{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "global_step": 625, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.0526315789473685e-06, "loss": 1.54, "step": 1 }, { "epoch": 0.0, "learning_rate": 2.105263157894737e-06, "loss": 1.5752, "step": 2 }, { "epoch": 0.0, "learning_rate": 3.157894736842105e-06, "loss": 1.5234, "step": 3 }, { "epoch": 0.01, "learning_rate": 4.210526315789474e-06, "loss": 1.5664, "step": 4 }, { "epoch": 0.01, "learning_rate": 5.263157894736842e-06, "loss": 1.5439, "step": 5 }, { "epoch": 0.01, "learning_rate": 6.31578947368421e-06, "loss": 1.5352, "step": 6 }, { "epoch": 0.01, "learning_rate": 7.368421052631579e-06, "loss": 1.5889, "step": 7 }, { "epoch": 0.01, "learning_rate": 8.421052631578948e-06, "loss": 1.5625, "step": 8 }, { "epoch": 0.01, "learning_rate": 9.473684210526315e-06, "loss": 1.584, "step": 9 }, { "epoch": 0.02, "learning_rate": 1.0526315789473684e-05, "loss": 1.5996, "step": 10 }, { "epoch": 0.02, "learning_rate": 1.1578947368421053e-05, "loss": 1.5752, "step": 11 }, { "epoch": 0.02, "learning_rate": 1.263157894736842e-05, "loss": 1.5566, "step": 12 }, { "epoch": 0.02, "learning_rate": 1.3684210526315791e-05, "loss": 1.6113, "step": 13 }, { "epoch": 0.02, "learning_rate": 1.4736842105263159e-05, "loss": 1.5469, "step": 14 }, { "epoch": 0.02, "learning_rate": 1.578947368421053e-05, "loss": 1.5479, "step": 15 }, { "epoch": 0.03, "learning_rate": 1.6842105263157896e-05, "loss": 1.5498, "step": 16 }, { "epoch": 0.03, "learning_rate": 1.7894736842105264e-05, "loss": 1.5293, "step": 17 }, { "epoch": 0.03, "learning_rate": 1.894736842105263e-05, "loss": 1.541, "step": 18 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 1.5439, "step": 19 }, { "epoch": 0.03, "learning_rate": 1.9999865623437014e-05, "loss": 1.5605, "step": 20 }, { "epoch": 0.03, "learning_rate": 1.9999462497359468e-05, "loss": 1.5244, "step": 21 }, { "epoch": 0.04, "learning_rate": 1.9998790632601496e-05, "loss": 1.5303, "step": 22 }, { "epoch": 0.04, "learning_rate": 1.999785004721968e-05, "loss": 1.4805, "step": 23 }, { "epoch": 0.04, "learning_rate": 1.9996640766492542e-05, "loss": 1.4893, "step": 24 }, { "epoch": 0.04, "learning_rate": 1.999516282291988e-05, "loss": 1.457, "step": 25 }, { "epoch": 0.04, "learning_rate": 1.9993416256221894e-05, "loss": 1.4189, "step": 26 }, { "epoch": 0.04, "learning_rate": 1.9991401113338103e-05, "loss": 1.4014, "step": 27 }, { "epoch": 0.04, "learning_rate": 1.998911744842611e-05, "loss": 1.3926, "step": 28 }, { "epoch": 0.05, "learning_rate": 1.9986565322860117e-05, "loss": 1.3848, "step": 29 }, { "epoch": 0.05, "learning_rate": 1.9983744805229296e-05, "loss": 1.3496, "step": 30 }, { "epoch": 0.05, "learning_rate": 1.9980655971335944e-05, "loss": 1.3369, "step": 31 }, { "epoch": 0.05, "learning_rate": 1.9977298904193438e-05, "loss": 1.3457, "step": 32 }, { "epoch": 0.05, "learning_rate": 1.9973673694024002e-05, "loss": 1.3086, "step": 33 }, { "epoch": 0.05, "learning_rate": 1.9969780438256295e-05, "loss": 1.3291, "step": 34 }, { "epoch": 0.06, "learning_rate": 1.996561924152278e-05, "loss": 1.3193, "step": 35 }, { "epoch": 0.06, "learning_rate": 1.996119021565693e-05, "loss": 1.2842, "step": 36 }, { "epoch": 0.06, "learning_rate": 1.995649347969019e-05, "loss": 1.2822, "step": 37 }, { "epoch": 0.06, "learning_rate": 1.9951529159848805e-05, "loss": 1.2939, "step": 38 }, { "epoch": 0.06, "learning_rate": 1.9946297389550433e-05, "loss": 1.2754, "step": 39 }, { "epoch": 0.06, "learning_rate": 1.9940798309400527e-05, "loss": 1.249, "step": 40 }, { "epoch": 0.07, "learning_rate": 1.993503206718859e-05, "loss": 1.2852, "step": 41 }, { "epoch": 0.07, "learning_rate": 1.9928998817884185e-05, "loss": 1.2715, "step": 42 }, { "epoch": 0.07, "learning_rate": 1.992269872363277e-05, "loss": 1.2695, "step": 43 }, { "epoch": 0.07, "learning_rate": 1.9916131953751342e-05, "loss": 1.2598, "step": 44 }, { "epoch": 0.07, "learning_rate": 1.9909298684723905e-05, "loss": 1.2637, "step": 45 }, { "epoch": 0.07, "learning_rate": 1.9902199100196697e-05, "loss": 1.2295, "step": 46 }, { "epoch": 0.08, "learning_rate": 1.9894833390973266e-05, "loss": 1.2393, "step": 47 }, { "epoch": 0.08, "learning_rate": 1.9887201755009358e-05, "loss": 1.2178, "step": 48 }, { "epoch": 0.08, "learning_rate": 1.987930439740757e-05, "loss": 1.249, "step": 49 }, { "epoch": 0.08, "learning_rate": 1.9871141530411854e-05, "loss": 1.2324, "step": 50 }, { "epoch": 0.08, "learning_rate": 1.986271337340182e-05, "loss": 1.1943, "step": 51 }, { "epoch": 0.08, "learning_rate": 1.9854020152886816e-05, "loss": 1.2002, "step": 52 }, { "epoch": 0.08, "learning_rate": 1.984506210249986e-05, "loss": 1.2285, "step": 53 }, { "epoch": 0.09, "learning_rate": 1.983583946299136e-05, "loss": 1.2002, "step": 54 }, { "epoch": 0.09, "learning_rate": 1.982635248222264e-05, "loss": 1.2061, "step": 55 }, { "epoch": 0.09, "learning_rate": 1.9816601415159266e-05, "loss": 1.2227, "step": 56 }, { "epoch": 0.09, "learning_rate": 1.9806586523864212e-05, "loss": 1.2471, "step": 57 }, { "epoch": 0.09, "learning_rate": 1.9796308077490817e-05, "loss": 1.2266, "step": 58 }, { "epoch": 0.09, "learning_rate": 1.978576635227554e-05, "loss": 1.1816, "step": 59 }, { "epoch": 0.1, "learning_rate": 1.9774961631530543e-05, "loss": 1.21, "step": 60 }, { "epoch": 0.1, "learning_rate": 1.976389420563607e-05, "loss": 1.1924, "step": 61 }, { "epoch": 0.1, "learning_rate": 1.9752564372032655e-05, "loss": 1.1543, "step": 62 }, { "epoch": 0.1, "learning_rate": 1.9740972435213114e-05, "loss": 1.1865, "step": 63 }, { "epoch": 0.1, "learning_rate": 1.9729118706714377e-05, "loss": 1.21, "step": 64 }, { "epoch": 0.1, "learning_rate": 1.9717003505109097e-05, "loss": 1.1982, "step": 65 }, { "epoch": 0.11, "learning_rate": 1.970462715599711e-05, "loss": 1.1729, "step": 66 }, { "epoch": 0.11, "learning_rate": 1.9691989991996663e-05, "loss": 1.1709, "step": 67 }, { "epoch": 0.11, "learning_rate": 1.967909235273549e-05, "loss": 1.1855, "step": 68 }, { "epoch": 0.11, "learning_rate": 1.966593458484168e-05, "loss": 1.1357, "step": 69 }, { "epoch": 0.11, "learning_rate": 1.9652517041934357e-05, "loss": 1.1387, "step": 70 }, { "epoch": 0.11, "learning_rate": 1.9638840084614182e-05, "loss": 1.1709, "step": 71 }, { "epoch": 0.12, "learning_rate": 1.9624904080453656e-05, "loss": 1.167, "step": 72 }, { "epoch": 0.12, "learning_rate": 1.9610709403987248e-05, "loss": 1.2031, "step": 73 }, { "epoch": 0.12, "learning_rate": 1.9596256436701324e-05, "loss": 1.1895, "step": 74 }, { "epoch": 0.12, "learning_rate": 1.95815455670239e-05, "loss": 1.1533, "step": 75 }, { "epoch": 0.12, "learning_rate": 1.95665771903142e-05, "loss": 1.1641, "step": 76 }, { "epoch": 0.12, "learning_rate": 1.955135170885202e-05, "loss": 1.1436, "step": 77 }, { "epoch": 0.12, "learning_rate": 1.9535869531826938e-05, "loss": 1.1748, "step": 78 }, { "epoch": 0.13, "learning_rate": 1.95201310753273e-05, "loss": 1.1562, "step": 79 }, { "epoch": 0.13, "learning_rate": 1.9504136762329046e-05, "loss": 1.1904, "step": 80 }, { "epoch": 0.13, "learning_rate": 1.9487887022684336e-05, "loss": 1.1855, "step": 81 }, { "epoch": 0.13, "learning_rate": 1.9471382293110004e-05, "loss": 1.1436, "step": 82 }, { "epoch": 0.13, "learning_rate": 1.9454623017175814e-05, "loss": 1.1885, "step": 83 }, { "epoch": 0.13, "learning_rate": 1.943760964529255e-05, "loss": 1.1318, "step": 84 }, { "epoch": 0.14, "learning_rate": 1.9420342634699893e-05, "loss": 1.1689, "step": 85 }, { "epoch": 0.14, "learning_rate": 1.9402822449454154e-05, "loss": 1.1689, "step": 86 }, { "epoch": 0.14, "learning_rate": 1.9385049560415794e-05, "loss": 1.1484, "step": 87 }, { "epoch": 0.14, "learning_rate": 1.936702444523675e-05, "loss": 1.1211, "step": 88 }, { "epoch": 0.14, "learning_rate": 1.9348747588347637e-05, "loss": 1.1494, "step": 89 }, { "epoch": 0.14, "learning_rate": 1.9330219480944693e-05, "loss": 1.1426, "step": 90 }, { "epoch": 0.15, "learning_rate": 1.9311440620976597e-05, "loss": 1.1787, "step": 91 }, { "epoch": 0.15, "learning_rate": 1.929241151313108e-05, "loss": 1.2158, "step": 92 }, { "epoch": 0.15, "learning_rate": 1.9273132668821363e-05, "loss": 1.167, "step": 93 }, { "epoch": 0.15, "learning_rate": 1.925360460617242e-05, "loss": 1.1533, "step": 94 }, { "epoch": 0.15, "learning_rate": 1.9233827850007028e-05, "loss": 1.1377, "step": 95 }, { "epoch": 0.15, "learning_rate": 1.9213802931831697e-05, "loss": 1.1553, "step": 96 }, { "epoch": 0.16, "learning_rate": 1.9193530389822364e-05, "loss": 1.1777, "step": 97 }, { "epoch": 0.16, "learning_rate": 1.9173010768809934e-05, "loss": 1.1494, "step": 98 }, { "epoch": 0.16, "learning_rate": 1.915224462026563e-05, "loss": 1.1211, "step": 99 }, { "epoch": 0.16, "learning_rate": 1.913123250228619e-05, "loss": 1.1709, "step": 100 }, { "epoch": 0.16, "learning_rate": 1.9109974979578852e-05, "loss": 1.1465, "step": 101 }, { "epoch": 0.16, "learning_rate": 1.9088472623446182e-05, "loss": 1.1465, "step": 102 }, { "epoch": 0.16, "learning_rate": 1.9066726011770725e-05, "loss": 1.1318, "step": 103 }, { "epoch": 0.17, "learning_rate": 1.9044735728999472e-05, "loss": 1.1172, "step": 104 }, { "epoch": 0.17, "learning_rate": 1.9022502366128136e-05, "loss": 1.1348, "step": 105 }, { "epoch": 0.17, "learning_rate": 1.90000265206853e-05, "loss": 1.1113, "step": 106 }, { "epoch": 0.17, "learning_rate": 1.897730879671634e-05, "loss": 1.127, "step": 107 }, { "epoch": 0.17, "learning_rate": 1.8954349804767185e-05, "loss": 1.1182, "step": 108 }, { "epoch": 0.17, "learning_rate": 1.8931150161867917e-05, "loss": 1.1328, "step": 109 }, { "epoch": 0.18, "learning_rate": 1.8907710491516197e-05, "loss": 1.1445, "step": 110 }, { "epoch": 0.18, "learning_rate": 1.8884031423660492e-05, "loss": 1.1201, "step": 111 }, { "epoch": 0.18, "learning_rate": 1.8860113594683148e-05, "loss": 1.1406, "step": 112 }, { "epoch": 0.18, "learning_rate": 1.8835957647383304e-05, "loss": 1.1113, "step": 113 }, { "epoch": 0.18, "learning_rate": 1.8811564230959585e-05, "loss": 1.125, "step": 114 }, { "epoch": 0.18, "learning_rate": 1.878693400099269e-05, "loss": 1.1504, "step": 115 }, { "epoch": 0.19, "learning_rate": 1.8762067619427745e-05, "loss": 1.1191, "step": 116 }, { "epoch": 0.19, "learning_rate": 1.8736965754556527e-05, "loss": 1.1387, "step": 117 }, { "epoch": 0.19, "learning_rate": 1.8711629080999506e-05, "loss": 1.127, "step": 118 }, { "epoch": 0.19, "learning_rate": 1.86860582796877e-05, "loss": 1.1162, "step": 119 }, { "epoch": 0.19, "learning_rate": 1.866025403784439e-05, "loss": 1.1016, "step": 120 }, { "epoch": 0.19, "learning_rate": 1.8634217048966638e-05, "loss": 1.1572, "step": 121 }, { "epoch": 0.2, "learning_rate": 1.8607948012806664e-05, "loss": 1.1377, "step": 122 }, { "epoch": 0.2, "learning_rate": 1.858144763535302e-05, "loss": 1.1328, "step": 123 }, { "epoch": 0.2, "learning_rate": 1.855471662881164e-05, "loss": 1.1133, "step": 124 }, { "epoch": 0.2, "learning_rate": 1.852775571158668e-05, "loss": 1.1045, "step": 125 }, { "epoch": 0.2, "learning_rate": 1.8500565608261215e-05, "loss": 1.1621, "step": 126 }, { "epoch": 0.2, "learning_rate": 1.8473147049577777e-05, "loss": 1.1211, "step": 127 }, { "epoch": 0.2, "learning_rate": 1.8445500772418697e-05, "loss": 1.1172, "step": 128 }, { "epoch": 0.21, "learning_rate": 1.8417627519786317e-05, "loss": 1.1211, "step": 129 }, { "epoch": 0.21, "learning_rate": 1.8389528040783014e-05, "loss": 1.1367, "step": 130 }, { "epoch": 0.21, "learning_rate": 1.836120309059107e-05, "loss": 1.1543, "step": 131 }, { "epoch": 0.21, "learning_rate": 1.8332653430452375e-05, "loss": 1.1406, "step": 132 }, { "epoch": 0.21, "learning_rate": 1.8303879827647977e-05, "loss": 1.1348, "step": 133 }, { "epoch": 0.21, "learning_rate": 1.8274883055477436e-05, "loss": 1.127, "step": 134 }, { "epoch": 0.22, "learning_rate": 1.8245663893238075e-05, "loss": 1.1279, "step": 135 }, { "epoch": 0.22, "learning_rate": 1.821622312620401e-05, "loss": 1.1094, "step": 136 }, { "epoch": 0.22, "learning_rate": 1.8186561545605055e-05, "loss": 1.1436, "step": 137 }, { "epoch": 0.22, "learning_rate": 1.815667994860547e-05, "loss": 1.1436, "step": 138 }, { "epoch": 0.22, "learning_rate": 1.8126579138282502e-05, "loss": 1.1035, "step": 139 }, { "epoch": 0.22, "learning_rate": 1.809625992360485e-05, "loss": 1.166, "step": 140 }, { "epoch": 0.23, "learning_rate": 1.8065723119410885e-05, "loss": 1.0996, "step": 141 }, { "epoch": 0.23, "learning_rate": 1.803496954638676e-05, "loss": 1.0771, "step": 142 }, { "epoch": 0.23, "learning_rate": 1.8004000031044363e-05, "loss": 1.1416, "step": 143 }, { "epoch": 0.23, "learning_rate": 1.7972815405699105e-05, "loss": 1.1455, "step": 144 }, { "epoch": 0.23, "learning_rate": 1.7941416508447537e-05, "loss": 1.1201, "step": 145 }, { "epoch": 0.23, "learning_rate": 1.7909804183144837e-05, "loss": 1.1367, "step": 146 }, { "epoch": 0.24, "learning_rate": 1.7877979279382135e-05, "loss": 1.1211, "step": 147 }, { "epoch": 0.24, "learning_rate": 1.784594265246366e-05, "loss": 1.0889, "step": 148 }, { "epoch": 0.24, "learning_rate": 1.781369516338378e-05, "loss": 1.0815, "step": 149 }, { "epoch": 0.24, "learning_rate": 1.7781237678803845e-05, "loss": 1.1191, "step": 150 }, { "epoch": 0.24, "learning_rate": 1.77485710710289e-05, "loss": 1.1221, "step": 151 }, { "epoch": 0.24, "learning_rate": 1.7715696217984233e-05, "loss": 1.0879, "step": 152 }, { "epoch": 0.24, "learning_rate": 1.7682614003191807e-05, "loss": 1.1006, "step": 153 }, { "epoch": 0.25, "learning_rate": 1.764932531574648e-05, "loss": 1.1201, "step": 154 }, { "epoch": 0.25, "learning_rate": 1.761583105029213e-05, "loss": 1.1191, "step": 155 }, { "epoch": 0.25, "learning_rate": 1.7582132106997615e-05, "loss": 1.1152, "step": 156 }, { "epoch": 0.25, "learning_rate": 1.7548229391532572e-05, "loss": 1.1045, "step": 157 }, { "epoch": 0.25, "learning_rate": 1.7514123815043073e-05, "loss": 1.0869, "step": 158 }, { "epoch": 0.25, "learning_rate": 1.747981629412715e-05, "loss": 1.082, "step": 159 }, { "epoch": 0.26, "learning_rate": 1.7445307750810153e-05, "loss": 1.1094, "step": 160 }, { "epoch": 0.26, "learning_rate": 1.741059911251997e-05, "loss": 1.1113, "step": 161 }, { "epoch": 0.26, "learning_rate": 1.7375691312062102e-05, "loss": 1.0889, "step": 162 }, { "epoch": 0.26, "learning_rate": 1.7340585287594605e-05, "loss": 1.1152, "step": 163 }, { "epoch": 0.26, "learning_rate": 1.730528198260285e-05, "loss": 1.0938, "step": 164 }, { "epoch": 0.26, "learning_rate": 1.7269782345874204e-05, "loss": 1.0869, "step": 165 }, { "epoch": 0.27, "learning_rate": 1.72340873314725e-05, "loss": 1.0625, "step": 166 }, { "epoch": 0.27, "learning_rate": 1.7198197898712402e-05, "loss": 1.1025, "step": 167 }, { "epoch": 0.27, "learning_rate": 1.7162115012133643e-05, "loss": 1.124, "step": 168 }, { "epoch": 0.27, "learning_rate": 1.7125839641475074e-05, "loss": 1.0825, "step": 169 }, { "epoch": 0.27, "learning_rate": 1.7089372761648617e-05, "loss": 1.0908, "step": 170 }, { "epoch": 0.27, "learning_rate": 1.7052715352713076e-05, "loss": 1.0635, "step": 171 }, { "epoch": 0.28, "learning_rate": 1.7015868399847768e-05, "loss": 1.1143, "step": 172 }, { "epoch": 0.28, "learning_rate": 1.6978832893326074e-05, "loss": 1.1182, "step": 173 }, { "epoch": 0.28, "learning_rate": 1.6941609828488806e-05, "loss": 1.0957, "step": 174 }, { "epoch": 0.28, "learning_rate": 1.690420020571747e-05, "loss": 1.0977, "step": 175 }, { "epoch": 0.28, "learning_rate": 1.686660503040737e-05, "loss": 1.1104, "step": 176 }, { "epoch": 0.28, "learning_rate": 1.6828825312940594e-05, "loss": 1.1045, "step": 177 }, { "epoch": 0.28, "learning_rate": 1.6790862068658863e-05, "loss": 1.1074, "step": 178 }, { "epoch": 0.29, "learning_rate": 1.675271631783623e-05, "loss": 1.1094, "step": 179 }, { "epoch": 0.29, "learning_rate": 1.671438908565167e-05, "loss": 1.0977, "step": 180 }, { "epoch": 0.29, "learning_rate": 1.667588140216154e-05, "loss": 1.1025, "step": 181 }, { "epoch": 0.29, "learning_rate": 1.663719430227186e-05, "loss": 1.0908, "step": 182 }, { "epoch": 0.29, "learning_rate": 1.6598328825710536e-05, "loss": 1.0684, "step": 183 }, { "epoch": 0.29, "learning_rate": 1.65592860169994e-05, "loss": 1.1289, "step": 184 }, { "epoch": 0.3, "learning_rate": 1.6520066925426146e-05, "loss": 1.0908, "step": 185 }, { "epoch": 0.3, "learning_rate": 1.648067260501611e-05, "loss": 1.0771, "step": 186 }, { "epoch": 0.3, "learning_rate": 1.644110411450398e-05, "loss": 1.0708, "step": 187 }, { "epoch": 0.3, "learning_rate": 1.6401362517305296e-05, "loss": 1.0928, "step": 188 }, { "epoch": 0.3, "learning_rate": 1.6361448881487913e-05, "loss": 1.0898, "step": 189 }, { "epoch": 0.3, "learning_rate": 1.6321364279743267e-05, "loss": 1.0674, "step": 190 }, { "epoch": 0.31, "learning_rate": 1.628110978935756e-05, "loss": 1.1162, "step": 191 }, { "epoch": 0.31, "learning_rate": 1.6240686492182806e-05, "loss": 1.0869, "step": 192 }, { "epoch": 0.31, "learning_rate": 1.6200095474607753e-05, "loss": 1.0908, "step": 193 }, { "epoch": 0.31, "learning_rate": 1.6159337827528686e-05, "loss": 1.0874, "step": 194 }, { "epoch": 0.31, "learning_rate": 1.6118414646320115e-05, "loss": 1.0908, "step": 195 }, { "epoch": 0.31, "learning_rate": 1.6077327030805318e-05, "loss": 1.0693, "step": 196 }, { "epoch": 0.32, "learning_rate": 1.6036076085226813e-05, "loss": 1.1025, "step": 197 }, { "epoch": 0.32, "learning_rate": 1.599466291821666e-05, "loss": 1.0918, "step": 198 }, { "epoch": 0.32, "learning_rate": 1.595308864276666e-05, "loss": 1.0752, "step": 199 }, { "epoch": 0.32, "learning_rate": 1.5911354376198468e-05, "loss": 1.0762, "step": 200 }, { "epoch": 0.32, "learning_rate": 1.586946124013354e-05, "loss": 1.1104, "step": 201 }, { "epoch": 0.32, "learning_rate": 1.582741036046301e-05, "loss": 1.0771, "step": 202 }, { "epoch": 0.32, "learning_rate": 1.578520286731741e-05, "loss": 1.1152, "step": 203 }, { "epoch": 0.33, "learning_rate": 1.5742839895036305e-05, "loss": 1.0889, "step": 204 }, { "epoch": 0.33, "learning_rate": 1.570032258213783e-05, "loss": 1.0771, "step": 205 }, { "epoch": 0.33, "learning_rate": 1.565765207128805e-05, "loss": 1.0977, "step": 206 }, { "epoch": 0.33, "learning_rate": 1.561482950927029e-05, "loss": 1.1006, "step": 207 }, { "epoch": 0.33, "learning_rate": 1.5571856046954284e-05, "loss": 1.0884, "step": 208 }, { "epoch": 0.33, "learning_rate": 1.5528732839265272e-05, "loss": 1.1006, "step": 209 }, { "epoch": 0.34, "learning_rate": 1.5485461045152937e-05, "loss": 1.0859, "step": 210 }, { "epoch": 0.34, "learning_rate": 1.5442041827560274e-05, "loss": 1.0713, "step": 211 }, { "epoch": 0.34, "learning_rate": 1.5398476353392323e-05, "loss": 1.0918, "step": 212 }, { "epoch": 0.34, "learning_rate": 1.5354765793484834e-05, "loss": 1.0928, "step": 213 }, { "epoch": 0.34, "learning_rate": 1.531091132257275e-05, "loss": 1.0967, "step": 214 }, { "epoch": 0.34, "learning_rate": 1.52669141192587e-05, "loss": 1.1191, "step": 215 }, { "epoch": 0.35, "learning_rate": 1.5222775365981272e-05, "loss": 1.0771, "step": 216 }, { "epoch": 0.35, "learning_rate": 1.5178496248983254e-05, "loss": 1.1025, "step": 217 }, { "epoch": 0.35, "learning_rate": 1.5134077958279764e-05, "loss": 1.1016, "step": 218 }, { "epoch": 0.35, "learning_rate": 1.5089521687626243e-05, "loss": 1.1182, "step": 219 }, { "epoch": 0.35, "learning_rate": 1.50448286344864e-05, "loss": 1.1201, "step": 220 }, { "epoch": 0.35, "learning_rate": 1.5000000000000002e-05, "loss": 1.0879, "step": 221 }, { "epoch": 0.36, "learning_rate": 1.4955036988950617e-05, "loss": 1.1045, "step": 222 }, { "epoch": 0.36, "learning_rate": 1.4909940809733223e-05, "loss": 1.0996, "step": 223 }, { "epoch": 0.36, "learning_rate": 1.4864712674321733e-05, "loss": 1.0703, "step": 224 }, { "epoch": 0.36, "learning_rate": 1.4819353798236427e-05, "loss": 1.0762, "step": 225 }, { "epoch": 0.36, "learning_rate": 1.477386540051127e-05, "loss": 1.085, "step": 226 }, { "epoch": 0.36, "learning_rate": 1.4728248703661183e-05, "loss": 1.0654, "step": 227 }, { "epoch": 0.36, "learning_rate": 1.4682504933649144e-05, "loss": 1.1191, "step": 228 }, { "epoch": 0.37, "learning_rate": 1.4636635319853274e-05, "loss": 1.0811, "step": 229 }, { "epoch": 0.37, "learning_rate": 1.4590641095033786e-05, "loss": 1.0742, "step": 230 }, { "epoch": 0.37, "learning_rate": 1.4544523495299843e-05, "loss": 1.0713, "step": 231 }, { "epoch": 0.37, "learning_rate": 1.4498283760076362e-05, "loss": 1.084, "step": 232 }, { "epoch": 0.37, "learning_rate": 1.445192313207067e-05, "loss": 1.0684, "step": 233 }, { "epoch": 0.37, "learning_rate": 1.4405442857239151e-05, "loss": 1.084, "step": 234 }, { "epoch": 0.38, "learning_rate": 1.4358844184753713e-05, "loss": 1.0947, "step": 235 }, { "epoch": 0.38, "learning_rate": 1.4312128366968244e-05, "loss": 1.1084, "step": 236 }, { "epoch": 0.38, "learning_rate": 1.4265296659384956e-05, "loss": 1.0908, "step": 237 }, { "epoch": 0.38, "learning_rate": 1.4218350320620625e-05, "loss": 1.1123, "step": 238 }, { "epoch": 0.38, "learning_rate": 1.4171290612372781e-05, "loss": 1.0664, "step": 239 }, { "epoch": 0.38, "learning_rate": 1.4124118799385797e-05, "loss": 1.0918, "step": 240 }, { "epoch": 0.39, "learning_rate": 1.4076836149416889e-05, "loss": 1.0938, "step": 241 }, { "epoch": 0.39, "learning_rate": 1.4029443933202059e-05, "loss": 1.0752, "step": 242 }, { "epoch": 0.39, "learning_rate": 1.3981943424421932e-05, "loss": 1.0957, "step": 243 }, { "epoch": 0.39, "learning_rate": 1.3934335899667526e-05, "loss": 1.0713, "step": 244 }, { "epoch": 0.39, "learning_rate": 1.3886622638405953e-05, "loss": 1.0854, "step": 245 }, { "epoch": 0.39, "learning_rate": 1.3838804922946027e-05, "loss": 1.1025, "step": 246 }, { "epoch": 0.4, "learning_rate": 1.3790884038403796e-05, "loss": 1.0654, "step": 247 }, { "epoch": 0.4, "learning_rate": 1.374286127266801e-05, "loss": 1.0996, "step": 248 }, { "epoch": 0.4, "learning_rate": 1.3694737916365517e-05, "loss": 1.1001, "step": 249 }, { "epoch": 0.4, "learning_rate": 1.3646515262826551e-05, "loss": 1.0996, "step": 250 }, { "epoch": 0.4, "learning_rate": 1.3598194608050011e-05, "loss": 1.0811, "step": 251 }, { "epoch": 0.4, "learning_rate": 1.354977725066859e-05, "loss": 1.1123, "step": 252 }, { "epoch": 0.4, "learning_rate": 1.3501264491913909e-05, "loss": 1.1006, "step": 253 }, { "epoch": 0.41, "learning_rate": 1.3452657635581521e-05, "loss": 1.0859, "step": 254 }, { "epoch": 0.41, "learning_rate": 1.3403957987995884e-05, "loss": 1.0938, "step": 255 }, { "epoch": 0.41, "learning_rate": 1.335516685797525e-05, "loss": 1.0762, "step": 256 }, { "epoch": 0.41, "learning_rate": 1.3306285556796494e-05, "loss": 1.0957, "step": 257 }, { "epoch": 0.41, "learning_rate": 1.3257315398159865e-05, "loss": 1.0444, "step": 258 }, { "epoch": 0.41, "learning_rate": 1.3208257698153677e-05, "loss": 1.0854, "step": 259 }, { "epoch": 0.42, "learning_rate": 1.3159113775218963e-05, "loss": 1.0605, "step": 260 }, { "epoch": 0.42, "learning_rate": 1.3109884950114007e-05, "loss": 1.0371, "step": 261 }, { "epoch": 0.42, "learning_rate": 1.3060572545878875e-05, "loss": 1.0854, "step": 262 }, { "epoch": 0.42, "learning_rate": 1.3011177887799846e-05, "loss": 1.0576, "step": 263 }, { "epoch": 0.42, "learning_rate": 1.2961702303373795e-05, "loss": 1.0664, "step": 264 }, { "epoch": 0.42, "learning_rate": 1.2912147122272523e-05, "loss": 1.0762, "step": 265 }, { "epoch": 0.43, "learning_rate": 1.2862513676307009e-05, "loss": 1.0859, "step": 266 }, { "epoch": 0.43, "learning_rate": 1.2812803299391629e-05, "loss": 1.0508, "step": 267 }, { "epoch": 0.43, "learning_rate": 1.2763017327508304e-05, "loss": 1.1104, "step": 268 }, { "epoch": 0.43, "learning_rate": 1.271315709867059e-05, "loss": 1.0928, "step": 269 }, { "epoch": 0.43, "learning_rate": 1.2663223952887724e-05, "loss": 1.1001, "step": 270 }, { "epoch": 0.43, "learning_rate": 1.2613219232128608e-05, "loss": 1.0688, "step": 271 }, { "epoch": 0.44, "learning_rate": 1.2563144280285742e-05, "loss": 1.0654, "step": 272 }, { "epoch": 0.44, "learning_rate": 1.2513000443139112e-05, "loss": 1.0596, "step": 273 }, { "epoch": 0.44, "learning_rate": 1.2462789068320016e-05, "loss": 1.1016, "step": 274 }, { "epoch": 0.44, "learning_rate": 1.2412511505274845e-05, "loss": 1.0791, "step": 275 }, { "epoch": 0.44, "learning_rate": 1.2362169105228828e-05, "loss": 1.0488, "step": 276 }, { "epoch": 0.44, "learning_rate": 1.23117632211497e-05, "loss": 1.0889, "step": 277 }, { "epoch": 0.44, "learning_rate": 1.2261295207711347e-05, "loss": 1.0645, "step": 278 }, { "epoch": 0.45, "learning_rate": 1.221076642125742e-05, "loss": 1.0605, "step": 279 }, { "epoch": 0.45, "learning_rate": 1.2160178219764838e-05, "loss": 1.1123, "step": 280 }, { "epoch": 0.45, "learning_rate": 1.2109531962807333e-05, "loss": 1.0811, "step": 281 }, { "epoch": 0.45, "learning_rate": 1.2058829011518896e-05, "loss": 1.0776, "step": 282 }, { "epoch": 0.45, "learning_rate": 1.2008070728557186e-05, "loss": 1.0576, "step": 283 }, { "epoch": 0.45, "learning_rate": 1.195725847806693e-05, "loss": 1.0723, "step": 284 }, { "epoch": 0.46, "learning_rate": 1.1906393625643244e-05, "loss": 1.0742, "step": 285 }, { "epoch": 0.46, "learning_rate": 1.1855477538294934e-05, "loss": 1.0898, "step": 286 }, { "epoch": 0.46, "learning_rate": 1.1804511584407763e-05, "loss": 1.0669, "step": 287 }, { "epoch": 0.46, "learning_rate": 1.1753497133707678e-05, "loss": 1.0566, "step": 288 }, { "epoch": 0.46, "learning_rate": 1.1702435557223988e-05, "loss": 1.1211, "step": 289 }, { "epoch": 0.46, "learning_rate": 1.1651328227252516e-05, "loss": 1.085, "step": 290 }, { "epoch": 0.47, "learning_rate": 1.1600176517318742e-05, "loss": 1.082, "step": 291 }, { "epoch": 0.47, "learning_rate": 1.1548981802140849e-05, "loss": 1.1201, "step": 292 }, { "epoch": 0.47, "learning_rate": 1.1497745457592817e-05, "loss": 1.0898, "step": 293 }, { "epoch": 0.47, "learning_rate": 1.1446468860667422e-05, "loss": 1.0679, "step": 294 }, { "epoch": 0.47, "learning_rate": 1.1395153389439232e-05, "loss": 1.0801, "step": 295 }, { "epoch": 0.47, "learning_rate": 1.1343800423027583e-05, "loss": 1.1113, "step": 296 }, { "epoch": 0.48, "learning_rate": 1.129241134155949e-05, "loss": 1.1162, "step": 297 }, { "epoch": 0.48, "learning_rate": 1.1240987526132595e-05, "loss": 1.0503, "step": 298 }, { "epoch": 0.48, "learning_rate": 1.1189530358778005e-05, "loss": 1.0664, "step": 299 }, { "epoch": 0.48, "learning_rate": 1.1138041222423177e-05, "loss": 1.0444, "step": 300 }, { "epoch": 0.48, "learning_rate": 1.1086521500854746e-05, "loss": 1.0625, "step": 301 }, { "epoch": 0.48, "learning_rate": 1.1034972578681338e-05, "loss": 1.0439, "step": 302 }, { "epoch": 0.48, "learning_rate": 1.0983395841296349e-05, "loss": 1.0767, "step": 303 }, { "epoch": 0.49, "learning_rate": 1.0931792674840718e-05, "loss": 1.0791, "step": 304 }, { "epoch": 0.49, "learning_rate": 1.0880164466165675e-05, "loss": 1.0635, "step": 305 }, { "epoch": 0.49, "learning_rate": 1.0828512602795462e-05, "loss": 1.0786, "step": 306 }, { "epoch": 0.49, "learning_rate": 1.0776838472890065e-05, "loss": 1.0771, "step": 307 }, { "epoch": 0.49, "learning_rate": 1.0725143465207868e-05, "loss": 1.0664, "step": 308 }, { "epoch": 0.49, "learning_rate": 1.0673428969068365e-05, "loss": 1.0454, "step": 309 }, { "epoch": 0.5, "learning_rate": 1.0621696374314807e-05, "loss": 1.0732, "step": 310 }, { "epoch": 0.5, "learning_rate": 1.0569947071276847e-05, "loss": 1.0645, "step": 311 }, { "epoch": 0.5, "learning_rate": 1.0518182450733185e-05, "loss": 1.0664, "step": 312 }, { "epoch": 0.5, "learning_rate": 1.0466403903874176e-05, "loss": 1.083, "step": 313 }, { "epoch": 0.5, "learning_rate": 1.0414612822264457e-05, "loss": 1.082, "step": 314 }, { "epoch": 0.5, "learning_rate": 1.0362810597805526e-05, "loss": 1.0898, "step": 315 }, { "epoch": 0.51, "learning_rate": 1.031099862269837e-05, "loss": 1.0757, "step": 316 }, { "epoch": 0.51, "learning_rate": 1.0259178289406011e-05, "loss": 1.083, "step": 317 }, { "epoch": 0.51, "learning_rate": 1.0207350990616107e-05, "loss": 1.0479, "step": 318 }, { "epoch": 0.51, "learning_rate": 1.0155518119203511e-05, "loss": 1.082, "step": 319 }, { "epoch": 0.51, "learning_rate": 1.0103681068192845e-05, "loss": 1.0703, "step": 320 }, { "epoch": 0.51, "learning_rate": 1.0051841230721065e-05, "loss": 1.0889, "step": 321 }, { "epoch": 0.52, "learning_rate": 1e-05, "loss": 1.0806, "step": 322 }, { "epoch": 0.52, "learning_rate": 9.948158769278939e-06, "loss": 1.0928, "step": 323 }, { "epoch": 0.52, "learning_rate": 9.896318931807155e-06, "loss": 1.0947, "step": 324 }, { "epoch": 0.52, "learning_rate": 9.844481880796492e-06, "loss": 1.0957, "step": 325 }, { "epoch": 0.52, "learning_rate": 9.7926490093839e-06, "loss": 1.0327, "step": 326 }, { "epoch": 0.52, "learning_rate": 9.740821710593989e-06, "loss": 1.0723, "step": 327 }, { "epoch": 0.52, "learning_rate": 9.689001377301634e-06, "loss": 1.0752, "step": 328 }, { "epoch": 0.53, "learning_rate": 9.637189402194477e-06, "loss": 1.0425, "step": 329 }, { "epoch": 0.53, "learning_rate": 9.585387177735548e-06, "loss": 1.0552, "step": 330 }, { "epoch": 0.53, "learning_rate": 9.533596096125826e-06, "loss": 1.0645, "step": 331 }, { "epoch": 0.53, "learning_rate": 9.481817549266817e-06, "loss": 1.0811, "step": 332 }, { "epoch": 0.53, "learning_rate": 9.430052928723153e-06, "loss": 1.0723, "step": 333 }, { "epoch": 0.53, "learning_rate": 9.378303625685196e-06, "loss": 1.0791, "step": 334 }, { "epoch": 0.54, "learning_rate": 9.326571030931636e-06, "loss": 1.0767, "step": 335 }, { "epoch": 0.54, "learning_rate": 9.274856534792138e-06, "loss": 1.0586, "step": 336 }, { "epoch": 0.54, "learning_rate": 9.223161527109938e-06, "loss": 1.0586, "step": 337 }, { "epoch": 0.54, "learning_rate": 9.17148739720454e-06, "loss": 1.0664, "step": 338 }, { "epoch": 0.54, "learning_rate": 9.119835533834332e-06, "loss": 1.0742, "step": 339 }, { "epoch": 0.54, "learning_rate": 9.068207325159285e-06, "loss": 1.0557, "step": 340 }, { "epoch": 0.55, "learning_rate": 9.016604158703654e-06, "loss": 1.0752, "step": 341 }, { "epoch": 0.55, "learning_rate": 8.965027421318666e-06, "loss": 1.0957, "step": 342 }, { "epoch": 0.55, "learning_rate": 8.913478499145255e-06, "loss": 1.1006, "step": 343 }, { "epoch": 0.55, "learning_rate": 8.861958777576826e-06, "loss": 1.0454, "step": 344 }, { "epoch": 0.55, "learning_rate": 8.810469641222001e-06, "loss": 1.0713, "step": 345 }, { "epoch": 0.55, "learning_rate": 8.759012473867407e-06, "loss": 1.0522, "step": 346 }, { "epoch": 0.56, "learning_rate": 8.707588658440511e-06, "loss": 1.1133, "step": 347 }, { "epoch": 0.56, "learning_rate": 8.656199576972424e-06, "loss": 1.0586, "step": 348 }, { "epoch": 0.56, "learning_rate": 8.604846610560771e-06, "loss": 1.0811, "step": 349 }, { "epoch": 0.56, "learning_rate": 8.553531139332583e-06, "loss": 1.0708, "step": 350 }, { "epoch": 0.56, "learning_rate": 8.502254542407186e-06, "loss": 1.0703, "step": 351 }, { "epoch": 0.56, "learning_rate": 8.451018197859153e-06, "loss": 1.0791, "step": 352 }, { "epoch": 0.56, "learning_rate": 8.399823482681263e-06, "loss": 1.0669, "step": 353 }, { "epoch": 0.57, "learning_rate": 8.348671772747488e-06, "loss": 1.0801, "step": 354 }, { "epoch": 0.57, "learning_rate": 8.297564442776014e-06, "loss": 1.0557, "step": 355 }, { "epoch": 0.57, "learning_rate": 8.246502866292324e-06, "loss": 1.0771, "step": 356 }, { "epoch": 0.57, "learning_rate": 8.195488415592238e-06, "loss": 1.0659, "step": 357 }, { "epoch": 0.57, "learning_rate": 8.144522461705067e-06, "loss": 1.0708, "step": 358 }, { "epoch": 0.57, "learning_rate": 8.09360637435676e-06, "loss": 1.0664, "step": 359 }, { "epoch": 0.58, "learning_rate": 8.042741521933071e-06, "loss": 1.0752, "step": 360 }, { "epoch": 0.58, "learning_rate": 7.991929271442817e-06, "loss": 1.0586, "step": 361 }, { "epoch": 0.58, "learning_rate": 7.941170988481108e-06, "loss": 1.0415, "step": 362 }, { "epoch": 0.58, "learning_rate": 7.89046803719267e-06, "loss": 1.0771, "step": 363 }, { "epoch": 0.58, "learning_rate": 7.839821780235168e-06, "loss": 1.0908, "step": 364 }, { "epoch": 0.58, "learning_rate": 7.789233578742583e-06, "loss": 1.0879, "step": 365 }, { "epoch": 0.59, "learning_rate": 7.738704792288654e-06, "loss": 1.0625, "step": 366 }, { "epoch": 0.59, "learning_rate": 7.688236778850307e-06, "loss": 1.0664, "step": 367 }, { "epoch": 0.59, "learning_rate": 7.637830894771176e-06, "loss": 1.0508, "step": 368 }, { "epoch": 0.59, "learning_rate": 7.587488494725157e-06, "loss": 1.0508, "step": 369 }, { "epoch": 0.59, "learning_rate": 7.537210931679988e-06, "loss": 1.0469, "step": 370 }, { "epoch": 0.59, "learning_rate": 7.48699955686089e-06, "loss": 1.0708, "step": 371 }, { "epoch": 0.6, "learning_rate": 7.4368557197142596e-06, "loss": 1.0425, "step": 372 }, { "epoch": 0.6, "learning_rate": 7.3867807678713965e-06, "loss": 1.0684, "step": 373 }, { "epoch": 0.6, "learning_rate": 7.336776047112277e-06, "loss": 1.0752, "step": 374 }, { "epoch": 0.6, "learning_rate": 7.286842901329413e-06, "loss": 1.0713, "step": 375 }, { "epoch": 0.6, "learning_rate": 7.236982672491699e-06, "loss": 1.0938, "step": 376 }, { "epoch": 0.6, "learning_rate": 7.187196700608373e-06, "loss": 1.0713, "step": 377 }, { "epoch": 0.6, "learning_rate": 7.137486323692994e-06, "loss": 1.0898, "step": 378 }, { "epoch": 0.61, "learning_rate": 7.0878528777274814e-06, "loss": 1.0474, "step": 379 }, { "epoch": 0.61, "learning_rate": 7.0382976966262065e-06, "loss": 1.0801, "step": 380 }, { "epoch": 0.61, "learning_rate": 6.988822112200157e-06, "loss": 1.0698, "step": 381 }, { "epoch": 0.61, "learning_rate": 6.939427454121128e-06, "loss": 1.0547, "step": 382 }, { "epoch": 0.61, "learning_rate": 6.890115049885995e-06, "loss": 1.0474, "step": 383 }, { "epoch": 0.61, "learning_rate": 6.840886224781039e-06, "loss": 1.0649, "step": 384 }, { "epoch": 0.62, "learning_rate": 6.791742301846325e-06, "loss": 1.0654, "step": 385 }, { "epoch": 0.62, "learning_rate": 6.742684601840142e-06, "loss": 1.0503, "step": 386 }, { "epoch": 0.62, "learning_rate": 6.693714443203507e-06, "loss": 1.0977, "step": 387 }, { "epoch": 0.62, "learning_rate": 6.644833142024752e-06, "loss": 1.0742, "step": 388 }, { "epoch": 0.62, "learning_rate": 6.59604201200412e-06, "loss": 1.0703, "step": 389 }, { "epoch": 0.62, "learning_rate": 6.547342364418482e-06, "loss": 1.0464, "step": 390 }, { "epoch": 0.63, "learning_rate": 6.498735508086094e-06, "loss": 1.0879, "step": 391 }, { "epoch": 0.63, "learning_rate": 6.450222749331414e-06, "loss": 1.0576, "step": 392 }, { "epoch": 0.63, "learning_rate": 6.40180539194999e-06, "loss": 1.0371, "step": 393 }, { "epoch": 0.63, "learning_rate": 6.35348473717345e-06, "loss": 1.0791, "step": 394 }, { "epoch": 0.63, "learning_rate": 6.305262083634488e-06, "loss": 1.0654, "step": 395 }, { "epoch": 0.63, "learning_rate": 6.2571387273319905e-06, "loss": 1.0928, "step": 396 }, { "epoch": 0.64, "learning_rate": 6.209115961596208e-06, "loss": 1.0762, "step": 397 }, { "epoch": 0.64, "learning_rate": 6.1611950770539766e-06, "loss": 1.0742, "step": 398 }, { "epoch": 0.64, "learning_rate": 6.113377361594048e-06, "loss": 1.0708, "step": 399 }, { "epoch": 0.64, "learning_rate": 6.065664100332478e-06, "loss": 1.0796, "step": 400 }, { "epoch": 0.64, "learning_rate": 6.018056575578075e-06, "loss": 1.0801, "step": 401 }, { "epoch": 0.64, "learning_rate": 5.970556066797941e-06, "loss": 1.0659, "step": 402 }, { "epoch": 0.64, "learning_rate": 5.923163850583114e-06, "loss": 1.0361, "step": 403 }, { "epoch": 0.65, "learning_rate": 5.875881200614208e-06, "loss": 1.0498, "step": 404 }, { "epoch": 0.65, "learning_rate": 5.828709387627219e-06, "loss": 1.0576, "step": 405 }, { "epoch": 0.65, "learning_rate": 5.781649679379379e-06, "loss": 1.082, "step": 406 }, { "epoch": 0.65, "learning_rate": 5.7347033406150494e-06, "loss": 1.0591, "step": 407 }, { "epoch": 0.65, "learning_rate": 5.687871633031754e-06, "loss": 1.0522, "step": 408 }, { "epoch": 0.65, "learning_rate": 5.64115581524629e-06, "loss": 1.0693, "step": 409 }, { "epoch": 0.66, "learning_rate": 5.594557142760853e-06, "loss": 1.085, "step": 410 }, { "epoch": 0.66, "learning_rate": 5.548076867929331e-06, "loss": 1.0781, "step": 411 }, { "epoch": 0.66, "learning_rate": 5.501716239923642e-06, "loss": 1.0918, "step": 412 }, { "epoch": 0.66, "learning_rate": 5.455476504700161e-06, "loss": 1.083, "step": 413 }, { "epoch": 0.66, "learning_rate": 5.4093589049662175e-06, "loss": 1.0542, "step": 414 }, { "epoch": 0.66, "learning_rate": 5.3633646801467255e-06, "loss": 1.082, "step": 415 }, { "epoch": 0.67, "learning_rate": 5.31749506635086e-06, "loss": 1.0786, "step": 416 }, { "epoch": 0.67, "learning_rate": 5.271751296338823e-06, "loss": 1.0674, "step": 417 }, { "epoch": 0.67, "learning_rate": 5.226134599488728e-06, "loss": 1.0879, "step": 418 }, { "epoch": 0.67, "learning_rate": 5.1806462017635775e-06, "loss": 1.0693, "step": 419 }, { "epoch": 0.67, "learning_rate": 5.135287325678271e-06, "loss": 1.0635, "step": 420 }, { "epoch": 0.67, "learning_rate": 5.090059190266779e-06, "loss": 1.0664, "step": 421 }, { "epoch": 0.68, "learning_rate": 5.044963011049384e-06, "loss": 1.0967, "step": 422 }, { "epoch": 0.68, "learning_rate": 5.000000000000003e-06, "loss": 1.0742, "step": 423 }, { "epoch": 0.68, "learning_rate": 4.955171365513603e-06, "loss": 1.0645, "step": 424 }, { "epoch": 0.68, "learning_rate": 4.910478312373757e-06, "loss": 1.0645, "step": 425 }, { "epoch": 0.68, "learning_rate": 4.865922041720239e-06, "loss": 1.0947, "step": 426 }, { "epoch": 0.68, "learning_rate": 4.821503751016746e-06, "loss": 1.0527, "step": 427 }, { "epoch": 0.68, "learning_rate": 4.777224634018732e-06, "loss": 1.0571, "step": 428 }, { "epoch": 0.69, "learning_rate": 4.733085880741301e-06, "loss": 1.0713, "step": 429 }, { "epoch": 0.69, "learning_rate": 4.689088677427249e-06, "loss": 1.0474, "step": 430 }, { "epoch": 0.69, "learning_rate": 4.645234206515171e-06, "loss": 1.0103, "step": 431 }, { "epoch": 0.69, "learning_rate": 4.601523646607675e-06, "loss": 1.0576, "step": 432 }, { "epoch": 0.69, "learning_rate": 4.557958172439726e-06, "loss": 1.0693, "step": 433 }, { "epoch": 0.69, "learning_rate": 4.5145389548470645e-06, "loss": 1.082, "step": 434 }, { "epoch": 0.7, "learning_rate": 4.471267160734731e-06, "loss": 1.0732, "step": 435 }, { "epoch": 0.7, "learning_rate": 4.4281439530457174e-06, "loss": 1.0503, "step": 436 }, { "epoch": 0.7, "learning_rate": 4.385170490729712e-06, "loss": 1.0576, "step": 437 }, { "epoch": 0.7, "learning_rate": 4.342347928711953e-06, "loss": 1.0498, "step": 438 }, { "epoch": 0.7, "learning_rate": 4.299677417862174e-06, "loss": 1.063, "step": 439 }, { "epoch": 0.7, "learning_rate": 4.257160104963695e-06, "loss": 1.0625, "step": 440 }, { "epoch": 0.71, "learning_rate": 4.214797132682597e-06, "loss": 1.0713, "step": 441 }, { "epoch": 0.71, "learning_rate": 4.172589639536992e-06, "loss": 1.0557, "step": 442 }, { "epoch": 0.71, "learning_rate": 4.130538759866457e-06, "loss": 1.0391, "step": 443 }, { "epoch": 0.71, "learning_rate": 4.088645623801534e-06, "loss": 1.0684, "step": 444 }, { "epoch": 0.71, "learning_rate": 4.046911357233343e-06, "loss": 1.0469, "step": 445 }, { "epoch": 0.71, "learning_rate": 4.00533708178334e-06, "loss": 1.0698, "step": 446 }, { "epoch": 0.72, "learning_rate": 3.9639239147731865e-06, "loss": 1.0645, "step": 447 }, { "epoch": 0.72, "learning_rate": 3.9226729691946865e-06, "loss": 1.0615, "step": 448 }, { "epoch": 0.72, "learning_rate": 3.881585353679891e-06, "loss": 1.0508, "step": 449 }, { "epoch": 0.72, "learning_rate": 3.840662172471315e-06, "loss": 1.0742, "step": 450 }, { "epoch": 0.72, "learning_rate": 3.799904525392251e-06, "loss": 1.0596, "step": 451 }, { "epoch": 0.72, "learning_rate": 3.759313507817196e-06, "loss": 1.0791, "step": 452 }, { "epoch": 0.72, "learning_rate": 3.718890210642442e-06, "loss": 1.1436, "step": 453 }, { "epoch": 0.73, "learning_rate": 3.6786357202567367e-06, "loss": 1.0771, "step": 454 }, { "epoch": 0.73, "learning_rate": 3.638551118512089e-06, "loss": 1.0454, "step": 455 }, { "epoch": 0.73, "learning_rate": 3.5986374826947067e-06, "loss": 1.0889, "step": 456 }, { "epoch": 0.73, "learning_rate": 3.558895885496023e-06, "loss": 1.0469, "step": 457 }, { "epoch": 0.73, "learning_rate": 3.519327394983888e-06, "loss": 1.0723, "step": 458 }, { "epoch": 0.73, "learning_rate": 3.479933074573858e-06, "loss": 1.0552, "step": 459 }, { "epoch": 0.74, "learning_rate": 3.440713983000601e-06, "loss": 1.0596, "step": 460 }, { "epoch": 0.74, "learning_rate": 3.401671174289469e-06, "loss": 1.063, "step": 461 }, { "epoch": 0.74, "learning_rate": 3.3628056977281456e-06, "loss": 1.0693, "step": 462 }, { "epoch": 0.74, "learning_rate": 3.3241185978384636e-06, "loss": 1.0957, "step": 463 }, { "epoch": 0.74, "learning_rate": 3.2856109143483316e-06, "loss": 1.0566, "step": 464 }, { "epoch": 0.74, "learning_rate": 3.2472836821637744e-06, "loss": 1.0503, "step": 465 }, { "epoch": 0.75, "learning_rate": 3.209137931341143e-06, "loss": 1.0918, "step": 466 }, { "epoch": 0.75, "learning_rate": 3.1711746870594083e-06, "loss": 1.0508, "step": 467 }, { "epoch": 0.75, "learning_rate": 3.1333949695926323e-06, "loss": 1.0713, "step": 468 }, { "epoch": 0.75, "learning_rate": 3.0957997942825337e-06, "loss": 1.0576, "step": 469 }, { "epoch": 0.75, "learning_rate": 3.0583901715111965e-06, "loss": 1.0562, "step": 470 }, { "epoch": 0.75, "learning_rate": 3.021167106673928e-06, "loss": 1.0596, "step": 471 }, { "epoch": 0.76, "learning_rate": 2.9841316001522345e-06, "loss": 1.0811, "step": 472 }, { "epoch": 0.76, "learning_rate": 2.94728464728693e-06, "loss": 1.0469, "step": 473 }, { "epoch": 0.76, "learning_rate": 2.910627238351383e-06, "loss": 1.0781, "step": 474 }, { "epoch": 0.76, "learning_rate": 2.8741603585249312e-06, "loss": 1.0459, "step": 475 }, { "epoch": 0.76, "learning_rate": 2.837884987866363e-06, "loss": 1.0518, "step": 476 }, { "epoch": 0.76, "learning_rate": 2.8018021012875994e-06, "loss": 1.0684, "step": 477 }, { "epoch": 0.76, "learning_rate": 2.7659126685275028e-06, "loss": 1.0688, "step": 478 }, { "epoch": 0.77, "learning_rate": 2.7302176541257984e-06, "loss": 1.0781, "step": 479 }, { "epoch": 0.77, "learning_rate": 2.694718017397151e-06, "loss": 1.0771, "step": 480 }, { "epoch": 0.77, "learning_rate": 2.6594147124053983e-06, "loss": 1.084, "step": 481 }, { "epoch": 0.77, "learning_rate": 2.6243086879379e-06, "loss": 1.0527, "step": 482 }, { "epoch": 0.77, "learning_rate": 2.5894008874800323e-06, "loss": 1.0234, "step": 483 }, { "epoch": 0.77, "learning_rate": 2.5546922491898497e-06, "loss": 1.0503, "step": 484 }, { "epoch": 0.78, "learning_rate": 2.5201837058728506e-06, "loss": 1.0884, "step": 485 }, { "epoch": 0.78, "learning_rate": 2.485876184956928e-06, "loss": 1.0688, "step": 486 }, { "epoch": 0.78, "learning_rate": 2.451770608467432e-06, "loss": 1.0635, "step": 487 }, { "epoch": 0.78, "learning_rate": 2.417867893002387e-06, "loss": 1.0645, "step": 488 }, { "epoch": 0.78, "learning_rate": 2.3841689497078746e-06, "loss": 1.0469, "step": 489 }, { "epoch": 0.78, "learning_rate": 2.3506746842535244e-06, "loss": 1.0601, "step": 490 }, { "epoch": 0.79, "learning_rate": 2.317385996808195e-06, "loss": 1.0586, "step": 491 }, { "epoch": 0.79, "learning_rate": 2.2843037820157678e-06, "loss": 1.0889, "step": 492 }, { "epoch": 0.79, "learning_rate": 2.251428928971102e-06, "loss": 1.0674, "step": 493 }, { "epoch": 0.79, "learning_rate": 2.218762321196156e-06, "loss": 1.0908, "step": 494 }, { "epoch": 0.79, "learning_rate": 2.186304836616221e-06, "loss": 1.0654, "step": 495 }, { "epoch": 0.79, "learning_rate": 2.1540573475363402e-06, "loss": 1.0596, "step": 496 }, { "epoch": 0.8, "learning_rate": 2.122020720617869e-06, "loss": 1.0811, "step": 497 }, { "epoch": 0.8, "learning_rate": 2.090195816855164e-06, "loss": 1.0371, "step": 498 }, { "epoch": 0.8, "learning_rate": 2.058583491552465e-06, "loss": 1.0322, "step": 499 }, { "epoch": 0.8, "learning_rate": 2.0271845943008984e-06, "loss": 1.0674, "step": 500 }, { "epoch": 0.8, "learning_rate": 1.9959999689556407e-06, "loss": 1.0547, "step": 501 }, { "epoch": 0.8, "learning_rate": 1.9650304536132426e-06, "loss": 1.0557, "step": 502 }, { "epoch": 0.8, "learning_rate": 1.9342768805891176e-06, "loss": 1.0864, "step": 503 }, { "epoch": 0.81, "learning_rate": 1.9037400763951508e-06, "loss": 1.0371, "step": 504 }, { "epoch": 0.81, "learning_rate": 1.8734208617174986e-06, "loss": 1.0601, "step": 505 }, { "epoch": 0.81, "learning_rate": 1.8433200513945338e-06, "loss": 1.0527, "step": 506 }, { "epoch": 0.81, "learning_rate": 1.813438454394948e-06, "loss": 1.0723, "step": 507 }, { "epoch": 0.81, "learning_rate": 1.7837768737959937e-06, "loss": 1.083, "step": 508 }, { "epoch": 0.81, "learning_rate": 1.7543361067619269e-06, "loss": 1.063, "step": 509 }, { "epoch": 0.82, "learning_rate": 1.7251169445225658e-06, "loss": 1.0391, "step": 510 }, { "epoch": 0.82, "learning_rate": 1.6961201723520248e-06, "loss": 1.0327, "step": 511 }, { "epoch": 0.82, "learning_rate": 1.6673465695476233e-06, "loss": 1.084, "step": 512 }, { "epoch": 0.82, "learning_rate": 1.6387969094089318e-06, "loss": 1.0684, "step": 513 }, { "epoch": 0.82, "learning_rate": 1.6104719592169905e-06, "loss": 1.0698, "step": 514 }, { "epoch": 0.82, "learning_rate": 1.5823724802136863e-06, "loss": 1.0977, "step": 515 }, { "epoch": 0.83, "learning_rate": 1.5544992275813053e-06, "loss": 1.0635, "step": 516 }, { "epoch": 0.83, "learning_rate": 1.5268529504222262e-06, "loss": 1.0747, "step": 517 }, { "epoch": 0.83, "learning_rate": 1.4994343917387854e-06, "loss": 1.0503, "step": 518 }, { "epoch": 0.83, "learning_rate": 1.4722442884133214e-06, "loss": 1.0605, "step": 519 }, { "epoch": 0.83, "learning_rate": 1.4452833711883629e-06, "loss": 1.0645, "step": 520 }, { "epoch": 0.83, "learning_rate": 1.4185523646469822e-06, "loss": 1.0742, "step": 521 }, { "epoch": 0.84, "learning_rate": 1.3920519871933425e-06, "loss": 1.0801, "step": 522 }, { "epoch": 0.84, "learning_rate": 1.3657829510333653e-06, "loss": 1.0386, "step": 523 }, { "epoch": 0.84, "learning_rate": 1.339745962155613e-06, "loss": 1.0889, "step": 524 }, { "epoch": 0.84, "learning_rate": 1.313941720312303e-06, "loss": 1.0454, "step": 525 }, { "epoch": 0.84, "learning_rate": 1.2883709190004956e-06, "loss": 1.0723, "step": 526 }, { "epoch": 0.84, "learning_rate": 1.263034245443473e-06, "loss": 1.043, "step": 527 }, { "epoch": 0.84, "learning_rate": 1.2379323805722575e-06, "loss": 1.0605, "step": 528 }, { "epoch": 0.85, "learning_rate": 1.2130659990073146e-06, "loss": 1.0532, "step": 529 }, { "epoch": 0.85, "learning_rate": 1.1884357690404157e-06, "loss": 1.0454, "step": 530 }, { "epoch": 0.85, "learning_rate": 1.1640423526166987e-06, "loss": 1.0483, "step": 531 }, { "epoch": 0.85, "learning_rate": 1.1398864053168534e-06, "loss": 1.0928, "step": 532 }, { "epoch": 0.85, "learning_rate": 1.1159685763395113e-06, "loss": 1.0645, "step": 533 }, { "epoch": 0.85, "learning_rate": 1.0922895084838036e-06, "loss": 1.0669, "step": 534 }, { "epoch": 0.86, "learning_rate": 1.0688498381320855e-06, "loss": 1.064, "step": 535 }, { "epoch": 0.86, "learning_rate": 1.0456501952328191e-06, "loss": 1.0601, "step": 536 }, { "epoch": 0.86, "learning_rate": 1.022691203283661e-06, "loss": 1.082, "step": 537 }, { "epoch": 0.86, "learning_rate": 9.999734793146998e-07, "loss": 1.0601, "step": 538 }, { "epoch": 0.86, "learning_rate": 9.77497633871868e-07, "loss": 1.0762, "step": 539 }, { "epoch": 0.86, "learning_rate": 9.5526427100053e-07, "loss": 1.0532, "step": 540 }, { "epoch": 0.87, "learning_rate": 9.332739882292752e-07, "loss": 1.0576, "step": 541 }, { "epoch": 0.87, "learning_rate": 9.115273765538202e-07, "loss": 1.1055, "step": 542 }, { "epoch": 0.87, "learning_rate": 8.900250204211513e-07, "loss": 1.0386, "step": 543 }, { "epoch": 0.87, "learning_rate": 8.687674977138116e-07, "loss": 1.0654, "step": 544 }, { "epoch": 0.87, "learning_rate": 8.477553797343729e-07, "loss": 1.0615, "step": 545 }, { "epoch": 0.87, "learning_rate": 8.269892311900696e-07, "loss": 1.0879, "step": 546 }, { "epoch": 0.88, "learning_rate": 8.06469610177636e-07, "loss": 1.0596, "step": 547 }, { "epoch": 0.88, "learning_rate": 7.861970681683051e-07, "loss": 1.0352, "step": 548 }, { "epoch": 0.88, "learning_rate": 7.661721499929753e-07, "loss": 1.0605, "step": 549 }, { "epoch": 0.88, "learning_rate": 7.463953938275859e-07, "loss": 1.0483, "step": 550 }, { "epoch": 0.88, "learning_rate": 7.268673311786378e-07, "loss": 1.0557, "step": 551 }, { "epoch": 0.88, "learning_rate": 7.07588486868922e-07, "loss": 1.0527, "step": 552 }, { "epoch": 0.88, "learning_rate": 6.885593790234057e-07, "loss": 1.0439, "step": 553 }, { "epoch": 0.89, "learning_rate": 6.697805190553086e-07, "loss": 1.084, "step": 554 }, { "epoch": 0.89, "learning_rate": 6.512524116523633e-07, "loss": 1.0552, "step": 555 }, { "epoch": 0.89, "learning_rate": 6.329755547632499e-07, "loss": 1.0508, "step": 556 }, { "epoch": 0.89, "learning_rate": 6.149504395842087e-07, "loss": 1.0869, "step": 557 }, { "epoch": 0.89, "learning_rate": 5.971775505458444e-07, "loss": 1.083, "step": 558 }, { "epoch": 0.89, "learning_rate": 5.796573653001091e-07, "loss": 1.0444, "step": 559 }, { "epoch": 0.9, "learning_rate": 5.62390354707455e-07, "loss": 1.0781, "step": 560 }, { "epoch": 0.9, "learning_rate": 5.453769828241872e-07, "loss": 1.0488, "step": 561 }, { "epoch": 0.9, "learning_rate": 5.286177068899989e-07, "loss": 1.0713, "step": 562 }, { "epoch": 0.9, "learning_rate": 5.121129773156663e-07, "loss": 1.0728, "step": 563 }, { "epoch": 0.9, "learning_rate": 4.95863237670956e-07, "loss": 1.0527, "step": 564 }, { "epoch": 0.9, "learning_rate": 4.798689246727006e-07, "loss": 1.0723, "step": 565 }, { "epoch": 0.91, "learning_rate": 4.6413046817306404e-07, "loss": 1.085, "step": 566 }, { "epoch": 0.91, "learning_rate": 4.4864829114798394e-07, "loss": 1.0674, "step": 567 }, { "epoch": 0.91, "learning_rate": 4.3342280968580287e-07, "loss": 1.0615, "step": 568 }, { "epoch": 0.91, "learning_rate": 4.184544329761009e-07, "loss": 1.0703, "step": 569 }, { "epoch": 0.91, "learning_rate": 4.037435632986786e-07, "loss": 1.0508, "step": 570 }, { "epoch": 0.91, "learning_rate": 3.8929059601275463e-07, "loss": 1.0586, "step": 571 }, { "epoch": 0.92, "learning_rate": 3.750959195463466e-07, "loss": 1.0762, "step": 572 }, { "epoch": 0.92, "learning_rate": 3.611599153858214e-07, "loss": 1.082, "step": 573 }, { "epoch": 0.92, "learning_rate": 3.474829580656436e-07, "loss": 1.0947, "step": 574 }, { "epoch": 0.92, "learning_rate": 3.3406541515832e-07, "loss": 1.0791, "step": 575 }, { "epoch": 0.92, "learning_rate": 3.209076472645112e-07, "loss": 1.0371, "step": 576 }, { "epoch": 0.92, "learning_rate": 3.080100080033388e-07, "loss": 1.0674, "step": 577 }, { "epoch": 0.92, "learning_rate": 2.9537284400289354e-07, "loss": 1.0977, "step": 578 }, { "epoch": 0.93, "learning_rate": 2.829964948909048e-07, "loss": 1.0664, "step": 579 }, { "epoch": 0.93, "learning_rate": 2.708812932856253e-07, "loss": 1.0493, "step": 580 }, { "epoch": 0.93, "learning_rate": 2.5902756478688674e-07, "loss": 1.0859, "step": 581 }, { "epoch": 0.93, "learning_rate": 2.474356279673462e-07, "loss": 1.0693, "step": 582 }, { "epoch": 0.93, "learning_rate": 2.3610579436392999e-07, "loss": 1.0234, "step": 583 }, { "epoch": 0.93, "learning_rate": 2.2503836846945792e-07, "loss": 1.1074, "step": 584 }, { "epoch": 0.94, "learning_rate": 2.1423364772445886e-07, "loss": 1.0664, "step": 585 }, { "epoch": 0.94, "learning_rate": 2.036919225091827e-07, "loss": 1.0732, "step": 586 }, { "epoch": 0.94, "learning_rate": 1.9341347613579086e-07, "loss": 1.0527, "step": 587 }, { "epoch": 0.94, "learning_rate": 1.8339858484073935e-07, "loss": 1.0586, "step": 588 }, { "epoch": 0.94, "learning_rate": 1.7364751777736334e-07, "loss": 1.0664, "step": 589 }, { "epoch": 0.94, "learning_rate": 1.6416053700863965e-07, "loss": 1.0601, "step": 590 }, { "epoch": 0.95, "learning_rate": 1.5493789750014032e-07, "loss": 1.0352, "step": 591 }, { "epoch": 0.95, "learning_rate": 1.459798471131868e-07, "loss": 1.0132, "step": 592 }, { "epoch": 0.95, "learning_rate": 1.3728662659818205e-07, "loss": 1.0317, "step": 593 }, { "epoch": 0.95, "learning_rate": 1.2885846958814673e-07, "loss": 1.0352, "step": 594 }, { "epoch": 0.95, "learning_rate": 1.206956025924333e-07, "loss": 1.0723, "step": 595 }, { "epoch": 0.95, "learning_rate": 1.1279824499064396e-07, "loss": 1.0513, "step": 596 }, { "epoch": 0.96, "learning_rate": 1.0516660902673448e-07, "loss": 1.0615, "step": 597 }, { "epoch": 0.96, "learning_rate": 9.780089980330643e-08, "loss": 1.0801, "step": 598 }, { "epoch": 0.96, "learning_rate": 9.070131527609604e-08, "loss": 1.0977, "step": 599 }, { "epoch": 0.96, "learning_rate": 8.386804624865851e-08, "loss": 1.0801, "step": 600 }, { "epoch": 0.96, "learning_rate": 7.730127636723539e-08, "loss": 1.0684, "step": 601 }, { "epoch": 0.96, "learning_rate": 7.100118211581852e-08, "loss": 1.0762, "step": 602 }, { "epoch": 0.96, "learning_rate": 6.496793281141056e-08, "loss": 1.0518, "step": 603 }, { "epoch": 0.97, "learning_rate": 5.920169059947412e-08, "loss": 1.0625, "step": 604 }, { "epoch": 0.97, "learning_rate": 5.37026104495697e-08, "loss": 1.0366, "step": 605 }, { "epoch": 0.97, "learning_rate": 4.8470840151195745e-08, "loss": 1.0703, "step": 606 }, { "epoch": 0.97, "learning_rate": 4.350652030981395e-08, "loss": 1.0576, "step": 607 }, { "epoch": 0.97, "learning_rate": 3.8809784343072364e-08, "loss": 1.0791, "step": 608 }, { "epoch": 0.97, "learning_rate": 3.438075847721933e-08, "loss": 1.0615, "step": 609 }, { "epoch": 0.98, "learning_rate": 3.0219561743707326e-08, "loss": 1.0986, "step": 610 }, { "epoch": 0.98, "learning_rate": 2.6326305976001054e-08, "loss": 1.0967, "step": 611 }, { "epoch": 0.98, "learning_rate": 2.2701095806565432e-08, "loss": 1.0566, "step": 612 }, { "epoch": 0.98, "learning_rate": 1.9344028664056715e-08, "loss": 1.0938, "step": 613 }, { "epoch": 0.98, "learning_rate": 1.6255194770704586e-08, "loss": 1.085, "step": 614 }, { "epoch": 0.98, "learning_rate": 1.3434677139885222e-08, "loss": 1.0454, "step": 615 }, { "epoch": 0.99, "learning_rate": 1.0882551573891953e-08, "loss": 1.082, "step": 616 }, { "epoch": 0.99, "learning_rate": 8.59888666189579e-09, "loss": 1.0601, "step": 617 }, { "epoch": 0.99, "learning_rate": 6.583743778106888e-09, "loss": 1.0386, "step": 618 }, { "epoch": 0.99, "learning_rate": 4.837177080119215e-09, "loss": 1.0674, "step": 619 }, { "epoch": 0.99, "learning_rate": 3.3592335074594805e-09, "loss": 1.0493, "step": 620 }, { "epoch": 0.99, "learning_rate": 2.149952780321485e-09, "loss": 1.082, "step": 621 }, { "epoch": 1.0, "learning_rate": 1.209367398504746e-09, "loss": 1.0449, "step": 622 }, { "epoch": 1.0, "learning_rate": 5.375026405352035e-10, "loss": 1.0698, "step": 623 }, { "epoch": 1.0, "learning_rate": 1.3437656298687096e-10, "loss": 1.0913, "step": 624 }, { "epoch": 1.0, "learning_rate": 0.0, "loss": 1.043, "step": 625 }, { "epoch": 1.0, "step": 625, "total_flos": 201129833791488.0, "train_loss": 1.113865625, "train_runtime": 3780.6253, "train_samples_per_second": 21.161, "train_steps_per_second": 0.165 } ], "max_steps": 625, "num_train_epochs": 1, "total_flos": 201129833791488.0, "trial_name": null, "trial_params": null }