| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 29.62962962962963, | |
| "eval_steps": 500, | |
| "global_step": 800, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.999970808109463e-05, | |
| "loss": 5.1275, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.9998814492475354e-05, | |
| "loss": 4.575, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.9997319220216543e-05, | |
| "loss": 3.7996, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.999522235429695e-05, | |
| "loss": 3.5835, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.9992524020896517e-05, | |
| "loss": 3.185, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.9989224382388813e-05, | |
| "loss": 2.7431, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.9985323637331214e-05, | |
| "loss": 2.6319, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.9980822020453007e-05, | |
| "loss": 2.4273, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.9975719802641226e-05, | |
| "loss": 2.3277, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.9970017290924365e-05, | |
| "loss": 2.2614, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.99637148284539e-05, | |
| "loss": 2.2018, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.9956812794483658e-05, | |
| "loss": 2.0017, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.9949311604346965e-05, | |
| "loss": 2.0208, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.994121170943167e-05, | |
| "loss": 1.8773, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.993251359715298e-05, | |
| "loss": 1.7932, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.992321779092414e-05, | |
| "loss": 1.806, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.991332485012491e-05, | |
| "loss": 1.731, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.9902835370067927e-05, | |
| "loss": 1.7323, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.989174998196288e-05, | |
| "loss": 1.7465, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.9880069352878508e-05, | |
| "loss": 1.7926, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.986779418570249e-05, | |
| "loss": 1.7291, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.9854925219099116e-05, | |
| "loss": 1.716, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.9841463227464862e-05, | |
| "loss": 1.5832, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.9827409020881775e-05, | |
| "loss": 1.6131, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.9812763445068725e-05, | |
| "loss": 1.6786, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.979752738133054e-05, | |
| "loss": 1.5621, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.9781701746504926e-05, | |
| "loss": 1.6118, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.9765287492907347e-05, | |
| "loss": 1.4956, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.974828560827368e-05, | |
| "loss": 1.5483, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.97306971157008e-05, | |
| "loss": 1.4634, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.9712523073585e-05, | |
| "loss": 1.5822, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.969376457555831e-05, | |
| "loss": 1.4255, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1.967442275042269e-05, | |
| "loss": 1.5571, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 1.9654498762082096e-05, | |
| "loss": 1.4493, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 1.963399380947244e-05, | |
| "loss": 1.4962, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 1.9612909126489457e-05, | |
| "loss": 1.445, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.9591245981914446e-05, | |
| "loss": 1.5671, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1.9569005679337905e-05, | |
| "loss": 1.4461, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.9546189557081124e-05, | |
| "loss": 1.4698, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 1.952279898811561e-05, | |
| "loss": 1.5083, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 1.949883537998049e-05, | |
| "loss": 1.382, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 1.947430017469783e-05, | |
| "loss": 1.3836, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.9449194848685807e-05, | |
| "loss": 1.386, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 1.9423520912669913e-05, | |
| "loss": 1.4059, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.9397279911592034e-05, | |
| "loss": 1.3551, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.9370473424517463e-05, | |
| "loss": 1.2524, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 1.9343103064539915e-05, | |
| "loss": 1.3327, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 1.9315170478684423e-05, | |
| "loss": 1.3513, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 1.928667734780825e-05, | |
| "loss": 1.3488, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 1.9257625386499716e-05, | |
| "loss": 1.372, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.922801634297507e-05, | |
| "loss": 1.298, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 1.9197851998973237e-05, | |
| "loss": 1.3144, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.916713416964864e-05, | |
| "loss": 1.3531, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.9135864703461943e-05, | |
| "loss": 1.3806, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 1.910404548206884e-05, | |
| "loss": 1.2281, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 1.907167842020683e-05, | |
| "loss": 1.2588, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 1.903876546557996e-05, | |
| "loss": 1.2666, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 1.9005308598741662e-05, | |
| "loss": 1.213, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 1.897130983297555e-05, | |
| "loss": 1.187, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 1.8936771214174292e-05, | |
| "loss": 1.1696, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 1.8901694820716456e-05, | |
| "loss": 1.1778, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 1.886608276334148e-05, | |
| "loss": 1.2148, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 1.882993718502265e-05, | |
| "loss": 1.1771, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 1.8793260260838127e-05, | |
| "loss": 1.2162, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 1.875605419784009e-05, | |
| "loss": 1.213, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 1.8718321234921895e-05, | |
| "loss": 1.2071, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 1.868006364268337e-05, | |
| "loss": 1.1868, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 1.8641283723294177e-05, | |
| "loss": 1.1492, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 1.8601983810355265e-05, | |
| "loss": 1.1094, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 1.8562166268758462e-05, | |
| "loss": 1.1069, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 1.8521833494544153e-05, | |
| "loss": 1.1004, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 1.848098791475711e-05, | |
| "loss": 1.0538, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 1.8439631987300418e-05, | |
| "loss": 1.0556, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 1.8397768200787613e-05, | |
| "loss": 1.0134, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 1.8355399074392875e-05, | |
| "loss": 1.0615, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 1.8312527157699488e-05, | |
| "loss": 1.0209, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 1.826915503054637e-05, | |
| "loss": 1.0104, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 1.8225285302872872e-05, | |
| "loss": 1.0017, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 1.818092061456169e-05, | |
| "loss": 1.0131, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 1.8136063635280036e-05, | |
| "loss": 1.0151, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 1.8090717064318958e-05, | |
| "loss": 0.9734, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 1.804488363043094e-05, | |
| "loss": 0.8721, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "learning_rate": 1.799856609166568e-05, | |
| "loss": 0.885, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 1.7951767235204142e-05, | |
| "loss": 0.8312, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 1.79044898771908e-05, | |
| "loss": 0.8543, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 1.7856736862564208e-05, | |
| "loss": 0.8782, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 1.7808511064885798e-05, | |
| "loss": 0.8549, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 1.7759815386166945e-05, | |
| "loss": 0.795, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 1.771065275669436e-05, | |
| "loss": 0.883, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 1.766102613485375e-05, | |
| "loss": 0.8116, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 1.761093850695178e-05, | |
| "loss": 0.8215, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 1.75603928870364e-05, | |
| "loss": 0.8416, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 1.7509392316715445e-05, | |
| "loss": 0.7994, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 1.7457939864973625e-05, | |
| "loss": 0.8312, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 1.7406038627987835e-05, | |
| "loss": 0.7368, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 1.7353691728940852e-05, | |
| "loss": 0.6201, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 7.19, | |
| "learning_rate": 1.7300902317833377e-05, | |
| "loss": 0.6334, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 1.724767357129451e-05, | |
| "loss": 0.6199, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 1.7194008692390566e-05, | |
| "loss": 0.6208, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 1.7139910910432353e-05, | |
| "loss": 0.6039, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 1.708538348078082e-05, | |
| "loss": 0.5737, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 7.56, | |
| "learning_rate": 1.7030429684651203e-05, | |
| "loss": 0.6267, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 7.63, | |
| "learning_rate": 1.6975052828915534e-05, | |
| "loss": 0.6654, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "learning_rate": 1.6919256245903675e-05, | |
| "loss": 0.6041, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 1.686304329320278e-05, | |
| "loss": 0.6349, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 7.85, | |
| "learning_rate": 1.6806417353455267e-05, | |
| "loss": 0.6032, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 7.93, | |
| "learning_rate": 1.674938183415525e-05, | |
| "loss": 0.6501, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 1.6691940167443487e-05, | |
| "loss": 0.5961, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "learning_rate": 1.6634095809900873e-05, | |
| "loss": 0.4455, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 1.657585224234042e-05, | |
| "loss": 0.4305, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "learning_rate": 1.6517212969597793e-05, | |
| "loss": 0.4517, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 1.6458181520320426e-05, | |
| "loss": 0.4096, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "learning_rate": 1.639876144675516e-05, | |
| "loss": 0.4079, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "learning_rate": 1.6338956324534514e-05, | |
| "loss": 0.4153, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 1.6278769752461477e-05, | |
| "loss": 0.3869, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 1.6218205352292984e-05, | |
| "loss": 0.3988, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 1.6157266768521967e-05, | |
| "loss": 0.4212, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 8.74, | |
| "learning_rate": 1.6095957668158043e-05, | |
| "loss": 0.4137, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 8.81, | |
| "learning_rate": 1.6034281740506846e-05, | |
| "loss": 0.4265, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 8.89, | |
| "learning_rate": 1.5972242696948024e-05, | |
| "loss": 0.3991, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 1.59098442707119e-05, | |
| "loss": 0.384, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 1.584709021665485e-05, | |
| "loss": 0.3623, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 1.5783984311033296e-05, | |
| "loss": 0.2732, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 1.5720530351276542e-05, | |
| "loss": 0.2461, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 1.565673215575819e-05, | |
| "loss": 0.266, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 9.33, | |
| "learning_rate": 1.5592593563566424e-05, | |
| "loss": 0.2524, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "learning_rate": 1.552811843427295e-05, | |
| "loss": 0.2587, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 1.5463310647700777e-05, | |
| "loss": 0.2538, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 1.5398174103690717e-05, | |
| "loss": 0.2541, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 9.63, | |
| "learning_rate": 1.5332712721866735e-05, | |
| "loss": 0.2695, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 9.7, | |
| "learning_rate": 1.5266930441400066e-05, | |
| "loss": 0.2887, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 9.78, | |
| "learning_rate": 1.5200831220772187e-05, | |
| "loss": 0.2753, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 1.5134419037536603e-05, | |
| "loss": 0.2613, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 9.93, | |
| "learning_rate": 1.5067697888079496e-05, | |
| "loss": 0.2655, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 1.5000671787379257e-05, | |
| "loss": 0.2828, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 10.07, | |
| "learning_rate": 1.4933344768764857e-05, | |
| "loss": 0.2013, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 10.15, | |
| "learning_rate": 1.4865720883673153e-05, | |
| "loss": 0.194, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 10.22, | |
| "learning_rate": 1.4797804201405099e-05, | |
| "loss": 0.1917, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 10.3, | |
| "learning_rate": 1.472959880888086e-05, | |
| "loss": 0.1783, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 10.37, | |
| "learning_rate": 1.4661108810393883e-05, | |
| "loss": 0.1992, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 10.44, | |
| "learning_rate": 1.4592338327363917e-05, | |
| "loss": 0.1804, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 10.52, | |
| "learning_rate": 1.4523291498089021e-05, | |
| "loss": 0.1834, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 10.59, | |
| "learning_rate": 1.4453972477496509e-05, | |
| "loss": 0.1938, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 10.67, | |
| "learning_rate": 1.4384385436892963e-05, | |
| "loss": 0.1756, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 10.74, | |
| "learning_rate": 1.4314534563713186e-05, | |
| "loss": 0.1928, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "learning_rate": 1.4244424061268239e-05, | |
| "loss": 0.2066, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 10.89, | |
| "learning_rate": 1.4174058148492515e-05, | |
| "loss": 0.2115, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 10.96, | |
| "learning_rate": 1.4103441059689843e-05, | |
| "loss": 0.2111, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 1.4032577044278695e-05, | |
| "loss": 0.1766, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 11.11, | |
| "learning_rate": 1.3961470366536472e-05, | |
| "loss": 0.1426, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 11.19, | |
| "learning_rate": 1.3890125305342905e-05, | |
| "loss": 0.1705, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 11.26, | |
| "learning_rate": 1.3818546153922567e-05, | |
| "loss": 0.1591, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 11.33, | |
| "learning_rate": 1.374673721958653e-05, | |
| "loss": 0.1579, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 11.41, | |
| "learning_rate": 1.3674702823473165e-05, | |
| "loss": 0.1579, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 11.48, | |
| "learning_rate": 1.3602447300288114e-05, | |
| "loss": 0.1607, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 11.56, | |
| "learning_rate": 1.3529974998043453e-05, | |
| "loss": 0.1569, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 11.63, | |
| "learning_rate": 1.345729027779605e-05, | |
| "loss": 0.1809, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 1.3384397513385126e-05, | |
| "loss": 0.1573, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 11.78, | |
| "learning_rate": 1.3311301091169067e-05, | |
| "loss": 0.1686, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 11.85, | |
| "learning_rate": 1.3238005409761468e-05, | |
| "loss": 0.1827, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 11.93, | |
| "learning_rate": 1.3164514879766443e-05, | |
| "loss": 0.1791, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 1.309083392351322e-05, | |
| "loss": 0.1692, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 12.07, | |
| "learning_rate": 1.3016966974790018e-05, | |
| "loss": 0.1352, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 12.15, | |
| "learning_rate": 1.294291847857725e-05, | |
| "loss": 0.1432, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 12.22, | |
| "learning_rate": 1.2868692890780036e-05, | |
| "loss": 0.139, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 12.3, | |
| "learning_rate": 1.2794294677960073e-05, | |
| "loss": 0.1614, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 12.37, | |
| "learning_rate": 1.2719728317066853e-05, | |
| "loss": 0.1562, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 12.44, | |
| "learning_rate": 1.2644998295168257e-05, | |
| "loss": 0.1419, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 12.52, | |
| "learning_rate": 1.2570109109180561e-05, | |
| "loss": 0.1649, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 12.59, | |
| "learning_rate": 1.249506526559781e-05, | |
| "loss": 0.1539, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 12.67, | |
| "learning_rate": 1.2419871280220646e-05, | |
| "loss": 0.1572, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 12.74, | |
| "learning_rate": 1.2344531677884572e-05, | |
| "loss": 0.1558, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 12.81, | |
| "learning_rate": 1.2269050992187666e-05, | |
| "loss": 0.1521, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 12.89, | |
| "learning_rate": 1.219343376521776e-05, | |
| "loss": 0.153, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 12.96, | |
| "learning_rate": 1.211768454727913e-05, | |
| "loss": 0.1582, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 1.2041807896618676e-05, | |
| "loss": 0.1504, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 13.11, | |
| "learning_rate": 1.1965808379151622e-05, | |
| "loss": 0.1386, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 13.19, | |
| "learning_rate": 1.1889690568186767e-05, | |
| "loss": 0.1438, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 13.26, | |
| "learning_rate": 1.1813459044151276e-05, | |
| "loss": 0.1299, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 13.33, | |
| "learning_rate": 1.1737118394315052e-05, | |
| "loss": 0.1398, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 13.41, | |
| "learning_rate": 1.1660673212514708e-05, | |
| "loss": 0.1417, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 13.48, | |
| "learning_rate": 1.1584128098877112e-05, | |
| "loss": 0.1463, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 13.56, | |
| "learning_rate": 1.1507487659542578e-05, | |
| "loss": 0.1483, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "learning_rate": 1.1430756506387699e-05, | |
| "loss": 0.1514, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 13.7, | |
| "learning_rate": 1.135393925674781e-05, | |
| "loss": 0.1428, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 13.78, | |
| "learning_rate": 1.1277040533139149e-05, | |
| "loss": 0.1588, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "learning_rate": 1.1200064962980692e-05, | |
| "loss": 0.156, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 13.93, | |
| "learning_rate": 1.1123017178315692e-05, | |
| "loss": 0.1455, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 1.104590181553294e-05, | |
| "loss": 0.1627, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 14.07, | |
| "learning_rate": 1.0968723515087786e-05, | |
| "loss": 0.1357, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "learning_rate": 1.0891486921222877e-05, | |
| "loss": 0.1405, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 14.22, | |
| "learning_rate": 1.0814196681688695e-05, | |
| "loss": 0.1307, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 14.3, | |
| "learning_rate": 1.0736857447463887e-05, | |
| "loss": 0.121, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 14.37, | |
| "learning_rate": 1.065947387247536e-05, | |
| "loss": 0.1381, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 14.44, | |
| "learning_rate": 1.0582050613318276e-05, | |
| "loss": 0.1364, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 14.52, | |
| "learning_rate": 1.0504592328975783e-05, | |
| "loss": 0.139, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 14.59, | |
| "learning_rate": 1.0427103680538706e-05, | |
| "loss": 0.137, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 14.67, | |
| "learning_rate": 1.0349589330925039e-05, | |
| "loss": 0.1381, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 14.74, | |
| "learning_rate": 1.0272053944599358e-05, | |
| "loss": 0.1438, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 14.81, | |
| "learning_rate": 1.0194502187292128e-05, | |
| "loss": 0.135, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 14.89, | |
| "learning_rate": 1.0116938725718956e-05, | |
| "loss": 0.1435, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 14.96, | |
| "learning_rate": 1.0039368227299753e-05, | |
| "loss": 0.1518, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 9.961795359877873e-06, | |
| "loss": 0.1295, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 15.11, | |
| "learning_rate": 9.884224791439229e-06, | |
| "loss": 0.1267, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 15.19, | |
| "learning_rate": 9.806661189831397e-06, | |
| "loss": 0.1278, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 15.26, | |
| "learning_rate": 9.729109222482708e-06, | |
| "loss": 0.1344, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 15.33, | |
| "learning_rate": 9.651573556121408e-06, | |
| "loss": 0.1359, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 15.41, | |
| "learning_rate": 9.574058856494819e-06, | |
| "loss": 0.1313, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 15.48, | |
| "learning_rate": 9.496569788088572e-06, | |
| "loss": 0.1369, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 15.56, | |
| "learning_rate": 9.419111013845939e-06, | |
| "loss": 0.1224, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 15.63, | |
| "learning_rate": 9.341687194887216e-06, | |
| "loss": 0.1433, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 15.7, | |
| "learning_rate": 9.264302990229253e-06, | |
| "loss": 0.1263, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 15.78, | |
| "learning_rate": 9.186963056505082e-06, | |
| "loss": 0.1449, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 15.85, | |
| "learning_rate": 9.109672047683723e-06, | |
| "loss": 0.1351, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 15.93, | |
| "learning_rate": 9.032434614790101e-06, | |
| "loss": 0.1391, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 8.955255405625195e-06, | |
| "loss": 0.1392, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 16.07, | |
| "learning_rate": 8.878139064486332e-06, | |
| "loss": 0.1294, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 16.15, | |
| "learning_rate": 8.801090231887733e-06, | |
| "loss": 0.1151, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 16.22, | |
| "learning_rate": 8.724113544281248e-06, | |
| "loss": 0.113, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 16.3, | |
| "learning_rate": 8.647213633777381e-06, | |
| "loss": 0.1272, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 16.37, | |
| "learning_rate": 8.570395127866521e-06, | |
| "loss": 0.1246, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 16.44, | |
| "learning_rate": 8.49366264914051e-06, | |
| "loss": 0.1322, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 16.52, | |
| "learning_rate": 8.417020815014442e-06, | |
| "loss": 0.1351, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 16.59, | |
| "learning_rate": 8.340474237448851e-06, | |
| "loss": 0.1286, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 8.264027522672134e-06, | |
| "loss": 0.1317, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 16.74, | |
| "learning_rate": 8.187685270903419e-06, | |
| "loss": 0.1277, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 16.81, | |
| "learning_rate": 8.1114520760757e-06, | |
| "loss": 0.1361, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 16.89, | |
| "learning_rate": 8.035332525559436e-06, | |
| "loss": 0.1389, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 16.96, | |
| "learning_rate": 7.959331199886464e-06, | |
| "loss": 0.1382, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 7.88345267247439e-06, | |
| "loss": 0.1342, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 17.11, | |
| "learning_rate": 7.807701509351382e-06, | |
| "loss": 0.1158, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 17.19, | |
| "learning_rate": 7.732082268881377e-06, | |
| "loss": 0.1207, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 17.26, | |
| "learning_rate": 7.656599501489827e-06, | |
| "loss": 0.116, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 17.33, | |
| "learning_rate": 7.581257749389826e-06, | |
| "loss": 0.1275, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 17.41, | |
| "learning_rate": 7.506061546308818e-06, | |
| "loss": 0.1182, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 17.48, | |
| "learning_rate": 7.431015417215751e-06, | |
| "loss": 0.1328, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 17.56, | |
| "learning_rate": 7.356123878048806e-06, | |
| "loss": 0.1206, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 17.63, | |
| "learning_rate": 7.2813914354436325e-06, | |
| "loss": 0.125, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 17.7, | |
| "learning_rate": 7.206822586462168e-06, | |
| "loss": 0.1229, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 17.78, | |
| "learning_rate": 7.132421818322015e-06, | |
| "loss": 0.1301, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 17.85, | |
| "learning_rate": 7.058193608126439e-06, | |
| "loss": 0.1367, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 17.93, | |
| "learning_rate": 6.984142422594935e-06, | |
| "loss": 0.1368, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 6.910272717794462e-06, | |
| "loss": 0.1279, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 18.07, | |
| "learning_rate": 6.836588938871271e-06, | |
| "loss": 0.1116, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 18.15, | |
| "learning_rate": 6.763095519783447e-06, | |
| "loss": 0.1124, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 18.22, | |
| "learning_rate": 6.689796883034059e-06, | |
| "loss": 0.1192, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 18.3, | |
| "learning_rate": 6.616697439405073e-06, | |
| "loss": 0.1209, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 18.37, | |
| "learning_rate": 6.54380158769189e-06, | |
| "loss": 0.1154, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 18.44, | |
| "learning_rate": 6.471113714438679e-06, | |
| "loss": 0.1261, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 18.52, | |
| "learning_rate": 6.398638193674396e-06, | |
| "loss": 0.1325, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 18.59, | |
| "learning_rate": 6.326379386649579e-06, | |
| "loss": 0.1204, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 18.67, | |
| "learning_rate": 6.2543416415739175e-06, | |
| "loss": 0.1184, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 18.74, | |
| "learning_rate": 6.182529293354578e-06, | |
| "loss": 0.1292, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 18.81, | |
| "learning_rate": 6.110946663335366e-06, | |
| "loss": 0.129, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 18.89, | |
| "learning_rate": 6.0395980590366685e-06, | |
| "loss": 0.1234, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 18.96, | |
| "learning_rate": 5.968487773896281e-06, | |
| "loss": 0.1337, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 5.897620087010998e-06, | |
| "loss": 0.1263, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 19.11, | |
| "learning_rate": 5.826999262879164e-06, | |
| "loss": 0.1153, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 19.19, | |
| "learning_rate": 5.7566295511440195e-06, | |
| "loss": 0.1218, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 19.26, | |
| "learning_rate": 5.68651518633801e-06, | |
| "loss": 0.1215, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 19.33, | |
| "learning_rate": 5.616660387627925e-06, | |
| "loss": 0.1173, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 19.41, | |
| "learning_rate": 5.547069358561064e-06, | |
| "loss": 0.1194, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 19.48, | |
| "learning_rate": 5.47774628681224e-06, | |
| "loss": 0.1169, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 19.56, | |
| "learning_rate": 5.4086953439318024e-06, | |
| "loss": 0.1179, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 19.63, | |
| "learning_rate": 5.339920685094613e-06, | |
| "loss": 0.1214, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 19.7, | |
| "learning_rate": 5.27142644885001e-06, | |
| "loss": 0.1222, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 19.78, | |
| "learning_rate": 5.203216756872746e-06, | |
| "loss": 0.1205, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 19.85, | |
| "learning_rate": 5.135295713714998e-06, | |
| "loss": 0.1186, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 19.93, | |
| "learning_rate": 5.067667406559349e-06, | |
| "loss": 0.1257, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 5.000335904972846e-06, | |
| "loss": 0.1201, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 20.07, | |
| "learning_rate": 4.933305260662111e-06, | |
| "loss": 0.117, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 20.15, | |
| "learning_rate": 4.8665795072295455e-06, | |
| "loss": 0.1169, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 20.22, | |
| "learning_rate": 4.800162659930563e-06, | |
| "loss": 0.1121, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 20.3, | |
| "learning_rate": 4.734058715432019e-06, | |
| "loss": 0.1111, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 20.37, | |
| "learning_rate": 4.668271651571691e-06, | |
| "loss": 0.1177, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 20.44, | |
| "learning_rate": 4.602805427118878e-06, | |
| "loss": 0.1091, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 20.52, | |
| "learning_rate": 4.537663981536239e-06, | |
| "loss": 0.1134, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 20.59, | |
| "learning_rate": 4.4728512347426835e-06, | |
| "loss": 0.1226, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 20.67, | |
| "learning_rate": 4.408371086877516e-06, | |
| "loss": 0.1233, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 20.74, | |
| "learning_rate": 4.344227418065728e-06, | |
| "loss": 0.121, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 20.81, | |
| "learning_rate": 4.280424088184534e-06, | |
| "loss": 0.114, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 20.89, | |
| "learning_rate": 4.216964936631054e-06, | |
| "loss": 0.1245, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 20.96, | |
| "learning_rate": 4.153853782091335e-06, | |
| "loss": 0.1194, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 21.04, | |
| "learning_rate": 4.0910944223105155e-06, | |
| "loss": 0.1161, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 21.11, | |
| "learning_rate": 4.028690633864312e-06, | |
| "loss": 0.1107, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 21.19, | |
| "learning_rate": 3.966646171931754e-06, | |
| "loss": 0.1113, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 21.26, | |
| "learning_rate": 3.904964770069233e-06, | |
| "loss": 0.114, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 21.33, | |
| "learning_rate": 3.843650139985807e-06, | |
| "loss": 0.1113, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 21.41, | |
| "learning_rate": 3.7827059713198646e-06, | |
| "loss": 0.1151, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 21.48, | |
| "learning_rate": 3.7221359314170933e-06, | |
| "loss": 0.1147, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 21.56, | |
| "learning_rate": 3.661943665109796e-06, | |
| "loss": 0.1212, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 21.63, | |
| "learning_rate": 3.602132794497558e-06, | |
| "loss": 0.1211, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 21.7, | |
| "learning_rate": 3.5427069187293007e-06, | |
| "loss": 0.1139, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 21.78, | |
| "learning_rate": 3.4836696137866755e-06, | |
| "loss": 0.1138, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 21.85, | |
| "learning_rate": 3.425024432268894e-06, | |
| "loss": 0.1219, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 21.93, | |
| "learning_rate": 3.3667749031789554e-06, | |
| "loss": 0.1186, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "learning_rate": 3.308924531711265e-06, | |
| "loss": 0.1188, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 22.07, | |
| "learning_rate": 3.251476799040725e-06, | |
| "loss": 0.1145, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 22.15, | |
| "learning_rate": 3.194435162113245e-06, | |
| "loss": 0.1152, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 22.22, | |
| "learning_rate": 3.137803053437725e-06, | |
| "loss": 0.1144, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 22.3, | |
| "learning_rate": 3.0815838808794907e-06, | |
| "loss": 0.1073, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 22.37, | |
| "learning_rate": 3.0257810274552444e-06, | |
| "loss": 0.1045, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 22.44, | |
| "learning_rate": 2.970397851129467e-06, | |
| "loss": 0.1181, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 22.52, | |
| "learning_rate": 2.9154376846123633e-06, | |
| "loss": 0.1146, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 22.59, | |
| "learning_rate": 2.86090383515931e-06, | |
| "loss": 0.1165, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 22.67, | |
| "learning_rate": 2.8067995843718544e-06, | |
| "loss": 0.1164, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 22.74, | |
| "learning_rate": 2.753128188000208e-06, | |
| "loss": 0.1131, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 22.81, | |
| "learning_rate": 2.6998928757473764e-06, | |
| "loss": 0.1138, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 22.89, | |
| "learning_rate": 2.64709685107477e-06, | |
| "loss": 0.1168, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 22.96, | |
| "learning_rate": 2.594743291009456e-06, | |
| "loss": 0.1178, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 23.04, | |
| "learning_rate": 2.54283534595297e-06, | |
| "loss": 0.1173, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 23.11, | |
| "learning_rate": 2.491376139491756e-06, | |
| "loss": 0.1119, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 23.19, | |
| "learning_rate": 2.440368768209167e-06, | |
| "loss": 0.1075, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 23.26, | |
| "learning_rate": 2.3898163014991694e-06, | |
| "loss": 0.1128, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 23.33, | |
| "learning_rate": 2.3397217813816107e-06, | |
| "loss": 0.1125, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 23.41, | |
| "learning_rate": 2.2900882223191766e-06, | |
| "loss": 0.118, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 23.48, | |
| "learning_rate": 2.240918611035988e-06, | |
| "loss": 0.107, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 23.56, | |
| "learning_rate": 2.1922159063378876e-06, | |
| "loss": 0.1201, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 23.63, | |
| "learning_rate": 2.1439830389343695e-06, | |
| "loss": 0.1083, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 23.7, | |
| "learning_rate": 2.0962229112622403e-06, | |
| "loss": 0.1152, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 23.78, | |
| "learning_rate": 2.0489383973109643e-06, | |
| "loss": 0.113, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 23.85, | |
| "learning_rate": 2.0021323424496962e-06, | |
| "loss": 0.1149, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 23.93, | |
| "learning_rate": 1.955807563256097e-06, | |
| "loss": 0.111, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "learning_rate": 1.9099668473468113e-06, | |
| "loss": 0.1164, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 24.07, | |
| "learning_rate": 1.8646129532097391e-06, | |
| "loss": 0.1129, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 24.15, | |
| "learning_rate": 1.8197486100380357e-06, | |
| "loss": 0.1089, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 24.22, | |
| "learning_rate": 1.775376517565891e-06, | |
| "loss": 0.116, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 24.3, | |
| "learning_rate": 1.7314993459060492e-06, | |
| "loss": 0.1121, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 24.37, | |
| "learning_rate": 1.6881197353891643e-06, | |
| "loss": 0.1118, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 24.44, | |
| "learning_rate": 1.645240296404892e-06, | |
| "loss": 0.11, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 24.52, | |
| "learning_rate": 1.6028636092448168e-06, | |
| "loss": 0.1134, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 24.59, | |
| "learning_rate": 1.5609922239471842e-06, | |
| "loss": 0.1079, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 24.67, | |
| "learning_rate": 1.5196286601434508e-06, | |
| "loss": 0.1128, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 24.74, | |
| "learning_rate": 1.4787754069066574e-06, | |
| "loss": 0.1102, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 24.81, | |
| "learning_rate": 1.4384349226016559e-06, | |
| "loss": 0.1159, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 24.89, | |
| "learning_rate": 1.3986096347371692e-06, | |
| "loss": 0.1145, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 24.96, | |
| "learning_rate": 1.3593019398197216e-06, | |
| "loss": 0.1113, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 25.04, | |
| "learning_rate": 1.3205142032094188e-06, | |
| "loss": 0.1113, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 25.11, | |
| "learning_rate": 1.2822487589776234e-06, | |
| "loss": 0.1088, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 25.19, | |
| "learning_rate": 1.2445079097664914e-06, | |
| "loss": 0.1127, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 25.26, | |
| "learning_rate": 1.2072939266504068e-06, | |
| "loss": 0.1066, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 25.33, | |
| "learning_rate": 1.1706090489993348e-06, | |
| "loss": 0.1193, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 25.41, | |
| "learning_rate": 1.134455484344047e-06, | |
| "loss": 0.108, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 25.48, | |
| "learning_rate": 1.098835408243295e-06, | |
| "loss": 0.1083, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 25.56, | |
| "learning_rate": 1.0637509641528876e-06, | |
| "loss": 0.112, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 25.63, | |
| "learning_rate": 1.0292042632967158e-06, | |
| "loss": 0.1157, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 25.7, | |
| "learning_rate": 9.95197384539698e-07, | |
| "loss": 0.1128, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 25.78, | |
| "learning_rate": 9.61732374262696e-07, | |
| "loss": 0.1123, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 25.85, | |
| "learning_rate": 9.288112462393605e-07, | |
| "loss": 0.1111, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 25.93, | |
| "learning_rate": 8.964359815149593e-07, | |
| "loss": 0.1194, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "learning_rate": 8.646085282871597e-07, | |
| "loss": 0.1103, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 26.07, | |
| "learning_rate": 8.33330801788812e-07, | |
| "loss": 0.113, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 26.15, | |
| "learning_rate": 8.026046841726687e-07, | |
| "loss": 0.1165, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 26.22, | |
| "learning_rate": 7.724320243981587e-07, | |
| "loss": 0.1231, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 26.3, | |
| "learning_rate": 7.42814638120104e-07, | |
| "loss": 0.1117, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 26.37, | |
| "learning_rate": 7.137543075794651e-07, | |
| "loss": 0.1102, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 26.44, | |
| "learning_rate": 6.852527814960986e-07, | |
| "loss": 0.1106, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 26.52, | |
| "learning_rate": 6.573117749635294e-07, | |
| "loss": 0.1136, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 26.59, | |
| "learning_rate": 6.299329693457279e-07, | |
| "loss": 0.1118, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 26.67, | |
| "learning_rate": 6.031180121759572e-07, | |
| "loss": 0.1072, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 26.74, | |
| "learning_rate": 5.768685170576105e-07, | |
| "loss": 0.1102, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 26.81, | |
| "learning_rate": 5.511860635671229e-07, | |
| "loss": 0.1089, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 26.89, | |
| "learning_rate": 5.260721971589211e-07, | |
| "loss": 0.1124, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 26.96, | |
| "learning_rate": 5.015284290724165e-07, | |
| "loss": 0.1043, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 27.04, | |
| "learning_rate": 4.775562362410724e-07, | |
| "loss": 0.1137, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 27.11, | |
| "learning_rate": 4.5415706120352287e-07, | |
| "loss": 0.1112, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 27.19, | |
| "learning_rate": 4.3133231201677894e-07, | |
| "loss": 0.1094, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 27.26, | |
| "learning_rate": 4.0908336217147957e-07, | |
| "loss": 0.1145, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 27.33, | |
| "learning_rate": 3.8741155050926237e-07, | |
| "loss": 0.1128, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 27.41, | |
| "learning_rate": 3.6631818114218007e-07, | |
| "loss": 0.1109, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 27.48, | |
| "learning_rate": 3.4580452337423465e-07, | |
| "loss": 0.1109, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 27.56, | |
| "learning_rate": 3.258718116249937e-07, | |
| "loss": 0.1139, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 27.63, | |
| "learning_rate": 3.0652124535531125e-07, | |
| "loss": 0.1066, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 27.7, | |
| "learning_rate": 2.8775398899514304e-07, | |
| "loss": 0.1104, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 27.78, | |
| "learning_rate": 2.695711718734839e-07, | |
| "loss": 0.1088, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 27.85, | |
| "learning_rate": 2.5197388815040634e-07, | |
| "loss": 0.1112, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 27.93, | |
| "learning_rate": 2.3496319675121892e-07, | |
| "loss": 0.1123, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "learning_rate": 2.185401213027405e-07, | |
| "loss": 0.1143, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 28.07, | |
| "learning_rate": 2.0270565007171726e-07, | |
| "loss": 0.1103, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 28.15, | |
| "learning_rate": 1.8746073590533487e-07, | |
| "loss": 0.1096, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 28.22, | |
| "learning_rate": 1.7280629617389187e-07, | |
| "loss": 0.1085, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 28.3, | |
| "learning_rate": 1.5874321271559634e-07, | |
| "loss": 0.1052, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 28.37, | |
| "learning_rate": 1.4527233178349255e-07, | |
| "loss": 0.1165, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 28.44, | |
| "learning_rate": 1.323944639945507e-07, | |
| "loss": 0.1179, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 28.52, | |
| "learning_rate": 1.2011038428087263e-07, | |
| "loss": 0.1173, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 28.59, | |
| "learning_rate": 1.0842083184307239e-07, | |
| "loss": 0.1095, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 28.67, | |
| "learning_rate": 9.732651010578408e-08, | |
| "loss": 0.1135, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 28.74, | |
| "learning_rate": 8.682808667534348e-08, | |
| "loss": 0.1166, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 28.81, | |
| "learning_rate": 7.692619329960571e-08, | |
| "loss": 0.1044, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 28.89, | |
| "learning_rate": 6.762142582993458e-08, | |
| "loss": 0.1179, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 28.96, | |
| "learning_rate": 5.8914344185344584e-08, | |
| "loss": 0.1084, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 29.04, | |
| "learning_rate": 5.0805472318806726e-08, | |
| "loss": 0.1074, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 29.11, | |
| "learning_rate": 4.329529818572265e-08, | |
| "loss": 0.1149, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 29.19, | |
| "learning_rate": 3.6384273714560324e-08, | |
| "loss": 0.1215, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 29.26, | |
| "learning_rate": 3.0072814779656956e-08, | |
| "loss": 0.1111, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 29.33, | |
| "learning_rate": 2.4361301176195617e-08, | |
| "loss": 0.1036, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 29.41, | |
| "learning_rate": 1.9250076597351343e-08, | |
| "loss": 0.1069, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 29.48, | |
| "learning_rate": 1.4739448613607655e-08, | |
| "loss": 0.1097, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 29.56, | |
| "learning_rate": 1.082968865424805e-08, | |
| "loss": 0.1123, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 29.63, | |
| "learning_rate": 7.521031991024608e-09, | |
| "loss": 0.1128, | |
| "step": 800 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 810, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 30, | |
| "save_steps": 200, | |
| "total_flos": 1.2489159543555686e+17, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |