2.1930000000000002e-05, "loss": 0.0149, "step": 5622 }, { "epoch": 7.36, "learning_rate": 2.1925e-05, "loss": 0.0102, "step": 5623 }, { "epoch": 7.36, "learning_rate": 2.192e-05, "loss": 0.0125, "step": 5624 }, { "epoch": 7.36, "learning_rate": 2.1915000000000003e-05, "loss": 0.0125, "step": 5625 }, { "epoch": 7.36, "learning_rate": 2.191e-05, "loss": 0.0112, "step": 5626 }, { "epoch": 7.37, "learning_rate": 2.1905e-05, "loss": 0.0105, "step": 5627 }, { "epoch": 7.37, "learning_rate": 2.19e-05, "loss": 0.0131, "step": 5628 }, { "epoch": 7.37, "learning_rate": 2.1895000000000003e-05, "loss": 0.0127, "step": 5629 }, { "epoch": 7.37, "learning_rate": 2.1890000000000002e-05, "loss": 0.0138, "step": 5630 }, { "epoch": 7.37, "learning_rate": 2.1885e-05, "loss": 0.0139, "step": 5631 }, { "epoch": 7.37, "learning_rate": 2.188e-05, "loss": 0.0209, "step": 5632 }, { "epoch": 7.37, "learning_rate": 2.1875e-05, "loss": 0.0104, "step": 5633 }, { "epoch": 7.37, "learning_rate": 2.1870000000000002e-05, "loss": 0.0108, "step": 5634 }, { "epoch": 7.38, "learning_rate": 2.1865e-05, "loss": 0.0152, "step": 5635 }, { "epoch": 7.38, "learning_rate": 2.186e-05, "loss": 0.0121, "step": 5636 }, { "epoch": 7.38, "learning_rate": 2.1855e-05, "loss": 0.0093, "step": 5637 }, { "epoch": 7.38, "learning_rate": 2.1850000000000003e-05, "loss": 0.0119, "step": 5638 }, { "epoch": 7.38, "learning_rate": 2.1845000000000002e-05, "loss": 0.0141, "step": 5639 }, { "epoch": 7.38, "learning_rate": 2.184e-05, "loss": 0.0087, "step": 5640 }, { "epoch": 7.38, "learning_rate": 2.1835e-05, "loss": 0.0119, "step": 5641 }, { "epoch": 7.38, "learning_rate": 2.183e-05, "loss": 0.0121, "step": 5642 }, { "epoch": 7.39, "learning_rate": 2.1825000000000002e-05, "loss": 0.0115, "step": 5643 }, { "epoch": 7.39, "learning_rate": 2.182e-05, "loss": 0.0117, "step": 5644 }, { "epoch": 7.39, "learning_rate": 2.1815000000000004e-05, "loss": 0.0107, "step": 5645 }, { "epoch": 7.39, "learning_rate": 2.181e-05, "loss": 0.0111, "step": 5646 }, { "epoch": 7.39, "learning_rate": 2.1805e-05, "loss": 0.0149, "step": 5647 }, { "epoch": 7.39, "learning_rate": 2.18e-05, "loss": 0.0113, "step": 5648 }, { "epoch": 7.39, "learning_rate": 2.1795e-05, "loss": 0.0917, "step": 5649 }, { "epoch": 7.4, "learning_rate": 2.1790000000000003e-05, "loss": 0.0144, "step": 5650 }, { "epoch": 7.4, "learning_rate": 2.1785e-05, "loss": 0.0136, "step": 5651 }, { "epoch": 7.4, "learning_rate": 2.178e-05, "loss": 0.0102, "step": 5652 }, { "epoch": 7.4, "learning_rate": 2.1775e-05, "loss": 0.0113, "step": 5653 }, { "epoch": 7.4, "learning_rate": 2.177e-05, "loss": 0.012, "step": 5654 }, { "epoch": 7.4, "learning_rate": 2.1765000000000003e-05, "loss": 0.0123, "step": 5655 }, { "epoch": 7.4, "learning_rate": 2.176e-05, "loss": 0.0141, "step": 5656 }, { "epoch": 7.4, "learning_rate": 2.1755e-05, "loss": 0.0138, "step": 5657 }, { "epoch": 7.41, "learning_rate": 2.175e-05, "loss": 0.0103, "step": 5658 }, { "epoch": 7.41, "learning_rate": 2.1745000000000003e-05, "loss": 0.0134, "step": 5659 }, { "epoch": 7.41, "learning_rate": 2.1740000000000002e-05, "loss": 0.0133, "step": 5660 }, { "epoch": 7.41, "learning_rate": 2.1735e-05, "loss": 0.0115, "step": 5661 }, { "epoch": 7.41, "learning_rate": 2.173e-05, "loss": 0.0134, "step": 5662 }, { "epoch": 7.41, "learning_rate": 2.1725e-05, "loss": 0.0126, "step": 5663 }, { "epoch": 7.41, "learning_rate": 2.1720000000000002e-05, "loss": 0.0153, "step": 5664 }, { "epoch": 7.41, "learning_rate": 2.1715e-05, "loss": 0.0098, "step": 5665 }, { "epoch": 7.42, "learning_rate": 2.171e-05, "loss": 0.0105, "step": 5666 }, { "epoch": 7.42, "learning_rate": 2.1705e-05, "loss": 0.0123, "step": 5667 }, { "epoch": 7.42, "learning_rate": 2.1700000000000002e-05, "loss": 0.0133, "step": 5668 }, { "epoch": 7.42, "learning_rate": 2.1695e-05, "loss": 0.0175, "step": 5669 }, { "epoch": 7.42, "learning_rate": 2.169e-05, "loss": 0.0123, "step": 5670 }, { "epoch": 7.42, "learning_rate": 2.1685e-05, "loss": 0.0103, "step": 5671 }, { "epoch": 7.42, "learning_rate": 2.168e-05, "loss": 0.013, "step": 5672 }, { "epoch": 7.43, "learning_rate": 2.1675e-05, "loss": 0.0119, "step": 5673 }, { "epoch": 7.43, "learning_rate": 2.167e-05, "loss": 0.0122, "step": 5674 }, { "epoch": 7.43, "learning_rate": 2.1665000000000003e-05, "loss": 0.0127, "step": 5675 }, { "epoch": 7.43, "learning_rate": 2.166e-05, "loss": 0.0131, "step": 5676 }, { "epoch": 7.43, "learning_rate": 2.166e-05, "loss": 0.0486, "step": 5677 }, { "epoch": 7.43, "learning_rate": 2.1655000000000002e-05, "loss": 0.0156, "step": 5678 }, { "epoch": 7.43, "learning_rate": 2.165e-05, "loss": 0.011, "step": 5679 }, { "epoch": 7.43, "learning_rate": 2.1645e-05, "loss": 0.0158, "step": 5680 }, { "epoch": 7.44, "learning_rate": 2.1640000000000003e-05, "loss": 0.0138, "step": 5681 }, { "epoch": 7.44, "learning_rate": 2.1635e-05, "loss": 0.0114, "step": 5682 }, { "epoch": 7.44, "learning_rate": 2.163e-05, "loss": 0.0117, "step": 5683 }, { "epoch": 7.44, "learning_rate": 2.1625e-05, "loss": 0.0131, "step": 5684 }, { "epoch": 7.44, "learning_rate": 2.162e-05, "loss": 0.0136, "step": 5685 }, { "epoch": 7.44, "learning_rate": 2.1615000000000002e-05, "loss": 0.014, "step": 5686 }, { "epoch": 7.44, "learning_rate": 2.1609999999999998e-05, "loss": 0.0113, "step": 5687 }, { "epoch": 7.45, "learning_rate": 2.1605e-05, "loss": 0.013, "step": 5688 }, { "epoch": 7.45, "learning_rate": 2.16e-05, "loss": 0.0134, "step": 5689 }, { "epoch": 7.45, "learning_rate": 2.1595000000000002e-05, "loss": 0.0143, "step": 5690 }, { "epoch": 7.45, "learning_rate": 2.159e-05, "loss": 0.0125, "step": 5691 }, { "epoch": 7.45, "learning_rate": 2.1585e-05, "loss": 0.0154, "step": 5692 }, { "epoch": 7.45, "learning_rate": 2.158e-05, "loss": 0.0127, "step": 5693 }, { "epoch": 7.45, "learning_rate": 2.1575e-05, "loss": 0.0143, "step": 5694 }, { "epoch": 7.45, "learning_rate": 2.1570000000000002e-05, "loss": 0.0136, "step": 5695 }, { "epoch": 7.46, "learning_rate": 2.1565e-05, "loss": 0.0122, "step": 5696 }, { "epoch": 7.46, "learning_rate": 2.1560000000000004e-05, "loss": 0.0131, "step": 5697 }, { "epoch": 7.46, "learning_rate": 2.1555e-05, "loss": 0.0115, "step": 5698 }, { "epoch": 7.46, "learning_rate": 2.1550000000000002e-05, "loss": 0.0136, "step": 5699 }, { "epoch": 7.46, "learning_rate": 2.1545e-05, "loss": 0.0101, "step": 5700 }, { "epoch": 7.46, "learning_rate": 2.154e-05, "loss": 0.0141, "step": 5701 }, { "epoch": 7.46, "learning_rate": 2.1535000000000003e-05, "loss": 0.0129, "step": 5702 }, { "epoch": 7.46, "learning_rate": 2.153e-05, "loss": 0.0125, "step": 5703 }, { "epoch": 7.47, "learning_rate": 2.1525e-05, "loss": 0.0128, "step": 5704 }, { "epoch": 7.47, "learning_rate": 2.152e-05, "loss": 0.0139, "step": 5705 }, { "epoch": 7.47, "learning_rate": 2.1515000000000003e-05, "loss": 0.0166, "step": 5706 }, { "epoch": 7.47, "learning_rate": 2.1510000000000002e-05, "loss": 0.0139, "step": 5707 }, { "epoch": 7.47, "learning_rate": 2.1505e-05, "loss": 0.0126, "step": 5708 }, { "epoch": 7.47, "learning_rate": 2.15e-05, "loss": 0.0123, "step": 5709 }, { "epoch": 7.47, "learning_rate": 2.1495e-05, "loss": 0.0136, "step": 5710 }, { "epoch": 7.48, "learning_rate": 2.1490000000000003e-05, "loss": 0.0137, "step": 5711 }, { "epoch": 7.48, "learning_rate": 2.1485000000000002e-05, "loss": 0.0102, "step": 5712 }, { "epoch": 7.48, "learning_rate": 2.148e-05, "loss": 0.014, "step": 5713 }, { "epoch": 7.48, "learning_rate": 2.1475e-05, "loss": 0.015, "step": 5714 }, { "epoch": 7.48, "learning_rate": 2.1470000000000003e-05, "loss": 0.0123, "step": 5715 }, { "epoch": 7.48, "learning_rate": 2.1465000000000002e-05, "loss": 0.0114, "step": 5716 }, { "epoch": 7.48, "learning_rate": 2.146e-05, "loss": 0.013, "step": 5717 }, { "epoch": 7.48, "learning_rate": 2.1455e-05, "loss": 0.0129, "step": 5718 }, { "epoch": 7.49, "learning_rate": 2.145e-05, "loss": 0.0113, "step": 5719 }, { "epoch": 7.49, "learning_rate": 2.1445000000000002e-05, "loss": 0.0162, "step": 5720 }, { "epoch": 7.49, "learning_rate": 2.144e-05, "loss": 0.0154, "step": 5721 }, { "epoch": 7.49, "learning_rate": 2.1435000000000004e-05, "loss": 0.0112, "step": 5722 }, { "epoch": 7.49, "learning_rate": 2.143e-05, "loss": 0.0122, "step": 5723 }, { "epoch": 7.49, "learning_rate": 2.1425e-05, "loss": 0.0139, "step": 5724 }, { "epoch": 7.49, "learning_rate": 2.142e-05, "loss": 0.0134, "step": 5725 }, { "epoch": 7.49, "learning_rate": 2.1415e-05, "loss": 0.0134, "step": 5726 }, { "epoch": 7.5, "learning_rate": 2.1410000000000003e-05, "loss": 0.0119, "step": 5727 }, { "epoch": 7.5, "learning_rate": 2.1405e-05, "loss": 0.0103, "step": 5728 }, { "epoch": 7.5, "learning_rate": 2.1400000000000002e-05, "loss": 0.0111, "step": 5729 }, { "epoch": 7.5, "learning_rate": 2.1395e-05, "loss": 0.0116, "step": 5730 }, { "epoch": 7.5, "learning_rate": 2.139e-05, "loss": 0.0135, "step": 5731 }, { "epoch": 7.5, "learning_rate": 2.1385000000000003e-05, "loss": 0.0108, "step": 5732 }, { "epoch": 7.5, "learning_rate": 2.138e-05, "loss": 0.0147, "step": 5733 }, { "epoch": 7.51, "learning_rate": 2.1375e-05, "loss": 0.0142, "step": 5734 }, { "epoch": 7.51, "learning_rate": 2.137e-05, "loss": 0.0108, "step": 5735 }, { "epoch": 7.51, "learning_rate": 2.1365000000000003e-05, "loss": 0.0124, "step": 5736 }, { "epoch": 7.51, "learning_rate": 2.1360000000000002e-05, "loss": 0.0109, "step": 5737 }, { "epoch": 7.51, "learning_rate": 2.1355e-05, "loss": 0.0129, "step": 5738 }, { "epoch": 7.51, "learning_rate": 2.135e-05, "loss": 0.0132, "step": 5739 }, { "epoch": 7.51, "learning_rate": 2.1345e-05, "loss": 0.0108, "step": 5740 }, { "epoch": 7.51, "learning_rate": 2.1340000000000002e-05, "loss": 0.0115, "step": 5741 }, { "epoch": 7.52, "learning_rate": 2.1335e-05, "loss": 0.0117, "step": 5742 }, { "epoch": 7.52, "learning_rate": 2.133e-05, "loss": 0.0127, "step": 5743 }, { "epoch": 7.52, "learning_rate": 2.1325e-05, "loss": 0.0133, "step": 5744 }, { "epoch": 7.52, "learning_rate": 2.1320000000000003e-05, "loss": 0.0167, "step": 5745 }, { "epoch": 7.52, "learning_rate": 2.1315000000000002e-05, "loss": 0.0126, "step": 5746 }, { "epoch": 7.52, "learning_rate": 2.131e-05, "loss": 0.0106, "step": 5747 }, { "epoch": 7.52, "learning_rate": 2.1305e-05, "loss": 0.0138, "step": 5748 }, { "epoch": 7.52, "learning_rate": 2.13e-05, "loss": 0.0135, "step": 5749 }, { "epoch": 7.53, "learning_rate": 2.1295000000000002e-05, "loss": 0.0097, "step": 5750 }, { "epoch": 7.53, "learning_rate": 2.129e-05, "loss": 0.0149, "step": 5751 }, { "epoch": 7.53, "learning_rate": 2.1285000000000004e-05, "loss": 0.0117, "step": 5752 }, { "epoch": 7.53, "learning_rate": 2.128e-05, "loss": 0.0108, "step": 5753 }, { "epoch": 7.53, "learning_rate": 2.1275000000000002e-05, "loss": 0.0141, "step": 5754 }, { "epoch": 7.53, "learning_rate": 2.127e-05, "loss": 0.014, "step": 5755 }, { "epoch": 7.53, "learning_rate": 2.1265e-05, "loss": 0.0127, "step": 5756 }, { "epoch": 7.54, "learning_rate": 2.1260000000000003e-05, "loss": 0.0105, "step": 5757 }, { "epoch": 7.54, "learning_rate": 2.1255e-05, "loss": 0.0163, "step": 5758 }, { "epoch": 7.54, "learning_rate": 2.125e-05, "loss": 0.0128, "step": 5759 }, { "epoch": 7.54, "learning_rate": 2.1245e-05, "loss": 0.0122, "step": 5760 }, { "epoch": 7.54, "learning_rate": 2.124e-05, "loss": 0.0121, "step": 5761 }, { "epoch": 7.54, "learning_rate": 2.1235000000000003e-05, "loss": 0.0134, "step": 5762 }, { "epoch": 7.54, "learning_rate": 2.123e-05, "loss": 0.0125, "step": 5763 }, { "epoch": 7.54, "learning_rate": 2.1225e-05, "loss": 0.0131, "step": 5764 }, { "epoch": 7.55, "learning_rate": 2.122e-05, "loss": 0.014, "step": 5765 }, { "epoch": 7.55, "learning_rate": 2.1215000000000003e-05, "loss": 0.0149, "step": 5766 }, { "epoch": 7.55, "learning_rate": 2.1210000000000002e-05, "loss": 0.0161, "step": 5767 }, { "epoch": 7.55, "learning_rate": 2.1205e-05, "loss": 0.0147, "step": 5768 }, { "epoch": 7.55, "learning_rate": 2.12e-05, "loss": 0.0153, "step": 5769 }, { "epoch": 7.55, "learning_rate": 2.1195e-05, "loss": 0.0144, "step": 5770 }, { "epoch": 7.55, "learning_rate": 2.1190000000000002e-05, "loss": 0.0138, "step": 5771 }, { "epoch": 7.55, "learning_rate": 2.1185e-05, "loss": 0.0167, "step": 5772 }, { "epoch": 7.56, "learning_rate": 2.118e-05, "loss": 0.0135, "step": 5773 }, { "epoch": 7.56, "learning_rate": 2.1175e-05, "loss": 0.0102, "step": 5774 }, { "epoch": 7.56, "learning_rate": 2.1170000000000002e-05, "loss": 0.0143, "step": 5775 }, { "epoch": 7.56, "learning_rate": 2.1165e-05, "loss": 0.0155, "step": 5776 }, { "epoch": 7.56, "learning_rate": 2.116e-05, "loss": 0.0154, "step": 5777 }, { "epoch": 7.56, "learning_rate": 2.1155e-05, "loss": 0.0136, "step": 5778 }, { "epoch": 7.56, "learning_rate": 2.115e-05, "loss": 0.0148, "step": 5779 }, { "epoch": 7.57, "learning_rate": 2.1145e-05, "loss": 0.0112, "step": 5780 }, { "epoch": 7.57, "learning_rate": 2.114e-05, "loss": 0.01, "step": 5781 }, { "epoch": 7.57, "learning_rate": 2.1135000000000003e-05, "loss": 0.0115, "step": 5782 }, { "epoch": 7.57, "learning_rate": 2.113e-05, "loss": 0.0109, "step": 5783 }, { "epoch": 7.57, "learning_rate": 2.1125000000000002e-05, "loss": 0.0259, "step": 5784 }, { "epoch": 7.57, "learning_rate": 2.112e-05, "loss": 0.0126, "step": 5785 }, { "epoch": 7.57, "learning_rate": 2.1115e-05, "loss": 0.0126, "step": 5786 }, { "epoch": 7.57, "learning_rate": 2.1110000000000003e-05, "loss": 0.0124, "step": 5787 }, { "epoch": 7.58, "learning_rate": 2.1105e-05, "loss": 0.0153, "step": 5788 }, { "epoch": 7.58, "learning_rate": 2.11e-05, "loss": 0.0182, "step": 5789 }, { "epoch": 7.58, "learning_rate": 2.1095e-05, "loss": 0.0115, "step": 5790 }, { "epoch": 7.58, "learning_rate": 2.1090000000000003e-05, "loss": 0.0137, "step": 5791 }, { "epoch": 7.58, "learning_rate": 2.1085000000000002e-05, "loss": 0.0119, "step": 5792 }, { "epoch": 7.58, "learning_rate": 2.1079999999999998e-05, "loss": 0.0124, "step": 5793 }, { "epoch": 7.58, "learning_rate": 2.1075e-05, "loss": 0.0143, "step": 5794 }, { "epoch": 7.59, "learning_rate": 2.107e-05, "loss": 0.0172, "step": 5795 }, { "epoch": 7.59, "learning_rate": 2.1065000000000002e-05, "loss": 0.0123, "step": 5796 }, { "epoch": 7.59, "learning_rate": 2.106e-05, "loss": 0.0136, "step": 5797 }, { "epoch": 7.59, "learning_rate": 2.1055e-05, "loss": 0.0109, "step": 5798 }, { "epoch": 7.59, "learning_rate": 2.105e-05, "loss": 0.0137, "step": 5799 }, { "epoch": 7.59, "learning_rate": 2.1045e-05, "loss": 0.0184, "step": 5800 }, { "epoch": 7.59, "learning_rate": 2.1040000000000002e-05, "loss": 0.0117, "step": 5801 }, { "epoch": 7.59, "learning_rate": 2.1035e-05, "loss": 0.0151, "step": 5802 }, { "epoch": 7.6, "learning_rate": 2.103e-05, "loss": 0.0146, "step": 5803 }, { "epoch": 7.6, "learning_rate": 2.1025e-05, "loss": 0.0129, "step": 5804 }, { "epoch": 7.6, "learning_rate": 2.1020000000000002e-05, "loss": 0.0129, "step": 5805 }, { "epoch": 7.6, "learning_rate": 2.1015e-05, "loss": 0.0131, "step": 5806 }, { "epoch": 7.6, "learning_rate": 2.101e-05, "loss": 0.0144, "step": 5807 }, { "epoch": 7.6, "learning_rate": 2.1005e-05, "loss": 0.0151, "step": 5808 }, { "epoch": 7.6, "learning_rate": 2.1e-05, "loss": 0.0161, "step": 5809 }, { "epoch": 7.6, "learning_rate": 2.0995e-05, "loss": 0.0141, "step": 5810 }, { "epoch": 7.61, "learning_rate": 2.099e-05, "loss": 0.015, "step": 5811 }, { "epoch": 7.61, "learning_rate": 2.0985000000000003e-05, "loss": 0.0172, "step": 5812 }, { "epoch": 7.61, "learning_rate": 2.098e-05, "loss": 0.017, "step": 5813 }, { "epoch": 7.61, "learning_rate": 2.0975e-05, "loss": 0.0119, "step": 5814 }, { "epoch": 7.61, "learning_rate": 2.097e-05, "loss": 0.0136, "step": 5815 }, { "epoch": 7.61, "learning_rate": 2.0965e-05, "loss": 0.0139, "step": 5816 }, { "epoch": 7.61, "learning_rate": 2.0960000000000003e-05, "loss": 0.0158, "step": 5817 }, { "epoch": 7.62, "learning_rate": 2.0955e-05, "loss": 0.0155, "step": 5818 }, { "epoch": 7.62, "learning_rate": 2.095e-05, "loss": 0.0127, "step": 5819 }, { "epoch": 7.62, "learning_rate": 2.0945e-05, "loss": 0.0136, "step": 5820 }, { "epoch": 7.62, "learning_rate": 2.0940000000000003e-05, "loss": 0.014, "step": 5821 }, { "epoch": 7.62, "learning_rate": 2.0935000000000002e-05, "loss": 0.0169, "step": 5822 }, { "epoch": 7.62, "learning_rate": 2.093e-05, "loss": 0.014, "step": 5823 }, { "epoch": 7.62, "learning_rate": 2.0925e-05, "loss": 0.0133, "step": 5824 }, { "epoch": 7.62, "learning_rate": 2.092e-05, "loss": 0.0125, "step": 5825 }, { "epoch": 7.63, "learning_rate": 2.0915000000000002e-05, "loss": 0.0148, "step": 5826 }, { "epoch": 7.63, "learning_rate": 2.091e-05, "loss": 0.0164, "step": 5827 }, { "epoch": 7.63, "learning_rate": 2.0905000000000004e-05, "loss": 0.0103, "step": 5828 }, { "epoch": 7.63, "learning_rate": 2.09e-05, "loss": 0.0158, "step": 5829 }, { "epoch": 7.63, "learning_rate": 2.0895e-05, "loss": 0.0131, "step": 5830 }, { "epoch": 7.63, "learning_rate": 2.089e-05, "loss": 0.0135, "step": 5831 }, { "epoch": 7.63, "learning_rate": 2.0885e-05, "loss": 0.0183, "step": 5832 }, { "epoch": 7.63, "learning_rate": 2.0880000000000003e-05, "loss": 0.0139, "step": 5833 }, { "epoch": 7.64, "learning_rate": 2.0875e-05, "loss": 0.0136, "step": 5834 }, { "epoch": 7.64, "learning_rate": 2.0870000000000002e-05, "loss": 0.0129, "step": 5835 }, { "epoch": 7.64, "learning_rate": 2.0865e-05, "loss": 0.0128, "step": 5836 }, { "epoch": 7.64, "learning_rate": 2.086e-05, "loss": 0.0139, "step": 5837 }, { "epoch": 7.64, "learning_rate": 2.0855000000000003e-05, "loss": 0.0139, "step": 5838 }, { "epoch": 7.64, "learning_rate": 2.085e-05, "loss": 0.016, "step": 5839 }, { "epoch": 7.64, "learning_rate": 2.0845e-05, "loss": 0.0124, "step": 5840 }, { "epoch": 7.65, "learning_rate": 2.084e-05, "loss": 0.0182, "step": 5841 }, { "epoch": 7.65, "learning_rate": 2.0835000000000003e-05, "loss": 0.0135, "step": 5842 }, { "epoch": 7.65, "learning_rate": 2.0830000000000002e-05, "loss": 0.014, "step": 5843 }, { "epoch": 7.65, "learning_rate": 2.0825e-05, "loss": 0.0137, "step": 5844 }, { "epoch": 7.65, "learning_rate": 2.082e-05, "loss": 0.0139, "step": 5845 }, { "epoch": 7.65, "learning_rate": 2.0815e-05, "loss": 0.0126, "step": 5846 }, { "epoch": 7.65, "learning_rate": 2.0810000000000002e-05, "loss": 0.0118, "step": 5847 }, { "epoch": 7.65, "learning_rate": 2.0805e-05, "loss": 0.014, "step": 5848 }, { "epoch": 7.66, "learning_rate": 2.08e-05, "loss": 0.0146, "step": 5849 }, { "epoch": 7.66, "learning_rate": 2.0795e-05, "loss": 0.015, "step": 5850 }, { "epoch": 7.66, "learning_rate": 2.0790000000000003e-05, "loss": 0.016, "step": 5851 }, { "epoch": 7.66, "learning_rate": 2.0785000000000002e-05, "loss": 0.0193, "step": 5852 }, { "epoch": 7.66, "learning_rate": 2.078e-05, "loss": 0.0149, "step": 5853 }, { "epoch": 7.66, "learning_rate": 2.0775e-05, "loss": 0.0148, "step": 5854 }, { "epoch": 7.66, "learning_rate": 2.077e-05, "loss": 0.0127, "step": 5855 }, { "epoch": 7.66, "learning_rate": 2.0765000000000002e-05, "loss": 0.012, "step": 5856 }, { "epoch": 7.67, "learning_rate": 2.076e-05, "loss": 0.0154, "step": 5857 }, { "epoch": 7.67, "learning_rate": 2.0755000000000004e-05, "loss": 0.0129, "step": 5858 }, { "epoch": 7.67, "learning_rate": 2.075e-05, "loss": 0.0123, "step": 5859 }, { "epoch": 7.67, "learning_rate": 2.0745000000000002e-05, "loss": 0.0149, "step": 5860 }, { "epoch": 7.67, "learning_rate": 2.074e-05, "loss": 0.0121, "step": 5861 }, { "epoch": 7.67, "learning_rate": 2.0735e-05, "loss": 0.0141, "step": 5862 }, { "epoch": 7.67, "learning_rate": 2.0730000000000003e-05, "loss": 0.0165, "step": 5863 }, { "epoch": 7.68, "learning_rate": 2.0725e-05, "loss": 0.0141, "step": 5864 }, { "epoch": 7.68, "learning_rate": 2.072e-05, "loss": 0.0136, "step": 5865 }, { "epoch": 7.68, "learning_rate": 2.0715e-05, "loss": 0.0112, "step": 5866 }, { "epoch": 7.68, "learning_rate": 2.0710000000000003e-05, "loss": 0.0143, "step": 5867 }, { "epoch": 7.68, "learning_rate": 2.0705000000000003e-05, "loss": 0.0122, "step": 5868 }, { "epoch": 7.68, "learning_rate": 2.07e-05, "loss": 0.0147, "step": 5869 }, { "epoch": 7.68, "learning_rate": 2.0695e-05, "loss": 0.0138, "step": 5870 }, { "epoch": 7.68, "learning_rate": 2.069e-05, "loss": 0.0163, "step": 5871 }, { "epoch": 7.69, "learning_rate": 2.0685000000000003e-05, "loss": 0.0139, "step": 5872 }, { "epoch": 7.69, "learning_rate": 2.0680000000000002e-05, "loss": 0.0134, "step": 5873 }, { "epoch": 7.69, "learning_rate": 2.0675e-05, "loss": 0.0117, "step": 5874 }, { "epoch": 7.69, "learning_rate": 2.067e-05, "loss": 0.0112, "step": 5875 }, { "epoch": 7.69, "learning_rate": 2.0665e-05, "loss": 0.0143, "step": 5876 }, { "epoch": 7.69, "learning_rate": 2.0660000000000002e-05, "loss": 0.0126, "step": 5877 }, { "epoch": 7.69, "learning_rate": 2.0655e-05, "loss": 0.0129, "step": 5878 }, { "epoch": 7.7, "learning_rate": 2.065e-05, "loss": 0.0167, "step": 5879 }, { "epoch": 7.7, "learning_rate": 2.0645e-05, "loss": 0.0138, "step": 5880 }, { "epoch": 7.7, "learning_rate": 2.0640000000000002e-05, "loss": 0.016, "step": 5881 }, { "epoch": 7.7, "learning_rate": 2.0635e-05, "loss": 0.0107, "step": 5882 }, { "epoch": 7.7, "learning_rate": 2.063e-05, "loss": 0.0113, "step": 5883 }, { "epoch": 7.7, "learning_rate": 2.0625e-05, "loss": 0.0153, "step": 5884 }, { "epoch": 7.7, "learning_rate": 2.062e-05, "loss": 0.013, "step": 5885 }, { "epoch": 7.7, "learning_rate": 2.0615e-05, "loss": 0.0144, "step": 5886 }, { "epoch": 7.71, "learning_rate": 2.061e-05, "loss": 0.0177, "step": 5887 }, { "epoch": 7.71, "learning_rate": 2.0605000000000003e-05, "loss": 0.0135, "step": 5888 }, { "epoch": 7.71, "learning_rate": 2.06e-05, "loss": 0.0144, "step": 5889 }, { "epoch": 7.71, "learning_rate": 2.0595000000000002e-05, "loss": 0.0132, "step": 5890 }, { "epoch": 7.71, "learning_rate": 2.059e-05, "loss": 0.0133, "step": 5891 }, { "epoch": 7.71, "learning_rate": 2.0585e-05, "loss": 0.0095, "step": 5892 }, { "epoch": 7.71, "learning_rate": 2.0580000000000003e-05, "loss": 0.0136, "step": 5893 }, { "epoch": 7.71, "learning_rate": 2.0575e-05, "loss": 0.0145, "step": 5894 }, { "epoch": 7.72, "learning_rate": 2.057e-05, "loss": 0.0129, "step": 5895 }, { "epoch": 7.72, "learning_rate": 2.0565e-05, "loss": 0.0161, "step": 5896 }, { "epoch": 7.72, "learning_rate": 2.0560000000000003e-05, "loss": 0.0156, "step": 5897 }, { "epoch": 7.72, "learning_rate": 2.0555000000000002e-05, "loss": 0.0143, "step": 5898 }, { "epoch": 7.72, "learning_rate": 2.055e-05, "loss": 0.0168, "step": 5899 }, { "epoch": 7.72, "learning_rate": 2.0545e-05, "loss": 0.0146, "step": 5900 }, { "epoch": 7.72, "learning_rate": 2.054e-05, "loss": 0.0149, "step": 5901 }, { "epoch": 7.73, "learning_rate": 2.0535000000000002e-05, "loss": 0.0151, "step": 5902 }, { "epoch": 7.73, "learning_rate": 2.053e-05, "loss": 0.0138, "step": 5903 }, { "epoch": 7.73, "learning_rate": 2.0525e-05, "loss": 0.0112, "step": 5904 }, { "epoch": 7.73, "learning_rate": 2.052e-05, "loss": 0.0135, "step": 5905 }, { "epoch": 7.73, "learning_rate": 2.0515e-05, "loss": 0.0139, "step": 5906 }, { "epoch": 7.73, "learning_rate": 2.0510000000000002e-05, "loss": 0.0116, "step": 5907 }, { "epoch": 7.73, "learning_rate": 2.0505e-05, "loss": 0.0137, "step": 5908 }, { "epoch": 7.73, "learning_rate": 2.05e-05, "loss": 0.0129, "step": 5909 }, { "epoch": 7.74, "learning_rate": 2.0495e-05, "loss": 0.0189, "step": 5910 }, { "epoch": 7.74, "learning_rate": 2.0490000000000002e-05, "loss": 0.0126, "step": 5911 }, { "epoch": 7.74, "learning_rate": 2.0485e-05, "loss": 0.0108, "step": 5912 }, { "epoch": 7.74, "learning_rate": 2.048e-05, "loss": 0.0156, "step": 5913 }, { "epoch": 7.74, "learning_rate": 2.0475e-05, "loss": 0.0122, "step": 5914 }, { "epoch": 7.74, "learning_rate": 2.047e-05, "loss": 0.0111, "step": 5915 }, { "epoch": 7.74, "learning_rate": 2.0465e-05, "loss": 0.0131, "step": 5916 }, { "epoch": 7.74, "learning_rate": 2.046e-05, "loss": 0.0102, "step": 5917 }, { "epoch": 7.75, "learning_rate": 2.0455000000000003e-05, "loss": 0.0127, "step": 5918 }, { "epoch": 7.75, "learning_rate": 2.045e-05, "loss": 0.0144, "step": 5919 }, { "epoch": 7.75, "learning_rate": 2.0445e-05, "loss": 0.0168, "step": 5920 }, { "epoch": 7.75, "learning_rate": 2.044e-05, "loss": 0.013, "step": 5921 }, { "epoch": 7.75, "learning_rate": 2.0435e-05, "loss": 0.0125, "step": 5922 }, { "epoch": 7.75, "learning_rate": 2.0430000000000003e-05, "loss": 0.0118, "step": 5923 }, { "epoch": 7.75, "learning_rate": 2.0425e-05, "loss": 0.0134, "step": 5924 }, { "epoch": 7.76, "learning_rate": 2.042e-05, "loss": 0.0109, "step": 5925 }, { "epoch": 7.76, "learning_rate": 2.0415e-05, "loss": 0.0162, "step": 5926 }, { "epoch": 7.76, "learning_rate": 2.0410000000000003e-05, "loss": 0.0125, "step": 5927 }, { "epoch": 7.76, "learning_rate": 2.0405000000000002e-05, "loss": 0.0156, "step": 5928 }, { "epoch": 7.76, "learning_rate": 2.04e-05, "loss": 0.0145, "step": 5929 }, { "epoch": 7.76, "learning_rate": 2.0395e-05, "loss": 0.0151, "step": 5930 }, { "epoch": 7.76, "learning_rate": 2.039e-05, "loss": 0.0142, "step": 5931 }, { "epoch": 7.76, "learning_rate": 2.0385000000000002e-05, "loss": 0.0138, "step": 5932 }, { "epoch": 7.77, "learning_rate": 2.038e-05, "loss": 0.0128, "step": 5933 }, { "epoch": 7.77, "learning_rate": 2.0375e-05, "loss": 0.0124, "step": 5934 }, { "epoch": 7.77, "learning_rate": 2.037e-05, "loss": 0.0126, "step": 5935 }, { "epoch": 7.77, "learning_rate": 2.0365000000000002e-05, "loss": 0.0132, "step": 5936 }, { "epoch": 7.77, "learning_rate": 2.036e-05, "loss": 0.013, "step": 5937 }, { "epoch": 7.77, "learning_rate": 2.0355e-05, "loss": 0.0131, "step": 5938 }, { "epoch": 7.77, "learning_rate": 2.035e-05, "loss": 0.0133, "step": 5939 }, { "epoch": 7.77, "learning_rate": 2.0345e-05, "loss": 0.0118, "step": 5940 }, { "epoch": 7.78, "learning_rate": 2.0340000000000002e-05, "loss": 0.0142, "step": 5941 }, { "epoch": 7.78, "learning_rate": 2.0335e-05, "loss": 0.0136, "step": 5942 }, { "epoch": 7.78, "learning_rate": 2.033e-05, "loss": 0.0118, "step": 5943 }, { "epoch": 7.78, "learning_rate": 2.0325e-05, "loss": 0.0225, "step": 5944 }, { "epoch": 7.78, "learning_rate": 2.032e-05, "loss": 0.0142, "step": 5945 }, { "epoch": 7.78, "learning_rate": 2.0315e-05, "loss": 0.0129, "step": 5946 }, { "epoch": 7.78, "learning_rate": 2.031e-05, "loss": 0.0115, "step": 5947 }, { "epoch": 7.79, "learning_rate": 2.0305000000000003e-05, "loss": 0.0157, "step": 5948 }, { "epoch": 7.79, "learning_rate": 2.0300000000000002e-05, "loss": 0.0133, "step": 5949 }, { "epoch": 7.79, "learning_rate": 2.0295e-05, "loss": 0.012, "step": 5950 }, { "epoch": 7.79, "learning_rate": 2.029e-05, "loss": 0.0165, "step": 5951 }, { "epoch": 7.79, "learning_rate": 2.0285e-05, "loss": 0.0142, "step": 5952 }, { "epoch": 7.79, "learning_rate": 2.0280000000000002e-05, "loss": 0.0159, "step": 5953 }, { "epoch": 7.79, "learning_rate": 2.0275e-05, "loss": 0.0161, "step": 5954 }, { "epoch": 7.79, "learning_rate": 2.027e-05, "loss": 0.0082, "step": 5955 }, { "epoch": 7.8, "learning_rate": 2.0265e-05, "loss": 0.0149, "step": 5956 }, { "epoch": 7.8, "learning_rate": 2.0260000000000003e-05, "loss": 0.0151, "step": 5957 }, { "epoch": 7.8, "learning_rate": 2.0255000000000002e-05, "loss": 0.0125, "step": 5958 }, { "epoch": 7.8, "learning_rate": 2.025e-05, "loss": 0.0108, "step": 5959 }, { "epoch": 7.8, "learning_rate": 2.0245e-05, "loss": 0.0141, "step": 5960 }, { "epoch": 7.8, "learning_rate": 2.024e-05, "loss": 0.0156, "step": 5961 }, { "epoch": 7.8, "learning_rate": 2.0235000000000002e-05, "loss": 0.0118, "step": 5962 }, { "epoch": 7.8, "learning_rate": 2.023e-05, "loss": 0.0119, "step": 5963 }, { "epoch": 7.81, "learning_rate": 2.0225000000000004e-05, "loss": 0.0174, "step": 5964 }, { "epoch": 7.81, "learning_rate": 2.022e-05, "loss": 0.0163, "step": 5965 }, { "epoch": 7.81, "learning_rate": 2.0215000000000002e-05, "loss": 0.0154, "step": 5966 }, { "epoch": 7.81, "learning_rate": 2.021e-05, "loss": 0.0187, "step": 5967 }, { "epoch": 7.81, "learning_rate": 2.0205e-05, "loss": 0.0133, "step": 5968 }, { "epoch": 7.81, "learning_rate": 2.0200000000000003e-05, "loss": 0.0184, "step": 5969 }, { "epoch": 7.81, "learning_rate": 2.0195e-05, "loss": 0.0139, "step": 5970 }, { "epoch": 7.82, "learning_rate": 2.019e-05, "loss": 0.0179, "step": 5971 }, { "epoch": 7.82, "learning_rate": 2.0185e-05, "loss": 0.0131, "step": 5972 }, { "epoch": 7.82, "learning_rate": 2.0180000000000003e-05, "loss": 0.0131, "step": 5973 }, { "epoch": 7.82, "learning_rate": 2.0175000000000003e-05, "loss": 0.0112, "step": 5974 }, { "epoch": 7.82, "learning_rate": 2.017e-05, "loss": 0.0146, "step": 5975 }, { "epoch": 7.82, "learning_rate": 2.0165e-05, "loss": 0.0146, "step": 5976 }, { "epoch": 7.82, "learning_rate": 2.016e-05, "loss": 0.0134, "step": 5977 }, { "epoch": 7.82, "learning_rate": 2.0155000000000003e-05, "loss": 0.0149, "step": 5978 }, { "epoch": 7.83, "learning_rate": 2.0150000000000002e-05, "loss": 0.0147, "step": 5979 }, { "epoch": 7.83, "learning_rate": 2.0145e-05, "loss": 0.0123, "step": 5980 }, { "epoch": 7.83, "learning_rate": 2.014e-05, "loss": 0.0152, "step": 5981 }, { "epoch": 7.83, "learning_rate": 2.0135e-05, "loss": 0.0155, "step": 5982 }, { "epoch": 7.83, "learning_rate": 2.0130000000000002e-05, "loss": 0.0151, "step": 5983 }, { "epoch": 7.83, "learning_rate": 2.0125e-05, "loss": 0.0152, "step": 5984 }, { "epoch": 7.83, "learning_rate": 2.012e-05, "loss": 0.0118, "step": 5985 }, { "epoch": 7.84, "learning_rate": 2.0115e-05, "loss": 0.0137, "step": 5986 }, { "epoch": 7.84, "learning_rate": 2.0110000000000002e-05, "loss": 0.0136, "step": 5987 }, { "epoch": 7.84, "learning_rate": 2.0105e-05, "loss": 0.0143, "step": 5988 }, { "epoch": 7.84, "learning_rate": 2.01e-05, "loss": 0.0167, "step": 5989 }, { "epoch": 7.84, "learning_rate": 2.0095e-05, "loss": 0.0155, "step": 5990 }, { "epoch": 7.84, "learning_rate": 2.009e-05, "loss": 0.0163, "step": 5991 }, { "epoch": 7.84, "learning_rate": 2.0085e-05, "loss": 0.0122, "step": 5992 }, { "epoch": 7.84, "learning_rate": 2.008e-05, "loss": 0.0151, "step": 5993 }, { "epoch": 7.85, "learning_rate": 2.0075000000000003e-05, "loss": 0.0146, "step": 5994 }, { "epoch": 7.85, "learning_rate": 2.007e-05, "loss": 0.0129, "step": 5995 }, { "epoch": 7.85, "learning_rate": 2.0065000000000002e-05, "loss": 0.0141, "step": 5996 }, { "epoch": 7.85, "learning_rate": 2.006e-05, "loss": 0.017, "step": 5997 }, { "epoch": 7.85, "learning_rate": 2.0055e-05, "loss": 0.015, "step": 5998 }, { "epoch": 7.85, "learning_rate": 2.0050000000000003e-05, "loss": 0.0124, "step": 5999 }, { "epoch": 7.85, "learning_rate": 2.0045e-05, "loss": 0.0161, "step": 6000 }, { "epoch": 7.85, "learning_rate": 2.004e-05, "loss": 0.0143, "step": 6001 }, { "epoch": 7.86, "learning_rate": 2.0035e-05, "loss": 0.0118, "step": 6002 }, { "epoch": 7.86, "learning_rate": 2.0030000000000003e-05, "loss": 0.0171, "step": 6003 }, { "epoch": 7.86, "learning_rate": 2.0025000000000002e-05, "loss": 0.0145, "step": 6004 }, { "epoch": 7.86, "learning_rate": 2.002e-05, "loss": 0.0136, "step": 6005 }, { "epoch": 7.86, "learning_rate": 2.0015e-05, "loss": 0.0136, "step": 6006 }, { "epoch": 7.86, "learning_rate": 2.001e-05, "loss": 0.014, "step": 6007 }, { "epoch": 7.86, "learning_rate": 2.0005000000000002e-05, "loss": 0.0111, "step": 6008 }, { "epoch": 7.87, "learning_rate": 2e-05, "loss": 0.0128, "step": 6009 }, { "epoch": 7.87, "learning_rate": 1.9995e-05, "loss": 0.0151, "step": 6010 }, { "epoch": 7.87, "learning_rate": 1.999e-05, "loss": 0.0126, "step": 6011 }, { "epoch": 7.87, "learning_rate": 1.9985000000000003e-05, "loss": 0.0128, "step": 6012 }, { "epoch": 7.87, "learning_rate": 1.9980000000000002e-05, "loss": 0.014, "step": 6013 }, { "epoch": 7.87, "learning_rate": 1.9975e-05, "loss": 0.0151, "step": 6014 }, { "epoch": 7.87, "learning_rate": 1.997e-05, "loss": 0.0117, "step": 6015 }, { "epoch": 7.87, "learning_rate": 1.9965e-05, "loss": 0.0122, "step": 6016 }, { "epoch": 7.88, "learning_rate": 1.9960000000000002e-05, "loss": 0.0112, "step": 6017 }, { "epoch": 7.88, "learning_rate": 1.9955e-05, "loss": 0.0174, "step": 6018 }, { "epoch": 7.88, "learning_rate": 1.995e-05, "loss": 0.0151, "step": 6019 }, { "epoch": 7.88, "learning_rate": 1.9945e-05, "loss": 0.0143, "step": 6020 }, { "epoch": 7.88, "learning_rate": 1.994e-05, "loss": 0.0186, "step": 6021 }, { "epoch": 7.88, "learning_rate": 1.9935e-05, "loss": 0.0148, "step": 6022 }, { "epoch": 7.88, "learning_rate": 1.993e-05, "loss": 0.0137, "step": 6023 }, { "epoch": 7.88, "learning_rate": 1.9925000000000003e-05, "loss": 0.0129, "step": 6024 }, { "epoch": 7.89, "learning_rate": 1.992e-05, "loss": 0.0143, "step": 6025 }, { "epoch": 7.89, "learning_rate": 1.9915e-05, "loss": 0.0119, "step": 6026 }, { "epoch": 7.89, "learning_rate": 1.991e-05, "loss": 0.0095, "step": 6027 }, { "epoch": 7.89, "learning_rate": 1.9905e-05, "loss": 0.0152, "step": 6028 }, { "epoch": 7.89, "learning_rate": 1.9900000000000003e-05, "loss": 0.0149, "step": 6029 }, { "epoch": 7.89, "learning_rate": 1.9895e-05, "loss": 0.0154, "step": 6030 }, { "epoch": 7.89, "learning_rate": 1.989e-05, "loss": 0.0136, "step": 6031 }, { "epoch": 7.9, "learning_rate": 1.9885e-05, "loss": 0.012, "step": 6032 }, { "epoch": 7.9, "learning_rate": 1.9880000000000003e-05, "loss": 0.0124, "step": 6033 }, { "epoch": 7.9, "learning_rate": 1.9875000000000002e-05, "loss": 0.0159, "step": 6034 }, { "epoch": 7.9, "learning_rate": 1.987e-05, "loss": 0.0147, "step": 6035 }, { "epoch": 7.9, "learning_rate": 1.9865e-05, "loss": 0.0124, "step": 6036 }, { "epoch": 7.9, "learning_rate": 1.986e-05, "loss": 0.011, "step": 6037 }, { "epoch": 7.9, "learning_rate": 1.9855000000000002e-05, "loss": 0.0177, "step": 6038 }, { "epoch": 7.9, "learning_rate": 1.985e-05, "loss": 0.0138, "step": 6039 }, { "epoch": 7.91, "learning_rate": 1.9845e-05, "loss": 0.0132, "step": 6040 }, { "epoch": 7.91, "learning_rate": 1.984e-05, "loss": 0.0129, "step": 6041 }, { "epoch": 7.91, "learning_rate": 1.9835000000000002e-05, "loss": 0.0161, "step": 6042 }, { "epoch": 7.91, "learning_rate": 1.983e-05, "loss": 0.0101, "step": 6043 }, { "epoch": 7.91, "learning_rate": 1.9825e-05, "loss": 0.0188, "step": 6044 }, { "epoch": 7.91, "learning_rate": 1.982e-05, "loss": 0.0113, "step": 6045 }, { "epoch": 7.91, "learning_rate": 1.9815e-05, "loss": 0.0168, "step": 6046 }, { "epoch": 7.91, "learning_rate": 1.9810000000000002e-05, "loss": 0.0231, "step": 6047 }, { "epoch": 7.92, "learning_rate": 1.9805e-05, "loss": 0.015, "step": 6048 }, { "epoch": 7.92, "learning_rate": 1.9800000000000004e-05, "loss": 0.0123, "step": 6049 }, { "epoch": 7.92, "learning_rate": 1.9795e-05, "loss": 0.0149, "step": 6050 }, { "epoch": 7.92, "learning_rate": 1.979e-05, "loss": 0.0122, "step": 6051 }, { "epoch": 7.92, "learning_rate": 1.9785e-05, "loss": 0.014, "step": 6052 }, { "epoch": 7.92, "learning_rate": 1.978e-05, "loss": 0.0168, "step": 6053 }, { "epoch": 7.92, "learning_rate": 1.9775000000000003e-05, "loss": 0.0123, "step": 6054 }, { "epoch": 7.93, "learning_rate": 1.977e-05, "loss": 0.0125, "step": 6055 }, { "epoch": 7.93, "learning_rate": 1.9765e-05, "loss": 0.0136, "step": 6056 }, { "epoch": 7.93, "learning_rate": 1.976e-05, "loss": 0.0151, "step": 6057 }, { "epoch": 7.93, "learning_rate": 1.9755e-05, "loss": 0.0144, "step": 6058 }, { "epoch": 7.93, "learning_rate": 1.9750000000000002e-05, "loss": 0.0135, "step": 6059 }, { "epoch": 7.93, "learning_rate": 1.9744999999999998e-05, "loss": 0.012, "step": 6060 }, { "epoch": 7.93, "learning_rate": 1.974e-05, "loss": 0.0125, "step": 6061 }, { "epoch": 7.93, "learning_rate": 1.9735e-05, "loss": 0.0164, "step": 6062 }, { "epoch": 7.94, "learning_rate": 1.9730000000000003e-05, "loss": 0.0155, "step": 6063 }, { "epoch": 7.94, "learning_rate": 1.9725000000000002e-05, "loss": 0.0118, "step": 6064 }, { "epoch": 7.94, "learning_rate": 1.972e-05, "loss": 0.0127, "step": 6065 }, { "epoch": 7.94, "learning_rate": 1.9715e-05, "loss": 0.0164, "step": 6066 }, { "epoch": 7.94, "learning_rate": 1.971e-05, "loss": 0.0177, "step": 6067 }, { "epoch": 7.94, "learning_rate": 1.9705000000000002e-05, "loss": 0.0146, "step": 6068 }, { "epoch": 7.94, "learning_rate": 1.97e-05, "loss": 0.0166, "step": 6069 }, { "epoch": 7.95, "learning_rate": 1.9695e-05, "loss": 0.0145, "step": 6070 }, { "epoch": 7.95, "learning_rate": 1.969e-05, "loss": 0.015, "step": 6071 }, { "epoch": 7.95, "learning_rate": 1.9685000000000002e-05, "loss": 0.0148, "step": 6072 }, { "epoch": 7.95, "learning_rate": 1.968e-05, "loss": 0.0157, "step": 6073 }, { "epoch": 7.95, "learning_rate": 1.9675e-05, "loss": 0.0159, "step": 6074 }, { "epoch": 7.95, "learning_rate": 1.9670000000000003e-05, "loss": 0.0117, "step": 6075 }, { "epoch": 7.95, "learning_rate": 1.9665e-05, "loss": 0.0144, "step": 6076 }, { "epoch": 7.95, "learning_rate": 1.966e-05, "loss": 0.0156, "step": 6077 }, { "epoch": 7.96, "learning_rate": 1.9655e-05, "loss": 0.0129, "step": 6078 }, { "epoch": 7.96, "learning_rate": 1.9650000000000003e-05, "loss": 0.0146, "step": 6079 }, { "epoch": 7.96, "learning_rate": 1.9645000000000002e-05, "loss": 0.0118, "step": 6080 }, { "epoch": 7.96, "learning_rate": 1.9640000000000002e-05, "loss": 0.0166, "step": 6081 }, { "epoch": 7.96, "learning_rate": 1.9635e-05, "loss": 0.0131, "step": 6082 }, { "epoch": 7.96, "learning_rate": 1.963e-05, "loss": 0.0152, "step": 6083 }, { "epoch": 7.96, "learning_rate": 1.9625000000000003e-05, "loss": 0.0159, "step": 6084 }, { "epoch": 7.96, "learning_rate": 1.9620000000000002e-05, "loss": 0.0153, "step": 6085 }, { "epoch": 7.97, "learning_rate": 1.9615e-05, "loss": 0.0097, "step": 6086 }, { "epoch": 7.97, "learning_rate": 1.961e-05, "loss": 0.0149, "step": 6087 }, { "epoch": 7.97, "learning_rate": 1.9605e-05, "loss": 0.0143, "step": 6088 }, { "epoch": 7.97, "learning_rate": 1.9600000000000002e-05, "loss": 0.0183, "step": 6089 }, { "epoch": 7.97, "learning_rate": 1.9595e-05, "loss": 0.0143, "step": 6090 }, { "epoch": 7.97, "learning_rate": 1.959e-05, "loss": 0.0147, "step": 6091 }, { "epoch": 7.97, "learning_rate": 1.9585e-05, "loss": 0.0131, "step": 6092 }, { "epoch": 7.98, "learning_rate": 1.9580000000000002e-05, "loss": 0.0145, "step": 6093 }, { "epoch": 7.98, "learning_rate": 1.9575e-05, "loss": 0.0137, "step": 6094 }, { "epoch": 7.98, "learning_rate": 1.957e-05, "loss": 0.0129, "step": 6095 }, { "epoch": 7.98, "learning_rate": 1.9565e-05, "loss": 0.0137, "step": 6096 }, { "epoch": 7.98, "learning_rate": 1.956e-05, "loss": 0.0148, "step": 6097 }, { "epoch": 7.98, "learning_rate": 1.9555e-05, "loss": 0.0182, "step": 6098 }, { "epoch": 7.98, "learning_rate": 1.955e-05, "loss": 0.012, "step": 6099 }, { "epoch": 7.98, "learning_rate": 1.9545000000000003e-05, "loss": 0.0128, "step": 6100 }, { "epoch": 7.99, "learning_rate": 1.954e-05, "loss": 0.0122, "step": 6101 }, { "epoch": 7.99, "learning_rate": 1.9535000000000002e-05, "loss": 0.0107, "step": 6102 }, { "epoch": 7.99, "learning_rate": 1.953e-05, "loss": 0.016, "step": 6103 }, { "epoch": 7.99, "learning_rate": 1.9525e-05, "loss": 0.0164, "step": 6104 }, { "epoch": 7.99, "learning_rate": 1.9520000000000003e-05, "loss": 0.017, "step": 6105 }, { "epoch": 7.99, "learning_rate": 1.9515e-05, "loss": 0.0149, "step": 6106 }, { "epoch": 7.99, "learning_rate": 1.951e-05, "loss": 0.0156, "step": 6107 }, { "epoch": 7.99, "learning_rate": 1.9505e-05, "loss": 0.0147, "step": 6108 }, { "epoch": 8.0, "learning_rate": 1.9500000000000003e-05, "loss": 0.0146, "step": 6109 }, { "epoch": 8.0, "learning_rate": 1.9495000000000002e-05, "loss": 0.0149, "step": 6110 }, { "epoch": 8.0, "learning_rate": 1.949e-05, "loss": 0.0209, "step": 6111 }, { "epoch": 8.0, "learning_rate": 1.9485e-05, "loss": 0.0116, "step": 6112 }, { "epoch": 8.0, "learning_rate": 1.948e-05, "loss": 0.0082, "step": 6113 }, { "epoch": 8.0, "learning_rate": 1.9475000000000002e-05, "loss": 0.0099, "step": 6114 }, { "epoch": 8.0, "learning_rate": 1.947e-05, "loss": 0.0075, "step": 6115 }, { "epoch": 8.01, "learning_rate": 1.9465e-05, "loss": 0.0091, "step": 6116 }, { "epoch": 8.01, "learning_rate": 1.946e-05, "loss": 0.0098, "step": 6117 }, { "epoch": 8.01, "learning_rate": 1.9455000000000003e-05, "loss": 0.0077, "step": 6118 }, { "epoch": 8.01, "learning_rate": 1.9450000000000002e-05, "loss": 0.008, "step": 6119 }, { "epoch": 8.01, "learning_rate": 1.9445e-05, "loss": 0.0085, "step": 6120 }, { "epoch": 8.01, "learning_rate": 1.944e-05, "loss": 0.0087, "step": 6121 }, { "epoch": 8.01, "learning_rate": 1.9435e-05, "loss": 0.0075, "step": 6122 }, { "epoch": 8.01, "learning_rate": 1.9430000000000002e-05, "loss": 0.0097, "step": 6123 }, { "epoch": 8.02, "learning_rate": 1.9425e-05, "loss": 0.0078, "step": 6124 }, { "epoch": 8.02, "learning_rate": 1.942e-05, "loss": 0.0095, "step": 6125 }, { "epoch": 8.02, "learning_rate": 1.9415e-05, "loss": 0.0099, "step": 6126 }, { "epoch": 8.02, "learning_rate": 1.941e-05, "loss": 0.0086, "step": 6127 }, { "epoch": 8.02, "learning_rate": 1.9405e-05, "loss": 0.0102, "step": 6128 }, { "epoch": 8.02, "learning_rate": 1.94e-05, "loss": 0.0089, "step": 6129 }, { "epoch": 8.02, "learning_rate": 1.9395000000000003e-05, "loss": 0.008, "step": 6130 }, { "epoch": 8.02, "learning_rate": 1.939e-05, "loss": 0.0108, "step": 6131 }, { "epoch": 8.03, "learning_rate": 1.9385e-05, "loss": 0.0093, "step": 6132 }, { "epoch": 8.03, "learning_rate": 1.938e-05, "loss": 0.0083, "step": 6133 }, { "epoch": 8.03, "learning_rate": 1.9375e-05, "loss": 0.0086, "step": 6134 }, { "epoch": 8.03, "learning_rate": 1.9370000000000003e-05, "loss": 0.007, "step": 6135 }, { "epoch": 8.03, "learning_rate": 1.9365e-05, "loss": 0.0134, "step": 6136 }, { "epoch": 8.03, "learning_rate": 1.936e-05, "loss": 0.0086, "step": 6137 }, { "epoch": 8.03, "learning_rate": 1.9355e-05, "loss": 0.0077, "step": 6138 }, { "epoch": 8.04, "learning_rate": 1.9350000000000003e-05, "loss": 0.0076, "step": 6139 }, { "epoch": 8.04, "learning_rate": 1.9345000000000002e-05, "loss": 0.0104, "step": 6140 }, { "epoch": 8.04, "learning_rate": 1.934e-05, "loss": 0.0085, "step": 6141 }, { "epoch": 8.04, "learning_rate": 1.9335e-05, "loss": 0.0076, "step": 6142 }, { "epoch": 8.04, "learning_rate": 1.933e-05, "loss": 0.0072, "step": 6143 }, { "epoch": 8.04, "learning_rate": 1.9325000000000002e-05, "loss": 0.0077, "step": 6144 }, { "epoch": 8.04, "learning_rate": 1.932e-05, "loss": 0.0087, "step": 6145 }, { "epoch": 8.04, "learning_rate": 1.9315e-05, "loss": 0.0074, "step": 6146 }, { "epoch": 8.05, "learning_rate": 1.931e-05, "loss": 0.0108, "step": 6147 }, { "epoch": 8.05, "learning_rate": 1.9305000000000002e-05, "loss": 0.0085, "step": 6148 }, { "epoch": 8.05, "learning_rate": 1.93e-05, "loss": 0.0082, "step": 6149 }, { "epoch": 8.05, "learning_rate": 1.9295e-05, "loss": 0.009, "step": 6150 }, { "epoch": 8.05, "learning_rate": 1.929e-05, "loss": 0.0119, "step": 6151 }, { "epoch": 8.05, "learning_rate": 1.9285e-05, "loss": 0.0088, "step": 6152 }, { "epoch": 8.05, "learning_rate": 1.9280000000000002e-05, "loss": 0.0106, "step": 6153 }, { "epoch": 8.05, "learning_rate": 1.9275e-05, "loss": 0.0105, "step": 6154 }, { "epoch": 8.06, "learning_rate": 1.9270000000000004e-05, "loss": 0.0083, "step": 6155 }, { "epoch": 8.06, "learning_rate": 1.9265e-05, "loss": 0.0104, "step": 6156 }, { "epoch": 8.06, "learning_rate": 1.9260000000000002e-05, "loss": 0.009, "step": 6157 }, { "epoch": 8.06, "learning_rate": 1.9255e-05, "loss": 0.0087, "step": 6158 }, { "epoch": 8.06, "learning_rate": 1.925e-05, "loss": 0.0094, "step": 6159 }, { "epoch": 8.06, "learning_rate": 1.9245000000000003e-05, "loss": 0.0095, "step": 6160 }, { "epoch": 8.06, "learning_rate": 1.924e-05, "loss": 0.009, "step": 6161 }, { "epoch": 8.07, "learning_rate": 1.9235e-05, "loss": 0.0092, "step": 6162 }, { "epoch": 8.07, "learning_rate": 1.923e-05, "loss": 0.0094, "step": 6163 }, { "epoch": 8.07, "learning_rate": 1.9225e-05, "loss": 0.009, "step": 6164 }, { "epoch": 8.07, "learning_rate": 1.9220000000000002e-05, "loss": 0.0077, "step": 6165 }, { "epoch": 8.07, "learning_rate": 1.9214999999999998e-05, "loss": 0.0088, "step": 6166 }, { "epoch": 8.07, "learning_rate": 1.921e-05, "loss": 0.0077, "step": 6167 }, { "epoch": 8.07, "learning_rate": 1.9205e-05, "loss": 0.0096, "step": 6168 }, { "epoch": 8.07, "learning_rate": 1.9200000000000003e-05, "loss": 0.0113, "step": 6169 }, { "epoch": 8.08, "learning_rate": 1.9195000000000002e-05, "loss": 0.0084, "step": 6170 }, { "epoch": 8.08, "learning_rate": 1.919e-05, "loss": 0.0086, "step": 6171 }, { "epoch": 8.08, "learning_rate": 1.9185e-05, "loss": 0.0065, "step": 6172 }, { "epoch": 8.08, "learning_rate": 1.918e-05, "loss": 0.0087, "step": 6173 }, { "epoch": 8.08, "learning_rate": 1.9175000000000002e-05, "loss": 0.0089, "step": 6174 }, { "epoch": 8.08, "learning_rate": 1.917e-05, "loss": 0.0107, "step": 6175 }, { "epoch": 8.08, "learning_rate": 1.9165e-05, "loss": 0.0083, "step": 6176 }, { "epoch": 8.09, "learning_rate": 1.916e-05, "loss": 0.0094, "step": 6177 }, { "epoch": 8.09, "learning_rate": 1.9155000000000002e-05, "loss": 0.0093, "step": 6178 }, { "epoch": 8.09, "learning_rate": 1.915e-05, "loss": 0.0084, "step": 6179 }, { "epoch": 8.09, "learning_rate": 1.9145e-05, "loss": 0.0085, "step": 6180 }, { "epoch": 8.09, "learning_rate": 1.914e-05, "loss": 0.0086, "step": 6181 }, { "epoch": 8.09, "learning_rate": 1.9135e-05, "loss": 0.0077, "step": 6182 }, { "epoch": 8.09, "learning_rate": 1.913e-05, "loss": 0.0092, "step": 6183 }, { "epoch": 8.09, "learning_rate": 1.9125e-05, "loss": 0.0092, "step": 6184 }, { "epoch": 8.1, "learning_rate": 1.9120000000000003e-05, "loss": 0.0093, "step": 6185 }, { "epoch": 8.1, "learning_rate": 1.9115e-05, "loss": 0.0102, "step": 6186 }, { "epoch": 8.1, "learning_rate": 1.911e-05, "loss": 0.0094, "step": 6187 }, { "epoch": 8.1, "learning_rate": 1.9105e-05, "loss": 0.0073, "step": 6188 }, { "epoch": 8.1, "learning_rate": 1.91e-05, "loss": 0.0093, "step": 6189 }, { "epoch": 8.1, "learning_rate": 1.9095000000000003e-05, "loss": 0.0112, "step": 6190 }, { "epoch": 8.1, "learning_rate": 1.909e-05, "loss": 0.0079, "step": 6191 }, { "epoch": 8.1, "learning_rate": 1.9085e-05, "loss": 0.0091, "step": 6192 }, { "epoch": 8.11, "learning_rate": 1.908e-05, "loss": 0.0099, "step": 6193 }, { "epoch": 8.11, "learning_rate": 1.9075000000000003e-05, "loss": 0.0084, "step": 6194 }, { "epoch": 8.11, "learning_rate": 1.9070000000000002e-05, "loss": 0.0098, "step": 6195 }, { "epoch": 8.11, "learning_rate": 1.9064999999999998e-05, "loss": 0.0089, "step": 6196 }, { "epoch": 8.11, "learning_rate": 1.906e-05, "loss": 0.0077, "step": 6197 }, { "epoch": 8.11, "learning_rate": 1.9055e-05, "loss": 0.0077, "step": 6198 }, { "epoch": 8.11, "learning_rate": 1.9050000000000002e-05, "loss": 0.009, "step": 6199 }, { "epoch": 8.12, "learning_rate": 1.9045e-05, "loss": 0.0159, "step": 6200 }, { "epoch": 8.12, "learning_rate": 1.904e-05, "loss": 0.0105, "step": 6201 }, { "epoch": 8.12, "learning_rate": 1.9035e-05, "loss": 0.0095, "step": 6202 }, { "epoch": 8.12, "learning_rate": 1.903e-05, "loss": 0.0108, "step": 6203 }, { "epoch": 8.12, "learning_rate": 1.9025e-05, "loss": 0.0099, "step": 6204 }, { "epoch": 8.12, "learning_rate": 1.902e-05, "loss": 0.009, "step": 6205 }, { "epoch": 8.12, "learning_rate": 1.9015000000000003e-05, "loss": 0.0093, "step": 6206 }, { "epoch": 8.12, "learning_rate": 1.901e-05, "loss": 0.0084, "step": 6207 }, { "epoch": 8.13, "learning_rate": 1.9005000000000002e-05, "loss": 0.01, "step": 6208 }, { "epoch": 8.13, "learning_rate": 1.9e-05, "loss": 0.0086, "step": 6209 }, { "epoch": 8.13, "learning_rate": 1.8995e-05, "loss": 0.0089, "step": 6210 }, { "epoch": 8.13, "learning_rate": 1.8990000000000003e-05, "loss": 0.0061, "step": 6211 }, { "epoch": 8.13, "learning_rate": 1.8985e-05, "loss": 0.0063, "step": 6212 }, { "epoch": 8.13, "learning_rate": 1.898e-05, "loss": 0.0072, "step": 6213 }, { "epoch": 8.13, "learning_rate": 1.8975e-05, "loss": 0.0117, "step": 6214 }, { "epoch": 8.13, "learning_rate": 1.8970000000000003e-05, "loss": 0.0098, "step": 6215 }, { "epoch": 8.14, "learning_rate": 1.8965000000000002e-05, "loss": 0.0071, "step": 6216 }, { "epoch": 8.14, "learning_rate": 1.896e-05, "loss": 0.0097, "step": 6217 }, { "epoch": 8.14, "learning_rate": 1.8955e-05, "loss": 0.0083, "step": 6218 }, { "epoch": 8.14, "learning_rate": 1.895e-05, "loss": 0.0105, "step": 6219 }, { "epoch": 8.14, "learning_rate": 1.8945000000000002e-05, "loss": 0.007, "step": 6220 }, { "epoch": 8.14, "learning_rate": 1.894e-05, "loss": 0.009, "step": 6221 }, { "epoch": 8.14, "learning_rate": 1.8935e-05, "loss": 0.0079, "step": 6222 }, { "epoch": 8.15, "learning_rate": 1.893e-05, "loss": 0.0083, "step": 6223 }, { "epoch": 8.15, "learning_rate": 1.8925000000000003e-05, "loss": 0.0068, "step": 6224 }, { "epoch": 8.15, "learning_rate": 1.8920000000000002e-05, "loss": 0.0108, "step": 6225 }, { "epoch": 8.15, "learning_rate": 1.8915e-05, "loss": 0.0077, "step": 6226 }, { "epoch": 8.15, "learning_rate": 1.891e-05, "loss": 0.0098, "step": 6227 }, { "epoch": 8.15, "learning_rate": 1.8905e-05, "loss": 0.0066, "step": 6228 }, { "epoch": 8.15, "learning_rate": 1.8900000000000002e-05, "loss": 0.0076, "step": 6229 }, { "epoch": 8.15, "learning_rate": 1.8895e-05, "loss": 0.0181, "step": 6230 }, { "epoch": 8.16, "learning_rate": 1.8890000000000004e-05, "loss": 0.0089, "step": 6231 }, { "epoch": 8.16, "learning_rate": 1.8885e-05, "loss": 0.008, "step": 6232 }, { "epoch": 8.16, "learning_rate": 1.888e-05, "loss": 0.0097, "step": 6233 }, { "epoch": 8.16, "learning_rate": 1.8875e-05, "loss": 0.0077, "step": 6234 }, { "epoch": 8.16, "learning_rate": 1.887e-05, "loss": 0.0124, "step": 6235 }, { "epoch": 8.16, "learning_rate": 1.8865000000000003e-05, "loss": 0.0079, "step": 6236 }, { "epoch": 8.16, "learning_rate": 1.886e-05, "loss": 0.0098, "step": 6237 }, { "epoch": 8.16, "learning_rate": 1.8855e-05, "loss": 0.0096, "step": 6238 }, { "epoch": 8.17, "learning_rate": 1.885e-05, "loss": 0.0087, "step": 6239 }, { "epoch": 8.17, "learning_rate": 1.8845e-05, "loss": 0.0078, "step": 6240 }, { "epoch": 8.17, "learning_rate": 1.8840000000000003e-05, "loss": 0.0121, "step": 6241 }, { "epoch": 8.17, "learning_rate": 1.8835e-05, "loss": 0.0078, "step": 6242 }, { "epoch": 8.17, "learning_rate": 1.883e-05, "loss": 0.008, "step": 6243 }, { "epoch": 8.17, "learning_rate": 1.8825e-05, "loss": 0.008, "step": 6244 }, { "epoch": 8.17, "learning_rate": 1.8820000000000003e-05, "loss": 0.0101, "step": 6245 }, { "epoch": 8.18, "learning_rate": 1.8815000000000002e-05, "loss": 0.008, "step": 6246 }, { "epoch": 8.18, "learning_rate": 1.881e-05, "loss": 0.0096, "step": 6247 }, { "epoch": 8.18, "learning_rate": 1.8805e-05, "loss": 0.0081, "step": 6248 }, { "epoch": 8.18, "learning_rate": 1.88e-05, "loss": 0.0112, "step": 6249 }, { "epoch": 8.18, "learning_rate": 1.8795000000000002e-05, "loss": 0.0085, "step": 6250 }, { "epoch": 8.18, "learning_rate": 1.879e-05, "loss": 0.0066, "step": 6251 }, { "epoch": 8.18, "learning_rate": 1.8785e-05, "loss": 0.0089, "step": 6252 }, { "epoch": 8.18, "learning_rate": 1.878e-05, "loss": 0.0079, "step": 6253 }, { "epoch": 8.19, "learning_rate": 1.8775000000000002e-05, "loss": 0.0084, "step": 6254 }, { "epoch": 8.19, "learning_rate": 1.877e-05, "loss": 0.0097, "step": 6255 }, { "epoch": 8.19, "learning_rate": 1.8765e-05, "loss": 0.0119, "step": 6256 }, { "epoch": 8.19, "learning_rate": 1.876e-05, "loss": 0.009, "step": 6257 }, { "epoch": 8.19, "learning_rate": 1.8755e-05, "loss": 0.01, "step": 6258 }, { "epoch": 8.19, "learning_rate": 1.8750000000000002e-05, "loss": 0.0093, "step": 6259 }, { "epoch": 8.19, "learning_rate": 1.8745e-05, "loss": 0.0085, "step": 6260 }, { "epoch": 8.2, "learning_rate": 1.8740000000000004e-05, "loss": 0.0084, "step": 6261 }, { "epoch": 8.2, "learning_rate": 1.8735e-05, "loss": 0.01, "step": 6262 }, { "epoch": 8.2, "learning_rate": 1.8730000000000002e-05, "loss": 0.0102, "step": 6263 }, { "epoch": 8.2, "learning_rate": 1.8725e-05, "loss": 0.008, "step": 6264 }, { "epoch": 8.2, "learning_rate": 1.872e-05, "loss": 0.0073, "step": 6265 }, { "epoch": 8.2, "learning_rate": 1.8715000000000003e-05, "loss": 0.0091, "step": 6266 }, { "epoch": 8.2, "learning_rate": 1.871e-05, "loss": 0.0096, "step": 6267 }, { "epoch": 8.2, "learning_rate": 1.8705e-05, "loss": 0.0074, "step": 6268 }, { "epoch": 8.21, "learning_rate": 1.87e-05, "loss": 0.0081, "step": 6269 }, { "epoch": 8.21, "learning_rate": 1.8695e-05, "loss": 0.0111, "step": 6270 }, { "epoch": 8.21, "learning_rate": 1.8690000000000002e-05, "loss": 0.0079, "step": 6271 }, { "epoch": 8.21, "learning_rate": 1.8684999999999998e-05, "loss": 0.0107, "step": 6272 }, { "epoch": 8.21, "learning_rate": 1.868e-05, "loss": 0.0083, "step": 6273 }, { "epoch": 8.21, "learning_rate": 1.8675e-05, "loss": 0.0079, "step": 6274 }, { "epoch": 8.21, "learning_rate": 1.8670000000000003e-05, "loss": 0.0094, "step": 6275 }, { "epoch": 8.21, "learning_rate": 1.8665000000000002e-05, "loss": 0.0104, "step": 6276 }, { "epoch": 8.22, "learning_rate": 1.866e-05, "loss": 0.01, "step": 6277 }, { "epoch": 8.22, "learning_rate": 1.8655e-05, "loss": 0.0077, "step": 6278 }, { "epoch": 8.22, "learning_rate": 1.865e-05, "loss": 0.0079, "step": 6279 }, { "epoch": 8.22, "learning_rate": 1.8645000000000002e-05, "loss": 0.0104, "step": 6280 }, { "epoch": 8.22, "learning_rate": 1.864e-05, "loss": 0.0076, "step": 6281 }, { "epoch": 8.22, "learning_rate": 1.8635e-05, "loss": 0.0087, "step": 6282 }, { "epoch": 8.22, "learning_rate": 1.863e-05, "loss": 0.0069, "step": 6283 }, { "epoch": 8.23, "learning_rate": 1.8625000000000002e-05, "loss": 0.0089, "step": 6284 }, { "epoch": 8.23, "learning_rate": 1.862e-05, "loss": 0.0077, "step": 6285 }, { "epoch": 8.23, "learning_rate": 1.8615e-05, "loss": 0.0079, "step": 6286 }, { "epoch": 8.23, "learning_rate": 1.861e-05, "loss": 0.008, "step": 6287 }, { "epoch": 8.23, "learning_rate": 1.8605e-05, "loss": 0.0089, "step": 6288 }, { "epoch": 8.23, "learning_rate": 1.86e-05, "loss": 0.0077, "step": 6289 }, { "epoch": 8.23, "learning_rate": 1.8595e-05, "loss": 0.0101, "step": 6290 }, { "epoch": 8.23, "learning_rate": 1.8590000000000003e-05, "loss": 0.0078, "step": 6291 }, { "epoch": 8.24, "learning_rate": 1.8585e-05, "loss": 0.0086, "step": 6292 }, { "epoch": 8.24, "learning_rate": 1.858e-05, "loss": 0.0091, "step": 6293 }, { "epoch": 8.24, "learning_rate": 1.8575e-05, "loss": 0.0079, "step": 6294 }, { "epoch": 8.24, "learning_rate": 1.857e-05, "loss": 0.0097, "step": 6295 }, { "epoch": 8.24, "learning_rate": 1.8565000000000003e-05, "loss": 0.0108, "step": 6296 }, { "epoch": 8.24, "learning_rate": 1.856e-05, "loss": 0.0085, "step": 6297 }, { "epoch": 8.24, "learning_rate": 1.8555e-05, "loss": 0.0093, "step": 6298 }, { "epoch": 8.24, "learning_rate": 1.855e-05, "loss": 0.0142, "step": 6299 }, { "epoch": 8.25, "learning_rate": 1.8545000000000003e-05, "loss": 0.0109, "step": 6300 }, { "epoch": 8.25, "learning_rate": 1.8540000000000002e-05, "loss": 0.0077, "step": 6301 }, { "epoch": 8.25, "learning_rate": 1.8535e-05, "loss": 0.0079, "step": 6302 }, { "epoch": 8.25, "learning_rate": 1.853e-05, "loss": 0.0129, "step": 6303 }, { "epoch": 8.25, "learning_rate": 1.8525e-05, "loss": 0.0076, "step": 6304 }, { "epoch": 8.25, "learning_rate": 1.8520000000000002e-05, "loss": 0.0086, "step": 6305 }, { "epoch": 8.25, "learning_rate": 1.8515e-05, "loss": 0.0075, "step": 6306 }, { "epoch": 8.26, "learning_rate": 1.851e-05, "loss": 0.0072, "step": 6307 }, { "epoch": 8.26, "learning_rate": 1.8505e-05, "loss": 0.0075, "step": 6308 }, { "epoch": 8.26, "learning_rate": 1.85e-05, "loss": 0.0088, "step": 6309 }, { "epoch": 8.26, "learning_rate": 1.8495e-05, "loss": 0.0088, "step": 6310 }, { "epoch": 8.26, "learning_rate": 1.849e-05, "loss": 0.008, "step": 6311 }, { "epoch": 8.26, "learning_rate": 1.8485e-05, "loss": 0.0078, "step": 6312 }, { "epoch": 8.26, "learning_rate": 1.848e-05, "loss": 0.0076, "step": 6313 }, { "epoch": 8.26, "learning_rate": 1.8475000000000002e-05, "loss": 0.0068, "step": 6314 }, { "epoch": 8.27, "learning_rate": 1.847e-05, "loss": 0.0085, "step": 6315 }, { "epoch": 8.27, "learning_rate": 1.8465e-05, "loss": 0.0081, "step": 6316 }, { "epoch": 8.27, "learning_rate": 1.846e-05, "loss": 0.0074, "step": 6317 }, { "epoch": 8.27, "learning_rate": 1.8455e-05, "loss": 0.0093, "step": 6318 }, { "epoch": 8.27, "learning_rate": 1.845e-05, "loss": 0.0087, "step": 6319 }, { "epoch": 8.27, "learning_rate": 1.8445e-05, "loss": 0.0078, "step": 6320 }, { "epoch": 8.27, "learning_rate": 1.8440000000000003e-05, "loss": 0.0151, "step": 6321 }, { "epoch": 8.27, "learning_rate": 1.8435000000000002e-05, "loss": 0.0108, "step": 6322 }, { "epoch": 8.28, "learning_rate": 1.843e-05, "loss": 0.0083, "step": 6323 }, { "epoch": 8.28, "learning_rate": 1.8425e-05, "loss": 0.0086, "step": 6324 }, { "epoch": 8.28, "learning_rate": 1.842e-05, "loss": 0.0104, "step": 6325 }, { "epoch": 8.28, "learning_rate": 1.8415000000000002e-05, "loss": 0.0087, "step": 6326 }, { "epoch": 8.28, "learning_rate": 1.841e-05, "loss": 0.0068, "step": 6327 }, { "epoch": 8.28, "learning_rate": 1.8405e-05, "loss": 0.0099, "step": 6328 }, { "epoch": 8.28, "learning_rate": 1.84e-05, "loss": 0.0099, "step": 6329 }, { "epoch": 8.29, "learning_rate": 1.8395000000000003e-05, "loss": 0.0103, "step": 6330 }, { "epoch": 8.29, "learning_rate": 1.8390000000000002e-05, "loss": 0.009, "step": 6331 }, { "epoch": 8.29, "learning_rate": 1.8385e-05, "loss": 0.0079, "step": 6332 }, { "epoch": 8.29, "learning_rate": 1.838e-05, "loss": 0.0091, "step": 6333 }, { "epoch": 8.29, "learning_rate": 1.8375e-05, "loss": 0.0101, "step": 6334 }, { "epoch": 8.29, "learning_rate": 1.8370000000000002e-05, "loss": 0.0067, "step": 6335 }, { "epoch": 8.29, "learning_rate": 1.8365e-05, "loss": 0.0071, "step": 6336 }, { "epoch": 8.29, "learning_rate": 1.8360000000000004e-05, "loss": 0.0074, "step": 6337 }, { "epoch": 8.3, "learning_rate": 1.8355e-05, "loss": 0.0076, "step": 6338 }, { "epoch": 8.3, "learning_rate": 1.8350000000000002e-05, "loss": 0.0091, "step": 6339 }, { "epoch": 8.3, "learning_rate": 1.8345e-05, "loss": 0.0074, "step": 6340 }, { "epoch": 8.3, "learning_rate": 1.834e-05, "loss": 0.0078, "step": 6341 }, { "epoch": 8.3, "learning_rate": 1.8335000000000003e-05, "loss": 0.0075, "step": 6342 }, { "epoch": 8.3, "learning_rate": 1.833e-05, "loss": 0.0731, "step": 6343 }, { "epoch": 8.3, "learning_rate": 1.8325e-05, "loss": 0.0107, "step": 6344 }, { "epoch": 8.3, "learning_rate": 1.832e-05, "loss": 0.0079, "step": 6345 }, { "epoch": 8.31, "learning_rate": 1.8315e-05, "loss": 0.0078, "step": 6346 }, { "epoch": 8.31, "learning_rate": 1.8310000000000003e-05, "loss": 0.0098, "step": 6347 }, { "epoch": 8.31, "learning_rate": 1.8305e-05, "loss": 0.0089, "step": 6348 }, { "epoch": 8.31, "learning_rate": 1.83e-05, "loss": 0.0085, "step": 6349 }, { "epoch": 8.31, "learning_rate": 1.8295e-05, "loss": 0.0087, "step": 6350 }, { "epoch": 8.31, "learning_rate": 1.8290000000000003e-05, "loss": 0.0066, "step": 6351 }, { "epoch": 8.31, "learning_rate": 1.8285000000000002e-05, "loss": 0.0098, "step": 6352 }, { "epoch": 8.32, "learning_rate": 1.828e-05, "loss": 0.0105, "step": 6353 }, { "epoch": 8.32, "learning_rate": 1.8275e-05, "loss": 0.0103, "step": 6354 }, { "epoch": 8.32, "learning_rate": 1.827e-05, "loss": 0.0064, "step": 6355 }, { "epoch": 8.32, "learning_rate": 1.8265000000000002e-05, "loss": 0.0068, "step": 6356 }, { "epoch": 8.32, "learning_rate": 1.826e-05, "loss": 0.0069, "step": 6357 }, { "epoch": 8.32, "learning_rate": 1.8255e-05, "loss": 0.0099, "step": 6358 }, { "epoch": 8.32, "learning_rate": 1.825e-05, "loss": 0.0077, "step": 6359 }, { "epoch": 8.32, "learning_rate": 1.8245000000000002e-05, "loss": 0.0087, "step": 6360 }, { "epoch": 8.33, "learning_rate": 1.824e-05, "loss": 0.0083, "step": 6361 }, { "epoch": 8.33, "learning_rate": 1.8235e-05, "loss": 0.0113, "step": 6362 }, { "epoch": 8.33, "learning_rate": 1.823e-05, "loss": 0.009, "step": 6363 }, { "epoch": 8.33, "learning_rate": 1.8225e-05, "loss": 0.0078, "step": 6364 }, { "epoch": 8.33, "learning_rate": 1.8220000000000002e-05, "loss": 0.0076, "step": 6365 }, { "epoch": 8.33, "learning_rate": 1.8215e-05, "loss": 0.0094, "step": 6366 }, { "epoch": 8.33, "learning_rate": 1.8210000000000004e-05, "loss": 0.0102, "step": 6367 }, { "epoch": 8.34, "learning_rate": 1.8205e-05, "loss": 0.0085, "step": 6368 }, { "epoch": 8.34, "learning_rate": 1.8200000000000002e-05, "loss": 0.0076, "step": 6369 }, { "epoch": 8.34, "learning_rate": 1.8195e-05, "loss": 0.01, "step": 6370 }, { "epoch": 8.34, "learning_rate": 1.819e-05, "loss": 0.0112, "step": 6371 }, { "epoch": 8.34, "learning_rate": 1.8185000000000003e-05, "loss": 0.0098, "step": 6372 }, { "epoch": 8.34, "learning_rate": 1.818e-05, "loss": 0.0097, "step": 6373 }, { "epoch": 8.34, "learning_rate": 1.8175e-05, "loss": 0.0103, "step": 6374 }, { "epoch": 8.34, "learning_rate": 1.817e-05, "loss": 0.0115, "step": 6375 }, { "epoch": 8.35, "learning_rate": 1.8165000000000003e-05, "loss": 0.0103, "step": 6376 }, { "epoch": 8.35, "learning_rate": 1.8160000000000002e-05, "loss": 0.0101, "step": 6377 }, { "epoch": 8.35, "learning_rate": 1.8154999999999998e-05, "loss": 0.0067, "step": 6378 }, { "epoch": 8.35, "learning_rate": 1.815e-05, "loss": 0.0085, "step": 6379 }, { "epoch": 8.35, "learning_rate": 1.8145e-05, "loss": 0.0071, "step": 6380 }, { "epoch": 8.35, "learning_rate": 1.8140000000000003e-05, "loss": 0.0086, "step": 6381 }, { "epoch": 8.35, "learning_rate": 1.8135000000000002e-05, "loss": 0.0095, "step": 6382 }, { "epoch": 8.35, "learning_rate": 1.813e-05, "loss": 0.0073, "step": 6383 }, { "epoch": 8.36, "learning_rate": 1.8125e-05, "loss": 0.0092, "step": 6384 }, { "epoch": 8.36, "learning_rate": 1.812e-05, "loss": 0.0088, "step": 6385 }, { "epoch": 8.36, "learning_rate": 1.8115000000000002e-05, "loss": 0.0099, "step": 6386 }, { "epoch": 8.36, "learning_rate": 1.811e-05, "loss": 0.0086, "step": 6387 }, { "epoch": 8.36, "learning_rate": 1.8105e-05, "loss": 0.0079, "step": 6388 }, { "epoch": 8.36, "learning_rate": 1.81e-05, "loss": 0.009, "step": 6389 }, { "epoch": 8.36, "learning_rate": 1.8095000000000002e-05, "loss": 0.0107, "step": 6390 }, { "epoch": 8.37, "learning_rate": 1.809e-05, "loss": 0.011, "step": 6391 }, { "epoch": 8.37, "learning_rate": 1.8085e-05, "loss": 0.0072, "step": 6392 }, { "epoch": 8.37, "learning_rate": 1.808e-05, "loss": 0.0083, "step": 6393 }, { "epoch": 8.37, "learning_rate": 1.8075e-05, "loss": 0.0098, "step": 6394 }, { "epoch": 8.37, "learning_rate": 1.807e-05, "loss": 0.009, "step": 6395 }, { "epoch": 8.37, "learning_rate": 1.8065e-05, "loss": 0.0079, "step": 6396 }, { "epoch": 8.37, "learning_rate": 1.8060000000000003e-05, "loss": 0.0095, "step": 6397 }, { "epoch": 8.37, "learning_rate": 1.8055e-05, "loss": 0.009, "step": 6398 }, { "epoch": 8.38, "learning_rate": 1.805e-05, "loss": 0.0115, "step": 6399 }, { "epoch": 8.38, "learning_rate": 1.8045e-05, "loss": 0.0091, "step": 6400 }, { "epoch": 8.38, "learning_rate": 1.804e-05, "loss": 0.0078, "step": 6401 }, { "epoch": 8.38, "learning_rate": 1.8035000000000003e-05, "loss": 0.0106, "step": 6402 }, { "epoch": 8.38, "learning_rate": 1.803e-05, "loss": 0.0084, "step": 6403 }, { "epoch": 8.38, "learning_rate": 1.8025e-05, "loss": 0.0101, "step": 6404 }, { "epoch": 8.38, "learning_rate": 1.802e-05, "loss": 0.0111, "step": 6405 }, { "epoch": 8.38, "learning_rate": 1.8015000000000003e-05, "loss": 0.0088, "step": 6406 }, { "epoch": 8.39, "learning_rate": 1.8010000000000002e-05, "loss": 0.0114, "step": 6407 }, { "epoch": 8.39, "learning_rate": 1.8005e-05, "loss": 0.0057, "step": 6408 }, { "epoch": 8.39, "learning_rate": 1.8e-05, "loss": 0.011, "step": 6409 }, { "epoch": 8.39, "learning_rate": 1.7995e-05, "loss": 0.0099, "step": 6410 }, { "epoch": 8.39, "learning_rate": 1.7990000000000002e-05, "loss": 0.0069, "step": 6411 }, { "epoch": 8.39, "learning_rate": 1.7985e-05, "loss": 0.0102, "step": 6412 }, { "epoch": 8.39, "learning_rate": 1.798e-05, "loss": 0.0099, "step": 6413 }, { "epoch": 8.4, "learning_rate": 1.7975e-05, "loss": 0.0094, "step": 6414 }, { "epoch": 8.4, "learning_rate": 1.797e-05, "loss": 0.0083, "step": 6415 }, { "epoch": 8.4, "learning_rate": 1.7965e-05, "loss": 0.0102, "step": 6416 }, { "epoch": 8.4, "learning_rate": 1.796e-05, "loss": 0.0072, "step": 6417 }, { "epoch": 8.4, "learning_rate": 1.7955e-05, "loss": 0.0101, "step": 6418 }, { "epoch": 8.4, "learning_rate": 1.795e-05, "loss": 0.009, "step": 6419 }, { "epoch": 8.4, "learning_rate": 1.7945000000000002e-05, "loss": 0.0084, "step": 6420 }, { "epoch": 8.4, "learning_rate": 1.794e-05, "loss": 0.0095, "step": 6421 }, { "epoch": 8.41, "learning_rate": 1.7935e-05, "loss": 0.0099, "step": 6422 }, { "epoch": 8.41, "learning_rate": 1.793e-05, "loss": 0.0106, "step": 6423 }, { "epoch": 8.41, "learning_rate": 1.7925e-05, "loss": 0.0075, "step": 6424 }, { "epoch": 8.41, "learning_rate": 1.792e-05, "loss": 0.0092, "step": 6425 }, { "epoch": 8.41, "learning_rate": 1.7915e-05, "loss": 0.01, "step": 6426 }, { "epoch": 8.41, "learning_rate": 1.7910000000000003e-05, "loss": 0.0093, "step": 6427 }, { "epoch": 8.41, "learning_rate": 1.7905e-05, "loss": 0.0091, "step": 6428 }, { "epoch": 8.41, "learning_rate": 1.79e-05, "loss": 0.0071, "step": 6429 }, { "epoch": 8.42, "learning_rate": 1.7895e-05, "loss": 0.0077, "step": 6430 }, { "epoch": 8.42, "learning_rate": 1.789e-05, "loss": 0.0082, "step": 6431 }, { "epoch": 8.42, "learning_rate": 1.7885000000000002e-05, "loss": 0.0116, "step": 6432 }, { "epoch": 8.42, "learning_rate": 1.7879999999999998e-05, "loss": 0.0099, "step": 6433 }, { "epoch": 8.42, "learning_rate": 1.7875e-05, "loss": 0.0081, "step": 6434 }, { "epoch": 8.42, "learning_rate": 1.787e-05, "loss": 0.0103, "step": 6435 }, { "epoch": 8.42, "learning_rate": 1.7865000000000003e-05, "loss": 0.0085, "step": 6436 }, { "epoch": 8.43, "learning_rate": 1.7860000000000002e-05, "loss": 0.0083, "step": 6437 }, { "epoch": 8.43, "learning_rate": 1.7855e-05, "loss": 0.0054, "step": 6438 }, { "epoch": 8.43, "learning_rate": 1.785e-05, "loss": 0.0089, "step": 6439 }, { "epoch": 8.43, "learning_rate": 1.7845e-05, "loss": 0.0085, "step": 6440 }, { "epoch": 8.43, "learning_rate": 1.7840000000000002e-05, "loss": 0.0098, "step": 6441 }, { "epoch": 8.43, "learning_rate": 1.7835e-05, "loss": 0.0095, "step": 6442 }, { "epoch": 8.43, "learning_rate": 1.783e-05, "loss": 0.0089, "step": 6443 }, { "epoch": 8.43, "learning_rate": 1.7825e-05, "loss": 0.008, "step": 6444 }, { "epoch": 8.44, "learning_rate": 1.7820000000000002e-05, "loss": 0.0091, "step": 6445 }, { "epoch": 8.44, "learning_rate": 1.7815e-05, "loss": 0.0085, "step": 6446 }, { "epoch": 8.44, "learning_rate": 1.781e-05, "loss": 0.0088, "step": 6447 }, { "epoch": 8.44, "learning_rate": 1.7805000000000003e-05, "loss": 0.0083, "step": 6448 }, { "epoch": 8.44, "learning_rate": 1.78e-05, "loss": 0.0076, "step": 6449 }, { "epoch": 8.44, "learning_rate": 1.7795e-05, "loss": 0.0102, "step": 6450 }, { "epoch": 8.44, "learning_rate": 1.779e-05, "loss": 0.0083, "step": 6451 }, { "epoch": 8.45, "learning_rate": 1.7785e-05, "loss": 0.0102, "step": 6452 }, { "epoch": 8.45, "learning_rate": 1.7780000000000003e-05, "loss": 0.0075, "step": 6453 }, { "epoch": 8.45, "learning_rate": 1.7775e-05, "loss": 0.0097, "step": 6454 }, { "epoch": 8.45, "learning_rate": 1.777e-05, "loss": 0.0092, "step": 6455 }, { "epoch": 8.45, "learning_rate": 1.7765e-05, "loss": 0.0081, "step": 6456 }, { "epoch": 8.45, "learning_rate": 1.7760000000000003e-05, "loss": 0.0091, "step": 6457 }, { "epoch": 8.45, "learning_rate": 1.7755000000000002e-05, "loss": 0.0082, "step": 6458 }, { "epoch": 8.45, "learning_rate": 1.775e-05, "loss": 0.0089, "step": 6459 }, { "epoch": 8.46, "learning_rate": 1.7745e-05, "loss": 0.0088, "step": 6460 }, { "epoch": 8.46, "learning_rate": 1.774e-05, "loss": 0.0095, "step": 6461 }, { "epoch": 8.46, "learning_rate": 1.7735000000000002e-05, "loss": 0.0074, "step": 6462 }, { "epoch": 8.46, "learning_rate": 1.773e-05, "loss": 0.009, "step": 6463 }, { "epoch": 8.46, "learning_rate": 1.7725e-05, "loss": 0.0082, "step": 6464 }, { "epoch": 8.46, "learning_rate": 1.772e-05, "loss": 0.0106, "step": 6465 }, { "epoch": 8.46, "learning_rate": 1.7715000000000002e-05, "loss": 0.0076, "step": 6466 }, { "epoch": 8.46, "learning_rate": 1.771e-05, "loss": 0.0091, "step": 6467 }, { "epoch": 8.47, "learning_rate": 1.7705e-05, "loss": 0.0118, "step": 6468 }, { "epoch": 8.47, "learning_rate": 1.77e-05, "loss": 0.009, "step": 6469 }, { "epoch": 8.47, "learning_rate": 1.7695e-05, "loss": 0.0086, "step": 6470 }, { "epoch": 8.47, "learning_rate": 1.7690000000000002e-05, "loss": 0.0082, "step": 6471 }, { "epoch": 8.47, "learning_rate": 1.7685e-05, "loss": 0.0123, "step": 6472 }, { "epoch": 8.47, "learning_rate": 1.7680000000000004e-05, "loss": 0.0094, "step": 6473 }, { "epoch": 8.47, "learning_rate": 1.7675e-05, "loss": 0.0087, "step": 6474 }, { "epoch": 8.48, "learning_rate": 1.7670000000000002e-05, "loss": 0.0077, "step": 6475 }, { "epoch": 8.48, "learning_rate": 1.7665e-05, "loss": 0.011, "step": 6476 }, { "epoch": 8.48, "learning_rate": 1.766e-05, "loss": 0.0089, "step": 6477 }, { "epoch": 8.48, "learning_rate": 1.7655000000000003e-05, "loss": 0.0073, "step": 6478 }, { "epoch": 8.48, "learning_rate": 1.765e-05, "loss": 0.0088, "step": 6479 }, { "epoch": 8.48, "learning_rate": 1.7645e-05, "loss": 0.0048, "step": 6480 }, { "epoch": 8.48, "learning_rate": 1.764e-05, "loss": 0.0063, "step": 6481 }, { "epoch": 8.48, "learning_rate": 1.7635000000000003e-05, "loss": 0.0076, "step": 6482 }, { "epoch": 8.49, "learning_rate": 1.7630000000000002e-05, "loss": 0.0074, "step": 6483 }, { "epoch": 8.49, "learning_rate": 1.7625e-05, "loss": 0.0103, "step": 6484 }, { "epoch": 8.49, "learning_rate": 1.762e-05, "loss": 0.0066, "step": 6485 }, { "epoch": 8.49, "learning_rate": 1.7615e-05, "loss": 0.0089, "step": 6486 }, { "epoch": 8.49, "learning_rate": 1.7610000000000002e-05, "loss": 0.0098, "step": 6487 }, { "epoch": 8.49, "learning_rate": 1.7605000000000002e-05, "loss": 0.0089, "step": 6488 }, { "epoch": 8.49, "learning_rate": 1.76e-05, "loss": 0.0091, "step": 6489 }, { "epoch": 8.49, "learning_rate": 1.7595e-05, "loss": 0.0084, "step": 6490 }, { "epoch": 8.5, "learning_rate": 1.759e-05, "loss": 0.0093, "step": 6491 }, { "epoch": 8.5, "learning_rate": 1.7585000000000002e-05, "loss": 0.0113, "step": 6492 }, { "epoch": 8.5, "learning_rate": 1.758e-05, "loss": 0.0081, "step": 6493 }, { "epoch": 8.5, "learning_rate": 1.7575e-05, "loss": 0.009, "step": 6494 }, { "epoch": 8.5, "learning_rate": 1.757e-05, "loss": 0.0127, "step": 6495 }, { "epoch": 8.5, "learning_rate": 1.7565000000000002e-05, "loss": 0.0091, "step": 6496 }, { "epoch": 8.5, "learning_rate": 1.756e-05, "loss": 0.009, "step": 6497 }, { "epoch": 8.51, "learning_rate": 1.7555e-05, "loss": 0.0088, "step": 6498 }, { "epoch": 8.51, "learning_rate": 1.755e-05, "loss": 0.0061, "step": 6499 }, { "epoch": 8.51, "learning_rate": 1.7545e-05, "loss": 0.0088, "step": 6500 }, { "epoch": 8.51, "learning_rate": 1.754e-05, "loss": 0.0083, "step": 6501 }, { "epoch": 8.51, "learning_rate": 1.7535e-05, "loss": 0.0096, "step": 6502 }, { "epoch": 8.51, "learning_rate": 1.7530000000000003e-05, "loss": 0.0096, "step": 6503 }, { "epoch": 8.51, "learning_rate": 1.7525e-05, "loss": 0.0073, "step": 6504 }, { "epoch": 8.51, "learning_rate": 1.752e-05, "loss": 0.0083, "step": 6505 }, { "epoch": 8.52, "learning_rate": 1.7515e-05, "loss": 0.011, "step": 6506 }, { "epoch": 8.52, "learning_rate": 1.751e-05, "loss": 0.01, "step": 6507 }, { "epoch": 8.52, "learning_rate": 1.7505000000000003e-05, "loss": 0.0781, "step": 6508 }, { "epoch": 8.52, "learning_rate": 1.75e-05, "loss": 0.0086, "step": 6509 }, { "epoch": 8.52, "learning_rate": 1.7495e-05, "loss": 0.0081, "step": 6510 }, { "epoch": 8.52, "learning_rate": 1.749e-05, "loss": 0.0117, "step": 6511 }, { "epoch": 8.52, "learning_rate": 1.7485000000000003e-05, "loss": 0.0068, "step": 6512 }, { "epoch": 8.52, "learning_rate": 1.7480000000000002e-05, "loss": 0.0111, "step": 6513 }, { "epoch": 8.53, "learning_rate": 1.7475e-05, "loss": 0.0114, "step": 6514 }, { "epoch": 8.53, "learning_rate": 1.747e-05, "loss": 0.012, "step": 6515 }, { "epoch": 8.53, "learning_rate": 1.7465e-05, "loss": 0.0095, "step": 6516 }, { "epoch": 8.53, "learning_rate": 1.7460000000000002e-05, "loss": 0.0224, "step": 6517 }, { "epoch": 8.53, "learning_rate": 1.7455e-05, "loss": 0.0091, "step": 6518 }, { "epoch": 8.53, "learning_rate": 1.745e-05, "loss": 0.0115, "step": 6519 }, { "epoch": 8.53, "learning_rate": 1.7445e-05, "loss": 0.0123, "step": 6520 }, { "epoch": 8.54, "learning_rate": 1.7440000000000002e-05, "loss": 0.0109, "step": 6521 }, { "epoch": 8.54, "learning_rate": 1.7435e-05, "loss": 0.0095, "step": 6522 }, { "epoch": 8.54, "learning_rate": 1.743e-05, "loss": 0.0125, "step": 6523 }, { "epoch": 8.54, "learning_rate": 1.7425e-05, "loss": 0.0097, "step": 6524 }, { "epoch": 8.54, "learning_rate": 1.742e-05, "loss": 0.0086, "step": 6525 }, { "epoch": 8.54, "learning_rate": 1.7415000000000002e-05, "loss": 0.0124, "step": 6526 }, { "epoch": 8.54, "learning_rate": 1.741e-05, "loss": 0.0137, "step": 6527 }, { "epoch": 8.54, "learning_rate": 1.7405e-05, "loss": 0.0082, "step": 6528 }, { "epoch": 8.55, "learning_rate": 1.74e-05, "loss": 0.0086, "step": 6529 }, { "epoch": 8.55, "learning_rate": 1.7395e-05, "loss": 0.0099, "step": 6530 }, { "epoch": 8.55, "learning_rate": 1.739e-05, "loss": 0.0108, "step": 6531 }, { "epoch": 8.55, "learning_rate": 1.7385e-05, "loss": 0.0125, "step": 6532 }, { "epoch": 8.55, "learning_rate": 1.7380000000000003e-05, "loss": 0.0078, "step": 6533 }, { "epoch": 8.55, "learning_rate": 1.7375e-05, "loss": 0.0082, "step": 6534 }, { "epoch": 8.55, "learning_rate": 1.737e-05, "loss": 0.0104, "step": 6535 }, { "epoch": 8.55, "learning_rate": 1.7365e-05, "loss": 0.0099, "step": 6536 }, { "epoch": 8.56, "learning_rate": 1.736e-05, "loss": 0.0095, "step": 6537 }, { "epoch": 8.56, "learning_rate": 1.7355000000000002e-05, "loss": 0.011, "step": 6538 }, { "epoch": 8.56, "learning_rate": 1.7349999999999998e-05, "loss": 0.0116, "step": 6539 }, { "epoch": 8.56, "learning_rate": 1.7345e-05, "loss": 0.0114, "step": 6540 }, { "epoch": 8.56, "learning_rate": 1.734e-05, "loss": 0.0078, "step": 6541 }, { "epoch": 8.56, "learning_rate": 1.7335000000000003e-05, "loss": 0.0091, "step": 6542 }, { "epoch": 8.56, "learning_rate": 1.7330000000000002e-05, "loss": 0.012, "step": 6543 }, { "epoch": 8.57, "learning_rate": 1.7325e-05, "loss": 0.0087, "step": 6544 }, { "epoch": 8.57, "learning_rate": 1.732e-05, "loss": 0.0101, "step": 6545 }, { "epoch": 8.57, "learning_rate": 1.7315e-05, "loss": 0.0089, "step": 6546 }, { "epoch": 8.57, "learning_rate": 1.7310000000000002e-05, "loss": 0.011, "step": 6547 }, { "epoch": 8.57, "learning_rate": 1.7305e-05, "loss": 0.0102, "step": 6548 }, { "epoch": 8.57, "learning_rate": 1.73e-05, "loss": 0.0098, "step": 6549 }, { "epoch": 8.57, "learning_rate": 1.7295e-05, "loss": 0.0093, "step": 6550 }, { "epoch": 8.57, "learning_rate": 1.7290000000000002e-05, "loss": 0.0104, "step": 6551 }, { "epoch": 8.58, "learning_rate": 1.7285e-05, "loss": 0.0096, "step": 6552 }, { "epoch": 8.58, "learning_rate": 1.728e-05, "loss": 0.0092, "step": 6553 }, { "epoch": 8.58, "learning_rate": 1.7275e-05, "loss": 0.0094, "step": 6554 }, { "epoch": 8.58, "learning_rate": 1.727e-05, "loss": 0.0095, "step": 6555 }, { "epoch": 8.58, "learning_rate": 1.7265e-05, "loss": 0.0086, "step": 6556 }, { "epoch": 8.58, "learning_rate": 1.726e-05, "loss": 0.0111, "step": 6557 }, { "epoch": 8.58, "learning_rate": 1.7255000000000003e-05, "loss": 0.0115, "step": 6558 }, { "epoch": 8.59, "learning_rate": 1.725e-05, "loss": 0.0083, "step": 6559 }, { "epoch": 8.59, "learning_rate": 1.7245e-05, "loss": 0.0077, "step": 6560 }, { "epoch": 8.59, "learning_rate": 1.724e-05, "loss": 0.0094, "step": 6561 }, { "epoch": 8.59, "learning_rate": 1.7235e-05, "loss": 0.0098, "step": 6562 }, { "epoch": 8.59, "learning_rate": 1.7230000000000003e-05, "loss": 0.0111, "step": 6563 }, { "epoch": 8.59, "learning_rate": 1.7225e-05, "loss": 0.0099, "step": 6564 }, { "epoch": 8.59, "learning_rate": 1.722e-05, "loss": 0.0099, "step": 6565 }, { "epoch": 8.59, "learning_rate": 1.7215e-05, "loss": 0.0086, "step": 6566 }, { "epoch": 8.6, "learning_rate": 1.721e-05, "loss": 0.0097, "step": 6567 }, { "epoch": 8.6, "learning_rate": 1.7205000000000002e-05, "loss": 0.0095, "step": 6568 }, { "epoch": 8.6, "learning_rate": 1.7199999999999998e-05, "loss": 0.0077, "step": 6569 }, { "epoch": 8.6, "learning_rate": 1.7195e-05, "loss": 0.0099, "step": 6570 }, { "epoch": 8.6, "learning_rate": 1.719e-05, "loss": 0.0096, "step": 6571 }, { "epoch": 8.6, "learning_rate": 1.7185000000000002e-05, "loss": 0.0081, "step": 6572 }, { "epoch": 8.6, "learning_rate": 1.718e-05, "loss": 0.0101, "step": 6573 }, { "epoch": 8.6, "learning_rate": 1.7175e-05, "loss": 0.0084, "step": 6574 }, { "epoch": 8.61, "learning_rate": 1.717e-05, "loss": 0.0094, "step": 6575 }, { "epoch": 8.61, "learning_rate": 1.7165e-05, "loss": 0.008, "step": 6576 }, { "epoch": 8.61, "learning_rate": 1.7160000000000002e-05, "loss": 0.0087, "step": 6577 }, { "epoch": 8.61, "learning_rate": 1.7155e-05, "loss": 0.009, "step": 6578 }, { "epoch": 8.61, "learning_rate": 1.7150000000000004e-05, "loss": 0.01, "step": 6579 }, { "epoch": 8.61, "learning_rate": 1.7145e-05, "loss": 0.0083, "step": 6580 }, { "epoch": 8.61, "learning_rate": 1.7140000000000002e-05, "loss": 0.0074, "step": 6581 }, { "epoch": 8.62, "learning_rate": 1.7135e-05, "loss": 0.0078, "step": 6582 }, { "epoch": 8.62, "learning_rate": 1.713e-05, "loss": 0.0126, "step": 6583 }, { "epoch": 8.62, "learning_rate": 1.7125000000000003e-05, "loss": 0.0104, "step": 6584 }, { "epoch": 8.62, "learning_rate": 1.712e-05, "loss": 0.0094, "step": 6585 }, { "epoch": 8.62, "learning_rate": 1.7115e-05, "loss": 0.0113, "step": 6586 }, { "epoch": 8.62, "learning_rate": 1.711e-05, "loss": 0.0066, "step": 6587 }, { "epoch": 8.62, "learning_rate": 1.7105000000000003e-05, "loss": 0.0079, "step": 6588 }, { "epoch": 8.62, "learning_rate": 1.7100000000000002e-05, "loss": 0.0099, "step": 6589 }, { "epoch": 8.63, "learning_rate": 1.7095e-05, "loss": 0.0089, "step": 6590 }, { "epoch": 8.63, "learning_rate": 1.709e-05, "loss": 0.007, "step": 6591 }, { "epoch": 8.63, "learning_rate": 1.7085e-05, "loss": 0.0112, "step": 6592 }, { "epoch": 8.63, "learning_rate": 1.7080000000000002e-05, "loss": 0.0071, "step": 6593 }, { "epoch": 8.63, "learning_rate": 1.7075e-05, "loss": 0.0092, "step": 6594 }, { "epoch": 8.63, "learning_rate": 1.707e-05, "loss": 0.0105, "step": 6595 }, { "epoch": 8.63, "learning_rate": 1.7065e-05, "loss": 0.0078, "step": 6596 }, { "epoch": 8.63, "learning_rate": 1.706e-05, "loss": 0.0087, "step": 6597 }, { "epoch": 8.64, "learning_rate": 1.7055000000000002e-05, "loss": 0.0089, "step": 6598 }, { "epoch": 8.64, "learning_rate": 1.705e-05, "loss": 0.0092, "step": 6599 }, { "epoch": 8.64, "learning_rate": 1.7045e-05, "loss": 0.0086, "step": 6600 }, { "epoch": 8.64, "learning_rate": 1.704e-05, "loss": 0.0077, "step": 6601 }, { "epoch": 8.64, "learning_rate": 1.7035000000000002e-05, "loss": 0.0097, "step": 6602 }, { "epoch": 8.64, "learning_rate": 1.703e-05, "loss": 0.0081, "step": 6603 }, { "epoch": 8.64, "learning_rate": 1.7025e-05, "loss": 0.0104, "step": 6604 }, { "epoch": 8.65, "learning_rate": 1.702e-05, "loss": 0.0087, "step": 6605 }, { "epoch": 8.65, "learning_rate": 1.7015e-05, "loss": 0.0115, "step": 6606 }, { "epoch": 8.65, "learning_rate": 1.701e-05, "loss": 0.0129, "step": 6607 }, { "epoch": 8.65, "learning_rate": 1.7005e-05, "loss": 0.0076, "step": 6608 }, { "epoch": 8.65, "learning_rate": 1.7000000000000003e-05, "loss": 0.0101, "step": 6609 }, { "epoch": 8.65, "learning_rate": 1.6995e-05, "loss": 0.0089, "step": 6610 }, { "epoch": 8.65, "learning_rate": 1.699e-05, "loss": 0.0091, "step": 6611 }, { "epoch": 8.65, "learning_rate": 1.6985e-05, "loss": 0.0095, "step": 6612 }, { "epoch": 8.66, "learning_rate": 1.698e-05, "loss": 0.01, "step": 6613 }, { "epoch": 8.66, "learning_rate": 1.6975000000000003e-05, "loss": 0.0083, "step": 6614 }, { "epoch": 8.66, "learning_rate": 1.697e-05, "loss": 0.0101, "step": 6615 }, { "epoch": 8.66, "learning_rate": 1.6965e-05, "loss": 0.0096, "step": 6616 }, { "epoch": 8.66, "learning_rate": 1.696e-05, "loss": 0.0076, "step": 6617 }, { "epoch": 8.66, "learning_rate": 1.6955000000000003e-05, "loss": 0.0093, "step": 6618 }, { "epoch": 8.66, "learning_rate": 1.6950000000000002e-05, "loss": 0.0098, "step": 6619 }, { "epoch": 8.66, "learning_rate": 1.6945e-05, "loss": 0.0083, "step": 6620 }, { "epoch": 8.67, "learning_rate": 1.694e-05, "loss": 0.0094, "step": 6621 }, { "epoch": 8.67, "learning_rate": 1.6935e-05, "loss": 0.0092, "step": 6622 }, { "epoch": 8.67, "learning_rate": 1.6930000000000002e-05, "loss": 0.0097, "step": 6623 }, { "epoch": 8.67, "learning_rate": 1.6925e-05, "loss": 0.0075, "step": 6624 }, { "epoch": 8.67, "learning_rate": 1.692e-05, "loss": 0.0125, "step": 6625 }, { "epoch": 8.67, "learning_rate": 1.6915e-05, "loss": 0.0093, "step": 6626 }, { "epoch": 8.67, "learning_rate": 1.6910000000000002e-05, "loss": 0.0119, "step": 6627 }, { "epoch": 8.68, "learning_rate": 1.6905e-05, "loss": 0.0093, "step": 6628 }, { "epoch": 8.68, "learning_rate": 1.69e-05, "loss": 0.0081, "step": 6629 }, { "epoch": 8.68, "learning_rate": 1.6895e-05, "loss": 0.0084, "step": 6630 }, { "epoch": 8.68, "learning_rate": 1.689e-05, "loss": 0.0086, "step": 6631 }, { "epoch": 8.68, "learning_rate": 1.6885000000000002e-05, "loss": 0.0106, "step": 6632 }, { "epoch": 8.68, "learning_rate": 1.688e-05, "loss": 0.0093, "step": 6633 }, { "epoch": 8.68, "learning_rate": 1.6875000000000004e-05, "loss": 0.0078, "step": 6634 }, { "epoch": 8.68, "learning_rate": 1.687e-05, "loss": 0.0103, "step": 6635 }, { "epoch": 8.69, "learning_rate": 1.6865e-05, "loss": 0.0073, "step": 6636 }, { "epoch": 8.69, "learning_rate": 1.686e-05, "loss": 0.0126, "step": 6637 }, { "epoch": 8.69, "learning_rate": 1.6855e-05, "loss": 0.0109, "step": 6638 }, { "epoch": 8.69, "learning_rate": 1.6850000000000003e-05, "loss": 0.0084, "step": 6639 }, { "epoch": 8.69, "learning_rate": 1.6845e-05, "loss": 0.0093, "step": 6640 }, { "epoch": 8.69, "learning_rate": 1.684e-05, "loss": 0.0097, "step": 6641 }, { "epoch": 8.69, "learning_rate": 1.6835e-05, "loss": 0.0071, "step": 6642 }, { "epoch": 8.7, "learning_rate": 1.683e-05, "loss": 0.0078, "step": 6643 }, { "epoch": 8.7, "learning_rate": 1.6825000000000002e-05, "loss": 0.0087, "step": 6644 }, { "epoch": 8.7, "learning_rate": 1.6819999999999998e-05, "loss": 0.0102, "step": 6645 }, { "epoch": 8.7, "learning_rate": 1.6815e-05, "loss": 0.0109, "step": 6646 }, { "epoch": 8.7, "learning_rate": 1.681e-05, "loss": 0.0088, "step": 6647 }, { "epoch": 8.7, "learning_rate": 1.6805000000000003e-05, "loss": 0.0166, "step": 6648 }, { "epoch": 8.7, "learning_rate": 1.6800000000000002e-05, "loss": 0.01, "step": 6649 }, { "epoch": 8.7, "learning_rate": 1.6795e-05, "loss": 0.0074, "step": 6650 }, { "epoch": 8.71, "learning_rate": 1.679e-05, "loss": 0.0096, "step": 6651 }, { "epoch": 8.71, "learning_rate": 1.6785e-05, "loss": 0.0073, "step": 6652 }, { "epoch": 8.71, "learning_rate": 1.6780000000000002e-05, "loss": 0.009, "step": 6653 }, { "epoch": 8.71, "learning_rate": 1.6775e-05, "loss": 0.0083, "step": 6654 }, { "epoch": 8.71, "learning_rate": 1.677e-05, "loss": 0.01, "step": 6655 }, { "epoch": 8.71, "learning_rate": 1.6765e-05, "loss": 0.0094, "step": 6656 }, { "epoch": 8.71, "learning_rate": 1.6760000000000002e-05, "loss": 0.0083, "step": 6657 }, { "epoch": 8.71, "learning_rate": 1.6755e-05, "loss": 0.0073, "step": 6658 }, { "epoch": 8.72, "learning_rate": 1.675e-05, "loss": 0.0084, "step": 6659 }, { "epoch": 8.72, "learning_rate": 1.6745e-05, "loss": 0.0072, "step": 6660 }, { "epoch": 8.72, "learning_rate": 1.674e-05, "loss": 0.0091, "step": 6661 }, { "epoch": 8.72, "learning_rate": 1.6735e-05, "loss": 0.008, "step": 6662 }, { "epoch": 8.72, "learning_rate": 1.673e-05, "loss": 0.012, "step": 6663 }, { "epoch": 8.72, "learning_rate": 1.6725000000000003e-05, "loss": 0.0092, "step": 6664 }, { "epoch": 8.72, "learning_rate": 1.672e-05, "loss": 0.0115, "step": 6665 }, { "epoch": 8.73, "learning_rate": 1.6715000000000002e-05, "loss": 0.0091, "step": 6666 }, { "epoch": 8.73, "learning_rate": 1.671e-05, "loss": 0.0094, "step": 6667 }, { "epoch": 8.73, "learning_rate": 1.6705e-05, "loss": 0.0079, "step": 6668 }, { "epoch": 8.73, "learning_rate": 1.6700000000000003e-05, "loss": 0.014, "step": 6669 }, { "epoch": 8.73, "learning_rate": 1.6695e-05, "loss": 0.0083, "step": 6670 }, { "epoch": 8.73, "learning_rate": 1.669e-05, "loss": 0.0094, "step": 6671 }, { "epoch": 8.73, "learning_rate": 1.6685e-05, "loss": 0.0131, "step": 6672 }, { "epoch": 8.73, "learning_rate": 1.668e-05, "loss": 0.0094, "step": 6673 }, { "epoch": 8.74, "learning_rate": 1.6675000000000002e-05, "loss": 0.0102, "step": 6674 }, { "epoch": 8.74, "learning_rate": 1.6669999999999998e-05, "loss": 0.009, "step": 6675 }, { "epoch": 8.74, "learning_rate": 1.6665e-05, "loss": 0.0092, "step": 6676 }, { "epoch": 8.74, "learning_rate": 1.666e-05, "loss": 0.0093, "step": 6677 }, { "epoch": 8.74, "learning_rate": 1.6655000000000002e-05, "loss": 0.0093, "step": 6678 }, { "epoch": 8.74, "learning_rate": 1.665e-05, "loss": 0.0092, "step": 6679 }, { "epoch": 8.74, "learning_rate": 1.6645e-05, "loss": 0.0099, "step": 6680 }, { "epoch": 8.74, "learning_rate": 1.664e-05, "loss": 0.0083, "step": 6681 }, { "epoch": 8.75, "learning_rate": 1.6635e-05, "loss": 0.0098, "step": 6682 }, { "epoch": 8.75, "learning_rate": 1.6630000000000002e-05, "loss": 0.0082, "step": 6683 }, { "epoch": 8.75, "learning_rate": 1.6625e-05, "loss": 0.0069, "step": 6684 }, { "epoch": 8.75, "learning_rate": 1.662e-05, "loss": 0.0073, "step": 6685 }, { "epoch": 8.75, "learning_rate": 1.6615e-05, "loss": 0.0079, "step": 6686 }, { "epoch": 8.75, "learning_rate": 1.6610000000000002e-05, "loss": 0.0083, "step": 6687 }, { "epoch": 8.75, "learning_rate": 1.6605e-05, "loss": 0.0079, "step": 6688 }, { "epoch": 8.76, "learning_rate": 1.66e-05, "loss": 0.0116, "step": 6689 }, { "epoch": 8.76, "learning_rate": 1.6595e-05, "loss": 0.0099, "step": 6690 }, { "epoch": 8.76, "learning_rate": 1.659e-05, "loss": 0.0067, "step": 6691 }, { "epoch": 8.76, "learning_rate": 1.6585e-05, "loss": 0.0095, "step": 6692 }, { "epoch": 8.76, "learning_rate": 1.658e-05, "loss": 0.0078, "step": 6693 }, { "epoch": 8.76, "learning_rate": 1.6575000000000003e-05, "loss": 0.0094, "step": 6694 }, { "epoch": 8.76, "learning_rate": 1.657e-05, "loss": 0.0098, "step": 6695 }, { "epoch": 8.76, "learning_rate": 1.6565e-05, "loss": 0.0103, "step": 6696 }, { "epoch": 8.77, "learning_rate": 1.656e-05, "loss": 0.0106, "step": 6697 }, { "epoch": 8.77, "learning_rate": 1.6555e-05, "loss": 0.0094, "step": 6698 }, { "epoch": 8.77, "learning_rate": 1.6550000000000002e-05, "loss": 0.0205, "step": 6699 }, { "epoch": 8.77, "learning_rate": 1.6545e-05, "loss": 0.0138, "step": 6700 }, { "epoch": 8.77, "learning_rate": 1.654e-05, "loss": 0.0083, "step": 6701 }, { "epoch": 8.77, "learning_rate": 1.6535e-05, "loss": 0.013, "step": 6702 }, { "epoch": 8.77, "learning_rate": 1.6530000000000003e-05, "loss": 0.0093, "step": 6703 }, { "epoch": 8.77, "learning_rate": 1.6525000000000002e-05, "loss": 0.0083, "step": 6704 }, { "epoch": 8.78, "learning_rate": 1.652e-05, "loss": 0.0083, "step": 6705 }, { "epoch": 8.78, "learning_rate": 1.6515e-05, "loss": 0.0098, "step": 6706 }, { "epoch": 8.78, "learning_rate": 1.651e-05, "loss": 0.0098, "step": 6707 }, { "epoch": 8.78, "learning_rate": 1.6505000000000002e-05, "loss": 0.0073, "step": 6708 }, { "epoch": 8.78, "learning_rate": 1.65e-05, "loss": 0.0101, "step": 6709 }, { "epoch": 8.78, "learning_rate": 1.6495e-05, "loss": 0.0095, "step": 6710 }, { "epoch": 8.78, "learning_rate": 1.649e-05, "loss": 0.0101, "step": 6711 }, { "epoch": 8.79, "learning_rate": 1.6485e-05, "loss": 0.0093, "step": 6712 }, { "epoch": 8.79, "learning_rate": 1.648e-05, "loss": 0.0104, "step": 6713 }, { "epoch": 8.79, "learning_rate": 1.6475e-05, "loss": 0.0072, "step": 6714 }, { "epoch": 8.79, "learning_rate": 1.6470000000000003e-05, "loss": 0.0119, "step": 6715 }, { "epoch": 8.79, "learning_rate": 1.6465e-05, "loss": 0.0084, "step": 6716 }, { "epoch": 8.79, "learning_rate": 1.646e-05, "loss": 0.0087, "step": 6717 }, { "epoch": 8.79, "learning_rate": 1.6455e-05, "loss": 0.0083, "step": 6718 }, { "epoch": 8.79, "learning_rate": 1.645e-05, "loss": 0.0165, "step": 6719 }, { "epoch": 8.8, "learning_rate": 1.6445000000000003e-05, "loss": 0.0091, "step": 6720 }, { "epoch": 8.8, "learning_rate": 1.644e-05, "loss": 0.0089, "step": 6721 }, { "epoch": 8.8, "learning_rate": 1.6435e-05, "loss": 0.01, "step": 6722 }, { "epoch": 8.8, "learning_rate": 1.643e-05, "loss": 0.0084, "step": 6723 }, { "epoch": 8.8, "learning_rate": 1.6425000000000003e-05, "loss": 0.0079, "step": 6724 }, { "epoch": 8.8, "learning_rate": 1.6420000000000002e-05, "loss": 0.0105, "step": 6725 }, { "epoch": 8.8, "learning_rate": 1.6415e-05, "loss": 0.0092, "step": 6726 }, { "epoch": 8.8, "learning_rate": 1.641e-05, "loss": 0.0095, "step": 6727 }, { "epoch": 8.81, "learning_rate": 1.6405e-05, "loss": 0.0084, "step": 6728 }, { "epoch": 8.81, "learning_rate": 1.6400000000000002e-05, "loss": 0.0104, "step": 6729 }, { "epoch": 8.81, "learning_rate": 1.6395e-05, "loss": 0.008, "step": 6730 }, { "epoch": 8.81, "learning_rate": 1.639e-05, "loss": 0.0067, "step": 6731 }, { "epoch": 8.81, "learning_rate": 1.6385e-05, "loss": 0.0072, "step": 6732 }, { "epoch": 8.81, "learning_rate": 1.6380000000000002e-05, "loss": 0.0076, "step": 6733 }, { "epoch": 8.81, "learning_rate": 1.6375e-05, "loss": 0.0105, "step": 6734 }, { "epoch": 8.82, "learning_rate": 1.637e-05, "loss": 0.0095, "step": 6735 }, { "epoch": 8.82, "learning_rate": 1.6365e-05, "loss": 0.008, "step": 6736 }, { "epoch": 8.82, "learning_rate": 1.636e-05, "loss": 0.0084, "step": 6737 }, { "epoch": 8.82, "learning_rate": 1.6355000000000002e-05, "loss": 0.0077, "step": 6738 }, { "epoch": 8.82, "learning_rate": 1.635e-05, "loss": 0.0087, "step": 6739 }, { "epoch": 8.82, "learning_rate": 1.6345000000000004e-05, "loss": 0.0091, "step": 6740 }, { "epoch": 8.82, "learning_rate": 1.634e-05, "loss": 0.0081, "step": 6741 }, { "epoch": 8.82, "learning_rate": 1.6335e-05, "loss": 0.0068, "step": 6742 }, { "epoch": 8.83, "learning_rate": 1.633e-05, "loss": 0.0094, "step": 6743 }, { "epoch": 8.83, "learning_rate": 1.6325e-05, "loss": 0.0111, "step": 6744 }, { "epoch": 8.83, "learning_rate": 1.6320000000000003e-05, "loss": 0.011, "step": 6745 }, { "epoch": 8.83, "learning_rate": 1.6315e-05, "loss": 0.013, "step": 6746 }, { "epoch": 8.83, "learning_rate": 1.631e-05, "loss": 0.0079, "step": 6747 }, { "epoch": 8.83, "learning_rate": 1.6305e-05, "loss": 0.0093, "step": 6748 }, { "epoch": 8.83, "learning_rate": 1.63e-05, "loss": 0.0098, "step": 6749 }, { "epoch": 8.84, "learning_rate": 1.6295000000000002e-05, "loss": 0.01, "step": 6750 }, { "epoch": 8.84, "learning_rate": 1.6289999999999998e-05, "loss": 0.0105, "step": 6751 }, { "epoch": 8.84, "learning_rate": 1.6285e-05, "loss": 0.0085, "step": 6752 }, { "epoch": 8.84, "learning_rate": 1.628e-05, "loss": 0.0091, "step": 6753 }, { "epoch": 8.84, "learning_rate": 1.6275000000000003e-05, "loss": 0.0117, "step": 6754 }, { "epoch": 8.84, "learning_rate": 1.6270000000000002e-05, "loss": 0.0072, "step": 6755 }, { "epoch": 8.84, "learning_rate": 1.6265e-05, "loss": 0.0072, "step": 6756 }, { "epoch": 8.84, "learning_rate": 1.626e-05, "loss": 0.0122, "step": 6757 }, { "epoch": 8.85, "learning_rate": 1.6255e-05, "loss": 0.0096, "step": 6758 }, { "epoch": 8.85, "learning_rate": 1.6250000000000002e-05, "loss": 0.0078, "step": 6759 }, { "epoch": 8.85, "learning_rate": 1.6245e-05, "loss": 0.0081, "step": 6760 }, { "epoch": 8.85, "learning_rate": 1.624e-05, "loss": 0.0106, "step": 6761 }, { "epoch": 8.85, "learning_rate": 1.6235e-05, "loss": 0.0088, "step": 6762 }, { "epoch": 8.85, "learning_rate": 1.6230000000000002e-05, "loss": 0.0093, "step": 6763 }, { "epoch": 8.85, "learning_rate": 1.6225e-05, "loss": 0.0113, "step": 6764 }, { "epoch": 8.85, "learning_rate": 1.622e-05, "loss": 0.006, "step": 6765 }, { "epoch": 8.86, "learning_rate": 1.6215e-05, "loss": 0.0099, "step": 6766 }, { "epoch": 8.86, "learning_rate": 1.621e-05, "loss": 0.0084, "step": 6767 }, { "epoch": 8.86, "learning_rate": 1.6205e-05, "loss": 0.0065, "step": 6768 }, { "epoch": 8.86, "learning_rate": 1.62e-05, "loss": 0.0095, "step": 6769 }, { "epoch": 8.86, "learning_rate": 1.6195000000000003e-05, "loss": 0.0068, "step": 6770 }, { "epoch": 8.86, "learning_rate": 1.619e-05, "loss": 0.0088, "step": 6771 }, { "epoch": 8.86, "learning_rate": 1.6185000000000002e-05, "loss": 0.0089, "step": 6772 }, { "epoch": 8.87, "learning_rate": 1.618e-05, "loss": 0.0129, "step": 6773 }, { "epoch": 8.87, "learning_rate": 1.6175e-05, "loss": 0.0077, "step": 6774 }, { "epoch": 8.87, "learning_rate": 1.6170000000000003e-05, "loss": 0.0076, "step": 6775 }, { "epoch": 8.87, "learning_rate": 1.6165e-05, "loss": 0.0094, "step": 6776 }, { "epoch": 8.87, "learning_rate": 1.616e-05, "loss": 0.0099, "step": 6777 }, { "epoch": 8.87, "learning_rate": 1.6155e-05, "loss": 0.0098, "step": 6778 }, { "epoch": 8.87, "learning_rate": 1.6150000000000003e-05, "loss": 0.0075, "step": 6779 }, { "epoch": 8.87, "learning_rate": 1.6145000000000002e-05, "loss": 0.0079, "step": 6780 }, { "epoch": 8.88, "learning_rate": 1.6139999999999998e-05, "loss": 0.0081, "step": 6781 }, { "epoch": 8.88, "learning_rate": 1.6135e-05, "loss": 0.0099, "step": 6782 }, { "epoch": 8.88, "learning_rate": 1.613e-05, "loss": 0.0074, "step": 6783 }, { "epoch": 8.88, "learning_rate": 1.6125000000000002e-05, "loss": 0.0093, "step": 6784 }, { "epoch": 8.88, "learning_rate": 1.612e-05, "loss": 0.0087, "step": 6785 }, { "epoch": 8.88, "learning_rate": 1.6115e-05, "loss": 0.0066, "step": 6786 }, { "epoch": 8.88, "learning_rate": 1.611e-05, "loss": 0.0078, "step": 6787 }, { "epoch": 8.88, "learning_rate": 1.6105e-05, "loss": 0.0077, "step": 6788 }, { "epoch": 8.89, "learning_rate": 1.6100000000000002e-05, "loss": 0.0098, "step": 6789 }, { "epoch": 8.89, "learning_rate": 1.6095e-05, "loss": 0.0115, "step": 6790 }, { "epoch": 8.89, "learning_rate": 1.609e-05, "loss": 0.0077, "step": 6791 }, { "epoch": 8.89, "learning_rate": 1.6085e-05, "loss": 0.0069, "step": 6792 }, { "epoch": 8.89, "learning_rate": 1.6080000000000002e-05, "loss": 0.0093, "step": 6793 }, { "epoch": 8.89, "learning_rate": 1.6075e-05, "loss": 0.0081, "step": 6794 }, { "epoch": 8.89, "learning_rate": 1.607e-05, "loss": 0.0093, "step": 6795 }, { "epoch": 8.9, "learning_rate": 1.6065e-05, "loss": 0.0079, "step": 6796 }, { "epoch": 8.9, "learning_rate": 1.606e-05, "loss": 0.0082, "step": 6797 }, { "epoch": 8.9, "learning_rate": 1.6055e-05, "loss": 0.0103, "step": 6798 }, { "epoch": 8.9, "learning_rate": 1.605e-05, "loss": 0.0099, "step": 6799 }, { "epoch": 8.9, "learning_rate": 1.6045000000000003e-05, "loss": 0.0096, "step": 6800 }, { "epoch": 8.9, "learning_rate": 1.604e-05, "loss": 0.0099, "step": 6801 }, { "epoch": 8.9, "learning_rate": 1.6035e-05, "loss": 0.0108, "step": 6802 }, { "epoch": 8.9, "learning_rate": 1.603e-05, "loss": 0.0097, "step": 6803 }, { "epoch": 8.91, "learning_rate": 1.6025e-05, "loss": 0.0066, "step": 6804 }, { "epoch": 8.91, "learning_rate": 1.6020000000000002e-05, "loss": 0.0088, "step": 6805 }, { "epoch": 8.91, "learning_rate": 1.6014999999999998e-05, "loss": 0.0084, "step": 6806 }, { "epoch": 8.91, "learning_rate": 1.601e-05, "loss": 0.007, "step": 6807 }, { "epoch": 8.91, "learning_rate": 1.6005e-05, "loss": 0.0099, "step": 6808 }, { "epoch": 8.91, "learning_rate": 1.6000000000000003e-05, "loss": 0.009, "step": 6809 }, { "epoch": 8.91, "learning_rate": 1.5995000000000002e-05, "loss": 0.0078, "step": 6810 }, { "epoch": 8.91, "learning_rate": 1.599e-05, "loss": 0.0089, "step": 6811 }, { "epoch": 8.92, "learning_rate": 1.5985e-05, "loss": 0.0057, "step": 6812 }, { "epoch": 8.92, "learning_rate": 1.598e-05, "loss": 0.009, "step": 6813 }, { "epoch": 8.92, "learning_rate": 1.5975000000000002e-05, "loss": 0.0099, "step": 6814 }, { "epoch": 8.92, "learning_rate": 1.597e-05, "loss": 0.0081, "step": 6815 }, { "epoch": 8.92, "learning_rate": 1.5965e-05, "loss": 0.0097, "step": 6816 }, { "epoch": 8.92, "learning_rate": 1.596e-05, "loss": 0.0086, "step": 6817 }, { "epoch": 8.92, "learning_rate": 1.5955e-05, "loss": 0.0106, "step": 6818 }, { "epoch": 8.93, "learning_rate": 1.595e-05, "loss": 0.0067, "step": 6819 }, { "epoch": 8.93, "learning_rate": 1.5945e-05, "loss": 0.0078, "step": 6820 }, { "epoch": 8.93, "learning_rate": 1.594e-05, "loss": 0.0093, "step": 6821 }, { "epoch": 8.93, "learning_rate": 1.5935e-05, "loss": 0.0094, "step": 6822 }, { "epoch": 8.93, "learning_rate": 1.593e-05, "loss": 0.0103, "step": 6823 }, { "epoch": 8.93, "learning_rate": 1.5925e-05, "loss": 0.0078, "step": 6824 }, { "epoch": 8.93, "learning_rate": 1.592e-05, "loss": 0.0086, "step": 6825 }, { "epoch": 8.93, "learning_rate": 1.5915000000000003e-05, "loss": 0.0113, "step": 6826 }, { "epoch": 8.94, "learning_rate": 1.591e-05, "loss": 0.0081, "step": 6827 }, { "epoch": 8.94, "learning_rate": 1.5905e-05, "loss": 0.0087, "step": 6828 }, { "epoch": 8.94, "learning_rate": 1.59e-05, "loss": 0.0077, "step": 6829 }, { "epoch": 8.94, "learning_rate": 1.5895000000000003e-05, "loss": 0.0089, "step": 6830 }, { "epoch": 8.94, "learning_rate": 1.5890000000000002e-05, "loss": 0.0085, "step": 6831 }, { "epoch": 8.94, "learning_rate": 1.5885e-05, "loss": 0.009, "step": 6832 }, { "epoch": 8.94, "learning_rate": 1.588e-05, "loss": 0.0089, "step": 6833 }, { "epoch": 8.95, "learning_rate": 1.5875e-05, "loss": 0.0091, "step": 6834 }, { "epoch": 8.95, "learning_rate": 1.5870000000000002e-05, "loss": 0.0084, "step": 6835 }, { "epoch": 8.95, "learning_rate": 1.5865e-05, "loss": 0.0073, "step": 6836 }, { "epoch": 8.95, "learning_rate": 1.586e-05, "loss": 0.0101, "step": 6837 }, { "epoch": 8.95, "learning_rate": 1.5855e-05, "loss": 0.0092, "step": 6838 }, { "epoch": 8.95, "learning_rate": 1.5850000000000002e-05, "loss": 0.0095, "step": 6839 }, { "epoch": 8.95, "learning_rate": 1.5845e-05, "loss": 0.0097, "step": 6840 }, { "epoch": 8.95, "learning_rate": 1.584e-05, "loss": 0.0118, "step": 6841 }, { "epoch": 8.96, "learning_rate": 1.5835e-05, "loss": 0.0094, "step": 6842 }, { "epoch": 8.96, "learning_rate": 1.583e-05, "loss": 0.0088, "step": 6843 }, { "epoch": 8.96, "learning_rate": 1.5825000000000002e-05, "loss": 0.0111, "step": 6844 }, { "epoch": 8.96, "learning_rate": 1.582e-05, "loss": 0.0074, "step": 6845 }, { "epoch": 8.96, "learning_rate": 1.5815000000000004e-05, "loss": 0.0093, "step": 6846 }, { "epoch": 8.96, "learning_rate": 1.581e-05, "loss": 0.009, "step": 6847 }, { "epoch": 8.96, "learning_rate": 1.5805000000000002e-05, "loss": 0.0081, "step": 6848 }, { "epoch": 8.96, "learning_rate": 1.58e-05, "loss": 0.0078, "step": 6849 }, { "epoch": 8.97, "learning_rate": 1.5795e-05, "loss": 0.0095, "step": 6850 }, { "epoch": 8.97, "learning_rate": 1.5790000000000003e-05, "loss": 0.0122, "step": 6851 }, { "epoch": 8.97, "learning_rate": 1.5785e-05, "loss": 0.0101, "step": 6852 }, { "epoch": 8.97, "learning_rate": 1.578e-05, "loss": 0.014, "step": 6853 }, { "epoch": 8.97, "learning_rate": 1.5775e-05, "loss": 0.009, "step": 6854 }, { "epoch": 8.97, "learning_rate": 1.577e-05, "loss": 0.0081, "step": 6855 }, { "epoch": 8.97, "learning_rate": 1.5765000000000002e-05, "loss": 0.0078, "step": 6856 }, { "epoch": 8.98, "learning_rate": 1.5759999999999998e-05, "loss": 0.0077, "step": 6857 }, { "epoch": 8.98, "learning_rate": 1.5755e-05, "loss": 0.0086, "step": 6858 }, { "epoch": 8.98, "learning_rate": 1.575e-05, "loss": 0.0081, "step": 6859 }, { "epoch": 8.98, "learning_rate": 1.5745000000000003e-05, "loss": 0.0091, "step": 6860 }, { "epoch": 8.98, "learning_rate": 1.5740000000000002e-05, "loss": 0.0089, "step": 6861 }, { "epoch": 8.98, "learning_rate": 1.5735e-05, "loss": 0.0102, "step": 6862 }, { "epoch": 8.98, "learning_rate": 1.573e-05, "loss": 0.0105, "step": 6863 }, { "epoch": 8.98, "learning_rate": 1.5725e-05, "loss": 0.0092, "step": 6864 }, { "epoch": 8.99, "learning_rate": 1.5720000000000002e-05, "loss": 0.0437, "step": 6865 }, { "epoch": 8.99, "learning_rate": 1.5715e-05, "loss": 0.0112, "step": 6866 }, { "epoch": 8.99, "learning_rate": 1.571e-05, "loss": 0.009, "step": 6867 }, { "epoch": 8.99, "learning_rate": 1.5705e-05, "loss": 0.0084, "step": 6868 }, { "epoch": 8.99, "learning_rate": 1.5700000000000002e-05, "loss": 0.0084, "step": 6869 }, { "epoch": 8.99, "learning_rate": 1.5695e-05, "loss": 0.0096, "step": 6870 }, { "epoch": 8.99, "learning_rate": 1.569e-05, "loss": 0.0075, "step": 6871 }, { "epoch": 8.99, "learning_rate": 1.5685e-05, "loss": 0.0103, "step": 6872 }, { "epoch": 9.0, "learning_rate": 1.568e-05, "loss": 0.0095, "step": 6873 }, { "epoch": 9.0, "learning_rate": 1.5675e-05, "loss": 0.0132, "step": 6874 }, { "epoch": 9.0, "learning_rate": 1.567e-05, "loss": 0.0093, "step": 6875 }, { "epoch": 9.0, "learning_rate": 1.5665000000000003e-05, "loss": 0.009, "step": 6876 }, { "epoch": 9.0, "learning_rate": 1.566e-05, "loss": 0.0093, "step": 6877 }, { "epoch": 9.0, "learning_rate": 1.5655000000000002e-05, "loss": 0.0061, "step": 6878 }, { "epoch": 9.0, "learning_rate": 1.565e-05, "loss": 0.0057, "step": 6879 }, { "epoch": 9.01, "learning_rate": 1.5645e-05, "loss": 0.0054, "step": 6880 }, { "epoch": 9.01, "learning_rate": 1.5640000000000003e-05, "loss": 0.0058, "step": 6881 }, { "epoch": 9.01, "learning_rate": 1.5635e-05, "loss": 0.0064, "step": 6882 }, { "epoch": 9.01, "learning_rate": 1.563e-05, "loss": 0.0055, "step": 6883 }, { "epoch": 9.01, "learning_rate": 1.5625e-05, "loss": 0.0054, "step": 6884 }, { "epoch": 9.01, "learning_rate": 1.5620000000000003e-05, "loss": 0.0048, "step": 6885 }, { "epoch": 9.01, "learning_rate": 1.5615000000000002e-05, "loss": 0.0057, "step": 6886 }, { "epoch": 9.01, "learning_rate": 1.561e-05, "loss": 0.0052, "step": 6887 }, { "epoch": 9.02, "learning_rate": 1.5605e-05, "loss": 0.0056, "step": 6888 }, { "epoch": 9.02, "learning_rate": 1.56e-05, "loss": 0.0055, "step": 6889 }, { "epoch": 9.02, "learning_rate": 1.5595000000000002e-05, "loss": 0.0068, "step": 6890 }, { "epoch": 9.02, "learning_rate": 1.559e-05, "loss": 0.0058, "step": 6891 }, { "epoch": 9.02, "learning_rate": 1.5585e-05, "loss": 0.0073, "step": 6892 }, { "epoch": 9.02, "learning_rate": 1.558e-05, "loss": 0.0072, "step": 6893 }, { "epoch": 9.02, "learning_rate": 1.5575e-05, "loss": 0.0074, "step": 6894 }, { "epoch": 9.02, "learning_rate": 1.5570000000000002e-05, "loss": 0.0055, "step": 6895 }, { "epoch": 9.03, "learning_rate": 1.5565e-05, "loss": 0.0049, "step": 6896 }, { "epoch": 9.03, "learning_rate": 1.556e-05, "loss": 0.0052, "step": 6897 }, { "epoch": 9.03, "learning_rate": 1.5555e-05, "loss": 0.0053, "step": 6898 }, { "epoch": 9.03, "learning_rate": 1.5550000000000002e-05, "loss": 0.0075, "step": 6899 }, { "epoch": 9.03, "learning_rate": 1.5545e-05, "loss": 0.0059, "step": 6900 }, { "epoch": 9.03, "learning_rate": 1.554e-05, "loss": 0.0049, "step": 6901 }, { "epoch": 9.03, "learning_rate": 1.5535e-05, "loss": 0.0064, "step": 6902 }, { "epoch": 9.04, "learning_rate": 1.553e-05, "loss": 0.005, "step": 6903 }, { "epoch": 9.04, "learning_rate": 1.5525e-05, "loss": 0.0047, "step": 6904 }, { "epoch": 9.04, "learning_rate": 1.552e-05, "loss": 0.0048, "step": 6905 }, { "epoch": 9.04, "learning_rate": 1.5515000000000003e-05, "loss": 0.0054, "step": 6906 }, { "epoch": 9.04, "learning_rate": 1.551e-05, "loss": 0.0053, "step": 6907 }, { "epoch": 9.04, "learning_rate": 1.5505e-05, "loss": 0.0053, "step": 6908 }, { "epoch": 9.04, "learning_rate": 1.55e-05, "loss": 0.0047, "step": 6909 }, { "epoch": 9.04, "learning_rate": 1.5495e-05, "loss": 0.005, "step": 6910 }, { "epoch": 9.05, "learning_rate": 1.5490000000000002e-05, "loss": 0.0056, "step": 6911 }, { "epoch": 9.05, "learning_rate": 1.5484999999999998e-05, "loss": 0.0075, "step": 6912 }, { "epoch": 9.05, "learning_rate": 1.548e-05, "loss": 0.0068, "step": 6913 }, { "epoch": 9.05, "learning_rate": 1.5475e-05, "loss": 0.0046, "step": 6914 }, { "epoch": 9.05, "learning_rate": 1.5470000000000003e-05, "loss": 0.0069, "step": 6915 }, { "epoch": 9.05, "learning_rate": 1.5465000000000002e-05, "loss": 0.0071, "step": 6916 }, { "epoch": 9.05, "learning_rate": 1.546e-05, "loss": 0.0054, "step": 6917 }, { "epoch": 9.05, "learning_rate": 1.5455e-05, "loss": 0.0053, "step": 6918 }, { "epoch": 9.06, "learning_rate": 1.545e-05, "loss": 0.0069, "step": 6919 }, { "epoch": 9.06, "learning_rate": 1.5445000000000002e-05, "loss": 0.0069, "step": 6920 }, { "epoch": 9.06, "learning_rate": 1.544e-05, "loss": 0.0047, "step": 6921 }, { "epoch": 9.06, "learning_rate": 1.5435e-05, "loss": 0.0055, "step": 6922 }, { "epoch": 9.06, "learning_rate": 1.543e-05, "loss": 0.0051, "step": 6923 }, { "epoch": 9.06, "learning_rate": 1.5425000000000002e-05, "loss": 0.0067, "step": 6924 }, { "epoch": 9.06, "learning_rate": 1.542e-05, "loss": 0.0058, "step": 6925 }, { "epoch": 9.07, "learning_rate": 1.5415e-05, "loss": 0.0047, "step": 6926 }, { "epoch": 9.07, "learning_rate": 1.541e-05, "loss": 0.0058, "step": 6927 }, { "epoch": 9.07, "learning_rate": 1.5405e-05, "loss": 0.0079, "step": 6928 }, { "epoch": 9.07, "learning_rate": 1.54e-05, "loss": 0.0043, "step": 6929 }, { "epoch": 9.07, "learning_rate": 1.5395e-05, "loss": 0.0045, "step": 6930 }, { "epoch": 9.07, "learning_rate": 1.539e-05, "loss": 0.0078, "step": 6931 }, { "epoch": 9.07, "learning_rate": 1.5385e-05, "loss": 0.0047, "step": 6932 }, { "epoch": 9.07, "learning_rate": 1.538e-05, "loss": 0.0066, "step": 6933 }, { "epoch": 9.08, "learning_rate": 1.5375e-05, "loss": 0.0049, "step": 6934 }, { "epoch": 9.08, "learning_rate": 1.537e-05, "loss": 0.0048, "step": 6935 }, { "epoch": 9.08, "learning_rate": 1.5365000000000003e-05, "loss": 0.0074, "step": 6936 }, { "epoch": 9.08, "learning_rate": 1.536e-05, "loss": 0.0061, "step": 6937 }, { "epoch": 9.08, "learning_rate": 1.5355e-05, "loss": 0.0046, "step": 6938 }, { "epoch": 9.08, "learning_rate": 1.535e-05, "loss": 0.0041, "step": 6939 }, { "epoch": 9.08, "learning_rate": 1.5345e-05, "loss": 0.0059, "step": 6940 }, { "epoch": 9.09, "learning_rate": 1.5340000000000002e-05, "loss": 0.0052, "step": 6941 }, { "epoch": 9.09, "learning_rate": 1.5334999999999998e-05, "loss": 0.006, "step": 6942 }, { "epoch": 9.09, "learning_rate": 1.533e-05, "loss": 0.0074, "step": 6943 }, { "epoch": 9.09, "learning_rate": 1.5325e-05, "loss": 0.0059, "step": 6944 }, { "epoch": 9.09, "learning_rate": 1.5320000000000002e-05, "loss": 0.0048, "step": 6945 }, { "epoch": 9.09, "learning_rate": 1.5315e-05, "loss": 0.0057, "step": 6946 }, { "epoch": 9.09, "learning_rate": 1.531e-05, "loss": 0.0055, "step": 6947 }, { "epoch": 9.09, "learning_rate": 1.5305e-05, "loss": 0.0038, "step": 6948 }, { "epoch": 9.1, "learning_rate": 1.53e-05, "loss": 0.0056, "step": 6949 }, { "epoch": 9.1, "learning_rate": 1.5295000000000002e-05, "loss": 0.0036, "step": 6950 }, { "epoch": 9.1, "learning_rate": 1.529e-05, "loss": 0.0053, "step": 6951 }, { "epoch": 9.1, "learning_rate": 1.5285000000000004e-05, "loss": 0.0052, "step": 6952 }, { "epoch": 9.1, "learning_rate": 1.528e-05, "loss": 0.0049, "step": 6953 }, { "epoch": 9.1, "learning_rate": 1.5275000000000002e-05, "loss": 0.0044, "step": 6954 }, { "epoch": 9.1, "learning_rate": 1.527e-05, "loss": 0.0059, "step": 6955 }, { "epoch": 9.1, "learning_rate": 1.5265e-05, "loss": 0.0038, "step": 6956 }, { "epoch": 9.11, "learning_rate": 1.5260000000000003e-05, "loss": 0.006, "step": 6957 }, { "epoch": 9.11, "learning_rate": 1.5255e-05, "loss": 0.0049, "step": 6958 }, { "epoch": 9.11, "learning_rate": 1.525e-05, "loss": 0.0053, "step": 6959 }, { "epoch": 9.11, "learning_rate": 1.5245e-05, "loss": 0.0068, "step": 6960 }, { "epoch": 9.11, "learning_rate": 1.5240000000000001e-05, "loss": 0.0064, "step": 6961 }, { "epoch": 9.11, "learning_rate": 1.5235000000000002e-05, "loss": 0.0082, "step": 6962 }, { "epoch": 9.11, "learning_rate": 1.523e-05, "loss": 0.0036, "step": 6963 }, { "epoch": 9.12, "learning_rate": 1.5225e-05, "loss": 0.0079, "step": 6964 }, { "epoch": 9.12, "learning_rate": 1.5220000000000002e-05, "loss": 0.0041, "step": 6965 }, { "epoch": 9.12, "learning_rate": 1.5215000000000001e-05, "loss": 0.0051, "step": 6966 }, { "epoch": 9.12, "learning_rate": 1.5210000000000002e-05, "loss": 0.0057, "step": 6967 }, { "epoch": 9.12, "learning_rate": 1.5205e-05, "loss": 0.0066, "step": 6968 }, { "epoch": 9.12, "learning_rate": 1.52e-05, "loss": 0.0051, "step": 6969 }, { "epoch": 9.12, "learning_rate": 1.5195000000000001e-05, "loss": 0.0054, "step": 6970 }, { "epoch": 9.12, "learning_rate": 1.5190000000000002e-05, "loss": 0.0039, "step": 6971 }, { "epoch": 9.13, "learning_rate": 1.5185000000000003e-05, "loss": 0.0072, "step": 6972 }, { "epoch": 9.13, "learning_rate": 1.518e-05, "loss": 0.0052, "step": 6973 }, { "epoch": 9.13, "learning_rate": 1.5175e-05, "loss": 0.0044, "step": 6974 }, { "epoch": 9.13, "learning_rate": 1.517e-05, "loss": 0.0045, "step": 6975 }, { "epoch": 9.13, "learning_rate": 1.5165000000000001e-05, "loss": 0.0073, "step": 6976 }, { "epoch": 9.13, "learning_rate": 1.5160000000000002e-05, "loss": 0.0055, "step": 6977 }, { "epoch": 9.13, "learning_rate": 1.5155e-05, "loss": 0.0064, "step": 6978 }, { "epoch": 9.13, "learning_rate": 1.515e-05, "loss": 0.006, "step": 6979 }, { "epoch": 9.14, "learning_rate": 1.5145000000000002e-05, "loss": 0.0038, "step": 6980 }, { "epoch": 9.14, "learning_rate": 1.514e-05, "loss": 0.0052, "step": 6981 }, { "epoch": 9.14, "learning_rate": 1.5135000000000002e-05, "loss": 0.0059, "step": 6982 }, { "epoch": 9.14, "learning_rate": 1.5129999999999999e-05, "loss": 0.0057, "step": 6983 }, { "epoch": 9.14, "learning_rate": 1.5125e-05, "loss": 0.0072, "step": 6984 }, { "epoch": 9.14, "learning_rate": 1.5120000000000001e-05, "loss": 0.0073, "step": 6985 }, { "epoch": 9.14, "learning_rate": 1.5115000000000002e-05, "loss": 0.0044, "step": 6986 }, { "epoch": 9.15, "learning_rate": 1.5110000000000003e-05, "loss": 0.004, "step": 6987 }, { "epoch": 9.15, "learning_rate": 1.5105e-05, "loss": 0.0061, "step": 6988 }, { "epoch": 9.15, "learning_rate": 1.51e-05, "loss": 0.004, "step": 6989 }, { "epoch": 9.15, "learning_rate": 1.5095e-05, "loss": 0.0051, "step": 6990 }, { "epoch": 9.15, "learning_rate": 1.5090000000000001e-05, "loss": 0.0047, "step": 6991 }, { "epoch": 9.15, "learning_rate": 1.5085000000000002e-05, "loss": 0.0044, "step": 6992 }, { "epoch": 9.15, "learning_rate": 1.508e-05, "loss": 0.007, "step": 6993 }, { "epoch": 9.15, "learning_rate": 1.5075e-05, "loss": 0.0055, "step": 6994 }, { "epoch": 9.16, "learning_rate": 1.5070000000000001e-05, "loss": 0.0058, "step": 6995 }, { "epoch": 9.16, "learning_rate": 1.5065e-05, "loss": 0.0078, "step": 6996 }, { "epoch": 9.16, "learning_rate": 1.5060000000000001e-05, "loss": 0.0037, "step": 6997 }, { "epoch": 9.16, "learning_rate": 1.5054999999999999e-05, "loss": 0.0056, "step": 6998 }, { "epoch": 9.16, "learning_rate": 1.505e-05, "loss": 0.008, "step": 6999 }, { "epoch": 9.16, "learning_rate": 1.5045e-05, "loss": 0.0069, "step": 7000 }, { "epoch": 9.16, "learning_rate": 1.5040000000000002e-05, "loss": 0.0064, "step": 7001 }, { "epoch": 9.16, "learning_rate": 1.5035000000000003e-05, "loss": 0.0056, "step": 7002 }, { "epoch": 9.17, "learning_rate": 1.503e-05, "loss": 0.0063, "step": 7003 }, { "epoch": 9.17, "learning_rate": 1.5025000000000001e-05, "loss": 0.0059, "step": 7004 }, { "epoch": 9.17, "learning_rate": 1.502e-05, "loss": 0.007, "step": 7005 }, { "epoch": 9.17, "learning_rate": 1.5015000000000001e-05, "loss": 0.0053, "step": 7006 }, { "epoch": 9.17, "learning_rate": 1.5010000000000002e-05, "loss": 0.0058, "step": 7007 }, { "epoch": 9.17, "learning_rate": 1.5005e-05, "loss": 0.0054, "step": 7008 }, { "epoch": 9.17, "learning_rate": 1.5e-05, "loss": 0.0038, "step": 7009 }, { "epoch": 9.18, "learning_rate": 1.4995000000000001e-05, "loss": 0.0046, "step": 7010 }, { "epoch": 9.18, "learning_rate": 1.499e-05, "loss": 0.0056, "step": 7011 }, { "epoch": 9.18, "learning_rate": 1.4985000000000001e-05, "loss": 0.0077, "step": 7012 }, { "epoch": 9.18, "learning_rate": 1.4979999999999999e-05, "loss": 0.0047, "step": 7013 }, { "epoch": 9.18, "learning_rate": 1.4975e-05, "loss": 0.0053, "step": 7014 }, { "epoch": 9.18, "learning_rate": 1.497e-05, "loss": 0.0051, "step": 7015 }, { "epoch": 9.18, "learning_rate": 1.4965000000000002e-05, "loss": 0.0141, "step": 7016 }, { "epoch": 9.18, "learning_rate": 1.4960000000000002e-05, "loss": 0.005, "step": 7017 }, { "epoch": 9.19, "learning_rate": 1.4955e-05, "loss": 0.0073, "step": 7018 }, { "epoch": 9.19, "learning_rate": 1.4950000000000001e-05, "loss": 0.0095, "step": 7019 }, { "epoch": 9.19, "learning_rate": 1.4945e-05, "loss": 0.0055, "step": 7020 }, { "epoch": 9.19, "learning_rate": 1.4940000000000001e-05, "loss": 0.0057, "step": 7021 }, { "epoch": 9.19, "learning_rate": 1.4935000000000002e-05, "loss": 0.0042, "step": 7022 }, { "epoch": 9.19, "learning_rate": 1.493e-05, "loss": 0.0057, "step": 7023 }, { "epoch": 9.19, "learning_rate": 1.4925e-05, "loss": 0.0056, "step": 7024 }, { "epoch": 9.2, "learning_rate": 1.4920000000000001e-05, "loss": 0.0054, "step": 7025 }, { "epoch": 9.2, "learning_rate": 1.4915000000000002e-05, "loss": 0.0052, "step": 7026 }, { "epoch": 9.2, "learning_rate": 1.4910000000000001e-05, "loss": 0.0058, "step": 7027 }, { "epoch": 9.2, "learning_rate": 1.4904999999999999e-05, "loss": 0.0054, "step": 7028 }, { "epoch": 9.2, "learning_rate": 1.49e-05, "loss": 0.0054, "step": 7029 }, { "epoch": 9.2, "learning_rate": 1.4895e-05, "loss": 0.0051, "step": 7030 }, { "epoch": 9.2, "learning_rate": 1.4890000000000001e-05, "loss": 0.0041, "step": 7031 }, { "epoch": 9.2, "learning_rate": 1.4885000000000002e-05, "loss": 0.0044, "step": 7032 }, { "epoch": 9.21, "learning_rate": 1.488e-05, "loss": 0.0051, "step": 7033 }, { "epoch": 9.21, "learning_rate": 1.4875e-05, "loss": 0.0058, "step": 7034 }, { "epoch": 9.21, "learning_rate": 1.487e-05, "loss": 0.0055, "step": 7035 }, { "epoch": 9.21, "learning_rate": 1.4865e-05, "loss": 0.0059, "step": 7036 }, { "epoch": 9.21, "learning_rate": 1.4860000000000002e-05, "loss": 0.0042, "step": 7037 }, { "epoch": 9.21, "learning_rate": 1.4855e-05, "loss": 0.0064, "step": 7038 }, { "epoch": 9.21, "learning_rate": 1.485e-05, "loss": 0.0056, "step": 7039 }, { "epoch": 9.21, "learning_rate": 1.4845000000000001e-05, "loss": 0.0044, "step": 7040 }, { "epoch": 9.22, "learning_rate": 1.4840000000000002e-05, "loss": 0.0051, "step": 7041 }, { "epoch": 9.22, "learning_rate": 1.4835000000000001e-05, "loss": 0.0063, "step": 7042 }, { "epoch": 9.22, "learning_rate": 1.4829999999999999e-05, "loss": 0.0047, "step": 7043 }, { "epoch": 9.22, "learning_rate": 1.4825e-05, "loss": 0.0048, "step": 7044 }, { "epoch": 9.22, "learning_rate": 1.482e-05, "loss": 0.0055, "step": 7045 }, { "epoch": 9.22, "learning_rate": 1.4815000000000001e-05, "loss": 0.0047, "step": 7046 }, { "epoch": 9.22, "learning_rate": 1.4810000000000002e-05, "loss": 0.0058, "step": 7047 }, { "epoch": 9.23, "learning_rate": 1.4805e-05, "loss": 0.005, "step": 7048 }, { "epoch": 9.23, "learning_rate": 1.48e-05, "loss": 0.0069, "step": 7049 }, { "epoch": 9.23, "learning_rate": 1.4795e-05, "loss": 0.0062, "step": 7050 }, { "epoch": 9.23, "learning_rate": 1.479e-05, "loss": 0.0047, "step": 7051 }, { "epoch": 9.23, "learning_rate": 1.4785000000000002e-05, "loss": 0.0053, "step": 7052 }, { "epoch": 9.23, "learning_rate": 1.4779999999999999e-05, "loss": 0.0043, "step": 7053 }, { "epoch": 9.23, "learning_rate": 1.4775e-05, "loss": 0.0053, "step": 7054 }, { "epoch": 9.23, "learning_rate": 1.4770000000000001e-05, "loss": 0.0038, "step": 7055 }, { "epoch": 9.24, "learning_rate": 1.4765000000000002e-05, "loss": 0.0068, "step": 7056 }, { "epoch": 9.24, "learning_rate": 1.4760000000000001e-05, "loss": 0.0049, "step": 7057 }, { "epoch": 9.24, "learning_rate": 1.4755e-05, "loss": 0.0058, "step": 7058 }, { "epoch": 9.24, "learning_rate": 1.475e-05, "loss": 0.0076, "step": 7059 }, { "epoch": 9.24, "learning_rate": 1.4745e-05, "loss": 0.0046, "step": 7060 }, { "epoch": 9.24, "learning_rate": 1.4740000000000001e-05, "loss": 0.0051, "step": 7061 }, { "epoch": 9.24, "learning_rate": 1.4735000000000002e-05, "loss": 0.0053, "step": 7062 }, { "epoch": 9.24, "learning_rate": 1.473e-05, "loss": 0.0053, "step": 7063 }, { "epoch": 9.25, "learning_rate": 1.4725e-05, "loss": 0.0067, "step": 7064 }, { "epoch": 9.25, "learning_rate": 1.472e-05, "loss": 0.0056, "step": 7065 }, { "epoch": 9.25, "learning_rate": 1.4715e-05, "loss": 0.0053, "step": 7066 }, { "epoch": 9.25, "learning_rate": 1.4710000000000001e-05, "loss": 0.006, "step": 7067 }, { "epoch": 9.25, "learning_rate": 1.4704999999999999e-05, "loss": 0.0055, "step": 7068 }, { "epoch": 9.25, "learning_rate": 1.47e-05, "loss": 0.0047, "step": 7069 }, { "epoch": 9.25, "learning_rate": 1.4695e-05, "loss": 0.0042, "step": 7070 }, { "epoch": 9.26, "learning_rate": 1.4690000000000002e-05, "loss": 0.0064, "step": 7071 }, { "epoch": 9.26, "learning_rate": 1.4685000000000001e-05, "loss": 0.0049, "step": 7072 }, { "epoch": 9.26, "learning_rate": 1.4680000000000002e-05, "loss": 0.0062, "step": 7073 }, { "epoch": 9.26, "learning_rate": 1.4675e-05, "loss": 0.0065, "step": 7074 }, { "epoch": 9.26, "learning_rate": 1.467e-05, "loss": 0.0055, "step": 7075 }, { "epoch": 9.26, "learning_rate": 1.4665000000000001e-05, "loss": 0.0046, "step": 7076 }, { "epoch": 9.26, "learning_rate": 1.4660000000000002e-05, "loss": 0.0038, "step": 7077 }, { "epoch": 9.26, "learning_rate": 1.4655000000000003e-05, "loss": 0.0048, "step": 7078 }, { "epoch": 9.27, "learning_rate": 1.465e-05, "loss": 0.0052, "step": 7079 }, { "epoch": 9.27, "learning_rate": 1.4645e-05, "loss": 0.0103, "step": 7080 }, { "epoch": 9.27, "learning_rate": 1.464e-05, "loss": 0.0067, "step": 7081 }, { "epoch": 9.27, "learning_rate": 1.4635000000000001e-05, "loss": 0.005, "step": 7082 }, { "epoch": 9.27, "learning_rate": 1.4630000000000002e-05, "loss": 0.0049, "step": 7083 }, { "epoch": 9.27, "learning_rate": 1.4625e-05, "loss": 0.0065, "step": 7084 }, { "epoch": 9.27, "learning_rate": 1.462e-05, "loss": 0.0054, "step": 7085 }, { "epoch": 9.27, "learning_rate": 1.4615000000000002e-05, "loss": 0.0074, "step": 7086 }, { "epoch": 9.28, "learning_rate": 1.461e-05, "loss": 0.0046, "step": 7087 }, { "epoch": 9.28, "learning_rate": 1.4605000000000002e-05, "loss": 0.0047, "step": 7088 }, { "epoch": 9.28, "learning_rate": 1.4599999999999999e-05, "loss": 0.0057, "step": 7089 }, { "epoch": 9.28, "learning_rate": 1.4595e-05, "loss": 0.0052, "step": 7090 }, { "epoch": 9.28, "learning_rate": 1.4590000000000001e-05, "loss": 0.0064, "step": 7091 }, { "epoch": 9.28, "learning_rate": 1.4585000000000002e-05, "loss": 0.0042, "step": 7092 }, { "epoch": 9.28, "learning_rate": 1.4580000000000003e-05, "loss": 0.006, "step": 7093 }, { "epoch": 9.29, "learning_rate": 1.4575e-05, "loss": 0.0073, "step": 7094 }, { "epoch": 9.29, "learning_rate": 1.4570000000000001e-05, "loss": 0.013, "step": 7095 }, { "epoch": 9.29, "learning_rate": 1.4565e-05, "loss": 0.006, "step": 7096 }, { "epoch": 9.29, "learning_rate": 1.4560000000000001e-05, "loss": 0.0059, "step": 7097 }, { "epoch": 9.29, "learning_rate": 1.4555000000000002e-05, "loss": 0.0044, "step": 7098 }, { "epoch": 9.29, "learning_rate": 1.455e-05, "loss": 0.0047, "step": 7099 }, { "epoch": 9.29, "learning_rate": 1.4545e-05, "loss": 0.0063, "step": 7100 }, { "epoch": 9.29, "learning_rate": 1.4540000000000001e-05, "loss": 0.0066, "step": 7101 }, { "epoch": 9.3, "learning_rate": 1.4535e-05, "loss": 0.006, "step": 7102 }, { "epoch": 9.3, "learning_rate": 1.4530000000000001e-05, "loss": 0.0057, "step": 7103 }, { "epoch": 9.3, "learning_rate": 1.4524999999999999e-05, "loss": 0.0055, "step": 7104 }, { "epoch": 9.3, "learning_rate": 1.452e-05, "loss": 0.0171, "step": 7105 }, { "epoch": 9.3, "learning_rate": 1.4515e-05, "loss": 0.0057, "step": 7106 }, { "epoch": 9.3, "learning_rate": 1.4510000000000002e-05, "loss": 0.0045, "step": 7107 }, { "epoch": 9.3, "learning_rate": 1.4505000000000003e-05, "loss": 0.0062, "step": 7108 }, { "epoch": 9.3, "learning_rate": 1.45e-05, "loss": 0.0063, "step": 7109 }, { "epoch": 9.31, "learning_rate": 1.4495000000000001e-05, "loss": 0.0034, "step": 7110 }, { "epoch": 9.31, "learning_rate": 1.449e-05, "loss": 0.0057, "step": 7111 }, { "epoch": 9.31, "learning_rate": 1.4485000000000001e-05, "loss": 0.0053, "step": 7112 }, { "epoch": 9.31, "learning_rate": 1.4480000000000002e-05, "loss": 0.0048, "step": 7113 }, { "epoch": 9.31, "learning_rate": 1.4475e-05, "loss": 0.0061, "step": 7114 }, { "epoch": 9.31, "learning_rate": 1.447e-05, "loss": 0.0059, "step": 7115 }, { "epoch": 9.31, "learning_rate": 1.4465000000000001e-05, "loss": 0.0034, "step": 7116 }, { "epoch": 9.32, "learning_rate": 1.4460000000000002e-05, "loss": 0.004, "step": 7117 }, { "epoch": 9.32, "learning_rate": 1.4455000000000001e-05, "loss": 0.0056, "step": 7118 }, { "epoch": 9.32, "learning_rate": 1.4449999999999999e-05, "loss": 0.0056, "step": 7119 }, { "epoch": 9.32, "learning_rate": 1.4445e-05, "loss": 0.0036, "step": 7120 }, { "epoch": 9.32, "learning_rate": 1.444e-05, "loss": 0.0058, "step": 7121 }, { "epoch": 9.32, "learning_rate": 1.4435000000000002e-05, "loss": 0.0047, "step": 7122 }, { "epoch": 9.32, "learning_rate": 1.4430000000000002e-05, "loss": 0.0059, "step": 7123 }, { "epoch": 9.32, "learning_rate": 1.4425e-05, "loss": 0.0038, "step": 7124 }, { "epoch": 9.33, "learning_rate": 1.4420000000000001e-05, "loss": 0.0066, "step": 7125 }, { "epoch": 9.33, "learning_rate": 1.4415e-05, "loss": 0.0048, "step": 7126 }, { "epoch": 9.33, "learning_rate": 1.4410000000000001e-05, "loss": 0.0062, "step": 7127 }, { "epoch": 9.33, "learning_rate": 1.4405000000000002e-05, "loss": 0.0065, "step": 7128 }, { "epoch": 9.33, "learning_rate": 1.44e-05, "loss": 0.0063, "step": 7129 }, { "epoch": 9.33, "learning_rate": 1.4395e-05, "loss": 0.0042, "step": 7130 }, { "epoch": 9.33, "learning_rate": 1.4390000000000001e-05, "loss": 0.0035, "step": 7131 }, { "epoch": 9.34, "learning_rate": 1.4385000000000002e-05, "loss": 0.004, "step": 7132 }, { "epoch": 9.34, "learning_rate": 1.4380000000000001e-05, "loss": 0.0059, "step": 7133 }, { "epoch": 9.34, "learning_rate": 1.4374999999999999e-05, "loss": 0.0045, "step": 7134 }, { "epoch": 9.34, "learning_rate": 1.437e-05, "loss": 0.0065, "step": 7135 }, { "epoch": 9.34, "learning_rate": 1.4365e-05, "loss": 0.0056, "step": 7136 }, { "epoch": 9.34, "learning_rate": 1.4360000000000001e-05, "loss": 0.0048, "step": 7137 }, { "epoch": 9.34, "learning_rate": 1.4355000000000002e-05, "loss": 0.0049, "step": 7138 }, { "epoch": 9.34, "learning_rate": 1.435e-05, "loss": 0.005, "step": 7139 }, { "epoch": 9.35, "learning_rate": 1.4345e-05, "loss": 0.0111, "step": 7140 }, { "epoch": 9.35, "learning_rate": 1.434e-05, "loss": 0.0044, "step": 7141 }, { "epoch": 9.35, "learning_rate": 1.4335e-05, "loss": 0.0058, "step": 7142 }, { "epoch": 9.35, "learning_rate": 1.4330000000000002e-05, "loss": 0.0081, "step": 7143 }, { "epoch": 9.35, "learning_rate": 1.4325e-05, "loss": 0.0039, "step": 7144 }, { "epoch": 9.35, "learning_rate": 1.432e-05, "loss": 0.0067, "step": 7145 }, { "epoch": 9.35, "learning_rate": 1.4315000000000001e-05, "loss": 0.0067, "step": 7146 }, { "epoch": 9.35, "learning_rate": 1.4310000000000002e-05, "loss": 0.004, "step": 7147 }, { "epoch": 9.36, "learning_rate": 1.4305000000000001e-05, "loss": 0.0049, "step": 7148 }, { "epoch": 9.36, "learning_rate": 1.43e-05, "loss": 0.0047, "step": 7149 }, { "epoch": 9.36, "learning_rate": 1.4295e-05, "loss": 0.0067, "step": 7150 }, { "epoch": 9.36, "learning_rate": 1.429e-05, "loss": 0.0046, "step": 7151 }, { "epoch": 9.36, "learning_rate": 1.4285000000000001e-05, "loss": 0.0038, "step": 7152 }, { "epoch": 9.36, "learning_rate": 1.4280000000000002e-05, "loss": 0.0062, "step": 7153 }, { "epoch": 9.36, "learning_rate": 1.4275e-05, "loss": 0.0051, "step": 7154 }, { "epoch": 9.37, "learning_rate": 1.427e-05, "loss": 0.0046, "step": 7155 }, { "epoch": 9.37, "learning_rate": 1.4265e-05, "loss": 0.0056, "step": 7156 }, { "epoch": 9.37, "learning_rate": 1.426e-05, "loss": 0.0054, "step": 7157 }, { "epoch": 9.37, "learning_rate": 1.4255000000000002e-05, "loss": 0.006, "step": 7158 }, { "epoch": 9.37, "learning_rate": 1.4249999999999999e-05, "loss": 0.0069, "step": 7159 }, { "epoch": 9.37, "learning_rate": 1.4245e-05, "loss": 0.0054, "step": 7160 }, { "epoch": 9.37, "learning_rate": 1.4240000000000001e-05, "loss": 0.0068, "step": 7161 }, { "epoch": 9.37, "learning_rate": 1.4235000000000002e-05, "loss": 0.0053, "step": 7162 }, { "epoch": 9.38, "learning_rate": 1.4230000000000001e-05, "loss": 0.0048, "step": 7163 }, { "epoch": 9.38, "learning_rate": 1.4225e-05, "loss": 0.0049, "step": 7164 }, { "epoch": 9.38, "learning_rate": 1.422e-05, "loss": 0.0074, "step": 7165 }, { "epoch": 9.38, "learning_rate": 1.4215e-05, "loss": 0.0054, "step": 7166 }, { "epoch": 9.38, "learning_rate": 1.4210000000000001e-05, "loss": 0.0058, "step": 7167 }, { "epoch": 9.38, "learning_rate": 1.4205000000000002e-05, "loss": 0.0048, "step": 7168 }, { "epoch": 9.38, "learning_rate": 1.42e-05, "loss": 0.0058, "step": 7169 }, { "epoch": 9.38, "learning_rate": 1.4195e-05, "loss": 0.0047, "step": 7170 }, { "epoch": 9.39, "learning_rate": 1.4190000000000001e-05, "loss": 0.0049, "step": 7171 }, { "epoch": 9.39, "learning_rate": 1.4185e-05, "loss": 0.0071, "step": 7172 }, { "epoch": 9.39, "learning_rate": 1.4180000000000001e-05, "loss": 0.0047, "step": 7173 }, { "epoch": 9.39, "learning_rate": 1.4174999999999999e-05, "loss": 0.0057, "step": 7174 }, { "epoch": 9.39, "learning_rate": 1.417e-05, "loss": 0.0071, "step": 7175 }, { "epoch": 9.39, "learning_rate": 1.4165e-05, "loss": 0.0063, "step": 7176 }, { "epoch": 9.39, "learning_rate": 1.4160000000000002e-05, "loss": 0.0071, "step": 7177 }, { "epoch": 9.4, "learning_rate": 1.4155000000000001e-05, "loss": 0.0045, "step": 7178 }, { "epoch": 9.4, "learning_rate": 1.415e-05, "loss": 0.0047, "step": 7179 }, { "epoch": 9.4, "learning_rate": 1.4145e-05, "loss": 0.0057, "step": 7180 }, { "epoch": 9.4, "learning_rate": 1.414e-05, "loss": 0.0042, "step": 7181 }, { "epoch": 9.4, "learning_rate": 1.4135000000000001e-05, "loss": 0.0054, "step": 7182 }, { "epoch": 9.4, "learning_rate": 1.4130000000000002e-05, "loss": 0.0034, "step": 7183 }, { "epoch": 9.4, "learning_rate": 1.4125e-05, "loss": 0.0048, "step": 7184 }, { "epoch": 9.4, "learning_rate": 1.412e-05, "loss": 0.0052, "step": 7185 }, { "epoch": 9.41, "learning_rate": 1.4115000000000001e-05, "loss": 0.0044, "step": 7186 }, { "epoch": 9.41, "learning_rate": 1.411e-05, "loss": 0.0053, "step": 7187 }, { "epoch": 9.41, "learning_rate": 1.4105000000000001e-05, "loss": 0.0322, "step": 7188 }, { "epoch": 9.41, "learning_rate": 1.4099999999999999e-05, "loss": 0.0044, "step": 7189 }, { "epoch": 9.41, "learning_rate": 1.4095e-05, "loss": 0.0053, "step": 7190 }, { "epoch": 9.41, "learning_rate": 1.409e-05, "loss": 0.0044, "step": 7191 }, { "epoch": 9.41, "learning_rate": 1.4085000000000002e-05, "loss": 0.0039, "step": 7192 }, { "epoch": 9.41, "learning_rate": 1.408e-05, "loss": 0.0047, "step": 7193 }, { "epoch": 9.42, "learning_rate": 1.4075e-05, "loss": 0.0052, "step": 7194 }, { "epoch": 9.42, "learning_rate": 1.4069999999999999e-05, "loss": 0.0054, "step": 7195 }, { "epoch": 9.42, "learning_rate": 1.4065e-05, "loss": 0.0057, "step": 7196 }, { "epoch": 9.42, "learning_rate": 1.4060000000000001e-05, "loss": 0.0056, "step": 7197 }, { "epoch": 9.42, "learning_rate": 1.4055000000000002e-05, "loss": 0.0052, "step": 7198 }, { "epoch": 9.42, "learning_rate": 1.4050000000000003e-05, "loss": 0.0056, "step": 7199 }, { "epoch": 9.42, "learning_rate": 1.4045e-05, "loss": 0.0051, "step": 7200 }, { "epoch": 9.43, "learning_rate": 1.4040000000000001e-05, "loss": 0.0062, "step": 7201 }, { "epoch": 9.43, "learning_rate": 1.4035e-05, "loss": 0.0072, "step": 7202 }, { "epoch": 9.43, "learning_rate": 1.4030000000000001e-05, "loss": 0.0039, "step": 7203 }, { "epoch": 9.43, "learning_rate": 1.4025000000000002e-05, "loss": 0.0051, "step": 7204 }, { "epoch": 9.43, "learning_rate": 1.402e-05, "loss": 0.0042, "step": 7205 }, { "epoch": 9.43, "learning_rate": 1.4015e-05, "loss": 0.0067, "step": 7206 }, { "epoch": 9.43, "learning_rate": 1.4010000000000001e-05, "loss": 0.007, "step": 7207 }, { "epoch": 9.43, "learning_rate": 1.4005000000000002e-05, "loss": 0.005, "step": 7208 }, { "epoch": 9.44, "learning_rate": 1.4000000000000001e-05, "loss": 0.0062, "step": 7209 }, { "epoch": 9.44, "learning_rate": 1.3994999999999999e-05, "loss": 0.0048, "step": 7210 }, { "epoch": 9.44, "learning_rate": 1.399e-05, "loss": 0.0057, "step": 7211 }, { "epoch": 9.44, "learning_rate": 1.3985e-05, "loss": 0.0052, "step": 7212 }, { "epoch": 9.44, "learning_rate": 1.3980000000000002e-05, "loss": 0.0055, "step": 7213 }, { "epoch": 9.44, "learning_rate": 1.3975000000000003e-05, "loss": 0.0062, "step": 7214 }, { "epoch": 9.44, "learning_rate": 1.397e-05, "loss": 0.0056, "step": 7215 }, { "epoch": 9.45, "learning_rate": 1.3965000000000001e-05, "loss": 0.0058, "step": 7216 }, { "epoch": 9.45, "learning_rate": 1.396e-05, "loss": 0.0052, "step": 7217 }, { "epoch": 9.45, "learning_rate": 1.3955000000000001e-05, "loss": 0.0054, "step": 7218 }, { "epoch": 9.45, "learning_rate": 1.3950000000000002e-05, "loss": 0.0053, "step": 7219 }, { "epoch": 9.45, "learning_rate": 1.3945e-05, "loss": 0.0049, "step": 7220 }, { "epoch": 9.45, "learning_rate": 1.394e-05, "loss": 0.0064, "step": 7221 }, { "epoch": 9.45, "learning_rate": 1.3935000000000001e-05, "loss": 0.0053, "step": 7222 }, { "epoch": 9.45, "learning_rate": 1.3930000000000002e-05, "loss": 0.0046, "step": 7223 }, { "epoch": 9.46, "learning_rate": 1.3925000000000001e-05, "loss": 0.0049, "step": 7224 }, { "epoch": 9.46, "learning_rate": 1.3919999999999999e-05, "loss": 0.005, "step": 7225 }, { "epoch": 9.46, "learning_rate": 1.3915e-05, "loss": 0.0049, "step": 7226 }, { "epoch": 9.46, "learning_rate": 1.391e-05, "loss": 0.0053, "step": 7227 }, { "epoch": 9.46, "learning_rate": 1.3905000000000002e-05, "loss": 0.0076, "step": 7228 }, { "epoch": 9.46, "learning_rate": 1.3900000000000002e-05, "loss": 0.0044, "step": 7229 }, { "epoch": 9.46, "learning_rate": 1.3895e-05, "loss": 0.0045, "step": 7230 }, { "epoch": 9.46, "learning_rate": 1.389e-05, "loss": 0.0047, "step": 7231 }, { "epoch": 9.47, "learning_rate": 1.3885e-05, "loss": 0.0046, "step": 7232 }, { "epoch": 9.47, "learning_rate": 1.3880000000000001e-05, "loss": 0.0055, "step": 7233 }, { "epoch": 9.47, "learning_rate": 1.3875000000000002e-05, "loss": 0.0056, "step": 7234 }, { "epoch": 9.47, "learning_rate": 1.387e-05, "loss": 0.012, "step": 7235 }, { "epoch": 9.47, "learning_rate": 1.3865e-05, "loss": 0.0045, "step": 7236 }, { "epoch": 9.47, "learning_rate": 1.3860000000000001e-05, "loss": 0.0083, "step": 7237 }, { "epoch": 9.47, "learning_rate": 1.3855000000000002e-05, "loss": 0.0044, "step": 7238 }, { "epoch": 9.48, "learning_rate": 1.3850000000000001e-05, "loss": 0.0071, "step": 7239 }, { "epoch": 9.48, "learning_rate": 1.3845e-05, "loss": 0.0069, "step": 7240 }, { "epoch": 9.48, "learning_rate": 1.384e-05, "loss": 0.006, "step": 7241 }, { "epoch": 9.48, "learning_rate": 1.3835e-05, "loss": 0.0054, "step": 7242 }, { "epoch": 9.48, "learning_rate": 1.3830000000000001e-05, "loss": 0.0047, "step": 7243 }, { "epoch": 9.48, "learning_rate": 1.3825000000000002e-05, "loss": 0.0063, "step": 7244 }, { "epoch": 9.48, "learning_rate": 1.382e-05, "loss": 0.0073, "step": 7245 }, { "epoch": 9.48, "learning_rate": 1.3815e-05, "loss": 0.0054, "step": 7246 }, { "epoch": 9.49, "learning_rate": 1.381e-05, "loss": 0.0038, "step": 7247 }, { "epoch": 9.49, "learning_rate": 1.3805e-05, "loss": 0.0096, "step": 7248 }, { "epoch": 9.49, "learning_rate": 1.3800000000000002e-05, "loss": 0.0035, "step": 7249 }, { "epoch": 9.49, "learning_rate": 1.3795e-05, "loss": 0.0053, "step": 7250 }, { "epoch": 9.49, "learning_rate": 1.379e-05, "loss": 0.0051, "step": 7251 }, { "epoch": 9.49, "learning_rate": 1.3785000000000001e-05, "loss": 0.0054, "step": 7252 }, { "epoch": 9.49, "learning_rate": 1.3780000000000002e-05, "loss": 0.0053, "step": 7253 }, { "epoch": 9.49, "learning_rate": 1.3775000000000001e-05, "loss": 0.0053, "step": 7254 }, { "epoch": 9.5, "learning_rate": 1.377e-05, "loss": 0.0056, "step": 7255 }, { "epoch": 9.5, "learning_rate": 1.3765e-05, "loss": 0.004, "step": 7256 }, { "epoch": 9.5, "learning_rate": 1.376e-05, "loss": 0.0069, "step": 7257 }, { "epoch": 9.5, "learning_rate": 1.3755000000000001e-05, "loss": 0.0064, "step": 7258 }, { "epoch": 9.5, "learning_rate": 1.3750000000000002e-05, "loss": 0.004, "step": 7259 }, { "epoch": 9.5, "learning_rate": 1.3745e-05, "loss": 0.0048, "step": 7260 }, { "epoch": 9.5, "learning_rate": 1.374e-05, "loss": 0.0053, "step": 7261 }, { "epoch": 9.51, "learning_rate": 1.3735000000000001e-05, "loss": 0.0044, "step": 7262 }, { "epoch": 9.51, "learning_rate": 1.373e-05, "loss": 0.0049, "step": 7263 }, { "epoch": 9.51, "learning_rate": 1.3725000000000002e-05, "loss": 0.0052, "step": 7264 }, { "epoch": 9.51, "learning_rate": 1.3719999999999999e-05, "loss": 0.0035, "step": 7265 }, { "epoch": 9.51, "learning_rate": 1.3715e-05, "loss": 0.0051, "step": 7266 }, { "epoch": 9.51, "learning_rate": 1.3710000000000001e-05, "loss": 0.0055, "step": 7267 }, { "epoch": 9.51, "learning_rate": 1.3705000000000002e-05, "loss": 0.006, "step": 7268 }, { "epoch": 9.51, "learning_rate": 1.3700000000000001e-05, "loss": 0.0045, "step": 7269 }, { "epoch": 9.52, "learning_rate": 1.3695e-05, "loss": 0.0046, "step": 7270 }, { "epoch": 9.52, "learning_rate": 1.369e-05, "loss": 0.0047, "step": 7271 }, { "epoch": 9.52, "learning_rate": 1.3685e-05, "loss": 0.0045, "step": 7272 }, { "epoch": 9.52, "learning_rate": 1.3680000000000001e-05, "loss": 0.0045, "step": 7273 }, { "epoch": 9.52, "learning_rate": 1.3675000000000002e-05, "loss": 0.0056, "step": 7274 }, { "epoch": 9.52, "learning_rate": 1.367e-05, "loss": 0.005, "step": 7275 }, { "epoch": 9.52, "learning_rate": 1.3665e-05, "loss": 0.0076, "step": 7276 }, { "epoch": 9.52, "learning_rate": 1.3660000000000001e-05, "loss": 0.0077, "step": 7277 }, { "epoch": 9.53, "learning_rate": 1.3655e-05, "loss": 0.0062, "step": 7278 }, { "epoch": 9.53, "learning_rate": 1.3650000000000001e-05, "loss": 0.0054, "step": 7279 }, { "epoch": 9.53, "learning_rate": 1.3644999999999999e-05, "loss": 0.009, "step": 7280 }, { "epoch": 9.53, "learning_rate": 1.364e-05, "loss": 0.0042, "step": 7281 }, { "epoch": 9.53, "learning_rate": 1.3635e-05, "loss": 0.0039, "step": 7282 }, { "epoch": 9.53, "learning_rate": 1.3630000000000002e-05, "loss": 0.0063, "step": 7283 }, { "epoch": 9.53, "learning_rate": 1.3625e-05, "loss": 0.0048, "step": 7284 }, { "epoch": 9.54, "learning_rate": 1.362e-05, "loss": 0.0064, "step": 7285 }, { "epoch": 9.54, "learning_rate": 1.3615e-05, "loss": 0.0072, "step": 7286 }, { "epoch": 9.54, "learning_rate": 1.361e-05, "loss": 0.0046, "step": 7287 }, { "epoch": 9.54, "learning_rate": 1.3605000000000001e-05, "loss": 0.0069, "step": 7288 }, { "epoch": 9.54, "learning_rate": 1.3600000000000002e-05, "loss": 0.0056, "step": 7289 }, { "epoch": 9.54, "learning_rate": 1.3595e-05, "loss": 0.0049, "step": 7290 }, { "epoch": 9.54, "learning_rate": 1.359e-05, "loss": 0.0041, "step": 7291 }, { "epoch": 9.54, "learning_rate": 1.3585000000000001e-05, "loss": 0.0071, "step": 7292 }, { "epoch": 9.55, "learning_rate": 1.358e-05, "loss": 0.0046, "step": 7293 }, { "epoch": 9.55, "learning_rate": 1.3575000000000001e-05, "loss": 0.0049, "step": 7294 }, { "epoch": 9.55, "learning_rate": 1.3569999999999999e-05, "loss": 0.0067, "step": 7295 }, { "epoch": 9.55, "learning_rate": 1.3565e-05, "loss": 0.0071, "step": 7296 }, { "epoch": 9.55, "learning_rate": 1.356e-05, "loss": 0.0065, "step": 7297 }, { "epoch": 9.55, "learning_rate": 1.3555000000000002e-05, "loss": 0.0068, "step": 7298 }, { "epoch": 9.55, "learning_rate": 1.3550000000000002e-05, "loss": 0.0045, "step": 7299 }, { "epoch": 9.55, "learning_rate": 1.3545e-05, "loss": 0.0074, "step": 7300 }, { "epoch": 9.56, "learning_rate": 1.3539999999999999e-05, "loss": 0.0055, "step": 7301 }, { "epoch": 9.56, "learning_rate": 1.3535e-05, "loss": 0.0054, "step": 7302 }, { "epoch": 9.56, "learning_rate": 1.3530000000000001e-05, "loss": 0.0095, "step": 7303 }, { "epoch": 9.56, "learning_rate": 1.3525000000000002e-05, "loss": 0.006, "step": 7304 }, { "epoch": 9.56, "learning_rate": 1.352e-05, "loss": 0.005, "step": 7305 }, { "epoch": 9.56, "learning_rate": 1.3515e-05, "loss": 0.0073, "step": 7306 }, { "epoch": 9.56, "learning_rate": 1.3510000000000001e-05, "loss": 0.0069, "step": 7307 }, { "epoch": 9.57, "learning_rate": 1.3505e-05, "loss": 0.0063, "step": 7308 }, { "epoch": 9.57, "learning_rate": 1.3500000000000001e-05, "loss": 0.0057, "step": 7309 }, { "epoch": 9.57, "learning_rate": 1.3494999999999999e-05, "loss": 0.0057, "step": 7310 }, { "epoch": 9.57, "learning_rate": 1.349e-05, "loss": 0.0037, "step": 7311 }, { "epoch": 9.57, "learning_rate": 1.3485e-05, "loss": 0.0047, "step": 7312 }, { "epoch": 9.57, "learning_rate": 1.3480000000000001e-05, "loss": 0.0052, "step": 7313 }, { "epoch": 9.57, "learning_rate": 1.3475000000000002e-05, "loss": 0.0045, "step": 7314 }, { "epoch": 9.57, "learning_rate": 1.347e-05, "loss": 0.0066, "step": 7315 }, { "epoch": 9.58, "learning_rate": 1.3465e-05, "loss": 0.0054, "step": 7316 }, { "epoch": 9.58, "learning_rate": 1.346e-05, "loss": 0.0063, "step": 7317 }, { "epoch": 9.58, "learning_rate": 1.3455e-05, "loss": 0.007, "step": 7318 }, { "epoch": 9.58, "learning_rate": 1.3450000000000002e-05, "loss": 0.0062, "step": 7319 }, { "epoch": 9.58, "learning_rate": 1.3445e-05, "loss": 0.0063, "step": 7320 }, { "epoch": 9.58, "learning_rate": 1.344e-05, "loss": 0.0055, "step": 7321 }, { "epoch": 9.58, "learning_rate": 1.3435000000000001e-05, "loss": 0.0063, "step": 7322 }, { "epoch": 9.59, "learning_rate": 1.343e-05, "loss": 0.0052, "step": 7323 }, { "epoch": 9.59, "learning_rate": 1.3425000000000001e-05, "loss": 0.0041, "step": 7324 }, { "epoch": 9.59, "learning_rate": 1.3420000000000002e-05, "loss": 0.0039, "step": 7325 }, { "epoch": 9.59, "learning_rate": 1.3415e-05, "loss": 0.0049, "step": 7326 }, { "epoch": 9.59, "learning_rate": 1.341e-05, "loss": 0.0056, "step": 7327 }, { "epoch": 9.59, "learning_rate": 1.3405000000000001e-05, "loss": 0.0044, "step": 7328 }, { "epoch": 9.59, "learning_rate": 1.3400000000000002e-05, "loss": 0.0059, "step": 7329 }, { "epoch": 9.59, "learning_rate": 1.3395000000000001e-05, "loss": 0.0065, "step": 7330 }, { "epoch": 9.6, "learning_rate": 1.339e-05, "loss": 0.0047, "step": 7331 }, { "epoch": 9.6, "learning_rate": 1.3385e-05, "loss": 0.0041, "step": 7332 }, { "epoch": 9.6, "learning_rate": 1.338e-05, "loss": 0.0075, "step": 7333 }, { "epoch": 9.6, "learning_rate": 1.3375000000000002e-05, "loss": 0.0083, "step": 7334 }, { "epoch": 9.6, "learning_rate": 1.3370000000000002e-05, "loss": 0.0058, "step": 7335 }, { "epoch": 9.6, "learning_rate": 1.3365e-05, "loss": 0.0061, "step": 7336 }, { "epoch": 9.6, "learning_rate": 1.336e-05, "loss": 0.0041, "step": 7337 }, { "epoch": 9.6, "learning_rate": 1.3355e-05, "loss": 0.005, "step": 7338 }, { "epoch": 9.61, "learning_rate": 1.3350000000000001e-05, "loss": 0.0059, "step": 7339 }, { "epoch": 9.61, "learning_rate": 1.3345000000000002e-05, "loss": 0.0047, "step": 7340 }, { "epoch": 9.61, "learning_rate": 1.334e-05, "loss": 0.0063, "step": 7341 }, { "epoch": 9.61, "learning_rate": 1.3335e-05, "loss": 0.0036, "step": 7342 }, { "epoch": 9.61, "learning_rate": 1.3330000000000001e-05, "loss": 0.0056, "step": 7343 }, { "epoch": 9.61, "learning_rate": 1.3325000000000002e-05, "loss": 0.0068, "step": 7344 }, { "epoch": 9.61, "learning_rate": 1.3320000000000001e-05, "loss": 0.006, "step": 7345 }, { "epoch": 9.62, "learning_rate": 1.3315e-05, "loss": 0.0068, "step": 7346 }, { "epoch": 9.62, "learning_rate": 1.331e-05, "loss": 0.0036, "step": 7347 }, { "epoch": 9.62, "learning_rate": 1.3305e-05, "loss": 0.0063, "step": 7348 }, { "epoch": 9.62, "learning_rate": 1.3300000000000001e-05, "loss": 0.007, "step": 7349 }, { "epoch": 9.62, "learning_rate": 1.3295000000000002e-05, "loss": 0.0042, "step": 7350 }, { "epoch": 9.62, "learning_rate": 1.329e-05, "loss": 0.0062, "step": 7351 }, { "epoch": 9.62, "learning_rate": 1.3285e-05, "loss": 0.0059, "step": 7352 }, { "epoch": 9.62, "learning_rate": 1.3280000000000002e-05, "loss": 0.0061, "step": 7353 }, { "epoch": 9.63, "learning_rate": 1.3275e-05, "loss": 0.0045, "step": 7354 }, { "epoch": 9.63, "learning_rate": 1.3270000000000002e-05, "loss": 0.0057, "step": 7355 }, { "epoch": 9.63, "learning_rate": 1.3265e-05, "loss": 0.005, "step": 7356 }, { "epoch": 9.63, "learning_rate": 1.326e-05, "loss": 0.0058, "step": 7357 }, { "epoch": 9.63, "learning_rate": 1.3255000000000001e-05, "loss": 0.0044, "step": 7358 }, { "epoch": 9.63, "learning_rate": 1.3250000000000002e-05, "loss": 0.0041, "step": 7359 }, { "epoch": 9.63, "learning_rate": 1.3245000000000001e-05, "loss": 0.0075, "step": 7360 }, { "epoch": 9.63, "learning_rate": 1.324e-05, "loss": 0.0042, "step": 7361 }, { "epoch": 9.64, "learning_rate": 1.3235e-05, "loss": 0.0054, "step": 7362 }, { "epoch": 9.64, "learning_rate": 1.323e-05, "loss": 0.0046, "step": 7363 }, { "epoch": 9.64, "learning_rate": 1.3225000000000001e-05, "loss": 0.0061, "step": 7364 }, { "epoch": 9.64, "learning_rate": 1.3220000000000002e-05, "loss": 0.007, "step": 7365 }, { "epoch": 9.64, "learning_rate": 1.3215e-05, "loss": 0.0054, "step": 7366 }, { "epoch": 9.64, "learning_rate": 1.321e-05, "loss": 0.0073, "step": 7367 }, { "epoch": 9.64, "learning_rate": 1.3205000000000001e-05, "loss": 0.0051, "step": 7368 }, { "epoch": 9.65, "learning_rate": 1.32e-05, "loss": 0.0038, "step": 7369 }, { "epoch": 9.65, "learning_rate": 1.3195000000000002e-05, "loss": 0.0063, "step": 7370 }, { "epoch": 9.65, "learning_rate": 1.3189999999999999e-05, "loss": 0.0047, "step": 7371 }, { "epoch": 9.65, "learning_rate": 1.3185e-05, "loss": 0.0032, "step": 7372 }, { "epoch": 9.65, "learning_rate": 1.3180000000000001e-05, "loss": 0.0046, "step": 7373 }, { "epoch": 9.65, "learning_rate": 1.3175000000000002e-05, "loss": 0.0104, "step": 7374 }, { "epoch": 9.65, "learning_rate": 1.3170000000000001e-05, "loss": 0.0053, "step": 7375 }, { "epoch": 9.65, "learning_rate": 1.3165e-05, "loss": 0.0047, "step": 7376 }, { "epoch": 9.66, "learning_rate": 1.316e-05, "loss": 0.0049, "step": 7377 }, { "epoch": 9.66, "learning_rate": 1.3155e-05, "loss": 0.0073, "step": 7378 }, { "epoch": 9.66, "learning_rate": 1.3150000000000001e-05, "loss": 0.0047, "step": 7379 }, { "epoch": 9.66, "learning_rate": 1.3145000000000002e-05, "loss": 0.0057, "step": 7380 }, { "epoch": 9.66, "learning_rate": 1.314e-05, "loss": 0.0048, "step": 7381 }, { "epoch": 9.66, "learning_rate": 1.3135e-05, "loss": 0.004, "step": 7382 }, { "epoch": 9.66, "learning_rate": 1.3130000000000001e-05, "loss": 0.0062, "step": 7383 }, { "epoch": 9.66, "learning_rate": 1.3125e-05, "loss": 0.0058, "step": 7384 }, { "epoch": 9.67, "learning_rate": 1.3120000000000001e-05, "loss": 0.0073, "step": 7385 }, { "epoch": 9.67, "learning_rate": 1.3114999999999999e-05, "loss": 0.006, "step": 7386 }, { "epoch": 9.67, "learning_rate": 1.311e-05, "loss": 0.0062, "step": 7387 }, { "epoch": 9.67, "learning_rate": 1.3105e-05, "loss": 0.007, "step": 7388 }, { "epoch": 9.67, "learning_rate": 1.3100000000000002e-05, "loss": 0.0042, "step": 7389 }, { "epoch": 9.67, "learning_rate": 1.3095000000000003e-05, "loss": 0.004, "step": 7390 }, { "epoch": 9.67, "learning_rate": 1.309e-05, "loss": 0.0068, "step": 7391 }, { "epoch": 9.68, "learning_rate": 1.3085e-05, "loss": 0.0047, "step": 7392 }, { "epoch": 9.68, "learning_rate": 1.308e-05, "loss": 0.0047, "step": 7393 }, { "epoch": 9.68, "learning_rate": 1.3075000000000001e-05, "loss": 0.0064, "step": 7394 }, { "epoch": 9.68, "learning_rate": 1.3070000000000002e-05, "loss": 0.0041, "step": 7395 }, { "epoch": 9.68, "learning_rate": 1.3065e-05, "loss": 0.005, "step": 7396 }, { "epoch": 9.68, "learning_rate": 1.306e-05, "loss": 0.0031, "step": 7397 }, { "epoch": 9.68, "learning_rate": 1.3055000000000001e-05, "loss": 0.0045, "step": 7398 }, { "epoch": 9.68, "learning_rate": 1.305e-05, "loss": 0.0067, "step": 7399 }, { "epoch": 9.69, "learning_rate": 1.3045000000000001e-05, "loss": 0.0055, "step": 7400 }, { "epoch": 9.69, "learning_rate": 1.3039999999999999e-05, "loss": 0.0058, "step": 7401 }, { "epoch": 9.69, "learning_rate": 1.3035e-05, "loss": 0.0043, "step": 7402 }, { "epoch": 9.69, "learning_rate": 1.303e-05, "loss": 0.0062, "step": 7403 }, { "epoch": 9.69, "learning_rate": 1.3025000000000002e-05, "loss": 0.0043, "step": 7404 }, { "epoch": 9.69, "learning_rate": 1.3020000000000002e-05, "loss": 0.0057, "step": 7405 }, { "epoch": 9.69, "learning_rate": 1.3015e-05, "loss": 0.0039, "step": 7406 }, { "epoch": 9.7, "learning_rate": 1.301e-05, "loss": 0.0037, "step": 7407 }, { "epoch": 9.7, "learning_rate": 1.3005e-05, "loss": 0.005, "step": 7408 }, { "epoch": 9.7, "learning_rate": 1.3000000000000001e-05, "loss": 0.0078, "step": 7409 }, { "epoch": 9.7, "learning_rate": 1.2995000000000002e-05, "loss": 0.0052, "step": 7410 }, { "epoch": 9.7, "learning_rate": 1.299e-05, "loss": 0.0054, "step": 7411 }, { "epoch": 9.7, "learning_rate": 1.2985e-05, "loss": 0.0039, "step": 7412 }, { "epoch": 9.7, "learning_rate": 1.2980000000000001e-05, "loss": 0.0061, "step": 7413 }, { "epoch": 9.7, "learning_rate": 1.2975e-05, "loss": 0.0051, "step": 7414 }, { "epoch": 9.71, "learning_rate": 1.2970000000000001e-05, "loss": 0.0059, "step": 7415 }, { "epoch": 9.71, "learning_rate": 1.2964999999999999e-05, "loss": 0.0045, "step": 7416 }, { "epoch": 9.71, "learning_rate": 1.296e-05, "loss": 0.006, "step": 7417 }, { "epoch": 9.71, "learning_rate": 1.2955e-05, "loss": 0.0047, "step": 7418 }, { "epoch": 9.71, "learning_rate": 1.2950000000000001e-05, "loss": 0.0043, "step": 7419 }, { "epoch": 9.71, "learning_rate": 1.2945000000000002e-05, "loss": 0.0054, "step": 7420 }, { "epoch": 9.71, "learning_rate": 1.294e-05, "loss": 0.0053, "step": 7421 }, { "epoch": 9.71, "learning_rate": 1.2935e-05, "loss": 0.0054, "step": 7422 }, { "epoch": 9.72, "learning_rate": 1.293e-05, "loss": 0.0051, "step": 7423 }, { "epoch": 9.72, "learning_rate": 1.2925e-05, "loss": 0.0092, "step": 7424 }, { "epoch": 9.72, "learning_rate": 1.2920000000000002e-05, "loss": 0.0056, "step": 7425 }, { "epoch": 9.72, "learning_rate": 1.2915e-05, "loss": 0.0043, "step": 7426 }, { "epoch": 9.72, "learning_rate": 1.291e-05, "loss": 0.0054, "step": 7427 }, { "epoch": 9.72, "learning_rate": 1.2905000000000001e-05, "loss": 0.0076, "step": 7428 }, { "epoch": 9.72, "learning_rate": 1.29e-05, "loss": 0.006, "step": 7429 }, { "epoch": 9.73, "learning_rate": 1.2895000000000001e-05, "loss": 0.0042, "step": 7430 }, { "epoch": 9.73, "learning_rate": 1.2889999999999999e-05, "loss": 0.0053, "step": 7431 }, { "epoch": 9.73, "learning_rate": 1.2885e-05, "loss": 0.0061, "step": 7432 }, { "epoch": 9.73, "learning_rate": 1.288e-05, "loss": 0.0056, "step": 7433 }, { "epoch": 9.73, "learning_rate": 1.2875000000000001e-05, "loss": 0.006, "step": 7434 }, { "epoch": 9.73, "learning_rate": 1.2870000000000002e-05, "loss": 0.0041, "step": 7435 }, { "epoch": 9.73, "learning_rate": 1.2865e-05, "loss": 0.0058, "step": 7436 }, { "epoch": 9.73, "learning_rate": 1.286e-05, "loss": 0.0056, "step": 7437 }, { "epoch": 9.74, "learning_rate": 1.2855e-05, "loss": 0.0099, "step": 7438 }, { "epoch": 9.74, "learning_rate": 1.285e-05, "loss": 0.005, "step": 7439 }, { "epoch": 9.74, "learning_rate": 1.2845000000000002e-05, "loss": 0.0059, "step": 7440 }, { "epoch": 9.74, "learning_rate": 1.2839999999999999e-05, "loss": 0.0043, "step": 7441 }, { "epoch": 9.74, "learning_rate": 1.2835e-05, "loss": 0.0058, "step": 7442 }, { "epoch": 9.74, "learning_rate": 1.283e-05, "loss": 0.0071, "step": 7443 }, { "epoch": 9.74, "learning_rate": 1.2825000000000002e-05, "loss": 0.0052, "step": 7444 }, { "epoch": 9.74, "learning_rate": 1.2820000000000001e-05, "loss": 0.0051, "step": 7445 }, { "epoch": 9.75, "learning_rate": 1.2814999999999998e-05, "loss": 0.0058, "step": 7446 }, { "epoch": 9.75, "learning_rate": 1.281e-05, "loss": 0.0059, "step": 7447 }, { "epoch": 9.75, "learning_rate": 1.2805e-05, "loss": 0.0052, "step": 7448 }, { "epoch": 9.75, "learning_rate": 1.2800000000000001e-05, "loss": 0.0064, "step": 7449 }, { "epoch": 9.75, "learning_rate": 1.2795000000000002e-05, "loss": 0.0056, "step": 7450 }, { "epoch": 9.75, "learning_rate": 1.2790000000000001e-05, "loss": 0.0039, "step": 7451 }, { "epoch": 9.75, "learning_rate": 1.2785e-05, "loss": 0.0062, "step": 7452 }, { "epoch": 9.76, "learning_rate": 1.278e-05, "loss": 0.005, "step": 7453 }, { "epoch": 9.76, "learning_rate": 1.2775e-05, "loss": 0.0056, "step": 7454 }, { "epoch": 9.76, "learning_rate": 1.2770000000000001e-05, "loss": 0.0039, "step": 7455 }, { "epoch": 9.76, "learning_rate": 1.2765000000000002e-05, "loss": 0.0052, "step": 7456 }, { "epoch": 9.76, "learning_rate": 1.276e-05, "loss": 0.0074, "step": 7457 }, { "epoch": 9.76, "learning_rate": 1.2755e-05, "loss": 0.0045, "step": 7458 }, { "epoch": 9.76, "learning_rate": 1.2750000000000002e-05, "loss": 0.0057, "step": 7459 }, { "epoch": 9.76, "learning_rate": 1.2745e-05, "loss": 0.0042, "step": 7460 }, { "epoch": 9.77, "learning_rate": 1.2740000000000002e-05, "loss": 0.0058, "step": 7461 }, { "epoch": 9.77, "learning_rate": 1.2735e-05, "loss": 0.0054, "step": 7462 }, { "epoch": 9.77, "learning_rate": 1.273e-05, "loss": 0.0052, "step": 7463 }, { "epoch": 9.77, "learning_rate": 1.2725000000000001e-05, "loss": 0.006, "step": 7464 }, { "epoch": 9.77, "learning_rate": 1.2720000000000002e-05, "loss": 0.0055, "step": 7465 }, { "epoch": 9.77, "learning_rate": 1.2715000000000001e-05, "loss": 0.0056, "step": 7466 }, { "epoch": 9.77, "learning_rate": 1.271e-05, "loss": 0.0074, "step": 7467 }, { "epoch": 9.77, "learning_rate": 1.2705e-05, "loss": 0.0059, "step": 7468 }, { "epoch": 9.78, "learning_rate": 1.27e-05, "loss": 0.0086, "step": 7469 }, { "epoch": 9.78, "learning_rate": 1.2695000000000001e-05, "loss": 0.0054, "step": 7470 }, { "epoch": 9.78, "learning_rate": 1.2690000000000002e-05, "loss": 0.0047, "step": 7471 }, { "epoch": 9.78, "learning_rate": 1.2685e-05, "loss": 0.0049, "step": 7472 }, { "epoch": 9.78, "learning_rate": 1.268e-05, "loss": 0.0068, "step": 7473 }, { "epoch": 9.78, "learning_rate": 1.2675000000000001e-05, "loss": 0.0077, "step": 7474 }, { "epoch": 9.78, "learning_rate": 1.267e-05, "loss": 0.0054, "step": 7475 }, { "epoch": 9.79, "learning_rate": 1.2665000000000002e-05, "loss": 0.007, "step": 7476 }, { "epoch": 9.79, "learning_rate": 1.2659999999999999e-05, "loss": 0.0048, "step": 7477 }, { "epoch": 9.79, "learning_rate": 1.2655e-05, "loss": 0.0063, "step": 7478 }, { "epoch": 9.79, "learning_rate": 1.2650000000000001e-05, "loss": 0.0054, "step": 7479 }, { "epoch": 9.79, "learning_rate": 1.2645000000000002e-05, "loss": 0.0037, "step": 7480 }, { "epoch": 9.79, "learning_rate": 1.2640000000000003e-05, "loss": 0.0052, "step": 7481 }, { "epoch": 9.79, "learning_rate": 1.2635e-05, "loss": 0.0045, "step": 7482 }, { "epoch": 9.79, "learning_rate": 1.263e-05, "loss": 0.0076, "step": 7483 }, { "epoch": 9.8, "learning_rate": 1.2625e-05, "loss": 0.0053, "step": 7484 }, { "epoch": 9.8, "learning_rate": 1.2620000000000001e-05, "loss": 0.007, "step": 7485 }, { "epoch": 9.8, "learning_rate": 1.2615000000000002e-05, "loss": 0.0051, "step": 7486 }, { "epoch": 9.8, "learning_rate": 1.261e-05, "loss": 0.0054, "step": 7487 }, { "epoch": 9.8, "learning_rate": 1.2605e-05, "loss": 0.0055, "step": 7488 }, { "epoch": 9.8, "learning_rate": 1.2600000000000001e-05, "loss": 0.0037, "step": 7489 }, { "epoch": 9.8, "learning_rate": 1.2595e-05, "loss": 0.0066, "step": 7490 }, { "epoch": 9.8, "learning_rate": 1.2590000000000001e-05, "loss": 0.0051, "step": 7491 }, { "epoch": 9.81, "learning_rate": 1.2584999999999999e-05, "loss": 0.0045, "step": 7492 }, { "epoch": 9.81, "learning_rate": 1.258e-05, "loss": 0.0058, "step": 7493 }, { "epoch": 9.81, "learning_rate": 1.2575e-05, "loss": 0.0046, "step": 7494 }, { "epoch": 9.81, "learning_rate": 1.2570000000000002e-05, "loss": 0.0052, "step": 7495 }, { "epoch": 9.81, "learning_rate": 1.2565000000000003e-05, "loss": 0.005, "step": 7496 }, { "epoch": 9.81, "learning_rate": 1.256e-05, "loss": 0.0042, "step": 7497 }, { "epoch": 9.81, "learning_rate": 1.2555000000000001e-05, "loss": 0.0053, "step": 7498 }, { "epoch": 9.82, "learning_rate": 1.255e-05, "loss": 0.0049, "step": 7499 }, { "epoch": 9.82, "learning_rate": 1.2545000000000001e-05, "loss": 0.0058, "step": 7500 }, { "epoch": 9.82, "learning_rate": 1.2540000000000002e-05, "loss": 0.0058, "step": 7501 }, { "epoch": 9.82, "learning_rate": 1.2535e-05, "loss": 0.0053, "step": 7502 }, { "epoch": 9.82, "learning_rate": 1.253e-05, "loss": 0.0074, "step": 7503 }, { "epoch": 9.82, "learning_rate": 1.2525000000000001e-05, "loss": 0.0046, "step": 7504 }, { "epoch": 9.82, "learning_rate": 1.252e-05, "loss": 0.0064, "step": 7505 }, { "epoch": 9.82, "learning_rate": 1.2515000000000001e-05, "loss": 0.0056, "step": 7506 }, { "epoch": 9.83, "learning_rate": 1.2509999999999999e-05, "loss": 0.0048, "step": 7507 }, { "epoch": 9.83, "learning_rate": 1.2505e-05, "loss": 0.0067, "step": 7508 }, { "epoch": 9.83, "learning_rate": 1.25e-05, "loss": 0.0051, "step": 7509 }, { "epoch": 9.83, "learning_rate": 1.2495000000000001e-05, "loss": 0.005, "step": 7510 }, { "epoch": 9.83, "learning_rate": 1.249e-05, "loss": 0.0064, "step": 7511 }, { "epoch": 9.83, "learning_rate": 1.2485000000000002e-05, "loss": 0.0046, "step": 7512 }, { "epoch": 9.83, "learning_rate": 1.248e-05, "loss": 0.0056, "step": 7513 }, { "epoch": 9.84, "learning_rate": 1.2475e-05, "loss": 0.0044, "step": 7514 }, { "epoch": 9.84, "learning_rate": 1.2470000000000001e-05, "loss": 0.0078, "step": 7515 }, { "epoch": 9.84, "learning_rate": 1.2465e-05, "loss": 0.0049, "step": 7516 }, { "epoch": 9.84, "learning_rate": 1.2460000000000001e-05, "loss": 0.0058, "step": 7517 }, { "epoch": 9.84, "learning_rate": 1.2455e-05, "loss": 0.0054, "step": 7518 }, { "epoch": 9.84, "learning_rate": 1.2450000000000001e-05, "loss": 0.009, "step": 7519 }, { "epoch": 9.84, "learning_rate": 1.2445e-05, "loss": 0.0078, "step": 7520 }, { "epoch": 9.84, "learning_rate": 1.244e-05, "loss": 0.0049, "step": 7521 }, { "epoch": 9.85, "learning_rate": 1.2435e-05, "loss": 0.0067, "step": 7522 }, { "epoch": 9.85, "learning_rate": 1.243e-05, "loss": 0.006, "step": 7523 }, { "epoch": 9.85, "learning_rate": 1.2425e-05, "loss": 0.0058, "step": 7524 }, { "epoch": 9.85, "learning_rate": 1.2420000000000001e-05, "loss": 0.0041, "step": 7525 }, { "epoch": 9.85, "learning_rate": 1.2415e-05, "loss": 0.0061, "step": 7526 }, { "epoch": 9.85, "learning_rate": 1.2410000000000001e-05, "loss": 0.0048, "step": 7527 }, { "epoch": 9.85, "learning_rate": 1.2405e-05, "loss": 0.0057, "step": 7528 }, { "epoch": 9.85, "learning_rate": 1.24e-05, "loss": 0.0051, "step": 7529 }, { "epoch": 9.86, "learning_rate": 1.2395e-05, "loss": 0.0057, "step": 7530 }, { "epoch": 9.86, "learning_rate": 1.239e-05, "loss": 0.0064, "step": 7531 }, { "epoch": 9.86, "learning_rate": 1.2385000000000001e-05, "loss": 0.0051, "step": 7532 }, { "epoch": 9.86, "learning_rate": 1.238e-05, "loss": 0.0047, "step": 7533 }, { "epoch": 9.86, "learning_rate": 1.2375000000000001e-05, "loss": 0.0051, "step": 7534 }, { "epoch": 9.86, "learning_rate": 1.2370000000000002e-05, "loss": 0.0056, "step": 7535 }, { "epoch": 9.86, "learning_rate": 1.2365e-05, "loss": 0.0049, "step": 7536 }, { "epoch": 9.87, "learning_rate": 1.236e-05, "loss": 0.0053, "step": 7537 }, { "epoch": 9.87, "learning_rate": 1.2355e-05, "loss": 0.0079, "step": 7538 }, { "epoch": 9.87, "learning_rate": 1.235e-05, "loss": 0.0068, "step": 7539 }, { "epoch": 9.87, "learning_rate": 1.2345000000000001e-05, "loss": 0.0058, "step": 7540 }, { "epoch": 9.87, "learning_rate": 1.234e-05, "loss": 0.0056, "step": 7541 }, { "epoch": 9.87, "learning_rate": 1.2335000000000001e-05, "loss": 0.0065, "step": 7542 }, { "epoch": 9.87, "learning_rate": 1.233e-05, "loss": 0.0042, "step": 7543 }, { "epoch": 9.87, "learning_rate": 1.2325e-05, "loss": 0.0079, "step": 7544 }, { "epoch": 9.88, "learning_rate": 1.232e-05, "loss": 0.0053, "step": 7545 }, { "epoch": 9.88, "learning_rate": 1.2315e-05, "loss": 0.0056, "step": 7546 }, { "epoch": 9.88, "learning_rate": 1.231e-05, "loss": 0.0038, "step": 7547 }, { "epoch": 9.88, "learning_rate": 1.2305000000000002e-05, "loss": 0.0059, "step": 7548 }, { "epoch": 9.88, "learning_rate": 1.23e-05, "loss": 0.0049, "step": 7549 }, { "epoch": 9.88, "learning_rate": 1.2295000000000002e-05, "loss": 0.0051, "step": 7550 }, { "epoch": 9.88, "learning_rate": 1.2290000000000001e-05, "loss": 0.0058, "step": 7551 }, { "epoch": 9.88, "learning_rate": 1.2285e-05, "loss": 0.0073, "step": 7552 }, { "epoch": 9.89, "learning_rate": 1.2280000000000001e-05, "loss": 0.005, "step": 7553 }, { "epoch": 9.89, "learning_rate": 1.2275e-05, "loss": 0.0049, "step": 7554 }, { "epoch": 9.89, "learning_rate": 1.2270000000000001e-05, "loss": 0.0057, "step": 7555 }, { "epoch": 9.89, "learning_rate": 1.2265e-05, "loss": 0.0049, "step": 7556 }, { "epoch": 9.89, "learning_rate": 1.2260000000000001e-05, "loss": 0.0064, "step": 7557 }, { "epoch": 9.89, "learning_rate": 1.2255e-05, "loss": 0.0055, "step": 7558 }, { "epoch": 9.89, "learning_rate": 1.225e-05, "loss": 0.0054, "step": 7559 }, { "epoch": 9.9, "learning_rate": 1.2245e-05, "loss": 0.0055, "step": 7560 }, { "epoch": 9.9, "learning_rate": 1.224e-05, "loss": 0.0063, "step": 7561 }, { "epoch": 9.9, "learning_rate": 1.2235e-05, "loss": 0.0067, "step": 7562 }, { "epoch": 9.9, "learning_rate": 1.2230000000000001e-05, "loss": 0.0047, "step": 7563 }, { "epoch": 9.9, "learning_rate": 1.2225e-05, "loss": 0.0057, "step": 7564 }, { "epoch": 9.9, "learning_rate": 1.2220000000000002e-05, "loss": 0.0045, "step": 7565 }, { "epoch": 9.9, "learning_rate": 1.2215e-05, "loss": 0.0059, "step": 7566 }, { "epoch": 9.9, "learning_rate": 1.221e-05, "loss": 0.0053, "step": 7567 }, { "epoch": 9.91, "learning_rate": 1.2205000000000001e-05, "loss": 0.0047, "step": 7568 }, { "epoch": 9.91, "learning_rate": 1.22e-05, "loss": 0.0055, "step": 7569 }, { "epoch": 9.91, "learning_rate": 1.2195000000000001e-05, "loss": 0.0048, "step": 7570 }, { "epoch": 9.91, "learning_rate": 1.219e-05, "loss": 0.0057, "step": 7571 }, { "epoch": 9.91, "learning_rate": 1.2185000000000001e-05, "loss": 0.0047, "step": 7572 }, { "epoch": 9.91, "learning_rate": 1.2180000000000002e-05, "loss": 0.0054, "step": 7573 }, { "epoch": 9.91, "learning_rate": 1.2175e-05, "loss": 0.005, "step": 7574 }, { "epoch": 9.91, "learning_rate": 1.217e-05, "loss": 0.0055, "step": 7575 }, { "epoch": 9.92, "learning_rate": 1.2165e-05, "loss": 0.0055, "step": 7576 }, { "epoch": 9.92, "learning_rate": 1.216e-05, "loss": 0.0051, "step": 7577 }, { "epoch": 9.92, "learning_rate": 1.2155000000000001e-05, "loss": 0.0053, "step": 7578 }, { "epoch": 9.92, "learning_rate": 1.215e-05, "loss": 0.0045, "step": 7579 }, { "epoch": 9.92, "learning_rate": 1.2145000000000001e-05, "loss": 0.01, "step": 7580 }, { "epoch": 9.92, "learning_rate": 1.214e-05, "loss": 0.0043, "step": 7581 }, { "epoch": 9.92, "learning_rate": 1.2135e-05, "loss": 0.0386, "step": 7582 }, { "epoch": 9.93, "learning_rate": 1.213e-05, "loss": 0.0051, "step": 7583 }, { "epoch": 9.93, "learning_rate": 1.2125e-05, "loss": 0.0048, "step": 7584 }, { "epoch": 9.93, "learning_rate": 1.2120000000000001e-05, "loss": 0.0055, "step": 7585 }, { "epoch": 9.93, "learning_rate": 1.2115e-05, "loss": 0.0059, "step": 7586 }, { "epoch": 9.93, "learning_rate": 1.2110000000000001e-05, "loss": 0.0066, "step": 7587 }, { "epoch": 9.93, "learning_rate": 1.2105000000000002e-05, "loss": 0.0063, "step": 7588 }, { "epoch": 9.93, "learning_rate": 1.2100000000000001e-05, "loss": 0.0077, "step": 7589 }, { "epoch": 9.93, "learning_rate": 1.2095e-05, "loss": 0.0053, "step": 7590 }, { "epoch": 9.94, "learning_rate": 1.209e-05, "loss": 0.005, "step": 7591 }, { "epoch": 9.94, "learning_rate": 1.2085e-05, "loss": 0.0053, "step": 7592 }, { "epoch": 9.94, "learning_rate": 1.2080000000000001e-05, "loss": 0.0055, "step": 7593 }, { "epoch": 9.94, "learning_rate": 1.2075e-05, "loss": 0.0053, "step": 7594 }, { "epoch": 9.94, "learning_rate": 1.2070000000000001e-05, "loss": 0.0044, "step": 7595 }, { "epoch": 9.94, "learning_rate": 1.2065e-05, "loss": 0.0062, "step": 7596 }, { "epoch": 9.94, "learning_rate": 1.206e-05, "loss": 0.0071, "step": 7597 }, { "epoch": 9.95, "learning_rate": 1.2055e-05, "loss": 0.005, "step": 7598 }, { "epoch": 9.95, "learning_rate": 1.205e-05, "loss": 0.006, "step": 7599 }, { "epoch": 9.95, "learning_rate": 1.2045e-05, "loss": 0.0138, "step": 7600 }, { "epoch": 9.95, "learning_rate": 1.204e-05, "loss": 0.0049, "step": 7601 }, { "epoch": 9.95, "learning_rate": 1.2035e-05, "loss": 0.0059, "step": 7602 }, { "epoch": 9.95, "learning_rate": 1.2030000000000002e-05, "loss": 0.0058, "step": 7603 }, { "epoch": 9.95, "learning_rate": 1.2025000000000001e-05, "loss": 0.0054, "step": 7604 }, { "epoch": 9.95, "learning_rate": 1.202e-05, "loss": 0.0059, "step": 7605 }, { "epoch": 9.96, "learning_rate": 1.2015000000000001e-05, "loss": 0.0063, "step": 7606 }, { "epoch": 9.96, "learning_rate": 1.201e-05, "loss": 0.0057, "step": 7607 }, { "epoch": 9.96, "learning_rate": 1.2005000000000001e-05, "loss": 0.0049, "step": 7608 }, { "epoch": 9.96, "learning_rate": 1.2e-05, "loss": 0.0075, "step": 7609 }, { "epoch": 9.96, "learning_rate": 1.1995000000000001e-05, "loss": 0.0053, "step": 7610 }, { "epoch": 9.96, "learning_rate": 1.199e-05, "loss": 0.0048, "step": 7611 }, { "epoch": 9.96, "learning_rate": 1.1985e-05, "loss": 0.0072, "step": 7612 }, { "epoch": 9.96, "learning_rate": 1.198e-05, "loss": 0.0074, "step": 7613 }, { "epoch": 9.97, "learning_rate": 1.1975e-05, "loss": 0.0053, "step": 7614 }, { "epoch": 9.97, "learning_rate": 1.197e-05, "loss": 0.0078, "step": 7615 }, { "epoch": 9.97, "learning_rate": 1.1965000000000001e-05, "loss": 0.0059, "step": 7616 }, { "epoch": 9.97, "learning_rate": 1.196e-05, "loss": 0.0042, "step": 7617 }, { "epoch": 9.97, "learning_rate": 1.1955000000000002e-05, "loss": 0.0078, "step": 7618 }, { "epoch": 9.97, "learning_rate": 1.195e-05, "loss": 0.007, "step": 7619 }, { "epoch": 9.97, "learning_rate": 1.1945e-05, "loss": 0.0051, "step": 7620 }, { "epoch": 9.98, "learning_rate": 1.1940000000000001e-05, "loss": 0.0062, "step": 7621 }, { "epoch": 9.98, "learning_rate": 1.1935e-05, "loss": 0.007, "step": 7622 }, { "epoch": 9.98, "learning_rate": 1.1930000000000001e-05, "loss": 0.008, "step": 7623 }, { "epoch": 9.98, "learning_rate": 1.1925e-05, "loss": 0.0089, "step": 7624 }, { "epoch": 9.98, "learning_rate": 1.1920000000000001e-05, "loss": 0.0058, "step": 7625 }, { "epoch": 9.98, "learning_rate": 1.1915000000000002e-05, "loss": 0.0088, "step": 7626 }, { "epoch": 9.98, "learning_rate": 1.1910000000000001e-05, "loss": 0.0055, "step": 7627 }, { "epoch": 9.98, "learning_rate": 1.1905e-05, "loss": 0.0053, "step": 7628 }, { "epoch": 9.99, "learning_rate": 1.19e-05, "loss": 0.006, "step": 7629 }, { "epoch": 9.99, "learning_rate": 1.1895e-05, "loss": 0.0051, "step": 7630 }, { "epoch": 9.99, "learning_rate": 1.1890000000000001e-05, "loss": 0.0057, "step": 7631 }, { "epoch": 9.99, "learning_rate": 1.1885e-05, "loss": 0.0046, "step": 7632 }, { "epoch": 9.99, "learning_rate": 1.1880000000000001e-05, "loss": 0.0045, "step": 7633 }, { "epoch": 9.99, "learning_rate": 1.1875e-05, "loss": 0.0055, "step": 7634 }, { "epoch": 9.99, "learning_rate": 1.187e-05, "loss": 0.0061, "step": 7635 }, { "epoch": 9.99, "learning_rate": 1.1865e-05, "loss": 0.005, "step": 7636 }, { "epoch": 10.0, "learning_rate": 1.186e-05, "loss": 0.0085, "step": 7637 }, { "epoch": 10.0, "learning_rate": 1.1855e-05, "loss": 0.0053, "step": 7638 }, { "epoch": 10.0, "learning_rate": 1.185e-05, "loss": 0.0049, "step": 7639 }, { "epoch": 10.0, "learning_rate": 1.1845000000000001e-05, "loss": 0.0077, "step": 7640 }, { "epoch": 10.0, "learning_rate": 1.1840000000000002e-05, "loss": 0.0042, "step": 7641 }, { "epoch": 10.0, "learning_rate": 1.1835000000000001e-05, "loss": 0.0035, "step": 7642 }, { "epoch": 10.0, "learning_rate": 1.183e-05, "loss": 0.0037, "step": 7643 }, { "epoch": 10.01, "learning_rate": 1.1825e-05, "loss": 0.0034, "step": 7644 }, { "epoch": 10.01, "learning_rate": 1.182e-05, "loss": 0.0042, "step": 7645 }, { "epoch": 10.01, "learning_rate": 1.1815000000000001e-05, "loss": 0.0035, "step": 7646 }, { "epoch": 10.01, "learning_rate": 1.181e-05, "loss": 0.0028, "step": 7647 }, { "epoch": 10.01, "learning_rate": 1.1805000000000001e-05, "loss": 0.003, "step": 7648 }, { "epoch": 10.01, "learning_rate": 1.18e-05, "loss": 0.0036, "step": 7649 }, { "epoch": 10.01, "learning_rate": 1.1795e-05, "loss": 0.0034, "step": 7650 }, { "epoch": 10.01, "learning_rate": 1.179e-05, "loss": 0.0045, "step": 7651 }, { "epoch": 10.02, "learning_rate": 1.1785e-05, "loss": 0.003, "step": 7652 }, { "epoch": 10.02, "learning_rate": 1.178e-05, "loss": 0.0041, "step": 7653 }, { "epoch": 10.02, "learning_rate": 1.1775e-05, "loss": 0.004, "step": 7654 }, { "epoch": 10.02, "learning_rate": 1.177e-05, "loss": 0.0044, "step": 7655 }, { "epoch": 10.02, "learning_rate": 1.1765000000000002e-05, "loss": 0.0032, "step": 7656 }, { "epoch": 10.02, "learning_rate": 1.1760000000000001e-05, "loss": 0.005, "step": 7657 }, { "epoch": 10.02, "learning_rate": 1.1755e-05, "loss": 0.0035, "step": 7658 }, { "epoch": 10.02, "learning_rate": 1.175e-05, "loss": 0.0046, "step": 7659 }, { "epoch": 10.03, "learning_rate": 1.1745e-05, "loss": 0.0039, "step": 7660 }, { "epoch": 10.03, "learning_rate": 1.1740000000000001e-05, "loss": 0.0043, "step": 7661 }, { "epoch": 10.03, "learning_rate": 1.1735e-05, "loss": 0.003, "step": 7662 }, { "epoch": 10.03, "learning_rate": 1.1730000000000001e-05, "loss": 0.0031, "step": 7663 }, { "epoch": 10.03, "learning_rate": 1.1725e-05, "loss": 0.0029, "step": 7664 }, { "epoch": 10.03, "learning_rate": 1.172e-05, "loss": 0.0039, "step": 7665 }, { "epoch": 10.03, "learning_rate": 1.1715e-05, "loss": 0.0031, "step": 7666 }, { "epoch": 10.04, "learning_rate": 1.171e-05, "loss": 0.0021, "step": 7667 }, { "epoch": 10.04, "learning_rate": 1.1705e-05, "loss": 0.0045, "step": 7668 }, { "epoch": 10.04, "learning_rate": 1.1700000000000001e-05, "loss": 0.0048, "step": 7669 }, { "epoch": 10.04, "learning_rate": 1.1695e-05, "loss": 0.0031, "step": 7670 }, { "epoch": 10.04, "learning_rate": 1.1690000000000002e-05, "loss": 0.0049, "step": 7671 }, { "epoch": 10.04, "learning_rate": 1.1685e-05, "loss": 0.0031, "step": 7672 }, { "epoch": 10.04, "learning_rate": 1.168e-05, "loss": 0.0036, "step": 7673 }, { "epoch": 10.04, "learning_rate": 1.1675000000000001e-05, "loss": 0.0033, "step": 7674 }, { "epoch": 10.05, "learning_rate": 1.167e-05, "loss": 0.0034, "step": 7675 }, { "epoch": 10.05, "learning_rate": 1.1665000000000001e-05, "loss": 0.003, "step": 7676 }, { "epoch": 10.05, "learning_rate": 1.166e-05, "loss": 0.0037, "step": 7677 }, { "epoch": 10.05, "learning_rate": 1.1655000000000001e-05, "loss": 0.0038, "step": 7678 }, { "epoch": 10.05, "learning_rate": 1.1650000000000002e-05, "loss": 0.0054, "step": 7679 }, { "epoch": 10.05, "learning_rate": 1.1645000000000001e-05, "loss": 0.0033, "step": 7680 }, { "epoch": 10.05, "learning_rate": 1.164e-05, "loss": 0.0043, "step": 7681 }, { "epoch": 10.05, "learning_rate": 1.1635e-05, "loss": 0.0043, "step": 7682 }, { "epoch": 10.06, "learning_rate": 1.163e-05, "loss": 0.0057, "step": 7683 }, { "epoch": 10.06, "learning_rate": 1.1625000000000001e-05, "loss": 0.0032, "step": 7684 }, { "epoch": 10.06, "learning_rate": 1.162e-05, "loss": 0.0061, "step": 7685 }, { "epoch": 10.06, "learning_rate": 1.1615000000000001e-05, "loss": 0.0026, "step": 7686 }, { "epoch": 10.06, "learning_rate": 1.161e-05, "loss": 0.0043, "step": 7687 }, { "epoch": 10.06, "learning_rate": 1.1605e-05, "loss": 0.0027, "step": 7688 }, { "epoch": 10.06, "learning_rate": 1.16e-05, "loss": 0.0041, "step": 7689 }, { "epoch": 10.07, "learning_rate": 1.1595e-05, "loss": 0.0029, "step": 7690 }, { "epoch": 10.07, "learning_rate": 1.159e-05, "loss": 0.0053, "step": 7691 }, { "epoch": 10.07, "learning_rate": 1.1585e-05, "loss": 0.0043, "step": 7692 }, { "epoch": 10.07, "learning_rate": 1.1580000000000001e-05, "loss": 0.0058, "step": 7693 }, { "epoch": 10.07, "learning_rate": 1.1575000000000002e-05, "loss": 0.0042, "step": 7694 }, { "epoch": 10.07, "learning_rate": 1.1570000000000001e-05, "loss": 0.0028, "step": 7695 }, { "epoch": 10.07, "learning_rate": 1.1565e-05, "loss": 0.0036, "step": 7696 }, { "epoch": 10.07, "learning_rate": 1.156e-05, "loss": 0.0079, "step": 7697 }, { "epoch": 10.08, "learning_rate": 1.1555e-05, "loss": 0.0044, "step": 7698 }, { "epoch": 10.08, "learning_rate": 1.1550000000000001e-05, "loss": 0.0041, "step": 7699 }, { "epoch": 10.08, "learning_rate": 1.1545e-05, "loss": 0.0049, "step": 7700 }, { "epoch": 10.08, "learning_rate": 1.1540000000000001e-05, "loss": 0.0039, "step": 7701 }, { "epoch": 10.08, "learning_rate": 1.1535e-05, "loss": 0.0036, "step": 7702 }, { "epoch": 10.08, "learning_rate": 1.153e-05, "loss": 0.0043, "step": 7703 }, { "epoch": 10.08, "learning_rate": 1.1525e-05, "loss": 0.003, "step": 7704 }, { "epoch": 10.09, "learning_rate": 1.152e-05, "loss": 0.0031, "step": 7705 }, { "epoch": 10.09, "learning_rate": 1.1515e-05, "loss": 0.0062, "step": 7706 }, { "epoch": 10.09, "learning_rate": 1.151e-05, "loss": 0.0025, "step": 7707 }, { "epoch": 10.09, "learning_rate": 1.1505e-05, "loss": 0.0031, "step": 7708 }, { "epoch": 10.09, "learning_rate": 1.1500000000000002e-05, "loss": 0.0034, "step": 7709 }, { "epoch": 10.09, "learning_rate": 1.1495000000000001e-05, "loss": 0.0048, "step": 7710 }, { "epoch": 10.09, "learning_rate": 1.149e-05, "loss": 0.0042, "step": 7711 }, { "epoch": 10.09, "learning_rate": 1.1485e-05, "loss": 0.0028, "step": 7712 }, { "epoch": 10.1, "learning_rate": 1.148e-05, "loss": 0.0043, "step": 7713 }, { "epoch": 10.1, "learning_rate": 1.1475000000000001e-05, "loss": 0.0048, "step": 7714 }, { "epoch": 10.1, "learning_rate": 1.147e-05, "loss": 0.0031, "step": 7715 }, { "epoch": 10.1, "learning_rate": 1.1465000000000001e-05, "loss": 0.0043, "step": 7716 }, { "epoch": 10.1, "learning_rate": 1.146e-05, "loss": 0.0033, "step": 7717 }, { "epoch": 10.1, "learning_rate": 1.1455000000000001e-05, "loss": 0.0038, "step": 7718 }, { "epoch": 10.1, "learning_rate": 1.145e-05, "loss": 0.0034, "step": 7719 }, { "epoch": 10.1, "learning_rate": 1.1445e-05, "loss": 0.0039, "step": 7720 }, { "epoch": 10.11, "learning_rate": 1.144e-05, "loss": 0.0029, "step": 7721 }, { "epoch": 10.11, "learning_rate": 1.1435e-05, "loss": 0.0041, "step": 7722 }, { "epoch": 10.11, "learning_rate": 1.143e-05, "loss": 0.0041, "step": 7723 }, { "epoch": 10.11, "learning_rate": 1.1425000000000002e-05, "loss": 0.0027, "step": 7724 }, { "epoch": 10.11, "learning_rate": 1.142e-05, "loss": 0.0036, "step": 7725 }, { "epoch": 10.11, "learning_rate": 1.1415e-05, "loss": 0.0027, "step": 7726 }, { "epoch": 10.11, "learning_rate": 1.141e-05, "loss": 0.0024, "step": 7727 }, { "epoch": 10.12, "learning_rate": 1.1405e-05, "loss": 0.0027, "step": 7728 }, { "epoch": 10.12, "learning_rate": 1.1400000000000001e-05, "loss": 0.0025, "step": 7729 }, { "epoch": 10.12, "learning_rate": 1.1395e-05, "loss": 0.0051, "step": 7730 }, { "epoch": 10.12, "learning_rate": 1.1390000000000001e-05, "loss": 0.0029, "step": 7731 }, { "epoch": 10.12, "learning_rate": 1.1385000000000002e-05, "loss": 0.0061, "step": 7732 }, { "epoch": 10.12, "learning_rate": 1.1380000000000001e-05, "loss": 0.0039, "step": 7733 }, { "epoch": 10.12, "learning_rate": 1.1375e-05, "loss": 0.0054, "step": 7734 }, { "epoch": 10.12, "learning_rate": 1.137e-05, "loss": 0.0036, "step": 7735 }, { "epoch": 10.13, "learning_rate": 1.1365e-05, "loss": 0.0028, "step": 7736 }, { "epoch": 10.13, "learning_rate": 1.1360000000000001e-05, "loss": 0.0039, "step": 7737 }, { "epoch": 10.13, "learning_rate": 1.1355e-05, "loss": 0.0044, "step": 7738 }, { "epoch": 10.13, "learning_rate": 1.1350000000000001e-05, "loss": 0.0051, "step": 7739 }, { "epoch": 10.13, "learning_rate": 1.1345e-05, "loss": 0.0023, "step": 7740 }, { "epoch": 10.13, "learning_rate": 1.134e-05, "loss": 0.0021, "step": 7741 }, { "epoch": 10.13, "learning_rate": 1.1335e-05, "loss": 0.0029, "step": 7742 }, { "epoch": 10.13, "learning_rate": 1.133e-05, "loss": 0.0022, "step": 7743 }, { "epoch": 10.14, "learning_rate": 1.1325e-05, "loss": 0.0052, "step": 7744 }, { "epoch": 10.14, "learning_rate": 1.132e-05, "loss": 0.0047, "step": 7745 }, { "epoch": 10.14, "learning_rate": 1.1315000000000001e-05, "loss": 0.003, "step": 7746 }, { "epoch": 10.14, "learning_rate": 1.1310000000000002e-05, "loss": 0.0082, "step": 7747 }, { "epoch": 10.14, "learning_rate": 1.1305000000000001e-05, "loss": 0.0025, "step": 7748 }, { "epoch": 10.14, "learning_rate": 1.13e-05, "loss": 0.0038, "step": 7749 }, { "epoch": 10.14, "learning_rate": 1.1295e-05, "loss": 0.0037, "step": 7750 }, { "epoch": 10.15, "learning_rate": 1.129e-05, "loss": 0.0034, "step": 7751 }, { "epoch": 10.15, "learning_rate": 1.1285000000000001e-05, "loss": 0.0035, "step": 7752 }, { "epoch": 10.15, "learning_rate": 1.128e-05, "loss": 0.0042, "step": 7753 }, { "epoch": 10.15, "learning_rate": 1.1275000000000001e-05, "loss": 0.0046, "step": 7754 }, { "epoch": 10.15, "learning_rate": 1.127e-05, "loss": 0.0037, "step": 7755 }, { "epoch": 10.15, "learning_rate": 1.1265e-05, "loss": 0.0049, "step": 7756 }, { "epoch": 10.15, "learning_rate": 1.126e-05, "loss": 0.004, "step": 7757 }, { "epoch": 10.15, "learning_rate": 1.1255e-05, "loss": 0.0022, "step": 7758 }, { "epoch": 10.16, "learning_rate": 1.125e-05, "loss": 0.0038, "step": 7759 }, { "epoch": 10.16, "learning_rate": 1.1245e-05, "loss": 0.003, "step": 7760 }, { "epoch": 10.16, "learning_rate": 1.124e-05, "loss": 0.0057, "step": 7761 }, { "epoch": 10.16, "learning_rate": 1.1235000000000002e-05, "loss": 0.0056, "step": 7762 }, { "epoch": 10.16, "learning_rate": 1.1230000000000001e-05, "loss": 0.0032, "step": 7763 }, { "epoch": 10.16, "learning_rate": 1.1225e-05, "loss": 0.0029, "step": 7764 }, { "epoch": 10.16, "learning_rate": 1.122e-05, "loss": 0.0041, "step": 7765 }, { "epoch": 10.16, "learning_rate": 1.1215e-05, "loss": 0.0032, "step": 7766 }, { "epoch": 10.17, "learning_rate": 1.1210000000000001e-05, "loss": 0.0033, "step": 7767 }, { "epoch": 10.17, "learning_rate": 1.1205e-05, "loss": 0.0034, "step": 7768 }, { "epoch": 10.17, "learning_rate": 1.1200000000000001e-05, "loss": 0.0036, "step": 7769 }, { "epoch": 10.17, "learning_rate": 1.1195e-05, "loss": 0.0028, "step": 7770 }, { "epoch": 10.17, "learning_rate": 1.1190000000000001e-05, "loss": 0.003, "step": 7771 }, { "epoch": 10.17, "learning_rate": 1.1185e-05, "loss": 0.003, "step": 7772 }, { "epoch": 10.17, "learning_rate": 1.118e-05, "loss": 0.004, "step": 7773 }, { "epoch": 10.18, "learning_rate": 1.1175e-05, "loss": 0.0044, "step": 7774 }, { "epoch": 10.18, "learning_rate": 1.117e-05, "loss": 0.0037, "step": 7775 }, { "epoch": 10.18, "learning_rate": 1.1165e-05, "loss": 0.0044, "step": 7776 }, { "epoch": 10.18, "learning_rate": 1.1160000000000002e-05, "loss": 0.0044, "step": 7777 }, { "epoch": 10.18, "learning_rate": 1.1155e-05, "loss": 0.0041, "step": 7778 }, { "epoch": 10.18, "learning_rate": 1.115e-05, "loss": 0.0036, "step": 7779 }, { "epoch": 10.18, "learning_rate": 1.1145e-05, "loss": 0.0041, "step": 7780 }, { "epoch": 10.18, "learning_rate": 1.114e-05, "loss": 0.0021, "step": 7781 }, { "epoch": 10.19, "learning_rate": 1.1135000000000001e-05, "loss": 0.0036, "step": 7782 }, { "epoch": 10.19, "learning_rate": 1.113e-05, "loss": 0.004, "step": 7783 }, { "epoch": 10.19, "learning_rate": 1.1125000000000001e-05, "loss": 0.0057, "step": 7784 }, { "epoch": 10.19, "learning_rate": 1.112e-05, "loss": 0.0041, "step": 7785 }, { "epoch": 10.19, "learning_rate": 1.1115000000000001e-05, "loss": 0.0029, "step": 7786 }, { "epoch": 10.19, "learning_rate": 1.111e-05, "loss": 0.0058, "step": 7787 }, { "epoch": 10.19, "learning_rate": 1.1105e-05, "loss": 0.0057, "step": 7788 }, { "epoch": 10.2, "learning_rate": 1.11e-05, "loss": 0.0042, "step": 7789 }, { "epoch": 10.2, "learning_rate": 1.1095e-05, "loss": 0.003, "step": 7790 }, { "epoch": 10.2, "learning_rate": 1.109e-05, "loss": 0.0029, "step": 7791 }, { "epoch": 10.2, "learning_rate": 1.1085000000000001e-05, "loss": 0.0034, "step": 7792 }, { "epoch": 10.2, "learning_rate": 1.108e-05, "loss": 0.0022, "step": 7793 }, { "epoch": 10.2, "learning_rate": 1.1075e-05, "loss": 0.0066, "step": 7794 }, { "epoch": 10.2, "learning_rate": 1.107e-05, "loss": 0.0036, "step": 7795 }, { "epoch": 10.2, "learning_rate": 1.1065e-05, "loss": 0.0036, "step": 7796 }, { "epoch": 10.21, "learning_rate": 1.106e-05, "loss": 0.0038, "step": 7797 }, { "epoch": 10.21, "learning_rate": 1.1055e-05, "loss": 0.0034, "step": 7798 }, { "epoch": 10.21, "learning_rate": 1.1050000000000001e-05, "loss": 0.0027, "step": 7799 }, { "epoch": 10.21, "learning_rate": 1.1045000000000002e-05, "loss": 0.0041, "step": 7800 }, { "epoch": 10.21, "learning_rate": 1.1040000000000001e-05, "loss": 0.0064, "step": 7801 }, { "epoch": 10.21, "learning_rate": 1.1035e-05, "loss": 0.0061, "step": 7802 }, { "epoch": 10.21, "learning_rate": 1.103e-05, "loss": 0.0045, "step": 7803 }, { "epoch": 10.21, "learning_rate": 1.1025e-05, "loss": 0.0035, "step": 7804 }, { "epoch": 10.22, "learning_rate": 1.1020000000000001e-05, "loss": 0.0042, "step": 7805 }, { "epoch": 10.22, "learning_rate": 1.1015e-05, "loss": 0.0035, "step": 7806 }, { "epoch": 10.22, "learning_rate": 1.1010000000000001e-05, "loss": 0.0036, "step": 7807 }, { "epoch": 10.22, "learning_rate": 1.1005e-05, "loss": 0.0056, "step": 7808 }, { "epoch": 10.22, "learning_rate": 1.1000000000000001e-05, "loss": 0.0064, "step": 7809 }, { "epoch": 10.22, "learning_rate": 1.0995e-05, "loss": 0.0035, "step": 7810 }, { "epoch": 10.22, "learning_rate": 1.099e-05, "loss": 0.0026, "step": 7811 }, { "epoch": 10.23, "learning_rate": 1.0985e-05, "loss": 0.0048, "step": 7812 }, { "epoch": 10.23, "learning_rate": 1.098e-05, "loss": 0.0045, "step": 7813 }, { "epoch": 10.23, "learning_rate": 1.0975e-05, "loss": 0.0046, "step": 7814 }, { "epoch": 10.23, "learning_rate": 1.0970000000000002e-05, "loss": 0.0037, "step": 7815 }, { "epoch": 10.23, "learning_rate": 1.0965000000000001e-05, "loss": 0.0037, "step": 7816 }, { "epoch": 10.23, "learning_rate": 1.096e-05, "loss": 0.0022, "step": 7817 }, { "epoch": 10.23, "learning_rate": 1.0955e-05, "loss": 0.0036, "step": 7818 }, { "epoch": 10.23, "learning_rate": 1.095e-05, "loss": 0.0019, "step": 7819 }, { "epoch": 10.24, "learning_rate": 1.0945000000000001e-05, "loss": 0.0038, "step": 7820 }, { "epoch": 10.24, "learning_rate": 1.094e-05, "loss": 0.0042, "step": 7821 }, { "epoch": 10.24, "learning_rate": 1.0935000000000001e-05, "loss": 0.0065, "step": 7822 }, { "epoch": 10.24, "learning_rate": 1.093e-05, "loss": 0.0027, "step": 7823 }, { "epoch": 10.24, "learning_rate": 1.0925000000000001e-05, "loss": 0.0052, "step": 7824 }, { "epoch": 10.24, "learning_rate": 1.092e-05, "loss": 0.007, "step": 7825 }, { "epoch": 10.24, "learning_rate": 1.0915e-05, "loss": 0.0034, "step": 7826 }, { "epoch": 10.24, "learning_rate": 1.091e-05, "loss": 0.0035, "step": 7827 }, { "epoch": 10.25, "learning_rate": 1.0905e-05, "loss": 0.0029, "step": 7828 }, { "epoch": 10.25, "learning_rate": 1.09e-05, "loss": 0.0037, "step": 7829 }, { "epoch": 10.25, "learning_rate": 1.0895000000000002e-05, "loss": 0.0047, "step": 7830 }, { "epoch": 10.25, "learning_rate": 1.089e-05, "loss": 0.0043, "step": 7831 }, { "epoch": 10.25, "learning_rate": 1.0885e-05, "loss": 0.0054, "step": 7832 }, { "epoch": 10.25, "learning_rate": 1.088e-05, "loss": 0.0022, "step": 7833 }, { "epoch": 10.25, "learning_rate": 1.0875e-05, "loss": 0.0027, "step": 7834 }, { "epoch": 10.26, "learning_rate": 1.0870000000000001e-05, "loss": 0.0037, "step": 7835 }, { "epoch": 10.26, "learning_rate": 1.0865e-05, "loss": 0.0041, "step": 7836 }, { "epoch": 10.26, "learning_rate": 1.0860000000000001e-05, "loss": 0.0038, "step": 7837 }, { "epoch": 10.26, "learning_rate": 1.0855e-05, "loss": 0.0029, "step": 7838 }, { "epoch": 10.26, "learning_rate": 1.0850000000000001e-05, "loss": 0.0043, "step": 7839 }, { "epoch": 10.26, "learning_rate": 1.0845e-05, "loss": 0.0036, "step": 7840 }, { "epoch": 10.26, "learning_rate": 1.084e-05, "loss": 0.0026, "step": 7841 }, { "epoch": 10.26, "learning_rate": 1.0835e-05, "loss": 0.0028, "step": 7842 }, { "epoch": 10.27, "learning_rate": 1.083e-05, "loss": 0.0036, "step": 7843 }, { "epoch": 10.27, "learning_rate": 1.0825e-05, "loss": 0.017, "step": 7844 }, { "epoch": 10.27, "learning_rate": 1.0820000000000001e-05, "loss": 0.0032, "step": 7845 }, { "epoch": 10.27, "learning_rate": 1.0815e-05, "loss": 0.0033, "step": 7846 }, { "epoch": 10.27, "learning_rate": 1.081e-05, "loss": 0.0037, "step": 7847 }, { "epoch": 10.27, "learning_rate": 1.0804999999999999e-05, "loss": 0.0036, "step": 7848 }, { "epoch": 10.27, "learning_rate": 1.08e-05, "loss": 0.0028, "step": 7849 }, { "epoch": 10.27, "learning_rate": 1.0795e-05, "loss": 0.0065, "step": 7850 }, { "epoch": 10.28, "learning_rate": 1.079e-05, "loss": 0.0066, "step": 7851 }, { "epoch": 10.28, "learning_rate": 1.0785000000000001e-05, "loss": 0.0036, "step": 7852 }, { "epoch": 10.28, "learning_rate": 1.0780000000000002e-05, "loss": 0.0031, "step": 7853 }, { "epoch": 10.28, "learning_rate": 1.0775000000000001e-05, "loss": 0.004, "step": 7854 }, { "epoch": 10.28, "learning_rate": 1.077e-05, "loss": 0.0034, "step": 7855 }, { "epoch": 10.28, "learning_rate": 1.0765e-05, "loss": 0.0037, "step": 7856 }, { "epoch": 10.28, "learning_rate": 1.076e-05, "loss": 0.0042, "step": 7857 }, { "epoch": 10.29, "learning_rate": 1.0755000000000001e-05, "loss": 0.0044, "step": 7858 }, { "epoch": 10.29, "learning_rate": 1.075e-05, "loss": 0.0032, "step": 7859 }, { "epoch": 10.29, "learning_rate": 1.0745000000000001e-05, "loss": 0.0047, "step": 7860 }, { "epoch": 10.29, "learning_rate": 1.074e-05, "loss": 0.0033, "step": 7861 }, { "epoch": 10.29, "learning_rate": 1.0735000000000001e-05, "loss": 0.0032, "step": 7862 }, { "epoch": 10.29, "learning_rate": 1.073e-05, "loss": 0.0056, "step": 7863 }, { "epoch": 10.29, "learning_rate": 1.0725e-05, "loss": 0.0026, "step": 7864 }, { "epoch": 10.29, "learning_rate": 1.072e-05, "loss": 0.0035, "step": 7865 }, { "epoch": 10.3, "learning_rate": 1.0715e-05, "loss": 0.0031, "step": 7866 }, { "epoch": 10.3, "learning_rate": 1.071e-05, "loss": 0.005, "step": 7867 }, { "epoch": 10.3, "learning_rate": 1.0705000000000002e-05, "loss": 0.0042, "step": 7868 }, { "epoch": 10.3, "learning_rate": 1.0700000000000001e-05, "loss": 0.0031, "step": 7869 }, { "epoch": 10.3, "learning_rate": 1.0695e-05, "loss": 0.0027, "step": 7870 }, { "epoch": 10.3, "learning_rate": 1.069e-05, "loss": 0.0044, "step": 7871 }, { "epoch": 10.3, "learning_rate": 1.0685e-05, "loss": 0.0058, "step": 7872 }, { "epoch": 10.3, "learning_rate": 1.0680000000000001e-05, "loss": 0.0027, "step": 7873 }, { "epoch": 10.31, "learning_rate": 1.0675e-05, "loss": 0.0043, "step": 7874 }, { "epoch": 10.31, "learning_rate": 1.0670000000000001e-05, "loss": 0.0034, "step": 7875 }, { "epoch": 10.31, "learning_rate": 1.0665e-05, "loss": 0.0044, "step": 7876 }, { "epoch": 10.31, "learning_rate": 1.0660000000000001e-05, "loss": 0.0036, "step": 7877 }, { "epoch": 10.31, "learning_rate": 1.0655e-05, "loss": 0.0039, "step": 7878 }, { "epoch": 10.31, "learning_rate": 1.065e-05, "loss": 0.0047, "step": 7879 }, { "epoch": 10.31, "learning_rate": 1.0645e-05, "loss": 0.0035, "step": 7880 }, { "epoch": 10.32, "learning_rate": 1.064e-05, "loss": 0.0035, "step": 7881 }, { "epoch": 10.32, "learning_rate": 1.0635e-05, "loss": 0.003, "step": 7882 }, { "epoch": 10.32, "learning_rate": 1.0630000000000002e-05, "loss": 0.0034, "step": 7883 }, { "epoch": 10.32, "learning_rate": 1.0625e-05, "loss": 0.0034, "step": 7884 }, { "epoch": 10.32, "learning_rate": 1.062e-05, "loss": 0.0034, "step": 7885 }, { "epoch": 10.32, "learning_rate": 1.0615e-05, "loss": 0.0032, "step": 7886 }, { "epoch": 10.32, "learning_rate": 1.061e-05, "loss": 0.0046, "step": 7887 }, { "epoch": 10.32, "learning_rate": 1.0605000000000001e-05, "loss": 0.0093, "step": 7888 }, { "epoch": 10.33, "learning_rate": 1.06e-05, "loss": 0.0039, "step": 7889 }, { "epoch": 10.33, "learning_rate": 1.0595000000000001e-05, "loss": 0.0051, "step": 7890 }, { "epoch": 10.33, "learning_rate": 1.059e-05, "loss": 0.0036, "step": 7891 }, { "epoch": 10.33, "learning_rate": 1.0585000000000001e-05, "loss": 0.0036, "step": 7892 }, { "epoch": 10.33, "learning_rate": 1.058e-05, "loss": 0.0032, "step": 7893 }, { "epoch": 10.33, "learning_rate": 1.0575e-05, "loss": 0.0039, "step": 7894 }, { "epoch": 10.33, "learning_rate": 1.057e-05, "loss": 0.0038, "step": 7895 }, { "epoch": 10.34, "learning_rate": 1.0565e-05, "loss": 0.0031, "step": 7896 }, { "epoch": 10.34, "learning_rate": 1.056e-05, "loss": 0.0035, "step": 7897 }, { "epoch": 10.34, "learning_rate": 1.0555000000000001e-05, "loss": 0.0042, "step": 7898 }, { "epoch": 10.34, "learning_rate": 1.055e-05, "loss": 0.0044, "step": 7899 }, { "epoch": 10.34, "learning_rate": 1.0545000000000002e-05, "loss": 0.0037, "step": 7900 }, { "epoch": 10.34, "learning_rate": 1.0539999999999999e-05, "loss": 0.0028, "step": 7901 }, { "epoch": 10.34, "learning_rate": 1.0535e-05, "loss": 0.0036, "step": 7902 }, { "epoch": 10.34, "learning_rate": 1.053e-05, "loss": 0.0043, "step": 7903 }, { "epoch": 10.35, "learning_rate": 1.0525e-05, "loss": 0.0038, "step": 7904 }, { "epoch": 10.35, "learning_rate": 1.0520000000000001e-05, "loss": 0.0042, "step": 7905 }, { "epoch": 10.35, "learning_rate": 1.0515e-05, "loss": 0.0033, "step": 7906 }, { "epoch": 10.35, "learning_rate": 1.0510000000000001e-05, "loss": 0.004, "step": 7907 }, { "epoch": 10.35, "learning_rate": 1.0505e-05, "loss": 0.0045, "step": 7908 }, { "epoch": 10.35, "learning_rate": 1.05e-05, "loss": 0.0042, "step": 7909 }, { "epoch": 10.35, "learning_rate": 1.0495e-05, "loss": 0.0043, "step": 7910 }, { "epoch": 10.35, "learning_rate": 1.049e-05, "loss": 0.0035, "step": 7911 }, { "epoch": 10.36, "learning_rate": 1.0485e-05, "loss": 0.0041, "step": 7912 }, { "epoch": 10.36, "learning_rate": 1.0480000000000001e-05, "loss": 0.0041, "step": 7913 }, { "epoch": 10.36, "learning_rate": 1.0475e-05, "loss": 0.0038, "step": 7914 }, { "epoch": 10.36, "learning_rate": 1.0470000000000001e-05, "loss": 0.0035, "step": 7915 }, { "epoch": 10.36, "learning_rate": 1.0465e-05, "loss": 0.0042, "step": 7916 }, { "epoch": 10.36, "learning_rate": 1.046e-05, "loss": 0.0044, "step": 7917 }, { "epoch": 10.36, "learning_rate": 1.0455e-05, "loss": 0.0047, "step": 7918 }, { "epoch": 10.37, "learning_rate": 1.045e-05, "loss": 0.0058, "step": 7919 }, { "epoch": 10.37, "learning_rate": 1.0445e-05, "loss": 0.0034, "step": 7920 }, { "epoch": 10.37, "learning_rate": 1.0440000000000002e-05, "loss": 0.0039, "step": 7921 }, { "epoch": 10.37, "learning_rate": 1.0435000000000001e-05, "loss": 0.003, "step": 7922 }, { "epoch": 10.37, "learning_rate": 1.043e-05, "loss": 0.0072, "step": 7923 }, { "epoch": 10.37, "learning_rate": 1.0425e-05, "loss": 0.0035, "step": 7924 }, { "epoch": 10.37, "learning_rate": 1.042e-05, "loss": 0.0045, "step": 7925 }, { "epoch": 10.37, "learning_rate": 1.0415000000000001e-05, "loss": 0.0031, "step": 7926 }, { "epoch": 10.38, "learning_rate": 1.041e-05, "loss": 0.0047, "step": 7927 }, { "epoch": 10.38, "learning_rate": 1.0405000000000001e-05, "loss": 0.0039, "step": 7928 }, { "epoch": 10.38, "learning_rate": 1.04e-05, "loss": 0.0025, "step": 7929 }, { "epoch": 10.38, "learning_rate": 1.0395000000000001e-05, "loss": 0.0049, "step": 7930 }, { "epoch": 10.38, "learning_rate": 1.039e-05, "loss": 0.0019, "step": 7931 }, { "epoch": 10.38, "learning_rate": 1.0385e-05, "loss": 0.0041, "step": 7932 }, { "epoch": 10.38, "learning_rate": 1.038e-05, "loss": 0.0074, "step": 7933 }, { "epoch": 10.38, "learning_rate": 1.0375e-05, "loss": 0.0057, "step": 7934 }, { "epoch": 10.39, "learning_rate": 1.037e-05, "loss": 0.0038, "step": 7935 }, { "epoch": 10.39, "learning_rate": 1.0365000000000002e-05, "loss": 0.0038, "step": 7936 }, { "epoch": 10.39, "learning_rate": 1.036e-05, "loss": 0.0042, "step": 7937 }, { "epoch": 10.39, "learning_rate": 1.0355000000000002e-05, "loss": 0.0036, "step": 7938 }, { "epoch": 10.39, "learning_rate": 1.035e-05, "loss": 0.0062, "step": 7939 }, { "epoch": 10.39, "learning_rate": 1.0345e-05, "loss": 0.0033, "step": 7940 }, { "epoch": 10.39, "learning_rate": 1.0340000000000001e-05, "loss": 0.0026, "step": 7941 }, { "epoch": 10.4, "learning_rate": 1.0335e-05, "loss": 0.0043, "step": 7942 }, { "epoch": 10.4, "learning_rate": 1.0330000000000001e-05, "loss": 0.0051, "step": 7943 }, { "epoch": 10.4, "learning_rate": 1.0325e-05, "loss": 0.0031, "step": 7944 }, { "epoch": 10.4, "learning_rate": 1.0320000000000001e-05, "loss": 0.0028, "step": 7945 }, { "epoch": 10.4, "learning_rate": 1.0315e-05, "loss": 0.0024, "step": 7946 }, { "epoch": 10.4, "learning_rate": 1.031e-05, "loss": 0.0031, "step": 7947 }, { "epoch": 10.4, "learning_rate": 1.0305e-05, "loss": 0.0036, "step": 7948 }, { "epoch": 10.4, "learning_rate": 1.03e-05, "loss": 0.003, "step": 7949 }, { "epoch": 10.41, "learning_rate": 1.0295e-05, "loss": 0.0029, "step": 7950 }, { "epoch": 10.41, "learning_rate": 1.0290000000000001e-05, "loss": 0.0034, "step": 7951 }, { "epoch": 10.41, "learning_rate": 1.0285e-05, "loss": 0.0042, "step": 7952 }, { "epoch": 10.41, "learning_rate": 1.0280000000000002e-05, "loss": 0.004, "step": 7953 }, { "epoch": 10.41, "learning_rate": 1.0275e-05, "loss": 0.0058, "step": 7954 }, { "epoch": 10.41, "learning_rate": 1.027e-05, "loss": 0.0038, "step": 7955 }, { "epoch": 10.41, "learning_rate": 1.0265e-05, "loss": 0.0036, "step": 7956 }, { "epoch": 10.41, "learning_rate": 1.026e-05, "loss": 0.0033, "step": 7957 }, { "epoch": 10.42, "learning_rate": 1.0255000000000001e-05, "loss": 0.003, "step": 7958 }, { "epoch": 10.42, "learning_rate": 1.025e-05, "loss": 0.0043, "step": 7959 }, { "epoch": 10.42, "learning_rate": 1.0245000000000001e-05, "loss": 0.0032, "step": 7960 }, { "epoch": 10.42, "learning_rate": 1.024e-05, "loss": 0.0029, "step": 7961 }, { "epoch": 10.42, "learning_rate": 1.0235e-05, "loss": 0.008, "step": 7962 }, { "epoch": 10.42, "learning_rate": 1.023e-05, "loss": 0.0039, "step": 7963 }, { "epoch": 10.42, "learning_rate": 1.0225e-05, "loss": 0.0044, "step": 7964 }, { "epoch": 10.43, "learning_rate": 1.022e-05, "loss": 0.0037, "step": 7965 }, { "epoch": 10.43, "learning_rate": 1.0215000000000001e-05, "loss": 0.004, "step": 7966 }, { "epoch": 10.43, "learning_rate": 1.021e-05, "loss": 0.0054, "step": 7967 }, { "epoch": 10.43, "learning_rate": 1.0205000000000001e-05, "loss": 0.0035, "step": 7968 }, { "epoch": 10.43, "learning_rate": 1.02e-05, "loss": 0.0043, "step": 7969 }, { "epoch": 10.43, "learning_rate": 1.0195e-05, "loss": 0.0029, "step": 7970 }, { "epoch": 10.43, "learning_rate": 1.019e-05, "loss": 0.0083, "step": 7971 }, { "epoch": 10.43, "learning_rate": 1.0185e-05, "loss": 0.004, "step": 7972 }, { "epoch": 10.44, "learning_rate": 1.018e-05, "loss": 0.003, "step": 7973 }, { "epoch": 10.44, "learning_rate": 1.0175e-05, "loss": 0.0054, "step": 7974 }, { "epoch": 10.44, "learning_rate": 1.0170000000000001e-05, "loss": 0.0034, "step": 7975 }, { "epoch": 10.44, "learning_rate": 1.0165e-05, "loss": 0.0038, "step": 7976 }, { "epoch": 10.44, "learning_rate": 1.016e-05, "loss": 0.0034, "step": 7977 }, { "epoch": 10.44, "learning_rate": 1.0155e-05, "loss": 0.0045, "step": 7978 }, { "epoch": 10.44, "learning_rate": 1.0150000000000001e-05, "loss": 0.0066, "step": 7979 }, { "epoch": 10.45, "learning_rate": 1.0145e-05, "loss": 0.0026, "step": 7980 }, { "epoch": 10.45, "learning_rate": 1.0140000000000001e-05, "loss": 0.0037, "step": 7981 }, { "epoch": 10.45, "learning_rate": 1.0135e-05, "loss": 0.0035, "step": 7982 }, { "epoch": 10.45, "learning_rate": 1.0130000000000001e-05, "loss": 0.0033, "step": 7983 }, { "epoch": 10.45, "learning_rate": 1.0125e-05, "loss": 0.0046, "step": 7984 }, { "epoch": 10.45, "learning_rate": 1.012e-05, "loss": 0.0028, "step": 7985 }, { "epoch": 10.45, "learning_rate": 1.0115e-05, "loss": 0.0028, "step": 7986 }, { "epoch": 10.45, "learning_rate": 1.011e-05, "loss": 0.0033, "step": 7987 }, { "epoch": 10.46, "learning_rate": 1.0105e-05, "loss": 0.0026, "step": 7988 }, { "epoch": 10.46, "learning_rate": 1.0100000000000002e-05, "loss": 0.006, "step": 7989 }, { "epoch": 10.46, "learning_rate": 1.0095e-05, "loss": 0.0041, "step": 7990 }, { "epoch": 10.46, "learning_rate": 1.0090000000000002e-05, "loss": 0.004, "step": 7991 }, { "epoch": 10.46, "learning_rate": 1.0085e-05, "loss": 0.0045, "step": 7992 }, { "epoch": 10.46, "learning_rate": 1.008e-05, "loss": 0.0032, "step": 7993 }, { "epoch": 10.46, "learning_rate": 1.0075000000000001e-05, "loss": 0.004, "step": 7994 }, { "epoch": 10.46, "learning_rate": 1.007e-05, "loss": 0.0037, "step": 7995 }, { "epoch": 10.47, "learning_rate": 1.0065000000000001e-05, "loss": 0.0145, "step": 7996 }, { "epoch": 10.47, "learning_rate": 1.006e-05, "loss": 0.0047, "step": 7997 }, { "epoch": 10.47, "learning_rate": 1.0055000000000001e-05, "loss": 0.0029, "step": 7998 }, { "epoch": 10.47, "learning_rate": 1.005e-05, "loss": 0.0036, "step": 7999 }, { "epoch": 10.47, "learning_rate": 1.0045e-05, "loss": 0.0053, "step": 8000 }, { "epoch": 10.47, "learning_rate": 1.004e-05, "loss": 0.0055, "step": 8001 }, { "epoch": 10.47, "learning_rate": 1.0035e-05, "loss": 0.0038, "step": 8002 }, { "epoch": 10.48, "learning_rate": 1.003e-05, "loss": 0.003, "step": 8003 }, { "epoch": 10.48, "learning_rate": 1.0025000000000001e-05, "loss": 0.0038, "step": 8004 }, { "epoch": 10.48, "learning_rate": 1.002e-05, "loss": 0.0031, "step": 8005 }, { "epoch": 10.48, "learning_rate": 1.0015000000000002e-05, "loss": 0.0031, "step": 8006 }, { "epoch": 10.48, "learning_rate": 1.001e-05, "loss": 0.0039, "step": 8007 }, { "epoch": 10.48, "learning_rate": 1.0005e-05, "loss": 0.0049, "step": 8008 }, { "epoch": 10.48, "learning_rate": 1e-05, "loss": 0.0038, "step": 8009 }, { "epoch": 10.48, "learning_rate": 9.995e-06, "loss": 0.0028, "step": 8010 }, { "epoch": 10.49, "learning_rate": 9.990000000000001e-06, "loss": 0.0045, "step": 8011 }, { "epoch": 10.49, "learning_rate": 9.985e-06, "loss": 0.0039, "step": 8012 }, { "epoch": 10.49, "learning_rate": 9.980000000000001e-06, "loss": 0.0034, "step": 8013 }, { "epoch": 10.49, "learning_rate": 9.975e-06, "loss": 0.0039, "step": 8014 }, { "epoch": 10.49, "learning_rate": 9.97e-06, "loss": 0.0041, "step": 8015 }, { "epoch": 10.49, "learning_rate": 9.965e-06, "loss": 0.0031, "step": 8016 }, { "epoch": 10.49, "learning_rate": 9.96e-06, "loss": 0.0043, "step": 8017 }, { "epoch": 10.49, "learning_rate": 9.955e-06, "loss": 0.0048, "step": 8018 }, { "epoch": 10.5, "learning_rate": 9.950000000000001e-06, "loss": 0.0023, "step": 8019 }, { "epoch": 10.5, "learning_rate": 9.945e-06, "loss": 0.0022, "step": 8020 }, { "epoch": 10.5, "learning_rate": 9.940000000000001e-06, "loss": 0.0034, "step": 8021 }, { "epoch": 10.5, "learning_rate": 9.935e-06, "loss": 0.0047, "step": 8022 }, { "epoch": 10.5, "learning_rate": 9.93e-06, "loss": 0.004, "step": 8023 }, { "epoch": 10.5, "learning_rate": 9.925e-06, "loss": 0.0051, "step": 8024 }, { "epoch": 10.5, "learning_rate": 9.92e-06, "loss": 0.0046, "step": 8025 }, { "epoch": 10.51, "learning_rate": 9.915e-06, "loss": 0.0047, "step": 8026 }, { "epoch": 10.51, "learning_rate": 9.91e-06, "loss": 0.0021, "step": 8027 }, { "epoch": 10.51, "learning_rate": 9.905000000000001e-06, "loss": 0.0044, "step": 8028 }, { "epoch": 10.51, "learning_rate": 9.900000000000002e-06, "loss": 0.0037, "step": 8029 }, { "epoch": 10.51, "learning_rate": 9.895e-06, "loss": 0.0035, "step": 8030 }, { "epoch": 10.51, "learning_rate": 9.89e-06, "loss": 0.0055, "step": 8031 }, { "epoch": 10.51, "learning_rate": 9.885e-06, "loss": 0.0041, "step": 8032 }, { "epoch": 10.51, "learning_rate": 9.88e-06, "loss": 0.0039, "step": 8033 }, { "epoch": 10.52, "learning_rate": 9.875000000000001e-06, "loss": 0.0039, "step": 8034 }, { "epoch": 10.52, "learning_rate": 9.87e-06, "loss": 0.0035, "step": 8035 }, { "epoch": 10.52, "learning_rate": 9.865000000000001e-06, "loss": 0.0039, "step": 8036 }, { "epoch": 10.52, "learning_rate": 9.86e-06, "loss": 0.0044, "step": 8037 }, { "epoch": 10.52, "learning_rate": 9.855e-06, "loss": 0.004, "step": 8038 }, { "epoch": 10.52, "learning_rate": 9.85e-06, "loss": 0.0031, "step": 8039 }, { "epoch": 10.52, "learning_rate": 9.845e-06, "loss": 0.0042, "step": 8040 }, { "epoch": 10.52, "learning_rate": 9.84e-06, "loss": 0.0045, "step": 8041 }, { "epoch": 10.53, "learning_rate": 9.835000000000002e-06, "loss": 0.0045, "step": 8042 }, { "epoch": 10.53, "learning_rate": 9.83e-06, "loss": 0.0022, "step": 8043 }, { "epoch": 10.53, "learning_rate": 9.825000000000002e-06, "loss": 0.0047, "step": 8044 }, { "epoch": 10.53, "learning_rate": 9.820000000000001e-06, "loss": 0.0047, "step": 8045 }, { "epoch": 10.53, "learning_rate": 9.815e-06, "loss": 0.0028, "step": 8046 }, { "epoch": 10.53, "learning_rate": 9.810000000000001e-06, "loss": 0.0036, "step": 8047 }, { "epoch": 10.53, "learning_rate": 9.805e-06, "loss": 0.0027, "step": 8048 }, { "epoch": 10.54, "learning_rate": 9.800000000000001e-06, "loss": 0.0033, "step": 8049 }, { "epoch": 10.54, "learning_rate": 9.795e-06, "loss": 0.0047, "step": 8050 }, { "epoch": 10.54, "learning_rate": 9.790000000000001e-06, "loss": 0.0039, "step": 8051 }, { "epoch": 10.54, "learning_rate": 9.785e-06, "loss": 0.0028, "step": 8052 }, { "epoch": 10.54, "learning_rate": 9.78e-06, "loss": 0.0034, "step": 8053 }, { "epoch": 10.54, "learning_rate": 9.775e-06, "loss": 0.0034, "step": 8054 }, { "epoch": 10.54, "learning_rate": 9.77e-06, "loss": 0.0039, "step": 8055 }, { "epoch": 10.54, "learning_rate": 9.765e-06, "loss": 0.0031, "step": 8056 }, { "epoch": 10.55, "learning_rate": 9.760000000000001e-06, "loss": 0.0038, "step": 8057 }, { "epoch": 10.55, "learning_rate": 9.755e-06, "loss": 0.0046, "step": 8058 }, { "epoch": 10.55, "learning_rate": 9.750000000000002e-06, "loss": 0.0053, "step": 8059 }, { "epoch": 10.55, "learning_rate": 9.745e-06, "loss": 0.0027, "step": 8060 }, { "epoch": 10.55, "learning_rate": 9.74e-06, "loss": 0.0025, "step": 8061 }, { "epoch": 10.55, "learning_rate": 9.735e-06, "loss": 0.0039, "step": 8062 }, { "epoch": 10.55, "learning_rate": 9.73e-06, "loss": 0.0034, "step": 8063 }, { "epoch": 10.55, "learning_rate": 9.725000000000001e-06, "loss": 0.0026, "step": 8064 }, { "epoch": 10.56, "learning_rate": 9.72e-06, "loss": 0.0035, "step": 8065 }, { "epoch": 10.56, "learning_rate": 9.715000000000001e-06, "loss": 0.0047, "step": 8066 }, { "epoch": 10.56, "learning_rate": 9.71e-06, "loss": 0.0043, "step": 8067 }, { "epoch": 10.56, "learning_rate": 9.705e-06, "loss": 0.0034, "step": 8068 }, { "epoch": 10.56, "learning_rate": 9.7e-06, "loss": 0.004, "step": 8069 }, { "epoch": 10.56, "learning_rate": 9.695e-06, "loss": 0.0026, "step": 8070 }, { "epoch": 10.56, "learning_rate": 9.69e-06, "loss": 0.0036, "step": 8071 }, { "epoch": 10.57, "learning_rate": 9.685000000000001e-06, "loss": 0.0045, "step": 8072 }, { "epoch": 10.57, "learning_rate": 9.68e-06, "loss": 0.0027, "step": 8073 }, { "epoch": 10.57, "learning_rate": 9.675000000000001e-06, "loss": 0.0072, "step": 8074 }, { "epoch": 10.57, "learning_rate": 9.67e-06, "loss": 0.0038, "step": 8075 }, { "epoch": 10.57, "learning_rate": 9.665e-06, "loss": 0.0032, "step": 8076 }, { "epoch": 10.57, "learning_rate": 9.66e-06, "loss": 0.0032, "step": 8077 }, { "epoch": 10.57, "learning_rate": 9.655e-06, "loss": 0.0035, "step": 8078 }, { "epoch": 10.57, "learning_rate": 9.65e-06, "loss": 0.0024, "step": 8079 }, { "epoch": 10.58, "learning_rate": 9.645e-06, "loss": 0.0025, "step": 8080 }, { "epoch": 10.58, "learning_rate": 9.640000000000001e-06, "loss": 0.0038, "step": 8081 }, { "epoch": 10.58, "learning_rate": 9.635000000000002e-06, "loss": 0.0042, "step": 8082 }, { "epoch": 10.58, "learning_rate": 9.630000000000001e-06, "loss": 0.0035, "step": 8083 }, { "epoch": 10.58, "learning_rate": 9.625e-06, "loss": 0.0036, "step": 8084 }, { "epoch": 10.58, "learning_rate": 9.62e-06, "loss": 0.0066, "step": 8085 }, { "epoch": 10.58, "learning_rate": 9.615e-06, "loss": 0.0033, "step": 8086 }, { "epoch": 10.59, "learning_rate": 9.610000000000001e-06, "loss": 0.0041, "step": 8087 }, { "epoch": 10.59, "learning_rate": 9.605e-06, "loss": 0.004, "step": 8088 }, { "epoch": 10.59, "learning_rate": 9.600000000000001e-06, "loss": 0.0045, "step": 8089 }, { "epoch": 10.59, "learning_rate": 9.595e-06, "loss": 0.0039, "step": 8090 }, { "epoch": 10.59, "learning_rate": 9.59e-06, "loss": 0.0044, "step": 8091 }, { "epoch": 10.59, "learning_rate": 9.585e-06, "loss": 0.0027, "step": 8092 }, { "epoch": 10.59, "learning_rate": 9.58e-06, "loss": 0.0037, "step": 8093 }, { "epoch": 10.59, "learning_rate": 9.575e-06, "loss": 0.0047, "step": 8094 }, { "epoch": 10.6, "learning_rate": 9.57e-06, "loss": 0.0036, "step": 8095 }, { "epoch": 10.6, "learning_rate": 9.565e-06, "loss": 0.0036, "step": 8096 }, { "epoch": 10.6, "learning_rate": 9.560000000000002e-06, "loss": 0.0029, "step": 8097 }, { "epoch": 10.6, "learning_rate": 9.555e-06, "loss": 0.0027, "step": 8098 }, { "epoch": 10.6, "learning_rate": 9.55e-06, "loss": 0.0049, "step": 8099 }, { "epoch": 10.6, "learning_rate": 9.545e-06, "loss": 0.0044, "step": 8100 }, { "epoch": 10.6, "learning_rate": 9.54e-06, "loss": 0.0043, "step": 8101 }, { "epoch": 10.6, "learning_rate": 9.535000000000001e-06, "loss": 0.0039, "step": 8102 }, { "epoch": 10.61, "learning_rate": 9.53e-06, "loss": 0.0027, "step": 8103 }, { "epoch": 10.61, "learning_rate": 9.525000000000001e-06, "loss": 0.0038, "step": 8104 }, { "epoch": 10.61, "learning_rate": 9.52e-06, "loss": 0.0039, "step": 8105 }, { "epoch": 10.61, "learning_rate": 9.515e-06, "loss": 0.0032, "step": 8106 }, { "epoch": 10.61, "learning_rate": 9.51e-06, "loss": 0.004, "step": 8107 }, { "epoch": 10.61, "learning_rate": 9.505e-06, "loss": 0.0034, "step": 8108 }, { "epoch": 10.61, "learning_rate": 9.5e-06, "loss": 0.0043, "step": 8109 }, { "epoch": 10.62, "learning_rate": 9.495000000000001e-06, "loss": 0.0048, "step": 8110 }, { "epoch": 10.62, "learning_rate": 9.49e-06, "loss": 0.005, "step": 8111 }, { "epoch": 10.62, "learning_rate": 9.485000000000002e-06, "loss": 0.004, "step": 8112 }, { "epoch": 10.62, "learning_rate": 9.48e-06, "loss": 0.0042, "step": 8113 }, { "epoch": 10.62, "learning_rate": 9.475e-06, "loss": 0.007, "step": 8114 }, { "epoch": 10.62, "learning_rate": 9.47e-06, "loss": 0.0034, "step": 8115 }, { "epoch": 10.62, "learning_rate": 9.465e-06, "loss": 0.0053, "step": 8116 }, { "epoch": 10.62, "learning_rate": 9.460000000000001e-06, "loss": 0.0055, "step": 8117 }, { "epoch": 10.63, "learning_rate": 9.455e-06, "loss": 0.0049, "step": 8118 }, { "epoch": 10.63, "learning_rate": 9.450000000000001e-06, "loss": 0.005, "step": 8119 }, { "epoch": 10.63, "learning_rate": 9.445000000000002e-06, "loss": 0.0049, "step": 8120 }, { "epoch": 10.63, "learning_rate": 9.44e-06, "loss": 0.0035, "step": 8121 }, { "epoch": 10.63, "learning_rate": 9.435e-06, "loss": 0.0041, "step": 8122 }, { "epoch": 10.63, "learning_rate": 9.43e-06, "loss": 0.0025, "step": 8123 }, { "epoch": 10.63, "learning_rate": 9.425e-06, "loss": 0.0045, "step": 8124 }, { "epoch": 10.63, "learning_rate": 9.420000000000001e-06, "loss": 0.0047, "step": 8125 }, { "epoch": 10.64, "learning_rate": 9.415e-06, "loss": 0.0056, "step": 8126 }, { "epoch": 10.64, "learning_rate": 9.410000000000001e-06, "loss": 0.0046, "step": 8127 }, { "epoch": 10.64, "learning_rate": 9.405e-06, "loss": 0.0044, "step": 8128 }, { "epoch": 10.64, "learning_rate": 9.4e-06, "loss": 0.0059, "step": 8129 }, { "epoch": 10.64, "learning_rate": 9.395e-06, "loss": 0.0042, "step": 8130 }, { "epoch": 10.64, "learning_rate": 9.39e-06, "loss": 0.0093, "step": 8131 }, { "epoch": 10.64, "learning_rate": 9.385e-06, "loss": 0.004, "step": 8132 }, { "epoch": 10.65, "learning_rate": 9.38e-06, "loss": 0.0047, "step": 8133 }, { "epoch": 10.65, "learning_rate": 9.375000000000001e-06, "loss": 0.0022, "step": 8134 }, { "epoch": 10.65, "learning_rate": 9.370000000000002e-06, "loss": 0.0035, "step": 8135 }, { "epoch": 10.65, "learning_rate": 9.365000000000001e-06, "loss": 0.0024, "step": 8136 }, { "epoch": 10.65, "learning_rate": 9.36e-06, "loss": 0.0051, "step": 8137 }, { "epoch": 10.65, "learning_rate": 9.355e-06, "loss": 0.0033, "step": 8138 }, { "epoch": 10.65, "learning_rate": 9.35e-06, "loss": 0.004, "step": 8139 }, { "epoch": 10.65, "learning_rate": 9.345000000000001e-06, "loss": 0.0027, "step": 8140 }, { "epoch": 10.66, "learning_rate": 9.34e-06, "loss": 0.0032, "step": 8141 }, { "epoch": 10.66, "learning_rate": 9.335000000000001e-06, "loss": 0.0034, "step": 8142 }, { "epoch": 10.66, "learning_rate": 9.33e-06, "loss": 0.0035, "step": 8143 }, { "epoch": 10.66, "learning_rate": 9.325e-06, "loss": 0.0048, "step": 8144 }, { "epoch": 10.66, "learning_rate": 9.32e-06, "loss": 0.0044, "step": 8145 }, { "epoch": 10.66, "learning_rate": 9.315e-06, "loss": 0.0048, "step": 8146 }, { "epoch": 10.66, "learning_rate": 9.31e-06, "loss": 0.0048, "step": 8147 }, { "epoch": 10.66, "learning_rate": 9.305e-06, "loss": 0.004, "step": 8148 }, { "epoch": 10.67, "learning_rate": 9.3e-06, "loss": 0.004, "step": 8149 }, { "epoch": 10.67, "learning_rate": 9.295000000000002e-06, "loss": 0.0043, "step": 8150 }, { "epoch": 10.67, "learning_rate": 9.29e-06, "loss": 0.0024, "step": 8151 }, { "epoch": 10.67, "learning_rate": 9.285e-06, "loss": 0.0038, "step": 8152 }, { "epoch": 10.67, "learning_rate": 9.28e-06, "loss": 0.0052, "step": 8153 }, { "epoch": 10.67, "learning_rate": 9.275e-06, "loss": 0.0049, "step": 8154 }, { "epoch": 10.67, "learning_rate": 9.270000000000001e-06, "loss": 0.0058, "step": 8155 }, { "epoch": 10.68, "learning_rate": 9.265e-06, "loss": 0.003, "step": 8156 }, { "epoch": 10.68, "learning_rate": 9.260000000000001e-06, "loss": 0.0037, "step": 8157 }, { "epoch": 10.68, "learning_rate": 9.255e-06, "loss": 0.0048, "step": 8158 }, { "epoch": 10.68, "learning_rate": 9.25e-06, "loss": 0.0047, "step": 8159 }, { "epoch": 10.68, "learning_rate": 9.245e-06, "loss": 0.0028, "step": 8160 }, { "epoch": 10.68, "learning_rate": 9.24e-06, "loss": 0.0034, "step": 8161 }, { "epoch": 10.68, "learning_rate": 9.235e-06, "loss": 0.0055, "step": 8162 }, { "epoch": 10.68, "learning_rate": 9.23e-06, "loss": 0.0043, "step": 8163 }, { "epoch": 10.69, "learning_rate": 9.225e-06, "loss": 0.0052, "step": 8164 }, { "epoch": 10.69, "learning_rate": 9.220000000000002e-06, "loss": 0.003, "step": 8165 }, { "epoch": 10.69, "learning_rate": 9.215e-06, "loss": 0.0038, "step": 8166 }, { "epoch": 10.69, "learning_rate": 9.21e-06, "loss": 0.0047, "step": 8167 }, { "epoch": 10.69, "learning_rate": 9.205e-06, "loss": 0.0036, "step": 8168 }, { "epoch": 10.69, "learning_rate": 9.2e-06, "loss": 0.006, "step": 8169 }, { "epoch": 10.69, "learning_rate": 9.195000000000001e-06, "loss": 0.0032, "step": 8170 }, { "epoch": 10.7, "learning_rate": 9.19e-06, "loss": 0.0054, "step": 8171 }, { "epoch": 10.7, "learning_rate": 9.185000000000001e-06, "loss": 0.0034, "step": 8172 }, { "epoch": 10.7, "learning_rate": 9.180000000000002e-06, "loss": 0.0046, "step": 8173 }, { "epoch": 10.7, "learning_rate": 9.175000000000001e-06, "loss": 0.0036, "step": 8174 }, { "epoch": 10.7, "learning_rate": 9.17e-06, "loss": 0.0056, "step": 8175 }, { "epoch": 10.7, "learning_rate": 9.165e-06, "loss": 0.0038, "step": 8176 }, { "epoch": 10.7, "learning_rate": 9.16e-06, "loss": 0.0052, "step": 8177 }, { "epoch": 10.7, "learning_rate": 9.155000000000001e-06, "loss": 0.0028, "step": 8178 }, { "epoch": 10.71, "learning_rate": 9.15e-06, "loss": 0.0042, "step": 8179 }, { "epoch": 10.71, "learning_rate": 9.145000000000001e-06, "loss": 0.004, "step": 8180 }, { "epoch": 10.71, "learning_rate": 9.14e-06, "loss": 0.0051, "step": 8181 }, { "epoch": 10.71, "learning_rate": 9.135e-06, "loss": 0.0033, "step": 8182 }, { "epoch": 10.71, "learning_rate": 9.13e-06, "loss": 0.0033, "step": 8183 }, { "epoch": 10.71, "learning_rate": 9.125e-06, "loss": 0.0032, "step": 8184 }, { "epoch": 10.71, "learning_rate": 9.12e-06, "loss": 0.0032, "step": 8185 }, { "epoch": 10.71, "learning_rate": 9.115e-06, "loss": 0.0032, "step": 8186 }, { "epoch": 10.72, "learning_rate": 9.110000000000001e-06, "loss": 0.0045, "step": 8187 }, { "epoch": 10.72, "learning_rate": 9.105000000000002e-06, "loss": 0.0028, "step": 8188 }, { "epoch": 10.72, "learning_rate": 9.100000000000001e-06, "loss": 0.0039, "step": 8189 }, { "epoch": 10.72, "learning_rate": 9.095e-06, "loss": 0.0029, "step": 8190 }, { "epoch": 10.72, "learning_rate": 9.09e-06, "loss": 0.004, "step": 8191 }, { "epoch": 10.72, "learning_rate": 9.085e-06, "loss": 0.0032, "step": 8192 }, { "epoch": 10.72, "learning_rate": 9.080000000000001e-06, "loss": 0.0032, "step": 8193 }, { "epoch": 10.73, "learning_rate": 9.075e-06, "loss": 0.0034, "step": 8194 }, { "epoch": 10.73, "learning_rate": 9.070000000000001e-06, "loss": 0.004, "step": 8195 }, { "epoch": 10.73, "learning_rate": 9.065e-06, "loss": 0.0034, "step": 8196 }, { "epoch": 10.73, "learning_rate": 9.06e-06, "loss": 0.0039, "step": 8197 }, { "epoch": 10.73, "learning_rate": 9.055e-06, "loss": 0.0466, "step": 8198 }, { "epoch": 10.73, "learning_rate": 9.05e-06, "loss": 0.003, "step": 8199 }, { "epoch": 10.73, "learning_rate": 9.045e-06, "loss": 0.0037, "step": 8200 }, { "epoch": 10.73, "learning_rate": 9.04e-06, "loss": 0.0034, "step": 8201 }, { "epoch": 10.74, "learning_rate": 9.035e-06, "loss": 0.0031, "step": 8202 }, { "epoch": 10.74, "learning_rate": 9.030000000000002e-06, "loss": 0.0046, "step": 8203 }, { "epoch": 10.74, "learning_rate": 9.025e-06, "loss": 0.0057, "step": 8204 }, { "epoch": 10.74, "learning_rate": 9.02e-06, "loss": 0.0039, "step": 8205 }, { "epoch": 10.74, "learning_rate": 9.015e-06, "loss": 0.0037, "step": 8206 }, { "epoch": 10.74, "learning_rate": 9.01e-06, "loss": 0.0045, "step": 8207 }, { "epoch": 10.74, "learning_rate": 9.005000000000001e-06, "loss": 0.0049, "step": 8208 }, { "epoch": 10.74, "learning_rate": 9e-06, "loss": 0.0046, "step": 8209 }, { "epoch": 10.75, "learning_rate": 8.995000000000001e-06, "loss": 0.004, "step": 8210 }, { "epoch": 10.75, "learning_rate": 8.99e-06, "loss": 0.0038, "step": 8211 }, { "epoch": 10.75, "learning_rate": 8.985e-06, "loss": 0.004, "step": 8212 }, { "epoch": 10.75, "learning_rate": 8.98e-06, "loss": 0.0036, "step": 8213 }, { "epoch": 10.75, "learning_rate": 8.975e-06, "loss": 0.0061, "step": 8214 }, { "epoch": 10.75, "learning_rate": 8.97e-06, "loss": 0.0074, "step": 8215 }, { "epoch": 10.75, "learning_rate": 8.965e-06, "loss": 0.0034, "step": 8216 }, { "epoch": 10.76, "learning_rate": 8.96e-06, "loss": 0.0025, "step": 8217 }, { "epoch": 10.76, "learning_rate": 8.955000000000002e-06, "loss": 0.0046, "step": 8218 }, { "epoch": 10.76, "learning_rate": 8.95e-06, "loss": 0.0027, "step": 8219 }, { "epoch": 10.76, "learning_rate": 8.945e-06, "loss": 0.0036, "step": 8220 }, { "epoch": 10.76, "learning_rate": 8.939999999999999e-06, "loss": 0.0059, "step": 8221 }, { "epoch": 10.76, "learning_rate": 8.935e-06, "loss": 0.0051, "step": 8222 }, { "epoch": 10.76, "learning_rate": 8.930000000000001e-06, "loss": 0.0032, "step": 8223 }, { "epoch": 10.76, "learning_rate": 8.925e-06, "loss": 0.0027, "step": 8224 }, { "epoch": 10.77, "learning_rate": 8.920000000000001e-06, "loss": 0.004, "step": 8225 }, { "epoch": 10.77, "learning_rate": 8.915e-06, "loss": 0.0065, "step": 8226 }, { "epoch": 10.77, "learning_rate": 8.910000000000001e-06, "loss": 0.0051, "step": 8227 }, { "epoch": 10.77, "learning_rate": 8.905e-06, "loss": 0.0035, "step": 8228 }, { "epoch": 10.77, "learning_rate": 8.9e-06, "loss": 0.0044, "step": 8229 }, { "epoch": 10.77, "learning_rate": 8.895e-06, "loss": 0.0045, "step": 8230 }, { "epoch": 10.77, "learning_rate": 8.890000000000001e-06, "loss": 0.0029, "step": 8231 }, { "epoch": 10.77, "learning_rate": 8.885e-06, "loss": 0.0037, "step": 8232 }, { "epoch": 10.78, "learning_rate": 8.880000000000001e-06, "loss": 0.0046, "step": 8233 }, { "epoch": 10.78, "learning_rate": 8.875e-06, "loss": 0.0034, "step": 8234 }, { "epoch": 10.78, "learning_rate": 8.87e-06, "loss": 0.0059, "step": 8235 }, { "epoch": 10.78, "learning_rate": 8.865e-06, "loss": 0.0037, "step": 8236 }, { "epoch": 10.78, "learning_rate": 8.86e-06, "loss": 0.0026, "step": 8237 }, { "epoch": 10.78, "learning_rate": 8.855e-06, "loss": 0.0044, "step": 8238 }, { "epoch": 10.78, "learning_rate": 8.85e-06, "loss": 0.0025, "step": 8239 }, { "epoch": 10.79, "learning_rate": 8.845000000000001e-06, "loss": 0.0031, "step": 8240 }, { "epoch": 10.79, "learning_rate": 8.840000000000002e-06, "loss": 0.0051, "step": 8241 }, { "epoch": 10.79, "learning_rate": 8.835000000000001e-06, "loss": 0.0063, "step": 8242 }, { "epoch": 10.79, "learning_rate": 8.83e-06, "loss": 0.0034, "step": 8243 }, { "epoch": 10.79, "learning_rate": 8.825e-06, "loss": 0.0036, "step": 8244 }, { "epoch": 10.79, "learning_rate": 8.82e-06, "loss": 0.0039, "step": 8245 }, { "epoch": 10.79, "learning_rate": 8.815000000000001e-06, "loss": 0.0035, "step": 8246 }, { "epoch": 10.79, "learning_rate": 8.81e-06, "loss": 0.0025, "step": 8247 }, { "epoch": 10.8, "learning_rate": 8.805000000000001e-06, "loss": 0.0044, "step": 8248 }, { "epoch": 10.8, "learning_rate": 8.8e-06, "loss": 0.0042, "step": 8249 }, { "epoch": 10.8, "learning_rate": 8.795e-06, "loss": 0.0043, "step": 8250 }, { "epoch": 10.8, "learning_rate": 8.79e-06, "loss": 0.0033, "step": 8251 }, { "epoch": 10.8, "learning_rate": 8.785e-06, "loss": 0.008, "step": 8252 }, { "epoch": 10.8, "learning_rate": 8.78e-06, "loss": 0.003, "step": 8253 }, { "epoch": 10.8, "learning_rate": 8.775e-06, "loss": 0.0034, "step": 8254 }, { "epoch": 10.8, "learning_rate": 8.77e-06, "loss": 0.0045, "step": 8255 }, { "epoch": 10.81, "learning_rate": 8.765000000000002e-06, "loss": 0.0031, "step": 8256 }, { "epoch": 10.81, "learning_rate": 8.76e-06, "loss": 0.0041, "step": 8257 }, { "epoch": 10.81, "learning_rate": 8.755e-06, "loss": 0.0031, "step": 8258 }, { "epoch": 10.81, "learning_rate": 8.75e-06, "loss": 0.0043, "step": 8259 }, { "epoch": 10.81, "learning_rate": 8.745e-06, "loss": 0.0044, "step": 8260 }, { "epoch": 10.81, "learning_rate": 8.740000000000001e-06, "loss": 0.0033, "step": 8261 }, { "epoch": 10.81, "learning_rate": 8.735e-06, "loss": 0.004, "step": 8262 }, { "epoch": 10.82, "learning_rate": 8.730000000000001e-06, "loss": 0.0034, "step": 8263 }, { "epoch": 10.82, "learning_rate": 8.725e-06, "loss": 0.0046, "step": 8264 }, { "epoch": 10.82, "learning_rate": 8.720000000000001e-06, "loss": 0.0038, "step": 8265 }, { "epoch": 10.82, "learning_rate": 8.715e-06, "loss": 0.0037, "step": 8266 }, { "epoch": 10.82, "learning_rate": 8.71e-06, "loss": 0.0028, "step": 8267 }, { "epoch": 10.82, "learning_rate": 8.705e-06, "loss": 0.0037, "step": 8268 }, { "epoch": 10.82, "learning_rate": 8.7e-06, "loss": 0.0044, "step": 8269 }, { "epoch": 10.82, "learning_rate": 8.695e-06, "loss": 0.0056, "step": 8270 }, { "epoch": 10.83, "learning_rate": 8.690000000000002e-06, "loss": 0.0039, "step": 8271 }, { "epoch": 10.83, "learning_rate": 8.685e-06, "loss": 0.0029, "step": 8272 }, { "epoch": 10.83, "learning_rate": 8.68e-06, "loss": 0.0035, "step": 8273 }, { "epoch": 10.83, "learning_rate": 8.674999999999999e-06, "loss": 0.0026, "step": 8274 }, { "epoch": 10.83, "learning_rate": 8.67e-06, "loss": 0.0035, "step": 8275 }, { "epoch": 10.83, "learning_rate": 8.665000000000001e-06, "loss": 0.0046, "step": 8276 }, { "epoch": 10.83, "learning_rate": 8.66e-06, "loss": 0.0036, "step": 8277 }, { "epoch": 10.84, "learning_rate": 8.655000000000001e-06, "loss": 0.0041, "step": 8278 }, { "epoch": 10.84, "learning_rate": 8.65e-06, "loss": 0.0044, "step": 8279 }, { "epoch": 10.84, "learning_rate": 8.645000000000001e-06, "loss": 0.0048, "step": 8280 }, { "epoch": 10.84, "learning_rate": 8.64e-06, "loss": 0.0057, "step": 8281 }, { "epoch": 10.84, "learning_rate": 8.635e-06, "loss": 0.0043, "step": 8282 }, { "epoch": 10.84, "learning_rate": 8.63e-06, "loss": 0.0022, "step": 8283 }, { "epoch": 10.84, "learning_rate": 8.625e-06, "loss": 0.0029, "step": 8284 }, { "epoch": 10.84, "learning_rate": 8.62e-06, "loss": 0.0031, "step": 8285 }, { "epoch": 10.85, "learning_rate": 8.615000000000001e-06, "loss": 0.0076, "step": 8286 }, { "epoch": 10.85, "learning_rate": 8.61e-06, "loss": 0.0043, "step": 8287 }, { "epoch": 10.85, "learning_rate": 8.605e-06, "loss": 0.0035, "step": 8288 }, { "epoch": 10.85, "learning_rate": 8.599999999999999e-06, "loss": 0.0025, "step": 8289 }, { "epoch": 10.85, "learning_rate": 8.595e-06, "loss": 0.0037, "step": 8290 }, { "epoch": 10.85, "learning_rate": 8.59e-06, "loss": 0.0032, "step": 8291 }, { "epoch": 10.85, "learning_rate": 8.585e-06, "loss": 0.0021, "step": 8292 }, { "epoch": 10.85, "learning_rate": 8.580000000000001e-06, "loss": 0.0027, "step": 8293 }, { "epoch": 10.86, "learning_rate": 8.575000000000002e-06, "loss": 0.0047, "step": 8294 }, { "epoch": 10.86, "learning_rate": 8.570000000000001e-06, "loss": 0.004, "step": 8295 }, { "epoch": 10.86, "learning_rate": 8.565e-06, "loss": 0.0051, "step": 8296 }, { "epoch": 10.86, "learning_rate": 8.56e-06, "loss": 0.0081, "step": 8297 }, { "epoch": 10.86, "learning_rate": 8.555e-06, "loss": 0.0045, "step": 8298 }, { "epoch": 10.86, "learning_rate": 8.550000000000001e-06, "loss": 0.0033, "step": 8299 }, { "epoch": 10.86, "learning_rate": 8.545e-06, "loss": 0.005, "step": 8300 }, { "epoch": 10.87, "learning_rate": 8.540000000000001e-06, "loss": 0.0036, "step": 8301 }, { "epoch": 10.87, "learning_rate": 8.535e-06, "loss": 0.0054, "step": 8302 }, { "epoch": 10.87, "learning_rate": 8.53e-06, "loss": 0.0032, "step": 8303 }, { "epoch": 10.87, "learning_rate": 8.525e-06, "loss": 0.0045, "step": 8304 }, { "epoch": 10.87, "learning_rate": 8.52e-06, "loss": 0.0044, "step": 8305 }, { "epoch": 10.87, "learning_rate": 8.515e-06, "loss": 0.0027, "step": 8306 }, { "epoch": 10.87, "learning_rate": 8.51e-06, "loss": 0.004, "step": 8307 }, { "epoch": 10.87, "learning_rate": 8.505e-06, "loss": 0.0048, "step": 8308 }, { "epoch": 10.88, "learning_rate": 8.500000000000002e-06, "loss": 0.0035, "step": 8309 }, { "epoch": 10.88, "learning_rate": 8.495e-06, "loss": 0.004, "step": 8310 }, { "epoch": 10.88, "learning_rate": 8.49e-06, "loss": 0.004, "step": 8311 }, { "epoch": 10.88, "learning_rate": 8.485e-06, "loss": 0.0032, "step": 8312 }, { "epoch": 10.88, "learning_rate": 8.48e-06, "loss": 0.004, "step": 8313 }, { "epoch": 10.88, "learning_rate": 8.475000000000001e-06, "loss": 0.0031, "step": 8314 }, { "epoch": 10.88, "learning_rate": 8.47e-06, "loss": 0.0036, "step": 8315 }, { "epoch": 10.88, "learning_rate": 8.465000000000001e-06, "loss": 0.0047, "step": 8316 }, { "epoch": 10.89, "learning_rate": 8.46e-06, "loss": 0.0031, "step": 8317 }, { "epoch": 10.89, "learning_rate": 8.455000000000001e-06, "loss": 0.0046, "step": 8318 }, { "epoch": 10.89, "learning_rate": 8.45e-06, "loss": 0.0116, "step": 8319 }, { "epoch": 10.89, "learning_rate": 8.445e-06, "loss": 0.004, "step": 8320 }, { "epoch": 10.89, "learning_rate": 8.44e-06, "loss": 0.003, "step": 8321 }, { "epoch": 10.89, "learning_rate": 8.435e-06, "loss": 0.0032, "step": 8322 }, { "epoch": 10.89, "learning_rate": 8.43e-06, "loss": 0.0045, "step": 8323 }, { "epoch": 10.9, "learning_rate": 8.425000000000001e-06, "loss": 0.0043, "step": 8324 }, { "epoch": 10.9, "learning_rate": 8.42e-06, "loss": 0.0031, "step": 8325 }, { "epoch": 10.9, "learning_rate": 8.415e-06, "loss": 0.0034, "step": 8326 }, { "epoch": 10.9, "learning_rate": 8.409999999999999e-06, "loss": 0.0039, "step": 8327 }, { "epoch": 10.9, "learning_rate": 8.405e-06, "loss": 0.0038, "step": 8328 }, { "epoch": 10.9, "learning_rate": 8.400000000000001e-06, "loss": 0.0044, "step": 8329 }, { "epoch": 10.9, "learning_rate": 8.395e-06, "loss": 0.0031, "step": 8330 }, { "epoch": 10.9, "learning_rate": 8.390000000000001e-06, "loss": 0.0034, "step": 8331 }, { "epoch": 10.91, "learning_rate": 8.385e-06, "loss": 0.0046, "step": 8332 }, { "epoch": 10.91, "learning_rate": 8.380000000000001e-06, "loss": 0.0045, "step": 8333 }, { "epoch": 10.91, "learning_rate": 8.375e-06, "loss": 0.0056, "step": 8334 }, { "epoch": 10.91, "learning_rate": 8.37e-06, "loss": 0.0053, "step": 8335 }, { "epoch": 10.91, "learning_rate": 8.365e-06, "loss": 0.0033, "step": 8336 }, { "epoch": 10.91, "learning_rate": 8.36e-06, "loss": 0.004, "step": 8337 }, { "epoch": 10.91, "learning_rate": 8.355e-06, "loss": 0.0027, "step": 8338 }, { "epoch": 10.91, "learning_rate": 8.350000000000001e-06, "loss": 0.0035, "step": 8339 }, { "epoch": 10.92, "learning_rate": 8.345e-06, "loss": 0.0045, "step": 8340 }, { "epoch": 10.92, "learning_rate": 8.34e-06, "loss": 0.0035, "step": 8341 }, { "epoch": 10.92, "learning_rate": 8.334999999999999e-06, "loss": 0.0035, "step": 8342 }, { "epoch": 10.92, "learning_rate": 8.33e-06, "loss": 0.0069, "step": 8343 }, { "epoch": 10.92, "learning_rate": 8.325e-06, "loss": 0.0047, "step": 8344 }, { "epoch": 10.92, "learning_rate": 8.32e-06, "loss": 0.0033, "step": 8345 }, { "epoch": 10.92, "learning_rate": 8.315000000000001e-06, "loss": 0.0044, "step": 8346 }, { "epoch": 10.93, "learning_rate": 8.31e-06, "loss": 0.0024, "step": 8347 }, { "epoch": 10.93, "learning_rate": 8.305000000000001e-06, "loss": 0.0034, "step": 8348 }, { "epoch": 10.93, "learning_rate": 8.3e-06, "loss": 0.0038, "step": 8349 }, { "epoch": 10.93, "learning_rate": 8.295e-06, "loss": 0.0029, "step": 8350 }, { "epoch": 10.93, "learning_rate": 8.29e-06, "loss": 0.0034, "step": 8351 }, { "epoch": 10.93, "learning_rate": 8.285e-06, "loss": 0.0036, "step": 8352 }, { "epoch": 10.93, "learning_rate": 8.28e-06, "loss": 0.0037, "step": 8353 }, { "epoch": 10.93, "learning_rate": 8.275000000000001e-06, "loss": 0.0025, "step": 8354 }, { "epoch": 10.94, "learning_rate": 8.27e-06, "loss": 0.0053, "step": 8355 }, { "epoch": 10.94, "learning_rate": 8.265000000000001e-06, "loss": 0.0024, "step": 8356 }, { "epoch": 10.94, "learning_rate": 8.26e-06, "loss": 0.0028, "step": 8357 }, { "epoch": 10.94, "learning_rate": 8.255e-06, "loss": 0.0028, "step": 8358 }, { "epoch": 10.94, "learning_rate": 8.25e-06, "loss": 0.0036, "step": 8359 }, { "epoch": 10.94, "learning_rate": 8.245e-06, "loss": 0.0048, "step": 8360 }, { "epoch": 10.94, "learning_rate": 8.24e-06, "loss": 0.0037, "step": 8361 }, { "epoch": 10.95, "learning_rate": 8.235000000000002e-06, "loss": 0.0031, "step": 8362 }, { "epoch": 10.95, "learning_rate": 8.23e-06, "loss": 0.0037, "step": 8363 }, { "epoch": 10.95, "learning_rate": 8.225e-06, "loss": 0.0028, "step": 8364 }, { "epoch": 10.95, "learning_rate": 8.22e-06, "loss": 0.0039, "step": 8365 }, { "epoch": 10.95, "learning_rate": 8.215e-06, "loss": 0.0018, "step": 8366 }, { "epoch": 10.95, "learning_rate": 8.210000000000001e-06, "loss": 0.0036, "step": 8367 }, { "epoch": 10.95, "learning_rate": 8.205e-06, "loss": 0.0034, "step": 8368 }, { "epoch": 10.95, "learning_rate": 8.200000000000001e-06, "loss": 0.003, "step": 8369 }, { "epoch": 10.96, "learning_rate": 8.195e-06, "loss": 0.004, "step": 8370 }, { "epoch": 10.96, "learning_rate": 8.190000000000001e-06, "loss": 0.0054, "step": 8371 }, { "epoch": 10.96, "learning_rate": 8.185e-06, "loss": 0.003, "step": 8372 }, { "epoch": 10.96, "learning_rate": 8.18e-06, "loss": 0.004, "step": 8373 }, { "epoch": 10.96, "learning_rate": 8.175e-06, "loss": 0.0033, "step": 8374 }, { "epoch": 10.96, "learning_rate": 8.17e-06, "loss": 0.0041, "step": 8375 }, { "epoch": 10.96, "learning_rate": 8.165e-06, "loss": 0.0032, "step": 8376 }, { "epoch": 10.96, "learning_rate": 8.160000000000001e-06, "loss": 0.0037, "step": 8377 }, { "epoch": 10.97, "learning_rate": 8.155e-06, "loss": 0.0024, "step": 8378 }, { "epoch": 10.97, "learning_rate": 8.15e-06, "loss": 0.0024, "step": 8379 }, { "epoch": 10.97, "learning_rate": 8.144999999999999e-06, "loss": 0.0033, "step": 8380 }, { "epoch": 10.97, "learning_rate": 8.14e-06, "loss": 0.0035, "step": 8381 }, { "epoch": 10.97, "learning_rate": 8.135000000000001e-06, "loss": 0.0042, "step": 8382 }, { "epoch": 10.97, "learning_rate": 8.13e-06, "loss": 0.0031, "step": 8383 }, { "epoch": 10.97, "learning_rate": 8.125000000000001e-06, "loss": 0.0022, "step": 8384 }, { "epoch": 10.98, "learning_rate": 8.12e-06, "loss": 0.0032, "step": 8385 }, { "epoch": 10.98, "learning_rate": 8.115000000000001e-06, "loss": 0.0025, "step": 8386 }, { "epoch": 10.98, "learning_rate": 8.11e-06, "loss": 0.0039, "step": 8387 }, { "epoch": 10.98, "learning_rate": 8.105e-06, "loss": 0.0032, "step": 8388 }, { "epoch": 10.98, "learning_rate": 8.1e-06, "loss": 0.003, "step": 8389 }, { "epoch": 10.98, "learning_rate": 8.095e-06, "loss": 0.0043, "step": 8390 }, { "epoch": 10.98, "learning_rate": 8.09e-06, "loss": 0.0036, "step": 8391 }, { "epoch": 10.98, "learning_rate": 8.085000000000001e-06, "loss": 0.0051, "step": 8392 }, { "epoch": 10.99, "learning_rate": 8.08e-06, "loss": 0.0045, "step": 8393 }, { "epoch": 10.99, "learning_rate": 8.075000000000001e-06, "loss": 0.0034, "step": 8394 }, { "epoch": 10.99, "learning_rate": 8.069999999999999e-06, "loss": 0.0038, "step": 8395 }, { "epoch": 10.99, "learning_rate": 8.065e-06, "loss": 0.0036, "step": 8396 }, { "epoch": 10.99, "learning_rate": 8.06e-06, "loss": 0.0044, "step": 8397 }, { "epoch": 10.99, "learning_rate": 8.055e-06, "loss": 0.0028, "step": 8398 }, { "epoch": 10.99, "learning_rate": 8.050000000000001e-06, "loss": 0.0028, "step": 8399 }, { "epoch": 10.99, "learning_rate": 8.045e-06, "loss": 0.0046, "step": 8400 }, { "epoch": 11.0, "learning_rate": 8.040000000000001e-06, "loss": 0.0029, "step": 8401 }, { "epoch": 11.0, "learning_rate": 8.035e-06, "loss": 0.002, "step": 8402 }, { "epoch": 11.0, "learning_rate": 8.03e-06, "loss": 0.0032, "step": 8403 }, { "epoch": 11.0, "learning_rate": 8.025e-06, "loss": 0.0036, "step": 8404 }, { "epoch": 11.0, "learning_rate": 8.02e-06, "loss": 0.0034, "step": 8405 }, { "epoch": 11.0, "learning_rate": 8.015e-06, "loss": 0.0039, "step": 8406 }, { "epoch": 11.0, "learning_rate": 8.010000000000001e-06, "loss": 0.0027, "step": 8407 }, { "epoch": 11.01, "learning_rate": 8.005e-06, "loss": 0.0034, "step": 8408 }, { "epoch": 11.01, "learning_rate": 8.000000000000001e-06, "loss": 0.0032, "step": 8409 }, { "epoch": 11.01, "learning_rate": 7.995e-06, "loss": 0.0032, "step": 8410 }, { "epoch": 11.01, "learning_rate": 7.99e-06, "loss": 0.003, "step": 8411 }, { "epoch": 11.01, "learning_rate": 7.985e-06, "loss": 0.003, "step": 8412 }, { "epoch": 11.01, "learning_rate": 7.98e-06, "loss": 0.0017, "step": 8413 }, { "epoch": 11.01, "learning_rate": 7.975e-06, "loss": 0.0037, "step": 8414 }, { "epoch": 11.01, "learning_rate": 7.97e-06, "loss": 0.0035, "step": 8415 }, { "epoch": 11.02, "learning_rate": 7.965e-06, "loss": 0.0016, "step": 8416 }, { "epoch": 11.02, "learning_rate": 7.96e-06, "loss": 0.0024, "step": 8417 }, { "epoch": 11.02, "learning_rate": 7.955e-06, "loss": 0.0027, "step": 8418 }, { "epoch": 11.02, "learning_rate": 7.95e-06, "loss": 0.0025, "step": 8419 }, { "epoch": 11.02, "learning_rate": 7.945000000000001e-06, "loss": 0.0023, "step": 8420 }, { "epoch": 11.02, "learning_rate": 7.94e-06, "loss": 0.0023, "step": 8421 }, { "epoch": 11.02, "learning_rate": 7.935000000000001e-06, "loss": 0.0034, "step": 8422 }, { "epoch": 11.02, "learning_rate": 7.93e-06, "loss": 0.0026, "step": 8423 }, { "epoch": 11.03, "learning_rate": 7.925000000000001e-06, "loss": 0.0021, "step": 8424 }, { "epoch": 11.03, "learning_rate": 7.92e-06, "loss": 0.0037, "step": 8425 }, { "epoch": 11.03, "learning_rate": 7.915e-06, "loss": 0.0027, "step": 8426 }, { "epoch": 11.03, "learning_rate": 7.91e-06, "loss": 0.0033, "step": 8427 }, { "epoch": 11.03, "learning_rate": 7.905e-06, "loss": 0.0035, "step": 8428 }, { "epoch": 11.03, "learning_rate": 7.9e-06, "loss": 0.003, "step": 8429 }, { "epoch": 11.03, "learning_rate": 7.895000000000001e-06, "loss": 0.0028, "step": 8430 }, { "epoch": 11.04, "learning_rate": 7.89e-06, "loss": 0.0035, "step": 8431 }, { "epoch": 11.04, "learning_rate": 7.885e-06, "loss": 0.0028, "step": 8432 }, { "epoch": 11.04, "learning_rate": 7.879999999999999e-06, "loss": 0.0022, "step": 8433 }, { "epoch": 11.04, "learning_rate": 7.875e-06, "loss": 0.0024, "step": 8434 }, { "epoch": 11.04, "learning_rate": 7.870000000000001e-06, "loss": 0.0032, "step": 8435 }, { "epoch": 11.04, "learning_rate": 7.865e-06, "loss": 0.0017, "step": 8436 }, { "epoch": 11.04, "learning_rate": 7.860000000000001e-06, "loss": 0.0024, "step": 8437 }, { "epoch": 11.04, "learning_rate": 7.855e-06, "loss": 0.0041, "step": 8438 }, { "epoch": 11.05, "learning_rate": 7.850000000000001e-06, "loss": 0.0055, "step": 8439 }, { "epoch": 11.05, "learning_rate": 7.845e-06, "loss": 0.0027, "step": 8440 }, { "epoch": 11.05, "learning_rate": 7.84e-06, "loss": 0.0028, "step": 8441 }, { "epoch": 11.05, "learning_rate": 7.835e-06, "loss": 0.0038, "step": 8442 }, { "epoch": 11.05, "learning_rate": 7.83e-06, "loss": 0.0037, "step": 8443 }, { "epoch": 11.05, "learning_rate": 7.825e-06, "loss": 0.0032, "step": 8444 }, { "epoch": 11.05, "learning_rate": 7.820000000000001e-06, "loss": 0.0028, "step": 8445 }, { "epoch": 11.05, "learning_rate": 7.815e-06, "loss": 0.0035, "step": 8446 }, { "epoch": 11.06, "learning_rate": 7.810000000000001e-06, "loss": 0.0024, "step": 8447 }, { "epoch": 11.06, "learning_rate": 7.805e-06, "loss": 0.0027, "step": 8448 }, { "epoch": 11.06, "learning_rate": 7.8e-06, "loss": 0.0021, "step": 8449 }, { "epoch": 11.06, "learning_rate": 7.795e-06, "loss": 0.0023, "step": 8450 }, { "epoch": 11.06, "learning_rate": 7.79e-06, "loss": 0.0025, "step": 8451 }, { "epoch": 11.06, "learning_rate": 7.785000000000001e-06, "loss": 0.0025, "step": 8452 }, { "epoch": 11.06, "learning_rate": 7.78e-06, "loss": 0.0032, "step": 8453 }, { "epoch": 11.07, "learning_rate": 7.775000000000001e-06, "loss": 0.002, "step": 8454 }, { "epoch": 11.07, "learning_rate": 7.77e-06, "loss": 0.0046, "step": 8455 }, { "epoch": 11.07, "learning_rate": 7.765e-06, "loss": 0.0023, "step": 8456 }, { "epoch": 11.07, "learning_rate": 7.76e-06, "loss": 0.0028, "step": 8457 }, { "epoch": 11.07, "learning_rate": 7.755e-06, "loss": 0.0027, "step": 8458 }, { "epoch": 11.07, "learning_rate": 7.75e-06, "loss": 0.0022, "step": 8459 }, { "epoch": 11.07, "learning_rate": 7.745000000000001e-06, "loss": 0.0041, "step": 8460 }, { "epoch": 11.07, "learning_rate": 7.74e-06, "loss": 0.0021, "step": 8461 }, { "epoch": 11.08, "learning_rate": 7.735000000000001e-06, "loss": 0.0021, "step": 8462 }, { "epoch": 11.08, "learning_rate": 7.73e-06, "loss": 0.0044, "step": 8463 }, { "epoch": 11.08, "learning_rate": 7.725e-06, "loss": 0.0022, "step": 8464 }, { "epoch": 11.08, "learning_rate": 7.72e-06, "loss": 0.0036, "step": 8465 }, { "epoch": 11.08, "learning_rate": 7.715e-06, "loss": 0.0029, "step": 8466 }, { "epoch": 11.08, "learning_rate": 7.71e-06, "loss": 0.0034, "step": 8467 }, { "epoch": 11.08, "learning_rate": 7.705e-06, "loss": 0.0028, "step": 8468 }, { "epoch": 11.09, "learning_rate": 7.7e-06, "loss": 0.0026, "step": 8469 }, { "epoch": 11.09, "learning_rate": 7.695e-06, "loss": 0.0019, "step": 8470 }, { "epoch": 11.09, "learning_rate": 7.69e-06, "loss": 0.0028, "step": 8471 }, { "epoch": 11.09, "learning_rate": 7.685e-06, "loss": 0.0024, "step": 8472 }, { "epoch": 11.09, "learning_rate": 7.68e-06, "loss": 0.0024, "step": 8473 }, { "epoch": 11.09, "learning_rate": 7.675e-06, "loss": 0.0031, "step": 8474 }, { "epoch": 11.09, "learning_rate": 7.670000000000001e-06, "loss": 0.0019, "step": 8475 }, { "epoch": 11.09, "learning_rate": 7.665e-06, "loss": 0.0029, "step": 8476 }, { "epoch": 11.1, "learning_rate": 7.660000000000001e-06, "loss": 0.0027, "step": 8477 }, { "epoch": 11.1, "learning_rate": 7.655e-06, "loss": 0.003, "step": 8478 }, { "epoch": 11.1, "learning_rate": 7.65e-06, "loss": 0.0036, "step": 8479 }, { "epoch": 11.1, "learning_rate": 7.645e-06, "loss": 0.0032, "step": 8480 }, { "epoch": 11.1, "learning_rate": 7.64e-06, "loss": 0.0018, "step": 8481 }, { "epoch": 11.1, "learning_rate": 7.635e-06, "loss": 0.0031, "step": 8482 }, { "epoch": 11.1, "learning_rate": 7.630000000000001e-06, "loss": 0.0038, "step": 8483 }, { "epoch": 11.1, "learning_rate": 7.625e-06, "loss": 0.0024, "step": 8484 }, { "epoch": 11.11, "learning_rate": 7.620000000000001e-06, "loss": 0.0023, "step": 8485 }, { "epoch": 11.11, "learning_rate": 7.615e-06, "loss": 0.0025, "step": 8486 }, { "epoch": 11.11, "learning_rate": 7.610000000000001e-06, "loss": 0.0038, "step": 8487 }, { "epoch": 11.11, "learning_rate": 7.605000000000001e-06, "loss": 0.003, "step": 8488 }, { "epoch": 11.11, "learning_rate": 7.6e-06, "loss": 0.0038, "step": 8489 }, { "epoch": 11.11, "learning_rate": 7.595000000000001e-06, "loss": 0.0029, "step": 8490 }, { "epoch": 11.11, "learning_rate": 7.59e-06, "loss": 0.0043, "step": 8491 }, { "epoch": 11.12, "learning_rate": 7.585e-06, "loss": 0.0028, "step": 8492 }, { "epoch": 11.12, "learning_rate": 7.580000000000001e-06, "loss": 0.0034, "step": 8493 }, { "epoch": 11.12, "learning_rate": 7.575e-06, "loss": 0.0036, "step": 8494 }, { "epoch": 11.12, "learning_rate": 7.57e-06, "loss": 0.0024, "step": 8495 }, { "epoch": 11.12, "learning_rate": 7.5649999999999996e-06, "loss": 0.0031, "step": 8496 }, { "epoch": 11.12, "learning_rate": 7.5600000000000005e-06, "loss": 0.0037, "step": 8497 }, { "epoch": 11.12, "learning_rate": 7.555000000000001e-06, "loss": 0.0033, "step": 8498 }, { "epoch": 11.12, "learning_rate": 7.55e-06, "loss": 0.0023, "step": 8499 }, { "epoch": 11.13, "learning_rate": 7.545000000000001e-06, "loss": 0.0017, "step": 8500 }, { "epoch": 11.13, "learning_rate": 7.54e-06, "loss": 0.004, "step": 8501 }, { "epoch": 11.13, "learning_rate": 7.535000000000001e-06, "loss": 0.0036, "step": 8502 }, { "epoch": 11.13, "learning_rate": 7.530000000000001e-06, "loss": 0.0022, "step": 8503 }, { "epoch": 11.13, "learning_rate": 7.525e-06, "loss": 0.0033, "step": 8504 }, { "epoch": 11.13, "learning_rate": 7.520000000000001e-06, "loss": 0.0028, "step": 8505 }, { "epoch": 11.13, "learning_rate": 7.515e-06, "loss": 0.002, "step": 8506 }, { "epoch": 11.13, "learning_rate": 7.51e-06, "loss": 0.0021, "step": 8507 }, { "epoch": 11.14, "learning_rate": 7.505000000000001e-06, "loss": 0.0029, "step": 8508 }, { "epoch": 11.14, "learning_rate": 7.5e-06, "loss": 0.0026, "step": 8509 }, { "epoch": 11.14, "learning_rate": 7.495e-06, "loss": 0.0024, "step": 8510 }, { "epoch": 11.14, "learning_rate": 7.4899999999999994e-06, "loss": 0.0043, "step": 8511 }, { "epoch": 11.14, "learning_rate": 7.485e-06, "loss": 0.0021, "step": 8512 }, { "epoch": 11.14, "learning_rate": 7.480000000000001e-06, "loss": 0.0029, "step": 8513 }, { "epoch": 11.14, "learning_rate": 7.4750000000000004e-06, "loss": 0.0035, "step": 8514 }, { "epoch": 11.15, "learning_rate": 7.4700000000000005e-06, "loss": 0.0035, "step": 8515 }, { "epoch": 11.15, "learning_rate": 7.465e-06, "loss": 0.0018, "step": 8516 }, { "epoch": 11.15, "learning_rate": 7.4600000000000006e-06, "loss": 0.0037, "step": 8517 }, { "epoch": 11.15, "learning_rate": 7.455000000000001e-06, "loss": 0.0041, "step": 8518 }, { "epoch": 11.15, "learning_rate": 7.45e-06, "loss": 0.0031, "step": 8519 }, { "epoch": 11.15, "learning_rate": 7.445000000000001e-06, "loss": 0.0018, "step": 8520 }, { "epoch": 11.15, "learning_rate": 7.44e-06, "loss": 0.002, "step": 8521 }, { "epoch": 11.15, "learning_rate": 7.435e-06, "loss": 0.0025, "step": 8522 }, { "epoch": 11.16, "learning_rate": 7.430000000000001e-06, "loss": 0.0027, "step": 8523 }, { "epoch": 11.16, "learning_rate": 7.425e-06, "loss": 0.0041, "step": 8524 }, { "epoch": 11.16, "learning_rate": 7.420000000000001e-06, "loss": 0.0024, "step": 8525 }, { "epoch": 11.16, "learning_rate": 7.414999999999999e-06, "loss": 0.0035, "step": 8526 }, { "epoch": 11.16, "learning_rate": 7.41e-06, "loss": 0.0035, "step": 8527 }, { "epoch": 11.16, "learning_rate": 7.405000000000001e-06, "loss": 0.0024, "step": 8528 }, { "epoch": 11.16, "learning_rate": 7.4e-06, "loss": 0.0038, "step": 8529 }, { "epoch": 11.16, "learning_rate": 7.395e-06, "loss": 0.003, "step": 8530 }, { "epoch": 11.17, "learning_rate": 7.3899999999999995e-06, "loss": 0.0029, "step": 8531 }, { "epoch": 11.17, "learning_rate": 7.3850000000000004e-06, "loss": 0.0022, "step": 8532 }, { "epoch": 11.17, "learning_rate": 7.3800000000000005e-06, "loss": 0.0022, "step": 8533 }, { "epoch": 11.17, "learning_rate": 7.375e-06, "loss": 0.0036, "step": 8534 }, { "epoch": 11.17, "learning_rate": 7.370000000000001e-06, "loss": 0.002, "step": 8535 }, { "epoch": 11.17, "learning_rate": 7.365e-06, "loss": 0.0029, "step": 8536 }, { "epoch": 11.17, "learning_rate": 7.36e-06, "loss": 0.0018, "step": 8537 }, { "epoch": 11.18, "learning_rate": 7.355000000000001e-06, "loss": 0.0027, "step": 8538 }, { "epoch": 11.18, "learning_rate": 7.35e-06, "loss": 0.0028, "step": 8539 }, { "epoch": 11.18, "learning_rate": 7.345000000000001e-06, "loss": 0.0032, "step": 8540 }, { "epoch": 11.18, "learning_rate": 7.340000000000001e-06, "loss": 0.0019, "step": 8541 }, { "epoch": 11.18, "learning_rate": 7.335e-06, "loss": 0.0027, "step": 8542 }, { "epoch": 11.18, "learning_rate": 7.330000000000001e-06, "loss": 0.002, "step": 8543 }, { "epoch": 11.18, "learning_rate": 7.325e-06, "loss": 0.0021, "step": 8544 }, { "epoch": 11.18, "learning_rate": 7.32e-06, "loss": 0.0026, "step": 8545 }, { "epoch": 11.19, "learning_rate": 7.315000000000001e-06, "loss": 0.0021, "step": 8546 }, { "epoch": 11.19, "learning_rate": 7.31e-06, "loss": 0.0018, "step": 8547 }, { "epoch": 11.19, "learning_rate": 7.305e-06, "loss": 0.003, "step": 8548 }, { "epoch": 11.19, "learning_rate": 7.2999999999999996e-06, "loss": 0.0018, "step": 8549 }, { "epoch": 11.19, "learning_rate": 7.2950000000000005e-06, "loss": 0.0029, "step": 8550 }, { "epoch": 11.19, "learning_rate": 7.290000000000001e-06, "loss": 0.0031, "step": 8551 }, { "epoch": 11.19, "learning_rate": 7.2850000000000006e-06, "loss": 0.0022, "step": 8552 }, { "epoch": 11.2, "learning_rate": 7.280000000000001e-06, "loss": 0.0025, "step": 8553 }, { "epoch": 11.2, "learning_rate": 7.275e-06, "loss": 0.0024, "step": 8554 }, { "epoch": 11.2, "learning_rate": 7.270000000000001e-06, "loss": 0.0017, "step": 8555 }, { "epoch": 11.2, "learning_rate": 7.265000000000001e-06, "loss": 0.0027, "step": 8556 }, { "epoch": 11.2, "learning_rate": 7.26e-06, "loss": 0.0026, "step": 8557 }, { "epoch": 11.2, "learning_rate": 7.255000000000001e-06, "loss": 0.0027, "step": 8558 }, { "epoch": 11.2, "learning_rate": 7.25e-06, "loss": 0.0031, "step": 8559 }, { "epoch": 11.2, "learning_rate": 7.245e-06, "loss": 0.0025, "step": 8560 }, { "epoch": 11.21, "learning_rate": 7.240000000000001e-06, "loss": 0.0022, "step": 8561 }, { "epoch": 11.21, "learning_rate": 7.235e-06, "loss": 0.0022, "step": 8562 }, { "epoch": 11.21, "learning_rate": 7.230000000000001e-06, "loss": 0.0022, "step": 8563 }, { "epoch": 11.21, "learning_rate": 7.2249999999999994e-06, "loss": 0.0027, "step": 8564 }, { "epoch": 11.21, "learning_rate": 7.22e-06, "loss": 0.0018, "step": 8565 }, { "epoch": 11.21, "learning_rate": 7.215000000000001e-06, "loss": 0.0026, "step": 8566 }, { "epoch": 11.21, "learning_rate": 7.2100000000000004e-06, "loss": 0.0033, "step": 8567 }, { "epoch": 11.21, "learning_rate": 7.2050000000000005e-06, "loss": 0.0039, "step": 8568 }, { "epoch": 11.22, "learning_rate": 7.2e-06, "loss": 0.0028, "step": 8569 }, { "epoch": 11.22, "learning_rate": 7.1950000000000006e-06, "loss": 0.0034, "step": 8570 }, { "epoch": 11.22, "learning_rate": 7.190000000000001e-06, "loss": 0.0015, "step": 8571 }, { "epoch": 11.22, "learning_rate": 7.185e-06, "loss": 0.0018, "step": 8572 }, { "epoch": 11.22, "learning_rate": 7.180000000000001e-06, "loss": 0.0038, "step": 8573 }, { "epoch": 11.22, "learning_rate": 7.175e-06, "loss": 0.0027, "step": 8574 }, { "epoch": 11.22, "learning_rate": 7.17e-06, "loss": 0.0017, "step": 8575 }, { "epoch": 11.23, "learning_rate": 7.165000000000001e-06, "loss": 0.005, "step": 8576 }, { "epoch": 11.23, "learning_rate": 7.16e-06, "loss": 0.0026, "step": 8577 }, { "epoch": 11.23, "learning_rate": 7.155000000000001e-06, "loss": 0.0039, "step": 8578 }, { "epoch": 11.23, "learning_rate": 7.15e-06, "loss": 0.0032, "step": 8579 }, { "epoch": 11.23, "learning_rate": 7.145e-06, "loss": 0.0024, "step": 8580 }, { "epoch": 11.23, "learning_rate": 7.140000000000001e-06, "loss": 0.0041, "step": 8581 }, { "epoch": 11.23, "learning_rate": 7.135e-06, "loss": 0.0032, "step": 8582 }, { "epoch": 11.23, "learning_rate": 7.13e-06, "loss": 0.0015, "step": 8583 }, { "epoch": 11.24, "learning_rate": 7.1249999999999995e-06, "loss": 0.003, "step": 8584 }, { "epoch": 11.24, "learning_rate": 7.1200000000000004e-06, "loss": 0.0046, "step": 8585 }, { "epoch": 11.24, "learning_rate": 7.1150000000000005e-06, "loss": 0.0034, "step": 8586 }, { "epoch": 11.24, "learning_rate": 7.11e-06, "loss": 0.0023, "step": 8587 }, { "epoch": 11.24, "learning_rate": 7.105000000000001e-06, "loss": 0.0022, "step": 8588 }, { "epoch": 11.24, "learning_rate": 7.1e-06, "loss": 0.003, "step": 8589 }, { "epoch": 11.24, "learning_rate": 7.095000000000001e-06, "loss": 0.0021, "step": 8590 }, { "epoch": 11.24, "learning_rate": 7.090000000000001e-06, "loss": 0.0025, "step": 8591 }, { "epoch": 11.25, "learning_rate": 7.085e-06, "loss": 0.0033, "step": 8592 }, { "epoch": 11.25, "learning_rate": 7.080000000000001e-06, "loss": 0.0026, "step": 8593 }, { "epoch": 11.25, "learning_rate": 7.075e-06, "loss": 0.0034, "step": 8594 }, { "epoch": 11.25, "learning_rate": 7.07e-06, "loss": 0.0029, "step": 8595 }, { "epoch": 11.25, "learning_rate": 7.065000000000001e-06, "loss": 0.0025, "step": 8596 }, { "epoch": 11.25, "learning_rate": 7.06e-06, "loss": 0.0034, "step": 8597 }, { "epoch": 11.25, "learning_rate": 7.055e-06, "loss": 0.0032, "step": 8598 }, { "epoch": 11.26, "learning_rate": 7.049999999999999e-06, "loss": 0.002, "step": 8599 }, { "epoch": 11.26, "learning_rate": 7.045e-06, "loss": 0.0034, "step": 8600 }, { "epoch": 11.26, "learning_rate": 7.04e-06, "loss": 0.0026, "step": 8601 }, { "epoch": 11.26, "learning_rate": 7.0349999999999996e-06, "loss": 0.0039, "step": 8602 }, { "epoch": 11.26, "learning_rate": 7.0300000000000005e-06, "loss": 0.0032, "step": 8603 }, { "epoch": 11.26, "learning_rate": 7.025000000000001e-06, "loss": 0.0037, "step": 8604 }, { "epoch": 11.26, "learning_rate": 7.0200000000000006e-06, "loss": 0.0021, "step": 8605 }, { "epoch": 11.26, "learning_rate": 7.015000000000001e-06, "loss": 0.0022, "step": 8606 }, { "epoch": 11.27, "learning_rate": 7.01e-06, "loss": 0.0018, "step": 8607 }, { "epoch": 11.27, "learning_rate": 7.005000000000001e-06, "loss": 0.0024, "step": 8608 }, { "epoch": 11.27, "learning_rate": 7.000000000000001e-06, "loss": 0.0027, "step": 8609 }, { "epoch": 11.27, "learning_rate": 6.995e-06, "loss": 0.0032, "step": 8610 }, { "epoch": 11.27, "learning_rate": 6.990000000000001e-06, "loss": 0.0033, "step": 8611 }, { "epoch": 11.27, "learning_rate": 6.985e-06, "loss": 0.0043, "step": 8612 }, { "epoch": 11.27, "learning_rate": 6.98e-06, "loss": 0.0043, "step": 8613 }, { "epoch": 11.27, "learning_rate": 6.975000000000001e-06, "loss": 0.0019, "step": 8614 }, { "epoch": 11.28, "learning_rate": 6.97e-06, "loss": 0.0016, "step": 8615 }, { "epoch": 11.28, "learning_rate": 6.965000000000001e-06, "loss": 0.0036, "step": 8616 }, { "epoch": 11.28, "learning_rate": 6.9599999999999994e-06, "loss": 0.0034, "step": 8617 }, { "epoch": 11.28, "learning_rate": 6.955e-06, "loss": 0.002, "step": 8618 }, { "epoch": 11.28, "learning_rate": 6.950000000000001e-06, "loss": 0.0022, "step": 8619 }, { "epoch": 11.28, "learning_rate": 6.945e-06, "loss": 0.003, "step": 8620 }, { "epoch": 11.28, "learning_rate": 6.9400000000000005e-06, "loss": 0.0018, "step": 8621 }, { "epoch": 11.29, "learning_rate": 6.935e-06, "loss": 0.0024, "step": 8622 }, { "epoch": 11.29, "learning_rate": 6.9300000000000006e-06, "loss": 0.0045, "step": 8623 }, { "epoch": 11.29, "learning_rate": 6.925000000000001e-06, "loss": 0.003, "step": 8624 }, { "epoch": 11.29, "learning_rate": 6.92e-06, "loss": 0.0048, "step": 8625 }, { "epoch": 11.29, "learning_rate": 6.915000000000001e-06, "loss": 0.002, "step": 8626 }, { "epoch": 11.29, "learning_rate": 6.91e-06, "loss": 0.0019, "step": 8627 }, { "epoch": 11.29, "learning_rate": 6.905e-06, "loss": 0.0027, "step": 8628 }, { "epoch": 11.29, "learning_rate": 6.900000000000001e-06, "loss": 0.0024, "step": 8629 }, { "epoch": 11.3, "learning_rate": 6.895e-06, "loss": 0.0028, "step": 8630 }, { "epoch": 11.3, "learning_rate": 6.890000000000001e-06, "loss": 0.0024, "step": 8631 }, { "epoch": 11.3, "learning_rate": 6.885e-06, "loss": 0.0032, "step": 8632 }, { "epoch": 11.3, "learning_rate": 6.88e-06, "loss": 0.0025, "step": 8633 }, { "epoch": 11.3, "learning_rate": 6.875000000000001e-06, "loss": 0.0037, "step": 8634 }, { "epoch": 11.3, "learning_rate": 6.87e-06, "loss": 0.0039, "step": 8635 }, { "epoch": 11.3, "learning_rate": 6.865e-06, "loss": 0.0023, "step": 8636 }, { "epoch": 11.3, "learning_rate": 6.8599999999999995e-06, "loss": 0.0039, "step": 8637 }, { "epoch": 11.31, "learning_rate": 6.8550000000000004e-06, "loss": 0.0026, "step": 8638 }, { "epoch": 11.31, "learning_rate": 6.8500000000000005e-06, "loss": 0.002, "step": 8639 }, { "epoch": 11.31, "learning_rate": 6.845e-06, "loss": 0.0024, "step": 8640 }, { "epoch": 11.31, "learning_rate": 6.840000000000001e-06, "loss": 0.0027, "step": 8641 }, { "epoch": 11.31, "learning_rate": 6.835e-06, "loss": 0.0038, "step": 8642 }, { "epoch": 11.31, "learning_rate": 6.830000000000001e-06, "loss": 0.0043, "step": 8643 }, { "epoch": 11.31, "learning_rate": 6.825000000000001e-06, "loss": 0.0027, "step": 8644 }, { "epoch": 11.32, "learning_rate": 6.82e-06, "loss": 0.0014, "step": 8645 }, { "epoch": 11.32, "learning_rate": 6.815000000000001e-06, "loss": 0.0026, "step": 8646 }, { "epoch": 11.32, "learning_rate": 6.81e-06, "loss": 0.0026, "step": 8647 }, { "epoch": 11.32, "learning_rate": 6.805e-06, "loss": 0.0012, "step": 8648 }, { "epoch": 11.32, "learning_rate": 6.800000000000001e-06, "loss": 0.0016, "step": 8649 }, { "epoch": 11.32, "learning_rate": 6.795e-06, "loss": 0.0016, "step": 8650 }, { "epoch": 11.32, "learning_rate": 6.79e-06, "loss": 0.0034, "step": 8651 }, { "epoch": 11.32, "learning_rate": 6.784999999999999e-06, "loss": 0.0015, "step": 8652 }, { "epoch": 11.33, "learning_rate": 6.78e-06, "loss": 0.0023, "step": 8653 }, { "epoch": 11.33, "learning_rate": 6.775000000000001e-06, "loss": 0.0025, "step": 8654 }, { "epoch": 11.33, "learning_rate": 6.7699999999999996e-06, "loss": 0.003, "step": 8655 }, { "epoch": 11.33, "learning_rate": 6.7650000000000005e-06, "loss": 0.0028, "step": 8656 }, { "epoch": 11.33, "learning_rate": 6.76e-06, "loss": 0.0029, "step": 8657 }, { "epoch": 11.33, "learning_rate": 6.7550000000000005e-06, "loss": 0.0018, "step": 8658 }, { "epoch": 11.33, "learning_rate": 6.750000000000001e-06, "loss": 0.0026, "step": 8659 }, { "epoch": 11.34, "learning_rate": 6.745e-06, "loss": 0.0024, "step": 8660 }, { "epoch": 11.34, "learning_rate": 6.740000000000001e-06, "loss": 0.0032, "step": 8661 }, { "epoch": 11.34, "learning_rate": 6.735e-06, "loss": 0.0054, "step": 8662 }, { "epoch": 11.34, "learning_rate": 6.73e-06, "loss": 0.003, "step": 8663 }, { "epoch": 11.34, "learning_rate": 6.725000000000001e-06, "loss": 0.0022, "step": 8664 }, { "epoch": 11.34, "learning_rate": 6.72e-06, "loss": 0.0027, "step": 8665 }, { "epoch": 11.34, "learning_rate": 6.715e-06, "loss": 0.0032, "step": 8666 }, { "epoch": 11.34, "learning_rate": 6.710000000000001e-06, "loss": 0.0027, "step": 8667 }, { "epoch": 11.35, "learning_rate": 6.705e-06, "loss": 0.0025, "step": 8668 }, { "epoch": 11.35, "learning_rate": 6.700000000000001e-06, "loss": 0.0021, "step": 8669 }, { "epoch": 11.35, "learning_rate": 6.695e-06, "loss": 0.0025, "step": 8670 }, { "epoch": 11.35, "learning_rate": 6.69e-06, "loss": 0.0089, "step": 8671 }, { "epoch": 11.35, "learning_rate": 6.685000000000001e-06, "loss": 0.0031, "step": 8672 }, { "epoch": 11.35, "learning_rate": 6.68e-06, "loss": 0.0027, "step": 8673 }, { "epoch": 11.35, "learning_rate": 6.6750000000000005e-06, "loss": 0.0022, "step": 8674 }, { "epoch": 11.35, "learning_rate": 6.67e-06, "loss": 0.0018, "step": 8675 }, { "epoch": 11.36, "learning_rate": 6.6650000000000006e-06, "loss": 0.0024, "step": 8676 }, { "epoch": 11.36, "learning_rate": 6.660000000000001e-06, "loss": 0.0018, "step": 8677 }, { "epoch": 11.36, "learning_rate": 6.660000000000001e-06, "loss": 0.0029, "step": 8678 }, { "epoch": 11.36, "learning_rate": 6.660000000000001e-06, "loss": 0.0027, "step": 8679 }, { "epoch": 11.36, "learning_rate": 6.655e-06, "loss": 0.0033, "step": 8680 }, { "epoch": 11.36, "learning_rate": 6.650000000000001e-06, "loss": 0.0031, "step": 8681 }, { "epoch": 11.36, "learning_rate": 6.645e-06, "loss": 0.0023, "step": 8682 }, { "epoch": 11.37, "learning_rate": 6.640000000000001e-06, "loss": 0.0025, "step": 8683 }, { "epoch": 11.37, "learning_rate": 6.635000000000001e-06, "loss": 0.0023, "step": 8684 }, { "epoch": 11.37, "learning_rate": 6.63e-06, "loss": 0.0014, "step": 8685 }, { "epoch": 11.37, "learning_rate": 6.625000000000001e-06, "loss": 0.0024, "step": 8686 }, { "epoch": 11.37, "learning_rate": 6.62e-06, "loss": 0.0028, "step": 8687 }, { "epoch": 11.37, "learning_rate": 6.615e-06, "loss": 0.0027, "step": 8688 }, { "epoch": 11.37, "learning_rate": 6.610000000000001e-06, "loss": 0.0017, "step": 8689 }, { "epoch": 11.37, "learning_rate": 6.605e-06, "loss": 0.0031, "step": 8690 }, { "epoch": 11.38, "learning_rate": 6.6e-06, "loss": 0.0068, "step": 8691 }, { "epoch": 11.38, "learning_rate": 6.5949999999999995e-06, "loss": 0.003, "step": 8692 }, { "epoch": 11.38, "learning_rate": 6.5900000000000004e-06, "loss": 0.003, "step": 8693 }, { "epoch": 11.38, "learning_rate": 6.5850000000000005e-06, "loss": 0.0026, "step": 8694 }, { "epoch": 11.38, "learning_rate": 6.58e-06, "loss": 0.0025, "step": 8695 }, { "epoch": 11.38, "learning_rate": 6.5750000000000006e-06, "loss": 0.0031, "step": 8696 }, { "epoch": 11.38, "learning_rate": 6.57e-06, "loss": 0.0026, "step": 8697 }, { "epoch": 11.38, "learning_rate": 6.565000000000001e-06, "loss": 0.0045, "step": 8698 }, { "epoch": 11.39, "learning_rate": 6.560000000000001e-06, "loss": 0.0024, "step": 8699 }, { "epoch": 11.39, "learning_rate": 6.555e-06, "loss": 0.0034, "step": 8700 }, { "epoch": 11.39, "learning_rate": 6.550000000000001e-06, "loss": 0.009, "step": 8701 }, { "epoch": 11.39, "learning_rate": 6.545e-06, "loss": 0.0021, "step": 8702 }, { "epoch": 11.39, "learning_rate": 6.54e-06, "loss": 0.0055, "step": 8703 }, { "epoch": 11.39, "learning_rate": 6.535000000000001e-06, "loss": 0.0034, "step": 8704 }, { "epoch": 11.39, "learning_rate": 6.53e-06, "loss": 0.0036, "step": 8705 }, { "epoch": 11.4, "learning_rate": 6.525e-06, "loss": 0.0036, "step": 8706 }, { "epoch": 11.4, "learning_rate": 6.519999999999999e-06, "loss": 0.0023, "step": 8707 }, { "epoch": 11.4, "learning_rate": 6.515e-06, "loss": 0.0017, "step": 8708 }, { "epoch": 11.4, "learning_rate": 6.510000000000001e-06, "loss": 0.0032, "step": 8709 }, { "epoch": 11.4, "learning_rate": 6.505e-06, "loss": 0.0031, "step": 8710 }, { "epoch": 11.4, "learning_rate": 6.5000000000000004e-06, "loss": 0.0041, "step": 8711 }, { "epoch": 11.4, "learning_rate": 6.495e-06, "loss": 0.0029, "step": 8712 }, { "epoch": 11.4, "learning_rate": 6.4900000000000005e-06, "loss": 0.0024, "step": 8713 }, { "epoch": 11.41, "learning_rate": 6.485000000000001e-06, "loss": 0.0029, "step": 8714 }, { "epoch": 11.41, "learning_rate": 6.48e-06, "loss": 0.0026, "step": 8715 }, { "epoch": 11.41, "learning_rate": 6.475000000000001e-06, "loss": 0.0024, "step": 8716 }, { "epoch": 11.41, "learning_rate": 6.47e-06, "loss": 0.0024, "step": 8717 }, { "epoch": 11.41, "learning_rate": 6.465e-06, "loss": 0.0016, "step": 8718 }, { "epoch": 11.41, "learning_rate": 6.460000000000001e-06, "loss": 0.0033, "step": 8719 }, { "epoch": 11.41, "learning_rate": 6.455e-06, "loss": 0.0024, "step": 8720 }, { "epoch": 11.41, "learning_rate": 6.45e-06, "loss": 0.0031, "step": 8721 }, { "epoch": 11.42, "learning_rate": 6.444999999999999e-06, "loss": 0.0019, "step": 8722 }, { "epoch": 11.42, "learning_rate": 6.44e-06, "loss": 0.0044, "step": 8723 }, { "epoch": 11.42, "learning_rate": 6.435000000000001e-06, "loss": 0.0028, "step": 8724 }, { "epoch": 11.42, "learning_rate": 6.43e-06, "loss": 0.0024, "step": 8725 }, { "epoch": 11.42, "learning_rate": 6.425e-06, "loss": 0.0023, "step": 8726 }, { "epoch": 11.42, "learning_rate": 6.4199999999999995e-06, "loss": 0.0057, "step": 8727 }, { "epoch": 11.42, "learning_rate": 6.415e-06, "loss": 0.0014, "step": 8728 }, { "epoch": 11.43, "learning_rate": 6.4100000000000005e-06, "loss": 0.0045, "step": 8729 }, { "epoch": 11.43, "learning_rate": 6.405e-06, "loss": 0.0022, "step": 8730 }, { "epoch": 11.43, "learning_rate": 6.4000000000000006e-06, "loss": 0.0029, "step": 8731 }, { "epoch": 11.43, "learning_rate": 6.395000000000001e-06, "loss": 0.003, "step": 8732 }, { "epoch": 11.43, "learning_rate": 6.39e-06, "loss": 0.0017, "step": 8733 }, { "epoch": 11.43, "learning_rate": 6.385000000000001e-06, "loss": 0.0033, "step": 8734 }, { "epoch": 11.43, "learning_rate": 6.38e-06, "loss": 0.0017, "step": 8735 }, { "epoch": 11.43, "learning_rate": 6.375000000000001e-06, "loss": 0.0031, "step": 8736 }, { "epoch": 11.44, "learning_rate": 6.370000000000001e-06, "loss": 0.0061, "step": 8737 }, { "epoch": 11.44, "learning_rate": 6.365e-06, "loss": 0.004, "step": 8738 }, { "epoch": 11.44, "learning_rate": 6.360000000000001e-06, "loss": 0.003, "step": 8739 }, { "epoch": 11.44, "learning_rate": 6.355e-06, "loss": 0.0018, "step": 8740 }, { "epoch": 11.44, "learning_rate": 6.35e-06, "loss": 0.0023, "step": 8741 }, { "epoch": 11.44, "learning_rate": 6.345000000000001e-06, "loss": 0.0021, "step": 8742 }, { "epoch": 11.44, "learning_rate": 6.34e-06, "loss": 0.0037, "step": 8743 }, { "epoch": 11.45, "learning_rate": 6.335e-06, "loss": 0.0024, "step": 8744 }, { "epoch": 11.45, "learning_rate": 6.3299999999999995e-06, "loss": 0.0023, "step": 8745 }, { "epoch": 11.45, "learning_rate": 6.3250000000000004e-06, "loss": 0.0051, "step": 8746 }, { "epoch": 11.45, "learning_rate": 6.320000000000001e-06, "loss": 0.0034, "step": 8747 }, { "epoch": 11.45, "learning_rate": 6.315e-06, "loss": 0.0026, "step": 8748 }, { "epoch": 11.45, "learning_rate": 6.3100000000000006e-06, "loss": 0.002, "step": 8749 }, { "epoch": 11.45, "learning_rate": 6.305e-06, "loss": 0.0035, "step": 8750 }, { "epoch": 11.45, "learning_rate": 6.300000000000001e-06, "loss": 0.0032, "step": 8751 }, { "epoch": 11.46, "learning_rate": 6.295000000000001e-06, "loss": 0.0029, "step": 8752 }, { "epoch": 11.46, "learning_rate": 6.29e-06, "loss": 0.0024, "step": 8753 }, { "epoch": 11.46, "learning_rate": 6.285000000000001e-06, "loss": 0.0026, "step": 8754 }, { "epoch": 11.46, "learning_rate": 6.28e-06, "loss": 0.0023, "step": 8755 }, { "epoch": 11.46, "learning_rate": 6.275e-06, "loss": 0.0046, "step": 8756 }, { "epoch": 11.46, "learning_rate": 6.270000000000001e-06, "loss": 0.002, "step": 8757 }, { "epoch": 11.46, "learning_rate": 6.265e-06, "loss": 0.0025, "step": 8758 }, { "epoch": 11.46, "learning_rate": 6.26e-06, "loss": 0.0022, "step": 8759 }, { "epoch": 11.47, "learning_rate": 6.254999999999999e-06, "loss": 0.0022, "step": 8760 }, { "epoch": 11.47, "learning_rate": 6.25e-06, "loss": 0.0029, "step": 8761 }, { "epoch": 11.47, "learning_rate": 6.245e-06, "loss": 0.0025, "step": 8762 }, { "epoch": 11.47, "learning_rate": 6.24e-06, "loss": 0.0013, "step": 8763 }, { "epoch": 11.47, "learning_rate": 6.2350000000000004e-06, "loss": 0.0031, "step": 8764 }, { "epoch": 11.47, "learning_rate": 6.2300000000000005e-06, "loss": 0.0025, "step": 8765 }, { "epoch": 11.47, "learning_rate": 6.2250000000000005e-06, "loss": 0.002, "step": 8766 }, { "epoch": 11.48, "learning_rate": 6.22e-06, "loss": 0.0028, "step": 8767 }, { "epoch": 11.48, "learning_rate": 6.215e-06, "loss": 0.0039, "step": 8768 }, { "epoch": 11.48, "learning_rate": 6.210000000000001e-06, "loss": 0.0024, "step": 8769 }, { "epoch": 11.48, "learning_rate": 6.205000000000001e-06, "loss": 0.0042, "step": 8770 }, { "epoch": 11.48, "learning_rate": 6.2e-06, "loss": 0.0024, "step": 8771 }, { "epoch": 11.48, "learning_rate": 6.195e-06, "loss": 0.0012, "step": 8772 }, { "epoch": 11.48, "learning_rate": 6.19e-06, "loss": 0.0019, "step": 8773 }, { "epoch": 11.48, "learning_rate": 6.185000000000001e-06, "loss": 0.0031, "step": 8774 }, { "epoch": 11.49, "learning_rate": 6.18e-06, "loss": 0.0028, "step": 8775 }, { "epoch": 11.49, "learning_rate": 6.175e-06, "loss": 0.0036, "step": 8776 }, { "epoch": 11.49, "learning_rate": 6.17e-06, "loss": 0.0027, "step": 8777 }, { "epoch": 11.49, "learning_rate": 6.165e-06, "loss": 0.0028, "step": 8778 }, { "epoch": 11.49, "learning_rate": 6.16e-06, "loss": 0.0031, "step": 8779 }, { "epoch": 11.49, "learning_rate": 6.155e-06, "loss": 0.0025, "step": 8780 }, { "epoch": 11.49, "learning_rate": 6.15e-06, "loss": 0.0024, "step": 8781 }, { "epoch": 11.49, "learning_rate": 6.1450000000000005e-06, "loss": 0.0028, "step": 8782 }, { "epoch": 11.5, "learning_rate": 6.1400000000000005e-06, "loss": 0.0028, "step": 8783 }, { "epoch": 11.5, "learning_rate": 6.1350000000000006e-06, "loss": 0.0024, "step": 8784 }, { "epoch": 11.5, "learning_rate": 6.130000000000001e-06, "loss": 0.0026, "step": 8785 }, { "epoch": 11.5, "learning_rate": 6.125e-06, "loss": 0.0034, "step": 8786 }, { "epoch": 11.5, "learning_rate": 6.12e-06, "loss": 0.0028, "step": 8787 }, { "epoch": 11.5, "learning_rate": 6.115000000000001e-06, "loss": 0.0032, "step": 8788 }, { "epoch": 11.5, "learning_rate": 6.110000000000001e-06, "loss": 0.0024, "step": 8789 }, { "epoch": 11.51, "learning_rate": 6.105e-06, "loss": 0.0024, "step": 8790 }, { "epoch": 11.51, "learning_rate": 6.1e-06, "loss": 0.0017, "step": 8791 }, { "epoch": 11.51, "learning_rate": 6.095e-06, "loss": 0.0036, "step": 8792 }, { "epoch": 11.51, "learning_rate": 6.090000000000001e-06, "loss": 0.0034, "step": 8793 }, { "epoch": 11.51, "learning_rate": 6.085e-06, "loss": 0.0028, "step": 8794 }, { "epoch": 11.51, "learning_rate": 6.08e-06, "loss": 0.0026, "step": 8795 }, { "epoch": 11.51, "learning_rate": 6.075e-06, "loss": 0.0035, "step": 8796 }, { "epoch": 11.51, "learning_rate": 6.07e-06, "loss": 0.0038, "step": 8797 }, { "epoch": 11.52, "learning_rate": 6.065e-06, "loss": 0.0022, "step": 8798 }, { "epoch": 11.52, "learning_rate": 6.0600000000000004e-06, "loss": 0.0021, "step": 8799 }, { "epoch": 11.52, "learning_rate": 6.0550000000000005e-06, "loss": 0.0021, "step": 8800 }, { "epoch": 11.52, "learning_rate": 6.0500000000000005e-06, "loss": 0.0033, "step": 8801 }, { "epoch": 11.52, "learning_rate": 6.045e-06, "loss": 0.004, "step": 8802 }, { "epoch": 11.52, "learning_rate": 6.040000000000001e-06, "loss": 0.0023, "step": 8803 }, { "epoch": 11.52, "learning_rate": 6.035000000000001e-06, "loss": 0.002, "step": 8804 }, { "epoch": 11.52, "learning_rate": 6.03e-06, "loss": 0.0048, "step": 8805 }, { "epoch": 11.53, "learning_rate": 6.025e-06, "loss": 0.0024, "step": 8806 }, { "epoch": 11.53, "learning_rate": 6.02e-06, "loss": 0.0016, "step": 8807 }, { "epoch": 11.53, "learning_rate": 6.015000000000001e-06, "loss": 0.0029, "step": 8808 }, { "epoch": 11.53, "learning_rate": 6.01e-06, "loss": 0.0022, "step": 8809 }, { "epoch": 11.53, "learning_rate": 6.005e-06, "loss": 0.0025, "step": 8810 }, { "epoch": 11.53, "learning_rate": 6e-06, "loss": 0.0048, "step": 8811 }, { "epoch": 11.53, "learning_rate": 5.995e-06, "loss": 0.0036, "step": 8812 }, { "epoch": 11.54, "learning_rate": 5.99e-06, "loss": 0.0049, "step": 8813 }, { "epoch": 11.54, "learning_rate": 5.985e-06, "loss": 0.002, "step": 8814 }, { "epoch": 11.54, "learning_rate": 5.98e-06, "loss": 0.0031, "step": 8815 }, { "epoch": 11.54, "learning_rate": 5.975e-06, "loss": 0.0018, "step": 8816 }, { "epoch": 11.54, "learning_rate": 5.9700000000000004e-06, "loss": 0.0043, "step": 8817 }, { "epoch": 11.54, "learning_rate": 5.9650000000000005e-06, "loss": 0.0022, "step": 8818 }, { "epoch": 11.54, "learning_rate": 5.9600000000000005e-06, "loss": 0.0023, "step": 8819 }, { "epoch": 11.54, "learning_rate": 5.955000000000001e-06, "loss": 0.0027, "step": 8820 }, { "epoch": 11.55, "learning_rate": 5.95e-06, "loss": 0.0014, "step": 8821 }, { "epoch": 11.55, "learning_rate": 5.945000000000001e-06, "loss": 0.0028, "step": 8822 }, { "epoch": 11.55, "learning_rate": 5.940000000000001e-06, "loss": 0.0034, "step": 8823 }, { "epoch": 11.55, "learning_rate": 5.935e-06, "loss": 0.0028, "step": 8824 }, { "epoch": 11.55, "learning_rate": 5.93e-06, "loss": 0.0037, "step": 8825 }, { "epoch": 11.55, "learning_rate": 5.925e-06, "loss": 0.0025, "step": 8826 }, { "epoch": 11.55, "learning_rate": 5.920000000000001e-06, "loss": 0.0035, "step": 8827 }, { "epoch": 11.55, "learning_rate": 5.915e-06, "loss": 0.0042, "step": 8828 }, { "epoch": 11.56, "learning_rate": 5.91e-06, "loss": 0.0034, "step": 8829 }, { "epoch": 11.56, "learning_rate": 5.905e-06, "loss": 0.0028, "step": 8830 }, { "epoch": 11.56, "learning_rate": 5.9e-06, "loss": 0.0026, "step": 8831 }, { "epoch": 11.56, "learning_rate": 5.895e-06, "loss": 0.0017, "step": 8832 }, { "epoch": 11.56, "learning_rate": 5.89e-06, "loss": 0.002, "step": 8833 }, { "epoch": 11.56, "learning_rate": 5.885e-06, "loss": 0.0041, "step": 8834 }, { "epoch": 11.56, "learning_rate": 5.8800000000000005e-06, "loss": 0.0025, "step": 8835 }, { "epoch": 11.57, "learning_rate": 5.875e-06, "loss": 0.0019, "step": 8836 }, { "epoch": 11.57, "learning_rate": 5.8700000000000005e-06, "loss": 0.0026, "step": 8837 }, { "epoch": 11.57, "learning_rate": 5.865000000000001e-06, "loss": 0.0025, "step": 8838 }, { "epoch": 11.57, "learning_rate": 5.86e-06, "loss": 0.0034, "step": 8839 }, { "epoch": 11.57, "learning_rate": 5.855e-06, "loss": 0.0021, "step": 8840 }, { "epoch": 11.57, "learning_rate": 5.850000000000001e-06, "loss": 0.0029, "step": 8841 }, { "epoch": 11.57, "learning_rate": 5.845000000000001e-06, "loss": 0.002, "step": 8842 }, { "epoch": 11.57, "learning_rate": 5.84e-06, "loss": 0.0018, "step": 8843 }, { "epoch": 11.58, "learning_rate": 5.835e-06, "loss": 0.0014, "step": 8844 }, { "epoch": 11.58, "learning_rate": 5.83e-06, "loss": 0.0018, "step": 8845 }, { "epoch": 11.58, "learning_rate": 5.825000000000001e-06, "loss": 0.0025, "step": 8846 }, { "epoch": 11.58, "learning_rate": 5.82e-06, "loss": 0.003, "step": 8847 }, { "epoch": 11.58, "learning_rate": 5.815e-06, "loss": 0.0018, "step": 8848 }, { "epoch": 11.58, "learning_rate": 5.81e-06, "loss": 0.0029, "step": 8849 }, { "epoch": 11.58, "learning_rate": 5.805e-06, "loss": 0.0032, "step": 8850 }, { "epoch": 11.59, "learning_rate": 5.8e-06, "loss": 0.0035, "step": 8851 }, { "epoch": 11.59, "learning_rate": 5.795e-06, "loss": 0.0019, "step": 8852 }, { "epoch": 11.59, "learning_rate": 5.7900000000000005e-06, "loss": 0.0047, "step": 8853 }, { "epoch": 11.59, "learning_rate": 5.7850000000000005e-06, "loss": 0.0032, "step": 8854 }, { "epoch": 11.59, "learning_rate": 5.78e-06, "loss": 0.0014, "step": 8855 }, { "epoch": 11.59, "learning_rate": 5.775000000000001e-06, "loss": 0.0036, "step": 8856 }, { "epoch": 11.59, "learning_rate": 5.770000000000001e-06, "loss": 0.0029, "step": 8857 }, { "epoch": 11.59, "learning_rate": 5.765e-06, "loss": 0.0023, "step": 8858 }, { "epoch": 11.6, "learning_rate": 5.76e-06, "loss": 0.0028, "step": 8859 }, { "epoch": 11.6, "learning_rate": 5.755e-06, "loss": 0.0043, "step": 8860 }, { "epoch": 11.6, "learning_rate": 5.750000000000001e-06, "loss": 0.0024, "step": 8861 }, { "epoch": 11.6, "learning_rate": 5.745e-06, "loss": 0.0024, "step": 8862 }, { "epoch": 11.6, "learning_rate": 5.74e-06, "loss": 0.0028, "step": 8863 }, { "epoch": 11.6, "learning_rate": 5.735e-06, "loss": 0.0029, "step": 8864 }, { "epoch": 11.6, "learning_rate": 5.73e-06, "loss": 0.0038, "step": 8865 }, { "epoch": 11.6, "learning_rate": 5.725e-06, "loss": 0.0031, "step": 8866 }, { "epoch": 11.61, "learning_rate": 5.72e-06, "loss": 0.0024, "step": 8867 }, { "epoch": 11.61, "learning_rate": 5.715e-06, "loss": 0.0025, "step": 8868 }, { "epoch": 11.61, "learning_rate": 5.71e-06, "loss": 0.0025, "step": 8869 }, { "epoch": 11.61, "learning_rate": 5.705e-06, "loss": 0.0022, "step": 8870 }, { "epoch": 11.61, "learning_rate": 5.7000000000000005e-06, "loss": 0.0059, "step": 8871 }, { "epoch": 11.61, "learning_rate": 5.6950000000000005e-06, "loss": 0.0018, "step": 8872 }, { "epoch": 11.61, "learning_rate": 5.690000000000001e-06, "loss": 0.0028, "step": 8873 }, { "epoch": 11.62, "learning_rate": 5.685e-06, "loss": 0.0028, "step": 8874 }, { "epoch": 11.62, "learning_rate": 5.680000000000001e-06, "loss": 0.0026, "step": 8875 }, { "epoch": 11.62, "learning_rate": 5.675000000000001e-06, "loss": 0.0024, "step": 8876 }, { "epoch": 11.62, "learning_rate": 5.67e-06, "loss": 0.0026, "step": 8877 }, { "epoch": 11.62, "learning_rate": 5.665e-06, "loss": 0.0022, "step": 8878 }, { "epoch": 11.62, "learning_rate": 5.66e-06, "loss": 0.0028, "step": 8879 }, { "epoch": 11.62, "learning_rate": 5.655000000000001e-06, "loss": 0.0047, "step": 8880 }, { "epoch": 11.62, "learning_rate": 5.65e-06, "loss": 0.0042, "step": 8881 }, { "epoch": 11.63, "learning_rate": 5.645e-06, "loss": 0.0041, "step": 8882 }, { "epoch": 11.63, "learning_rate": 5.64e-06, "loss": 0.0023, "step": 8883 }, { "epoch": 11.63, "learning_rate": 5.635e-06, "loss": 0.0025, "step": 8884 }, { "epoch": 11.63, "learning_rate": 5.63e-06, "loss": 0.0031, "step": 8885 }, { "epoch": 11.63, "learning_rate": 5.625e-06, "loss": 0.003, "step": 8886 }, { "epoch": 11.63, "learning_rate": 5.62e-06, "loss": 0.0024, "step": 8887 }, { "epoch": 11.63, "learning_rate": 5.6150000000000005e-06, "loss": 0.0025, "step": 8888 }, { "epoch": 11.63, "learning_rate": 5.61e-06, "loss": 0.0024, "step": 8889 }, { "epoch": 11.64, "learning_rate": 5.6050000000000005e-06, "loss": 0.0026, "step": 8890 }, { "epoch": 11.64, "learning_rate": 5.600000000000001e-06, "loss": 0.0032, "step": 8891 }, { "epoch": 11.64, "learning_rate": 5.595000000000001e-06, "loss": 0.0027, "step": 8892 }, { "epoch": 11.64, "learning_rate": 5.59e-06, "loss": 0.0019, "step": 8893 }, { "epoch": 11.64, "learning_rate": 5.585e-06, "loss": 0.0023, "step": 8894 }, { "epoch": 11.64, "learning_rate": 5.580000000000001e-06, "loss": 0.0027, "step": 8895 }, { "epoch": 11.64, "learning_rate": 5.575e-06, "loss": 0.0022, "step": 8896 }, { "epoch": 11.65, "learning_rate": 5.57e-06, "loss": 0.002, "step": 8897 }, { "epoch": 11.65, "learning_rate": 5.565e-06, "loss": 0.0051, "step": 8898 }, { "epoch": 11.65, "learning_rate": 5.56e-06, "loss": 0.0036, "step": 8899 }, { "epoch": 11.65, "learning_rate": 5.555e-06, "loss": 0.0031, "step": 8900 }, { "epoch": 11.65, "learning_rate": 5.55e-06, "loss": 0.0023, "step": 8901 }, { "epoch": 11.65, "learning_rate": 5.545e-06, "loss": 0.002, "step": 8902 }, { "epoch": 11.65, "learning_rate": 5.54e-06, "loss": 0.0019, "step": 8903 }, { "epoch": 11.65, "learning_rate": 5.535e-06, "loss": 0.0014, "step": 8904 }, { "epoch": 11.66, "learning_rate": 5.53e-06, "loss": 0.0067, "step": 8905 }, { "epoch": 11.66, "learning_rate": 5.5250000000000005e-06, "loss": 0.0024, "step": 8906 }, { "epoch": 11.66, "learning_rate": 5.5200000000000005e-06, "loss": 0.0025, "step": 8907 }, { "epoch": 11.66, "learning_rate": 5.515e-06, "loss": 0.0023, "step": 8908 }, { "epoch": 11.66, "learning_rate": 5.510000000000001e-06, "loss": 0.0025, "step": 8909 }, { "epoch": 11.66, "learning_rate": 5.505000000000001e-06, "loss": 0.0021, "step": 8910 }, { "epoch": 11.66, "learning_rate": 5.500000000000001e-06, "loss": 0.0032, "step": 8911 }, { "epoch": 11.66, "learning_rate": 5.495e-06, "loss": 0.0027, "step": 8912 }, { "epoch": 11.67, "learning_rate": 5.49e-06, "loss": 0.0031, "step": 8913 }, { "epoch": 11.67, "learning_rate": 5.485000000000001e-06, "loss": 0.0019, "step": 8914 }, { "epoch": 11.67, "learning_rate": 5.48e-06, "loss": 0.0013, "step": 8915 }, { "epoch": 11.67, "learning_rate": 5.475e-06, "loss": 0.0019, "step": 8916 }, { "epoch": 11.67, "learning_rate": 5.47e-06, "loss": 0.0029, "step": 8917 }, { "epoch": 11.67, "learning_rate": 5.465e-06, "loss": 0.0027, "step": 8918 }, { "epoch": 11.67, "learning_rate": 5.46e-06, "loss": 0.0035, "step": 8919 }, { "epoch": 11.68, "learning_rate": 5.455e-06, "loss": 0.0026, "step": 8920 }, { "epoch": 11.68, "learning_rate": 5.45e-06, "loss": 0.0018, "step": 8921 }, { "epoch": 11.68, "learning_rate": 5.445e-06, "loss": 0.0027, "step": 8922 }, { "epoch": 11.68, "learning_rate": 5.44e-06, "loss": 0.0017, "step": 8923 }, { "epoch": 11.68, "learning_rate": 5.4350000000000005e-06, "loss": 0.0024, "step": 8924 }, { "epoch": 11.68, "learning_rate": 5.4300000000000005e-06, "loss": 0.0051, "step": 8925 }, { "epoch": 11.68, "learning_rate": 5.4250000000000006e-06, "loss": 0.0021, "step": 8926 }, { "epoch": 11.68, "learning_rate": 5.42e-06, "loss": 0.0078, "step": 8927 }, { "epoch": 11.69, "learning_rate": 5.415e-06, "loss": 0.0024, "step": 8928 }, { "epoch": 11.69, "learning_rate": 5.410000000000001e-06, "loss": 0.002, "step": 8929 }, { "epoch": 11.69, "learning_rate": 5.405e-06, "loss": 0.0027, "step": 8930 }, { "epoch": 11.69, "learning_rate": 5.4e-06, "loss": 0.0019, "step": 8931 }, { "epoch": 11.69, "learning_rate": 5.395e-06, "loss": 0.0026, "step": 8932 }, { "epoch": 11.69, "learning_rate": 5.390000000000001e-06, "loss": 0.0029, "step": 8933 }, { "epoch": 11.69, "learning_rate": 5.385e-06, "loss": 0.0038, "step": 8934 }, { "epoch": 11.7, "learning_rate": 5.38e-06, "loss": 0.0025, "step": 8935 }, { "epoch": 11.7, "learning_rate": 5.375e-06, "loss": 0.0027, "step": 8936 }, { "epoch": 11.7, "learning_rate": 5.37e-06, "loss": 0.0033, "step": 8937 }, { "epoch": 11.7, "learning_rate": 5.365e-06, "loss": 0.0026, "step": 8938 }, { "epoch": 11.7, "learning_rate": 5.36e-06, "loss": 0.0149, "step": 8939 }, { "epoch": 11.7, "learning_rate": 5.355e-06, "loss": 0.002, "step": 8940 }, { "epoch": 11.7, "learning_rate": 5.3500000000000004e-06, "loss": 0.0017, "step": 8941 }, { "epoch": 11.7, "learning_rate": 5.345e-06, "loss": 0.0028, "step": 8942 }, { "epoch": 11.71, "learning_rate": 5.3400000000000005e-06, "loss": 0.0016, "step": 8943 }, { "epoch": 11.71, "learning_rate": 5.335000000000001e-06, "loss": 0.0022, "step": 8944 }, { "epoch": 11.71, "learning_rate": 5.330000000000001e-06, "loss": 0.0026, "step": 8945 }, { "epoch": 11.71, "learning_rate": 5.325e-06, "loss": 0.0025, "step": 8946 }, { "epoch": 11.71, "learning_rate": 5.32e-06, "loss": 0.0024, "step": 8947 }, { "epoch": 11.71, "learning_rate": 5.315000000000001e-06, "loss": 0.0028, "step": 8948 }, { "epoch": 11.71, "learning_rate": 5.31e-06, "loss": 0.0036, "step": 8949 }, { "epoch": 11.71, "learning_rate": 5.305e-06, "loss": 0.0016, "step": 8950 }, { "epoch": 11.72, "learning_rate": 5.3e-06, "loss": 0.004, "step": 8951 }, { "epoch": 11.72, "learning_rate": 5.295e-06, "loss": 0.003, "step": 8952 }, { "epoch": 11.72, "learning_rate": 5.29e-06, "loss": 0.0021, "step": 8953 }, { "epoch": 11.72, "learning_rate": 5.285e-06, "loss": 0.0025, "step": 8954 }, { "epoch": 11.72, "learning_rate": 5.28e-06, "loss": 0.0035, "step": 8955 }, { "epoch": 11.72, "learning_rate": 5.275e-06, "loss": 0.003, "step": 8956 }, { "epoch": 11.72, "learning_rate": 5.2699999999999995e-06, "loss": 0.0019, "step": 8957 }, { "epoch": 11.73, "learning_rate": 5.265e-06, "loss": 0.0037, "step": 8958 }, { "epoch": 11.73, "learning_rate": 5.2600000000000005e-06, "loss": 0.0032, "step": 8959 }, { "epoch": 11.73, "learning_rate": 5.2550000000000005e-06, "loss": 0.002, "step": 8960 }, { "epoch": 11.73, "learning_rate": 5.25e-06, "loss": 0.0023, "step": 8961 }, { "epoch": 11.73, "learning_rate": 5.245e-06, "loss": 0.0026, "step": 8962 }, { "epoch": 11.73, "learning_rate": 5.240000000000001e-06, "loss": 0.0026, "step": 8963 }, { "epoch": 11.73, "learning_rate": 5.235000000000001e-06, "loss": 0.0052, "step": 8964 }, { "epoch": 11.73, "learning_rate": 5.23e-06, "loss": 0.0031, "step": 8965 }, { "epoch": 11.74, "learning_rate": 5.225e-06, "loss": 0.0029, "step": 8966 }, { "epoch": 11.74, "learning_rate": 5.220000000000001e-06, "loss": 0.0027, "step": 8967 }, { "epoch": 11.74, "learning_rate": 5.215e-06, "loss": 0.0027, "step": 8968 }, { "epoch": 11.74, "learning_rate": 5.21e-06, "loss": 0.0022, "step": 8969 }, { "epoch": 11.74, "learning_rate": 5.205e-06, "loss": 0.0031, "step": 8970 }, { "epoch": 11.74, "learning_rate": 5.2e-06, "loss": 0.0018, "step": 8971 }, { "epoch": 11.74, "learning_rate": 5.195e-06, "loss": 0.0035, "step": 8972 }, { "epoch": 11.74, "learning_rate": 5.19e-06, "loss": 0.0022, "step": 8973 }, { "epoch": 11.75, "learning_rate": 5.185e-06, "loss": 0.0025, "step": 8974 }, { "epoch": 11.75, "learning_rate": 5.18e-06, "loss": 0.0016, "step": 8975 }, { "epoch": 11.75, "learning_rate": 5.175e-06, "loss": 0.0019, "step": 8976 }, { "epoch": 11.75, "learning_rate": 5.1700000000000005e-06, "loss": 0.004, "step": 8977 }, { "epoch": 11.75, "learning_rate": 5.1650000000000005e-06, "loss": 0.0031, "step": 8978 }, { "epoch": 11.75, "learning_rate": 5.1600000000000006e-06, "loss": 0.0031, "step": 8979 }, { "epoch": 11.75, "learning_rate": 5.155e-06, "loss": 0.0034, "step": 8980 }, { "epoch": 11.76, "learning_rate": 5.15e-06, "loss": 0.0015, "step": 8981 }, { "epoch": 11.76, "learning_rate": 5.145000000000001e-06, "loss": 0.0025, "step": 8982 }, { "epoch": 11.76, "learning_rate": 5.140000000000001e-06, "loss": 0.0051, "step": 8983 }, { "epoch": 11.76, "learning_rate": 5.135e-06, "loss": 0.0047, "step": 8984 }, { "epoch": 11.76, "learning_rate": 5.13e-06, "loss": 0.002, "step": 8985 }, { "epoch": 11.76, "learning_rate": 5.125e-06, "loss": 0.0025, "step": 8986 }, { "epoch": 11.76, "learning_rate": 5.12e-06, "loss": 0.0038, "step": 8987 }, { "epoch": 11.76, "learning_rate": 5.115e-06, "loss": 0.0028, "step": 8988 }, { "epoch": 11.77, "learning_rate": 5.11e-06, "loss": 0.0037, "step": 8989 }, { "epoch": 11.77, "learning_rate": 5.105e-06, "loss": 0.002, "step": 8990 }, { "epoch": 11.77, "learning_rate": 5.1e-06, "loss": 0.0014, "step": 8991 }, { "epoch": 11.77, "learning_rate": 5.095e-06, "loss": 0.0051, "step": 8992 }, { "epoch": 11.77, "learning_rate": 5.09e-06, "loss": 0.0024, "step": 8993 }, { "epoch": 11.77, "learning_rate": 5.0850000000000004e-06, "loss": 0.0028, "step": 8994 }, { "epoch": 11.77, "learning_rate": 5.08e-06, "loss": 0.0024, "step": 8995 }, { "epoch": 11.77, "learning_rate": 5.0750000000000005e-06, "loss": 0.0026, "step": 8996 }, { "epoch": 11.78, "learning_rate": 5.070000000000001e-06, "loss": 0.0022, "step": 8997 }, { "epoch": 11.78, "learning_rate": 5.065000000000001e-06, "loss": 0.0016, "step": 8998 }, { "epoch": 11.78, "learning_rate": 5.06e-06, "loss": 0.004, "step": 8999 }, { "epoch": 11.78, "learning_rate": 5.055e-06, "loss": 0.0029, "step": 9000 }, { "epoch": 11.78, "learning_rate": 5.050000000000001e-06, "loss": 0.0027, "step": 9001 }, { "epoch": 11.78, "learning_rate": 5.045000000000001e-06, "loss": 0.0032, "step": 9002 }, { "epoch": 11.78, "learning_rate": 5.04e-06, "loss": 0.003, "step": 9003 }, { "epoch": 11.79, "learning_rate": 5.035e-06, "loss": 0.0026, "step": 9004 }, { "epoch": 11.79, "learning_rate": 5.03e-06, "loss": 0.0036, "step": 9005 }, { "epoch": 11.79, "learning_rate": 5.025e-06, "loss": 0.0018, "step": 9006 }, { "epoch": 11.79, "learning_rate": 5.02e-06, "loss": 0.0022, "step": 9007 }, { "epoch": 11.79, "learning_rate": 5.015e-06, "loss": 0.0033, "step": 9008 }, { "epoch": 11.79, "learning_rate": 5.01e-06, "loss": 0.0032, "step": 9009 }, { "epoch": 11.79, "learning_rate": 5.005e-06, "loss": 0.0021, "step": 9010 }, { "epoch": 11.79, "learning_rate": 5e-06, "loss": 0.0031, "step": 9011 }, { "epoch": 11.8, "learning_rate": 4.9950000000000005e-06, "loss": 0.0034, "step": 9012 }, { "epoch": 11.8, "learning_rate": 4.9900000000000005e-06, "loss": 0.0031, "step": 9013 }, { "epoch": 11.8, "learning_rate": 4.985e-06, "loss": 0.0032, "step": 9014 }, { "epoch": 11.8, "learning_rate": 4.98e-06, "loss": 0.0026, "step": 9015 }, { "epoch": 11.8, "learning_rate": 4.975000000000001e-06, "loss": 0.002, "step": 9016 }, { "epoch": 11.8, "learning_rate": 4.970000000000001e-06, "loss": 0.0016, "step": 9017 }, { "epoch": 11.8, "learning_rate": 4.965e-06, "loss": 0.0025, "step": 9018 }, { "epoch": 11.8, "learning_rate": 4.96e-06, "loss": 0.0039, "step": 9019 }, { "epoch": 11.81, "learning_rate": 4.955e-06, "loss": 0.0021, "step": 9020 }, { "epoch": 11.81, "learning_rate": 4.950000000000001e-06, "loss": 0.0027, "step": 9021 }, { "epoch": 11.81, "learning_rate": 4.945e-06, "loss": 0.0033, "step": 9022 }, { "epoch": 11.81, "learning_rate": 4.94e-06, "loss": 0.0033, "step": 9023 }, { "epoch": 11.81, "learning_rate": 4.935e-06, "loss": 0.0029, "step": 9024 }, { "epoch": 11.81, "learning_rate": 4.93e-06, "loss": 0.003, "step": 9025 }, { "epoch": 11.81, "learning_rate": 4.925e-06, "loss": 0.0028, "step": 9026 }, { "epoch": 11.82, "learning_rate": 4.92e-06, "loss": 0.0036, "step": 9027 }, { "epoch": 11.82, "learning_rate": 4.915e-06, "loss": 0.002, "step": 9028 }, { "epoch": 11.82, "learning_rate": 4.9100000000000004e-06, "loss": 0.0028, "step": 9029 }, { "epoch": 11.82, "learning_rate": 4.9050000000000005e-06, "loss": 0.0023, "step": 9030 }, { "epoch": 11.82, "learning_rate": 4.9000000000000005e-06, "loss": 0.0031, "step": 9031 }, { "epoch": 11.82, "learning_rate": 4.8950000000000006e-06, "loss": 0.0019, "step": 9032 }, { "epoch": 11.82, "learning_rate": 4.89e-06, "loss": 0.0029, "step": 9033 }, { "epoch": 11.82, "learning_rate": 4.885e-06, "loss": 0.0017, "step": 9034 }, { "epoch": 11.83, "learning_rate": 4.880000000000001e-06, "loss": 0.0028, "step": 9035 }, { "epoch": 11.83, "learning_rate": 4.875000000000001e-06, "loss": 0.0035, "step": 9036 }, { "epoch": 11.83, "learning_rate": 4.87e-06, "loss": 0.0025, "step": 9037 }, { "epoch": 11.83, "learning_rate": 4.865e-06, "loss": 0.0031, "step": 9038 }, { "epoch": 11.83, "learning_rate": 4.86e-06, "loss": 0.0024, "step": 9039 }, { "epoch": 11.83, "learning_rate": 4.855e-06, "loss": 0.0024, "step": 9040 }, { "epoch": 11.83, "learning_rate": 4.85e-06, "loss": 0.0031, "step": 9041 }, { "epoch": 11.84, "learning_rate": 4.845e-06, "loss": 0.0031, "step": 9042 }, { "epoch": 11.84, "learning_rate": 4.84e-06, "loss": 0.0021, "step": 9043 }, { "epoch": 11.84, "learning_rate": 4.835e-06, "loss": 0.003, "step": 9044 }, { "epoch": 11.84, "learning_rate": 4.83e-06, "loss": 0.0029, "step": 9045 }, { "epoch": 11.84, "learning_rate": 4.825e-06, "loss": 0.0033, "step": 9046 }, { "epoch": 11.84, "learning_rate": 4.8200000000000004e-06, "loss": 0.0024, "step": 9047 }, { "epoch": 11.84, "learning_rate": 4.8150000000000005e-06, "loss": 0.0034, "step": 9048 }, { "epoch": 11.84, "learning_rate": 4.81e-06, "loss": 0.0029, "step": 9049 }, { "epoch": 11.85, "learning_rate": 4.805000000000001e-06, "loss": 0.0018, "step": 9050 }, { "epoch": 11.85, "learning_rate": 4.800000000000001e-06, "loss": 0.0029, "step": 9051 }, { "epoch": 11.85, "learning_rate": 4.795e-06, "loss": 0.0022, "step": 9052 }, { "epoch": 11.85, "learning_rate": 4.79e-06, "loss": 0.003, "step": 9053 }, { "epoch": 11.85, "learning_rate": 4.785e-06, "loss": 0.004, "step": 9054 }, { "epoch": 11.85, "learning_rate": 4.780000000000001e-06, "loss": 0.0022, "step": 9055 }, { "epoch": 11.85, "learning_rate": 4.775e-06, "loss": 0.0023, "step": 9056 }, { "epoch": 11.85, "learning_rate": 4.77e-06, "loss": 0.0019, "step": 9057 }, { "epoch": 11.86, "learning_rate": 4.765e-06, "loss": 0.0026, "step": 9058 }, { "epoch": 11.86, "learning_rate": 4.76e-06, "loss": 0.0034, "step": 9059 }, { "epoch": 11.86, "learning_rate": 4.755e-06, "loss": 0.0028, "step": 9060 }, { "epoch": 11.86, "learning_rate": 4.75e-06, "loss": 0.0031, "step": 9061 }, { "epoch": 11.86, "learning_rate": 4.745e-06, "loss": 0.0031, "step": 9062 }, { "epoch": 11.86, "learning_rate": 4.74e-06, "loss": 0.0027, "step": 9063 }, { "epoch": 11.86, "learning_rate": 4.735e-06, "loss": 0.0028, "step": 9064 }, { "epoch": 11.87, "learning_rate": 4.7300000000000005e-06, "loss": 0.0021, "step": 9065 }, { "epoch": 11.87, "learning_rate": 4.7250000000000005e-06, "loss": 0.0034, "step": 9066 }, { "epoch": 11.87, "learning_rate": 4.72e-06, "loss": 0.0016, "step": 9067 }, { "epoch": 11.87, "learning_rate": 4.715e-06, "loss": 0.0029, "step": 9068 }, { "epoch": 11.87, "learning_rate": 4.710000000000001e-06, "loss": 0.0021, "step": 9069 }, { "epoch": 11.87, "learning_rate": 4.705000000000001e-06, "loss": 0.0031, "step": 9070 }, { "epoch": 11.87, "learning_rate": 4.7e-06, "loss": 0.0019, "step": 9071 }, { "epoch": 11.87, "learning_rate": 4.695e-06, "loss": 0.0023, "step": 9072 }, { "epoch": 11.88, "learning_rate": 4.69e-06, "loss": 0.0031, "step": 9073 }, { "epoch": 11.88, "learning_rate": 4.685000000000001e-06, "loss": 0.0031, "step": 9074 }, { "epoch": 11.88, "learning_rate": 4.68e-06, "loss": 0.0025, "step": 9075 }, { "epoch": 11.88, "learning_rate": 4.675e-06, "loss": 0.0029, "step": 9076 }, { "epoch": 11.88, "learning_rate": 4.67e-06, "loss": 0.0036, "step": 9077 }, { "epoch": 11.88, "learning_rate": 4.665e-06, "loss": 0.0013, "step": 9078 }, { "epoch": 11.88, "learning_rate": 4.66e-06, "loss": 0.0016, "step": 9079 }, { "epoch": 11.88, "learning_rate": 4.655e-06, "loss": 0.0023, "step": 9080 }, { "epoch": 11.89, "learning_rate": 4.65e-06, "loss": 0.0015, "step": 9081 }, { "epoch": 11.89, "learning_rate": 4.645e-06, "loss": 0.0019, "step": 9082 }, { "epoch": 11.89, "learning_rate": 4.64e-06, "loss": 0.001, "step": 9083 }, { "epoch": 11.89, "learning_rate": 4.6350000000000005e-06, "loss": 0.0026, "step": 9084 }, { "epoch": 11.89, "learning_rate": 4.6300000000000006e-06, "loss": 0.0036, "step": 9085 }, { "epoch": 11.89, "learning_rate": 4.625e-06, "loss": 0.0027, "step": 9086 }, { "epoch": 11.89, "learning_rate": 4.62e-06, "loss": 0.0041, "step": 9087 }, { "epoch": 11.9, "learning_rate": 4.615e-06, "loss": 0.0038, "step": 9088 }, { "epoch": 11.9, "learning_rate": 4.610000000000001e-06, "loss": 0.0028, "step": 9089 }, { "epoch": 11.9, "learning_rate": 4.605e-06, "loss": 0.0028, "step": 9090 }, { "epoch": 11.9, "learning_rate": 4.6e-06, "loss": 0.0028, "step": 9091 }, { "epoch": 11.9, "learning_rate": 4.595e-06, "loss": 0.0035, "step": 9092 }, { "epoch": 11.9, "learning_rate": 4.590000000000001e-06, "loss": 0.0022, "step": 9093 }, { "epoch": 11.9, "learning_rate": 4.585e-06, "loss": 0.002, "step": 9094 }, { "epoch": 11.9, "learning_rate": 4.58e-06, "loss": 0.0025, "step": 9095 }, { "epoch": 11.91, "learning_rate": 4.575e-06, "loss": 0.0021, "step": 9096 }, { "epoch": 11.91, "learning_rate": 4.57e-06, "loss": 0.0033, "step": 9097 }, { "epoch": 11.91, "learning_rate": 4.565e-06, "loss": 0.0045, "step": 9098 }, { "epoch": 11.91, "learning_rate": 4.56e-06, "loss": 0.0027, "step": 9099 }, { "epoch": 11.91, "learning_rate": 4.5550000000000004e-06, "loss": 0.0034, "step": 9100 }, { "epoch": 11.91, "learning_rate": 4.5500000000000005e-06, "loss": 0.0013, "step": 9101 }, { "epoch": 11.91, "learning_rate": 4.545e-06, "loss": 0.0024, "step": 9102 }, { "epoch": 11.91, "learning_rate": 4.540000000000001e-06, "loss": 0.0015, "step": 9103 }, { "epoch": 11.92, "learning_rate": 4.535000000000001e-06, "loss": 0.0022, "step": 9104 }, { "epoch": 11.92, "learning_rate": 4.53e-06, "loss": 0.0023, "step": 9105 }, { "epoch": 11.92, "learning_rate": 4.525e-06, "loss": 0.0022, "step": 9106 }, { "epoch": 11.92, "learning_rate": 4.52e-06, "loss": 0.0032, "step": 9107 }, { "epoch": 11.92, "learning_rate": 4.515000000000001e-06, "loss": 0.0022, "step": 9108 }, { "epoch": 11.92, "learning_rate": 4.51e-06, "loss": 0.0023, "step": 9109 }, { "epoch": 11.92, "learning_rate": 4.505e-06, "loss": 0.0024, "step": 9110 }, { "epoch": 11.93, "learning_rate": 4.5e-06, "loss": 0.0045, "step": 9111 }, { "epoch": 11.93, "learning_rate": 4.495e-06, "loss": 0.0024, "step": 9112 }, { "epoch": 11.93, "learning_rate": 4.49e-06, "loss": 0.0024, "step": 9113 }, { "epoch": 11.93, "learning_rate": 4.485e-06, "loss": 0.0026, "step": 9114 }, { "epoch": 11.93, "learning_rate": 4.48e-06, "loss": 0.0026, "step": 9115 }, { "epoch": 11.93, "learning_rate": 4.475e-06, "loss": 0.0036, "step": 9116 }, { "epoch": 11.93, "learning_rate": 4.4699999999999996e-06, "loss": 0.0028, "step": 9117 }, { "epoch": 11.93, "learning_rate": 4.4650000000000004e-06, "loss": 0.0036, "step": 9118 }, { "epoch": 11.94, "learning_rate": 4.4600000000000005e-06, "loss": 0.0015, "step": 9119 }, { "epoch": 11.94, "learning_rate": 4.4550000000000005e-06, "loss": 0.003, "step": 9120 }, { "epoch": 11.94, "learning_rate": 4.45e-06, "loss": 0.0028, "step": 9121 }, { "epoch": 11.94, "learning_rate": 4.445000000000001e-06, "loss": 0.0021, "step": 9122 }, { "epoch": 11.94, "learning_rate": 4.440000000000001e-06, "loss": 0.0034, "step": 9123 }, { "epoch": 11.94, "learning_rate": 4.435e-06, "loss": 0.0041, "step": 9124 }, { "epoch": 11.94, "learning_rate": 4.43e-06, "loss": 0.0021, "step": 9125 }, { "epoch": 11.95, "learning_rate": 4.425e-06, "loss": 0.0021, "step": 9126 }, { "epoch": 11.95, "learning_rate": 4.420000000000001e-06, "loss": 0.003, "step": 9127 }, { "epoch": 11.95, "learning_rate": 4.415e-06, "loss": 0.0025, "step": 9128 }, { "epoch": 11.95, "learning_rate": 4.41e-06, "loss": 0.0043, "step": 9129 }, { "epoch": 11.95, "learning_rate": 4.405e-06, "loss": 0.0022, "step": 9130 }, { "epoch": 11.95, "learning_rate": 4.4e-06, "loss": 0.0035, "step": 9131 }, { "epoch": 11.95, "learning_rate": 4.395e-06, "loss": 0.0021, "step": 9132 }, { "epoch": 11.95, "learning_rate": 4.39e-06, "loss": 0.0022, "step": 9133 }, { "epoch": 11.96, "learning_rate": 4.385e-06, "loss": 0.0014, "step": 9134 }, { "epoch": 11.96, "learning_rate": 4.38e-06, "loss": 0.0031, "step": 9135 }, { "epoch": 11.96, "learning_rate": 4.375e-06, "loss": 0.0014, "step": 9136 }, { "epoch": 11.96, "learning_rate": 4.3700000000000005e-06, "loss": 0.0026, "step": 9137 }, { "epoch": 11.96, "learning_rate": 4.3650000000000006e-06, "loss": 0.0025, "step": 9138 }, { "epoch": 11.96, "learning_rate": 4.360000000000001e-06, "loss": 0.0042, "step": 9139 }, { "epoch": 11.96, "learning_rate": 4.355e-06, "loss": 0.0021, "step": 9140 }, { "epoch": 11.96, "learning_rate": 4.35e-06, "loss": 0.0022, "step": 9141 }, { "epoch": 11.97, "learning_rate": 4.345000000000001e-06, "loss": 0.004, "step": 9142 }, { "epoch": 11.97, "learning_rate": 4.34e-06, "loss": 0.0027, "step": 9143 }, { "epoch": 11.97, "learning_rate": 4.335e-06, "loss": 0.0038, "step": 9144 }, { "epoch": 11.97, "learning_rate": 4.33e-06, "loss": 0.0024, "step": 9145 }, { "epoch": 11.97, "learning_rate": 4.325e-06, "loss": 0.0021, "step": 9146 }, { "epoch": 11.97, "learning_rate": 4.32e-06, "loss": 0.0042, "step": 9147 }, { "epoch": 11.97, "learning_rate": 4.315e-06, "loss": 0.0016, "step": 9148 }, { "epoch": 11.98, "learning_rate": 4.31e-06, "loss": 0.0037, "step": 9149 }, { "epoch": 11.98, "learning_rate": 4.305e-06, "loss": 0.0019, "step": 9150 }, { "epoch": 11.98, "learning_rate": 4.2999999999999995e-06, "loss": 0.0026, "step": 9151 }, { "epoch": 11.98, "learning_rate": 4.295e-06, "loss": 0.0042, "step": 9152 }, { "epoch": 11.98, "learning_rate": 4.2900000000000004e-06, "loss": 0.0029, "step": 9153 }, { "epoch": 11.98, "learning_rate": 4.2850000000000005e-06, "loss": 0.0031, "step": 9154 }, { "epoch": 11.98, "learning_rate": 4.28e-06, "loss": 0.0038, "step": 9155 }, { "epoch": 11.98, "learning_rate": 4.2750000000000006e-06, "loss": 0.002, "step": 9156 }, { "epoch": 11.99, "learning_rate": 4.270000000000001e-06, "loss": 0.0022, "step": 9157 }, { "epoch": 11.99, "learning_rate": 4.265e-06, "loss": 0.0043, "step": 9158 }, { "epoch": 11.99, "learning_rate": 4.26e-06, "loss": 0.0025, "step": 9159 }, { "epoch": 11.99, "learning_rate": 4.255e-06, "loss": 0.0025, "step": 9160 }, { "epoch": 11.99, "learning_rate": 4.250000000000001e-06, "loss": 0.0024, "step": 9161 }, { "epoch": 11.99, "learning_rate": 4.245e-06, "loss": 0.0027, "step": 9162 }, { "epoch": 11.99, "learning_rate": 4.24e-06, "loss": 0.0026, "step": 9163 }, { "epoch": 11.99, "learning_rate": 4.235e-06, "loss": 0.003, "step": 9164 }, { "epoch": 12.0, "learning_rate": 4.23e-06, "loss": 0.0035, "step": 9165 }, { "epoch": 12.0, "learning_rate": 4.225e-06, "loss": 0.0032, "step": 9166 }, { "epoch": 12.0, "learning_rate": 4.22e-06, "loss": 0.0026, "step": 9167 }, { "epoch": 12.0, "learning_rate": 4.215e-06, "loss": 0.0026, "step": 9168 }, { "epoch": 12.0, "learning_rate": 4.21e-06, "loss": 0.0013, "step": 9169 }, { "epoch": 12.0, "learning_rate": 4.2049999999999996e-06, "loss": 0.003, "step": 9170 }, { "epoch": 12.0, "learning_rate": 4.2000000000000004e-06, "loss": 0.0039, "step": 9171 }, { "epoch": 12.01, "learning_rate": 4.1950000000000005e-06, "loss": 0.0034, "step": 9172 }, { "epoch": 12.01, "learning_rate": 4.1900000000000005e-06, "loss": 0.0018, "step": 9173 }, { "epoch": 12.01, "learning_rate": 4.185e-06, "loss": 0.0011, "step": 9174 }, { "epoch": 12.01, "learning_rate": 4.18e-06, "loss": 0.0024, "step": 9175 }, { "epoch": 12.01, "learning_rate": 4.175000000000001e-06, "loss": 0.0031, "step": 9176 }, { "epoch": 12.01, "learning_rate": 4.17e-06, "loss": 0.002, "step": 9177 }, { "epoch": 12.01, "learning_rate": 4.165e-06, "loss": 0.0012, "step": 9178 }, { "epoch": 12.01, "learning_rate": 4.16e-06, "loss": 0.003, "step": 9179 }, { "epoch": 12.02, "learning_rate": 4.155e-06, "loss": 0.0014, "step": 9180 }, { "epoch": 12.02, "learning_rate": 4.15e-06, "loss": 0.0034, "step": 9181 }, { "epoch": 12.02, "learning_rate": 4.145e-06, "loss": 0.0015, "step": 9182 }, { "epoch": 12.02, "learning_rate": 4.14e-06, "loss": 0.0022, "step": 9183 }, { "epoch": 12.02, "learning_rate": 4.135e-06, "loss": 0.0032, "step": 9184 }, { "epoch": 12.02, "learning_rate": 4.13e-06, "loss": 0.0026, "step": 9185 }, { "epoch": 12.02, "learning_rate": 4.125e-06, "loss": 0.0018, "step": 9186 }, { "epoch": 12.02, "learning_rate": 4.12e-06, "loss": 0.002, "step": 9187 }, { "epoch": 12.03, "learning_rate": 4.115e-06, "loss": 0.0025, "step": 9188 }, { "epoch": 12.03, "learning_rate": 4.11e-06, "loss": 0.0023, "step": 9189 }, { "epoch": 12.03, "learning_rate": 4.1050000000000005e-06, "loss": 0.0014, "step": 9190 }, { "epoch": 12.03, "learning_rate": 4.1000000000000006e-06, "loss": 0.0028, "step": 9191 }, { "epoch": 12.03, "learning_rate": 4.095000000000001e-06, "loss": 0.0018, "step": 9192 }, { "epoch": 12.03, "learning_rate": 4.09e-06, "loss": 0.0025, "step": 9193 }, { "epoch": 12.03, "learning_rate": 4.085e-06, "loss": 0.001, "step": 9194 }, { "epoch": 12.04, "learning_rate": 4.080000000000001e-06, "loss": 0.0029, "step": 9195 }, { "epoch": 12.04, "learning_rate": 4.075e-06, "loss": 0.0016, "step": 9196 }, { "epoch": 12.04, "learning_rate": 4.07e-06, "loss": 0.0022, "step": 9197 }, { "epoch": 12.04, "learning_rate": 4.065e-06, "loss": 0.0022, "step": 9198 }, { "epoch": 12.04, "learning_rate": 4.06e-06, "loss": 0.0027, "step": 9199 }, { "epoch": 12.04, "learning_rate": 4.055e-06, "loss": 0.0028, "step": 9200 }, { "epoch": 12.04, "learning_rate": 4.05e-06, "loss": 0.0043, "step": 9201 }, { "epoch": 12.04, "learning_rate": 4.045e-06, "loss": 0.0022, "step": 9202 }, { "epoch": 12.05, "learning_rate": 4.04e-06, "loss": 0.0015, "step": 9203 }, { "epoch": 12.05, "learning_rate": 4.0349999999999995e-06, "loss": 0.0024, "step": 9204 }, { "epoch": 12.05, "learning_rate": 4.03e-06, "loss": 0.0013, "step": 9205 }, { "epoch": 12.05, "learning_rate": 4.0250000000000004e-06, "loss": 0.002, "step": 9206 }, { "epoch": 12.05, "learning_rate": 4.0200000000000005e-06, "loss": 0.0028, "step": 9207 }, { "epoch": 12.05, "learning_rate": 4.015e-06, "loss": 0.002, "step": 9208 }, { "epoch": 12.05, "learning_rate": 4.01e-06, "loss": 0.0023, "step": 9209 }, { "epoch": 12.05, "learning_rate": 4.005000000000001e-06, "loss": 0.0008, "step": 9210 }, { "epoch": 12.06, "learning_rate": 4.000000000000001e-06, "loss": 0.0027, "step": 9211 }, { "epoch": 12.06, "learning_rate": 3.995e-06, "loss": 0.0039, "step": 9212 }, { "epoch": 12.06, "learning_rate": 3.99e-06, "loss": 0.0036, "step": 9213 }, { "epoch": 12.06, "learning_rate": 3.985e-06, "loss": 0.0017, "step": 9214 }, { "epoch": 12.06, "learning_rate": 3.98e-06, "loss": 0.0023, "step": 9215 }, { "epoch": 12.06, "learning_rate": 3.975e-06, "loss": 0.0027, "step": 9216 }, { "epoch": 12.06, "learning_rate": 3.97e-06, "loss": 0.0015, "step": 9217 }, { "epoch": 12.07, "learning_rate": 3.965e-06, "loss": 0.0018, "step": 9218 }, { "epoch": 12.07, "learning_rate": 3.96e-06, "loss": 0.0013, "step": 9219 }, { "epoch": 12.07, "learning_rate": 3.955e-06, "loss": 0.0041, "step": 9220 }, { "epoch": 12.07, "learning_rate": 3.95e-06, "loss": 0.0021, "step": 9221 }, { "epoch": 12.07, "learning_rate": 3.945e-06, "loss": 0.0029, "step": 9222 }, { "epoch": 12.07, "learning_rate": 3.9399999999999995e-06, "loss": 0.0027, "step": 9223 }, { "epoch": 12.07, "learning_rate": 3.9350000000000004e-06, "loss": 0.0023, "step": 9224 }, { "epoch": 12.07, "learning_rate": 3.9300000000000005e-06, "loss": 0.0019, "step": 9225 }, { "epoch": 12.08, "learning_rate": 3.9250000000000005e-06, "loss": 0.0024, "step": 9226 }, { "epoch": 12.08, "learning_rate": 3.92e-06, "loss": 0.002, "step": 9227 }, { "epoch": 12.08, "learning_rate": 3.915e-06, "loss": 0.0029, "step": 9228 }, { "epoch": 12.08, "learning_rate": 3.910000000000001e-06, "loss": 0.0021, "step": 9229 }, { "epoch": 12.08, "learning_rate": 3.905000000000001e-06, "loss": 0.0018, "step": 9230 }, { "epoch": 12.08, "learning_rate": 3.9e-06, "loss": 0.0021, "step": 9231 }, { "epoch": 12.08, "learning_rate": 3.895e-06, "loss": 0.0015, "step": 9232 }, { "epoch": 12.09, "learning_rate": 3.89e-06, "loss": 0.0016, "step": 9233 }, { "epoch": 12.09, "learning_rate": 3.885e-06, "loss": 0.0019, "step": 9234 }, { "epoch": 12.09, "learning_rate": 3.88e-06, "loss": 0.0036, "step": 9235 }, { "epoch": 12.09, "learning_rate": 3.875e-06, "loss": 0.0041, "step": 9236 }, { "epoch": 12.09, "learning_rate": 3.87e-06, "loss": 0.0025, "step": 9237 }, { "epoch": 12.09, "learning_rate": 3.865e-06, "loss": 0.0025, "step": 9238 }, { "epoch": 12.09, "learning_rate": 3.86e-06, "loss": 0.0023, "step": 9239 }, { "epoch": 12.09, "learning_rate": 3.855e-06, "loss": 0.0028, "step": 9240 }, { "epoch": 12.1, "learning_rate": 3.85e-06, "loss": 0.0017, "step": 9241 }, { "epoch": 12.1, "learning_rate": 3.845e-06, "loss": 0.0036, "step": 9242 }, { "epoch": 12.1, "learning_rate": 3.84e-06, "loss": 0.0018, "step": 9243 }, { "epoch": 12.1, "learning_rate": 3.8350000000000006e-06, "loss": 0.0021, "step": 9244 }, { "epoch": 12.1, "learning_rate": 3.830000000000001e-06, "loss": 0.0016, "step": 9245 }, { "epoch": 12.1, "learning_rate": 3.825e-06, "loss": 0.0021, "step": 9246 }, { "epoch": 12.1, "learning_rate": 3.82e-06, "loss": 0.0029, "step": 9247 }, { "epoch": 12.1, "learning_rate": 3.815000000000001e-06, "loss": 0.0026, "step": 9248 }, { "epoch": 12.11, "learning_rate": 3.8100000000000004e-06, "loss": 0.0017, "step": 9249 }, { "epoch": 12.11, "learning_rate": 3.8050000000000004e-06, "loss": 0.0018, "step": 9250 }, { "epoch": 12.11, "learning_rate": 3.8e-06, "loss": 0.0022, "step": 9251 }, { "epoch": 12.11, "learning_rate": 3.795e-06, "loss": 0.003, "step": 9252 }, { "epoch": 12.11, "learning_rate": 3.7900000000000006e-06, "loss": 0.0023, "step": 9253 }, { "epoch": 12.11, "learning_rate": 3.785e-06, "loss": 0.0026, "step": 9254 }, { "epoch": 12.11, "learning_rate": 3.7800000000000002e-06, "loss": 0.0021, "step": 9255 }, { "epoch": 12.12, "learning_rate": 3.775e-06, "loss": 0.0028, "step": 9256 }, { "epoch": 12.12, "learning_rate": 3.77e-06, "loss": 0.002, "step": 9257 }, { "epoch": 12.12, "learning_rate": 3.7650000000000004e-06, "loss": 0.0016, "step": 9258 }, { "epoch": 12.12, "learning_rate": 3.7600000000000004e-06, "loss": 0.002, "step": 9259 }, { "epoch": 12.12, "learning_rate": 3.755e-06, "loss": 0.003, "step": 9260 }, { "epoch": 12.12, "learning_rate": 3.75e-06, "loss": 0.0022, "step": 9261 }, { "epoch": 12.12, "learning_rate": 3.7449999999999997e-06, "loss": 0.0017, "step": 9262 }, { "epoch": 12.12, "learning_rate": 3.7400000000000006e-06, "loss": 0.0023, "step": 9263 }, { "epoch": 12.13, "learning_rate": 3.7350000000000002e-06, "loss": 0.0006, "step": 9264 }, { "epoch": 12.13, "learning_rate": 3.7300000000000003e-06, "loss": 0.0023, "step": 9265 }, { "epoch": 12.13, "learning_rate": 3.725e-06, "loss": 0.0019, "step": 9266 }, { "epoch": 12.13, "learning_rate": 3.72e-06, "loss": 0.0015, "step": 9267 }, { "epoch": 12.13, "learning_rate": 3.7150000000000004e-06, "loss": 0.0015, "step": 9268 }, { "epoch": 12.13, "learning_rate": 3.7100000000000005e-06, "loss": 0.0017, "step": 9269 }, { "epoch": 12.13, "learning_rate": 3.705e-06, "loss": 0.0021, "step": 9270 }, { "epoch": 12.13, "learning_rate": 3.7e-06, "loss": 0.0018, "step": 9271 }, { "epoch": 12.14, "learning_rate": 3.6949999999999998e-06, "loss": 0.0025, "step": 9272 }, { "epoch": 12.14, "learning_rate": 3.6900000000000002e-06, "loss": 0.002, "step": 9273 }, { "epoch": 12.14, "learning_rate": 3.6850000000000003e-06, "loss": 0.0029, "step": 9274 }, { "epoch": 12.14, "learning_rate": 3.68e-06, "loss": 0.0029, "step": 9275 }, { "epoch": 12.14, "learning_rate": 3.675e-06, "loss": 0.0024, "step": 9276 }, { "epoch": 12.14, "learning_rate": 3.6700000000000004e-06, "loss": 0.001, "step": 9277 }, { "epoch": 12.14, "learning_rate": 3.6650000000000005e-06, "loss": 0.0027, "step": 9278 }, { "epoch": 12.15, "learning_rate": 3.66e-06, "loss": 0.0018, "step": 9279 }, { "epoch": 12.15, "learning_rate": 3.655e-06, "loss": 0.0023, "step": 9280 }, { "epoch": 12.15, "learning_rate": 3.6499999999999998e-06, "loss": 0.0017, "step": 9281 }, { "epoch": 12.15, "learning_rate": 3.6450000000000007e-06, "loss": 0.0024, "step": 9282 }, { "epoch": 12.15, "learning_rate": 3.6400000000000003e-06, "loss": 0.001, "step": 9283 }, { "epoch": 12.15, "learning_rate": 3.6350000000000003e-06, "loss": 0.0016, "step": 9284 }, { "epoch": 12.15, "learning_rate": 3.63e-06, "loss": 0.0018, "step": 9285 }, { "epoch": 12.15, "learning_rate": 3.625e-06, "loss": 0.0037, "step": 9286 }, { "epoch": 12.16, "learning_rate": 3.6200000000000005e-06, "loss": 0.002, "step": 9287 }, { "epoch": 12.16, "learning_rate": 3.6150000000000005e-06, "loss": 0.0018, "step": 9288 }, { "epoch": 12.16, "learning_rate": 3.61e-06, "loss": 0.0017, "step": 9289 }, { "epoch": 12.16, "learning_rate": 3.6050000000000002e-06, "loss": 0.0013, "step": 9290 }, { "epoch": 12.16, "learning_rate": 3.6e-06, "loss": 0.0028, "step": 9291 }, { "epoch": 12.16, "learning_rate": 3.5950000000000003e-06, "loss": 0.0023, "step": 9292 }, { "epoch": 12.16, "learning_rate": 3.5900000000000004e-06, "loss": 0.0012, "step": 9293 }, { "epoch": 12.16, "learning_rate": 3.585e-06, "loss": 0.0025, "step": 9294 }, { "epoch": 12.17, "learning_rate": 3.58e-06, "loss": 0.0019, "step": 9295 }, { "epoch": 12.17, "learning_rate": 3.575e-06, "loss": 0.0031, "step": 9296 }, { "epoch": 12.17, "learning_rate": 3.5700000000000005e-06, "loss": 0.0018, "step": 9297 }, { "epoch": 12.17, "learning_rate": 3.565e-06, "loss": 0.0019, "step": 9298 }, { "epoch": 12.17, "learning_rate": 3.5600000000000002e-06, "loss": 0.002, "step": 9299 }, { "epoch": 12.17, "learning_rate": 3.555e-06, "loss": 0.002, "step": 9300 }, { "epoch": 12.17, "learning_rate": 3.55e-06, "loss": 0.0033, "step": 9301 }, { "epoch": 12.18, "learning_rate": 3.5450000000000004e-06, "loss": 0.0025, "step": 9302 }, { "epoch": 12.18, "learning_rate": 3.5400000000000004e-06, "loss": 0.002, "step": 9303 }, { "epoch": 12.18, "learning_rate": 3.535e-06, "loss": 0.0013, "step": 9304 }, { "epoch": 12.18, "learning_rate": 3.53e-06, "loss": 0.0026, "step": 9305 }, { "epoch": 12.18, "learning_rate": 3.5249999999999997e-06, "loss": 0.0019, "step": 9306 }, { "epoch": 12.18, "learning_rate": 3.52e-06, "loss": 0.0042, "step": 9307 }, { "epoch": 12.18, "learning_rate": 3.5150000000000002e-06, "loss": 0.0019, "step": 9308 }, { "epoch": 12.18, "learning_rate": 3.5100000000000003e-06, "loss": 0.0018, "step": 9309 }, { "epoch": 12.19, "learning_rate": 3.505e-06, "loss": 0.003, "step": 9310 }, { "epoch": 12.19, "learning_rate": 3.5000000000000004e-06, "loss": 0.0021, "step": 9311 }, { "epoch": 12.19, "learning_rate": 3.4950000000000004e-06, "loss": 0.0031, "step": 9312 }, { "epoch": 12.19, "learning_rate": 3.49e-06, "loss": 0.0026, "step": 9313 }, { "epoch": 12.19, "learning_rate": 3.485e-06, "loss": 0.0017, "step": 9314 }, { "epoch": 12.19, "learning_rate": 3.4799999999999997e-06, "loss": 0.0023, "step": 9315 }, { "epoch": 12.19, "learning_rate": 3.4750000000000006e-06, "loss": 0.0027, "step": 9316 }, { "epoch": 12.2, "learning_rate": 3.4700000000000002e-06, "loss": 0.0025, "step": 9317 }, { "epoch": 12.2, "learning_rate": 3.4650000000000003e-06, "loss": 0.0018, "step": 9318 }, { "epoch": 12.2, "learning_rate": 3.46e-06, "loss": 0.0045, "step": 9319 }, { "epoch": 12.2, "learning_rate": 3.455e-06, "loss": 0.0019, "step": 9320 }, { "epoch": 12.2, "learning_rate": 3.4500000000000004e-06, "loss": 0.0021, "step": 9321 }, { "epoch": 12.2, "learning_rate": 3.4450000000000005e-06, "loss": 0.0017, "step": 9322 }, { "epoch": 12.2, "learning_rate": 3.44e-06, "loss": 0.0018, "step": 9323 }, { "epoch": 12.2, "learning_rate": 3.435e-06, "loss": 0.0024, "step": 9324 }, { "epoch": 12.21, "learning_rate": 3.4299999999999998e-06, "loss": 0.0026, "step": 9325 }, { "epoch": 12.21, "learning_rate": 3.4250000000000002e-06, "loss": 0.0016, "step": 9326 }, { "epoch": 12.21, "learning_rate": 3.4200000000000003e-06, "loss": 0.0032, "step": 9327 }, { "epoch": 12.21, "learning_rate": 3.4150000000000003e-06, "loss": 0.0018, "step": 9328 }, { "epoch": 12.21, "learning_rate": 3.41e-06, "loss": 0.0018, "step": 9329 }, { "epoch": 12.21, "learning_rate": 3.405e-06, "loss": 0.0014, "step": 9330 }, { "epoch": 12.21, "learning_rate": 3.4000000000000005e-06, "loss": 0.0024, "step": 9331 }, { "epoch": 12.21, "learning_rate": 3.395e-06, "loss": 0.0011, "step": 9332 }, { "epoch": 12.22, "learning_rate": 3.39e-06, "loss": 0.0021, "step": 9333 }, { "epoch": 12.22, "learning_rate": 3.3849999999999998e-06, "loss": 0.0021, "step": 9334 }, { "epoch": 12.22, "learning_rate": 3.38e-06, "loss": 0.0022, "step": 9335 }, { "epoch": 12.22, "learning_rate": 3.3750000000000003e-06, "loss": 0.0023, "step": 9336 }, { "epoch": 12.22, "learning_rate": 3.3700000000000003e-06, "loss": 0.0022, "step": 9337 }, { "epoch": 12.22, "learning_rate": 3.365e-06, "loss": 0.0026, "step": 9338 }, { "epoch": 12.22, "learning_rate": 3.36e-06, "loss": 0.0017, "step": 9339 }, { "epoch": 12.23, "learning_rate": 3.3550000000000005e-06, "loss": 0.0022, "step": 9340 }, { "epoch": 12.23, "learning_rate": 3.3500000000000005e-06, "loss": 0.0025, "step": 9341 }, { "epoch": 12.23, "learning_rate": 3.345e-06, "loss": 0.0025, "step": 9342 }, { "epoch": 12.23, "learning_rate": 3.34e-06, "loss": 0.0018, "step": 9343 }, { "epoch": 12.23, "learning_rate": 3.335e-06, "loss": 0.002, "step": 9344 }, { "epoch": 12.23, "learning_rate": 3.3300000000000003e-06, "loss": 0.0019, "step": 9345 }, { "epoch": 12.23, "learning_rate": 3.3250000000000004e-06, "loss": 0.0023, "step": 9346 }, { "epoch": 12.23, "learning_rate": 3.3200000000000004e-06, "loss": 0.0011, "step": 9347 }, { "epoch": 12.24, "learning_rate": 3.315e-06, "loss": 0.0014, "step": 9348 }, { "epoch": 12.24, "learning_rate": 3.31e-06, "loss": 0.0092, "step": 9349 }, { "epoch": 12.24, "learning_rate": 3.3050000000000005e-06, "loss": 0.0016, "step": 9350 }, { "epoch": 12.24, "learning_rate": 3.3e-06, "loss": 0.0027, "step": 9351 }, { "epoch": 12.24, "learning_rate": 3.2950000000000002e-06, "loss": 0.0016, "step": 9352 }, { "epoch": 12.24, "learning_rate": 3.29e-06, "loss": 0.0011, "step": 9353 }, { "epoch": 12.24, "learning_rate": 3.285e-06, "loss": 0.0012, "step": 9354 }, { "epoch": 12.24, "learning_rate": 3.2800000000000004e-06, "loss": 0.0027, "step": 9355 }, { "epoch": 12.25, "learning_rate": 3.2750000000000004e-06, "loss": 0.0018, "step": 9356 }, { "epoch": 12.25, "learning_rate": 3.27e-06, "loss": 0.0011, "step": 9357 }, { "epoch": 12.25, "learning_rate": 3.265e-06, "loss": 0.0029, "step": 9358 }, { "epoch": 12.25, "learning_rate": 3.2599999999999997e-06, "loss": 0.0016, "step": 9359 }, { "epoch": 12.25, "learning_rate": 3.2550000000000006e-06, "loss": 0.0037, "step": 9360 }, { "epoch": 12.25, "learning_rate": 3.2500000000000002e-06, "loss": 0.0025, "step": 9361 }, { "epoch": 12.25, "learning_rate": 3.2450000000000003e-06, "loss": 0.0028, "step": 9362 }, { "epoch": 12.26, "learning_rate": 3.24e-06, "loss": 0.0014, "step": 9363 }, { "epoch": 12.26, "learning_rate": 3.235e-06, "loss": 0.0021, "step": 9364 }, { "epoch": 12.26, "learning_rate": 3.2300000000000004e-06, "loss": 0.0024, "step": 9365 }, { "epoch": 12.26, "learning_rate": 3.225e-06, "loss": 0.0013, "step": 9366 }, { "epoch": 12.26, "learning_rate": 3.22e-06, "loss": 0.0015, "step": 9367 }, { "epoch": 12.26, "learning_rate": 3.215e-06, "loss": 0.0019, "step": 9368 }, { "epoch": 12.26, "learning_rate": 3.2099999999999998e-06, "loss": 0.001, "step": 9369 }, { "epoch": 12.26, "learning_rate": 3.2050000000000002e-06, "loss": 0.002, "step": 9370 }, { "epoch": 12.27, "learning_rate": 3.2000000000000003e-06, "loss": 0.0026, "step": 9371 }, { "epoch": 12.27, "learning_rate": 3.195e-06, "loss": 0.0027, "step": 9372 }, { "epoch": 12.27, "learning_rate": 3.19e-06, "loss": 0.0023, "step": 9373 }, { "epoch": 12.27, "learning_rate": 3.1850000000000004e-06, "loss": 0.0024, "step": 9374 }, { "epoch": 12.27, "learning_rate": 3.1800000000000005e-06, "loss": 0.0026, "step": 9375 }, { "epoch": 12.27, "learning_rate": 3.175e-06, "loss": 0.0028, "step": 9376 }, { "epoch": 12.27, "learning_rate": 3.17e-06, "loss": 0.0024, "step": 9377 }, { "epoch": 12.27, "learning_rate": 3.1649999999999998e-06, "loss": 0.0017, "step": 9378 }, { "epoch": 12.28, "learning_rate": 3.1600000000000007e-06, "loss": 0.0014, "step": 9379 }, { "epoch": 12.28, "learning_rate": 3.1550000000000003e-06, "loss": 0.0013, "step": 9380 }, { "epoch": 12.28, "learning_rate": 3.1500000000000003e-06, "loss": 0.0023, "step": 9381 }, { "epoch": 12.28, "learning_rate": 3.145e-06, "loss": 0.0023, "step": 9382 }, { "epoch": 12.28, "learning_rate": 3.14e-06, "loss": 0.0012, "step": 9383 }, { "epoch": 12.28, "learning_rate": 3.1350000000000005e-06, "loss": 0.0015, "step": 9384 }, { "epoch": 12.28, "learning_rate": 3.13e-06, "loss": 0.0021, "step": 9385 }, { "epoch": 12.29, "learning_rate": 3.125e-06, "loss": 0.0019, "step": 9386 }, { "epoch": 12.29, "learning_rate": 3.12e-06, "loss": 0.0026, "step": 9387 }, { "epoch": 12.29, "learning_rate": 3.1150000000000002e-06, "loss": 0.0022, "step": 9388 }, { "epoch": 12.29, "learning_rate": 3.11e-06, "loss": 0.0033, "step": 9389 }, { "epoch": 12.29, "learning_rate": 3.1050000000000003e-06, "loss": 0.0027, "step": 9390 }, { "epoch": 12.29, "learning_rate": 3.1e-06, "loss": 0.0015, "step": 9391 }, { "epoch": 12.29, "learning_rate": 3.095e-06, "loss": 0.0012, "step": 9392 }, { "epoch": 12.29, "learning_rate": 3.09e-06, "loss": 0.0035, "step": 9393 }, { "epoch": 12.3, "learning_rate": 3.085e-06, "loss": 0.0015, "step": 9394 }, { "epoch": 12.3, "learning_rate": 3.08e-06, "loss": 0.0018, "step": 9395 }, { "epoch": 12.3, "learning_rate": 3.075e-06, "loss": 0.0017, "step": 9396 }, { "epoch": 12.3, "learning_rate": 3.0700000000000003e-06, "loss": 0.0015, "step": 9397 }, { "epoch": 12.3, "learning_rate": 3.0650000000000003e-06, "loss": 0.0016, "step": 9398 }, { "epoch": 12.3, "learning_rate": 3.06e-06, "loss": 0.0023, "step": 9399 }, { "epoch": 12.3, "learning_rate": 3.0550000000000004e-06, "loss": 0.0024, "step": 9400 }, { "epoch": 12.3, "learning_rate": 3.05e-06, "loss": 0.002, "step": 9401 }, { "epoch": 12.31, "learning_rate": 3.0450000000000005e-06, "loss": 0.002, "step": 9402 }, { "epoch": 12.31, "learning_rate": 3.04e-06, "loss": 0.0026, "step": 9403 }, { "epoch": 12.31, "learning_rate": 3.035e-06, "loss": 0.002, "step": 9404 }, { "epoch": 12.31, "learning_rate": 3.0300000000000002e-06, "loss": 0.006, "step": 9405 }, { "epoch": 12.31, "learning_rate": 3.0250000000000003e-06, "loss": 0.0021, "step": 9406 }, { "epoch": 12.31, "learning_rate": 3.0200000000000003e-06, "loss": 0.0029, "step": 9407 }, { "epoch": 12.31, "learning_rate": 3.015e-06, "loss": 0.0018, "step": 9408 }, { "epoch": 12.32, "learning_rate": 3.01e-06, "loss": 0.0024, "step": 9409 }, { "epoch": 12.32, "learning_rate": 3.005e-06, "loss": 0.0023, "step": 9410 }, { "epoch": 12.32, "learning_rate": 3e-06, "loss": 0.0022, "step": 9411 }, { "epoch": 12.32, "learning_rate": 2.995e-06, "loss": 0.0025, "step": 9412 }, { "epoch": 12.32, "learning_rate": 2.99e-06, "loss": 0.0042, "step": 9413 }, { "epoch": 12.32, "learning_rate": 2.9850000000000002e-06, "loss": 0.0028, "step": 9414 }, { "epoch": 12.32, "learning_rate": 2.9800000000000003e-06, "loss": 0.0015, "step": 9415 }, { "epoch": 12.32, "learning_rate": 2.975e-06, "loss": 0.0026, "step": 9416 }, { "epoch": 12.33, "learning_rate": 2.9700000000000004e-06, "loss": 0.0016, "step": 9417 }, { "epoch": 12.33, "learning_rate": 2.965e-06, "loss": 0.0025, "step": 9418 }, { "epoch": 12.33, "learning_rate": 2.9600000000000005e-06, "loss": 0.0023, "step": 9419 }, { "epoch": 12.33, "learning_rate": 2.955e-06, "loss": 0.0014, "step": 9420 }, { "epoch": 12.33, "learning_rate": 2.95e-06, "loss": 0.0015, "step": 9421 }, { "epoch": 12.33, "learning_rate": 2.945e-06, "loss": 0.0019, "step": 9422 }, { "epoch": 12.33, "learning_rate": 2.9400000000000002e-06, "loss": 0.0019, "step": 9423 }, { "epoch": 12.34, "learning_rate": 2.9350000000000003e-06, "loss": 0.0022, "step": 9424 }, { "epoch": 12.34, "learning_rate": 2.93e-06, "loss": 0.0021, "step": 9425 }, { "epoch": 12.34, "learning_rate": 2.9250000000000004e-06, "loss": 0.0012, "step": 9426 }, { "epoch": 12.34, "learning_rate": 2.92e-06, "loss": 0.0026, "step": 9427 }, { "epoch": 12.34, "learning_rate": 2.915e-06, "loss": 0.0019, "step": 9428 }, { "epoch": 12.34, "learning_rate": 2.91e-06, "loss": 0.0023, "step": 9429 }, { "epoch": 12.34, "learning_rate": 2.905e-06, "loss": 0.0027, "step": 9430 }, { "epoch": 12.34, "learning_rate": 2.9e-06, "loss": 0.0017, "step": 9431 }, { "epoch": 12.35, "learning_rate": 2.8950000000000002e-06, "loss": 0.002, "step": 9432 }, { "epoch": 12.35, "learning_rate": 2.89e-06, "loss": 0.0023, "step": 9433 }, { "epoch": 12.35, "learning_rate": 2.8850000000000003e-06, "loss": 0.002, "step": 9434 }, { "epoch": 12.35, "learning_rate": 2.88e-06, "loss": 0.0029, "step": 9435 }, { "epoch": 12.35, "learning_rate": 2.8750000000000004e-06, "loss": 0.0017, "step": 9436 }, { "epoch": 12.35, "learning_rate": 2.87e-06, "loss": 0.0021, "step": 9437 }, { "epoch": 12.35, "learning_rate": 2.865e-06, "loss": 0.0026, "step": 9438 }, { "epoch": 12.35, "learning_rate": 2.86e-06, "loss": 0.0023, "step": 9439 }, { "epoch": 12.36, "learning_rate": 2.855e-06, "loss": 0.0018, "step": 9440 }, { "epoch": 12.36, "learning_rate": 2.8500000000000002e-06, "loss": 0.0019, "step": 9441 }, { "epoch": 12.36, "learning_rate": 2.8450000000000003e-06, "loss": 0.0016, "step": 9442 }, { "epoch": 12.36, "learning_rate": 2.8400000000000003e-06, "loss": 0.0017, "step": 9443 }, { "epoch": 12.36, "learning_rate": 2.835e-06, "loss": 0.0028, "step": 9444 }, { "epoch": 12.36, "learning_rate": 2.83e-06, "loss": 0.0023, "step": 9445 }, { "epoch": 12.36, "learning_rate": 2.825e-06, "loss": 0.0028, "step": 9446 }, { "epoch": 12.37, "learning_rate": 2.82e-06, "loss": 0.0024, "step": 9447 }, { "epoch": 12.37, "learning_rate": 2.815e-06, "loss": 0.0024, "step": 9448 }, { "epoch": 12.37, "learning_rate": 2.81e-06, "loss": 0.0022, "step": 9449 }, { "epoch": 12.37, "learning_rate": 2.805e-06, "loss": 0.0024, "step": 9450 }, { "epoch": 12.37, "learning_rate": 2.8000000000000003e-06, "loss": 0.0029, "step": 9451 }, { "epoch": 12.37, "learning_rate": 2.795e-06, "loss": 0.0022, "step": 9452 }, { "epoch": 12.37, "learning_rate": 2.7900000000000004e-06, "loss": 0.0027, "step": 9453 }, { "epoch": 12.37, "learning_rate": 2.785e-06, "loss": 0.0028, "step": 9454 }, { "epoch": 12.38, "learning_rate": 2.78e-06, "loss": 0.0019, "step": 9455 }, { "epoch": 12.38, "learning_rate": 2.775e-06, "loss": 0.0024, "step": 9456 }, { "epoch": 12.38, "learning_rate": 2.77e-06, "loss": 0.0022, "step": 9457 }, { "epoch": 12.38, "learning_rate": 2.765e-06, "loss": 0.0014, "step": 9458 }, { "epoch": 12.38, "learning_rate": 2.7600000000000003e-06, "loss": 0.0014, "step": 9459 }, { "epoch": 12.38, "learning_rate": 2.7550000000000003e-06, "loss": 0.0019, "step": 9460 }, { "epoch": 12.38, "learning_rate": 2.7500000000000004e-06, "loss": 0.0022, "step": 9461 }, { "epoch": 12.38, "learning_rate": 2.745e-06, "loss": 0.0019, "step": 9462 }, { "epoch": 12.39, "learning_rate": 2.74e-06, "loss": 0.0024, "step": 9463 }, { "epoch": 12.39, "learning_rate": 2.735e-06, "loss": 0.002, "step": 9464 }, { "epoch": 12.39, "learning_rate": 2.73e-06, "loss": 0.0026, "step": 9465 }, { "epoch": 12.39, "learning_rate": 2.725e-06, "loss": 0.0028, "step": 9466 }, { "epoch": 12.39, "learning_rate": 2.72e-06, "loss": 0.002, "step": 9467 }, { "epoch": 12.39, "learning_rate": 2.7150000000000003e-06, "loss": 0.0018, "step": 9468 }, { "epoch": 12.39, "learning_rate": 2.71e-06, "loss": 0.0041, "step": 9469 }, { "epoch": 12.4, "learning_rate": 2.7050000000000004e-06, "loss": 0.0022, "step": 9470 }, { "epoch": 12.4, "learning_rate": 2.7e-06, "loss": 0.0009, "step": 9471 }, { "epoch": 12.4, "learning_rate": 2.6950000000000005e-06, "loss": 0.002, "step": 9472 }, { "epoch": 12.4, "learning_rate": 2.69e-06, "loss": 0.0022, "step": 9473 }, { "epoch": 12.4, "learning_rate": 2.685e-06, "loss": 0.0023, "step": 9474 }, { "epoch": 12.4, "learning_rate": 2.68e-06, "loss": 0.0025, "step": 9475 }, { "epoch": 12.4, "learning_rate": 2.6750000000000002e-06, "loss": 0.0018, "step": 9476 }, { "epoch": 12.4, "learning_rate": 2.6700000000000003e-06, "loss": 0.0016, "step": 9477 }, { "epoch": 12.41, "learning_rate": 2.6650000000000003e-06, "loss": 0.0025, "step": 9478 }, { "epoch": 12.41, "learning_rate": 2.66e-06, "loss": 0.0016, "step": 9479 }, { "epoch": 12.41, "learning_rate": 2.655e-06, "loss": 0.0018, "step": 9480 }, { "epoch": 12.41, "learning_rate": 2.65e-06, "loss": 0.0031, "step": 9481 }, { "epoch": 12.41, "learning_rate": 2.645e-06, "loss": 0.0037, "step": 9482 }, { "epoch": 12.41, "learning_rate": 2.64e-06, "loss": 0.0046, "step": 9483 }, { "epoch": 12.41, "learning_rate": 2.6349999999999998e-06, "loss": 0.0018, "step": 9484 }, { "epoch": 12.41, "learning_rate": 2.6300000000000002e-06, "loss": 0.0018, "step": 9485 }, { "epoch": 12.42, "learning_rate": 2.625e-06, "loss": 0.002, "step": 9486 }, { "epoch": 12.42, "learning_rate": 2.6200000000000003e-06, "loss": 0.0024, "step": 9487 }, { "epoch": 12.42, "learning_rate": 2.615e-06, "loss": 0.0012, "step": 9488 }, { "epoch": 12.42, "learning_rate": 2.6100000000000004e-06, "loss": 0.0025, "step": 9489 }, { "epoch": 12.42, "learning_rate": 2.605e-06, "loss": 0.0024, "step": 9490 }, { "epoch": 12.42, "learning_rate": 2.6e-06, "loss": 0.0018, "step": 9491 }, { "epoch": 12.42, "learning_rate": 2.595e-06, "loss": 0.002, "step": 9492 }, { "epoch": 12.43, "learning_rate": 2.59e-06, "loss": 0.0031, "step": 9493 }, { "epoch": 12.43, "learning_rate": 2.5850000000000002e-06, "loss": 0.0012, "step": 9494 }, { "epoch": 12.43, "learning_rate": 2.5800000000000003e-06, "loss": 0.003, "step": 9495 }, { "epoch": 12.43, "learning_rate": 2.575e-06, "loss": 0.0014, "step": 9496 }, { "epoch": 12.43, "learning_rate": 2.5700000000000004e-06, "loss": 0.0018, "step": 9497 }, { "epoch": 12.43, "learning_rate": 2.565e-06, "loss": 0.0019, "step": 9498 }, { "epoch": 12.43, "learning_rate": 2.56e-06, "loss": 0.0018, "step": 9499 }, { "epoch": 12.43, "learning_rate": 2.555e-06, "loss": 0.0012, "step": 9500 }, { "epoch": 12.44, "learning_rate": 2.55e-06, "loss": 0.0015, "step": 9501 }, { "epoch": 12.44, "learning_rate": 2.545e-06, "loss": 0.003, "step": 9502 }, { "epoch": 12.44, "learning_rate": 2.54e-06, "loss": 0.0015, "step": 9503 }, { "epoch": 12.44, "learning_rate": 2.5350000000000003e-06, "loss": 0.0029, "step": 9504 }, { "epoch": 12.44, "learning_rate": 2.53e-06, "loss": 0.0029, "step": 9505 }, { "epoch": 12.44, "learning_rate": 2.5250000000000004e-06, "loss": 0.0023, "step": 9506 }, { "epoch": 12.44, "learning_rate": 2.52e-06, "loss": 0.0024, "step": 9507 }, { "epoch": 12.45, "learning_rate": 2.515e-06, "loss": 0.0009, "step": 9508 }, { "epoch": 12.45, "learning_rate": 2.51e-06, "loss": 0.0021, "step": 9509 }, { "epoch": 12.45, "learning_rate": 2.505e-06, "loss": 0.0017, "step": 9510 }, { "epoch": 12.45, "learning_rate": 2.5e-06, "loss": 0.0024, "step": 9511 }, { "epoch": 12.45, "learning_rate": 2.4950000000000003e-06, "loss": 0.0031, "step": 9512 }, { "epoch": 12.45, "learning_rate": 2.49e-06, "loss": 0.0037, "step": 9513 }, { "epoch": 12.45, "learning_rate": 2.4850000000000003e-06, "loss": 0.0014, "step": 9514 }, { "epoch": 12.45, "learning_rate": 2.48e-06, "loss": 0.0015, "step": 9515 }, { "epoch": 12.46, "learning_rate": 2.4750000000000004e-06, "loss": 0.0027, "step": 9516 }, { "epoch": 12.46, "learning_rate": 2.47e-06, "loss": 0.0029, "step": 9517 }, { "epoch": 12.46, "learning_rate": 2.465e-06, "loss": 0.001, "step": 9518 }, { "epoch": 12.46, "learning_rate": 2.46e-06, "loss": 0.0028, "step": 9519 }, { "epoch": 12.46, "learning_rate": 2.4550000000000002e-06, "loss": 0.0027, "step": 9520 }, { "epoch": 12.46, "learning_rate": 2.4500000000000003e-06, "loss": 0.0018, "step": 9521 }, { "epoch": 12.46, "learning_rate": 2.445e-06, "loss": 0.0015, "step": 9522 }, { "epoch": 12.46, "learning_rate": 2.4400000000000004e-06, "loss": 0.0034, "step": 9523 }, { "epoch": 12.47, "learning_rate": 2.435e-06, "loss": 0.0019, "step": 9524 }, { "epoch": 12.47, "learning_rate": 2.43e-06, "loss": 0.003, "step": 9525 }, { "epoch": 12.47, "learning_rate": 2.425e-06, "loss": 0.0023, "step": 9526 }, { "epoch": 12.47, "learning_rate": 2.42e-06, "loss": 0.0017, "step": 9527 }, { "epoch": 12.47, "learning_rate": 2.415e-06, "loss": 0.0036, "step": 9528 }, { "epoch": 12.47, "learning_rate": 2.4100000000000002e-06, "loss": 0.0029, "step": 9529 }, { "epoch": 12.47, "learning_rate": 2.405e-06, "loss": 0.0022, "step": 9530 }, { "epoch": 12.48, "learning_rate": 2.4000000000000003e-06, "loss": 0.0016, "step": 9531 }, { "epoch": 12.48, "learning_rate": 2.395e-06, "loss": 0.0023, "step": 9532 }, { "epoch": 12.48, "learning_rate": 2.3900000000000004e-06, "loss": 0.0019, "step": 9533 }, { "epoch": 12.48, "learning_rate": 2.385e-06, "loss": 0.0018, "step": 9534 }, { "epoch": 12.48, "learning_rate": 2.38e-06, "loss": 0.001, "step": 9535 }, { "epoch": 12.48, "learning_rate": 2.375e-06, "loss": 0.0024, "step": 9536 }, { "epoch": 12.48, "learning_rate": 2.37e-06, "loss": 0.0019, "step": 9537 }, { "epoch": 12.48, "learning_rate": 2.3650000000000002e-06, "loss": 0.0017, "step": 9538 }, { "epoch": 12.49, "learning_rate": 2.36e-06, "loss": 0.0023, "step": 9539 }, { "epoch": 12.49, "learning_rate": 2.3550000000000003e-06, "loss": 0.0026, "step": 9540 }, { "epoch": 12.49, "learning_rate": 2.35e-06, "loss": 0.0025, "step": 9541 }, { "epoch": 12.49, "learning_rate": 2.345e-06, "loss": 0.0029, "step": 9542 }, { "epoch": 12.49, "learning_rate": 2.34e-06, "loss": 0.0036, "step": 9543 }, { "epoch": 12.49, "learning_rate": 2.335e-06, "loss": 0.0013, "step": 9544 }, { "epoch": 12.49, "learning_rate": 2.33e-06, "loss": 0.0021, "step": 9545 }, { "epoch": 12.49, "learning_rate": 2.325e-06, "loss": 0.0028, "step": 9546 }, { "epoch": 12.5, "learning_rate": 2.32e-06, "loss": 0.0019, "step": 9547 }, { "epoch": 12.5, "learning_rate": 2.3150000000000003e-06, "loss": 0.0013, "step": 9548 }, { "epoch": 12.5, "learning_rate": 2.31e-06, "loss": 0.0024, "step": 9549 }, { "epoch": 12.5, "learning_rate": 2.3050000000000004e-06, "loss": 0.0013, "step": 9550 }, { "epoch": 12.5, "learning_rate": 2.3e-06, "loss": 0.0017, "step": 9551 }, { "epoch": 12.5, "learning_rate": 2.2950000000000005e-06, "loss": 0.0025, "step": 9552 }, { "epoch": 12.5, "learning_rate": 2.29e-06, "loss": 0.0016, "step": 9553 }, { "epoch": 12.51, "learning_rate": 2.285e-06, "loss": 0.0028, "step": 9554 }, { "epoch": 12.51, "learning_rate": 2.28e-06, "loss": 0.0025, "step": 9555 }, { "epoch": 12.51, "learning_rate": 2.2750000000000002e-06, "loss": 0.0024, "step": 9556 }, { "epoch": 12.51, "learning_rate": 2.2700000000000003e-06, "loss": 0.0013, "step": 9557 }, { "epoch": 12.51, "learning_rate": 2.265e-06, "loss": 0.0023, "step": 9558 }, { "epoch": 12.51, "learning_rate": 2.26e-06, "loss": 0.0021, "step": 9559 }, { "epoch": 12.51, "learning_rate": 2.255e-06, "loss": 0.0023, "step": 9560 }, { "epoch": 12.51, "learning_rate": 2.25e-06, "loss": 0.0024, "step": 9561 }, { "epoch": 12.52, "learning_rate": 2.245e-06, "loss": 0.0023, "step": 9562 }, { "epoch": 12.52, "learning_rate": 2.24e-06, "loss": 0.0021, "step": 9563 }, { "epoch": 12.52, "learning_rate": 2.2349999999999998e-06, "loss": 0.004, "step": 9564 }, { "epoch": 12.52, "learning_rate": 2.2300000000000002e-06, "loss": 0.0021, "step": 9565 }, { "epoch": 12.52, "learning_rate": 2.225e-06, "loss": 0.0028, "step": 9566 }, { "epoch": 12.52, "learning_rate": 2.2200000000000003e-06, "loss": 0.002, "step": 9567 }, { "epoch": 12.52, "learning_rate": 2.215e-06, "loss": 0.0018, "step": 9568 }, { "epoch": 12.52, "learning_rate": 2.2100000000000004e-06, "loss": 0.0027, "step": 9569 }, { "epoch": 12.53, "learning_rate": 2.205e-06, "loss": 0.001, "step": 9570 }, { "epoch": 12.53, "learning_rate": 2.2e-06, "loss": 0.0014, "step": 9571 }, { "epoch": 12.53, "learning_rate": 2.195e-06, "loss": 0.0018, "step": 9572 }, { "epoch": 12.53, "learning_rate": 2.19e-06, "loss": 0.0028, "step": 9573 }, { "epoch": 12.53, "learning_rate": 2.1850000000000003e-06, "loss": 0.0012, "step": 9574 }, { "epoch": 12.53, "learning_rate": 2.1800000000000003e-06, "loss": 0.0027, "step": 9575 }, { "epoch": 12.53, "learning_rate": 2.175e-06, "loss": 0.0026, "step": 9576 }, { "epoch": 12.54, "learning_rate": 2.17e-06, "loss": 0.0025, "step": 9577 }, { "epoch": 12.54, "learning_rate": 2.165e-06, "loss": 0.0018, "step": 9578 }, { "epoch": 12.54, "learning_rate": 2.16e-06, "loss": 0.0039, "step": 9579 }, { "epoch": 12.54, "learning_rate": 2.155e-06, "loss": 0.0016, "step": 9580 }, { "epoch": 12.54, "learning_rate": 2.1499999999999997e-06, "loss": 0.0022, "step": 9581 }, { "epoch": 12.54, "learning_rate": 2.1450000000000002e-06, "loss": 0.0024, "step": 9582 }, { "epoch": 12.54, "learning_rate": 2.14e-06, "loss": 0.0019, "step": 9583 }, { "epoch": 12.54, "learning_rate": 2.1350000000000003e-06, "loss": 0.002, "step": 9584 }, { "epoch": 12.55, "learning_rate": 2.13e-06, "loss": 0.0022, "step": 9585 }, { "epoch": 12.55, "learning_rate": 2.1250000000000004e-06, "loss": 0.0008, "step": 9586 }, { "epoch": 12.55, "learning_rate": 2.12e-06, "loss": 0.0024, "step": 9587 }, { "epoch": 12.55, "learning_rate": 2.115e-06, "loss": 0.0017, "step": 9588 }, { "epoch": 12.55, "learning_rate": 2.11e-06, "loss": 0.0014, "step": 9589 }, { "epoch": 12.55, "learning_rate": 2.105e-06, "loss": 0.0018, "step": 9590 }, { "epoch": 12.55, "learning_rate": 2.1000000000000002e-06, "loss": 0.0027, "step": 9591 }, { "epoch": 12.55, "learning_rate": 2.0950000000000003e-06, "loss": 0.0016, "step": 9592 }, { "epoch": 12.56, "learning_rate": 2.09e-06, "loss": 0.0034, "step": 9593 }, { "epoch": 12.56, "learning_rate": 2.085e-06, "loss": 0.0033, "step": 9594 }, { "epoch": 12.56, "learning_rate": 2.08e-06, "loss": 0.0025, "step": 9595 }, { "epoch": 12.56, "learning_rate": 2.075e-06, "loss": 0.0014, "step": 9596 }, { "epoch": 12.56, "learning_rate": 2.07e-06, "loss": 0.0034, "step": 9597 }, { "epoch": 12.56, "learning_rate": 2.065e-06, "loss": 0.002, "step": 9598 }, { "epoch": 12.56, "learning_rate": 2.06e-06, "loss": 0.0027, "step": 9599 }, { "epoch": 12.57, "learning_rate": 2.055e-06, "loss": 0.0015, "step": 9600 }, { "epoch": 12.57, "learning_rate": 2.0500000000000003e-06, "loss": 0.0016, "step": 9601 }, { "epoch": 12.57, "learning_rate": 2.045e-06, "loss": 0.0025, "step": 9602 }, { "epoch": 12.57, "learning_rate": 2.0400000000000004e-06, "loss": 0.0019, "step": 9603 }, { "epoch": 12.57, "learning_rate": 2.035e-06, "loss": 0.0015, "step": 9604 }, { "epoch": 12.57, "learning_rate": 2.03e-06, "loss": 0.0026, "step": 9605 }, { "epoch": 12.57, "learning_rate": 2.025e-06, "loss": 0.0017, "step": 9606 }, { "epoch": 12.57, "learning_rate": 2.02e-06, "loss": 0.0037, "step": 9607 }, { "epoch": 12.58, "learning_rate": 2.015e-06, "loss": 0.0017, "step": 9608 }, { "epoch": 12.58, "learning_rate": 2.0100000000000002e-06, "loss": 0.001, "step": 9609 }, { "epoch": 12.58, "learning_rate": 2.005e-06, "loss": 0.0021, "step": 9610 }, { "epoch": 12.58, "learning_rate": 2.0000000000000003e-06, "loss": 0.0023, "step": 9611 }, { "epoch": 12.58, "learning_rate": 1.995e-06, "loss": 0.0012, "step": 9612 }, { "epoch": 12.58, "learning_rate": 1.99e-06, "loss": 0.0025, "step": 9613 }, { "epoch": 12.58, "learning_rate": 1.985e-06, "loss": 0.002, "step": 9614 }, { "epoch": 12.59, "learning_rate": 1.98e-06, "loss": 0.0014, "step": 9615 }, { "epoch": 12.59, "learning_rate": 1.975e-06, "loss": 0.0015, "step": 9616 }, { "epoch": 12.59, "learning_rate": 1.9699999999999998e-06, "loss": 0.0034, "step": 9617 }, { "epoch": 12.59, "learning_rate": 1.9650000000000002e-06, "loss": 0.0013, "step": 9618 }, { "epoch": 12.59, "learning_rate": 1.96e-06, "loss": 0.002, "step": 9619 }, { "epoch": 12.59, "learning_rate": 1.9550000000000003e-06, "loss": 0.0026, "step": 9620 }, { "epoch": 12.59, "learning_rate": 1.95e-06, "loss": 0.0044, "step": 9621 }, { "epoch": 12.59, "learning_rate": 1.945e-06, "loss": 0.0036, "step": 9622 }, { "epoch": 12.6, "learning_rate": 1.94e-06, "loss": 0.0029, "step": 9623 }, { "epoch": 12.6, "learning_rate": 1.935e-06, "loss": 0.0022, "step": 9624 }, { "epoch": 12.6, "learning_rate": 1.93e-06, "loss": 0.0026, "step": 9625 }, { "epoch": 12.6, "learning_rate": 1.925e-06, "loss": 0.0017, "step": 9626 }, { "epoch": 12.6, "learning_rate": 1.92e-06, "loss": 0.0025, "step": 9627 }, { "epoch": 12.6, "learning_rate": 1.9150000000000003e-06, "loss": 0.0018, "step": 9628 }, { "epoch": 12.6, "learning_rate": 1.91e-06, "loss": 0.0034, "step": 9629 }, { "epoch": 12.6, "learning_rate": 1.9050000000000002e-06, "loss": 0.0035, "step": 9630 }, { "epoch": 12.61, "learning_rate": 1.9e-06, "loss": 0.0017, "step": 9631 }, { "epoch": 12.61, "learning_rate": 1.8950000000000003e-06, "loss": 0.0022, "step": 9632 }, { "epoch": 12.61, "learning_rate": 1.8900000000000001e-06, "loss": 0.0025, "step": 9633 }, { "epoch": 12.61, "learning_rate": 1.885e-06, "loss": 0.0019, "step": 9634 }, { "epoch": 12.61, "learning_rate": 1.8800000000000002e-06, "loss": 0.0017, "step": 9635 }, { "epoch": 12.61, "learning_rate": 1.875e-06, "loss": 0.002, "step": 9636 }, { "epoch": 12.61, "learning_rate": 1.8700000000000003e-06, "loss": 0.0024, "step": 9637 }, { "epoch": 12.62, "learning_rate": 1.8650000000000001e-06, "loss": 0.0026, "step": 9638 }, { "epoch": 12.62, "learning_rate": 1.86e-06, "loss": 0.0019, "step": 9639 }, { "epoch": 12.62, "learning_rate": 1.8550000000000002e-06, "loss": 0.0021, "step": 9640 }, { "epoch": 12.62, "learning_rate": 1.85e-06, "loss": 0.0016, "step": 9641 }, { "epoch": 12.62, "learning_rate": 1.8450000000000001e-06, "loss": 0.0021, "step": 9642 }, { "epoch": 12.62, "learning_rate": 1.84e-06, "loss": 0.0023, "step": 9643 }, { "epoch": 12.62, "learning_rate": 1.8350000000000002e-06, "loss": 0.003, "step": 9644 }, { "epoch": 12.62, "learning_rate": 1.83e-06, "loss": 0.0018, "step": 9645 }, { "epoch": 12.63, "learning_rate": 1.8249999999999999e-06, "loss": 0.0019, "step": 9646 }, { "epoch": 12.63, "learning_rate": 1.8200000000000002e-06, "loss": 0.0025, "step": 9647 }, { "epoch": 12.63, "learning_rate": 1.815e-06, "loss": 0.002, "step": 9648 }, { "epoch": 12.63, "learning_rate": 1.8100000000000002e-06, "loss": 0.0024, "step": 9649 }, { "epoch": 12.63, "learning_rate": 1.805e-06, "loss": 0.0021, "step": 9650 }, { "epoch": 12.63, "learning_rate": 1.8e-06, "loss": 0.0019, "step": 9651 }, { "epoch": 12.63, "learning_rate": 1.7950000000000002e-06, "loss": 0.0021, "step": 9652 }, { "epoch": 12.63, "learning_rate": 1.79e-06, "loss": 0.0016, "step": 9653 }, { "epoch": 12.64, "learning_rate": 1.7850000000000003e-06, "loss": 0.0029, "step": 9654 }, { "epoch": 12.64, "learning_rate": 1.7800000000000001e-06, "loss": 0.0048, "step": 9655 }, { "epoch": 12.64, "learning_rate": 1.775e-06, "loss": 0.002, "step": 9656 }, { "epoch": 12.64, "learning_rate": 1.7700000000000002e-06, "loss": 0.0025, "step": 9657 }, { "epoch": 12.64, "learning_rate": 1.765e-06, "loss": 0.0025, "step": 9658 }, { "epoch": 12.64, "learning_rate": 1.76e-06, "loss": 0.0025, "step": 9659 }, { "epoch": 12.64, "learning_rate": 1.7550000000000001e-06, "loss": 0.0028, "step": 9660 }, { "epoch": 12.65, "learning_rate": 1.7500000000000002e-06, "loss": 0.0021, "step": 9661 }, { "epoch": 12.65, "learning_rate": 1.745e-06, "loss": 0.0018, "step": 9662 }, { "epoch": 12.65, "learning_rate": 1.7399999999999999e-06, "loss": 0.0022, "step": 9663 }, { "epoch": 12.65, "learning_rate": 1.7350000000000001e-06, "loss": 0.0026, "step": 9664 }, { "epoch": 12.65, "learning_rate": 1.73e-06, "loss": 0.0012, "step": 9665 }, { "epoch": 12.65, "learning_rate": 1.7250000000000002e-06, "loss": 0.0014, "step": 9666 }, { "epoch": 12.65, "learning_rate": 1.72e-06, "loss": 0.0027, "step": 9667 }, { "epoch": 12.65, "learning_rate": 1.7149999999999999e-06, "loss": 0.0013, "step": 9668 }, { "epoch": 12.66, "learning_rate": 1.7100000000000001e-06, "loss": 0.0031, "step": 9669 }, { "epoch": 12.66, "learning_rate": 1.705e-06, "loss": 0.0014, "step": 9670 }, { "epoch": 12.66, "learning_rate": 1.7000000000000002e-06, "loss": 0.0027, "step": 9671 }, { "epoch": 12.66, "learning_rate": 1.695e-06, "loss": 0.003, "step": 9672 }, { "epoch": 12.66, "learning_rate": 1.69e-06, "loss": 0.0018, "step": 9673 }, { "epoch": 12.66, "learning_rate": 1.6850000000000002e-06, "loss": 0.0018, "step": 9674 }, { "epoch": 12.66, "learning_rate": 1.68e-06, "loss": 0.0018, "step": 9675 }, { "epoch": 12.66, "learning_rate": 1.6750000000000003e-06, "loss": 0.0013, "step": 9676 }, { "epoch": 12.67, "learning_rate": 1.67e-06, "loss": 0.001, "step": 9677 }, { "epoch": 12.67, "learning_rate": 1.6650000000000002e-06, "loss": 0.0028, "step": 9678 }, { "epoch": 12.67, "learning_rate": 1.6600000000000002e-06, "loss": 0.0035, "step": 9679 }, { "epoch": 12.67, "learning_rate": 1.6600000000000002e-06, "loss": 0.0031, "step": 9680 }, { "epoch": 12.67, "learning_rate": 1.6600000000000002e-06, "loss": 0.0021, "step": 9681 }, { "epoch": 12.67, "learning_rate": 1.655e-06, "loss": 0.0021, "step": 9682 }, { "epoch": 12.67, "learning_rate": 1.65e-06, "loss": 0.002, "step": 9683 }, { "epoch": 12.68, "learning_rate": 1.645e-06, "loss": 0.0016, "step": 9684 }, { "epoch": 12.68, "learning_rate": 1.6400000000000002e-06, "loss": 0.0018, "step": 9685 }, { "epoch": 12.68, "learning_rate": 1.635e-06, "loss": 0.0021, "step": 9686 }, { "epoch": 12.68, "learning_rate": 1.6299999999999999e-06, "loss": 0.0033, "step": 9687 }, { "epoch": 12.68, "learning_rate": 1.6250000000000001e-06, "loss": 0.0015, "step": 9688 }, { "epoch": 12.68, "learning_rate": 1.62e-06, "loss": 0.002, "step": 9689 }, { "epoch": 12.68, "learning_rate": 1.6150000000000002e-06, "loss": 0.0022, "step": 9690 }, { "epoch": 12.68, "learning_rate": 1.61e-06, "loss": 0.0014, "step": 9691 }, { "epoch": 12.69, "learning_rate": 1.6049999999999999e-06, "loss": 0.0016, "step": 9692 }, { "epoch": 12.69, "learning_rate": 1.6000000000000001e-06, "loss": 0.0021, "step": 9693 }, { "epoch": 12.69, "learning_rate": 1.595e-06, "loss": 0.002, "step": 9694 }, { "epoch": 12.69, "learning_rate": 1.5900000000000002e-06, "loss": 0.0015, "step": 9695 }, { "epoch": 12.69, "learning_rate": 1.585e-06, "loss": 0.0045, "step": 9696 }, { "epoch": 12.69, "learning_rate": 1.5800000000000003e-06, "loss": 0.0016, "step": 9697 }, { "epoch": 12.69, "learning_rate": 1.5750000000000002e-06, "loss": 0.0019, "step": 9698 }, { "epoch": 12.7, "learning_rate": 1.57e-06, "loss": 0.0022, "step": 9699 }, { "epoch": 12.7, "learning_rate": 1.565e-06, "loss": 0.002, "step": 9700 }, { "epoch": 12.7, "learning_rate": 1.56e-06, "loss": 0.0021, "step": 9701 }, { "epoch": 12.7, "learning_rate": 1.555e-06, "loss": 0.0021, "step": 9702 }, { "epoch": 12.7, "learning_rate": 1.55e-06, "loss": 0.0027, "step": 9703 }, { "epoch": 12.7, "learning_rate": 1.545e-06, "loss": 0.0014, "step": 9704 }, { "epoch": 12.7, "learning_rate": 1.54e-06, "loss": 0.0015, "step": 9705 }, { "epoch": 12.7, "learning_rate": 1.5350000000000001e-06, "loss": 0.0011, "step": 9706 }, { "epoch": 12.71, "learning_rate": 1.53e-06, "loss": 0.0016, "step": 9707 }, { "epoch": 12.71, "learning_rate": 1.525e-06, "loss": 0.0022, "step": 9708 }, { "epoch": 12.71, "learning_rate": 1.52e-06, "loss": 0.0026, "step": 9709 }, { "epoch": 12.71, "learning_rate": 1.5150000000000001e-06, "loss": 0.0019, "step": 9710 }, { "epoch": 12.71, "learning_rate": 1.5100000000000002e-06, "loss": 0.0028, "step": 9711 }, { "epoch": 12.71, "learning_rate": 1.505e-06, "loss": 0.0014, "step": 9712 }, { "epoch": 12.71, "learning_rate": 1.5e-06, "loss": 0.0015, "step": 9713 }, { "epoch": 12.71, "learning_rate": 1.495e-06, "loss": 0.0024, "step": 9714 }, { "epoch": 12.72, "learning_rate": 1.4900000000000001e-06, "loss": 0.0024, "step": 9715 }, { "epoch": 12.72, "learning_rate": 1.4850000000000002e-06, "loss": 0.0029, "step": 9716 }, { "epoch": 12.72, "learning_rate": 1.4800000000000002e-06, "loss": 0.0021, "step": 9717 }, { "epoch": 12.72, "learning_rate": 1.475e-06, "loss": 0.0024, "step": 9718 }, { "epoch": 12.72, "learning_rate": 1.4700000000000001e-06, "loss": 0.0019, "step": 9719 }, { "epoch": 12.72, "learning_rate": 1.465e-06, "loss": 0.0017, "step": 9720 }, { "epoch": 12.72, "learning_rate": 1.46e-06, "loss": 0.002, "step": 9721 }, { "epoch": 12.73, "learning_rate": 1.455e-06, "loss": 0.0017, "step": 9722 }, { "epoch": 12.73, "learning_rate": 1.45e-06, "loss": 0.0018, "step": 9723 }, { "epoch": 12.73, "learning_rate": 1.445e-06, "loss": 0.0023, "step": 9724 }, { "epoch": 12.73, "learning_rate": 1.44e-06, "loss": 0.0017, "step": 9725 }, { "epoch": 12.73, "learning_rate": 1.435e-06, "loss": 0.0016, "step": 9726 }, { "epoch": 12.73, "learning_rate": 1.43e-06, "loss": 0.0027, "step": 9727 }, { "epoch": 12.73, "learning_rate": 1.4250000000000001e-06, "loss": 0.0019, "step": 9728 }, { "epoch": 12.73, "learning_rate": 1.4200000000000002e-06, "loss": 0.0021, "step": 9729 }, { "epoch": 12.74, "learning_rate": 1.415e-06, "loss": 0.0018, "step": 9730 }, { "epoch": 12.74, "learning_rate": 1.41e-06, "loss": 0.0031, "step": 9731 }, { "epoch": 12.74, "learning_rate": 1.405e-06, "loss": 0.0022, "step": 9732 }, { "epoch": 12.74, "learning_rate": 1.4000000000000001e-06, "loss": 0.0019, "step": 9733 }, { "epoch": 12.74, "learning_rate": 1.3950000000000002e-06, "loss": 0.0019, "step": 9734 }, { "epoch": 12.74, "learning_rate": 1.39e-06, "loss": 0.0016, "step": 9735 }, { "epoch": 12.74, "learning_rate": 1.385e-06, "loss": 0.0023, "step": 9736 }, { "epoch": 12.74, "learning_rate": 1.3800000000000001e-06, "loss": 0.0022, "step": 9737 }, { "epoch": 12.75, "learning_rate": 1.3750000000000002e-06, "loss": 0.0016, "step": 9738 }, { "epoch": 12.75, "learning_rate": 1.37e-06, "loss": 0.0019, "step": 9739 }, { "epoch": 12.75, "learning_rate": 1.365e-06, "loss": 0.003, "step": 9740 }, { "epoch": 12.75, "learning_rate": 1.36e-06, "loss": 0.0027, "step": 9741 }, { "epoch": 12.75, "learning_rate": 1.355e-06, "loss": 0.0021, "step": 9742 }, { "epoch": 12.75, "learning_rate": 1.35e-06, "loss": 0.0025, "step": 9743 }, { "epoch": 12.75, "learning_rate": 1.345e-06, "loss": 0.0013, "step": 9744 }, { "epoch": 12.76, "learning_rate": 1.34e-06, "loss": 0.0024, "step": 9745 }, { "epoch": 12.76, "learning_rate": 1.3350000000000001e-06, "loss": 0.003, "step": 9746 }, { "epoch": 12.76, "learning_rate": 1.33e-06, "loss": 0.0018, "step": 9747 }, { "epoch": 12.76, "learning_rate": 1.325e-06, "loss": 0.0026, "step": 9748 }, { "epoch": 12.76, "learning_rate": 1.32e-06, "loss": 0.0026, "step": 9749 }, { "epoch": 12.76, "learning_rate": 1.3150000000000001e-06, "loss": 0.0022, "step": 9750 }, { "epoch": 12.76, "learning_rate": 1.3100000000000002e-06, "loss": 0.002, "step": 9751 }, { "epoch": 12.76, "learning_rate": 1.3050000000000002e-06, "loss": 0.0022, "step": 9752 }, { "epoch": 12.77, "learning_rate": 1.3e-06, "loss": 0.0018, "step": 9753 }, { "epoch": 12.77, "learning_rate": 1.295e-06, "loss": 0.0018, "step": 9754 }, { "epoch": 12.77, "learning_rate": 1.2900000000000001e-06, "loss": 0.002, "step": 9755 }, { "epoch": 12.77, "learning_rate": 1.2850000000000002e-06, "loss": 0.002, "step": 9756 }, { "epoch": 12.77, "learning_rate": 1.28e-06, "loss": 0.0025, "step": 9757 }, { "epoch": 12.77, "learning_rate": 1.275e-06, "loss": 0.0019, "step": 9758 }, { "epoch": 12.77, "learning_rate": 1.27e-06, "loss": 0.0016, "step": 9759 }, { "epoch": 12.77, "learning_rate": 1.265e-06, "loss": 0.0015, "step": 9760 }, { "epoch": 12.78, "learning_rate": 1.26e-06, "loss": 0.002, "step": 9761 }, { "epoch": 12.78, "learning_rate": 1.255e-06, "loss": 0.0019, "step": 9762 }, { "epoch": 12.78, "learning_rate": 1.25e-06, "loss": 0.0014, "step": 9763 }, { "epoch": 12.78, "learning_rate": 1.245e-06, "loss": 0.0019, "step": 9764 }, { "epoch": 12.78, "learning_rate": 1.24e-06, "loss": 0.0018, "step": 9765 }, { "epoch": 12.78, "learning_rate": 1.235e-06, "loss": 0.0015, "step": 9766 }, { "epoch": 12.78, "learning_rate": 1.23e-06, "loss": 0.002, "step": 9767 }, { "epoch": 12.79, "learning_rate": 1.2250000000000001e-06, "loss": 0.002, "step": 9768 }, { "epoch": 12.79, "learning_rate": 1.2200000000000002e-06, "loss": 0.0016, "step": 9769 }, { "epoch": 12.79, "learning_rate": 1.215e-06, "loss": 0.0021, "step": 9770 }, { "epoch": 12.79, "learning_rate": 1.21e-06, "loss": 0.0015, "step": 9771 }, { "epoch": 12.79, "learning_rate": 1.2050000000000001e-06, "loss": 0.0029, "step": 9772 }, { "epoch": 12.79, "learning_rate": 1.2000000000000002e-06, "loss": 0.0009, "step": 9773 }, { "epoch": 12.79, "learning_rate": 1.1950000000000002e-06, "loss": 0.003, "step": 9774 }, { "epoch": 12.79, "learning_rate": 1.19e-06, "loss": 0.0016, "step": 9775 }, { "epoch": 12.8, "learning_rate": 1.185e-06, "loss": 0.0019, "step": 9776 }, { "epoch": 12.8, "learning_rate": 1.18e-06, "loss": 0.0022, "step": 9777 }, { "epoch": 12.8, "learning_rate": 1.175e-06, "loss": 0.0012, "step": 9778 }, { "epoch": 12.8, "learning_rate": 1.17e-06, "loss": 0.0042, "step": 9779 }, { "epoch": 12.8, "learning_rate": 1.165e-06, "loss": 0.0021, "step": 9780 }, { "epoch": 12.8, "learning_rate": 1.16e-06, "loss": 0.0019, "step": 9781 }, { "epoch": 12.8, "learning_rate": 1.155e-06, "loss": 0.0027, "step": 9782 }, { "epoch": 12.8, "learning_rate": 1.15e-06, "loss": 0.0023, "step": 9783 }, { "epoch": 12.81, "learning_rate": 1.145e-06, "loss": 0.0016, "step": 9784 }, { "epoch": 12.81, "learning_rate": 1.14e-06, "loss": 0.002, "step": 9785 }, { "epoch": 12.81, "learning_rate": 1.1350000000000001e-06, "loss": 0.0035, "step": 9786 }, { "epoch": 12.81, "learning_rate": 1.13e-06, "loss": 0.0023, "step": 9787 }, { "epoch": 12.81, "learning_rate": 1.125e-06, "loss": 0.0032, "step": 9788 }, { "epoch": 12.81, "learning_rate": 1.12e-06, "loss": 0.0017, "step": 9789 }, { "epoch": 12.81, "learning_rate": 1.1150000000000001e-06, "loss": 0.0028, "step": 9790 }, { "epoch": 12.82, "learning_rate": 1.1100000000000002e-06, "loss": 0.0023, "step": 9791 }, { "epoch": 12.82, "learning_rate": 1.1050000000000002e-06, "loss": 0.0025, "step": 9792 }, { "epoch": 12.82, "learning_rate": 1.1e-06, "loss": 0.0018, "step": 9793 }, { "epoch": 12.82, "learning_rate": 1.095e-06, "loss": 0.0017, "step": 9794 }, { "epoch": 12.82, "learning_rate": 1.0900000000000002e-06, "loss": 0.004, "step": 9795 }, { "epoch": 12.82, "learning_rate": 1.085e-06, "loss": 0.002, "step": 9796 }, { "epoch": 12.82, "learning_rate": 1.08e-06, "loss": 0.0013, "step": 9797 }, { "epoch": 12.82, "learning_rate": 1.0749999999999999e-06, "loss": 0.0018, "step": 9798 }, { "epoch": 12.83, "learning_rate": 1.07e-06, "loss": 0.002, "step": 9799 }, { "epoch": 12.83, "learning_rate": 1.065e-06, "loss": 0.0023, "step": 9800 }, { "epoch": 12.83, "learning_rate": 1.06e-06, "loss": 0.0014, "step": 9801 }, { "epoch": 12.83, "learning_rate": 1.055e-06, "loss": 0.0025, "step": 9802 }, { "epoch": 12.83, "learning_rate": 1.0500000000000001e-06, "loss": 0.0019, "step": 9803 }, { "epoch": 12.83, "learning_rate": 1.045e-06, "loss": 0.0019, "step": 9804 }, { "epoch": 12.83, "learning_rate": 1.04e-06, "loss": 0.0011, "step": 9805 }, { "epoch": 12.84, "learning_rate": 1.035e-06, "loss": 0.0014, "step": 9806 }, { "epoch": 12.84, "learning_rate": 1.03e-06, "loss": 0.0026, "step": 9807 }, { "epoch": 12.84, "learning_rate": 1.0250000000000001e-06, "loss": 0.0017, "step": 9808 }, { "epoch": 12.84, "learning_rate": 1.0200000000000002e-06, "loss": 0.0012, "step": 9809 }, { "epoch": 12.84, "learning_rate": 1.015e-06, "loss": 0.0023, "step": 9810 }, { "epoch": 12.84, "learning_rate": 1.01e-06, "loss": 0.0027, "step": 9811 }, { "epoch": 12.84, "learning_rate": 1.0050000000000001e-06, "loss": 0.0027, "step": 9812 }, { "epoch": 12.84, "learning_rate": 1.0000000000000002e-06, "loss": 0.0011, "step": 9813 }, { "epoch": 12.85, "learning_rate": 9.95e-07, "loss": 0.0022, "step": 9814 }, { "epoch": 12.85, "learning_rate": 9.9e-07, "loss": 0.0018, "step": 9815 }, { "epoch": 12.85, "learning_rate": 9.849999999999999e-07, "loss": 0.0018, "step": 9816 }, { "epoch": 12.85, "learning_rate": 9.8e-07, "loss": 0.0028, "step": 9817 }, { "epoch": 12.85, "learning_rate": 9.75e-07, "loss": 0.0024, "step": 9818 }, { "epoch": 12.85, "learning_rate": 9.7e-07, "loss": 0.0016, "step": 9819 }, { "epoch": 12.85, "learning_rate": 9.65e-07, "loss": 0.0017, "step": 9820 }, { "epoch": 12.85, "learning_rate": 9.6e-07, "loss": 0.0021, "step": 9821 }, { "epoch": 12.86, "learning_rate": 9.55e-07, "loss": 0.0027, "step": 9822 }, { "epoch": 12.86, "learning_rate": 9.5e-07, "loss": 0.0012, "step": 9823 }, { "epoch": 12.86, "learning_rate": 9.450000000000001e-07, "loss": 0.0034, "step": 9824 }, { "epoch": 12.86, "learning_rate": 9.400000000000001e-07, "loss": 0.0013, "step": 9825 }, { "epoch": 12.86, "learning_rate": 9.350000000000002e-07, "loss": 0.0017, "step": 9826 }, { "epoch": 12.86, "learning_rate": 9.3e-07, "loss": 0.0008, "step": 9827 }, { "epoch": 12.86, "learning_rate": 9.25e-07, "loss": 0.0021, "step": 9828 }, { "epoch": 12.87, "learning_rate": 9.2e-07, "loss": 0.0021, "step": 9829 }, { "epoch": 12.87, "learning_rate": 9.15e-07, "loss": 0.0021, "step": 9830 }, { "epoch": 12.87, "learning_rate": 9.100000000000001e-07, "loss": 0.0024, "step": 9831 }, { "epoch": 12.87, "learning_rate": 9.050000000000001e-07, "loss": 0.0015, "step": 9832 }, { "epoch": 12.87, "learning_rate": 9e-07, "loss": 0.0041, "step": 9833 }, { "epoch": 12.87, "learning_rate": 8.95e-07, "loss": 0.0018, "step": 9834 }, { "epoch": 12.87, "learning_rate": 8.900000000000001e-07, "loss": 0.0011, "step": 9835 }, { "epoch": 12.87, "learning_rate": 8.850000000000001e-07, "loss": 0.0007, "step": 9836 }, { "epoch": 12.88, "learning_rate": 8.8e-07, "loss": 0.0024, "step": 9837 }, { "epoch": 12.88, "learning_rate": 8.750000000000001e-07, "loss": 0.0017, "step": 9838 }, { "epoch": 12.88, "learning_rate": 8.699999999999999e-07, "loss": 0.0025, "step": 9839 }, { "epoch": 12.88, "learning_rate": 8.65e-07, "loss": 0.0025, "step": 9840 }, { "epoch": 12.88, "learning_rate": 8.6e-07, "loss": 0.0027, "step": 9841 }, { "epoch": 12.88, "learning_rate": 8.550000000000001e-07, "loss": 0.0025, "step": 9842 }, { "epoch": 12.88, "learning_rate": 8.500000000000001e-07, "loss": 0.0019, "step": 9843 }, { "epoch": 12.88, "learning_rate": 8.45e-07, "loss": 0.0021, "step": 9844 }, { "epoch": 12.89, "learning_rate": 8.4e-07, "loss": 0.0027, "step": 9845 }, { "epoch": 12.89, "learning_rate": 8.35e-07, "loss": 0.0029, "step": 9846 }, { "epoch": 12.89, "learning_rate": 8.300000000000001e-07, "loss": 0.0022, "step": 9847 }, { "epoch": 12.89, "learning_rate": 8.25e-07, "loss": 0.0018, "step": 9848 }, { "epoch": 12.89, "learning_rate": 8.200000000000001e-07, "loss": 0.0053, "step": 9849 }, { "epoch": 12.89, "learning_rate": 8.149999999999999e-07, "loss": 0.0032, "step": 9850 }, { "epoch": 12.89, "learning_rate": 8.1e-07, "loss": 0.0025, "step": 9851 }, { "epoch": 12.9, "learning_rate": 8.05e-07, "loss": 0.0013, "step": 9852 }, { "epoch": 12.9, "learning_rate": 8.000000000000001e-07, "loss": 0.0021, "step": 9853 }, { "epoch": 12.9, "learning_rate": 7.950000000000001e-07, "loss": 0.0021, "step": 9854 }, { "epoch": 12.9, "learning_rate": 7.900000000000002e-07, "loss": 0.0025, "step": 9855 }, { "epoch": 12.9, "learning_rate": 7.85e-07, "loss": 0.002, "step": 9856 }, { "epoch": 12.9, "learning_rate": 7.8e-07, "loss": 0.0023, "step": 9857 }, { "epoch": 12.9, "learning_rate": 7.75e-07, "loss": 0.0014, "step": 9858 }, { "epoch": 12.9, "learning_rate": 7.7e-07, "loss": 0.0017, "step": 9859 }, { "epoch": 12.91, "learning_rate": 7.65e-07, "loss": 0.0021, "step": 9860 }, { "epoch": 12.91, "learning_rate": 7.6e-07, "loss": 0.002, "step": 9861 }, { "epoch": 12.91, "learning_rate": 7.550000000000001e-07, "loss": 0.0017, "step": 9862 }, { "epoch": 12.91, "learning_rate": 7.5e-07, "loss": 0.0014, "step": 9863 }, { "epoch": 12.91, "learning_rate": 7.450000000000001e-07, "loss": 0.002, "step": 9864 }, { "epoch": 12.91, "learning_rate": 7.400000000000001e-07, "loss": 0.0036, "step": 9865 }, { "epoch": 12.91, "learning_rate": 7.350000000000001e-07, "loss": 0.0014, "step": 9866 }, { "epoch": 12.91, "learning_rate": 7.3e-07, "loss": 0.0024, "step": 9867 }, { "epoch": 12.92, "learning_rate": 7.25e-07, "loss": 0.0019, "step": 9868 }, { "epoch": 12.92, "learning_rate": 7.2e-07, "loss": 0.002, "step": 9869 }, { "epoch": 12.92, "learning_rate": 7.15e-07, "loss": 0.0025, "step": 9870 }, { "epoch": 12.92, "learning_rate": 7.100000000000001e-07, "loss": 0.0019, "step": 9871 }, { "epoch": 12.92, "learning_rate": 7.05e-07, "loss": 0.0021, "step": 9872 }, { "epoch": 12.92, "learning_rate": 7.000000000000001e-07, "loss": 0.0019, "step": 9873 }, { "epoch": 12.92, "learning_rate": 6.95e-07, "loss": 0.0017, "step": 9874 }, { "epoch": 12.93, "learning_rate": 6.900000000000001e-07, "loss": 0.0012, "step": 9875 }, { "epoch": 12.93, "learning_rate": 6.85e-07, "loss": 0.0014, "step": 9876 }, { "epoch": 12.93, "learning_rate": 6.8e-07, "loss": 0.0035, "step": 9877 }, { "epoch": 12.93, "learning_rate": 6.75e-07, "loss": 0.0026, "step": 9878 }, { "epoch": 12.93, "learning_rate": 6.7e-07, "loss": 0.0026, "step": 9879 }, { "epoch": 12.93, "learning_rate": 6.65e-07, "loss": 0.002, "step": 9880 }, { "epoch": 12.93, "learning_rate": 6.6e-07, "loss": 0.0021, "step": 9881 }, { "epoch": 12.93, "learning_rate": 6.550000000000001e-07, "loss": 0.0029, "step": 9882 }, { "epoch": 12.94, "learning_rate": 6.5e-07, "loss": 0.0014, "step": 9883 }, { "epoch": 12.94, "learning_rate": 6.450000000000001e-07, "loss": 0.0023, "step": 9884 }, { "epoch": 12.94, "learning_rate": 6.4e-07, "loss": 0.0016, "step": 9885 }, { "epoch": 12.94, "learning_rate": 6.35e-07, "loss": 0.0013, "step": 9886 }, { "epoch": 12.94, "learning_rate": 6.3e-07, "loss": 0.002, "step": 9887 }, { "epoch": 12.94, "learning_rate": 6.25e-07, "loss": 0.0027, "step": 9888 }, { "epoch": 12.94, "learning_rate": 6.2e-07, "loss": 0.0019, "step": 9889 }, { "epoch": 12.95, "learning_rate": 6.15e-07, "loss": 0.0027, "step": 9890 }, { "epoch": 12.95, "learning_rate": 6.100000000000001e-07, "loss": 0.0021, "step": 9891 }, { "epoch": 12.95, "learning_rate": 6.05e-07, "loss": 0.0017, "step": 9892 }, { "epoch": 12.95, "learning_rate": 6.000000000000001e-07, "loss": 0.0018, "step": 9893 }, { "epoch": 12.95, "learning_rate": 5.95e-07, "loss": 0.0023, "step": 9894 }, { "epoch": 12.95, "learning_rate": 5.9e-07, "loss": 0.002, "step": 9895 }, { "epoch": 12.95, "learning_rate": 5.85e-07, "loss": 0.0017, "step": 9896 }, { "epoch": 12.95, "learning_rate": 5.8e-07, "loss": 0.0023, "step": 9897 }, { "epoch": 12.96, "learning_rate": 5.75e-07, "loss": 0.0026, "step": 9898 }, { "epoch": 12.96, "learning_rate": 5.7e-07, "loss": 0.0024, "step": 9899 }, { "epoch": 12.96, "learning_rate": 5.65e-07, "loss": 0.0021, "step": 9900 }, { "epoch": 12.96, "learning_rate": 5.6e-07, "loss": 0.0019, "step": 9901 }, { "epoch": 12.96, "learning_rate": 5.550000000000001e-07, "loss": 0.003, "step": 9902 }, { "epoch": 12.96, "learning_rate": 5.5e-07, "loss": 0.0023, "step": 9903 }, { "epoch": 12.96, "learning_rate": 5.450000000000001e-07, "loss": 0.0027, "step": 9904 }, { "epoch": 12.96, "learning_rate": 5.4e-07, "loss": 0.0031, "step": 9905 }, { "epoch": 12.97, "learning_rate": 5.35e-07, "loss": 0.0026, "step": 9906 }, { "epoch": 12.97, "learning_rate": 5.3e-07, "loss": 0.0012, "step": 9907 }, { "epoch": 12.97, "learning_rate": 5.250000000000001e-07, "loss": 0.0023, "step": 9908 }, { "epoch": 12.97, "learning_rate": 5.2e-07, "loss": 0.0018, "step": 9909 }, { "epoch": 12.97, "learning_rate": 5.15e-07, "loss": 0.0021, "step": 9910 }, { "epoch": 12.97, "learning_rate": 5.100000000000001e-07, "loss": 0.0013, "step": 9911 }, { "epoch": 12.97, "learning_rate": 5.05e-07, "loss": 0.002, "step": 9912 }, { "epoch": 12.98, "learning_rate": 5.000000000000001e-07, "loss": 0.0024, "step": 9913 }, { "epoch": 12.98, "learning_rate": 4.95e-07, "loss": 0.0034, "step": 9914 }, { "epoch": 12.98, "learning_rate": 4.9e-07, "loss": 0.0022, "step": 9915 }, { "epoch": 12.98, "learning_rate": 4.85e-07, "loss": 0.0036, "step": 9916 }, { "epoch": 12.98, "learning_rate": 4.8e-07, "loss": 0.0016, "step": 9917 }, { "epoch": 12.98, "learning_rate": 4.75e-07, "loss": 0.002, "step": 9918 }, { "epoch": 12.98, "learning_rate": 4.7000000000000005e-07, "loss": 0.002, "step": 9919 }, { "epoch": 12.98, "learning_rate": 4.65e-07, "loss": 0.0025, "step": 9920 }, { "epoch": 12.99, "learning_rate": 4.6e-07, "loss": 0.0024, "step": 9921 }, { "epoch": 12.99, "learning_rate": 4.5500000000000004e-07, "loss": 0.0029, "step": 9922 }, { "epoch": 12.99, "learning_rate": 4.5e-07, "loss": 0.0017, "step": 9923 }, { "epoch": 12.99, "learning_rate": 4.4500000000000003e-07, "loss": 0.0037, "step": 9924 }, { "epoch": 12.99, "learning_rate": 4.4e-07, "loss": 0.0019, "step": 9925 }, { "epoch": 12.99, "learning_rate": 4.3499999999999996e-07, "loss": 0.0016, "step": 9926 }, { "epoch": 12.99, "learning_rate": 4.3e-07, "loss": 0.0024, "step": 9927 }, { "epoch": 12.99, "learning_rate": 4.2500000000000006e-07, "loss": 0.0013, "step": 9928 }, { "epoch": 13.0, "learning_rate": 4.2e-07, "loss": 0.0019, "step": 9929 }, { "epoch": 13.0, "learning_rate": 4.1500000000000005e-07, "loss": 0.0044, "step": 9930 }, { "epoch": 13.0, "learning_rate": 4.1000000000000004e-07, "loss": 0.0021, "step": 9931 }, { "epoch": 13.0, "learning_rate": 4.05e-07, "loss": 0.0016, "step": 9932 }, { "epoch": 13.0, "learning_rate": 4.0000000000000003e-07, "loss": 0.0014, "step": 9933 }, { "epoch": 13.0, "learning_rate": 3.950000000000001e-07, "loss": 0.0014, "step": 9934 }, { "epoch": 13.0, "learning_rate": 3.9e-07, "loss": 0.0022, "step": 9935 }, { "epoch": 13.01, "learning_rate": 3.85e-07, "loss": 0.0017, "step": 9936 }, { "epoch": 13.01, "learning_rate": 3.8e-07, "loss": 0.0011, "step": 9937 }, { "epoch": 13.01, "learning_rate": 3.75e-07, "loss": 0.0016, "step": 9938 }, { "epoch": 13.01, "learning_rate": 3.7000000000000006e-07, "loss": 0.0016, "step": 9939 }, { "epoch": 13.01, "learning_rate": 3.65e-07, "loss": 0.0016, "step": 9940 }, { "epoch": 13.01, "learning_rate": 3.6e-07, "loss": 0.0021, "step": 9941 }, { "epoch": 13.01, "learning_rate": 3.5500000000000004e-07, "loss": 0.0034, "step": 9942 }, { "epoch": 13.01, "learning_rate": 3.5000000000000004e-07, "loss": 0.0014, "step": 9943 }, { "epoch": 13.02, "learning_rate": 3.4500000000000003e-07, "loss": 0.0021, "step": 9944 }, { "epoch": 13.02, "learning_rate": 3.4e-07, "loss": 0.0022, "step": 9945 }, { "epoch": 13.02, "learning_rate": 3.35e-07, "loss": 0.0012, "step": 9946 }, { "epoch": 13.02, "learning_rate": 3.3e-07, "loss": 0.002, "step": 9947 }, { "epoch": 13.02, "learning_rate": 3.25e-07, "loss": 0.0025, "step": 9948 }, { "epoch": 13.02, "learning_rate": 3.2e-07, "loss": 0.0016, "step": 9949 }, { "epoch": 13.02, "learning_rate": 3.15e-07, "loss": 0.0015, "step": 9950 }, { "epoch": 13.02, "learning_rate": 3.1e-07, "loss": 0.0014, "step": 9951 }, { "epoch": 13.03, "learning_rate": 3.0500000000000004e-07, "loss": 0.0018, "step": 9952 }, { "epoch": 13.03, "learning_rate": 3.0000000000000004e-07, "loss": 0.0022, "step": 9953 }, { "epoch": 13.03, "learning_rate": 2.95e-07, "loss": 0.0024, "step": 9954 }, { "epoch": 13.03, "learning_rate": 2.9e-07, "loss": 0.0017, "step": 9955 }, { "epoch": 13.03, "learning_rate": 2.85e-07, "loss": 0.003, "step": 9956 }, { "epoch": 13.03, "learning_rate": 2.8e-07, "loss": 0.0026, "step": 9957 }, { "epoch": 13.03, "learning_rate": 2.75e-07, "loss": 0.0019, "step": 9958 }, { "epoch": 13.04, "learning_rate": 2.7e-07, "loss": 0.0021, "step": 9959 }, { "epoch": 13.04, "learning_rate": 2.65e-07, "loss": 0.0039, "step": 9960 }, { "epoch": 13.04, "learning_rate": 2.6e-07, "loss": 0.0017, "step": 9961 }, { "epoch": 13.04, "learning_rate": 2.5500000000000005e-07, "loss": 0.0019, "step": 9962 }, { "epoch": 13.04, "learning_rate": 2.5000000000000004e-07, "loss": 0.0011, "step": 9963 }, { "epoch": 13.04, "learning_rate": 2.45e-07, "loss": 0.0023, "step": 9964 }, { "epoch": 13.04, "learning_rate": 2.4e-07, "loss": 0.0027, "step": 9965 }, { "epoch": 13.04, "learning_rate": 2.3500000000000003e-07, "loss": 0.0014, "step": 9966 }, { "epoch": 13.05, "learning_rate": 2.3e-07, "loss": 0.0024, "step": 9967 }, { "epoch": 13.05, "learning_rate": 2.25e-07, "loss": 0.0011, "step": 9968 }, { "epoch": 13.05, "learning_rate": 2.2e-07, "loss": 0.0018, "step": 9969 }, { "epoch": 13.05, "learning_rate": 2.15e-07, "loss": 0.0019, "step": 9970 }, { "epoch": 13.05, "learning_rate": 2.1e-07, "loss": 0.0029, "step": 9971 }, { "epoch": 13.05, "learning_rate": 2.0500000000000002e-07, "loss": 0.0014, "step": 9972 }, { "epoch": 13.05, "learning_rate": 2.0000000000000002e-07, "loss": 0.0014, "step": 9973 }, { "epoch": 13.05, "learning_rate": 1.95e-07, "loss": 0.0025, "step": 9974 }, { "epoch": 13.06, "learning_rate": 1.9e-07, "loss": 0.0013, "step": 9975 }, { "epoch": 13.06, "learning_rate": 1.8500000000000003e-07, "loss": 0.0022, "step": 9976 }, { "epoch": 13.06, "learning_rate": 1.8e-07, "loss": 0.0017, "step": 9977 }, { "epoch": 13.06, "learning_rate": 1.7500000000000002e-07, "loss": 0.0026, "step": 9978 }, { "epoch": 13.06, "learning_rate": 1.7e-07, "loss": 0.0016, "step": 9979 }, { "epoch": 13.06, "learning_rate": 1.65e-07, "loss": 0.0019, "step": 9980 }, { "epoch": 13.06, "learning_rate": 1.6e-07, "loss": 0.0021, "step": 9981 }, { "epoch": 13.07, "learning_rate": 1.55e-07, "loss": 0.003, "step": 9982 }, { "epoch": 13.07, "learning_rate": 1.5000000000000002e-07, "loss": 0.0017, "step": 9983 }, { "epoch": 13.07, "learning_rate": 1.45e-07, "loss": 0.0018, "step": 9984 }, { "epoch": 13.07, "learning_rate": 1.4e-07, "loss": 0.0018, "step": 9985 }, { "epoch": 13.07, "learning_rate": 1.35e-07, "loss": 0.0012, "step": 9986 }, { "epoch": 13.07, "learning_rate": 1.3e-07, "loss": 0.0029, "step": 9987 }, { "epoch": 13.07, "learning_rate": 1.2500000000000002e-07, "loss": 0.0019, "step": 9988 }, { "epoch": 13.07, "learning_rate": 1.2e-07, "loss": 0.0021, "step": 9989 }, { "epoch": 13.08, "learning_rate": 1.15e-07, "loss": 0.002, "step": 9990 }, { "epoch": 13.08, "learning_rate": 1.1e-07, "loss": 0.0013, "step": 9991 }, { "epoch": 13.08, "learning_rate": 1.05e-07, "loss": 0.0006, "step": 9992 }, { "epoch": 13.08, "learning_rate": 1.0000000000000001e-07, "loss": 0.0023, "step": 9993 }, { "epoch": 13.08, "learning_rate": 9.5e-08, "loss": 0.0017, "step": 9994 }, { "epoch": 13.08, "learning_rate": 9e-08, "loss": 0.0024, "step": 9995 }, { "epoch": 13.08, "learning_rate": 8.5e-08, "loss": 0.0024, "step": 9996 }, { "epoch": 13.09, "learning_rate": 8e-08, "loss": 0.0011, "step": 9997 }, { "epoch": 13.09, "learning_rate": 7.500000000000001e-08, "loss": 0.0015, "step": 9998 }, { "epoch": 13.09, "learning_rate": 7e-08, "loss": 0.0018, "step": 9999 }, { "epoch": 13.09, "learning_rate": 6.5e-08, "loss": 0.0033, "step": 10000 }, { "epoch": 13.09, "step": 10000, "total_flos": 9.399866847600522e+19, "train_loss": 0.17524216154664754, "train_runtime": 141322.6962, "train_samples_per_second": 4.529, "train_steps_per_second": 0.071 } ], "max_steps": 10000, "num_train_epochs": 14, "total_flos": 9.399866847600522e+19, "trial_name": null, "trial_params": null }