{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.999807932392202, "global_step": 5206, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 9.569377990430622e-08, "loss": 1.6837, "step": 1 }, { "epoch": 0.0, "learning_rate": 1.9138755980861244e-07, "loss": 1.6405, "step": 2 }, { "epoch": 0.0, "learning_rate": 2.870813397129187e-07, "loss": 1.6922, "step": 3 }, { "epoch": 0.0, "learning_rate": 3.827751196172249e-07, "loss": 1.6333, "step": 4 }, { "epoch": 0.0, "learning_rate": 4.784688995215311e-07, "loss": 1.7213, "step": 5 }, { "epoch": 0.0, "learning_rate": 5.741626794258374e-07, "loss": 1.6693, "step": 6 }, { "epoch": 0.0, "learning_rate": 6.698564593301436e-07, "loss": 1.744, "step": 7 }, { "epoch": 0.0, "learning_rate": 7.655502392344498e-07, "loss": 1.6218, "step": 8 }, { "epoch": 0.0, "learning_rate": 8.612440191387561e-07, "loss": 1.6985, "step": 9 }, { "epoch": 0.0, "learning_rate": 9.569377990430622e-07, "loss": 1.6495, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.0526315789473685e-06, "loss": 1.7264, "step": 11 }, { "epoch": 0.0, "learning_rate": 1.1483253588516749e-06, "loss": 1.7072, "step": 12 }, { "epoch": 0.0, "learning_rate": 1.244019138755981e-06, "loss": 1.6678, "step": 13 }, { "epoch": 0.01, "learning_rate": 1.3397129186602873e-06, "loss": 1.5876, "step": 14 }, { "epoch": 0.01, "learning_rate": 1.4354066985645934e-06, "loss": 1.7304, "step": 15 }, { "epoch": 0.01, "learning_rate": 1.5311004784688995e-06, "loss": 1.6002, "step": 16 }, { "epoch": 0.01, "learning_rate": 1.6267942583732059e-06, "loss": 1.7426, "step": 17 }, { "epoch": 0.01, "learning_rate": 1.7224880382775122e-06, "loss": 1.7507, "step": 18 }, { "epoch": 0.01, "learning_rate": 1.8181818181818183e-06, "loss": 1.6407, "step": 19 }, { "epoch": 0.01, "learning_rate": 1.9138755980861244e-06, "loss": 1.6488, "step": 20 }, { "epoch": 0.01, "learning_rate": 2.0095693779904305e-06, "loss": 1.6789, "step": 21 }, { "epoch": 0.01, "learning_rate": 2.105263157894737e-06, "loss": 1.7015, "step": 22 }, { "epoch": 0.01, "learning_rate": 2.200956937799043e-06, "loss": 1.7146, "step": 23 }, { "epoch": 0.01, "learning_rate": 2.2966507177033497e-06, "loss": 1.6881, "step": 24 }, { "epoch": 0.01, "learning_rate": 2.392344497607656e-06, "loss": 1.6851, "step": 25 }, { "epoch": 0.01, "learning_rate": 2.488038277511962e-06, "loss": 1.722, "step": 26 }, { "epoch": 0.01, "learning_rate": 2.5837320574162685e-06, "loss": 1.7729, "step": 27 }, { "epoch": 0.01, "learning_rate": 2.6794258373205746e-06, "loss": 1.7067, "step": 28 }, { "epoch": 0.01, "learning_rate": 2.7751196172248807e-06, "loss": 1.6695, "step": 29 }, { "epoch": 0.01, "learning_rate": 2.870813397129187e-06, "loss": 1.7693, "step": 30 }, { "epoch": 0.01, "learning_rate": 2.966507177033493e-06, "loss": 1.7355, "step": 31 }, { "epoch": 0.01, "learning_rate": 3.062200956937799e-06, "loss": 1.6723, "step": 32 }, { "epoch": 0.01, "learning_rate": 3.157894736842105e-06, "loss": 1.6447, "step": 33 }, { "epoch": 0.01, "learning_rate": 3.2535885167464117e-06, "loss": 1.6678, "step": 34 }, { "epoch": 0.01, "learning_rate": 3.3492822966507182e-06, "loss": 1.6832, "step": 35 }, { "epoch": 0.01, "learning_rate": 3.4449760765550244e-06, "loss": 1.6746, "step": 36 }, { "epoch": 0.01, "learning_rate": 3.5406698564593305e-06, "loss": 1.6562, "step": 37 }, { "epoch": 0.01, "learning_rate": 3.6363636363636366e-06, "loss": 1.6835, "step": 38 }, { "epoch": 0.01, "learning_rate": 3.732057416267943e-06, "loss": 1.5717, "step": 39 }, { "epoch": 0.02, "learning_rate": 3.827751196172249e-06, "loss": 1.6407, "step": 40 }, { "epoch": 0.02, "learning_rate": 3.923444976076555e-06, "loss": 1.7743, "step": 41 }, { "epoch": 0.02, "learning_rate": 4.019138755980861e-06, "loss": 1.7158, "step": 42 }, { "epoch": 0.02, "learning_rate": 4.114832535885168e-06, "loss": 1.6319, "step": 43 }, { "epoch": 0.02, "learning_rate": 4.210526315789474e-06, "loss": 1.7498, "step": 44 }, { "epoch": 0.02, "learning_rate": 4.30622009569378e-06, "loss": 1.6349, "step": 45 }, { "epoch": 0.02, "learning_rate": 4.401913875598086e-06, "loss": 1.7408, "step": 46 }, { "epoch": 0.02, "learning_rate": 4.4976076555023925e-06, "loss": 1.6522, "step": 47 }, { "epoch": 0.02, "learning_rate": 4.593301435406699e-06, "loss": 1.6731, "step": 48 }, { "epoch": 0.02, "learning_rate": 4.6889952153110055e-06, "loss": 1.6394, "step": 49 }, { "epoch": 0.02, "learning_rate": 4.784688995215312e-06, "loss": 1.719, "step": 50 }, { "epoch": 0.02, "learning_rate": 4.880382775119618e-06, "loss": 1.6637, "step": 51 }, { "epoch": 0.02, "learning_rate": 4.976076555023924e-06, "loss": 1.6214, "step": 52 }, { "epoch": 0.02, "learning_rate": 5.07177033492823e-06, "loss": 1.7144, "step": 53 }, { "epoch": 0.02, "learning_rate": 5.167464114832537e-06, "loss": 1.6644, "step": 54 }, { "epoch": 0.02, "learning_rate": 5.263157894736842e-06, "loss": 1.705, "step": 55 }, { "epoch": 0.02, "learning_rate": 5.358851674641149e-06, "loss": 1.644, "step": 56 }, { "epoch": 0.02, "learning_rate": 5.4545454545454545e-06, "loss": 1.6952, "step": 57 }, { "epoch": 0.02, "learning_rate": 5.550239234449761e-06, "loss": 1.7306, "step": 58 }, { "epoch": 0.02, "learning_rate": 5.645933014354067e-06, "loss": 1.6748, "step": 59 }, { "epoch": 0.02, "learning_rate": 5.741626794258374e-06, "loss": 1.6571, "step": 60 }, { "epoch": 0.02, "learning_rate": 5.837320574162681e-06, "loss": 1.6663, "step": 61 }, { "epoch": 0.02, "learning_rate": 5.933014354066986e-06, "loss": 1.7043, "step": 62 }, { "epoch": 0.02, "learning_rate": 6.028708133971293e-06, "loss": 1.6267, "step": 63 }, { "epoch": 0.02, "learning_rate": 6.124401913875598e-06, "loss": 1.7082, "step": 64 }, { "epoch": 0.02, "learning_rate": 6.220095693779905e-06, "loss": 1.7672, "step": 65 }, { "epoch": 0.03, "learning_rate": 6.31578947368421e-06, "loss": 1.6969, "step": 66 }, { "epoch": 0.03, "learning_rate": 6.411483253588517e-06, "loss": 1.6767, "step": 67 }, { "epoch": 0.03, "learning_rate": 6.507177033492823e-06, "loss": 1.689, "step": 68 }, { "epoch": 0.03, "learning_rate": 6.6028708133971295e-06, "loss": 1.5895, "step": 69 }, { "epoch": 0.03, "learning_rate": 6.6985645933014365e-06, "loss": 1.733, "step": 70 }, { "epoch": 0.03, "learning_rate": 6.794258373205742e-06, "loss": 1.6752, "step": 71 }, { "epoch": 0.03, "learning_rate": 6.889952153110049e-06, "loss": 1.6481, "step": 72 }, { "epoch": 0.03, "learning_rate": 6.985645933014355e-06, "loss": 1.6958, "step": 73 }, { "epoch": 0.03, "learning_rate": 7.081339712918661e-06, "loss": 1.6746, "step": 74 }, { "epoch": 0.03, "learning_rate": 7.177033492822967e-06, "loss": 1.676, "step": 75 }, { "epoch": 0.03, "learning_rate": 7.272727272727273e-06, "loss": 1.6444, "step": 76 }, { "epoch": 0.03, "learning_rate": 7.368421052631579e-06, "loss": 1.6601, "step": 77 }, { "epoch": 0.03, "learning_rate": 7.464114832535886e-06, "loss": 1.6562, "step": 78 }, { "epoch": 0.03, "learning_rate": 7.5598086124401915e-06, "loss": 1.5893, "step": 79 }, { "epoch": 0.03, "learning_rate": 7.655502392344498e-06, "loss": 1.601, "step": 80 }, { "epoch": 0.03, "learning_rate": 7.751196172248805e-06, "loss": 1.5885, "step": 81 }, { "epoch": 0.03, "learning_rate": 7.84688995215311e-06, "loss": 1.6632, "step": 82 }, { "epoch": 0.03, "learning_rate": 7.942583732057418e-06, "loss": 1.6344, "step": 83 }, { "epoch": 0.03, "learning_rate": 8.038277511961722e-06, "loss": 1.6156, "step": 84 }, { "epoch": 0.03, "learning_rate": 8.13397129186603e-06, "loss": 1.6659, "step": 85 }, { "epoch": 0.03, "learning_rate": 8.229665071770336e-06, "loss": 1.6798, "step": 86 }, { "epoch": 0.03, "learning_rate": 8.325358851674642e-06, "loss": 1.5726, "step": 87 }, { "epoch": 0.03, "learning_rate": 8.421052631578948e-06, "loss": 1.5745, "step": 88 }, { "epoch": 0.03, "learning_rate": 8.516746411483254e-06, "loss": 1.6673, "step": 89 }, { "epoch": 0.03, "learning_rate": 8.61244019138756e-06, "loss": 1.6137, "step": 90 }, { "epoch": 0.03, "learning_rate": 8.708133971291867e-06, "loss": 1.6613, "step": 91 }, { "epoch": 0.04, "learning_rate": 8.803827751196173e-06, "loss": 1.6584, "step": 92 }, { "epoch": 0.04, "learning_rate": 8.899521531100479e-06, "loss": 1.6229, "step": 93 }, { "epoch": 0.04, "learning_rate": 8.995215311004785e-06, "loss": 1.6926, "step": 94 }, { "epoch": 0.04, "learning_rate": 9.090909090909091e-06, "loss": 1.6805, "step": 95 }, { "epoch": 0.04, "learning_rate": 9.186602870813399e-06, "loss": 1.6663, "step": 96 }, { "epoch": 0.04, "learning_rate": 9.282296650717703e-06, "loss": 1.6402, "step": 97 }, { "epoch": 0.04, "learning_rate": 9.377990430622011e-06, "loss": 1.5803, "step": 98 }, { "epoch": 0.04, "learning_rate": 9.473684210526315e-06, "loss": 1.6429, "step": 99 }, { "epoch": 0.04, "learning_rate": 9.569377990430623e-06, "loss": 1.5797, "step": 100 }, { "epoch": 0.04, "learning_rate": 9.66507177033493e-06, "loss": 1.603, "step": 101 }, { "epoch": 0.04, "learning_rate": 9.760765550239236e-06, "loss": 1.6353, "step": 102 }, { "epoch": 0.04, "learning_rate": 9.856459330143542e-06, "loss": 1.5553, "step": 103 }, { "epoch": 0.04, "learning_rate": 9.952153110047848e-06, "loss": 1.5825, "step": 104 }, { "epoch": 0.04, "learning_rate": 1.0047846889952154e-05, "loss": 1.5568, "step": 105 }, { "epoch": 0.04, "learning_rate": 1.014354066985646e-05, "loss": 1.5873, "step": 106 }, { "epoch": 0.04, "learning_rate": 1.0239234449760766e-05, "loss": 1.5779, "step": 107 }, { "epoch": 0.04, "learning_rate": 1.0334928229665074e-05, "loss": 1.5795, "step": 108 }, { "epoch": 0.04, "learning_rate": 1.0430622009569378e-05, "loss": 1.6138, "step": 109 }, { "epoch": 0.04, "learning_rate": 1.0526315789473684e-05, "loss": 1.5775, "step": 110 }, { "epoch": 0.04, "learning_rate": 1.062200956937799e-05, "loss": 1.6587, "step": 111 }, { "epoch": 0.04, "learning_rate": 1.0717703349282298e-05, "loss": 1.4842, "step": 112 }, { "epoch": 0.04, "learning_rate": 1.0813397129186604e-05, "loss": 1.5138, "step": 113 }, { "epoch": 0.04, "learning_rate": 1.0909090909090909e-05, "loss": 1.5523, "step": 114 }, { "epoch": 0.04, "learning_rate": 1.1004784688995217e-05, "loss": 1.5822, "step": 115 }, { "epoch": 0.04, "learning_rate": 1.1100478468899523e-05, "loss": 1.4537, "step": 116 }, { "epoch": 0.04, "learning_rate": 1.1196172248803829e-05, "loss": 1.5656, "step": 117 }, { "epoch": 0.05, "learning_rate": 1.1291866028708133e-05, "loss": 1.595, "step": 118 }, { "epoch": 0.05, "learning_rate": 1.1387559808612441e-05, "loss": 1.4627, "step": 119 }, { "epoch": 0.05, "learning_rate": 1.1483253588516747e-05, "loss": 1.5568, "step": 120 }, { "epoch": 0.05, "learning_rate": 1.1578947368421053e-05, "loss": 1.4923, "step": 121 }, { "epoch": 0.05, "learning_rate": 1.1674641148325361e-05, "loss": 1.5321, "step": 122 }, { "epoch": 0.05, "learning_rate": 1.1770334928229666e-05, "loss": 1.5087, "step": 123 }, { "epoch": 0.05, "learning_rate": 1.1866028708133972e-05, "loss": 1.5368, "step": 124 }, { "epoch": 0.05, "learning_rate": 1.196172248803828e-05, "loss": 1.4817, "step": 125 }, { "epoch": 0.05, "learning_rate": 1.2057416267942586e-05, "loss": 1.4701, "step": 126 }, { "epoch": 0.05, "learning_rate": 1.215311004784689e-05, "loss": 1.5298, "step": 127 }, { "epoch": 0.05, "learning_rate": 1.2248803827751196e-05, "loss": 1.5237, "step": 128 }, { "epoch": 0.05, "learning_rate": 1.2344497607655504e-05, "loss": 1.4189, "step": 129 }, { "epoch": 0.05, "learning_rate": 1.244019138755981e-05, "loss": 1.5034, "step": 130 }, { "epoch": 0.05, "learning_rate": 1.2535885167464116e-05, "loss": 1.4792, "step": 131 }, { "epoch": 0.05, "learning_rate": 1.263157894736842e-05, "loss": 1.4959, "step": 132 }, { "epoch": 0.05, "learning_rate": 1.2727272727272728e-05, "loss": 1.521, "step": 133 }, { "epoch": 0.05, "learning_rate": 1.2822966507177035e-05, "loss": 1.5176, "step": 134 }, { "epoch": 0.05, "learning_rate": 1.2918660287081342e-05, "loss": 1.4526, "step": 135 }, { "epoch": 0.05, "learning_rate": 1.3014354066985647e-05, "loss": 1.4623, "step": 136 }, { "epoch": 0.05, "learning_rate": 1.3110047846889953e-05, "loss": 1.4572, "step": 137 }, { "epoch": 0.05, "learning_rate": 1.3205741626794259e-05, "loss": 1.4457, "step": 138 }, { "epoch": 0.05, "learning_rate": 1.3301435406698567e-05, "loss": 1.4371, "step": 139 }, { "epoch": 0.05, "learning_rate": 1.3397129186602873e-05, "loss": 1.4097, "step": 140 }, { "epoch": 0.05, "learning_rate": 1.3492822966507177e-05, "loss": 1.4738, "step": 141 }, { "epoch": 0.05, "learning_rate": 1.3588516746411483e-05, "loss": 1.4385, "step": 142 }, { "epoch": 0.05, "learning_rate": 1.3684210526315791e-05, "loss": 1.5098, "step": 143 }, { "epoch": 0.06, "learning_rate": 1.3779904306220097e-05, "loss": 1.4362, "step": 144 }, { "epoch": 0.06, "learning_rate": 1.3875598086124402e-05, "loss": 1.4807, "step": 145 }, { "epoch": 0.06, "learning_rate": 1.397129186602871e-05, "loss": 1.4597, "step": 146 }, { "epoch": 0.06, "learning_rate": 1.4066985645933016e-05, "loss": 1.3928, "step": 147 }, { "epoch": 0.06, "learning_rate": 1.4162679425837322e-05, "loss": 1.4799, "step": 148 }, { "epoch": 0.06, "learning_rate": 1.4258373205741626e-05, "loss": 1.4959, "step": 149 }, { "epoch": 0.06, "learning_rate": 1.4354066985645934e-05, "loss": 1.4005, "step": 150 }, { "epoch": 0.06, "learning_rate": 1.444976076555024e-05, "loss": 1.4363, "step": 151 }, { "epoch": 0.06, "learning_rate": 1.4545454545454546e-05, "loss": 1.4938, "step": 152 }, { "epoch": 0.06, "learning_rate": 1.4641148325358854e-05, "loss": 1.431, "step": 153 }, { "epoch": 0.06, "learning_rate": 1.4736842105263159e-05, "loss": 1.4179, "step": 154 }, { "epoch": 0.06, "learning_rate": 1.4832535885167465e-05, "loss": 1.4423, "step": 155 }, { "epoch": 0.06, "learning_rate": 1.4928229665071772e-05, "loss": 1.3846, "step": 156 }, { "epoch": 0.06, "learning_rate": 1.5023923444976079e-05, "loss": 1.4503, "step": 157 }, { "epoch": 0.06, "learning_rate": 1.5119617224880383e-05, "loss": 1.449, "step": 158 }, { "epoch": 0.06, "learning_rate": 1.5215311004784689e-05, "loss": 1.4239, "step": 159 }, { "epoch": 0.06, "learning_rate": 1.5311004784688995e-05, "loss": 1.4865, "step": 160 }, { "epoch": 0.06, "learning_rate": 1.5406698564593305e-05, "loss": 1.4174, "step": 161 }, { "epoch": 0.06, "learning_rate": 1.550239234449761e-05, "loss": 1.4288, "step": 162 }, { "epoch": 0.06, "learning_rate": 1.5598086124401914e-05, "loss": 1.3891, "step": 163 }, { "epoch": 0.06, "learning_rate": 1.569377990430622e-05, "loss": 1.3881, "step": 164 }, { "epoch": 0.06, "learning_rate": 1.578947368421053e-05, "loss": 1.4854, "step": 165 }, { "epoch": 0.06, "learning_rate": 1.5885167464114835e-05, "loss": 1.4002, "step": 166 }, { "epoch": 0.06, "learning_rate": 1.5980861244019138e-05, "loss": 1.3995, "step": 167 }, { "epoch": 0.06, "learning_rate": 1.6076555023923444e-05, "loss": 1.4179, "step": 168 }, { "epoch": 0.06, "learning_rate": 1.6172248803827754e-05, "loss": 1.4052, "step": 169 }, { "epoch": 0.07, "learning_rate": 1.626794258373206e-05, "loss": 1.3586, "step": 170 }, { "epoch": 0.07, "learning_rate": 1.6363636363636366e-05, "loss": 1.3775, "step": 171 }, { "epoch": 0.07, "learning_rate": 1.6459330143540672e-05, "loss": 1.3696, "step": 172 }, { "epoch": 0.07, "learning_rate": 1.6555023923444978e-05, "loss": 1.3819, "step": 173 }, { "epoch": 0.07, "learning_rate": 1.6650717703349284e-05, "loss": 1.3907, "step": 174 }, { "epoch": 0.07, "learning_rate": 1.674641148325359e-05, "loss": 1.3675, "step": 175 }, { "epoch": 0.07, "learning_rate": 1.6842105263157896e-05, "loss": 1.3933, "step": 176 }, { "epoch": 0.07, "learning_rate": 1.6937799043062203e-05, "loss": 1.4095, "step": 177 }, { "epoch": 0.07, "learning_rate": 1.703349282296651e-05, "loss": 1.4225, "step": 178 }, { "epoch": 0.07, "learning_rate": 1.7129186602870815e-05, "loss": 1.3619, "step": 179 }, { "epoch": 0.07, "learning_rate": 1.722488038277512e-05, "loss": 1.3901, "step": 180 }, { "epoch": 0.07, "learning_rate": 1.7320574162679427e-05, "loss": 1.4437, "step": 181 }, { "epoch": 0.07, "learning_rate": 1.7416267942583733e-05, "loss": 1.4093, "step": 182 }, { "epoch": 0.07, "learning_rate": 1.751196172248804e-05, "loss": 1.3617, "step": 183 }, { "epoch": 0.07, "learning_rate": 1.7607655502392345e-05, "loss": 1.4122, "step": 184 }, { "epoch": 0.07, "learning_rate": 1.770334928229665e-05, "loss": 1.4035, "step": 185 }, { "epoch": 0.07, "learning_rate": 1.7799043062200958e-05, "loss": 1.4257, "step": 186 }, { "epoch": 0.07, "learning_rate": 1.7894736842105264e-05, "loss": 1.343, "step": 187 }, { "epoch": 0.07, "learning_rate": 1.799043062200957e-05, "loss": 1.3792, "step": 188 }, { "epoch": 0.07, "learning_rate": 1.8086124401913876e-05, "loss": 1.4001, "step": 189 }, { "epoch": 0.07, "learning_rate": 1.8181818181818182e-05, "loss": 1.3949, "step": 190 }, { "epoch": 0.07, "learning_rate": 1.8277511961722488e-05, "loss": 1.2751, "step": 191 }, { "epoch": 0.07, "learning_rate": 1.8373205741626798e-05, "loss": 1.3927, "step": 192 }, { "epoch": 0.07, "learning_rate": 1.8468899521531104e-05, "loss": 1.3736, "step": 193 }, { "epoch": 0.07, "learning_rate": 1.8564593301435407e-05, "loss": 1.407, "step": 194 }, { "epoch": 0.07, "learning_rate": 1.8660287081339713e-05, "loss": 1.293, "step": 195 }, { "epoch": 0.08, "learning_rate": 1.8755980861244022e-05, "loss": 1.4352, "step": 196 }, { "epoch": 0.08, "learning_rate": 1.8851674641148328e-05, "loss": 1.3686, "step": 197 }, { "epoch": 0.08, "learning_rate": 1.894736842105263e-05, "loss": 1.3603, "step": 198 }, { "epoch": 0.08, "learning_rate": 1.904306220095694e-05, "loss": 1.3622, "step": 199 }, { "epoch": 0.08, "learning_rate": 1.9138755980861247e-05, "loss": 1.3213, "step": 200 }, { "epoch": 0.08, "learning_rate": 1.9234449760765553e-05, "loss": 1.3544, "step": 201 }, { "epoch": 0.08, "learning_rate": 1.933014354066986e-05, "loss": 1.4098, "step": 202 }, { "epoch": 0.08, "learning_rate": 1.9425837320574165e-05, "loss": 1.3632, "step": 203 }, { "epoch": 0.08, "learning_rate": 1.952153110047847e-05, "loss": 1.3763, "step": 204 }, { "epoch": 0.08, "learning_rate": 1.9617224880382777e-05, "loss": 1.422, "step": 205 }, { "epoch": 0.08, "learning_rate": 1.9712918660287083e-05, "loss": 1.4294, "step": 206 }, { "epoch": 0.08, "learning_rate": 1.980861244019139e-05, "loss": 1.3162, "step": 207 }, { "epoch": 0.08, "learning_rate": 1.9904306220095696e-05, "loss": 1.4011, "step": 208 }, { "epoch": 0.08, "learning_rate": 2e-05, "loss": 1.3204, "step": 209 }, { "epoch": 0.08, "learning_rate": 1.9999998023708347e-05, "loss": 1.2975, "step": 210 }, { "epoch": 0.08, "learning_rate": 1.9999992094834168e-05, "loss": 1.3114, "step": 211 }, { "epoch": 0.08, "learning_rate": 1.9999982213379806e-05, "loss": 1.3969, "step": 212 }, { "epoch": 0.08, "learning_rate": 1.9999968379349164e-05, "loss": 1.4279, "step": 213 }, { "epoch": 0.08, "learning_rate": 1.9999950592747715e-05, "loss": 1.2922, "step": 214 }, { "epoch": 0.08, "learning_rate": 1.999992885358249e-05, "loss": 1.3326, "step": 215 }, { "epoch": 0.08, "learning_rate": 1.9999903161862077e-05, "loss": 1.3672, "step": 216 }, { "epoch": 0.08, "learning_rate": 1.9999873517596627e-05, "loss": 1.4222, "step": 217 }, { "epoch": 0.08, "learning_rate": 1.9999839920797866e-05, "loss": 1.3635, "step": 218 }, { "epoch": 0.08, "learning_rate": 1.9999802371479076e-05, "loss": 1.3334, "step": 219 }, { "epoch": 0.08, "learning_rate": 1.9999760869655085e-05, "loss": 1.3158, "step": 220 }, { "epoch": 0.08, "learning_rate": 1.999971541534231e-05, "loss": 1.3863, "step": 221 }, { "epoch": 0.09, "learning_rate": 1.999966600855871e-05, "loss": 1.35, "step": 222 }, { "epoch": 0.09, "learning_rate": 1.9999612649323813e-05, "loss": 1.3968, "step": 223 }, { "epoch": 0.09, "learning_rate": 1.9999555337658715e-05, "loss": 1.3218, "step": 224 }, { "epoch": 0.09, "learning_rate": 1.9999494073586066e-05, "loss": 1.2996, "step": 225 }, { "epoch": 0.09, "learning_rate": 1.999942885713008e-05, "loss": 1.2679, "step": 226 }, { "epoch": 0.09, "learning_rate": 1.999935968831654e-05, "loss": 1.3808, "step": 227 }, { "epoch": 0.09, "learning_rate": 1.999928656717278e-05, "loss": 1.3592, "step": 228 }, { "epoch": 0.09, "learning_rate": 1.9999209493727696e-05, "loss": 1.2891, "step": 229 }, { "epoch": 0.09, "learning_rate": 1.9999128468011764e-05, "loss": 1.2974, "step": 230 }, { "epoch": 0.09, "learning_rate": 1.9999043490057004e-05, "loss": 1.3625, "step": 231 }, { "epoch": 0.09, "learning_rate": 1.9998954559897005e-05, "loss": 1.3183, "step": 232 }, { "epoch": 0.09, "learning_rate": 1.999886167756692e-05, "loss": 1.341, "step": 233 }, { "epoch": 0.09, "learning_rate": 1.9998764843103457e-05, "loss": 1.3337, "step": 234 }, { "epoch": 0.09, "learning_rate": 1.999866405654489e-05, "loss": 1.3103, "step": 235 }, { "epoch": 0.09, "learning_rate": 1.9998559317931062e-05, "loss": 1.3444, "step": 236 }, { "epoch": 0.09, "learning_rate": 1.9998450627303366e-05, "loss": 1.2612, "step": 237 }, { "epoch": 0.09, "learning_rate": 1.9998337984704765e-05, "loss": 1.2397, "step": 238 }, { "epoch": 0.09, "learning_rate": 1.9998221390179783e-05, "loss": 1.376, "step": 239 }, { "epoch": 0.09, "learning_rate": 1.9998100843774504e-05, "loss": 1.2608, "step": 240 }, { "epoch": 0.09, "learning_rate": 1.9997976345536572e-05, "loss": 1.2986, "step": 241 }, { "epoch": 0.09, "learning_rate": 1.9997847895515204e-05, "loss": 1.2709, "step": 242 }, { "epoch": 0.09, "learning_rate": 1.9997715493761164e-05, "loss": 1.3278, "step": 243 }, { "epoch": 0.09, "learning_rate": 1.9997579140326785e-05, "loss": 1.3189, "step": 244 }, { "epoch": 0.09, "learning_rate": 1.9997438835265965e-05, "loss": 1.3316, "step": 245 }, { "epoch": 0.09, "learning_rate": 1.9997294578634155e-05, "loss": 1.2944, "step": 246 }, { "epoch": 0.09, "learning_rate": 1.9997146370488384e-05, "loss": 1.2699, "step": 247 }, { "epoch": 0.1, "learning_rate": 1.9996994210887225e-05, "loss": 1.348, "step": 248 }, { "epoch": 0.1, "learning_rate": 1.9996838099890818e-05, "loss": 1.3526, "step": 249 }, { "epoch": 0.1, "learning_rate": 1.9996678037560874e-05, "loss": 1.2874, "step": 250 }, { "epoch": 0.1, "learning_rate": 1.9996514023960657e-05, "loss": 1.2618, "step": 251 }, { "epoch": 0.1, "learning_rate": 1.999634605915499e-05, "loss": 1.2833, "step": 252 }, { "epoch": 0.1, "learning_rate": 1.9996174143210268e-05, "loss": 1.3532, "step": 253 }, { "epoch": 0.1, "learning_rate": 1.9995998276194443e-05, "loss": 1.366, "step": 254 }, { "epoch": 0.1, "learning_rate": 1.9995818458177023e-05, "loss": 1.3189, "step": 255 }, { "epoch": 0.1, "learning_rate": 1.9995634689229085e-05, "loss": 1.3693, "step": 256 }, { "epoch": 0.1, "learning_rate": 1.9995446969423263e-05, "loss": 1.3197, "step": 257 }, { "epoch": 0.1, "learning_rate": 1.9995255298833757e-05, "loss": 1.3405, "step": 258 }, { "epoch": 0.1, "learning_rate": 1.999505967753633e-05, "loss": 1.3305, "step": 259 }, { "epoch": 0.1, "learning_rate": 1.9994860105608296e-05, "loss": 1.2923, "step": 260 }, { "epoch": 0.1, "learning_rate": 1.999465658312854e-05, "loss": 1.3011, "step": 261 }, { "epoch": 0.1, "learning_rate": 1.999444911017751e-05, "loss": 1.2757, "step": 262 }, { "epoch": 0.1, "learning_rate": 1.9994237686837204e-05, "loss": 1.3422, "step": 263 }, { "epoch": 0.1, "learning_rate": 1.9994022313191193e-05, "loss": 1.2992, "step": 264 }, { "epoch": 0.1, "learning_rate": 1.999380298932461e-05, "loss": 1.2972, "step": 265 }, { "epoch": 0.1, "learning_rate": 1.9993579715324135e-05, "loss": 1.3537, "step": 266 }, { "epoch": 0.1, "learning_rate": 1.9993352491278028e-05, "loss": 1.3767, "step": 267 }, { "epoch": 0.1, "learning_rate": 1.9993121317276093e-05, "loss": 1.3174, "step": 268 }, { "epoch": 0.1, "learning_rate": 1.9992886193409705e-05, "loss": 1.3133, "step": 269 }, { "epoch": 0.1, "learning_rate": 1.9992647119771805e-05, "loss": 1.2588, "step": 270 }, { "epoch": 0.1, "learning_rate": 1.9992404096456886e-05, "loss": 1.3204, "step": 271 }, { "epoch": 0.1, "learning_rate": 1.9992157123561e-05, "loss": 1.2913, "step": 272 }, { "epoch": 0.1, "learning_rate": 1.999190620118177e-05, "loss": 1.3679, "step": 273 }, { "epoch": 0.11, "learning_rate": 1.9991651329418374e-05, "loss": 1.2399, "step": 274 }, { "epoch": 0.11, "learning_rate": 1.999139250837155e-05, "loss": 1.3594, "step": 275 }, { "epoch": 0.11, "learning_rate": 1.9991129738143608e-05, "loss": 1.3225, "step": 276 }, { "epoch": 0.11, "learning_rate": 1.9990863018838396e-05, "loss": 1.32, "step": 277 }, { "epoch": 0.11, "learning_rate": 1.9990592350561353e-05, "loss": 1.309, "step": 278 }, { "epoch": 0.11, "learning_rate": 1.9990317733419443e-05, "loss": 1.2683, "step": 279 }, { "epoch": 0.11, "learning_rate": 1.999003916752123e-05, "loss": 1.2796, "step": 280 }, { "epoch": 0.11, "learning_rate": 1.998975665297681e-05, "loss": 1.302, "step": 281 }, { "epoch": 0.11, "learning_rate": 1.998947018989785e-05, "loss": 1.2965, "step": 282 }, { "epoch": 0.11, "learning_rate": 1.9989179778397577e-05, "loss": 1.2684, "step": 283 }, { "epoch": 0.11, "learning_rate": 1.998888541859078e-05, "loss": 1.3246, "step": 284 }, { "epoch": 0.11, "learning_rate": 1.998858711059381e-05, "loss": 1.2903, "step": 285 }, { "epoch": 0.11, "learning_rate": 1.998828485452457e-05, "loss": 1.304, "step": 286 }, { "epoch": 0.11, "learning_rate": 1.998797865050253e-05, "loss": 1.2447, "step": 287 }, { "epoch": 0.11, "learning_rate": 1.998766849864872e-05, "loss": 1.2812, "step": 288 }, { "epoch": 0.11, "learning_rate": 1.998735439908573e-05, "loss": 1.317, "step": 289 }, { "epoch": 0.11, "learning_rate": 1.998703635193772e-05, "loss": 1.3165, "step": 290 }, { "epoch": 0.11, "learning_rate": 1.9986714357330384e-05, "loss": 1.2869, "step": 291 }, { "epoch": 0.11, "learning_rate": 1.9986388415391005e-05, "loss": 1.2873, "step": 292 }, { "epoch": 0.11, "learning_rate": 1.998605852624841e-05, "loss": 1.2754, "step": 293 }, { "epoch": 0.11, "learning_rate": 1.998572469003299e-05, "loss": 1.2467, "step": 294 }, { "epoch": 0.11, "learning_rate": 1.99853869068767e-05, "loss": 1.2853, "step": 295 }, { "epoch": 0.11, "learning_rate": 1.998504517691305e-05, "loss": 1.2631, "step": 296 }, { "epoch": 0.11, "learning_rate": 1.998469950027711e-05, "loss": 1.3287, "step": 297 }, { "epoch": 0.11, "learning_rate": 1.9984349877105512e-05, "loss": 1.2648, "step": 298 }, { "epoch": 0.11, "learning_rate": 1.9983996307536446e-05, "loss": 1.2791, "step": 299 }, { "epoch": 0.12, "learning_rate": 1.9983638791709668e-05, "loss": 1.2357, "step": 300 }, { "epoch": 0.12, "learning_rate": 1.9983277329766487e-05, "loss": 1.3593, "step": 301 }, { "epoch": 0.12, "learning_rate": 1.998291192184977e-05, "loss": 1.3516, "step": 302 }, { "epoch": 0.12, "learning_rate": 1.9982542568103956e-05, "loss": 1.3107, "step": 303 }, { "epoch": 0.12, "learning_rate": 1.9982169268675024e-05, "loss": 1.3086, "step": 304 }, { "epoch": 0.12, "learning_rate": 1.9981792023710533e-05, "loss": 1.3019, "step": 305 }, { "epoch": 0.12, "learning_rate": 1.9981410833359588e-05, "loss": 1.3181, "step": 306 }, { "epoch": 0.12, "learning_rate": 1.998102569777286e-05, "loss": 1.24, "step": 307 }, { "epoch": 0.12, "learning_rate": 1.9980636617102576e-05, "loss": 1.2333, "step": 308 }, { "epoch": 0.12, "learning_rate": 1.998024359150252e-05, "loss": 1.3341, "step": 309 }, { "epoch": 0.12, "learning_rate": 1.9979846621128044e-05, "loss": 1.2919, "step": 310 }, { "epoch": 0.12, "learning_rate": 1.997944570613605e-05, "loss": 1.2703, "step": 311 }, { "epoch": 0.12, "learning_rate": 1.9979040846685005e-05, "loss": 1.2633, "step": 312 }, { "epoch": 0.12, "learning_rate": 1.997863204293493e-05, "loss": 1.2974, "step": 313 }, { "epoch": 0.12, "learning_rate": 1.9978219295047413e-05, "loss": 1.2869, "step": 314 }, { "epoch": 0.12, "learning_rate": 1.9977802603185595e-05, "loss": 1.2932, "step": 315 }, { "epoch": 0.12, "learning_rate": 1.9977381967514176e-05, "loss": 1.2652, "step": 316 }, { "epoch": 0.12, "learning_rate": 1.9976957388199412e-05, "loss": 1.3481, "step": 317 }, { "epoch": 0.12, "learning_rate": 1.9976528865409128e-05, "loss": 1.2928, "step": 318 }, { "epoch": 0.12, "learning_rate": 1.9976096399312696e-05, "loss": 1.3031, "step": 319 }, { "epoch": 0.12, "learning_rate": 1.9975659990081054e-05, "loss": 1.3057, "step": 320 }, { "epoch": 0.12, "learning_rate": 1.99752196378867e-05, "loss": 1.2982, "step": 321 }, { "epoch": 0.12, "learning_rate": 1.9974775342903676e-05, "loss": 1.2699, "step": 322 }, { "epoch": 0.12, "learning_rate": 1.9974327105307605e-05, "loss": 1.2937, "step": 323 }, { "epoch": 0.12, "learning_rate": 1.9973874925275648e-05, "loss": 1.2447, "step": 324 }, { "epoch": 0.12, "learning_rate": 1.9973418802986543e-05, "loss": 1.3163, "step": 325 }, { "epoch": 0.13, "learning_rate": 1.997295873862057e-05, "loss": 1.2932, "step": 326 }, { "epoch": 0.13, "learning_rate": 1.997249473235957e-05, "loss": 1.2513, "step": 327 }, { "epoch": 0.13, "learning_rate": 1.997202678438695e-05, "loss": 1.2928, "step": 328 }, { "epoch": 0.13, "learning_rate": 1.9971554894887667e-05, "loss": 1.3077, "step": 329 }, { "epoch": 0.13, "learning_rate": 1.997107906404824e-05, "loss": 1.2587, "step": 330 }, { "epoch": 0.13, "learning_rate": 1.997059929205675e-05, "loss": 1.275, "step": 331 }, { "epoch": 0.13, "learning_rate": 1.9970115579102828e-05, "loss": 1.2301, "step": 332 }, { "epoch": 0.13, "learning_rate": 1.996962792537766e-05, "loss": 1.2772, "step": 333 }, { "epoch": 0.13, "learning_rate": 1.9969136331074005e-05, "loss": 1.2451, "step": 334 }, { "epoch": 0.13, "learning_rate": 1.9968640796386164e-05, "loss": 1.2746, "step": 335 }, { "epoch": 0.13, "learning_rate": 1.996814132151e-05, "loss": 1.2599, "step": 336 }, { "epoch": 0.13, "learning_rate": 1.996763790664294e-05, "loss": 1.2817, "step": 337 }, { "epoch": 0.13, "learning_rate": 1.9967130551983958e-05, "loss": 1.2183, "step": 338 }, { "epoch": 0.13, "learning_rate": 1.996661925773359e-05, "loss": 1.232, "step": 339 }, { "epoch": 0.13, "learning_rate": 1.9966104024093933e-05, "loss": 1.2529, "step": 340 }, { "epoch": 0.13, "learning_rate": 1.9965584851268634e-05, "loss": 1.236, "step": 341 }, { "epoch": 0.13, "learning_rate": 1.9965061739462903e-05, "loss": 1.2564, "step": 342 }, { "epoch": 0.13, "learning_rate": 1.9964534688883505e-05, "loss": 1.2145, "step": 343 }, { "epoch": 0.13, "learning_rate": 1.9964003699738758e-05, "loss": 1.2763, "step": 344 }, { "epoch": 0.13, "learning_rate": 1.9963468772238537e-05, "loss": 1.2779, "step": 345 }, { "epoch": 0.13, "learning_rate": 1.9962929906594282e-05, "loss": 1.2727, "step": 346 }, { "epoch": 0.13, "learning_rate": 1.9962387103018988e-05, "loss": 1.2599, "step": 347 }, { "epoch": 0.13, "learning_rate": 1.996184036172719e-05, "loss": 1.2426, "step": 348 }, { "epoch": 0.13, "learning_rate": 1.9961289682935006e-05, "loss": 1.2829, "step": 349 }, { "epoch": 0.13, "learning_rate": 1.9960735066860088e-05, "loss": 1.3243, "step": 350 }, { "epoch": 0.13, "learning_rate": 1.996017651372165e-05, "loss": 1.2661, "step": 351 }, { "epoch": 0.14, "learning_rate": 1.9959614023740473e-05, "loss": 1.261, "step": 352 }, { "epoch": 0.14, "learning_rate": 1.9959047597138883e-05, "loss": 1.2647, "step": 353 }, { "epoch": 0.14, "learning_rate": 1.9958477234140763e-05, "loss": 1.2303, "step": 354 }, { "epoch": 0.14, "learning_rate": 1.9957902934971555e-05, "loss": 1.2686, "step": 355 }, { "epoch": 0.14, "learning_rate": 1.9957324699858257e-05, "loss": 1.2533, "step": 356 }, { "epoch": 0.14, "learning_rate": 1.9956742529029416e-05, "loss": 1.2382, "step": 357 }, { "epoch": 0.14, "learning_rate": 1.9956156422715143e-05, "loss": 1.3174, "step": 358 }, { "epoch": 0.14, "learning_rate": 1.9955566381147106e-05, "loss": 1.3166, "step": 359 }, { "epoch": 0.14, "learning_rate": 1.9954972404558516e-05, "loss": 1.2643, "step": 360 }, { "epoch": 0.14, "learning_rate": 1.9954374493184153e-05, "loss": 1.2519, "step": 361 }, { "epoch": 0.14, "learning_rate": 1.995377264726034e-05, "loss": 1.2592, "step": 362 }, { "epoch": 0.14, "learning_rate": 1.9953166867024972e-05, "loss": 1.2011, "step": 363 }, { "epoch": 0.14, "learning_rate": 1.995255715271748e-05, "loss": 1.217, "step": 364 }, { "epoch": 0.14, "learning_rate": 1.995194350457886e-05, "loss": 1.2183, "step": 365 }, { "epoch": 0.14, "learning_rate": 1.9951325922851664e-05, "loss": 1.291, "step": 366 }, { "epoch": 0.14, "learning_rate": 1.9950704407779994e-05, "loss": 1.2728, "step": 367 }, { "epoch": 0.14, "learning_rate": 1.995007895960951e-05, "loss": 1.1918, "step": 368 }, { "epoch": 0.14, "learning_rate": 1.9949449578587427e-05, "loss": 1.3104, "step": 369 }, { "epoch": 0.14, "learning_rate": 1.994881626496251e-05, "loss": 1.2163, "step": 370 }, { "epoch": 0.14, "learning_rate": 1.9948179018985084e-05, "loss": 1.2922, "step": 371 }, { "epoch": 0.14, "learning_rate": 1.9947537840907025e-05, "loss": 1.3297, "step": 372 }, { "epoch": 0.14, "learning_rate": 1.9946892730981765e-05, "loss": 1.2488, "step": 373 }, { "epoch": 0.14, "learning_rate": 1.9946243689464287e-05, "loss": 1.2676, "step": 374 }, { "epoch": 0.14, "learning_rate": 1.994559071661113e-05, "loss": 1.2635, "step": 375 }, { "epoch": 0.14, "learning_rate": 1.994493381268039e-05, "loss": 1.2504, "step": 376 }, { "epoch": 0.14, "learning_rate": 1.9944272977931713e-05, "loss": 1.2318, "step": 377 }, { "epoch": 0.15, "learning_rate": 1.9943608212626295e-05, "loss": 1.263, "step": 378 }, { "epoch": 0.15, "learning_rate": 1.9942939517026893e-05, "loss": 1.1912, "step": 379 }, { "epoch": 0.15, "learning_rate": 1.9942266891397817e-05, "loss": 1.2222, "step": 380 }, { "epoch": 0.15, "learning_rate": 1.9941590336004924e-05, "loss": 1.2438, "step": 381 }, { "epoch": 0.15, "learning_rate": 1.9940909851115635e-05, "loss": 1.2879, "step": 382 }, { "epoch": 0.15, "learning_rate": 1.9940225436998905e-05, "loss": 1.2461, "step": 383 }, { "epoch": 0.15, "learning_rate": 1.9939537093925267e-05, "loss": 1.2197, "step": 384 }, { "epoch": 0.15, "learning_rate": 1.9938844822166784e-05, "loss": 1.2557, "step": 385 }, { "epoch": 0.15, "learning_rate": 1.9938148621997092e-05, "loss": 1.2463, "step": 386 }, { "epoch": 0.15, "learning_rate": 1.9937448493691362e-05, "loss": 1.2646, "step": 387 }, { "epoch": 0.15, "learning_rate": 1.993674443752633e-05, "loss": 1.2705, "step": 388 }, { "epoch": 0.15, "learning_rate": 1.993603645378028e-05, "loss": 1.2058, "step": 389 }, { "epoch": 0.15, "learning_rate": 1.993532454273304e-05, "loss": 1.2253, "step": 390 }, { "epoch": 0.15, "learning_rate": 1.9934608704666013e-05, "loss": 1.2477, "step": 391 }, { "epoch": 0.15, "learning_rate": 1.993388893986213e-05, "loss": 1.294, "step": 392 }, { "epoch": 0.15, "learning_rate": 1.993316524860589e-05, "loss": 1.2414, "step": 393 }, { "epoch": 0.15, "learning_rate": 1.993243763118333e-05, "loss": 1.2975, "step": 394 }, { "epoch": 0.15, "learning_rate": 1.9931706087882054e-05, "loss": 1.2749, "step": 395 }, { "epoch": 0.15, "learning_rate": 1.9930970618991207e-05, "loss": 1.2805, "step": 396 }, { "epoch": 0.15, "learning_rate": 1.9930231224801493e-05, "loss": 1.2792, "step": 397 }, { "epoch": 0.15, "learning_rate": 1.992948790560516e-05, "loss": 1.2556, "step": 398 }, { "epoch": 0.15, "learning_rate": 1.992874066169601e-05, "loss": 1.2951, "step": 399 }, { "epoch": 0.15, "learning_rate": 1.9927989493369403e-05, "loss": 1.2108, "step": 400 }, { "epoch": 0.15, "eval_loss": 1.261375069618225, "eval_runtime": 688.8775, "eval_samples_per_second": 4.937, "eval_steps_per_second": 0.309, "step": 400 }, { "epoch": 0.15, "learning_rate": 1.992723440092224e-05, "loss": 1.2489, "step": 401 }, { "epoch": 0.15, "learning_rate": 1.992647538465298e-05, "loss": 1.2346, "step": 402 }, { "epoch": 0.15, "learning_rate": 1.992571244486163e-05, "loss": 1.2032, "step": 403 }, { "epoch": 0.16, "learning_rate": 1.9924945581849743e-05, "loss": 1.264, "step": 404 }, { "epoch": 0.16, "learning_rate": 1.992417479592044e-05, "loss": 1.2275, "step": 405 }, { "epoch": 0.16, "learning_rate": 1.9923400087378363e-05, "loss": 1.2306, "step": 406 }, { "epoch": 0.16, "learning_rate": 1.992262145652974e-05, "loss": 1.2718, "step": 407 }, { "epoch": 0.16, "learning_rate": 1.992183890368232e-05, "loss": 1.2977, "step": 408 }, { "epoch": 0.16, "learning_rate": 1.992105242914542e-05, "loss": 1.229, "step": 409 }, { "epoch": 0.16, "learning_rate": 1.9920262033229894e-05, "loss": 1.2637, "step": 410 }, { "epoch": 0.16, "learning_rate": 1.9919467716248158e-05, "loss": 1.2093, "step": 411 }, { "epoch": 0.16, "learning_rate": 1.991866947851417e-05, "loss": 1.3048, "step": 412 }, { "epoch": 0.16, "learning_rate": 1.9917867320343443e-05, "loss": 1.2533, "step": 413 }, { "epoch": 0.16, "learning_rate": 1.9917061242053034e-05, "loss": 1.2907, "step": 414 }, { "epoch": 0.16, "learning_rate": 1.991625124396155e-05, "loss": 1.2571, "step": 415 }, { "epoch": 0.16, "learning_rate": 1.991543732638915e-05, "loss": 1.1929, "step": 416 }, { "epoch": 0.16, "learning_rate": 1.991461948965755e-05, "loss": 1.2488, "step": 417 }, { "epoch": 0.16, "learning_rate": 1.991379773409e-05, "loss": 1.2886, "step": 418 }, { "epoch": 0.16, "learning_rate": 1.99129720600113e-05, "loss": 1.2497, "step": 419 }, { "epoch": 0.16, "learning_rate": 1.9912142467747816e-05, "loss": 1.2662, "step": 420 }, { "epoch": 0.16, "learning_rate": 1.9911308957627446e-05, "loss": 1.219, "step": 421 }, { "epoch": 0.16, "learning_rate": 1.9910471529979637e-05, "loss": 1.2629, "step": 422 }, { "epoch": 0.16, "learning_rate": 1.99096301851354e-05, "loss": 1.2653, "step": 423 }, { "epoch": 0.16, "learning_rate": 1.9908784923427277e-05, "loss": 1.2442, "step": 424 }, { "epoch": 0.16, "learning_rate": 1.9907935745189364e-05, "loss": 1.2288, "step": 425 }, { "epoch": 0.16, "learning_rate": 1.9907082650757306e-05, "loss": 1.2696, "step": 426 }, { "epoch": 0.16, "learning_rate": 1.99062256404683e-05, "loss": 1.2495, "step": 427 }, { "epoch": 0.16, "learning_rate": 1.990536471466108e-05, "loss": 1.2875, "step": 428 }, { "epoch": 0.16, "learning_rate": 1.990449987367594e-05, "loss": 1.2496, "step": 429 }, { "epoch": 0.17, "learning_rate": 1.9903631117854713e-05, "loss": 1.2547, "step": 430 }, { "epoch": 0.17, "learning_rate": 1.990275844754078e-05, "loss": 1.2191, "step": 431 }, { "epoch": 0.17, "learning_rate": 1.9901881863079076e-05, "loss": 1.1999, "step": 432 }, { "epoch": 0.17, "learning_rate": 1.9901001364816075e-05, "loss": 1.2628, "step": 433 }, { "epoch": 0.17, "learning_rate": 1.99001169530998e-05, "loss": 1.2974, "step": 434 }, { "epoch": 0.17, "learning_rate": 1.9899228628279825e-05, "loss": 1.231, "step": 435 }, { "epoch": 0.17, "learning_rate": 1.9898336390707266e-05, "loss": 1.1927, "step": 436 }, { "epoch": 0.17, "learning_rate": 1.989744024073479e-05, "loss": 1.3103, "step": 437 }, { "epoch": 0.17, "learning_rate": 1.9896540178716604e-05, "loss": 1.269, "step": 438 }, { "epoch": 0.17, "learning_rate": 1.989563620500847e-05, "loss": 1.2436, "step": 439 }, { "epoch": 0.17, "learning_rate": 1.9894728319967683e-05, "loss": 1.2356, "step": 440 }, { "epoch": 0.17, "learning_rate": 1.9893816523953097e-05, "loss": 1.2951, "step": 441 }, { "epoch": 0.17, "learning_rate": 1.989290081732511e-05, "loss": 1.2473, "step": 442 }, { "epoch": 0.17, "learning_rate": 1.989198120044566e-05, "loss": 1.2311, "step": 443 }, { "epoch": 0.17, "learning_rate": 1.9891057673678235e-05, "loss": 1.2616, "step": 444 }, { "epoch": 0.17, "learning_rate": 1.989013023738786e-05, "loss": 1.2255, "step": 445 }, { "epoch": 0.17, "learning_rate": 1.988919889194112e-05, "loss": 1.2582, "step": 446 }, { "epoch": 0.17, "learning_rate": 1.9888263637706128e-05, "loss": 1.1917, "step": 447 }, { "epoch": 0.17, "learning_rate": 1.9887324475052564e-05, "loss": 1.2724, "step": 448 }, { "epoch": 0.17, "learning_rate": 1.9886381404351626e-05, "loss": 1.2663, "step": 449 }, { "epoch": 0.17, "learning_rate": 1.9885434425976082e-05, "loss": 1.2185, "step": 450 }, { "epoch": 0.17, "learning_rate": 1.9884483540300226e-05, "loss": 1.2272, "step": 451 }, { "epoch": 0.17, "learning_rate": 1.9883528747699906e-05, "loss": 1.2101, "step": 452 }, { "epoch": 0.17, "learning_rate": 1.988257004855251e-05, "loss": 1.1917, "step": 453 }, { "epoch": 0.17, "learning_rate": 1.9881607443236978e-05, "loss": 1.1786, "step": 454 }, { "epoch": 0.17, "learning_rate": 1.988064093213378e-05, "loss": 1.2896, "step": 455 }, { "epoch": 0.18, "learning_rate": 1.9879670515624937e-05, "loss": 1.1622, "step": 456 }, { "epoch": 0.18, "learning_rate": 1.987869619409402e-05, "loss": 1.285, "step": 457 }, { "epoch": 0.18, "learning_rate": 1.9877717967926132e-05, "loss": 1.2792, "step": 458 }, { "epoch": 0.18, "learning_rate": 1.9876735837507934e-05, "loss": 1.1994, "step": 459 }, { "epoch": 0.18, "learning_rate": 1.9875749803227613e-05, "loss": 1.2053, "step": 460 }, { "epoch": 0.18, "learning_rate": 1.9874759865474914e-05, "loss": 1.2735, "step": 461 }, { "epoch": 0.18, "learning_rate": 1.987376602464111e-05, "loss": 1.2156, "step": 462 }, { "epoch": 0.18, "learning_rate": 1.9872768281119027e-05, "loss": 1.1824, "step": 463 }, { "epoch": 0.18, "learning_rate": 1.9871766635303037e-05, "loss": 1.1847, "step": 464 }, { "epoch": 0.18, "learning_rate": 1.9870761087589044e-05, "loss": 1.2168, "step": 465 }, { "epoch": 0.18, "learning_rate": 1.9869751638374502e-05, "loss": 1.2297, "step": 466 }, { "epoch": 0.18, "learning_rate": 1.98687382880584e-05, "loss": 1.2365, "step": 467 }, { "epoch": 0.18, "learning_rate": 1.986772103704128e-05, "loss": 1.2129, "step": 468 }, { "epoch": 0.18, "learning_rate": 1.986669988572521e-05, "loss": 1.1969, "step": 469 }, { "epoch": 0.18, "learning_rate": 1.9865674834513815e-05, "loss": 1.1989, "step": 470 }, { "epoch": 0.18, "learning_rate": 1.9864645883812252e-05, "loss": 1.2236, "step": 471 }, { "epoch": 0.18, "learning_rate": 1.9863613034027224e-05, "loss": 1.2694, "step": 472 }, { "epoch": 0.18, "learning_rate": 1.9862576285566973e-05, "loss": 1.1584, "step": 473 }, { "epoch": 0.18, "learning_rate": 1.9861535638841284e-05, "loss": 1.2553, "step": 474 }, { "epoch": 0.18, "learning_rate": 1.986049109426148e-05, "loss": 1.185, "step": 475 }, { "epoch": 0.18, "learning_rate": 1.9859442652240424e-05, "loss": 1.2013, "step": 476 }, { "epoch": 0.18, "learning_rate": 1.9858390313192525e-05, "loss": 1.2487, "step": 477 }, { "epoch": 0.18, "learning_rate": 1.9857334077533727e-05, "loss": 1.2287, "step": 478 }, { "epoch": 0.18, "learning_rate": 1.9856273945681516e-05, "loss": 1.2406, "step": 479 }, { "epoch": 0.18, "learning_rate": 1.9855209918054916e-05, "loss": 1.2862, "step": 480 }, { "epoch": 0.18, "learning_rate": 1.9854141995074495e-05, "loss": 1.1864, "step": 481 }, { "epoch": 0.19, "learning_rate": 1.9853070177162358e-05, "loss": 1.2574, "step": 482 }, { "epoch": 0.19, "learning_rate": 1.9851994464742153e-05, "loss": 1.2301, "step": 483 }, { "epoch": 0.19, "learning_rate": 1.985091485823906e-05, "loss": 1.2024, "step": 484 }, { "epoch": 0.19, "learning_rate": 1.98498313580798e-05, "loss": 1.2676, "step": 485 }, { "epoch": 0.19, "learning_rate": 1.9848743964692643e-05, "loss": 1.1986, "step": 486 }, { "epoch": 0.19, "learning_rate": 1.9847652678507386e-05, "loss": 1.222, "step": 487 }, { "epoch": 0.19, "learning_rate": 1.984655749995537e-05, "loss": 1.2552, "step": 488 }, { "epoch": 0.19, "learning_rate": 1.9845458429469472e-05, "loss": 1.1667, "step": 489 }, { "epoch": 0.19, "learning_rate": 1.984435546748411e-05, "loss": 1.1648, "step": 490 }, { "epoch": 0.19, "learning_rate": 1.9843248614435238e-05, "loss": 1.236, "step": 491 }, { "epoch": 0.19, "learning_rate": 1.9842137870760348e-05, "loss": 1.2445, "step": 492 }, { "epoch": 0.19, "learning_rate": 1.9841023236898474e-05, "loss": 1.2038, "step": 493 }, { "epoch": 0.19, "learning_rate": 1.9839904713290186e-05, "loss": 1.224, "step": 494 }, { "epoch": 0.19, "learning_rate": 1.9838782300377583e-05, "loss": 1.3006, "step": 495 }, { "epoch": 0.19, "learning_rate": 1.983765599860431e-05, "loss": 1.2, "step": 496 }, { "epoch": 0.19, "learning_rate": 1.9836525808415554e-05, "loss": 1.2567, "step": 497 }, { "epoch": 0.19, "learning_rate": 1.9835391730258022e-05, "loss": 1.153, "step": 498 }, { "epoch": 0.19, "learning_rate": 1.9834253764579974e-05, "loss": 1.2293, "step": 499 }, { "epoch": 0.19, "learning_rate": 1.9833111911831196e-05, "loss": 1.212, "step": 500 }, { "epoch": 0.19, "learning_rate": 1.9831966172463023e-05, "loss": 1.2189, "step": 501 }, { "epoch": 0.19, "learning_rate": 1.9830816546928307e-05, "loss": 1.252, "step": 502 }, { "epoch": 0.19, "learning_rate": 1.9829663035681456e-05, "loss": 1.1805, "step": 503 }, { "epoch": 0.19, "learning_rate": 1.9828505639178403e-05, "loss": 1.2674, "step": 504 }, { "epoch": 0.19, "learning_rate": 1.9827344357876614e-05, "loss": 1.2015, "step": 505 }, { "epoch": 0.19, "learning_rate": 1.9826179192235102e-05, "loss": 1.2284, "step": 506 }, { "epoch": 0.19, "learning_rate": 1.9825010142714404e-05, "loss": 1.1657, "step": 507 }, { "epoch": 0.2, "learning_rate": 1.9823837209776596e-05, "loss": 1.2319, "step": 508 }, { "epoch": 0.2, "learning_rate": 1.9822660393885292e-05, "loss": 1.1924, "step": 509 }, { "epoch": 0.2, "learning_rate": 1.9821479695505633e-05, "loss": 1.2573, "step": 510 }, { "epoch": 0.2, "learning_rate": 1.982029511510431e-05, "loss": 1.261, "step": 511 }, { "epoch": 0.2, "learning_rate": 1.9819106653149533e-05, "loss": 1.2171, "step": 512 }, { "epoch": 0.2, "learning_rate": 1.9817914310111044e-05, "loss": 1.1973, "step": 513 }, { "epoch": 0.2, "learning_rate": 1.9816718086460138e-05, "loss": 1.2034, "step": 514 }, { "epoch": 0.2, "learning_rate": 1.981551798266963e-05, "loss": 1.218, "step": 515 }, { "epoch": 0.2, "learning_rate": 1.9814313999213865e-05, "loss": 1.2739, "step": 516 }, { "epoch": 0.2, "learning_rate": 1.9813106136568733e-05, "loss": 1.2838, "step": 517 }, { "epoch": 0.2, "learning_rate": 1.981189439521165e-05, "loss": 1.2433, "step": 518 }, { "epoch": 0.2, "learning_rate": 1.9810678775621564e-05, "loss": 1.2558, "step": 519 }, { "epoch": 0.2, "learning_rate": 1.9809459278278965e-05, "loss": 1.2454, "step": 520 }, { "epoch": 0.2, "learning_rate": 1.9808235903665865e-05, "loss": 1.2282, "step": 521 }, { "epoch": 0.2, "learning_rate": 1.9807008652265816e-05, "loss": 1.2266, "step": 522 }, { "epoch": 0.2, "learning_rate": 1.9805777524563896e-05, "loss": 1.2719, "step": 523 }, { "epoch": 0.2, "learning_rate": 1.980454252104672e-05, "loss": 1.2095, "step": 524 }, { "epoch": 0.2, "learning_rate": 1.9803303642202432e-05, "loss": 1.2454, "step": 525 }, { "epoch": 0.2, "learning_rate": 1.980206088852071e-05, "loss": 1.2077, "step": 526 }, { "epoch": 0.2, "learning_rate": 1.9800814260492766e-05, "loss": 1.2403, "step": 527 }, { "epoch": 0.2, "learning_rate": 1.979956375861134e-05, "loss": 1.2541, "step": 528 }, { "epoch": 0.2, "learning_rate": 1.9798309383370698e-05, "loss": 1.2614, "step": 529 }, { "epoch": 0.2, "learning_rate": 1.9797051135266642e-05, "loss": 1.2365, "step": 530 }, { "epoch": 0.2, "learning_rate": 1.979578901479651e-05, "loss": 1.2958, "step": 531 }, { "epoch": 0.2, "learning_rate": 1.9794523022459168e-05, "loss": 1.2385, "step": 532 }, { "epoch": 0.2, "learning_rate": 1.9793253158754998e-05, "loss": 1.2699, "step": 533 }, { "epoch": 0.21, "learning_rate": 1.9791979424185937e-05, "loss": 1.1559, "step": 534 }, { "epoch": 0.21, "learning_rate": 1.9790701819255433e-05, "loss": 1.2517, "step": 535 }, { "epoch": 0.21, "learning_rate": 1.978942034446847e-05, "loss": 1.2142, "step": 536 }, { "epoch": 0.21, "learning_rate": 1.9788135000331566e-05, "loss": 1.1642, "step": 537 }, { "epoch": 0.21, "learning_rate": 1.978684578735276e-05, "loss": 1.2663, "step": 538 }, { "epoch": 0.21, "learning_rate": 1.978555270604162e-05, "loss": 1.2663, "step": 539 }, { "epoch": 0.21, "learning_rate": 1.9784255756909255e-05, "loss": 1.2936, "step": 540 }, { "epoch": 0.21, "learning_rate": 1.9782954940468292e-05, "loss": 1.1335, "step": 541 }, { "epoch": 0.21, "learning_rate": 1.9781650257232892e-05, "loss": 1.2182, "step": 542 }, { "epoch": 0.21, "learning_rate": 1.9780341707718738e-05, "loss": 1.1952, "step": 543 }, { "epoch": 0.21, "learning_rate": 1.9779029292443045e-05, "loss": 1.1698, "step": 544 }, { "epoch": 0.21, "learning_rate": 1.9777713011924558e-05, "loss": 1.2487, "step": 545 }, { "epoch": 0.21, "learning_rate": 1.9776392866683547e-05, "loss": 1.1959, "step": 546 }, { "epoch": 0.21, "learning_rate": 1.9775068857241814e-05, "loss": 1.2181, "step": 547 }, { "epoch": 0.21, "learning_rate": 1.9773740984122675e-05, "loss": 1.2124, "step": 548 }, { "epoch": 0.21, "learning_rate": 1.9772409247850994e-05, "loss": 1.2779, "step": 549 }, { "epoch": 0.21, "learning_rate": 1.9771073648953148e-05, "loss": 1.1686, "step": 550 }, { "epoch": 0.21, "learning_rate": 1.976973418795704e-05, "loss": 1.1696, "step": 551 }, { "epoch": 0.21, "learning_rate": 1.9768390865392103e-05, "loss": 1.242, "step": 552 }, { "epoch": 0.21, "learning_rate": 1.97670436817893e-05, "loss": 1.2169, "step": 553 }, { "epoch": 0.21, "learning_rate": 1.9765692637681115e-05, "loss": 1.2756, "step": 554 }, { "epoch": 0.21, "learning_rate": 1.9764337733601562e-05, "loss": 1.1093, "step": 555 }, { "epoch": 0.21, "learning_rate": 1.9762978970086174e-05, "loss": 1.2024, "step": 556 }, { "epoch": 0.21, "learning_rate": 1.9761616347672013e-05, "loss": 1.1661, "step": 557 }, { "epoch": 0.21, "learning_rate": 1.976024986689767e-05, "loss": 1.1878, "step": 558 }, { "epoch": 0.21, "learning_rate": 1.975887952830326e-05, "loss": 1.2344, "step": 559 }, { "epoch": 0.22, "learning_rate": 1.9757505332430415e-05, "loss": 1.2396, "step": 560 }, { "epoch": 0.22, "learning_rate": 1.97561272798223e-05, "loss": 1.2388, "step": 561 }, { "epoch": 0.22, "learning_rate": 1.97547453710236e-05, "loss": 1.2512, "step": 562 }, { "epoch": 0.22, "learning_rate": 1.975335960658053e-05, "loss": 1.2595, "step": 563 }, { "epoch": 0.22, "learning_rate": 1.9751969987040827e-05, "loss": 1.2593, "step": 564 }, { "epoch": 0.22, "learning_rate": 1.975057651295374e-05, "loss": 1.241, "step": 565 }, { "epoch": 0.22, "learning_rate": 1.9749179184870057e-05, "loss": 1.1925, "step": 566 }, { "epoch": 0.22, "learning_rate": 1.9747778003342084e-05, "loss": 1.1714, "step": 567 }, { "epoch": 0.22, "learning_rate": 1.974637296892365e-05, "loss": 1.2442, "step": 568 }, { "epoch": 0.22, "learning_rate": 1.9744964082170103e-05, "loss": 1.2262, "step": 569 }, { "epoch": 0.22, "learning_rate": 1.9743551343638324e-05, "loss": 1.2131, "step": 570 }, { "epoch": 0.22, "learning_rate": 1.97421347538867e-05, "loss": 1.2208, "step": 571 }, { "epoch": 0.22, "learning_rate": 1.974071431347516e-05, "loss": 1.234, "step": 572 }, { "epoch": 0.22, "learning_rate": 1.973929002296514e-05, "loss": 1.1714, "step": 573 }, { "epoch": 0.22, "learning_rate": 1.97378618829196e-05, "loss": 1.2325, "step": 574 }, { "epoch": 0.22, "learning_rate": 1.973642989390303e-05, "loss": 1.2128, "step": 575 }, { "epoch": 0.22, "learning_rate": 1.973499405648143e-05, "loss": 1.2027, "step": 576 }, { "epoch": 0.22, "learning_rate": 1.973355437122233e-05, "loss": 1.2385, "step": 577 }, { "epoch": 0.22, "learning_rate": 1.9732110838694776e-05, "loss": 1.1782, "step": 578 }, { "epoch": 0.22, "learning_rate": 1.9730663459469342e-05, "loss": 1.2361, "step": 579 }, { "epoch": 0.22, "learning_rate": 1.9729212234118108e-05, "loss": 1.2181, "step": 580 }, { "epoch": 0.22, "learning_rate": 1.9727757163214687e-05, "loss": 1.203, "step": 581 }, { "epoch": 0.22, "learning_rate": 1.972629824733421e-05, "loss": 1.1753, "step": 582 }, { "epoch": 0.22, "learning_rate": 1.972483548705332e-05, "loss": 1.1721, "step": 583 }, { "epoch": 0.22, "learning_rate": 1.972336888295019e-05, "loss": 1.1944, "step": 584 }, { "epoch": 0.22, "learning_rate": 1.9721898435604507e-05, "loss": 1.2166, "step": 585 }, { "epoch": 0.23, "learning_rate": 1.9720424145597477e-05, "loss": 1.2197, "step": 586 }, { "epoch": 0.23, "learning_rate": 1.9718946013511827e-05, "loss": 1.2692, "step": 587 }, { "epoch": 0.23, "learning_rate": 1.9717464039931797e-05, "loss": 1.2583, "step": 588 }, { "epoch": 0.23, "learning_rate": 1.9715978225443147e-05, "loss": 1.1466, "step": 589 }, { "epoch": 0.23, "learning_rate": 1.9714488570633167e-05, "loss": 1.2399, "step": 590 }, { "epoch": 0.23, "learning_rate": 1.9712995076090652e-05, "loss": 1.1788, "step": 591 }, { "epoch": 0.23, "learning_rate": 1.9711497742405914e-05, "loss": 1.2216, "step": 592 }, { "epoch": 0.23, "learning_rate": 1.9709996570170787e-05, "loss": 1.2018, "step": 593 }, { "epoch": 0.23, "learning_rate": 1.970849155997863e-05, "loss": 1.2161, "step": 594 }, { "epoch": 0.23, "learning_rate": 1.97069827124243e-05, "loss": 1.1946, "step": 595 }, { "epoch": 0.23, "learning_rate": 1.970547002810419e-05, "loss": 1.1743, "step": 596 }, { "epoch": 0.23, "learning_rate": 1.9703953507616194e-05, "loss": 1.2005, "step": 597 }, { "epoch": 0.23, "learning_rate": 1.9702433151559734e-05, "loss": 1.249, "step": 598 }, { "epoch": 0.23, "learning_rate": 1.9700908960535747e-05, "loss": 1.2067, "step": 599 }, { "epoch": 0.23, "learning_rate": 1.9699380935146673e-05, "loss": 1.2253, "step": 600 }, { "epoch": 0.23, "learning_rate": 1.969784907599648e-05, "loss": 1.2322, "step": 601 }, { "epoch": 0.23, "learning_rate": 1.9696313383690653e-05, "loss": 1.225, "step": 602 }, { "epoch": 0.23, "learning_rate": 1.969477385883618e-05, "loss": 1.2689, "step": 603 }, { "epoch": 0.23, "learning_rate": 1.969323050204158e-05, "loss": 1.1557, "step": 604 }, { "epoch": 0.23, "learning_rate": 1.9691683313916868e-05, "loss": 1.1723, "step": 605 }, { "epoch": 0.23, "learning_rate": 1.969013229507359e-05, "loss": 1.2585, "step": 606 }, { "epoch": 0.23, "learning_rate": 1.9688577446124793e-05, "loss": 1.2282, "step": 607 }, { "epoch": 0.23, "learning_rate": 1.9687018767685048e-05, "loss": 1.22, "step": 608 }, { "epoch": 0.23, "learning_rate": 1.9685456260370436e-05, "loss": 1.2148, "step": 609 }, { "epoch": 0.23, "learning_rate": 1.968388992479855e-05, "loss": 1.1696, "step": 610 }, { "epoch": 0.23, "learning_rate": 1.9682319761588497e-05, "loss": 1.1694, "step": 611 }, { "epoch": 0.24, "learning_rate": 1.96807457713609e-05, "loss": 1.1621, "step": 612 }, { "epoch": 0.24, "learning_rate": 1.9679167954737885e-05, "loss": 1.1856, "step": 613 }, { "epoch": 0.24, "learning_rate": 1.9677586312343102e-05, "loss": 1.2265, "step": 614 }, { "epoch": 0.24, "learning_rate": 1.967600084480171e-05, "loss": 1.1323, "step": 615 }, { "epoch": 0.24, "learning_rate": 1.9674411552740375e-05, "loss": 1.1765, "step": 616 }, { "epoch": 0.24, "learning_rate": 1.967281843678728e-05, "loss": 1.2363, "step": 617 }, { "epoch": 0.24, "learning_rate": 1.967122149757211e-05, "loss": 1.2093, "step": 618 }, { "epoch": 0.24, "learning_rate": 1.966962073572608e-05, "loss": 1.1752, "step": 619 }, { "epoch": 0.24, "learning_rate": 1.9668016151881897e-05, "loss": 1.2025, "step": 620 }, { "epoch": 0.24, "learning_rate": 1.9666407746673788e-05, "loss": 1.1415, "step": 621 }, { "epoch": 0.24, "learning_rate": 1.966479552073749e-05, "loss": 1.1899, "step": 622 }, { "epoch": 0.24, "learning_rate": 1.9663179474710248e-05, "loss": 1.2198, "step": 623 }, { "epoch": 0.24, "learning_rate": 1.9661559609230815e-05, "loss": 1.1625, "step": 624 }, { "epoch": 0.24, "learning_rate": 1.965993592493946e-05, "loss": 1.1888, "step": 625 }, { "epoch": 0.24, "learning_rate": 1.9658308422477957e-05, "loss": 1.1439, "step": 626 }, { "epoch": 0.24, "learning_rate": 1.965667710248959e-05, "loss": 1.2258, "step": 627 }, { "epoch": 0.24, "learning_rate": 1.9655041965619145e-05, "loss": 1.2261, "step": 628 }, { "epoch": 0.24, "learning_rate": 1.9653403012512935e-05, "loss": 1.2088, "step": 629 }, { "epoch": 0.24, "learning_rate": 1.9651760243818764e-05, "loss": 1.2444, "step": 630 }, { "epoch": 0.24, "learning_rate": 1.965011366018595e-05, "loss": 1.2635, "step": 631 }, { "epoch": 0.24, "learning_rate": 1.9648463262265316e-05, "loss": 1.219, "step": 632 }, { "epoch": 0.24, "learning_rate": 1.96468090507092e-05, "loss": 1.2458, "step": 633 }, { "epoch": 0.24, "learning_rate": 1.9645151026171443e-05, "loss": 1.1852, "step": 634 }, { "epoch": 0.24, "learning_rate": 1.964348918930739e-05, "loss": 1.2324, "step": 635 }, { "epoch": 0.24, "learning_rate": 1.96418235407739e-05, "loss": 1.194, "step": 636 }, { "epoch": 0.24, "learning_rate": 1.964015408122933e-05, "loss": 1.2767, "step": 637 }, { "epoch": 0.25, "learning_rate": 1.963848081133355e-05, "loss": 1.1912, "step": 638 }, { "epoch": 0.25, "learning_rate": 1.9636803731747934e-05, "loss": 1.186, "step": 639 }, { "epoch": 0.25, "learning_rate": 1.963512284313536e-05, "loss": 1.1964, "step": 640 }, { "epoch": 0.25, "learning_rate": 1.963343814616022e-05, "loss": 1.1906, "step": 641 }, { "epoch": 0.25, "learning_rate": 1.9631749641488393e-05, "loss": 1.1312, "step": 642 }, { "epoch": 0.25, "learning_rate": 1.9630057329787283e-05, "loss": 1.2527, "step": 643 }, { "epoch": 0.25, "learning_rate": 1.9628361211725785e-05, "loss": 1.2341, "step": 644 }, { "epoch": 0.25, "learning_rate": 1.962666128797431e-05, "loss": 1.2221, "step": 645 }, { "epoch": 0.25, "learning_rate": 1.9624957559204763e-05, "loss": 1.1799, "step": 646 }, { "epoch": 0.25, "learning_rate": 1.962325002609056e-05, "loss": 1.1514, "step": 647 }, { "epoch": 0.25, "learning_rate": 1.9621538689306606e-05, "loss": 1.3231, "step": 648 }, { "epoch": 0.25, "learning_rate": 1.961982354952934e-05, "loss": 1.1969, "step": 649 }, { "epoch": 0.25, "learning_rate": 1.9618104607436673e-05, "loss": 1.1474, "step": 650 }, { "epoch": 0.25, "learning_rate": 1.9616381863708033e-05, "loss": 1.1693, "step": 651 }, { "epoch": 0.25, "learning_rate": 1.9614655319024352e-05, "loss": 1.2187, "step": 652 }, { "epoch": 0.25, "learning_rate": 1.9612924974068054e-05, "loss": 1.1745, "step": 653 }, { "epoch": 0.25, "learning_rate": 1.9611190829523084e-05, "loss": 1.1978, "step": 654 }, { "epoch": 0.25, "learning_rate": 1.9609452886074865e-05, "loss": 1.179, "step": 655 }, { "epoch": 0.25, "learning_rate": 1.960771114441034e-05, "loss": 1.1839, "step": 656 }, { "epoch": 0.25, "learning_rate": 1.9605965605217945e-05, "loss": 1.1795, "step": 657 }, { "epoch": 0.25, "learning_rate": 1.9604216269187624e-05, "loss": 1.1962, "step": 658 }, { "epoch": 0.25, "learning_rate": 1.960246313701081e-05, "loss": 1.163, "step": 659 }, { "epoch": 0.25, "learning_rate": 1.9600706209380443e-05, "loss": 1.1832, "step": 660 }, { "epoch": 0.25, "learning_rate": 1.9598945486990967e-05, "loss": 1.1598, "step": 661 }, { "epoch": 0.25, "learning_rate": 1.9597180970538325e-05, "loss": 1.2693, "step": 662 }, { "epoch": 0.25, "learning_rate": 1.9595412660719944e-05, "loss": 1.1502, "step": 663 }, { "epoch": 0.26, "learning_rate": 1.959364055823478e-05, "loss": 1.1614, "step": 664 }, { "epoch": 0.26, "learning_rate": 1.959186466378326e-05, "loss": 1.1894, "step": 665 }, { "epoch": 0.26, "learning_rate": 1.9590084978067325e-05, "loss": 1.2362, "step": 666 }, { "epoch": 0.26, "learning_rate": 1.958830150179041e-05, "loss": 1.2169, "step": 667 }, { "epoch": 0.26, "learning_rate": 1.958651423565745e-05, "loss": 1.1937, "step": 668 }, { "epoch": 0.26, "learning_rate": 1.9584723180374872e-05, "loss": 1.213, "step": 669 }, { "epoch": 0.26, "learning_rate": 1.9582928336650614e-05, "loss": 1.1167, "step": 670 }, { "epoch": 0.26, "learning_rate": 1.9581129705194093e-05, "loss": 1.1645, "step": 671 }, { "epoch": 0.26, "learning_rate": 1.9579327286716243e-05, "loss": 1.2271, "step": 672 }, { "epoch": 0.26, "learning_rate": 1.9577521081929473e-05, "loss": 1.1785, "step": 673 }, { "epoch": 0.26, "learning_rate": 1.9575711091547714e-05, "loss": 1.1881, "step": 674 }, { "epoch": 0.26, "learning_rate": 1.957389731628637e-05, "loss": 1.2261, "step": 675 }, { "epoch": 0.26, "learning_rate": 1.9572079756862354e-05, "loss": 1.1762, "step": 676 }, { "epoch": 0.26, "learning_rate": 1.957025841399407e-05, "loss": 1.1436, "step": 677 }, { "epoch": 0.26, "learning_rate": 1.9568433288401424e-05, "loss": 1.1939, "step": 678 }, { "epoch": 0.26, "learning_rate": 1.9566604380805807e-05, "loss": 1.1866, "step": 679 }, { "epoch": 0.26, "learning_rate": 1.9564771691930112e-05, "loss": 1.197, "step": 680 }, { "epoch": 0.26, "learning_rate": 1.9562935222498724e-05, "loss": 1.1756, "step": 681 }, { "epoch": 0.26, "learning_rate": 1.9561094973237524e-05, "loss": 1.2379, "step": 682 }, { "epoch": 0.26, "learning_rate": 1.9559250944873883e-05, "loss": 1.2383, "step": 683 }, { "epoch": 0.26, "learning_rate": 1.9557403138136672e-05, "loss": 1.1893, "step": 684 }, { "epoch": 0.26, "learning_rate": 1.9555551553756248e-05, "loss": 1.2241, "step": 685 }, { "epoch": 0.26, "learning_rate": 1.9553696192464474e-05, "loss": 1.1891, "step": 686 }, { "epoch": 0.26, "learning_rate": 1.955183705499468e-05, "loss": 1.1673, "step": 687 }, { "epoch": 0.26, "learning_rate": 1.9549974142081723e-05, "loss": 1.1967, "step": 688 }, { "epoch": 0.26, "learning_rate": 1.9548107454461927e-05, "loss": 1.1504, "step": 689 }, { "epoch": 0.27, "learning_rate": 1.954623699287312e-05, "loss": 1.2405, "step": 690 }, { "epoch": 0.27, "learning_rate": 1.9544362758054608e-05, "loss": 1.2022, "step": 691 }, { "epoch": 0.27, "learning_rate": 1.954248475074721e-05, "loss": 1.1706, "step": 692 }, { "epoch": 0.27, "learning_rate": 1.9540602971693215e-05, "loss": 1.1845, "step": 693 }, { "epoch": 0.27, "learning_rate": 1.9538717421636412e-05, "loss": 1.2365, "step": 694 }, { "epoch": 0.27, "learning_rate": 1.9536828101322087e-05, "loss": 1.2525, "step": 695 }, { "epoch": 0.27, "learning_rate": 1.953493501149701e-05, "loss": 1.1922, "step": 696 }, { "epoch": 0.27, "learning_rate": 1.9533038152909428e-05, "loss": 1.2053, "step": 697 }, { "epoch": 0.27, "learning_rate": 1.9531137526309104e-05, "loss": 1.2055, "step": 698 }, { "epoch": 0.27, "learning_rate": 1.9529233132447275e-05, "loss": 1.203, "step": 699 }, { "epoch": 0.27, "learning_rate": 1.9527324972076658e-05, "loss": 1.2084, "step": 700 }, { "epoch": 0.27, "learning_rate": 1.9525413045951476e-05, "loss": 1.1638, "step": 701 }, { "epoch": 0.27, "learning_rate": 1.952349735482744e-05, "loss": 1.1726, "step": 702 }, { "epoch": 0.27, "learning_rate": 1.9521577899461732e-05, "loss": 1.1643, "step": 703 }, { "epoch": 0.27, "learning_rate": 1.951965468061304e-05, "loss": 1.2236, "step": 704 }, { "epoch": 0.27, "learning_rate": 1.9517727699041528e-05, "loss": 1.1786, "step": 705 }, { "epoch": 0.27, "learning_rate": 1.951579695550885e-05, "loss": 1.1789, "step": 706 }, { "epoch": 0.27, "learning_rate": 1.9513862450778157e-05, "loss": 1.1746, "step": 707 }, { "epoch": 0.27, "learning_rate": 1.951192418561407e-05, "loss": 1.1787, "step": 708 }, { "epoch": 0.27, "learning_rate": 1.950998216078271e-05, "loss": 1.1081, "step": 709 }, { "epoch": 0.27, "learning_rate": 1.950803637705167e-05, "loss": 1.2522, "step": 710 }, { "epoch": 0.27, "learning_rate": 1.950608683519005e-05, "loss": 1.2191, "step": 711 }, { "epoch": 0.27, "learning_rate": 1.950413353596841e-05, "loss": 1.17, "step": 712 }, { "epoch": 0.27, "learning_rate": 1.9502176480158816e-05, "loss": 1.2141, "step": 713 }, { "epoch": 0.27, "learning_rate": 1.950021566853481e-05, "loss": 1.2216, "step": 714 }, { "epoch": 0.27, "learning_rate": 1.949825110187141e-05, "loss": 1.1939, "step": 715 }, { "epoch": 0.28, "learning_rate": 1.9496282780945145e-05, "loss": 1.2477, "step": 716 }, { "epoch": 0.28, "learning_rate": 1.949431070653399e-05, "loss": 1.174, "step": 717 }, { "epoch": 0.28, "learning_rate": 1.9492334879417442e-05, "loss": 1.263, "step": 718 }, { "epoch": 0.28, "learning_rate": 1.949035530037645e-05, "loss": 1.175, "step": 719 }, { "epoch": 0.28, "learning_rate": 1.948837197019346e-05, "loss": 1.1882, "step": 720 }, { "epoch": 0.28, "learning_rate": 1.948638488965241e-05, "loss": 1.1283, "step": 721 }, { "epoch": 0.28, "learning_rate": 1.94843940595387e-05, "loss": 1.1622, "step": 722 }, { "epoch": 0.28, "learning_rate": 1.9482399480639226e-05, "loss": 1.156, "step": 723 }, { "epoch": 0.28, "learning_rate": 1.9480401153742365e-05, "loss": 1.1979, "step": 724 }, { "epoch": 0.28, "learning_rate": 1.9478399079637967e-05, "loss": 1.1627, "step": 725 }, { "epoch": 0.28, "learning_rate": 1.9476393259117373e-05, "loss": 1.1787, "step": 726 }, { "epoch": 0.28, "learning_rate": 1.9474383692973397e-05, "loss": 1.2115, "step": 727 }, { "epoch": 0.28, "learning_rate": 1.9472370382000334e-05, "loss": 1.1858, "step": 728 }, { "epoch": 0.28, "learning_rate": 1.9470353326993966e-05, "loss": 1.2394, "step": 729 }, { "epoch": 0.28, "learning_rate": 1.9468332528751554e-05, "loss": 1.2074, "step": 730 }, { "epoch": 0.28, "learning_rate": 1.946630798807183e-05, "loss": 1.1511, "step": 731 }, { "epoch": 0.28, "learning_rate": 1.9464279705755013e-05, "loss": 1.1469, "step": 732 }, { "epoch": 0.28, "learning_rate": 1.94622476826028e-05, "loss": 1.159, "step": 733 }, { "epoch": 0.28, "learning_rate": 1.946021191941836e-05, "loss": 1.2198, "step": 734 }, { "epoch": 0.28, "learning_rate": 1.9458172417006347e-05, "loss": 1.1847, "step": 735 }, { "epoch": 0.28, "learning_rate": 1.9456129176172896e-05, "loss": 1.16, "step": 736 }, { "epoch": 0.28, "learning_rate": 1.945408219772561e-05, "loss": 1.1704, "step": 737 }, { "epoch": 0.28, "learning_rate": 1.9452031482473576e-05, "loss": 1.1264, "step": 738 }, { "epoch": 0.28, "learning_rate": 1.9449977031227356e-05, "loss": 1.1334, "step": 739 }, { "epoch": 0.28, "learning_rate": 1.944791884479899e-05, "loss": 1.1818, "step": 740 }, { "epoch": 0.28, "learning_rate": 1.944585692400199e-05, "loss": 1.1869, "step": 741 }, { "epoch": 0.29, "learning_rate": 1.944379126965135e-05, "loss": 1.185, "step": 742 }, { "epoch": 0.29, "learning_rate": 1.9441721882563538e-05, "loss": 1.2473, "step": 743 }, { "epoch": 0.29, "learning_rate": 1.9439648763556492e-05, "loss": 1.2658, "step": 744 }, { "epoch": 0.29, "learning_rate": 1.9437571913449637e-05, "loss": 1.2407, "step": 745 }, { "epoch": 0.29, "learning_rate": 1.943549133306386e-05, "loss": 1.2107, "step": 746 }, { "epoch": 0.29, "learning_rate": 1.9433407023221532e-05, "loss": 1.1669, "step": 747 }, { "epoch": 0.29, "learning_rate": 1.943131898474649e-05, "loss": 1.2241, "step": 748 }, { "epoch": 0.29, "learning_rate": 1.9429227218464045e-05, "loss": 1.2147, "step": 749 }, { "epoch": 0.29, "learning_rate": 1.942713172520099e-05, "loss": 1.1293, "step": 750 }, { "epoch": 0.29, "learning_rate": 1.9425032505785587e-05, "loss": 1.2133, "step": 751 }, { "epoch": 0.29, "learning_rate": 1.942292956104757e-05, "loss": 1.2267, "step": 752 }, { "epoch": 0.29, "learning_rate": 1.9420822891818143e-05, "loss": 1.1842, "step": 753 }, { "epoch": 0.29, "learning_rate": 1.9418712498929986e-05, "loss": 1.117, "step": 754 }, { "epoch": 0.29, "learning_rate": 1.941659838321725e-05, "loss": 1.1729, "step": 755 }, { "epoch": 0.29, "learning_rate": 1.941448054551555e-05, "loss": 1.1569, "step": 756 }, { "epoch": 0.29, "learning_rate": 1.941235898666199e-05, "loss": 1.1886, "step": 757 }, { "epoch": 0.29, "learning_rate": 1.941023370749513e-05, "loss": 1.1623, "step": 758 }, { "epoch": 0.29, "learning_rate": 1.9408104708855e-05, "loss": 1.2279, "step": 759 }, { "epoch": 0.29, "learning_rate": 1.940597199158311e-05, "loss": 1.2121, "step": 760 }, { "epoch": 0.29, "learning_rate": 1.9403835556522426e-05, "loss": 1.1943, "step": 761 }, { "epoch": 0.29, "learning_rate": 1.94016954045174e-05, "loss": 1.2125, "step": 762 }, { "epoch": 0.29, "learning_rate": 1.9399551536413943e-05, "loss": 1.1554, "step": 763 }, { "epoch": 0.29, "learning_rate": 1.9397403953059436e-05, "loss": 1.1661, "step": 764 }, { "epoch": 0.29, "learning_rate": 1.939525265530273e-05, "loss": 1.1793, "step": 765 }, { "epoch": 0.29, "learning_rate": 1.939309764399414e-05, "loss": 1.2267, "step": 766 }, { "epoch": 0.29, "learning_rate": 1.9390938919985457e-05, "loss": 1.1737, "step": 767 }, { "epoch": 0.3, "learning_rate": 1.9388776484129932e-05, "loss": 1.153, "step": 768 }, { "epoch": 0.3, "learning_rate": 1.9386610337282284e-05, "loss": 1.1453, "step": 769 }, { "epoch": 0.3, "learning_rate": 1.9384440480298706e-05, "loss": 1.1621, "step": 770 }, { "epoch": 0.3, "learning_rate": 1.9382266914036848e-05, "loss": 1.2469, "step": 771 }, { "epoch": 0.3, "learning_rate": 1.938008963935583e-05, "loss": 1.1882, "step": 772 }, { "epoch": 0.3, "learning_rate": 1.937790865711624e-05, "loss": 1.2329, "step": 773 }, { "epoch": 0.3, "learning_rate": 1.9375723968180127e-05, "loss": 1.1778, "step": 774 }, { "epoch": 0.3, "learning_rate": 1.9373535573411008e-05, "loss": 1.2023, "step": 775 }, { "epoch": 0.3, "learning_rate": 1.9371343473673868e-05, "loss": 1.2037, "step": 776 }, { "epoch": 0.3, "learning_rate": 1.9369147669835148e-05, "loss": 1.1827, "step": 777 }, { "epoch": 0.3, "learning_rate": 1.9366948162762758e-05, "loss": 1.1976, "step": 778 }, { "epoch": 0.3, "learning_rate": 1.9364744953326077e-05, "loss": 1.1825, "step": 779 }, { "epoch": 0.3, "learning_rate": 1.9362538042395934e-05, "loss": 1.1324, "step": 780 }, { "epoch": 0.3, "learning_rate": 1.9360327430844636e-05, "loss": 1.1728, "step": 781 }, { "epoch": 0.3, "learning_rate": 1.935811311954594e-05, "loss": 1.21, "step": 782 }, { "epoch": 0.3, "learning_rate": 1.9355895109375073e-05, "loss": 1.1454, "step": 783 }, { "epoch": 0.3, "learning_rate": 1.9353673401208726e-05, "loss": 1.1422, "step": 784 }, { "epoch": 0.3, "learning_rate": 1.935144799592504e-05, "loss": 1.1762, "step": 785 }, { "epoch": 0.3, "learning_rate": 1.934921889440363e-05, "loss": 1.1752, "step": 786 }, { "epoch": 0.3, "learning_rate": 1.9346986097525572e-05, "loss": 1.2365, "step": 787 }, { "epoch": 0.3, "learning_rate": 1.9344749606173384e-05, "loss": 1.2373, "step": 788 }, { "epoch": 0.3, "learning_rate": 1.9342509421231068e-05, "loss": 1.201, "step": 789 }, { "epoch": 0.3, "learning_rate": 1.9340265543584074e-05, "loss": 1.2608, "step": 790 }, { "epoch": 0.3, "learning_rate": 1.933801797411931e-05, "loss": 1.1715, "step": 791 }, { "epoch": 0.3, "learning_rate": 1.9335766713725153e-05, "loss": 1.2136, "step": 792 }, { "epoch": 0.3, "learning_rate": 1.9333511763291425e-05, "loss": 1.1846, "step": 793 }, { "epoch": 0.31, "learning_rate": 1.933125312370942e-05, "loss": 1.1516, "step": 794 }, { "epoch": 0.31, "learning_rate": 1.932899079587188e-05, "loss": 1.2197, "step": 795 }, { "epoch": 0.31, "learning_rate": 1.9326724780673013e-05, "loss": 1.22, "step": 796 }, { "epoch": 0.31, "learning_rate": 1.9324455079008473e-05, "loss": 1.1328, "step": 797 }, { "epoch": 0.31, "learning_rate": 1.9322181691775387e-05, "loss": 1.216, "step": 798 }, { "epoch": 0.31, "learning_rate": 1.9319904619872326e-05, "loss": 1.2601, "step": 799 }, { "epoch": 0.31, "learning_rate": 1.9317623864199323e-05, "loss": 1.1916, "step": 800 }, { "epoch": 0.31, "eval_loss": 1.1902344226837158, "eval_runtime": 688.5762, "eval_samples_per_second": 4.939, "eval_steps_per_second": 0.309, "step": 800 }, { "epoch": 0.31, "learning_rate": 1.9315339425657866e-05, "loss": 1.1631, "step": 801 }, { "epoch": 0.31, "learning_rate": 1.9313051305150893e-05, "loss": 1.1274, "step": 802 }, { "epoch": 0.31, "learning_rate": 1.931075950358281e-05, "loss": 1.19, "step": 803 }, { "epoch": 0.31, "learning_rate": 1.930846402185947e-05, "loss": 1.1627, "step": 804 }, { "epoch": 0.31, "learning_rate": 1.9306164860888173e-05, "loss": 1.1731, "step": 805 }, { "epoch": 0.31, "learning_rate": 1.930386202157769e-05, "loss": 1.1767, "step": 806 }, { "epoch": 0.31, "learning_rate": 1.9301555504838235e-05, "loss": 1.1277, "step": 807 }, { "epoch": 0.31, "learning_rate": 1.929924531158148e-05, "loss": 1.2038, "step": 808 }, { "epoch": 0.31, "learning_rate": 1.9296931442720543e-05, "loss": 1.1958, "step": 809 }, { "epoch": 0.31, "learning_rate": 1.9294613899170006e-05, "loss": 1.1761, "step": 810 }, { "epoch": 0.31, "learning_rate": 1.929229268184589e-05, "loss": 1.2013, "step": 811 }, { "epoch": 0.31, "learning_rate": 1.9289967791665683e-05, "loss": 1.1871, "step": 812 }, { "epoch": 0.31, "learning_rate": 1.9287639229548312e-05, "loss": 1.1975, "step": 813 }, { "epoch": 0.31, "learning_rate": 1.9285306996414163e-05, "loss": 1.2225, "step": 814 }, { "epoch": 0.31, "learning_rate": 1.928297109318507e-05, "loss": 1.1226, "step": 815 }, { "epoch": 0.31, "learning_rate": 1.928063152078432e-05, "loss": 1.2049, "step": 816 }, { "epoch": 0.31, "learning_rate": 1.9278288280136647e-05, "loss": 1.1523, "step": 817 }, { "epoch": 0.31, "learning_rate": 1.9275941372168234e-05, "loss": 1.2018, "step": 818 }, { "epoch": 0.31, "learning_rate": 1.927359079780672e-05, "loss": 1.1617, "step": 819 }, { "epoch": 0.31, "learning_rate": 1.9271236557981187e-05, "loss": 1.1392, "step": 820 }, { "epoch": 0.32, "learning_rate": 1.926887865362217e-05, "loss": 1.1532, "step": 821 }, { "epoch": 0.32, "learning_rate": 1.9266517085661644e-05, "loss": 1.2109, "step": 822 }, { "epoch": 0.32, "learning_rate": 1.9264151855033045e-05, "loss": 1.1942, "step": 823 }, { "epoch": 0.32, "learning_rate": 1.926178296267125e-05, "loss": 1.1592, "step": 824 }, { "epoch": 0.32, "learning_rate": 1.925941040951258e-05, "loss": 1.1505, "step": 825 }, { "epoch": 0.32, "learning_rate": 1.9257034196494807e-05, "loss": 1.1498, "step": 826 }, { "epoch": 0.32, "learning_rate": 1.9254654324557148e-05, "loss": 1.1406, "step": 827 }, { "epoch": 0.32, "learning_rate": 1.9252270794640273e-05, "loss": 1.2221, "step": 828 }, { "epoch": 0.32, "learning_rate": 1.924988360768629e-05, "loss": 1.1537, "step": 829 }, { "epoch": 0.32, "learning_rate": 1.9247492764638745e-05, "loss": 1.1908, "step": 830 }, { "epoch": 0.32, "learning_rate": 1.924509826644265e-05, "loss": 1.1763, "step": 831 }, { "epoch": 0.32, "learning_rate": 1.924270011404445e-05, "loss": 1.2013, "step": 832 }, { "epoch": 0.32, "learning_rate": 1.924029830839203e-05, "loss": 1.1869, "step": 833 }, { "epoch": 0.32, "learning_rate": 1.9237892850434722e-05, "loss": 1.1425, "step": 834 }, { "epoch": 0.32, "learning_rate": 1.9235483741123308e-05, "loss": 1.1912, "step": 835 }, { "epoch": 0.32, "learning_rate": 1.9233070981410007e-05, "loss": 1.2219, "step": 836 }, { "epoch": 0.32, "learning_rate": 1.9230654572248483e-05, "loss": 1.179, "step": 837 }, { "epoch": 0.32, "learning_rate": 1.922823451459384e-05, "loss": 1.1821, "step": 838 }, { "epoch": 0.32, "learning_rate": 1.9225810809402626e-05, "loss": 1.1434, "step": 839 }, { "epoch": 0.32, "learning_rate": 1.9223383457632833e-05, "loss": 1.2142, "step": 840 }, { "epoch": 0.32, "learning_rate": 1.922095246024389e-05, "loss": 1.1873, "step": 841 }, { "epoch": 0.32, "learning_rate": 1.921851781819667e-05, "loss": 1.1965, "step": 842 }, { "epoch": 0.32, "learning_rate": 1.9216079532453486e-05, "loss": 1.1994, "step": 843 }, { "epoch": 0.32, "learning_rate": 1.921363760397809e-05, "loss": 1.212, "step": 844 }, { "epoch": 0.32, "learning_rate": 1.9211192033735674e-05, "loss": 1.2031, "step": 845 }, { "epoch": 0.32, "learning_rate": 1.9208742822692872e-05, "loss": 1.1642, "step": 846 }, { "epoch": 0.33, "learning_rate": 1.9206289971817752e-05, "loss": 1.2395, "step": 847 }, { "epoch": 0.33, "learning_rate": 1.9203833482079824e-05, "loss": 1.2109, "step": 848 }, { "epoch": 0.33, "learning_rate": 1.920137335445004e-05, "loss": 1.1418, "step": 849 }, { "epoch": 0.33, "learning_rate": 1.9198909589900783e-05, "loss": 1.2393, "step": 850 }, { "epoch": 0.33, "learning_rate": 1.9196442189405874e-05, "loss": 1.1993, "step": 851 }, { "epoch": 0.33, "learning_rate": 1.919397115394058e-05, "loss": 1.1886, "step": 852 }, { "epoch": 0.33, "learning_rate": 1.9191496484481592e-05, "loss": 1.1387, "step": 853 }, { "epoch": 0.33, "learning_rate": 1.918901818200705e-05, "loss": 1.1627, "step": 854 }, { "epoch": 0.33, "learning_rate": 1.918653624749652e-05, "loss": 1.1681, "step": 855 }, { "epoch": 0.33, "learning_rate": 1.9184050681931004e-05, "loss": 1.1888, "step": 856 }, { "epoch": 0.33, "learning_rate": 1.918156148629295e-05, "loss": 1.1901, "step": 857 }, { "epoch": 0.33, "learning_rate": 1.9179068661566223e-05, "loss": 1.1334, "step": 858 }, { "epoch": 0.33, "learning_rate": 1.9176572208736147e-05, "loss": 1.1193, "step": 859 }, { "epoch": 0.33, "learning_rate": 1.9174072128789452e-05, "loss": 1.1846, "step": 860 }, { "epoch": 0.33, "learning_rate": 1.9171568422714324e-05, "loss": 1.1666, "step": 861 }, { "epoch": 0.33, "learning_rate": 1.916906109150037e-05, "loss": 1.1867, "step": 862 }, { "epoch": 0.33, "learning_rate": 1.9166550136138636e-05, "loss": 1.1881, "step": 863 }, { "epoch": 0.33, "learning_rate": 1.9164035557621593e-05, "loss": 1.1563, "step": 864 }, { "epoch": 0.33, "learning_rate": 1.9161517356943156e-05, "loss": 1.1588, "step": 865 }, { "epoch": 0.33, "learning_rate": 1.9158995535098662e-05, "loss": 1.1769, "step": 866 }, { "epoch": 0.33, "learning_rate": 1.915647009308488e-05, "loss": 1.1593, "step": 867 }, { "epoch": 0.33, "learning_rate": 1.915394103190001e-05, "loss": 1.2002, "step": 868 }, { "epoch": 0.33, "learning_rate": 1.915140835254369e-05, "loss": 1.1851, "step": 869 }, { "epoch": 0.33, "learning_rate": 1.914887205601698e-05, "loss": 1.2028, "step": 870 }, { "epoch": 0.33, "learning_rate": 1.914633214332238e-05, "loss": 1.1612, "step": 871 }, { "epoch": 0.33, "learning_rate": 1.9143788615463797e-05, "loss": 1.1785, "step": 872 }, { "epoch": 0.34, "learning_rate": 1.9141241473446594e-05, "loss": 1.187, "step": 873 }, { "epoch": 0.34, "learning_rate": 1.9138690718277542e-05, "loss": 1.1699, "step": 874 }, { "epoch": 0.34, "learning_rate": 1.9136136350964852e-05, "loss": 1.1401, "step": 875 }, { "epoch": 0.34, "learning_rate": 1.913357837251816e-05, "loss": 1.1833, "step": 876 }, { "epoch": 0.34, "learning_rate": 1.9131016783948526e-05, "loss": 1.182, "step": 877 }, { "epoch": 0.34, "learning_rate": 1.9128451586268438e-05, "loss": 1.2363, "step": 878 }, { "epoch": 0.34, "learning_rate": 1.9125882780491814e-05, "loss": 1.091, "step": 879 }, { "epoch": 0.34, "learning_rate": 1.9123310367633998e-05, "loss": 1.178, "step": 880 }, { "epoch": 0.34, "learning_rate": 1.9120734348711756e-05, "loss": 1.209, "step": 881 }, { "epoch": 0.34, "learning_rate": 1.9118154724743277e-05, "loss": 1.186, "step": 882 }, { "epoch": 0.34, "learning_rate": 1.9115571496748182e-05, "loss": 1.1695, "step": 883 }, { "epoch": 0.34, "learning_rate": 1.9112984665747515e-05, "loss": 1.2019, "step": 884 }, { "epoch": 0.34, "learning_rate": 1.9110394232763742e-05, "loss": 1.1378, "step": 885 }, { "epoch": 0.34, "learning_rate": 1.910780019882075e-05, "loss": 1.1942, "step": 886 }, { "epoch": 0.34, "learning_rate": 1.9105202564943857e-05, "loss": 1.1744, "step": 887 }, { "epoch": 0.34, "learning_rate": 1.9102601332159794e-05, "loss": 1.1549, "step": 888 }, { "epoch": 0.34, "learning_rate": 1.9099996501496725e-05, "loss": 1.1409, "step": 889 }, { "epoch": 0.34, "learning_rate": 1.909738807398423e-05, "loss": 1.1752, "step": 890 }, { "epoch": 0.34, "learning_rate": 1.909477605065331e-05, "loss": 1.2044, "step": 891 }, { "epoch": 0.34, "learning_rate": 1.909216043253639e-05, "loss": 1.1767, "step": 892 }, { "epoch": 0.34, "learning_rate": 1.9089541220667312e-05, "loss": 1.2186, "step": 893 }, { "epoch": 0.34, "learning_rate": 1.9086918416081345e-05, "loss": 1.167, "step": 894 }, { "epoch": 0.34, "learning_rate": 1.9084292019815175e-05, "loss": 1.1237, "step": 895 }, { "epoch": 0.34, "learning_rate": 1.90816620329069e-05, "loss": 1.2133, "step": 896 }, { "epoch": 0.34, "learning_rate": 1.9079028456396054e-05, "loss": 1.1758, "step": 897 }, { "epoch": 0.34, "learning_rate": 1.9076391291323577e-05, "loss": 1.159, "step": 898 }, { "epoch": 0.35, "learning_rate": 1.9073750538731826e-05, "loss": 1.1567, "step": 899 }, { "epoch": 0.35, "learning_rate": 1.9071106199664586e-05, "loss": 1.1883, "step": 900 }, { "epoch": 0.35, "learning_rate": 1.906845827516705e-05, "loss": 1.1653, "step": 901 }, { "epoch": 0.35, "learning_rate": 1.9065806766285834e-05, "loss": 1.1664, "step": 902 }, { "epoch": 0.35, "learning_rate": 1.906315167406897e-05, "loss": 1.148, "step": 903 }, { "epoch": 0.35, "learning_rate": 1.90604929995659e-05, "loss": 1.1541, "step": 904 }, { "epoch": 0.35, "learning_rate": 1.9057830743827493e-05, "loss": 1.2086, "step": 905 }, { "epoch": 0.35, "learning_rate": 1.905516490790603e-05, "loss": 1.173, "step": 906 }, { "epoch": 0.35, "learning_rate": 1.9052495492855198e-05, "loss": 1.2603, "step": 907 }, { "epoch": 0.35, "learning_rate": 1.9049822499730108e-05, "loss": 1.2005, "step": 908 }, { "epoch": 0.35, "learning_rate": 1.9047145929587285e-05, "loss": 1.2174, "step": 909 }, { "epoch": 0.35, "learning_rate": 1.904446578348466e-05, "loss": 1.1838, "step": 910 }, { "epoch": 0.35, "learning_rate": 1.904178206248159e-05, "loss": 1.1636, "step": 911 }, { "epoch": 0.35, "learning_rate": 1.9039094767638834e-05, "loss": 1.1506, "step": 912 }, { "epoch": 0.35, "learning_rate": 1.903640390001857e-05, "loss": 1.1355, "step": 913 }, { "epoch": 0.35, "learning_rate": 1.9033709460684383e-05, "loss": 1.1611, "step": 914 }, { "epoch": 0.35, "learning_rate": 1.9031011450701276e-05, "loss": 1.1426, "step": 915 }, { "epoch": 0.35, "learning_rate": 1.9028309871135656e-05, "loss": 1.1317, "step": 916 }, { "epoch": 0.35, "learning_rate": 1.9025604723055348e-05, "loss": 1.2381, "step": 917 }, { "epoch": 0.35, "learning_rate": 1.902289600752958e-05, "loss": 1.1778, "step": 918 }, { "epoch": 0.35, "learning_rate": 1.9020183725629004e-05, "loss": 1.1924, "step": 919 }, { "epoch": 0.35, "learning_rate": 1.901746787842566e-05, "loss": 1.1866, "step": 920 }, { "epoch": 0.35, "learning_rate": 1.901474846699302e-05, "loss": 1.1765, "step": 921 }, { "epoch": 0.35, "learning_rate": 1.9012025492405943e-05, "loss": 1.204, "step": 922 }, { "epoch": 0.35, "learning_rate": 1.9009298955740718e-05, "loss": 1.16, "step": 923 }, { "epoch": 0.35, "learning_rate": 1.9006568858075022e-05, "loss": 1.155, "step": 924 }, { "epoch": 0.36, "learning_rate": 1.9003835200487957e-05, "loss": 1.1702, "step": 925 }, { "epoch": 0.36, "learning_rate": 1.900109798406002e-05, "loss": 1.168, "step": 926 }, { "epoch": 0.36, "learning_rate": 1.8998357209873118e-05, "loss": 1.1978, "step": 927 }, { "epoch": 0.36, "learning_rate": 1.8995612879010566e-05, "loss": 1.1529, "step": 928 }, { "epoch": 0.36, "learning_rate": 1.8992864992557084e-05, "loss": 1.14, "step": 929 }, { "epoch": 0.36, "learning_rate": 1.8990113551598794e-05, "loss": 1.1781, "step": 930 }, { "epoch": 0.36, "learning_rate": 1.8987358557223232e-05, "loss": 1.1616, "step": 931 }, { "epoch": 0.36, "learning_rate": 1.8984600010519324e-05, "loss": 1.0774, "step": 932 }, { "epoch": 0.36, "learning_rate": 1.898183791257742e-05, "loss": 1.1265, "step": 933 }, { "epoch": 0.36, "learning_rate": 1.8979072264489252e-05, "loss": 1.1686, "step": 934 }, { "epoch": 0.36, "learning_rate": 1.8976303067347967e-05, "loss": 1.1682, "step": 935 }, { "epoch": 0.36, "learning_rate": 1.897353032224812e-05, "loss": 1.2033, "step": 936 }, { "epoch": 0.36, "learning_rate": 1.8970754030285655e-05, "loss": 1.1478, "step": 937 }, { "epoch": 0.36, "learning_rate": 1.896797419255793e-05, "loss": 1.1904, "step": 938 }, { "epoch": 0.36, "learning_rate": 1.896519081016369e-05, "loss": 1.2032, "step": 939 }, { "epoch": 0.36, "learning_rate": 1.8962403884203096e-05, "loss": 1.1986, "step": 940 }, { "epoch": 0.36, "learning_rate": 1.8959613415777706e-05, "loss": 1.2091, "step": 941 }, { "epoch": 0.36, "learning_rate": 1.895681940599047e-05, "loss": 1.1719, "step": 942 }, { "epoch": 0.36, "learning_rate": 1.895402185594575e-05, "loss": 1.2119, "step": 943 }, { "epoch": 0.36, "learning_rate": 1.8951220766749295e-05, "loss": 1.1589, "step": 944 }, { "epoch": 0.36, "learning_rate": 1.8948416139508265e-05, "loss": 1.1527, "step": 945 }, { "epoch": 0.36, "learning_rate": 1.8945607975331204e-05, "loss": 1.1479, "step": 946 }, { "epoch": 0.36, "learning_rate": 1.894279627532807e-05, "loss": 1.1366, "step": 947 }, { "epoch": 0.36, "learning_rate": 1.8939981040610208e-05, "loss": 1.2044, "step": 948 }, { "epoch": 0.36, "learning_rate": 1.8937162272290364e-05, "loss": 1.1931, "step": 949 }, { "epoch": 0.36, "learning_rate": 1.8934339971482676e-05, "loss": 1.1824, "step": 950 }, { "epoch": 0.37, "learning_rate": 1.8931514139302688e-05, "loss": 1.1711, "step": 951 }, { "epoch": 0.37, "learning_rate": 1.8928684776867327e-05, "loss": 1.1922, "step": 952 }, { "epoch": 0.37, "learning_rate": 1.8925851885294927e-05, "loss": 1.1502, "step": 953 }, { "epoch": 0.37, "learning_rate": 1.8923015465705212e-05, "loss": 1.1555, "step": 954 }, { "epoch": 0.37, "learning_rate": 1.8920175519219298e-05, "loss": 1.1577, "step": 955 }, { "epoch": 0.37, "learning_rate": 1.8917332046959696e-05, "loss": 1.1778, "step": 956 }, { "epoch": 0.37, "learning_rate": 1.891448505005032e-05, "loss": 1.158, "step": 957 }, { "epoch": 0.37, "learning_rate": 1.8911634529616454e-05, "loss": 1.1668, "step": 958 }, { "epoch": 0.37, "learning_rate": 1.8908780486784806e-05, "loss": 1.1627, "step": 959 }, { "epoch": 0.37, "learning_rate": 1.8905922922683454e-05, "loss": 1.158, "step": 960 }, { "epoch": 0.37, "learning_rate": 1.8903061838441874e-05, "loss": 1.1435, "step": 961 }, { "epoch": 0.37, "learning_rate": 1.890019723519093e-05, "loss": 1.1358, "step": 962 }, { "epoch": 0.37, "learning_rate": 1.889732911406289e-05, "loss": 1.1522, "step": 963 }, { "epoch": 0.37, "learning_rate": 1.889445747619139e-05, "loss": 1.141, "step": 964 }, { "epoch": 0.37, "learning_rate": 1.8891582322711474e-05, "loss": 1.1323, "step": 965 }, { "epoch": 0.37, "learning_rate": 1.8888703654759574e-05, "loss": 1.2075, "step": 966 }, { "epoch": 0.37, "learning_rate": 1.8885821473473507e-05, "loss": 1.1875, "step": 967 }, { "epoch": 0.37, "learning_rate": 1.8882935779992476e-05, "loss": 1.1376, "step": 968 }, { "epoch": 0.37, "learning_rate": 1.8880046575457072e-05, "loss": 1.1428, "step": 969 }, { "epoch": 0.37, "learning_rate": 1.8877153861009287e-05, "loss": 1.1851, "step": 970 }, { "epoch": 0.37, "learning_rate": 1.8874257637792484e-05, "loss": 1.1515, "step": 971 }, { "epoch": 0.37, "learning_rate": 1.887135790695142e-05, "loss": 1.2325, "step": 972 }, { "epoch": 0.37, "learning_rate": 1.8868454669632242e-05, "loss": 1.1756, "step": 973 }, { "epoch": 0.37, "learning_rate": 1.886554792698247e-05, "loss": 1.2175, "step": 974 }, { "epoch": 0.37, "learning_rate": 1.8862637680151023e-05, "loss": 1.2087, "step": 975 }, { "epoch": 0.37, "learning_rate": 1.8859723930288207e-05, "loss": 1.1704, "step": 976 }, { "epoch": 0.38, "learning_rate": 1.8856806678545692e-05, "loss": 1.108, "step": 977 }, { "epoch": 0.38, "learning_rate": 1.8853885926076556e-05, "loss": 1.1799, "step": 978 }, { "epoch": 0.38, "learning_rate": 1.885096167403525e-05, "loss": 1.1184, "step": 979 }, { "epoch": 0.38, "learning_rate": 1.8848033923577604e-05, "loss": 1.1964, "step": 980 }, { "epoch": 0.38, "learning_rate": 1.8845102675860842e-05, "loss": 1.13, "step": 981 }, { "epoch": 0.38, "learning_rate": 1.884216793204356e-05, "loss": 1.1475, "step": 982 }, { "epoch": 0.38, "learning_rate": 1.883922969328574e-05, "loss": 1.1874, "step": 983 }, { "epoch": 0.38, "learning_rate": 1.883628796074875e-05, "loss": 1.1786, "step": 984 }, { "epoch": 0.38, "learning_rate": 1.8833342735595327e-05, "loss": 1.1511, "step": 985 }, { "epoch": 0.38, "learning_rate": 1.88303940189896e-05, "loss": 1.1476, "step": 986 }, { "epoch": 0.38, "learning_rate": 1.882744181209707e-05, "loss": 1.1393, "step": 987 }, { "epoch": 0.38, "learning_rate": 1.8824486116084633e-05, "loss": 1.1747, "step": 988 }, { "epoch": 0.38, "learning_rate": 1.8821526932120538e-05, "loss": 1.1493, "step": 989 }, { "epoch": 0.38, "learning_rate": 1.8818564261374433e-05, "loss": 1.1594, "step": 990 }, { "epoch": 0.38, "learning_rate": 1.881559810501734e-05, "loss": 1.1703, "step": 991 }, { "epoch": 0.38, "learning_rate": 1.881262846422166e-05, "loss": 1.1952, "step": 992 }, { "epoch": 0.38, "learning_rate": 1.8809655340161158e-05, "loss": 1.1407, "step": 993 }, { "epoch": 0.38, "learning_rate": 1.880667873401099e-05, "loss": 1.1407, "step": 994 }, { "epoch": 0.38, "learning_rate": 1.880369864694769e-05, "loss": 1.0874, "step": 995 }, { "epoch": 0.38, "learning_rate": 1.8800715080149156e-05, "loss": 1.1695, "step": 996 }, { "epoch": 0.38, "learning_rate": 1.8797728034794675e-05, "loss": 1.2008, "step": 997 }, { "epoch": 0.38, "learning_rate": 1.879473751206489e-05, "loss": 1.1392, "step": 998 }, { "epoch": 0.38, "learning_rate": 1.8791743513141842e-05, "loss": 1.2187, "step": 999 }, { "epoch": 0.38, "learning_rate": 1.8788746039208926e-05, "loss": 1.103, "step": 1000 }, { "epoch": 0.38, "learning_rate": 1.8785745091450924e-05, "loss": 1.1651, "step": 1001 }, { "epoch": 0.38, "learning_rate": 1.878274067105398e-05, "loss": 1.1794, "step": 1002 }, { "epoch": 0.39, "learning_rate": 1.877973277920562e-05, "loss": 1.1769, "step": 1003 }, { "epoch": 0.39, "learning_rate": 1.8776721417094737e-05, "loss": 1.1769, "step": 1004 }, { "epoch": 0.39, "learning_rate": 1.8773706585911598e-05, "loss": 1.1607, "step": 1005 }, { "epoch": 0.39, "learning_rate": 1.8770688286847837e-05, "loss": 1.1209, "step": 1006 }, { "epoch": 0.39, "learning_rate": 1.876766652109647e-05, "loss": 1.1663, "step": 1007 }, { "epoch": 0.39, "learning_rate": 1.876464128985186e-05, "loss": 1.1103, "step": 1008 }, { "epoch": 0.39, "learning_rate": 1.8761612594309774e-05, "loss": 1.1306, "step": 1009 }, { "epoch": 0.39, "learning_rate": 1.8758580435667313e-05, "loss": 1.1685, "step": 1010 }, { "epoch": 0.39, "learning_rate": 1.875554481512297e-05, "loss": 1.1561, "step": 1011 }, { "epoch": 0.39, "learning_rate": 1.87525057338766e-05, "loss": 1.1616, "step": 1012 }, { "epoch": 0.39, "learning_rate": 1.8749463193129424e-05, "loss": 1.1162, "step": 1013 }, { "epoch": 0.39, "learning_rate": 1.874641719408403e-05, "loss": 1.1934, "step": 1014 }, { "epoch": 0.39, "learning_rate": 1.874336773794438e-05, "loss": 1.1313, "step": 1015 }, { "epoch": 0.39, "learning_rate": 1.8740314825915788e-05, "loss": 1.1098, "step": 1016 }, { "epoch": 0.39, "learning_rate": 1.873725845920495e-05, "loss": 1.1621, "step": 1017 }, { "epoch": 0.39, "learning_rate": 1.8734198639019916e-05, "loss": 1.1354, "step": 1018 }, { "epoch": 0.39, "learning_rate": 1.873113536657011e-05, "loss": 1.1532, "step": 1019 }, { "epoch": 0.39, "learning_rate": 1.8728068643066313e-05, "loss": 1.1547, "step": 1020 }, { "epoch": 0.39, "learning_rate": 1.8724998469720673e-05, "loss": 1.1849, "step": 1021 }, { "epoch": 0.39, "learning_rate": 1.8721924847746705e-05, "loss": 1.1847, "step": 1022 }, { "epoch": 0.39, "learning_rate": 1.8718847778359276e-05, "loss": 1.1887, "step": 1023 }, { "epoch": 0.39, "learning_rate": 1.871576726277463e-05, "loss": 1.1625, "step": 1024 }, { "epoch": 0.39, "learning_rate": 1.8712683302210366e-05, "loss": 1.1541, "step": 1025 }, { "epoch": 0.39, "learning_rate": 1.870959589788544e-05, "loss": 1.171, "step": 1026 }, { "epoch": 0.39, "learning_rate": 1.8706505051020184e-05, "loss": 1.1468, "step": 1027 }, { "epoch": 0.39, "learning_rate": 1.8703410762836268e-05, "loss": 1.1704, "step": 1028 }, { "epoch": 0.4, "learning_rate": 1.870031303455675e-05, "loss": 1.2187, "step": 1029 }, { "epoch": 0.4, "learning_rate": 1.869721186740602e-05, "loss": 1.1857, "step": 1030 }, { "epoch": 0.4, "learning_rate": 1.8694107262609843e-05, "loss": 1.2476, "step": 1031 }, { "epoch": 0.4, "learning_rate": 1.8690999221395346e-05, "loss": 1.1824, "step": 1032 }, { "epoch": 0.4, "learning_rate": 1.8687887744991e-05, "loss": 1.156, "step": 1033 }, { "epoch": 0.4, "learning_rate": 1.868477283462665e-05, "loss": 1.1862, "step": 1034 }, { "epoch": 0.4, "learning_rate": 1.868165449153349e-05, "loss": 1.1742, "step": 1035 }, { "epoch": 0.4, "learning_rate": 1.867853271694406e-05, "loss": 1.1325, "step": 1036 }, { "epoch": 0.4, "learning_rate": 1.8675407512092278e-05, "loss": 1.18, "step": 1037 }, { "epoch": 0.4, "learning_rate": 1.8672278878213403e-05, "loss": 1.1233, "step": 1038 }, { "epoch": 0.4, "learning_rate": 1.866914681654406e-05, "loss": 1.1088, "step": 1039 }, { "epoch": 0.4, "learning_rate": 1.8666011328322212e-05, "loss": 1.161, "step": 1040 }, { "epoch": 0.4, "learning_rate": 1.8662872414787193e-05, "loss": 1.138, "step": 1041 }, { "epoch": 0.4, "learning_rate": 1.8659730077179686e-05, "loss": 1.182, "step": 1042 }, { "epoch": 0.4, "learning_rate": 1.8656584316741723e-05, "loss": 1.114, "step": 1043 }, { "epoch": 0.4, "learning_rate": 1.865343513471669e-05, "loss": 1.2254, "step": 1044 }, { "epoch": 0.4, "learning_rate": 1.8650282532349332e-05, "loss": 1.1803, "step": 1045 }, { "epoch": 0.4, "learning_rate": 1.864712651088574e-05, "loss": 1.1625, "step": 1046 }, { "epoch": 0.4, "learning_rate": 1.8643967071573362e-05, "loss": 1.1869, "step": 1047 }, { "epoch": 0.4, "learning_rate": 1.8640804215660986e-05, "loss": 1.1752, "step": 1048 }, { "epoch": 0.4, "learning_rate": 1.8637637944398758e-05, "loss": 1.1564, "step": 1049 }, { "epoch": 0.4, "learning_rate": 1.863446825903817e-05, "loss": 1.1282, "step": 1050 }, { "epoch": 0.4, "learning_rate": 1.863129516083208e-05, "loss": 1.2232, "step": 1051 }, { "epoch": 0.4, "learning_rate": 1.8628118651034668e-05, "loss": 1.1227, "step": 1052 }, { "epoch": 0.4, "learning_rate": 1.862493873090148e-05, "loss": 1.1647, "step": 1053 }, { "epoch": 0.4, "learning_rate": 1.862175540168941e-05, "loss": 1.1831, "step": 1054 }, { "epoch": 0.41, "learning_rate": 1.8618568664656685e-05, "loss": 1.1102, "step": 1055 }, { "epoch": 0.41, "learning_rate": 1.8615378521062903e-05, "loss": 1.1932, "step": 1056 }, { "epoch": 0.41, "learning_rate": 1.8612184972168984e-05, "loss": 1.1507, "step": 1057 }, { "epoch": 0.41, "learning_rate": 1.860898801923721e-05, "loss": 1.1518, "step": 1058 }, { "epoch": 0.41, "learning_rate": 1.8605787663531206e-05, "loss": 1.1465, "step": 1059 }, { "epoch": 0.41, "learning_rate": 1.860258390631593e-05, "loss": 1.2038, "step": 1060 }, { "epoch": 0.41, "learning_rate": 1.85993767488577e-05, "loss": 1.1775, "step": 1061 }, { "epoch": 0.41, "learning_rate": 1.859616619242417e-05, "loss": 1.1547, "step": 1062 }, { "epoch": 0.41, "learning_rate": 1.8592952238284346e-05, "loss": 1.128, "step": 1063 }, { "epoch": 0.41, "learning_rate": 1.8589734887708556e-05, "loss": 1.1074, "step": 1064 }, { "epoch": 0.41, "learning_rate": 1.85865141419685e-05, "loss": 1.1773, "step": 1065 }, { "epoch": 0.41, "learning_rate": 1.8583290002337192e-05, "loss": 1.1452, "step": 1066 }, { "epoch": 0.41, "learning_rate": 1.858006247008901e-05, "loss": 1.2302, "step": 1067 }, { "epoch": 0.41, "learning_rate": 1.8576831546499653e-05, "loss": 1.1816, "step": 1068 }, { "epoch": 0.41, "learning_rate": 1.857359723284618e-05, "loss": 1.162, "step": 1069 }, { "epoch": 0.41, "learning_rate": 1.8570359530406975e-05, "loss": 1.1638, "step": 1070 }, { "epoch": 0.41, "learning_rate": 1.8567118440461767e-05, "loss": 1.1611, "step": 1071 }, { "epoch": 0.41, "learning_rate": 1.8563873964291628e-05, "loss": 1.1623, "step": 1072 }, { "epoch": 0.41, "learning_rate": 1.8560626103178954e-05, "loss": 1.1682, "step": 1073 }, { "epoch": 0.41, "learning_rate": 1.8557374858407504e-05, "loss": 1.1364, "step": 1074 }, { "epoch": 0.41, "learning_rate": 1.855412023126235e-05, "loss": 1.1217, "step": 1075 }, { "epoch": 0.41, "learning_rate": 1.8550862223029913e-05, "loss": 1.1504, "step": 1076 }, { "epoch": 0.41, "learning_rate": 1.8547600834997946e-05, "loss": 1.134, "step": 1077 }, { "epoch": 0.41, "learning_rate": 1.854433606845554e-05, "loss": 1.1421, "step": 1078 }, { "epoch": 0.41, "learning_rate": 1.8541067924693123e-05, "loss": 1.1474, "step": 1079 }, { "epoch": 0.41, "learning_rate": 1.8537796405002457e-05, "loss": 1.1391, "step": 1080 }, { "epoch": 0.42, "learning_rate": 1.8534521510676632e-05, "loss": 1.2057, "step": 1081 }, { "epoch": 0.42, "learning_rate": 1.8531243243010087e-05, "loss": 1.1315, "step": 1082 }, { "epoch": 0.42, "learning_rate": 1.8527961603298572e-05, "loss": 1.1164, "step": 1083 }, { "epoch": 0.42, "learning_rate": 1.8524676592839193e-05, "loss": 1.0955, "step": 1084 }, { "epoch": 0.42, "learning_rate": 1.8521388212930374e-05, "loss": 1.1102, "step": 1085 }, { "epoch": 0.42, "learning_rate": 1.8518096464871872e-05, "loss": 1.1269, "step": 1086 }, { "epoch": 0.42, "learning_rate": 1.851480134996478e-05, "loss": 1.1632, "step": 1087 }, { "epoch": 0.42, "learning_rate": 1.8511502869511524e-05, "loss": 1.1405, "step": 1088 }, { "epoch": 0.42, "learning_rate": 1.850820102481585e-05, "loss": 1.2323, "step": 1089 }, { "epoch": 0.42, "learning_rate": 1.8504895817182838e-05, "loss": 1.1772, "step": 1090 }, { "epoch": 0.42, "learning_rate": 1.8501587247918904e-05, "loss": 1.2141, "step": 1091 }, { "epoch": 0.42, "learning_rate": 1.8498275318331784e-05, "loss": 1.1974, "step": 1092 }, { "epoch": 0.42, "learning_rate": 1.849496002973055e-05, "loss": 1.178, "step": 1093 }, { "epoch": 0.42, "learning_rate": 1.8491641383425594e-05, "loss": 1.1724, "step": 1094 }, { "epoch": 0.42, "learning_rate": 1.848831938072864e-05, "loss": 1.1624, "step": 1095 }, { "epoch": 0.42, "learning_rate": 1.8484994022952736e-05, "loss": 1.1508, "step": 1096 }, { "epoch": 0.42, "learning_rate": 1.8481665311412263e-05, "loss": 1.1649, "step": 1097 }, { "epoch": 0.42, "learning_rate": 1.847833324742291e-05, "loss": 1.1389, "step": 1098 }, { "epoch": 0.42, "learning_rate": 1.8474997832301716e-05, "loss": 1.2207, "step": 1099 }, { "epoch": 0.42, "learning_rate": 1.847165906736702e-05, "loss": 1.1455, "step": 1100 }, { "epoch": 0.42, "learning_rate": 1.8468316953938505e-05, "loss": 1.1796, "step": 1101 }, { "epoch": 0.42, "learning_rate": 1.8464971493337167e-05, "loss": 1.1288, "step": 1102 }, { "epoch": 0.42, "learning_rate": 1.846162268688533e-05, "loss": 1.1294, "step": 1103 }, { "epoch": 0.42, "learning_rate": 1.8458270535906626e-05, "loss": 1.1196, "step": 1104 }, { "epoch": 0.42, "learning_rate": 1.8454915041726035e-05, "loss": 1.1731, "step": 1105 }, { "epoch": 0.42, "learning_rate": 1.845155620566984e-05, "loss": 1.1843, "step": 1106 }, { "epoch": 0.43, "learning_rate": 1.8448194029065642e-05, "loss": 1.1094, "step": 1107 }, { "epoch": 0.43, "learning_rate": 1.8444828513242375e-05, "loss": 1.2172, "step": 1108 }, { "epoch": 0.43, "learning_rate": 1.844145965953029e-05, "loss": 1.1723, "step": 1109 }, { "epoch": 0.43, "learning_rate": 1.8438087469260945e-05, "loss": 1.1259, "step": 1110 }, { "epoch": 0.43, "learning_rate": 1.843471194376724e-05, "loss": 1.176, "step": 1111 }, { "epoch": 0.43, "learning_rate": 1.8431333084383365e-05, "loss": 1.0905, "step": 1112 }, { "epoch": 0.43, "learning_rate": 1.842795089244485e-05, "loss": 1.1822, "step": 1113 }, { "epoch": 0.43, "learning_rate": 1.8424565369288537e-05, "loss": 1.1446, "step": 1114 }, { "epoch": 0.43, "learning_rate": 1.8421176516252578e-05, "loss": 1.1154, "step": 1115 }, { "epoch": 0.43, "learning_rate": 1.8417784334676444e-05, "loss": 1.1206, "step": 1116 }, { "epoch": 0.43, "learning_rate": 1.8414388825900927e-05, "loss": 1.1416, "step": 1117 }, { "epoch": 0.43, "learning_rate": 1.841098999126813e-05, "loss": 1.1904, "step": 1118 }, { "epoch": 0.43, "learning_rate": 1.8407587832121465e-05, "loss": 1.1845, "step": 1119 }, { "epoch": 0.43, "learning_rate": 1.8404182349805674e-05, "loss": 1.1543, "step": 1120 }, { "epoch": 0.43, "learning_rate": 1.8400773545666788e-05, "loss": 1.165, "step": 1121 }, { "epoch": 0.43, "learning_rate": 1.8397361421052178e-05, "loss": 1.1424, "step": 1122 }, { "epoch": 0.43, "learning_rate": 1.839394597731051e-05, "loss": 1.1931, "step": 1123 }, { "epoch": 0.43, "learning_rate": 1.8390527215791767e-05, "loss": 1.21, "step": 1124 }, { "epoch": 0.43, "learning_rate": 1.8387105137847242e-05, "loss": 1.1405, "step": 1125 }, { "epoch": 0.43, "learning_rate": 1.8383679744829543e-05, "loss": 1.1549, "step": 1126 }, { "epoch": 0.43, "learning_rate": 1.838025103809258e-05, "loss": 1.153, "step": 1127 }, { "epoch": 0.43, "learning_rate": 1.837681901899158e-05, "loss": 1.1318, "step": 1128 }, { "epoch": 0.43, "learning_rate": 1.8373383688883076e-05, "loss": 1.1575, "step": 1129 }, { "epoch": 0.43, "learning_rate": 1.8369945049124914e-05, "loss": 1.161, "step": 1130 }, { "epoch": 0.43, "learning_rate": 1.8366503101076248e-05, "loss": 1.0787, "step": 1131 }, { "epoch": 0.43, "learning_rate": 1.8363057846097528e-05, "loss": 1.1132, "step": 1132 }, { "epoch": 0.44, "learning_rate": 1.835960928555052e-05, "loss": 1.1162, "step": 1133 }, { "epoch": 0.44, "learning_rate": 1.8356157420798305e-05, "loss": 1.1346, "step": 1134 }, { "epoch": 0.44, "learning_rate": 1.8352702253205256e-05, "loss": 1.1099, "step": 1135 }, { "epoch": 0.44, "learning_rate": 1.834924378413706e-05, "loss": 1.1075, "step": 1136 }, { "epoch": 0.44, "learning_rate": 1.83457820149607e-05, "loss": 1.1029, "step": 1137 }, { "epoch": 0.44, "learning_rate": 1.8342316947044468e-05, "loss": 1.1168, "step": 1138 }, { "epoch": 0.44, "learning_rate": 1.8338848581757972e-05, "loss": 1.1641, "step": 1139 }, { "epoch": 0.44, "learning_rate": 1.8335376920472098e-05, "loss": 1.1197, "step": 1140 }, { "epoch": 0.44, "learning_rate": 1.833190196455906e-05, "loss": 1.1622, "step": 1141 }, { "epoch": 0.44, "learning_rate": 1.8328423715392364e-05, "loss": 1.1878, "step": 1142 }, { "epoch": 0.44, "learning_rate": 1.8324942174346807e-05, "loss": 1.1426, "step": 1143 }, { "epoch": 0.44, "learning_rate": 1.8321457342798503e-05, "loss": 1.1975, "step": 1144 }, { "epoch": 0.44, "learning_rate": 1.831796922212486e-05, "loss": 1.0903, "step": 1145 }, { "epoch": 0.44, "learning_rate": 1.831447781370459e-05, "loss": 1.215, "step": 1146 }, { "epoch": 0.44, "learning_rate": 1.8310983118917695e-05, "loss": 1.1202, "step": 1147 }, { "epoch": 0.44, "learning_rate": 1.830748513914549e-05, "loss": 1.1339, "step": 1148 }, { "epoch": 0.44, "learning_rate": 1.8303983875770573e-05, "loss": 1.1819, "step": 1149 }, { "epoch": 0.44, "learning_rate": 1.830047933017685e-05, "loss": 1.1475, "step": 1150 }, { "epoch": 0.44, "learning_rate": 1.8296971503749523e-05, "loss": 1.1407, "step": 1151 }, { "epoch": 0.44, "learning_rate": 1.829346039787509e-05, "loss": 1.2158, "step": 1152 }, { "epoch": 0.44, "learning_rate": 1.8289946013941344e-05, "loss": 1.1381, "step": 1153 }, { "epoch": 0.44, "learning_rate": 1.8286428353337376e-05, "loss": 1.1616, "step": 1154 }, { "epoch": 0.44, "learning_rate": 1.8282907417453568e-05, "loss": 1.1321, "step": 1155 }, { "epoch": 0.44, "learning_rate": 1.82793832076816e-05, "loss": 1.1109, "step": 1156 }, { "epoch": 0.44, "learning_rate": 1.8275855725414445e-05, "loss": 1.1941, "step": 1157 }, { "epoch": 0.44, "learning_rate": 1.827232497204637e-05, "loss": 1.0973, "step": 1158 }, { "epoch": 0.45, "learning_rate": 1.8268790948972942e-05, "loss": 1.1891, "step": 1159 }, { "epoch": 0.45, "learning_rate": 1.8265253657590997e-05, "loss": 1.1619, "step": 1160 }, { "epoch": 0.45, "learning_rate": 1.8261713099298696e-05, "loss": 1.1615, "step": 1161 }, { "epoch": 0.45, "learning_rate": 1.825816927549546e-05, "loss": 1.1615, "step": 1162 }, { "epoch": 0.45, "learning_rate": 1.8254622187582022e-05, "loss": 1.185, "step": 1163 }, { "epoch": 0.45, "learning_rate": 1.82510718369604e-05, "loss": 1.1513, "step": 1164 }, { "epoch": 0.45, "learning_rate": 1.8247518225033896e-05, "loss": 1.1307, "step": 1165 }, { "epoch": 0.45, "learning_rate": 1.8243961353207105e-05, "loss": 1.2261, "step": 1166 }, { "epoch": 0.45, "learning_rate": 1.824040122288591e-05, "loss": 1.1608, "step": 1167 }, { "epoch": 0.45, "learning_rate": 1.8236837835477486e-05, "loss": 1.1391, "step": 1168 }, { "epoch": 0.45, "learning_rate": 1.8233271192390284e-05, "loss": 1.112, "step": 1169 }, { "epoch": 0.45, "learning_rate": 1.8229701295034056e-05, "loss": 1.1845, "step": 1170 }, { "epoch": 0.45, "learning_rate": 1.8226128144819833e-05, "loss": 1.1183, "step": 1171 }, { "epoch": 0.45, "learning_rate": 1.822255174315993e-05, "loss": 1.173, "step": 1172 }, { "epoch": 0.45, "learning_rate": 1.8218972091467954e-05, "loss": 1.1192, "step": 1173 }, { "epoch": 0.45, "learning_rate": 1.8215389191158784e-05, "loss": 1.148, "step": 1174 }, { "epoch": 0.45, "learning_rate": 1.8211803043648596e-05, "loss": 1.101, "step": 1175 }, { "epoch": 0.45, "learning_rate": 1.8208213650354846e-05, "loss": 1.2277, "step": 1176 }, { "epoch": 0.45, "learning_rate": 1.820462101269627e-05, "loss": 1.1635, "step": 1177 }, { "epoch": 0.45, "learning_rate": 1.820102513209289e-05, "loss": 1.1735, "step": 1178 }, { "epoch": 0.45, "learning_rate": 1.8197426009966004e-05, "loss": 1.1453, "step": 1179 }, { "epoch": 0.45, "learning_rate": 1.8193823647738195e-05, "loss": 1.1567, "step": 1180 }, { "epoch": 0.45, "learning_rate": 1.819021804683333e-05, "loss": 1.1434, "step": 1181 }, { "epoch": 0.45, "learning_rate": 1.8186609208676552e-05, "loss": 1.1259, "step": 1182 }, { "epoch": 0.45, "learning_rate": 1.8182997134694285e-05, "loss": 1.1988, "step": 1183 }, { "epoch": 0.45, "learning_rate": 1.817938182631423e-05, "loss": 1.1797, "step": 1184 }, { "epoch": 0.46, "learning_rate": 1.8175763284965365e-05, "loss": 1.1333, "step": 1185 }, { "epoch": 0.46, "learning_rate": 1.8172141512077953e-05, "loss": 1.1549, "step": 1186 }, { "epoch": 0.46, "learning_rate": 1.8168516509083526e-05, "loss": 1.1666, "step": 1187 }, { "epoch": 0.46, "learning_rate": 1.81648882774149e-05, "loss": 1.1898, "step": 1188 }, { "epoch": 0.46, "learning_rate": 1.8161256818506166e-05, "loss": 1.1653, "step": 1189 }, { "epoch": 0.46, "learning_rate": 1.8157622133792687e-05, "loss": 1.1646, "step": 1190 }, { "epoch": 0.46, "learning_rate": 1.8153984224711094e-05, "loss": 1.0998, "step": 1191 }, { "epoch": 0.46, "learning_rate": 1.815034309269931e-05, "loss": 1.1669, "step": 1192 }, { "epoch": 0.46, "learning_rate": 1.814669873919652e-05, "loss": 1.1316, "step": 1193 }, { "epoch": 0.46, "learning_rate": 1.8143051165643183e-05, "loss": 1.1727, "step": 1194 }, { "epoch": 0.46, "learning_rate": 1.813940037348104e-05, "loss": 1.1107, "step": 1195 }, { "epoch": 0.46, "learning_rate": 1.813574636415309e-05, "loss": 1.1595, "step": 1196 }, { "epoch": 0.46, "learning_rate": 1.8132089139103612e-05, "loss": 1.1181, "step": 1197 }, { "epoch": 0.46, "learning_rate": 1.8128428699778154e-05, "loss": 1.1176, "step": 1198 }, { "epoch": 0.46, "learning_rate": 1.812476504762354e-05, "loss": 1.1063, "step": 1199 }, { "epoch": 0.46, "learning_rate": 1.8121098184087855e-05, "loss": 1.1629, "step": 1200 }, { "epoch": 0.46, "eval_loss": 1.1579093933105469, "eval_runtime": 688.3127, "eval_samples_per_second": 4.941, "eval_steps_per_second": 0.309, "step": 1200 }, { "epoch": 0.46, "learning_rate": 1.811742811062046e-05, "loss": 1.253, "step": 1201 }, { "epoch": 0.46, "learning_rate": 1.8113754828671973e-05, "loss": 1.1748, "step": 1202 }, { "epoch": 0.46, "learning_rate": 1.81100783396943e-05, "loss": 1.2049, "step": 1203 }, { "epoch": 0.46, "learning_rate": 1.8106398645140602e-05, "loss": 1.1107, "step": 1204 }, { "epoch": 0.46, "learning_rate": 1.8102715746465308e-05, "loss": 1.153, "step": 1205 }, { "epoch": 0.46, "learning_rate": 1.8099029645124108e-05, "loss": 1.1897, "step": 1206 }, { "epoch": 0.46, "learning_rate": 1.8095340342573973e-05, "loss": 1.1904, "step": 1207 }, { "epoch": 0.46, "learning_rate": 1.809164784027313e-05, "loss": 1.2165, "step": 1208 }, { "epoch": 0.46, "learning_rate": 1.8087952139681065e-05, "loss": 1.167, "step": 1209 }, { "epoch": 0.46, "learning_rate": 1.8084253242258542e-05, "loss": 1.1951, "step": 1210 }, { "epoch": 0.47, "learning_rate": 1.808055114946757e-05, "loss": 1.1704, "step": 1211 }, { "epoch": 0.47, "learning_rate": 1.8076845862771445e-05, "loss": 1.0904, "step": 1212 }, { "epoch": 0.47, "learning_rate": 1.8073137383634706e-05, "loss": 1.1694, "step": 1213 }, { "epoch": 0.47, "learning_rate": 1.8069425713523162e-05, "loss": 1.1605, "step": 1214 }, { "epoch": 0.47, "learning_rate": 1.8065710853903878e-05, "loss": 1.1637, "step": 1215 }, { "epoch": 0.47, "learning_rate": 1.8061992806245186e-05, "loss": 1.188, "step": 1216 }, { "epoch": 0.47, "learning_rate": 1.8058271572016677e-05, "loss": 1.1275, "step": 1217 }, { "epoch": 0.47, "learning_rate": 1.8054547152689195e-05, "loss": 1.1519, "step": 1218 }, { "epoch": 0.47, "learning_rate": 1.805081954973485e-05, "loss": 1.1428, "step": 1219 }, { "epoch": 0.47, "learning_rate": 1.804708876462701e-05, "loss": 1.146, "step": 1220 }, { "epoch": 0.47, "learning_rate": 1.8043354798840298e-05, "loss": 1.094, "step": 1221 }, { "epoch": 0.47, "learning_rate": 1.8039617653850596e-05, "loss": 1.1454, "step": 1222 }, { "epoch": 0.47, "learning_rate": 1.8035877331135037e-05, "loss": 1.1743, "step": 1223 }, { "epoch": 0.47, "learning_rate": 1.8032133832172022e-05, "loss": 1.1183, "step": 1224 }, { "epoch": 0.47, "learning_rate": 1.8028387158441192e-05, "loss": 1.1823, "step": 1225 }, { "epoch": 0.47, "learning_rate": 1.802463731142346e-05, "loss": 1.1447, "step": 1226 }, { "epoch": 0.47, "learning_rate": 1.8020884292600974e-05, "loss": 1.2102, "step": 1227 }, { "epoch": 0.47, "learning_rate": 1.8017128103457154e-05, "loss": 1.1573, "step": 1228 }, { "epoch": 0.47, "learning_rate": 1.801336874547666e-05, "loss": 1.1617, "step": 1229 }, { "epoch": 0.47, "learning_rate": 1.8009606220145416e-05, "loss": 1.178, "step": 1230 }, { "epoch": 0.47, "learning_rate": 1.8005840528950587e-05, "loss": 1.122, "step": 1231 }, { "epoch": 0.47, "learning_rate": 1.800207167338059e-05, "loss": 1.143, "step": 1232 }, { "epoch": 0.47, "learning_rate": 1.7998299654925106e-05, "loss": 1.0925, "step": 1233 }, { "epoch": 0.47, "learning_rate": 1.799452447507505e-05, "loss": 1.1963, "step": 1234 }, { "epoch": 0.47, "learning_rate": 1.7990746135322593e-05, "loss": 1.1538, "step": 1235 }, { "epoch": 0.47, "learning_rate": 1.7986964637161162e-05, "loss": 1.1639, "step": 1236 }, { "epoch": 0.48, "learning_rate": 1.7983179982085415e-05, "loss": 1.1556, "step": 1237 }, { "epoch": 0.48, "learning_rate": 1.7979392171591278e-05, "loss": 1.1381, "step": 1238 }, { "epoch": 0.48, "learning_rate": 1.797560120717591e-05, "loss": 1.147, "step": 1239 }, { "epoch": 0.48, "learning_rate": 1.7971807090337723e-05, "loss": 1.1444, "step": 1240 }, { "epoch": 0.48, "learning_rate": 1.796800982257637e-05, "loss": 1.1407, "step": 1241 }, { "epoch": 0.48, "learning_rate": 1.7964209405392755e-05, "loss": 1.1638, "step": 1242 }, { "epoch": 0.48, "learning_rate": 1.796040584028903e-05, "loss": 1.1723, "step": 1243 }, { "epoch": 0.48, "learning_rate": 1.7956599128768575e-05, "loss": 1.1279, "step": 1244 }, { "epoch": 0.48, "learning_rate": 1.795278927233603e-05, "loss": 1.1088, "step": 1245 }, { "epoch": 0.48, "learning_rate": 1.7948976272497277e-05, "loss": 1.1818, "step": 1246 }, { "epoch": 0.48, "learning_rate": 1.794516013075943e-05, "loss": 1.1807, "step": 1247 }, { "epoch": 0.48, "learning_rate": 1.7941340848630848e-05, "loss": 1.1292, "step": 1248 }, { "epoch": 0.48, "learning_rate": 1.7937518427621143e-05, "loss": 1.1363, "step": 1249 }, { "epoch": 0.48, "learning_rate": 1.793369286924115e-05, "loss": 1.1114, "step": 1250 }, { "epoch": 0.48, "learning_rate": 1.792986417500296e-05, "loss": 1.166, "step": 1251 }, { "epoch": 0.48, "learning_rate": 1.7926032346419892e-05, "loss": 1.1504, "step": 1252 }, { "epoch": 0.48, "learning_rate": 1.792219738500651e-05, "loss": 1.1592, "step": 1253 }, { "epoch": 0.48, "learning_rate": 1.7918359292278614e-05, "loss": 1.1492, "step": 1254 }, { "epoch": 0.48, "learning_rate": 1.791451806975324e-05, "loss": 1.1766, "step": 1255 }, { "epoch": 0.48, "learning_rate": 1.7910673718948668e-05, "loss": 1.1389, "step": 1256 }, { "epoch": 0.48, "learning_rate": 1.7906826241384404e-05, "loss": 1.1399, "step": 1257 }, { "epoch": 0.48, "learning_rate": 1.79029756385812e-05, "loss": 1.0732, "step": 1258 }, { "epoch": 0.48, "learning_rate": 1.7899121912061036e-05, "loss": 1.1314, "step": 1259 }, { "epoch": 0.48, "learning_rate": 1.789526506334713e-05, "loss": 1.0987, "step": 1260 }, { "epoch": 0.48, "learning_rate": 1.789140509396394e-05, "loss": 1.1488, "step": 1261 }, { "epoch": 0.48, "learning_rate": 1.7887542005437138e-05, "loss": 1.1091, "step": 1262 }, { "epoch": 0.49, "learning_rate": 1.7883675799293653e-05, "loss": 1.1225, "step": 1263 }, { "epoch": 0.49, "learning_rate": 1.7879806477061632e-05, "loss": 1.1408, "step": 1264 }, { "epoch": 0.49, "learning_rate": 1.7875934040270453e-05, "loss": 1.1909, "step": 1265 }, { "epoch": 0.49, "learning_rate": 1.7872058490450735e-05, "loss": 1.2358, "step": 1266 }, { "epoch": 0.49, "learning_rate": 1.7868179829134316e-05, "loss": 1.1166, "step": 1267 }, { "epoch": 0.49, "learning_rate": 1.7864298057854274e-05, "loss": 1.1568, "step": 1268 }, { "epoch": 0.49, "learning_rate": 1.7860413178144908e-05, "loss": 1.1744, "step": 1269 }, { "epoch": 0.49, "learning_rate": 1.7856525191541752e-05, "loss": 1.1539, "step": 1270 }, { "epoch": 0.49, "learning_rate": 1.7852634099581563e-05, "loss": 1.1627, "step": 1271 }, { "epoch": 0.49, "learning_rate": 1.784873990380233e-05, "loss": 1.126, "step": 1272 }, { "epoch": 0.49, "learning_rate": 1.7844842605743256e-05, "loss": 1.1569, "step": 1273 }, { "epoch": 0.49, "learning_rate": 1.7840942206944795e-05, "loss": 1.1802, "step": 1274 }, { "epoch": 0.49, "learning_rate": 1.7837038708948606e-05, "loss": 1.1276, "step": 1275 }, { "epoch": 0.49, "learning_rate": 1.7833132113297577e-05, "loss": 1.1501, "step": 1276 }, { "epoch": 0.49, "learning_rate": 1.7829222421535825e-05, "loss": 1.1709, "step": 1277 }, { "epoch": 0.49, "learning_rate": 1.782530963520869e-05, "loss": 1.1013, "step": 1278 }, { "epoch": 0.49, "learning_rate": 1.7821393755862728e-05, "loss": 1.2144, "step": 1279 }, { "epoch": 0.49, "learning_rate": 1.7817474785045727e-05, "loss": 1.1947, "step": 1280 }, { "epoch": 0.49, "learning_rate": 1.781355272430669e-05, "loss": 1.1596, "step": 1281 }, { "epoch": 0.49, "learning_rate": 1.7809627575195846e-05, "loss": 1.1699, "step": 1282 }, { "epoch": 0.49, "learning_rate": 1.7805699339264646e-05, "loss": 1.1807, "step": 1283 }, { "epoch": 0.49, "learning_rate": 1.7801768018065755e-05, "loss": 1.1621, "step": 1284 }, { "epoch": 0.49, "learning_rate": 1.779783361315306e-05, "loss": 1.1364, "step": 1285 }, { "epoch": 0.49, "learning_rate": 1.7793896126081666e-05, "loss": 1.1501, "step": 1286 }, { "epoch": 0.49, "learning_rate": 1.7789955558407898e-05, "loss": 1.1326, "step": 1287 }, { "epoch": 0.49, "learning_rate": 1.7786011911689302e-05, "loss": 1.1189, "step": 1288 }, { "epoch": 0.5, "learning_rate": 1.7782065187484635e-05, "loss": 1.1021, "step": 1289 }, { "epoch": 0.5, "learning_rate": 1.777811538735387e-05, "loss": 1.1713, "step": 1290 }, { "epoch": 0.5, "learning_rate": 1.7774162512858198e-05, "loss": 1.1483, "step": 1291 }, { "epoch": 0.5, "learning_rate": 1.7770206565560034e-05, "loss": 1.1244, "step": 1292 }, { "epoch": 0.5, "learning_rate": 1.776624754702299e-05, "loss": 1.1915, "step": 1293 }, { "epoch": 0.5, "learning_rate": 1.7762285458811906e-05, "loss": 1.1404, "step": 1294 }, { "epoch": 0.5, "learning_rate": 1.775832030249283e-05, "loss": 1.1548, "step": 1295 }, { "epoch": 0.5, "learning_rate": 1.775435207963302e-05, "loss": 1.1346, "step": 1296 }, { "epoch": 0.5, "learning_rate": 1.7750380791800948e-05, "loss": 1.1495, "step": 1297 }, { "epoch": 0.5, "learning_rate": 1.7746406440566305e-05, "loss": 1.1015, "step": 1298 }, { "epoch": 0.5, "learning_rate": 1.7742429027499986e-05, "loss": 1.1466, "step": 1299 }, { "epoch": 0.5, "learning_rate": 1.773844855417409e-05, "loss": 1.1523, "step": 1300 }, { "epoch": 0.5, "learning_rate": 1.7734465022161935e-05, "loss": 1.1428, "step": 1301 }, { "epoch": 0.5, "learning_rate": 1.7730478433038048e-05, "loss": 1.1339, "step": 1302 }, { "epoch": 0.5, "learning_rate": 1.772648878837816e-05, "loss": 1.0725, "step": 1303 }, { "epoch": 0.5, "learning_rate": 1.7722496089759208e-05, "loss": 1.1557, "step": 1304 }, { "epoch": 0.5, "learning_rate": 1.7718500338759348e-05, "loss": 1.1227, "step": 1305 }, { "epoch": 0.5, "learning_rate": 1.7714501536957923e-05, "loss": 1.2001, "step": 1306 }, { "epoch": 0.5, "learning_rate": 1.77104996859355e-05, "loss": 1.1278, "step": 1307 }, { "epoch": 0.5, "learning_rate": 1.7706494787273842e-05, "loss": 1.1305, "step": 1308 }, { "epoch": 0.5, "learning_rate": 1.7702486842555918e-05, "loss": 1.1462, "step": 1309 }, { "epoch": 0.5, "learning_rate": 1.7698475853365903e-05, "loss": 1.123, "step": 1310 }, { "epoch": 0.5, "learning_rate": 1.769446182128917e-05, "loss": 1.153, "step": 1311 }, { "epoch": 0.5, "learning_rate": 1.7690444747912304e-05, "loss": 1.1601, "step": 1312 }, { "epoch": 0.5, "learning_rate": 1.7686424634823086e-05, "loss": 1.1509, "step": 1313 }, { "epoch": 0.5, "learning_rate": 1.7682401483610495e-05, "loss": 1.1919, "step": 1314 }, { "epoch": 0.51, "learning_rate": 1.7678375295864717e-05, "loss": 1.1255, "step": 1315 }, { "epoch": 0.51, "learning_rate": 1.767434607317714e-05, "loss": 1.1836, "step": 1316 }, { "epoch": 0.51, "learning_rate": 1.767031381714034e-05, "loss": 1.1642, "step": 1317 }, { "epoch": 0.51, "learning_rate": 1.7666278529348108e-05, "loss": 1.1463, "step": 1318 }, { "epoch": 0.51, "learning_rate": 1.766224021139542e-05, "loss": 1.1298, "step": 1319 }, { "epoch": 0.51, "learning_rate": 1.765819886487846e-05, "loss": 1.1528, "step": 1320 }, { "epoch": 0.51, "learning_rate": 1.7654154491394596e-05, "loss": 1.1519, "step": 1321 }, { "epoch": 0.51, "learning_rate": 1.7650107092542403e-05, "loss": 1.1175, "step": 1322 }, { "epoch": 0.51, "learning_rate": 1.764605666992165e-05, "loss": 1.1314, "step": 1323 }, { "epoch": 0.51, "learning_rate": 1.7642003225133307e-05, "loss": 1.1785, "step": 1324 }, { "epoch": 0.51, "learning_rate": 1.7637946759779523e-05, "loss": 1.0907, "step": 1325 }, { "epoch": 0.51, "learning_rate": 1.763388727546365e-05, "loss": 1.1351, "step": 1326 }, { "epoch": 0.51, "learning_rate": 1.7629824773790236e-05, "loss": 1.144, "step": 1327 }, { "epoch": 0.51, "learning_rate": 1.762575925636502e-05, "loss": 1.158, "step": 1328 }, { "epoch": 0.51, "learning_rate": 1.762169072479493e-05, "loss": 1.1254, "step": 1329 }, { "epoch": 0.51, "learning_rate": 1.7617619180688087e-05, "loss": 1.1472, "step": 1330 }, { "epoch": 0.51, "learning_rate": 1.76135446256538e-05, "loss": 1.1018, "step": 1331 }, { "epoch": 0.51, "learning_rate": 1.760946706130257e-05, "loss": 1.1425, "step": 1332 }, { "epoch": 0.51, "learning_rate": 1.7605386489246096e-05, "loss": 1.165, "step": 1333 }, { "epoch": 0.51, "learning_rate": 1.7601302911097257e-05, "loss": 1.1455, "step": 1334 }, { "epoch": 0.51, "learning_rate": 1.7597216328470112e-05, "loss": 1.1545, "step": 1335 }, { "epoch": 0.51, "learning_rate": 1.7593126742979927e-05, "loss": 1.155, "step": 1336 }, { "epoch": 0.51, "learning_rate": 1.7589034156243135e-05, "loss": 1.1642, "step": 1337 }, { "epoch": 0.51, "learning_rate": 1.7584938569877376e-05, "loss": 1.1466, "step": 1338 }, { "epoch": 0.51, "learning_rate": 1.7580839985501454e-05, "loss": 1.1309, "step": 1339 }, { "epoch": 0.51, "learning_rate": 1.757673840473538e-05, "loss": 1.1352, "step": 1340 }, { "epoch": 0.52, "learning_rate": 1.7572633829200327e-05, "loss": 1.1567, "step": 1341 }, { "epoch": 0.52, "learning_rate": 1.756852626051867e-05, "loss": 1.0795, "step": 1342 }, { "epoch": 0.52, "learning_rate": 1.7564415700313954e-05, "loss": 1.1117, "step": 1343 }, { "epoch": 0.52, "learning_rate": 1.7560302150210916e-05, "loss": 1.1059, "step": 1344 }, { "epoch": 0.52, "learning_rate": 1.755618561183547e-05, "loss": 1.1237, "step": 1345 }, { "epoch": 0.52, "learning_rate": 1.7552066086814715e-05, "loss": 1.1762, "step": 1346 }, { "epoch": 0.52, "learning_rate": 1.7547943576776924e-05, "loss": 1.1343, "step": 1347 }, { "epoch": 0.52, "learning_rate": 1.754381808335155e-05, "loss": 1.1895, "step": 1348 }, { "epoch": 0.52, "learning_rate": 1.753968960816924e-05, "loss": 1.125, "step": 1349 }, { "epoch": 0.52, "learning_rate": 1.7535558152861797e-05, "loss": 1.1236, "step": 1350 }, { "epoch": 0.52, "learning_rate": 1.7531423719062216e-05, "loss": 1.1397, "step": 1351 }, { "epoch": 0.52, "learning_rate": 1.7527286308404668e-05, "loss": 1.1401, "step": 1352 }, { "epoch": 0.52, "learning_rate": 1.75231459225245e-05, "loss": 1.1124, "step": 1353 }, { "epoch": 0.52, "learning_rate": 1.7519002563058235e-05, "loss": 1.1979, "step": 1354 }, { "epoch": 0.52, "learning_rate": 1.7514856231643565e-05, "loss": 1.2207, "step": 1355 }, { "epoch": 0.52, "learning_rate": 1.7510706929919364e-05, "loss": 1.092, "step": 1356 }, { "epoch": 0.52, "learning_rate": 1.7506554659525684e-05, "loss": 1.1641, "step": 1357 }, { "epoch": 0.52, "learning_rate": 1.7502399422103735e-05, "loss": 1.1574, "step": 1358 }, { "epoch": 0.52, "learning_rate": 1.749824121929592e-05, "loss": 1.0818, "step": 1359 }, { "epoch": 0.52, "learning_rate": 1.749408005274579e-05, "loss": 1.119, "step": 1360 }, { "epoch": 0.52, "learning_rate": 1.748991592409809e-05, "loss": 1.1609, "step": 1361 }, { "epoch": 0.52, "learning_rate": 1.7485748834998726e-05, "loss": 1.2478, "step": 1362 }, { "epoch": 0.52, "learning_rate": 1.748157878709477e-05, "loss": 1.1156, "step": 1363 }, { "epoch": 0.52, "learning_rate": 1.7477405782034475e-05, "loss": 1.1909, "step": 1364 }, { "epoch": 0.52, "learning_rate": 1.747322982146725e-05, "loss": 1.2021, "step": 1365 }, { "epoch": 0.52, "learning_rate": 1.746905090704368e-05, "loss": 1.1566, "step": 1366 }, { "epoch": 0.53, "learning_rate": 1.746486904041552e-05, "loss": 1.1509, "step": 1367 }, { "epoch": 0.53, "learning_rate": 1.746068422323568e-05, "loss": 1.1378, "step": 1368 }, { "epoch": 0.53, "learning_rate": 1.745649645715825e-05, "loss": 1.1768, "step": 1369 }, { "epoch": 0.53, "learning_rate": 1.7452305743838475e-05, "loss": 1.1444, "step": 1370 }, { "epoch": 0.53, "learning_rate": 1.7448112084932772e-05, "loss": 1.1072, "step": 1371 }, { "epoch": 0.53, "learning_rate": 1.744391548209872e-05, "loss": 1.1743, "step": 1372 }, { "epoch": 0.53, "learning_rate": 1.7439715936995062e-05, "loss": 1.1042, "step": 1373 }, { "epoch": 0.53, "learning_rate": 1.7435513451281697e-05, "loss": 1.1562, "step": 1374 }, { "epoch": 0.53, "learning_rate": 1.7431308026619703e-05, "loss": 1.199, "step": 1375 }, { "epoch": 0.53, "learning_rate": 1.7427099664671302e-05, "loss": 1.1957, "step": 1376 }, { "epoch": 0.53, "learning_rate": 1.742288836709988e-05, "loss": 1.1817, "step": 1377 }, { "epoch": 0.53, "learning_rate": 1.7418674135569994e-05, "loss": 1.1355, "step": 1378 }, { "epoch": 0.53, "learning_rate": 1.7414456971747355e-05, "loss": 1.1754, "step": 1379 }, { "epoch": 0.53, "learning_rate": 1.741023687729883e-05, "loss": 1.1607, "step": 1380 }, { "epoch": 0.53, "learning_rate": 1.740601385389245e-05, "loss": 1.1905, "step": 1381 }, { "epoch": 0.53, "learning_rate": 1.740178790319739e-05, "loss": 1.0949, "step": 1382 }, { "epoch": 0.53, "learning_rate": 1.7397559026884005e-05, "loss": 1.144, "step": 1383 }, { "epoch": 0.53, "learning_rate": 1.7393327226623782e-05, "loss": 1.1753, "step": 1384 }, { "epoch": 0.53, "learning_rate": 1.7389092504089382e-05, "loss": 1.0686, "step": 1385 }, { "epoch": 0.53, "learning_rate": 1.7384854860954616e-05, "loss": 1.1827, "step": 1386 }, { "epoch": 0.53, "learning_rate": 1.7380614298894443e-05, "loss": 1.1307, "step": 1387 }, { "epoch": 0.53, "learning_rate": 1.737637081958498e-05, "loss": 1.1161, "step": 1388 }, { "epoch": 0.53, "learning_rate": 1.7372124424703504e-05, "loss": 1.1746, "step": 1389 }, { "epoch": 0.53, "learning_rate": 1.736787511592843e-05, "loss": 1.195, "step": 1390 }, { "epoch": 0.53, "learning_rate": 1.736362289493934e-05, "loss": 1.1594, "step": 1391 }, { "epoch": 0.53, "learning_rate": 1.7359367763416953e-05, "loss": 1.0859, "step": 1392 }, { "epoch": 0.54, "learning_rate": 1.735510972304315e-05, "loss": 1.1208, "step": 1393 }, { "epoch": 0.54, "learning_rate": 1.735084877550096e-05, "loss": 1.1173, "step": 1394 }, { "epoch": 0.54, "learning_rate": 1.7346584922474544e-05, "loss": 1.1737, "step": 1395 }, { "epoch": 0.54, "learning_rate": 1.734231816564924e-05, "loss": 1.1277, "step": 1396 }, { "epoch": 0.54, "learning_rate": 1.7338048506711515e-05, "loss": 1.1569, "step": 1397 }, { "epoch": 0.54, "learning_rate": 1.733377594734899e-05, "loss": 1.1556, "step": 1398 }, { "epoch": 0.54, "learning_rate": 1.732950048925042e-05, "loss": 1.0957, "step": 1399 }, { "epoch": 0.54, "learning_rate": 1.7325222134105724e-05, "loss": 1.1677, "step": 1400 }, { "epoch": 0.54, "learning_rate": 1.732094088360595e-05, "loss": 1.1388, "step": 1401 }, { "epoch": 0.54, "learning_rate": 1.7316656739443306e-05, "loss": 1.1591, "step": 1402 }, { "epoch": 0.54, "learning_rate": 1.7312369703311133e-05, "loss": 1.1528, "step": 1403 }, { "epoch": 0.54, "learning_rate": 1.730807977690392e-05, "loss": 1.1133, "step": 1404 }, { "epoch": 0.54, "learning_rate": 1.7303786961917293e-05, "loss": 1.1816, "step": 1405 }, { "epoch": 0.54, "learning_rate": 1.729949126004802e-05, "loss": 1.1455, "step": 1406 }, { "epoch": 0.54, "learning_rate": 1.7295192672994017e-05, "loss": 1.1532, "step": 1407 }, { "epoch": 0.54, "learning_rate": 1.7290891202454335e-05, "loss": 1.1582, "step": 1408 }, { "epoch": 0.54, "learning_rate": 1.728658685012917e-05, "loss": 1.2305, "step": 1409 }, { "epoch": 0.54, "learning_rate": 1.728227961771985e-05, "loss": 1.1239, "step": 1410 }, { "epoch": 0.54, "learning_rate": 1.7277969506928837e-05, "loss": 1.1104, "step": 1411 }, { "epoch": 0.54, "learning_rate": 1.7273656519459753e-05, "loss": 1.1337, "step": 1412 }, { "epoch": 0.54, "learning_rate": 1.726934065701733e-05, "loss": 1.1278, "step": 1413 }, { "epoch": 0.54, "learning_rate": 1.7265021921307456e-05, "loss": 1.1274, "step": 1414 }, { "epoch": 0.54, "learning_rate": 1.7260700314037146e-05, "loss": 1.1918, "step": 1415 }, { "epoch": 0.54, "learning_rate": 1.7256375836914547e-05, "loss": 1.1647, "step": 1416 }, { "epoch": 0.54, "learning_rate": 1.7252048491648946e-05, "loss": 1.118, "step": 1417 }, { "epoch": 0.54, "learning_rate": 1.7247718279950766e-05, "loss": 1.1422, "step": 1418 }, { "epoch": 0.55, "learning_rate": 1.724338520353156e-05, "loss": 1.0953, "step": 1419 }, { "epoch": 0.55, "learning_rate": 1.7239049264104005e-05, "loss": 1.1557, "step": 1420 }, { "epoch": 0.55, "learning_rate": 1.7234710463381924e-05, "loss": 1.1616, "step": 1421 }, { "epoch": 0.55, "learning_rate": 1.723036880308026e-05, "loss": 1.1118, "step": 1422 }, { "epoch": 0.55, "learning_rate": 1.7226024284915097e-05, "loss": 1.1239, "step": 1423 }, { "epoch": 0.55, "learning_rate": 1.722167691060363e-05, "loss": 1.1126, "step": 1424 }, { "epoch": 0.55, "learning_rate": 1.721732668186421e-05, "loss": 1.135, "step": 1425 }, { "epoch": 0.55, "learning_rate": 1.7212973600416285e-05, "loss": 1.148, "step": 1426 }, { "epoch": 0.55, "learning_rate": 1.7208617667980458e-05, "loss": 1.1112, "step": 1427 }, { "epoch": 0.55, "learning_rate": 1.7204258886278446e-05, "loss": 1.1277, "step": 1428 }, { "epoch": 0.55, "learning_rate": 1.7199897257033093e-05, "loss": 1.1086, "step": 1429 }, { "epoch": 0.55, "learning_rate": 1.7195532781968368e-05, "loss": 1.143, "step": 1430 }, { "epoch": 0.55, "learning_rate": 1.7191165462809366e-05, "loss": 1.1497, "step": 1431 }, { "epoch": 0.55, "learning_rate": 1.7186795301282307e-05, "loss": 1.1306, "step": 1432 }, { "epoch": 0.55, "learning_rate": 1.7182422299114537e-05, "loss": 1.1426, "step": 1433 }, { "epoch": 0.55, "learning_rate": 1.7178046458034513e-05, "loss": 1.1173, "step": 1434 }, { "epoch": 0.55, "learning_rate": 1.717366777977183e-05, "loss": 1.1256, "step": 1435 }, { "epoch": 0.55, "learning_rate": 1.7169286266057196e-05, "loss": 1.1466, "step": 1436 }, { "epoch": 0.55, "learning_rate": 1.716490191862244e-05, "loss": 1.1593, "step": 1437 }, { "epoch": 0.55, "learning_rate": 1.716051473920051e-05, "loss": 1.1011, "step": 1438 }, { "epoch": 0.55, "learning_rate": 1.7156124729525478e-05, "loss": 1.0508, "step": 1439 }, { "epoch": 0.55, "learning_rate": 1.7151731891332532e-05, "loss": 1.1036, "step": 1440 }, { "epoch": 0.55, "learning_rate": 1.7147336226357977e-05, "loss": 1.1706, "step": 1441 }, { "epoch": 0.55, "learning_rate": 1.7142937736339232e-05, "loss": 1.1484, "step": 1442 }, { "epoch": 0.55, "learning_rate": 1.7138536423014844e-05, "loss": 1.1499, "step": 1443 }, { "epoch": 0.55, "learning_rate": 1.7134132288124464e-05, "loss": 1.0698, "step": 1444 }, { "epoch": 0.56, "learning_rate": 1.7129725333408866e-05, "loss": 1.0988, "step": 1445 }, { "epoch": 0.56, "learning_rate": 1.7125315560609933e-05, "loss": 1.1256, "step": 1446 }, { "epoch": 0.56, "learning_rate": 1.7120902971470665e-05, "loss": 1.1496, "step": 1447 }, { "epoch": 0.56, "learning_rate": 1.7116487567735175e-05, "loss": 1.1137, "step": 1448 }, { "epoch": 0.56, "learning_rate": 1.711206935114869e-05, "loss": 1.0523, "step": 1449 }, { "epoch": 0.56, "learning_rate": 1.7107648323457546e-05, "loss": 1.0813, "step": 1450 }, { "epoch": 0.56, "learning_rate": 1.710322448640919e-05, "loss": 1.1713, "step": 1451 }, { "epoch": 0.56, "learning_rate": 1.7098797841752176e-05, "loss": 1.2123, "step": 1452 }, { "epoch": 0.56, "learning_rate": 1.709436839123618e-05, "loss": 1.1163, "step": 1453 }, { "epoch": 0.56, "learning_rate": 1.7089936136611976e-05, "loss": 1.1544, "step": 1454 }, { "epoch": 0.56, "learning_rate": 1.708550107963145e-05, "loss": 1.1316, "step": 1455 }, { "epoch": 0.56, "learning_rate": 1.7081063222047594e-05, "loss": 1.1707, "step": 1456 }, { "epoch": 0.56, "learning_rate": 1.707662256561451e-05, "loss": 1.1711, "step": 1457 }, { "epoch": 0.56, "learning_rate": 1.70721791120874e-05, "loss": 1.1357, "step": 1458 }, { "epoch": 0.56, "learning_rate": 1.7067732863222583e-05, "loss": 1.0989, "step": 1459 }, { "epoch": 0.56, "learning_rate": 1.706328382077747e-05, "loss": 1.164, "step": 1460 }, { "epoch": 0.56, "learning_rate": 1.7058831986510588e-05, "loss": 1.1656, "step": 1461 }, { "epoch": 0.56, "learning_rate": 1.7054377362181555e-05, "loss": 1.1175, "step": 1462 }, { "epoch": 0.56, "learning_rate": 1.7049919949551103e-05, "loss": 1.131, "step": 1463 }, { "epoch": 0.56, "learning_rate": 1.704545975038106e-05, "loss": 1.1814, "step": 1464 }, { "epoch": 0.56, "learning_rate": 1.7040996766434355e-05, "loss": 1.1326, "step": 1465 }, { "epoch": 0.56, "learning_rate": 1.703653099947502e-05, "loss": 1.1008, "step": 1466 }, { "epoch": 0.56, "learning_rate": 1.7032062451268192e-05, "loss": 1.2119, "step": 1467 }, { "epoch": 0.56, "learning_rate": 1.7027591123580093e-05, "loss": 1.1173, "step": 1468 }, { "epoch": 0.56, "learning_rate": 1.7023117018178062e-05, "loss": 1.1261, "step": 1469 }, { "epoch": 0.56, "learning_rate": 1.7018640136830518e-05, "loss": 1.1237, "step": 1470 }, { "epoch": 0.57, "learning_rate": 1.701416048130699e-05, "loss": 1.0964, "step": 1471 }, { "epoch": 0.57, "learning_rate": 1.7009678053378103e-05, "loss": 1.1342, "step": 1472 }, { "epoch": 0.57, "learning_rate": 1.7005192854815566e-05, "loss": 1.1488, "step": 1473 }, { "epoch": 0.57, "learning_rate": 1.7000704887392196e-05, "loss": 1.096, "step": 1474 }, { "epoch": 0.57, "learning_rate": 1.6996214152881895e-05, "loss": 1.1106, "step": 1475 }, { "epoch": 0.57, "learning_rate": 1.6991720653059668e-05, "loss": 1.1288, "step": 1476 }, { "epoch": 0.57, "learning_rate": 1.698722438970161e-05, "loss": 1.153, "step": 1477 }, { "epoch": 0.57, "learning_rate": 1.69827253645849e-05, "loss": 1.1413, "step": 1478 }, { "epoch": 0.57, "learning_rate": 1.697822357948782e-05, "loss": 1.1283, "step": 1479 }, { "epoch": 0.57, "learning_rate": 1.6973719036189732e-05, "loss": 1.1538, "step": 1480 }, { "epoch": 0.57, "learning_rate": 1.6969211736471102e-05, "loss": 1.1567, "step": 1481 }, { "epoch": 0.57, "learning_rate": 1.6964701682113477e-05, "loss": 1.1285, "step": 1482 }, { "epoch": 0.57, "learning_rate": 1.6960188874899485e-05, "loss": 1.0663, "step": 1483 }, { "epoch": 0.57, "learning_rate": 1.6955673316612863e-05, "loss": 1.1429, "step": 1484 }, { "epoch": 0.57, "learning_rate": 1.6951155009038412e-05, "loss": 1.1199, "step": 1485 }, { "epoch": 0.57, "learning_rate": 1.6946633953962036e-05, "loss": 1.1481, "step": 1486 }, { "epoch": 0.57, "learning_rate": 1.6942110153170722e-05, "loss": 1.1301, "step": 1487 }, { "epoch": 0.57, "learning_rate": 1.6937583608452533e-05, "loss": 1.096, "step": 1488 }, { "epoch": 0.57, "learning_rate": 1.6933054321596634e-05, "loss": 1.0767, "step": 1489 }, { "epoch": 0.57, "learning_rate": 1.692852229439325e-05, "loss": 1.1589, "step": 1490 }, { "epoch": 0.57, "learning_rate": 1.6923987528633713e-05, "loss": 1.0666, "step": 1491 }, { "epoch": 0.57, "learning_rate": 1.6919450026110424e-05, "loss": 1.1091, "step": 1492 }, { "epoch": 0.57, "learning_rate": 1.6914909788616864e-05, "loss": 1.1014, "step": 1493 }, { "epoch": 0.57, "learning_rate": 1.6910366817947605e-05, "loss": 1.108, "step": 1494 }, { "epoch": 0.57, "learning_rate": 1.6905821115898296e-05, "loss": 1.0631, "step": 1495 }, { "epoch": 0.57, "learning_rate": 1.6901272684265657e-05, "loss": 1.1235, "step": 1496 }, { "epoch": 0.58, "learning_rate": 1.6896721524847497e-05, "loss": 1.1491, "step": 1497 }, { "epoch": 0.58, "learning_rate": 1.6892167639442702e-05, "loss": 1.1565, "step": 1498 }, { "epoch": 0.58, "learning_rate": 1.6887611029851227e-05, "loss": 1.1075, "step": 1499 }, { "epoch": 0.58, "learning_rate": 1.6883051697874115e-05, "loss": 1.1273, "step": 1500 }, { "epoch": 0.58, "learning_rate": 1.687848964531348e-05, "loss": 1.165, "step": 1501 }, { "epoch": 0.58, "learning_rate": 1.687392487397251e-05, "loss": 1.1362, "step": 1502 }, { "epoch": 0.58, "learning_rate": 1.6869357385655467e-05, "loss": 1.1123, "step": 1503 }, { "epoch": 0.58, "learning_rate": 1.686478718216769e-05, "loss": 1.1114, "step": 1504 }, { "epoch": 0.58, "learning_rate": 1.6860214265315595e-05, "loss": 1.0815, "step": 1505 }, { "epoch": 0.58, "learning_rate": 1.6855638636906654e-05, "loss": 1.1901, "step": 1506 }, { "epoch": 0.58, "learning_rate": 1.6851060298749435e-05, "loss": 1.1406, "step": 1507 }, { "epoch": 0.58, "learning_rate": 1.684647925265355e-05, "loss": 1.1007, "step": 1508 }, { "epoch": 0.58, "learning_rate": 1.6841895500429713e-05, "loss": 1.1008, "step": 1509 }, { "epoch": 0.58, "learning_rate": 1.683730904388968e-05, "loss": 1.1207, "step": 1510 }, { "epoch": 0.58, "learning_rate": 1.6832719884846285e-05, "loss": 1.093, "step": 1511 }, { "epoch": 0.58, "learning_rate": 1.6828128025113435e-05, "loss": 1.0912, "step": 1512 }, { "epoch": 0.58, "learning_rate": 1.68235334665061e-05, "loss": 1.1773, "step": 1513 }, { "epoch": 0.58, "learning_rate": 1.6818936210840315e-05, "loss": 1.168, "step": 1514 }, { "epoch": 0.58, "learning_rate": 1.681433625993319e-05, "loss": 1.1598, "step": 1515 }, { "epoch": 0.58, "learning_rate": 1.6809733615602888e-05, "loss": 1.13, "step": 1516 }, { "epoch": 0.58, "learning_rate": 1.6805128279668642e-05, "loss": 1.1425, "step": 1517 }, { "epoch": 0.58, "learning_rate": 1.6800520253950755e-05, "loss": 1.1346, "step": 1518 }, { "epoch": 0.58, "learning_rate": 1.679590954027058e-05, "loss": 1.1511, "step": 1519 }, { "epoch": 0.58, "learning_rate": 1.6791296140450547e-05, "loss": 1.1625, "step": 1520 }, { "epoch": 0.58, "learning_rate": 1.6786680056314137e-05, "loss": 1.1458, "step": 1521 }, { "epoch": 0.58, "learning_rate": 1.67820612896859e-05, "loss": 1.1405, "step": 1522 }, { "epoch": 0.59, "learning_rate": 1.6777439842391437e-05, "loss": 1.1579, "step": 1523 }, { "epoch": 0.59, "learning_rate": 1.6772815716257414e-05, "loss": 1.1464, "step": 1524 }, { "epoch": 0.59, "learning_rate": 1.6768188913111557e-05, "loss": 1.1824, "step": 1525 }, { "epoch": 0.59, "learning_rate": 1.6763559434782646e-05, "loss": 1.1005, "step": 1526 }, { "epoch": 0.59, "learning_rate": 1.6758927283100526e-05, "loss": 1.0909, "step": 1527 }, { "epoch": 0.59, "learning_rate": 1.675429245989609e-05, "loss": 1.087, "step": 1528 }, { "epoch": 0.59, "learning_rate": 1.6749654967001287e-05, "loss": 1.1558, "step": 1529 }, { "epoch": 0.59, "learning_rate": 1.6745014806249132e-05, "loss": 1.1683, "step": 1530 }, { "epoch": 0.59, "learning_rate": 1.6740371979473684e-05, "loss": 1.1273, "step": 1531 }, { "epoch": 0.59, "learning_rate": 1.673572648851006e-05, "loss": 1.1671, "step": 1532 }, { "epoch": 0.59, "learning_rate": 1.6731078335194423e-05, "loss": 1.1572, "step": 1533 }, { "epoch": 0.59, "learning_rate": 1.6726427521364005e-05, "loss": 1.1209, "step": 1534 }, { "epoch": 0.59, "learning_rate": 1.6721774048857072e-05, "loss": 1.1436, "step": 1535 }, { "epoch": 0.59, "learning_rate": 1.6717117919512946e-05, "loss": 1.1427, "step": 1536 }, { "epoch": 0.59, "learning_rate": 1.6712459135172005e-05, "loss": 1.185, "step": 1537 }, { "epoch": 0.59, "learning_rate": 1.6707797697675673e-05, "loss": 1.1649, "step": 1538 }, { "epoch": 0.59, "learning_rate": 1.6703133608866415e-05, "loss": 1.1623, "step": 1539 }, { "epoch": 0.59, "learning_rate": 1.669846687058776e-05, "loss": 1.1534, "step": 1540 }, { "epoch": 0.59, "learning_rate": 1.669379748468427e-05, "loss": 1.0903, "step": 1541 }, { "epoch": 0.59, "learning_rate": 1.668912545300156e-05, "loss": 1.1981, "step": 1542 }, { "epoch": 0.59, "learning_rate": 1.6684450777386286e-05, "loss": 1.1276, "step": 1543 }, { "epoch": 0.59, "learning_rate": 1.667977345968616e-05, "loss": 1.1578, "step": 1544 }, { "epoch": 0.59, "learning_rate": 1.6675093501749927e-05, "loss": 1.1229, "step": 1545 }, { "epoch": 0.59, "learning_rate": 1.6670410905427378e-05, "loss": 1.1324, "step": 1546 }, { "epoch": 0.59, "learning_rate": 1.6665725672569348e-05, "loss": 1.1537, "step": 1547 }, { "epoch": 0.59, "learning_rate": 1.666103780502772e-05, "loss": 1.0611, "step": 1548 }, { "epoch": 0.6, "learning_rate": 1.6656347304655403e-05, "loss": 1.0827, "step": 1549 }, { "epoch": 0.6, "learning_rate": 1.6651654173306366e-05, "loss": 1.0847, "step": 1550 }, { "epoch": 0.6, "learning_rate": 1.6646958412835595e-05, "loss": 1.1355, "step": 1551 }, { "epoch": 0.6, "learning_rate": 1.6642260025099144e-05, "loss": 1.1875, "step": 1552 }, { "epoch": 0.6, "learning_rate": 1.6637559011954084e-05, "loss": 1.1541, "step": 1553 }, { "epoch": 0.6, "learning_rate": 1.6632855375258525e-05, "loss": 1.1234, "step": 1554 }, { "epoch": 0.6, "learning_rate": 1.6628149116871622e-05, "loss": 1.2131, "step": 1555 }, { "epoch": 0.6, "learning_rate": 1.6623440238653566e-05, "loss": 1.1558, "step": 1556 }, { "epoch": 0.6, "learning_rate": 1.661872874246558e-05, "loss": 1.1919, "step": 1557 }, { "epoch": 0.6, "learning_rate": 1.6614014630169916e-05, "loss": 1.128, "step": 1558 }, { "epoch": 0.6, "learning_rate": 1.660929790362987e-05, "loss": 1.2177, "step": 1559 }, { "epoch": 0.6, "learning_rate": 1.6604578564709764e-05, "loss": 1.1265, "step": 1560 }, { "epoch": 0.6, "learning_rate": 1.6599856615274965e-05, "loss": 1.1456, "step": 1561 }, { "epoch": 0.6, "learning_rate": 1.6595132057191855e-05, "loss": 1.1382, "step": 1562 }, { "epoch": 0.6, "learning_rate": 1.6590404892327855e-05, "loss": 1.165, "step": 1563 }, { "epoch": 0.6, "learning_rate": 1.658567512255142e-05, "loss": 1.0808, "step": 1564 }, { "epoch": 0.6, "learning_rate": 1.658094274973203e-05, "loss": 1.1443, "step": 1565 }, { "epoch": 0.6, "learning_rate": 1.6576207775740195e-05, "loss": 1.1127, "step": 1566 }, { "epoch": 0.6, "learning_rate": 1.6571470202447447e-05, "loss": 1.0815, "step": 1567 }, { "epoch": 0.6, "learning_rate": 1.656673003172636e-05, "loss": 1.1312, "step": 1568 }, { "epoch": 0.6, "learning_rate": 1.6561987265450523e-05, "loss": 1.1391, "step": 1569 }, { "epoch": 0.6, "learning_rate": 1.655724190549455e-05, "loss": 1.2001, "step": 1570 }, { "epoch": 0.6, "learning_rate": 1.6552493953734093e-05, "loss": 1.1495, "step": 1571 }, { "epoch": 0.6, "learning_rate": 1.654774341204581e-05, "loss": 1.17, "step": 1572 }, { "epoch": 0.6, "learning_rate": 1.6542990282307394e-05, "loss": 1.1036, "step": 1573 }, { "epoch": 0.6, "learning_rate": 1.6538234566397562e-05, "loss": 1.1767, "step": 1574 }, { "epoch": 0.61, "learning_rate": 1.653347626619605e-05, "loss": 1.1064, "step": 1575 }, { "epoch": 0.61, "learning_rate": 1.6528715383583614e-05, "loss": 1.1123, "step": 1576 }, { "epoch": 0.61, "learning_rate": 1.6523951920442032e-05, "loss": 1.0914, "step": 1577 }, { "epoch": 0.61, "learning_rate": 1.651918587865411e-05, "loss": 1.0962, "step": 1578 }, { "epoch": 0.61, "learning_rate": 1.6514417260103654e-05, "loss": 1.0886, "step": 1579 }, { "epoch": 0.61, "learning_rate": 1.6509646066675507e-05, "loss": 1.1313, "step": 1580 }, { "epoch": 0.61, "learning_rate": 1.6504872300255526e-05, "loss": 1.1243, "step": 1581 }, { "epoch": 0.61, "learning_rate": 1.6500095962730575e-05, "loss": 1.1508, "step": 1582 }, { "epoch": 0.61, "learning_rate": 1.6495317055988543e-05, "loss": 1.1663, "step": 1583 }, { "epoch": 0.61, "learning_rate": 1.6490535581918336e-05, "loss": 1.1819, "step": 1584 }, { "epoch": 0.61, "learning_rate": 1.6485751542409867e-05, "loss": 1.093, "step": 1585 }, { "epoch": 0.61, "learning_rate": 1.6480964939354072e-05, "loss": 1.1607, "step": 1586 }, { "epoch": 0.61, "learning_rate": 1.647617577464289e-05, "loss": 1.0956, "step": 1587 }, { "epoch": 0.61, "learning_rate": 1.6471384050169286e-05, "loss": 1.0721, "step": 1588 }, { "epoch": 0.61, "learning_rate": 1.646658976782722e-05, "loss": 1.1016, "step": 1589 }, { "epoch": 0.61, "learning_rate": 1.6461792929511683e-05, "loss": 1.1255, "step": 1590 }, { "epoch": 0.61, "learning_rate": 1.645699353711865e-05, "loss": 1.078, "step": 1591 }, { "epoch": 0.61, "learning_rate": 1.645219159254514e-05, "loss": 1.1717, "step": 1592 }, { "epoch": 0.61, "learning_rate": 1.6447387097689146e-05, "loss": 1.0949, "step": 1593 }, { "epoch": 0.61, "learning_rate": 1.6442580054449693e-05, "loss": 1.1667, "step": 1594 }, { "epoch": 0.61, "learning_rate": 1.64377704647268e-05, "loss": 1.0828, "step": 1595 }, { "epoch": 0.61, "learning_rate": 1.6432958330421497e-05, "loss": 1.0963, "step": 1596 }, { "epoch": 0.61, "learning_rate": 1.6428143653435826e-05, "loss": 1.0947, "step": 1597 }, { "epoch": 0.61, "learning_rate": 1.6423326435672823e-05, "loss": 1.1273, "step": 1598 }, { "epoch": 0.61, "learning_rate": 1.6418506679036537e-05, "loss": 1.0999, "step": 1599 }, { "epoch": 0.61, "learning_rate": 1.6413684385432014e-05, "loss": 1.1295, "step": 1600 }, { "epoch": 0.61, "eval_loss": 1.1390260457992554, "eval_runtime": 688.6432, "eval_samples_per_second": 4.939, "eval_steps_per_second": 0.309, "step": 1600 }, { "epoch": 0.62, "learning_rate": 1.640885955676531e-05, "loss": 1.1517, "step": 1601 }, { "epoch": 0.62, "learning_rate": 1.640403219494347e-05, "loss": 1.0478, "step": 1602 }, { "epoch": 0.62, "learning_rate": 1.6399202301874556e-05, "loss": 1.1834, "step": 1603 }, { "epoch": 0.62, "learning_rate": 1.6394369879467623e-05, "loss": 1.0593, "step": 1604 }, { "epoch": 0.62, "learning_rate": 1.6389534929632726e-05, "loss": 1.0406, "step": 1605 }, { "epoch": 0.62, "learning_rate": 1.638469745428092e-05, "loss": 1.1154, "step": 1606 }, { "epoch": 0.62, "learning_rate": 1.6379857455324253e-05, "loss": 1.1349, "step": 1607 }, { "epoch": 0.62, "learning_rate": 1.6375014934675778e-05, "loss": 1.1466, "step": 1608 }, { "epoch": 0.62, "learning_rate": 1.6370169894249545e-05, "loss": 1.0757, "step": 1609 }, { "epoch": 0.62, "learning_rate": 1.636532233596059e-05, "loss": 1.1128, "step": 1610 }, { "epoch": 0.62, "learning_rate": 1.6360472261724955e-05, "loss": 1.1102, "step": 1611 }, { "epoch": 0.62, "learning_rate": 1.635561967345967e-05, "loss": 1.1495, "step": 1612 }, { "epoch": 0.62, "learning_rate": 1.635076457308276e-05, "loss": 1.1222, "step": 1613 }, { "epoch": 0.62, "learning_rate": 1.634590696251325e-05, "loss": 1.1171, "step": 1614 }, { "epoch": 0.62, "learning_rate": 1.6341046843671145e-05, "loss": 1.1079, "step": 1615 }, { "epoch": 0.62, "learning_rate": 1.6336184218477445e-05, "loss": 1.1671, "step": 1616 }, { "epoch": 0.62, "learning_rate": 1.6331319088854155e-05, "loss": 1.1554, "step": 1617 }, { "epoch": 0.62, "learning_rate": 1.632645145672425e-05, "loss": 1.0443, "step": 1618 }, { "epoch": 0.62, "learning_rate": 1.63215813240117e-05, "loss": 1.108, "step": 1619 }, { "epoch": 0.62, "learning_rate": 1.6316708692641468e-05, "loss": 1.1443, "step": 1620 }, { "epoch": 0.62, "learning_rate": 1.63118335645395e-05, "loss": 1.1595, "step": 1621 }, { "epoch": 0.62, "learning_rate": 1.630695594163274e-05, "loss": 1.0954, "step": 1622 }, { "epoch": 0.62, "learning_rate": 1.6302075825849096e-05, "loss": 1.1407, "step": 1623 }, { "epoch": 0.62, "learning_rate": 1.6297193219117484e-05, "loss": 1.1136, "step": 1624 }, { "epoch": 0.62, "learning_rate": 1.629230812336779e-05, "loss": 1.0541, "step": 1625 }, { "epoch": 0.62, "learning_rate": 1.6287420540530895e-05, "loss": 1.1217, "step": 1626 }, { "epoch": 0.62, "learning_rate": 1.6282530472538645e-05, "loss": 1.1496, "step": 1627 }, { "epoch": 0.63, "learning_rate": 1.6277637921323896e-05, "loss": 1.1318, "step": 1628 }, { "epoch": 0.63, "learning_rate": 1.6272742888820453e-05, "loss": 1.1416, "step": 1629 }, { "epoch": 0.63, "learning_rate": 1.6267845376963136e-05, "loss": 1.0649, "step": 1630 }, { "epoch": 0.63, "learning_rate": 1.626294538768771e-05, "loss": 1.1263, "step": 1631 }, { "epoch": 0.63, "learning_rate": 1.625804292293095e-05, "loss": 1.1191, "step": 1632 }, { "epoch": 0.63, "learning_rate": 1.6253137984630586e-05, "loss": 1.1027, "step": 1633 }, { "epoch": 0.63, "learning_rate": 1.624823057472534e-05, "loss": 1.102, "step": 1634 }, { "epoch": 0.63, "learning_rate": 1.624332069515491e-05, "loss": 1.142, "step": 1635 }, { "epoch": 0.63, "learning_rate": 1.6238408347859964e-05, "loss": 1.1503, "step": 1636 }, { "epoch": 0.63, "learning_rate": 1.6233493534782146e-05, "loss": 1.0893, "step": 1637 }, { "epoch": 0.63, "learning_rate": 1.622857625786408e-05, "loss": 1.1671, "step": 1638 }, { "epoch": 0.63, "learning_rate": 1.6223656519049357e-05, "loss": 1.109, "step": 1639 }, { "epoch": 0.63, "learning_rate": 1.621873432028255e-05, "loss": 1.1428, "step": 1640 }, { "epoch": 0.63, "learning_rate": 1.6213809663509197e-05, "loss": 1.1177, "step": 1641 }, { "epoch": 0.63, "learning_rate": 1.6208882550675805e-05, "loss": 1.0548, "step": 1642 }, { "epoch": 0.63, "learning_rate": 1.6203952983729863e-05, "loss": 1.2012, "step": 1643 }, { "epoch": 0.63, "learning_rate": 1.619902096461982e-05, "loss": 1.1296, "step": 1644 }, { "epoch": 0.63, "learning_rate": 1.61940864952951e-05, "loss": 1.1782, "step": 1645 }, { "epoch": 0.63, "learning_rate": 1.6189149577706087e-05, "loss": 1.1707, "step": 1646 }, { "epoch": 0.63, "learning_rate": 1.6184210213804147e-05, "loss": 1.0981, "step": 1647 }, { "epoch": 0.63, "learning_rate": 1.61792684055416e-05, "loss": 1.1649, "step": 1648 }, { "epoch": 0.63, "learning_rate": 1.6174324154871735e-05, "loss": 1.1348, "step": 1649 }, { "epoch": 0.63, "learning_rate": 1.6169377463748816e-05, "loss": 1.126, "step": 1650 }, { "epoch": 0.63, "learning_rate": 1.6164428334128054e-05, "loss": 1.0887, "step": 1651 }, { "epoch": 0.63, "learning_rate": 1.615947676796564e-05, "loss": 1.1514, "step": 1652 }, { "epoch": 0.63, "learning_rate": 1.6154522767218726e-05, "loss": 1.1301, "step": 1653 }, { "epoch": 0.64, "learning_rate": 1.614956633384541e-05, "loss": 1.142, "step": 1654 }, { "epoch": 0.64, "learning_rate": 1.6144607469804775e-05, "loss": 1.0931, "step": 1655 }, { "epoch": 0.64, "learning_rate": 1.613964617705684e-05, "loss": 1.1787, "step": 1656 }, { "epoch": 0.64, "learning_rate": 1.6134682457562615e-05, "loss": 1.1022, "step": 1657 }, { "epoch": 0.64, "learning_rate": 1.612971631328404e-05, "loss": 1.1258, "step": 1658 }, { "epoch": 0.64, "learning_rate": 1.6124747746184022e-05, "loss": 1.1314, "step": 1659 }, { "epoch": 0.64, "learning_rate": 1.6119776758226437e-05, "loss": 1.114, "step": 1660 }, { "epoch": 0.64, "learning_rate": 1.6114803351376106e-05, "loss": 1.1268, "step": 1661 }, { "epoch": 0.64, "learning_rate": 1.610982752759881e-05, "loss": 1.1549, "step": 1662 }, { "epoch": 0.64, "learning_rate": 1.6104849288861287e-05, "loss": 1.151, "step": 1663 }, { "epoch": 0.64, "learning_rate": 1.609986863713122e-05, "loss": 1.1514, "step": 1664 }, { "epoch": 0.64, "learning_rate": 1.6094885574377263e-05, "loss": 1.1344, "step": 1665 }, { "epoch": 0.64, "learning_rate": 1.6089900102569e-05, "loss": 1.1281, "step": 1666 }, { "epoch": 0.64, "learning_rate": 1.6084912223676995e-05, "loss": 1.128, "step": 1667 }, { "epoch": 0.64, "learning_rate": 1.6079921939672736e-05, "loss": 1.1414, "step": 1668 }, { "epoch": 0.64, "learning_rate": 1.6074929252528684e-05, "loss": 1.1425, "step": 1669 }, { "epoch": 0.64, "learning_rate": 1.6069934164218235e-05, "loss": 1.1075, "step": 1670 }, { "epoch": 0.64, "learning_rate": 1.606493667671574e-05, "loss": 1.096, "step": 1671 }, { "epoch": 0.64, "learning_rate": 1.60599367919965e-05, "loss": 1.1653, "step": 1672 }, { "epoch": 0.64, "learning_rate": 1.605493451203676e-05, "loss": 1.1538, "step": 1673 }, { "epoch": 0.64, "learning_rate": 1.604992983881371e-05, "loss": 1.0941, "step": 1674 }, { "epoch": 0.64, "learning_rate": 1.604492277430549e-05, "loss": 1.1597, "step": 1675 }, { "epoch": 0.64, "learning_rate": 1.6039913320491186e-05, "loss": 1.0691, "step": 1676 }, { "epoch": 0.64, "learning_rate": 1.6034901479350823e-05, "loss": 1.1512, "step": 1677 }, { "epoch": 0.64, "learning_rate": 1.602988725286538e-05, "loss": 1.0862, "step": 1678 }, { "epoch": 0.64, "learning_rate": 1.6024870643016762e-05, "loss": 1.1525, "step": 1679 }, { "epoch": 0.65, "learning_rate": 1.6019851651787833e-05, "loss": 1.0927, "step": 1680 }, { "epoch": 0.65, "learning_rate": 1.601483028116239e-05, "loss": 1.0788, "step": 1681 }, { "epoch": 0.65, "learning_rate": 1.6009806533125166e-05, "loss": 1.2008, "step": 1682 }, { "epoch": 0.65, "learning_rate": 1.6004780409661845e-05, "loss": 1.1299, "step": 1683 }, { "epoch": 0.65, "learning_rate": 1.5999751912759043e-05, "loss": 1.1687, "step": 1684 }, { "epoch": 0.65, "learning_rate": 1.599472104440432e-05, "loss": 1.124, "step": 1685 }, { "epoch": 0.65, "learning_rate": 1.5989687806586157e-05, "loss": 1.0882, "step": 1686 }, { "epoch": 0.65, "learning_rate": 1.5984652201293993e-05, "loss": 1.1129, "step": 1687 }, { "epoch": 0.65, "learning_rate": 1.5979614230518187e-05, "loss": 1.1682, "step": 1688 }, { "epoch": 0.65, "learning_rate": 1.5974573896250048e-05, "loss": 1.1452, "step": 1689 }, { "epoch": 0.65, "learning_rate": 1.59695312004818e-05, "loss": 1.1187, "step": 1690 }, { "epoch": 0.65, "learning_rate": 1.5964486145206612e-05, "loss": 1.1539, "step": 1691 }, { "epoch": 0.65, "learning_rate": 1.595943873241859e-05, "loss": 1.1355, "step": 1692 }, { "epoch": 0.65, "learning_rate": 1.595438896411276e-05, "loss": 1.1331, "step": 1693 }, { "epoch": 0.65, "learning_rate": 1.594933684228509e-05, "loss": 1.1755, "step": 1694 }, { "epoch": 0.65, "learning_rate": 1.5944282368932468e-05, "loss": 1.0534, "step": 1695 }, { "epoch": 0.65, "learning_rate": 1.5939225546052722e-05, "loss": 1.1796, "step": 1696 }, { "epoch": 0.65, "learning_rate": 1.59341663756446e-05, "loss": 1.0855, "step": 1697 }, { "epoch": 0.65, "learning_rate": 1.5929104859707783e-05, "loss": 1.0894, "step": 1698 }, { "epoch": 0.65, "learning_rate": 1.5924041000242872e-05, "loss": 1.109, "step": 1699 }, { "epoch": 0.65, "learning_rate": 1.591897479925141e-05, "loss": 1.1085, "step": 1700 }, { "epoch": 0.65, "learning_rate": 1.5913906258735846e-05, "loss": 1.1513, "step": 1701 }, { "epoch": 0.65, "learning_rate": 1.5908835380699566e-05, "loss": 1.1397, "step": 1702 }, { "epoch": 0.65, "learning_rate": 1.5903762167146876e-05, "loss": 1.1174, "step": 1703 }, { "epoch": 0.65, "learning_rate": 1.5898686620083008e-05, "loss": 1.1441, "step": 1704 }, { "epoch": 0.65, "learning_rate": 1.5893608741514114e-05, "loss": 1.086, "step": 1705 }, { "epoch": 0.66, "learning_rate": 1.5888528533447264e-05, "loss": 1.1428, "step": 1706 }, { "epoch": 0.66, "learning_rate": 1.5883445997890458e-05, "loss": 1.1745, "step": 1707 }, { "epoch": 0.66, "learning_rate": 1.5878361136852606e-05, "loss": 1.1087, "step": 1708 }, { "epoch": 0.66, "learning_rate": 1.5873273952343543e-05, "loss": 1.1193, "step": 1709 }, { "epoch": 0.66, "learning_rate": 1.586818444637402e-05, "loss": 1.1392, "step": 1710 }, { "epoch": 0.66, "learning_rate": 1.5863092620955712e-05, "loss": 1.0491, "step": 1711 }, { "epoch": 0.66, "learning_rate": 1.5857998478101202e-05, "loss": 1.0988, "step": 1712 }, { "epoch": 0.66, "learning_rate": 1.585290201982399e-05, "loss": 1.0937, "step": 1713 }, { "epoch": 0.66, "learning_rate": 1.5847803248138494e-05, "loss": 1.0731, "step": 1714 }, { "epoch": 0.66, "learning_rate": 1.5842702165060052e-05, "loss": 1.1235, "step": 1715 }, { "epoch": 0.66, "learning_rate": 1.5837598772604904e-05, "loss": 1.1533, "step": 1716 }, { "epoch": 0.66, "learning_rate": 1.583249307279021e-05, "loss": 1.1201, "step": 1717 }, { "epoch": 0.66, "learning_rate": 1.582738506763404e-05, "loss": 1.1111, "step": 1718 }, { "epoch": 0.66, "learning_rate": 1.582227475915537e-05, "loss": 1.1673, "step": 1719 }, { "epoch": 0.66, "learning_rate": 1.5817162149374102e-05, "loss": 1.1128, "step": 1720 }, { "epoch": 0.66, "learning_rate": 1.5812047240311032e-05, "loss": 1.2026, "step": 1721 }, { "epoch": 0.66, "learning_rate": 1.5806930033987873e-05, "loss": 1.1048, "step": 1722 }, { "epoch": 0.66, "learning_rate": 1.580181053242724e-05, "loss": 1.143, "step": 1723 }, { "epoch": 0.66, "learning_rate": 1.5796688737652663e-05, "loss": 1.1591, "step": 1724 }, { "epoch": 0.66, "learning_rate": 1.5791564651688574e-05, "loss": 1.1165, "step": 1725 }, { "epoch": 0.66, "learning_rate": 1.5786438276560305e-05, "loss": 1.1097, "step": 1726 }, { "epoch": 0.66, "learning_rate": 1.5781309614294103e-05, "loss": 1.1039, "step": 1727 }, { "epoch": 0.66, "learning_rate": 1.577617866691711e-05, "loss": 1.0748, "step": 1728 }, { "epoch": 0.66, "learning_rate": 1.577104543645738e-05, "loss": 1.1344, "step": 1729 }, { "epoch": 0.66, "learning_rate": 1.5765909924943867e-05, "loss": 1.1205, "step": 1730 }, { "epoch": 0.66, "learning_rate": 1.5760772134406422e-05, "loss": 1.082, "step": 1731 }, { "epoch": 0.67, "learning_rate": 1.5755632066875795e-05, "loss": 1.0431, "step": 1732 }, { "epoch": 0.67, "learning_rate": 1.5750489724383648e-05, "loss": 1.0953, "step": 1733 }, { "epoch": 0.67, "learning_rate": 1.5745345108962535e-05, "loss": 1.1497, "step": 1734 }, { "epoch": 0.67, "learning_rate": 1.5740198222645897e-05, "loss": 1.1407, "step": 1735 }, { "epoch": 0.67, "learning_rate": 1.5735049067468097e-05, "loss": 1.1698, "step": 1736 }, { "epoch": 0.67, "learning_rate": 1.5729897645464376e-05, "loss": 1.1631, "step": 1737 }, { "epoch": 0.67, "learning_rate": 1.5724743958670873e-05, "loss": 1.1457, "step": 1738 }, { "epoch": 0.67, "learning_rate": 1.571958800912463e-05, "loss": 1.1515, "step": 1739 }, { "epoch": 0.67, "learning_rate": 1.571442979886358e-05, "loss": 1.059, "step": 1740 }, { "epoch": 0.67, "learning_rate": 1.5709269329926544e-05, "loss": 1.2055, "step": 1741 }, { "epoch": 0.67, "learning_rate": 1.5704106604353245e-05, "loss": 1.1595, "step": 1742 }, { "epoch": 0.67, "learning_rate": 1.569894162418429e-05, "loss": 1.1106, "step": 1743 }, { "epoch": 0.67, "learning_rate": 1.569377439146118e-05, "loss": 1.1024, "step": 1744 }, { "epoch": 0.67, "learning_rate": 1.568860490822631e-05, "loss": 1.1621, "step": 1745 }, { "epoch": 0.67, "learning_rate": 1.568343317652296e-05, "loss": 1.1967, "step": 1746 }, { "epoch": 0.67, "learning_rate": 1.56782591983953e-05, "loss": 1.1139, "step": 1747 }, { "epoch": 0.67, "learning_rate": 1.5673082975888388e-05, "loss": 1.192, "step": 1748 }, { "epoch": 0.67, "learning_rate": 1.5667904511048168e-05, "loss": 1.1558, "step": 1749 }, { "epoch": 0.67, "learning_rate": 1.5662723805921473e-05, "loss": 1.1785, "step": 1750 }, { "epoch": 0.67, "learning_rate": 1.565754086255602e-05, "loss": 1.0859, "step": 1751 }, { "epoch": 0.67, "learning_rate": 1.5652355683000405e-05, "loss": 1.1405, "step": 1752 }, { "epoch": 0.67, "learning_rate": 1.564716826930412e-05, "loss": 1.1626, "step": 1753 }, { "epoch": 0.67, "learning_rate": 1.5641978623517534e-05, "loss": 1.1081, "step": 1754 }, { "epoch": 0.67, "learning_rate": 1.563678674769189e-05, "loss": 1.145, "step": 1755 }, { "epoch": 0.67, "learning_rate": 1.563159264387933e-05, "loss": 1.0925, "step": 1756 }, { "epoch": 0.67, "learning_rate": 1.562639631413286e-05, "loss": 1.1186, "step": 1757 }, { "epoch": 0.68, "learning_rate": 1.5621197760506377e-05, "loss": 1.115, "step": 1758 }, { "epoch": 0.68, "learning_rate": 1.561599698505465e-05, "loss": 1.163, "step": 1759 }, { "epoch": 0.68, "learning_rate": 1.5610793989833328e-05, "loss": 1.1138, "step": 1760 }, { "epoch": 0.68, "learning_rate": 1.560558877689894e-05, "loss": 1.0886, "step": 1761 }, { "epoch": 0.68, "learning_rate": 1.5600381348308888e-05, "loss": 1.167, "step": 1762 }, { "epoch": 0.68, "learning_rate": 1.5595171706121454e-05, "loss": 1.1389, "step": 1763 }, { "epoch": 0.68, "learning_rate": 1.558995985239579e-05, "loss": 1.1482, "step": 1764 }, { "epoch": 0.68, "learning_rate": 1.5584745789191927e-05, "loss": 1.1133, "step": 1765 }, { "epoch": 0.68, "learning_rate": 1.5579529518570767e-05, "loss": 1.1457, "step": 1766 }, { "epoch": 0.68, "learning_rate": 1.557431104259408e-05, "loss": 1.0789, "step": 1767 }, { "epoch": 0.68, "learning_rate": 1.5569090363324516e-05, "loss": 1.089, "step": 1768 }, { "epoch": 0.68, "learning_rate": 1.5563867482825593e-05, "loss": 1.1305, "step": 1769 }, { "epoch": 0.68, "learning_rate": 1.5558642403161695e-05, "loss": 1.1192, "step": 1770 }, { "epoch": 0.68, "learning_rate": 1.555341512639808e-05, "loss": 1.1095, "step": 1771 }, { "epoch": 0.68, "learning_rate": 1.5548185654600872e-05, "loss": 1.1266, "step": 1772 }, { "epoch": 0.68, "learning_rate": 1.5542953989837062e-05, "loss": 1.1057, "step": 1773 }, { "epoch": 0.68, "learning_rate": 1.5537720134174515e-05, "loss": 1.1234, "step": 1774 }, { "epoch": 0.68, "learning_rate": 1.5532484089681948e-05, "loss": 1.1421, "step": 1775 }, { "epoch": 0.68, "learning_rate": 1.5527245858428958e-05, "loss": 1.1193, "step": 1776 }, { "epoch": 0.68, "learning_rate": 1.552200544248599e-05, "loss": 1.0929, "step": 1777 }, { "epoch": 0.68, "learning_rate": 1.5516762843924373e-05, "loss": 1.1902, "step": 1778 }, { "epoch": 0.68, "learning_rate": 1.551151806481628e-05, "loss": 1.1374, "step": 1779 }, { "epoch": 0.68, "learning_rate": 1.5506271107234762e-05, "loss": 1.1749, "step": 1780 }, { "epoch": 0.68, "learning_rate": 1.5501021973253715e-05, "loss": 1.1005, "step": 1781 }, { "epoch": 0.68, "learning_rate": 1.5495770664947904e-05, "loss": 1.1307, "step": 1782 }, { "epoch": 0.68, "learning_rate": 1.5490517184392957e-05, "loss": 1.0993, "step": 1783 }, { "epoch": 0.69, "learning_rate": 1.548526153366535e-05, "loss": 1.0704, "step": 1784 }, { "epoch": 0.69, "learning_rate": 1.5480003714842427e-05, "loss": 1.1443, "step": 1785 }, { "epoch": 0.69, "learning_rate": 1.5474743730002383e-05, "loss": 1.1067, "step": 1786 }, { "epoch": 0.69, "learning_rate": 1.5469481581224274e-05, "loss": 1.1311, "step": 1787 }, { "epoch": 0.69, "learning_rate": 1.5464217270588e-05, "loss": 1.1107, "step": 1788 }, { "epoch": 0.69, "learning_rate": 1.545895080017433e-05, "loss": 1.0691, "step": 1789 }, { "epoch": 0.69, "learning_rate": 1.545368217206488e-05, "loss": 1.0702, "step": 1790 }, { "epoch": 0.69, "learning_rate": 1.544841138834212e-05, "loss": 1.123, "step": 1791 }, { "epoch": 0.69, "learning_rate": 1.5443138451089366e-05, "loss": 1.1346, "step": 1792 }, { "epoch": 0.69, "learning_rate": 1.5437863362390794e-05, "loss": 1.1303, "step": 1793 }, { "epoch": 0.69, "learning_rate": 1.5432586124331427e-05, "loss": 1.132, "step": 1794 }, { "epoch": 0.69, "learning_rate": 1.5427306738997138e-05, "loss": 1.0676, "step": 1795 }, { "epoch": 0.69, "learning_rate": 1.5422025208474643e-05, "loss": 1.1031, "step": 1796 }, { "epoch": 0.69, "learning_rate": 1.5416741534851524e-05, "loss": 1.1679, "step": 1797 }, { "epoch": 0.69, "learning_rate": 1.541145572021618e-05, "loss": 1.1066, "step": 1798 }, { "epoch": 0.69, "learning_rate": 1.5406167766657884e-05, "loss": 1.1661, "step": 1799 }, { "epoch": 0.69, "learning_rate": 1.5400877676266742e-05, "loss": 1.0855, "step": 1800 }, { "epoch": 0.69, "learning_rate": 1.5395585451133703e-05, "loss": 1.1565, "step": 1801 }, { "epoch": 0.69, "learning_rate": 1.5390291093350565e-05, "loss": 1.1008, "step": 1802 }, { "epoch": 0.69, "learning_rate": 1.538499460500997e-05, "loss": 1.105, "step": 1803 }, { "epoch": 0.69, "learning_rate": 1.537969598820539e-05, "loss": 1.1733, "step": 1804 }, { "epoch": 0.69, "learning_rate": 1.537439524503116e-05, "loss": 1.1335, "step": 1805 }, { "epoch": 0.69, "learning_rate": 1.5369092377582436e-05, "loss": 1.1241, "step": 1806 }, { "epoch": 0.69, "learning_rate": 1.5363787387955216e-05, "loss": 1.1289, "step": 1807 }, { "epoch": 0.69, "learning_rate": 1.5358480278246353e-05, "loss": 1.1098, "step": 1808 }, { "epoch": 0.69, "learning_rate": 1.535317105055352e-05, "loss": 1.1494, "step": 1809 }, { "epoch": 0.7, "learning_rate": 1.5347859706975227e-05, "loss": 1.2182, "step": 1810 }, { "epoch": 0.7, "learning_rate": 1.534254624961084e-05, "loss": 1.1481, "step": 1811 }, { "epoch": 0.7, "learning_rate": 1.533723068056054e-05, "loss": 1.1679, "step": 1812 }, { "epoch": 0.7, "learning_rate": 1.5331913001925343e-05, "loss": 1.1034, "step": 1813 }, { "epoch": 0.7, "learning_rate": 1.532659321580712e-05, "loss": 1.1451, "step": 1814 }, { "epoch": 0.7, "learning_rate": 1.5321271324308552e-05, "loss": 1.1314, "step": 1815 }, { "epoch": 0.7, "learning_rate": 1.5315947329533163e-05, "loss": 1.142, "step": 1816 }, { "epoch": 0.7, "learning_rate": 1.5310621233585306e-05, "loss": 1.1668, "step": 1817 }, { "epoch": 0.7, "learning_rate": 1.5305293038570165e-05, "loss": 1.1161, "step": 1818 }, { "epoch": 0.7, "learning_rate": 1.5299962746593754e-05, "loss": 1.1827, "step": 1819 }, { "epoch": 0.7, "learning_rate": 1.5294630359762912e-05, "loss": 1.1265, "step": 1820 }, { "epoch": 0.7, "learning_rate": 1.5289295880185318e-05, "loss": 1.142, "step": 1821 }, { "epoch": 0.7, "learning_rate": 1.528395930996946e-05, "loss": 1.1118, "step": 1822 }, { "epoch": 0.7, "learning_rate": 1.5278620651224664e-05, "loss": 1.1367, "step": 1823 }, { "epoch": 0.7, "learning_rate": 1.5273279906061082e-05, "loss": 1.1068, "step": 1824 }, { "epoch": 0.7, "learning_rate": 1.5267937076589686e-05, "loss": 1.097, "step": 1825 }, { "epoch": 0.7, "learning_rate": 1.5262592164922275e-05, "loss": 1.1512, "step": 1826 }, { "epoch": 0.7, "learning_rate": 1.5257245173171471e-05, "loss": 1.0805, "step": 1827 }, { "epoch": 0.7, "learning_rate": 1.5251896103450717e-05, "loss": 1.1127, "step": 1828 }, { "epoch": 0.7, "learning_rate": 1.5246544957874272e-05, "loss": 1.1257, "step": 1829 }, { "epoch": 0.7, "learning_rate": 1.5241191738557227e-05, "loss": 1.1445, "step": 1830 }, { "epoch": 0.7, "learning_rate": 1.5235836447615482e-05, "loss": 1.1157, "step": 1831 }, { "epoch": 0.7, "learning_rate": 1.5230479087165765e-05, "loss": 1.1113, "step": 1832 }, { "epoch": 0.7, "learning_rate": 1.5225119659325618e-05, "loss": 1.1732, "step": 1833 }, { "epoch": 0.7, "learning_rate": 1.521975816621339e-05, "loss": 1.0896, "step": 1834 }, { "epoch": 0.7, "learning_rate": 1.5214394609948264e-05, "loss": 1.1163, "step": 1835 }, { "epoch": 0.71, "learning_rate": 1.520902899265023e-05, "loss": 1.1072, "step": 1836 }, { "epoch": 0.71, "learning_rate": 1.5203661316440093e-05, "loss": 1.0643, "step": 1837 }, { "epoch": 0.71, "learning_rate": 1.5198291583439469e-05, "loss": 1.1058, "step": 1838 }, { "epoch": 0.71, "learning_rate": 1.5192919795770791e-05, "loss": 1.1216, "step": 1839 }, { "epoch": 0.71, "learning_rate": 1.5187545955557302e-05, "loss": 1.1515, "step": 1840 }, { "epoch": 0.71, "learning_rate": 1.518217006492306e-05, "loss": 1.106, "step": 1841 }, { "epoch": 0.71, "learning_rate": 1.5176792125992925e-05, "loss": 1.1114, "step": 1842 }, { "epoch": 0.71, "learning_rate": 1.5171412140892577e-05, "loss": 1.1403, "step": 1843 }, { "epoch": 0.71, "learning_rate": 1.51660301117485e-05, "loss": 1.1464, "step": 1844 }, { "epoch": 0.71, "learning_rate": 1.5160646040687988e-05, "loss": 1.1044, "step": 1845 }, { "epoch": 0.71, "learning_rate": 1.5155259929839129e-05, "loss": 1.1456, "step": 1846 }, { "epoch": 0.71, "learning_rate": 1.5149871781330839e-05, "loss": 1.1617, "step": 1847 }, { "epoch": 0.71, "learning_rate": 1.5144481597292823e-05, "loss": 1.15, "step": 1848 }, { "epoch": 0.71, "learning_rate": 1.5139089379855603e-05, "loss": 1.1233, "step": 1849 }, { "epoch": 0.71, "learning_rate": 1.5133695131150487e-05, "loss": 1.0926, "step": 1850 }, { "epoch": 0.71, "learning_rate": 1.5128298853309607e-05, "loss": 1.1589, "step": 1851 }, { "epoch": 0.71, "learning_rate": 1.5122900548465882e-05, "loss": 1.1501, "step": 1852 }, { "epoch": 0.71, "learning_rate": 1.511750021875304e-05, "loss": 1.1325, "step": 1853 }, { "epoch": 0.71, "learning_rate": 1.5112097866305599e-05, "loss": 1.171, "step": 1854 }, { "epoch": 0.71, "learning_rate": 1.5106693493258893e-05, "loss": 1.1353, "step": 1855 }, { "epoch": 0.71, "learning_rate": 1.5101287101749038e-05, "loss": 1.0816, "step": 1856 }, { "epoch": 0.71, "learning_rate": 1.5095878693912962e-05, "loss": 1.0924, "step": 1857 }, { "epoch": 0.71, "learning_rate": 1.5090468271888379e-05, "loss": 1.1975, "step": 1858 }, { "epoch": 0.71, "learning_rate": 1.5085055837813803e-05, "loss": 1.1158, "step": 1859 }, { "epoch": 0.71, "learning_rate": 1.5079641393828544e-05, "loss": 1.0813, "step": 1860 }, { "epoch": 0.71, "learning_rate": 1.5074224942072706e-05, "loss": 1.124, "step": 1861 }, { "epoch": 0.72, "learning_rate": 1.5068806484687191e-05, "loss": 1.136, "step": 1862 }, { "epoch": 0.72, "learning_rate": 1.5063386023813685e-05, "loss": 1.1106, "step": 1863 }, { "epoch": 0.72, "learning_rate": 1.5057963561594668e-05, "loss": 1.1241, "step": 1864 }, { "epoch": 0.72, "learning_rate": 1.505253910017342e-05, "loss": 1.1404, "step": 1865 }, { "epoch": 0.72, "learning_rate": 1.5047112641694002e-05, "loss": 1.1441, "step": 1866 }, { "epoch": 0.72, "learning_rate": 1.5041684188301262e-05, "loss": 1.1526, "step": 1867 }, { "epoch": 0.72, "learning_rate": 1.503625374214085e-05, "loss": 1.095, "step": 1868 }, { "epoch": 0.72, "learning_rate": 1.5030821305359192e-05, "loss": 1.1206, "step": 1869 }, { "epoch": 0.72, "learning_rate": 1.5025386880103496e-05, "loss": 1.164, "step": 1870 }, { "epoch": 0.72, "learning_rate": 1.5019950468521775e-05, "loss": 1.1619, "step": 1871 }, { "epoch": 0.72, "learning_rate": 1.501451207276281e-05, "loss": 1.0673, "step": 1872 }, { "epoch": 0.72, "learning_rate": 1.5009071694976173e-05, "loss": 1.087, "step": 1873 }, { "epoch": 0.72, "learning_rate": 1.5003629337312222e-05, "loss": 1.1532, "step": 1874 }, { "epoch": 0.72, "learning_rate": 1.499818500192209e-05, "loss": 1.0868, "step": 1875 }, { "epoch": 0.72, "learning_rate": 1.4992738690957696e-05, "loss": 1.1018, "step": 1876 }, { "epoch": 0.72, "learning_rate": 1.4987290406571741e-05, "loss": 1.1697, "step": 1877 }, { "epoch": 0.72, "learning_rate": 1.4981840150917707e-05, "loss": 1.1429, "step": 1878 }, { "epoch": 0.72, "learning_rate": 1.4976387926149852e-05, "loss": 1.1218, "step": 1879 }, { "epoch": 0.72, "learning_rate": 1.497093373442321e-05, "loss": 1.1259, "step": 1880 }, { "epoch": 0.72, "learning_rate": 1.49654775778936e-05, "loss": 1.0789, "step": 1881 }, { "epoch": 0.72, "learning_rate": 1.4960019458717607e-05, "loss": 1.1296, "step": 1882 }, { "epoch": 0.72, "learning_rate": 1.4954559379052606e-05, "loss": 1.0972, "step": 1883 }, { "epoch": 0.72, "learning_rate": 1.4949097341056736e-05, "loss": 1.1138, "step": 1884 }, { "epoch": 0.72, "learning_rate": 1.4943633346888909e-05, "loss": 1.0961, "step": 1885 }, { "epoch": 0.72, "learning_rate": 1.4938167398708817e-05, "loss": 1.0586, "step": 1886 }, { "epoch": 0.72, "learning_rate": 1.4932699498676927e-05, "loss": 1.1348, "step": 1887 }, { "epoch": 0.73, "learning_rate": 1.4927229648954463e-05, "loss": 1.089, "step": 1888 }, { "epoch": 0.73, "learning_rate": 1.4921757851703432e-05, "loss": 1.0247, "step": 1889 }, { "epoch": 0.73, "learning_rate": 1.491628410908661e-05, "loss": 1.1053, "step": 1890 }, { "epoch": 0.73, "learning_rate": 1.4910808423267537e-05, "loss": 1.1726, "step": 1891 }, { "epoch": 0.73, "learning_rate": 1.4905330796410524e-05, "loss": 1.1414, "step": 1892 }, { "epoch": 0.73, "learning_rate": 1.4899851230680652e-05, "loss": 1.0555, "step": 1893 }, { "epoch": 0.73, "learning_rate": 1.4894369728243759e-05, "loss": 1.1373, "step": 1894 }, { "epoch": 0.73, "learning_rate": 1.4888886291266454e-05, "loss": 1.1214, "step": 1895 }, { "epoch": 0.73, "learning_rate": 1.4883400921916119e-05, "loss": 1.1352, "step": 1896 }, { "epoch": 0.73, "learning_rate": 1.4877913622360884e-05, "loss": 1.1127, "step": 1897 }, { "epoch": 0.73, "learning_rate": 1.4872424394769655e-05, "loss": 1.0621, "step": 1898 }, { "epoch": 0.73, "learning_rate": 1.4866933241312094e-05, "loss": 1.1697, "step": 1899 }, { "epoch": 0.73, "learning_rate": 1.4861440164158623e-05, "loss": 1.1069, "step": 1900 }, { "epoch": 0.73, "learning_rate": 1.4855945165480428e-05, "loss": 1.1154, "step": 1901 }, { "epoch": 0.73, "learning_rate": 1.4850448247449451e-05, "loss": 1.0975, "step": 1902 }, { "epoch": 0.73, "learning_rate": 1.4844949412238396e-05, "loss": 1.1746, "step": 1903 }, { "epoch": 0.73, "learning_rate": 1.4839448662020728e-05, "loss": 1.1438, "step": 1904 }, { "epoch": 0.73, "learning_rate": 1.4833945998970658e-05, "loss": 1.1008, "step": 1905 }, { "epoch": 0.73, "learning_rate": 1.482844142526316e-05, "loss": 1.0959, "step": 1906 }, { "epoch": 0.73, "learning_rate": 1.4822934943073966e-05, "loss": 1.1164, "step": 1907 }, { "epoch": 0.73, "learning_rate": 1.4817426554579555e-05, "loss": 1.1417, "step": 1908 }, { "epoch": 0.73, "learning_rate": 1.4811916261957168e-05, "loss": 1.146, "step": 1909 }, { "epoch": 0.73, "learning_rate": 1.4806404067384792e-05, "loss": 1.0857, "step": 1910 }, { "epoch": 0.73, "learning_rate": 1.4800889973041164e-05, "loss": 1.1266, "step": 1911 }, { "epoch": 0.73, "learning_rate": 1.4795373981105783e-05, "loss": 1.0548, "step": 1912 }, { "epoch": 0.73, "learning_rate": 1.4789856093758883e-05, "loss": 1.0776, "step": 1913 }, { "epoch": 0.74, "learning_rate": 1.4784336313181462e-05, "loss": 1.1658, "step": 1914 }, { "epoch": 0.74, "learning_rate": 1.4778814641555255e-05, "loss": 1.1218, "step": 1915 }, { "epoch": 0.74, "learning_rate": 1.477329108106275e-05, "loss": 1.0962, "step": 1916 }, { "epoch": 0.74, "learning_rate": 1.4767765633887181e-05, "loss": 1.0967, "step": 1917 }, { "epoch": 0.74, "learning_rate": 1.4762238302212526e-05, "loss": 1.1061, "step": 1918 }, { "epoch": 0.74, "learning_rate": 1.4756709088223508e-05, "loss": 1.093, "step": 1919 }, { "epoch": 0.74, "learning_rate": 1.4751177994105596e-05, "loss": 1.1185, "step": 1920 }, { "epoch": 0.74, "learning_rate": 1.4745645022045003e-05, "loss": 1.1007, "step": 1921 }, { "epoch": 0.74, "learning_rate": 1.4740110174228677e-05, "loss": 1.1334, "step": 1922 }, { "epoch": 0.74, "learning_rate": 1.4734573452844316e-05, "loss": 1.0637, "step": 1923 }, { "epoch": 0.74, "learning_rate": 1.4729034860080359e-05, "loss": 1.0634, "step": 1924 }, { "epoch": 0.74, "learning_rate": 1.4723494398125974e-05, "loss": 1.1294, "step": 1925 }, { "epoch": 0.74, "learning_rate": 1.4717952069171078e-05, "loss": 1.0951, "step": 1926 }, { "epoch": 0.74, "learning_rate": 1.4712407875406322e-05, "loss": 1.1018, "step": 1927 }, { "epoch": 0.74, "learning_rate": 1.4706861819023099e-05, "loss": 1.1244, "step": 1928 }, { "epoch": 0.74, "learning_rate": 1.470131390221353e-05, "loss": 1.135, "step": 1929 }, { "epoch": 0.74, "learning_rate": 1.4695764127170473e-05, "loss": 1.127, "step": 1930 }, { "epoch": 0.74, "learning_rate": 1.4690212496087526e-05, "loss": 1.1182, "step": 1931 }, { "epoch": 0.74, "learning_rate": 1.4684659011159017e-05, "loss": 1.1792, "step": 1932 }, { "epoch": 0.74, "learning_rate": 1.4679103674580009e-05, "loss": 1.0551, "step": 1933 }, { "epoch": 0.74, "learning_rate": 1.4673546488546292e-05, "loss": 1.1806, "step": 1934 }, { "epoch": 0.74, "learning_rate": 1.4667987455254392e-05, "loss": 1.0983, "step": 1935 }, { "epoch": 0.74, "learning_rate": 1.466242657690156e-05, "loss": 1.1168, "step": 1936 }, { "epoch": 0.74, "learning_rate": 1.4656863855685784e-05, "loss": 1.1584, "step": 1937 }, { "epoch": 0.74, "learning_rate": 1.4651299293805774e-05, "loss": 1.0957, "step": 1938 }, { "epoch": 0.74, "learning_rate": 1.4645732893460966e-05, "loss": 1.0973, "step": 1939 }, { "epoch": 0.75, "learning_rate": 1.4640164656851532e-05, "loss": 1.101, "step": 1940 }, { "epoch": 0.75, "learning_rate": 1.4634594586178366e-05, "loss": 1.1546, "step": 1941 }, { "epoch": 0.75, "learning_rate": 1.4629022683643075e-05, "loss": 1.1273, "step": 1942 }, { "epoch": 0.75, "learning_rate": 1.4623448951448006e-05, "loss": 1.1192, "step": 1943 }, { "epoch": 0.75, "learning_rate": 1.461787339179622e-05, "loss": 1.1415, "step": 1944 }, { "epoch": 0.75, "learning_rate": 1.4612296006891509e-05, "loss": 1.1263, "step": 1945 }, { "epoch": 0.75, "learning_rate": 1.4606716798938377e-05, "loss": 1.0612, "step": 1946 }, { "epoch": 0.75, "learning_rate": 1.4601135770142051e-05, "loss": 1.1752, "step": 1947 }, { "epoch": 0.75, "learning_rate": 1.459555292270848e-05, "loss": 1.1468, "step": 1948 }, { "epoch": 0.75, "learning_rate": 1.4589968258844334e-05, "loss": 1.137, "step": 1949 }, { "epoch": 0.75, "learning_rate": 1.4584381780756997e-05, "loss": 1.1374, "step": 1950 }, { "epoch": 0.75, "learning_rate": 1.4578793490654568e-05, "loss": 1.1304, "step": 1951 }, { "epoch": 0.75, "learning_rate": 1.4573203390745866e-05, "loss": 1.0995, "step": 1952 }, { "epoch": 0.75, "learning_rate": 1.456761148324043e-05, "loss": 1.137, "step": 1953 }, { "epoch": 0.75, "learning_rate": 1.45620177703485e-05, "loss": 1.1887, "step": 1954 }, { "epoch": 0.75, "learning_rate": 1.4556422254281044e-05, "loss": 1.1184, "step": 1955 }, { "epoch": 0.75, "learning_rate": 1.455082493724973e-05, "loss": 1.0645, "step": 1956 }, { "epoch": 0.75, "learning_rate": 1.4545225821466951e-05, "loss": 1.1376, "step": 1957 }, { "epoch": 0.75, "learning_rate": 1.4539624909145798e-05, "loss": 1.1059, "step": 1958 }, { "epoch": 0.75, "learning_rate": 1.4534022202500084e-05, "loss": 1.1177, "step": 1959 }, { "epoch": 0.75, "learning_rate": 1.4528417703744322e-05, "loss": 1.0624, "step": 1960 }, { "epoch": 0.75, "learning_rate": 1.4522811415093733e-05, "loss": 1.1078, "step": 1961 }, { "epoch": 0.75, "learning_rate": 1.4517203338764258e-05, "loss": 1.1572, "step": 1962 }, { "epoch": 0.75, "learning_rate": 1.4511593476972534e-05, "loss": 1.1635, "step": 1963 }, { "epoch": 0.75, "learning_rate": 1.45059818319359e-05, "loss": 1.1504, "step": 1964 }, { "epoch": 0.75, "learning_rate": 1.4500368405872411e-05, "loss": 1.1179, "step": 1965 }, { "epoch": 0.76, "learning_rate": 1.4494753201000815e-05, "loss": 1.0805, "step": 1966 }, { "epoch": 0.76, "learning_rate": 1.4489136219540574e-05, "loss": 1.1196, "step": 1967 }, { "epoch": 0.76, "learning_rate": 1.4483517463711842e-05, "loss": 1.1401, "step": 1968 }, { "epoch": 0.76, "learning_rate": 1.4477896935735484e-05, "loss": 1.1129, "step": 1969 }, { "epoch": 0.76, "learning_rate": 1.4472274637833057e-05, "loss": 1.1114, "step": 1970 }, { "epoch": 0.76, "learning_rate": 1.446665057222682e-05, "loss": 1.099, "step": 1971 }, { "epoch": 0.76, "learning_rate": 1.4461024741139738e-05, "loss": 1.1127, "step": 1972 }, { "epoch": 0.76, "learning_rate": 1.445539714679546e-05, "loss": 1.1782, "step": 1973 }, { "epoch": 0.76, "learning_rate": 1.4449767791418343e-05, "loss": 1.0845, "step": 1974 }, { "epoch": 0.76, "learning_rate": 1.4444136677233437e-05, "loss": 1.1186, "step": 1975 }, { "epoch": 0.76, "learning_rate": 1.443850380646649e-05, "loss": 1.1109, "step": 1976 }, { "epoch": 0.76, "learning_rate": 1.4432869181343935e-05, "loss": 1.1424, "step": 1977 }, { "epoch": 0.76, "learning_rate": 1.4427232804092906e-05, "loss": 1.1114, "step": 1978 }, { "epoch": 0.76, "learning_rate": 1.4421594676941228e-05, "loss": 1.1162, "step": 1979 }, { "epoch": 0.76, "learning_rate": 1.4415954802117424e-05, "loss": 1.1779, "step": 1980 }, { "epoch": 0.76, "learning_rate": 1.4410313181850693e-05, "loss": 1.0835, "step": 1981 }, { "epoch": 0.76, "learning_rate": 1.4404669818370933e-05, "loss": 1.1003, "step": 1982 }, { "epoch": 0.76, "learning_rate": 1.4399024713908736e-05, "loss": 1.1257, "step": 1983 }, { "epoch": 0.76, "learning_rate": 1.4393377870695372e-05, "loss": 1.1052, "step": 1984 }, { "epoch": 0.76, "learning_rate": 1.4387729290962806e-05, "loss": 1.0986, "step": 1985 }, { "epoch": 0.76, "learning_rate": 1.4382078976943685e-05, "loss": 1.1535, "step": 1986 }, { "epoch": 0.76, "learning_rate": 1.437642693087134e-05, "loss": 1.132, "step": 1987 }, { "epoch": 0.76, "learning_rate": 1.4370773154979793e-05, "loss": 1.1041, "step": 1988 }, { "epoch": 0.76, "learning_rate": 1.4365117651503749e-05, "loss": 1.0931, "step": 1989 }, { "epoch": 0.76, "learning_rate": 1.4359460422678584e-05, "loss": 1.1115, "step": 1990 }, { "epoch": 0.76, "learning_rate": 1.4353801470740369e-05, "loss": 1.1026, "step": 1991 }, { "epoch": 0.77, "learning_rate": 1.4348140797925851e-05, "loss": 1.1385, "step": 1992 }, { "epoch": 0.77, "learning_rate": 1.4342478406472465e-05, "loss": 1.0798, "step": 1993 }, { "epoch": 0.77, "learning_rate": 1.4336814298618307e-05, "loss": 1.1345, "step": 1994 }, { "epoch": 0.77, "learning_rate": 1.433114847660217e-05, "loss": 1.1111, "step": 1995 }, { "epoch": 0.77, "learning_rate": 1.4325480942663518e-05, "loss": 1.0864, "step": 1996 }, { "epoch": 0.77, "learning_rate": 1.431981169904249e-05, "loss": 1.1431, "step": 1997 }, { "epoch": 0.77, "learning_rate": 1.4314140747979896e-05, "loss": 1.0612, "step": 1998 }, { "epoch": 0.77, "learning_rate": 1.4308468091717233e-05, "loss": 1.1511, "step": 1999 }, { "epoch": 0.77, "learning_rate": 1.4302793732496665e-05, "loss": 1.147, "step": 2000 }, { "epoch": 0.77, "eval_loss": 1.127143383026123, "eval_runtime": 688.4915, "eval_samples_per_second": 4.94, "eval_steps_per_second": 0.309, "step": 2000 }, { "epoch": 0.77, "learning_rate": 1.429711767256103e-05, "loss": 1.1022, "step": 2001 }, { "epoch": 0.77, "learning_rate": 1.4291439914153837e-05, "loss": 1.0921, "step": 2002 }, { "epoch": 0.77, "learning_rate": 1.4285760459519264e-05, "loss": 1.1053, "step": 2003 }, { "epoch": 0.77, "learning_rate": 1.428007931090217e-05, "loss": 1.0994, "step": 2004 }, { "epoch": 0.77, "learning_rate": 1.4274396470548067e-05, "loss": 1.0602, "step": 2005 }, { "epoch": 0.77, "learning_rate": 1.4268711940703153e-05, "loss": 1.1486, "step": 2006 }, { "epoch": 0.77, "learning_rate": 1.4263025723614283e-05, "loss": 1.1064, "step": 2007 }, { "epoch": 0.77, "learning_rate": 1.4257337821528981e-05, "loss": 1.114, "step": 2008 }, { "epoch": 0.77, "learning_rate": 1.4251648236695436e-05, "loss": 1.1279, "step": 2009 }, { "epoch": 0.77, "learning_rate": 1.4245956971362512e-05, "loss": 1.1018, "step": 2010 }, { "epoch": 0.77, "learning_rate": 1.4240264027779716e-05, "loss": 1.1868, "step": 2011 }, { "epoch": 0.77, "learning_rate": 1.4234569408197244e-05, "loss": 1.1103, "step": 2012 }, { "epoch": 0.77, "learning_rate": 1.4228873114865933e-05, "loss": 1.1621, "step": 2013 }, { "epoch": 0.77, "learning_rate": 1.4223175150037297e-05, "loss": 1.0997, "step": 2014 }, { "epoch": 0.77, "learning_rate": 1.4217475515963497e-05, "loss": 1.1674, "step": 2015 }, { "epoch": 0.77, "learning_rate": 1.4211774214897367e-05, "loss": 1.13, "step": 2016 }, { "epoch": 0.77, "learning_rate": 1.420607124909239e-05, "loss": 1.084, "step": 2017 }, { "epoch": 0.78, "learning_rate": 1.4200366620802713e-05, "loss": 1.0808, "step": 2018 }, { "epoch": 0.78, "learning_rate": 1.4194660332283136e-05, "loss": 1.1352, "step": 2019 }, { "epoch": 0.78, "learning_rate": 1.418895238578912e-05, "loss": 1.0579, "step": 2020 }, { "epoch": 0.78, "learning_rate": 1.4183242783576771e-05, "loss": 1.124, "step": 2021 }, { "epoch": 0.78, "learning_rate": 1.417753152790287e-05, "loss": 1.1082, "step": 2022 }, { "epoch": 0.78, "learning_rate": 1.4171818621024826e-05, "loss": 1.1253, "step": 2023 }, { "epoch": 0.78, "learning_rate": 1.416610406520072e-05, "loss": 1.1353, "step": 2024 }, { "epoch": 0.78, "learning_rate": 1.4160387862689272e-05, "loss": 1.0772, "step": 2025 }, { "epoch": 0.78, "learning_rate": 1.4154670015749866e-05, "loss": 1.0834, "step": 2026 }, { "epoch": 0.78, "learning_rate": 1.4148950526642523e-05, "loss": 1.1051, "step": 2027 }, { "epoch": 0.78, "learning_rate": 1.414322939762792e-05, "loss": 1.1118, "step": 2028 }, { "epoch": 0.78, "learning_rate": 1.4137506630967381e-05, "loss": 1.0814, "step": 2029 }, { "epoch": 0.78, "learning_rate": 1.4131782228922877e-05, "loss": 1.0939, "step": 2030 }, { "epoch": 0.78, "learning_rate": 1.412605619375703e-05, "loss": 1.144, "step": 2031 }, { "epoch": 0.78, "learning_rate": 1.4120328527733096e-05, "loss": 1.155, "step": 2032 }, { "epoch": 0.78, "learning_rate": 1.4114599233114988e-05, "loss": 1.1609, "step": 2033 }, { "epoch": 0.78, "learning_rate": 1.4108868312167253e-05, "loss": 1.0778, "step": 2034 }, { "epoch": 0.78, "learning_rate": 1.4103135767155092e-05, "loss": 1.1142, "step": 2035 }, { "epoch": 0.78, "learning_rate": 1.4097401600344331e-05, "loss": 1.1548, "step": 2036 }, { "epoch": 0.78, "learning_rate": 1.4091665814001458e-05, "loss": 1.11, "step": 2037 }, { "epoch": 0.78, "learning_rate": 1.4085928410393584e-05, "loss": 1.1109, "step": 2038 }, { "epoch": 0.78, "learning_rate": 1.4080189391788465e-05, "loss": 1.0948, "step": 2039 }, { "epoch": 0.78, "learning_rate": 1.40744487604545e-05, "loss": 1.0844, "step": 2040 }, { "epoch": 0.78, "learning_rate": 1.406870651866072e-05, "loss": 1.0908, "step": 2041 }, { "epoch": 0.78, "learning_rate": 1.4062962668676789e-05, "loss": 1.1166, "step": 2042 }, { "epoch": 0.78, "learning_rate": 1.4057217212773018e-05, "loss": 1.0612, "step": 2043 }, { "epoch": 0.79, "learning_rate": 1.4051470153220346e-05, "loss": 1.1384, "step": 2044 }, { "epoch": 0.79, "learning_rate": 1.4045721492290342e-05, "loss": 1.118, "step": 2045 }, { "epoch": 0.79, "learning_rate": 1.4039971232255214e-05, "loss": 1.1016, "step": 2046 }, { "epoch": 0.79, "learning_rate": 1.4034219375387801e-05, "loss": 1.1169, "step": 2047 }, { "epoch": 0.79, "learning_rate": 1.4028465923961574e-05, "loss": 1.136, "step": 2048 }, { "epoch": 0.79, "learning_rate": 1.402271088025063e-05, "loss": 1.0782, "step": 2049 }, { "epoch": 0.79, "learning_rate": 1.4016954246529697e-05, "loss": 1.1532, "step": 2050 }, { "epoch": 0.79, "learning_rate": 1.4011196025074135e-05, "loss": 1.0832, "step": 2051 }, { "epoch": 0.79, "learning_rate": 1.4005436218159927e-05, "loss": 1.1388, "step": 2052 }, { "epoch": 0.79, "learning_rate": 1.3999674828063683e-05, "loss": 1.1233, "step": 2053 }, { "epoch": 0.79, "learning_rate": 1.3993911857062646e-05, "loss": 1.1069, "step": 2054 }, { "epoch": 0.79, "learning_rate": 1.3988147307434673e-05, "loss": 1.1439, "step": 2055 }, { "epoch": 0.79, "learning_rate": 1.3982381181458255e-05, "loss": 1.1523, "step": 2056 }, { "epoch": 0.79, "learning_rate": 1.3976613481412497e-05, "loss": 1.1277, "step": 2057 }, { "epoch": 0.79, "learning_rate": 1.3970844209577133e-05, "loss": 1.0814, "step": 2058 }, { "epoch": 0.79, "learning_rate": 1.3965073368232514e-05, "loss": 1.1437, "step": 2059 }, { "epoch": 0.79, "learning_rate": 1.3959300959659619e-05, "loss": 1.135, "step": 2060 }, { "epoch": 0.79, "learning_rate": 1.3953526986140032e-05, "loss": 1.0757, "step": 2061 }, { "epoch": 0.79, "learning_rate": 1.3947751449955967e-05, "loss": 1.1204, "step": 2062 }, { "epoch": 0.79, "learning_rate": 1.3941974353390255e-05, "loss": 1.0542, "step": 2063 }, { "epoch": 0.79, "learning_rate": 1.3936195698726338e-05, "loss": 1.089, "step": 2064 }, { "epoch": 0.79, "learning_rate": 1.3930415488248284e-05, "loss": 1.1255, "step": 2065 }, { "epoch": 0.79, "learning_rate": 1.3924633724240762e-05, "loss": 1.0951, "step": 2066 }, { "epoch": 0.79, "learning_rate": 1.3918850408989064e-05, "loss": 1.1157, "step": 2067 }, { "epoch": 0.79, "learning_rate": 1.3913065544779096e-05, "loss": 1.1204, "step": 2068 }, { "epoch": 0.79, "learning_rate": 1.3907279133897378e-05, "loss": 1.1354, "step": 2069 }, { "epoch": 0.8, "learning_rate": 1.3901491178631027e-05, "loss": 1.1268, "step": 2070 }, { "epoch": 0.8, "learning_rate": 1.3895701681267784e-05, "loss": 1.1342, "step": 2071 }, { "epoch": 0.8, "learning_rate": 1.3889910644096e-05, "loss": 1.0834, "step": 2072 }, { "epoch": 0.8, "learning_rate": 1.3884118069404627e-05, "loss": 1.1117, "step": 2073 }, { "epoch": 0.8, "learning_rate": 1.387832395948323e-05, "loss": 1.1457, "step": 2074 }, { "epoch": 0.8, "learning_rate": 1.3872528316621978e-05, "loss": 1.118, "step": 2075 }, { "epoch": 0.8, "learning_rate": 1.3866731143111646e-05, "loss": 1.1577, "step": 2076 }, { "epoch": 0.8, "learning_rate": 1.386093244124362e-05, "loss": 1.0851, "step": 2077 }, { "epoch": 0.8, "learning_rate": 1.3855132213309882e-05, "loss": 1.123, "step": 2078 }, { "epoch": 0.8, "learning_rate": 1.384933046160302e-05, "loss": 1.1039, "step": 2079 }, { "epoch": 0.8, "learning_rate": 1.3843527188416224e-05, "loss": 1.1559, "step": 2080 }, { "epoch": 0.8, "learning_rate": 1.383772239604329e-05, "loss": 1.1039, "step": 2081 }, { "epoch": 0.8, "learning_rate": 1.3831916086778604e-05, "loss": 1.1531, "step": 2082 }, { "epoch": 0.8, "learning_rate": 1.3826108262917167e-05, "loss": 1.1261, "step": 2083 }, { "epoch": 0.8, "learning_rate": 1.382029892675456e-05, "loss": 1.1285, "step": 2084 }, { "epoch": 0.8, "learning_rate": 1.381448808058698e-05, "loss": 1.1514, "step": 2085 }, { "epoch": 0.8, "learning_rate": 1.3808675726711208e-05, "loss": 1.1452, "step": 2086 }, { "epoch": 0.8, "learning_rate": 1.3802861867424625e-05, "loss": 1.1269, "step": 2087 }, { "epoch": 0.8, "learning_rate": 1.3797046505025207e-05, "loss": 1.1177, "step": 2088 }, { "epoch": 0.8, "learning_rate": 1.3791229641811525e-05, "loss": 1.1055, "step": 2089 }, { "epoch": 0.8, "learning_rate": 1.3785411280082747e-05, "loss": 1.1935, "step": 2090 }, { "epoch": 0.8, "learning_rate": 1.3779591422138623e-05, "loss": 1.1607, "step": 2091 }, { "epoch": 0.8, "learning_rate": 1.3773770070279502e-05, "loss": 1.1568, "step": 2092 }, { "epoch": 0.8, "learning_rate": 1.3767947226806322e-05, "loss": 1.0869, "step": 2093 }, { "epoch": 0.8, "learning_rate": 1.3762122894020613e-05, "loss": 1.0567, "step": 2094 }, { "epoch": 0.8, "learning_rate": 1.3756297074224487e-05, "loss": 1.1351, "step": 2095 }, { "epoch": 0.81, "learning_rate": 1.3750469769720652e-05, "loss": 1.0951, "step": 2096 }, { "epoch": 0.81, "learning_rate": 1.3744640982812396e-05, "loss": 1.1299, "step": 2097 }, { "epoch": 0.81, "learning_rate": 1.3738810715803595e-05, "loss": 1.1939, "step": 2098 }, { "epoch": 0.81, "learning_rate": 1.3732978970998711e-05, "loss": 1.1267, "step": 2099 }, { "epoch": 0.81, "learning_rate": 1.3727145750702791e-05, "loss": 1.0938, "step": 2100 }, { "epoch": 0.81, "learning_rate": 1.3721311057221462e-05, "loss": 1.1256, "step": 2101 }, { "epoch": 0.81, "learning_rate": 1.3715474892860936e-05, "loss": 1.1779, "step": 2102 }, { "epoch": 0.81, "learning_rate": 1.3709637259928007e-05, "loss": 1.1214, "step": 2103 }, { "epoch": 0.81, "learning_rate": 1.3703798160730047e-05, "loss": 1.1035, "step": 2104 }, { "epoch": 0.81, "learning_rate": 1.369795759757501e-05, "loss": 1.1459, "step": 2105 }, { "epoch": 0.81, "learning_rate": 1.3692115572771422e-05, "loss": 1.0865, "step": 2106 }, { "epoch": 0.81, "learning_rate": 1.3686272088628401e-05, "loss": 1.1048, "step": 2107 }, { "epoch": 0.81, "learning_rate": 1.3680427147455623e-05, "loss": 1.0981, "step": 2108 }, { "epoch": 0.81, "learning_rate": 1.3674580751563357e-05, "loss": 1.1519, "step": 2109 }, { "epoch": 0.81, "learning_rate": 1.3668732903262436e-05, "loss": 1.1313, "step": 2110 }, { "epoch": 0.81, "learning_rate": 1.366288360486427e-05, "loss": 1.0458, "step": 2111 }, { "epoch": 0.81, "learning_rate": 1.3657032858680847e-05, "loss": 1.1271, "step": 2112 }, { "epoch": 0.81, "learning_rate": 1.365118066702472e-05, "loss": 1.1211, "step": 2113 }, { "epoch": 0.81, "learning_rate": 1.3645327032209017e-05, "loss": 1.1028, "step": 2114 }, { "epoch": 0.81, "learning_rate": 1.3639471956547437e-05, "loss": 1.1728, "step": 2115 }, { "epoch": 0.81, "learning_rate": 1.3633615442354246e-05, "loss": 1.1281, "step": 2116 }, { "epoch": 0.81, "learning_rate": 1.362775749194428e-05, "loss": 1.1014, "step": 2117 }, { "epoch": 0.81, "learning_rate": 1.3621898107632945e-05, "loss": 1.0837, "step": 2118 }, { "epoch": 0.81, "learning_rate": 1.3616037291736208e-05, "loss": 1.1357, "step": 2119 }, { "epoch": 0.81, "learning_rate": 1.361017504657061e-05, "loss": 1.0948, "step": 2120 }, { "epoch": 0.81, "learning_rate": 1.360431137445325e-05, "loss": 1.1277, "step": 2121 }, { "epoch": 0.82, "learning_rate": 1.3598446277701789e-05, "loss": 1.1134, "step": 2122 }, { "epoch": 0.82, "learning_rate": 1.3592579758634461e-05, "loss": 1.1255, "step": 2123 }, { "epoch": 0.82, "learning_rate": 1.3586711819570055e-05, "loss": 1.1241, "step": 2124 }, { "epoch": 0.82, "learning_rate": 1.3580842462827919e-05, "loss": 1.1825, "step": 2125 }, { "epoch": 0.82, "learning_rate": 1.3574971690727972e-05, "loss": 1.141, "step": 2126 }, { "epoch": 0.82, "learning_rate": 1.3569099505590679e-05, "loss": 1.1165, "step": 2127 }, { "epoch": 0.82, "learning_rate": 1.3563225909737076e-05, "loss": 1.1107, "step": 2128 }, { "epoch": 0.82, "learning_rate": 1.3557350905488749e-05, "loss": 1.0982, "step": 2129 }, { "epoch": 0.82, "learning_rate": 1.3551474495167835e-05, "loss": 1.125, "step": 2130 }, { "epoch": 0.82, "learning_rate": 1.3545596681097047e-05, "loss": 1.1099, "step": 2131 }, { "epoch": 0.82, "learning_rate": 1.3539717465599632e-05, "loss": 1.1568, "step": 2132 }, { "epoch": 0.82, "learning_rate": 1.35338368509994e-05, "loss": 1.1198, "step": 2133 }, { "epoch": 0.82, "learning_rate": 1.3527954839620712e-05, "loss": 1.0997, "step": 2134 }, { "epoch": 0.82, "learning_rate": 1.3522071433788482e-05, "loss": 1.111, "step": 2135 }, { "epoch": 0.82, "learning_rate": 1.3516186635828176e-05, "loss": 1.1042, "step": 2136 }, { "epoch": 0.82, "learning_rate": 1.3510300448065814e-05, "loss": 1.1327, "step": 2137 }, { "epoch": 0.82, "learning_rate": 1.3504412872827952e-05, "loss": 1.0687, "step": 2138 }, { "epoch": 0.82, "learning_rate": 1.349852391244171e-05, "loss": 1.14, "step": 2139 }, { "epoch": 0.82, "learning_rate": 1.3492633569234745e-05, "loss": 1.1093, "step": 2140 }, { "epoch": 0.82, "learning_rate": 1.3486741845535267e-05, "loss": 1.1188, "step": 2141 }, { "epoch": 0.82, "learning_rate": 1.3480848743672027e-05, "loss": 1.1305, "step": 2142 }, { "epoch": 0.82, "learning_rate": 1.347495426597432e-05, "loss": 1.0791, "step": 2143 }, { "epoch": 0.82, "learning_rate": 1.3469058414771993e-05, "loss": 1.1278, "step": 2144 }, { "epoch": 0.82, "learning_rate": 1.3463161192395431e-05, "loss": 1.0969, "step": 2145 }, { "epoch": 0.82, "learning_rate": 1.3457262601175556e-05, "loss": 1.1241, "step": 2146 }, { "epoch": 0.82, "learning_rate": 1.3451362643443832e-05, "loss": 1.1117, "step": 2147 }, { "epoch": 0.83, "learning_rate": 1.3445461321532274e-05, "loss": 1.0366, "step": 2148 }, { "epoch": 0.83, "learning_rate": 1.3439558637773421e-05, "loss": 1.1287, "step": 2149 }, { "epoch": 0.83, "learning_rate": 1.343365459450037e-05, "loss": 1.13, "step": 2150 }, { "epoch": 0.83, "learning_rate": 1.3427749194046728e-05, "loss": 1.0887, "step": 2151 }, { "epoch": 0.83, "learning_rate": 1.3421842438746665e-05, "loss": 1.11, "step": 2152 }, { "epoch": 0.83, "learning_rate": 1.341593433093487e-05, "loss": 1.0986, "step": 2153 }, { "epoch": 0.83, "learning_rate": 1.3410024872946576e-05, "loss": 1.1222, "step": 2154 }, { "epoch": 0.83, "learning_rate": 1.3404114067117542e-05, "loss": 1.1058, "step": 2155 }, { "epoch": 0.83, "learning_rate": 1.3398201915784063e-05, "loss": 1.1557, "step": 2156 }, { "epoch": 0.83, "learning_rate": 1.3392288421282969e-05, "loss": 1.1014, "step": 2157 }, { "epoch": 0.83, "learning_rate": 1.3386373585951618e-05, "loss": 1.1238, "step": 2158 }, { "epoch": 0.83, "learning_rate": 1.3380457412127891e-05, "loss": 1.1389, "step": 2159 }, { "epoch": 0.83, "learning_rate": 1.3374539902150212e-05, "loss": 1.0802, "step": 2160 }, { "epoch": 0.83, "learning_rate": 1.3368621058357524e-05, "loss": 1.0762, "step": 2161 }, { "epoch": 0.83, "learning_rate": 1.3362700883089303e-05, "loss": 1.1168, "step": 2162 }, { "epoch": 0.83, "learning_rate": 1.335677937868554e-05, "loss": 1.076, "step": 2163 }, { "epoch": 0.83, "learning_rate": 1.3350856547486767e-05, "loss": 1.1088, "step": 2164 }, { "epoch": 0.83, "learning_rate": 1.3344932391834022e-05, "loss": 1.2226, "step": 2165 }, { "epoch": 0.83, "learning_rate": 1.333900691406889e-05, "loss": 1.1249, "step": 2166 }, { "epoch": 0.83, "learning_rate": 1.3333080116533456e-05, "loss": 1.168, "step": 2167 }, { "epoch": 0.83, "learning_rate": 1.3327152001570339e-05, "loss": 1.1056, "step": 2168 }, { "epoch": 0.83, "learning_rate": 1.3321222571522676e-05, "loss": 1.131, "step": 2169 }, { "epoch": 0.83, "learning_rate": 1.3315291828734123e-05, "loss": 1.1076, "step": 2170 }, { "epoch": 0.83, "learning_rate": 1.3309359775548858e-05, "loss": 1.1038, "step": 2171 }, { "epoch": 0.83, "learning_rate": 1.3303426414311568e-05, "loss": 1.108, "step": 2172 }, { "epoch": 0.83, "learning_rate": 1.329749174736747e-05, "loss": 1.1458, "step": 2173 }, { "epoch": 0.84, "learning_rate": 1.3291555777062288e-05, "loss": 1.1757, "step": 2174 }, { "epoch": 0.84, "learning_rate": 1.3285618505742266e-05, "loss": 1.0885, "step": 2175 }, { "epoch": 0.84, "learning_rate": 1.3279679935754155e-05, "loss": 1.1212, "step": 2176 }, { "epoch": 0.84, "learning_rate": 1.3273740069445229e-05, "loss": 1.1341, "step": 2177 }, { "epoch": 0.84, "learning_rate": 1.3267798909163268e-05, "loss": 1.0716, "step": 2178 }, { "epoch": 0.84, "learning_rate": 1.3261856457256563e-05, "loss": 1.113, "step": 2179 }, { "epoch": 0.84, "learning_rate": 1.3255912716073923e-05, "loss": 1.1487, "step": 2180 }, { "epoch": 0.84, "learning_rate": 1.3249967687964656e-05, "loss": 1.1226, "step": 2181 }, { "epoch": 0.84, "learning_rate": 1.3244021375278584e-05, "loss": 1.0534, "step": 2182 }, { "epoch": 0.84, "learning_rate": 1.3238073780366039e-05, "loss": 1.1416, "step": 2183 }, { "epoch": 0.84, "learning_rate": 1.323212490557786e-05, "loss": 1.1252, "step": 2184 }, { "epoch": 0.84, "learning_rate": 1.3226174753265381e-05, "loss": 1.1517, "step": 2185 }, { "epoch": 0.84, "learning_rate": 1.3220223325780456e-05, "loss": 1.1346, "step": 2186 }, { "epoch": 0.84, "learning_rate": 1.3214270625475437e-05, "loss": 1.1585, "step": 2187 }, { "epoch": 0.84, "learning_rate": 1.3208316654703175e-05, "loss": 1.1295, "step": 2188 }, { "epoch": 0.84, "learning_rate": 1.3202361415817026e-05, "loss": 1.1331, "step": 2189 }, { "epoch": 0.84, "learning_rate": 1.319640491117085e-05, "loss": 1.0896, "step": 2190 }, { "epoch": 0.84, "learning_rate": 1.3190447143119007e-05, "loss": 1.053, "step": 2191 }, { "epoch": 0.84, "learning_rate": 1.318448811401635e-05, "loss": 1.0979, "step": 2192 }, { "epoch": 0.84, "learning_rate": 1.3178527826218237e-05, "loss": 1.1519, "step": 2193 }, { "epoch": 0.84, "learning_rate": 1.3172566282080518e-05, "loss": 1.0956, "step": 2194 }, { "epoch": 0.84, "learning_rate": 1.316660348395955e-05, "loss": 1.1101, "step": 2195 }, { "epoch": 0.84, "learning_rate": 1.3160639434212174e-05, "loss": 1.1137, "step": 2196 }, { "epoch": 0.84, "learning_rate": 1.3154674135195734e-05, "loss": 1.1194, "step": 2197 }, { "epoch": 0.84, "learning_rate": 1.3148707589268056e-05, "loss": 1.1651, "step": 2198 }, { "epoch": 0.84, "learning_rate": 1.3142739798787474e-05, "loss": 1.0882, "step": 2199 }, { "epoch": 0.85, "learning_rate": 1.3136770766112807e-05, "loss": 1.107, "step": 2200 }, { "epoch": 0.85, "learning_rate": 1.3130800493603359e-05, "loss": 1.1321, "step": 2201 }, { "epoch": 0.85, "learning_rate": 1.3124828983618936e-05, "loss": 1.0836, "step": 2202 }, { "epoch": 0.85, "learning_rate": 1.3118856238519826e-05, "loss": 1.113, "step": 2203 }, { "epoch": 0.85, "learning_rate": 1.3112882260666805e-05, "loss": 1.1448, "step": 2204 }, { "epoch": 0.85, "learning_rate": 1.310690705242114e-05, "loss": 1.0891, "step": 2205 }, { "epoch": 0.85, "learning_rate": 1.3100930616144579e-05, "loss": 1.135, "step": 2206 }, { "epoch": 0.85, "learning_rate": 1.3094952954199355e-05, "loss": 1.0952, "step": 2207 }, { "epoch": 0.85, "learning_rate": 1.3088974068948194e-05, "loss": 1.1156, "step": 2208 }, { "epoch": 0.85, "learning_rate": 1.3082993962754303e-05, "loss": 1.0679, "step": 2209 }, { "epoch": 0.85, "learning_rate": 1.3077012637981362e-05, "loss": 1.1227, "step": 2210 }, { "epoch": 0.85, "learning_rate": 1.307103009699354e-05, "loss": 1.0885, "step": 2211 }, { "epoch": 0.85, "learning_rate": 1.3065046342155488e-05, "loss": 1.1117, "step": 2212 }, { "epoch": 0.85, "learning_rate": 1.3059061375832339e-05, "loss": 1.1096, "step": 2213 }, { "epoch": 0.85, "learning_rate": 1.3053075200389693e-05, "loss": 1.1558, "step": 2214 }, { "epoch": 0.85, "learning_rate": 1.3047087818193642e-05, "loss": 1.1224, "step": 2215 }, { "epoch": 0.85, "learning_rate": 1.3041099231610743e-05, "loss": 1.1031, "step": 2216 }, { "epoch": 0.85, "learning_rate": 1.3035109443008045e-05, "loss": 1.1776, "step": 2217 }, { "epoch": 0.85, "learning_rate": 1.3029118454753051e-05, "loss": 1.1658, "step": 2218 }, { "epoch": 0.85, "learning_rate": 1.302312626921375e-05, "loss": 1.1386, "step": 2219 }, { "epoch": 0.85, "learning_rate": 1.301713288875861e-05, "loss": 1.132, "step": 2220 }, { "epoch": 0.85, "learning_rate": 1.301113831575656e-05, "loss": 1.0887, "step": 2221 }, { "epoch": 0.85, "learning_rate": 1.3005142552577005e-05, "loss": 1.1095, "step": 2222 }, { "epoch": 0.85, "learning_rate": 1.2999145601589822e-05, "loss": 1.1176, "step": 2223 }, { "epoch": 0.85, "learning_rate": 1.2993147465165353e-05, "loss": 1.1154, "step": 2224 }, { "epoch": 0.85, "learning_rate": 1.2987148145674415e-05, "loss": 1.107, "step": 2225 }, { "epoch": 0.86, "learning_rate": 1.2981147645488288e-05, "loss": 1.1191, "step": 2226 }, { "epoch": 0.86, "learning_rate": 1.2975145966978717e-05, "loss": 1.1257, "step": 2227 }, { "epoch": 0.86, "learning_rate": 1.2969143112517917e-05, "loss": 1.1219, "step": 2228 }, { "epoch": 0.86, "learning_rate": 1.2963139084478568e-05, "loss": 1.1388, "step": 2229 }, { "epoch": 0.86, "learning_rate": 1.2957133885233813e-05, "loss": 1.1136, "step": 2230 }, { "epoch": 0.86, "learning_rate": 1.2951127517157251e-05, "loss": 1.0898, "step": 2231 }, { "epoch": 0.86, "learning_rate": 1.2945119982622954e-05, "loss": 1.1065, "step": 2232 }, { "epoch": 0.86, "learning_rate": 1.2939111284005447e-05, "loss": 1.1309, "step": 2233 }, { "epoch": 0.86, "learning_rate": 1.2933101423679723e-05, "loss": 1.1397, "step": 2234 }, { "epoch": 0.86, "learning_rate": 1.2927090404021226e-05, "loss": 1.0979, "step": 2235 }, { "epoch": 0.86, "learning_rate": 1.2921078227405862e-05, "loss": 1.1366, "step": 2236 }, { "epoch": 0.86, "learning_rate": 1.291506489620999e-05, "loss": 1.0919, "step": 2237 }, { "epoch": 0.86, "learning_rate": 1.2909050412810438e-05, "loss": 1.1078, "step": 2238 }, { "epoch": 0.86, "learning_rate": 1.2903034779584474e-05, "loss": 1.0786, "step": 2239 }, { "epoch": 0.86, "learning_rate": 1.2897017998909826e-05, "loss": 1.0847, "step": 2240 }, { "epoch": 0.86, "learning_rate": 1.2891000073164684e-05, "loss": 1.1317, "step": 2241 }, { "epoch": 0.86, "learning_rate": 1.2884981004727676e-05, "loss": 1.1211, "step": 2242 }, { "epoch": 0.86, "learning_rate": 1.2878960795977894e-05, "loss": 1.0708, "step": 2243 }, { "epoch": 0.86, "learning_rate": 1.2872939449294874e-05, "loss": 1.0893, "step": 2244 }, { "epoch": 0.86, "learning_rate": 1.2866916967058601e-05, "loss": 1.1438, "step": 2245 }, { "epoch": 0.86, "learning_rate": 1.2860893351649513e-05, "loss": 1.111, "step": 2246 }, { "epoch": 0.86, "learning_rate": 1.2854868605448495e-05, "loss": 1.1483, "step": 2247 }, { "epoch": 0.86, "learning_rate": 1.2848842730836879e-05, "loss": 1.1775, "step": 2248 }, { "epoch": 0.86, "learning_rate": 1.2842815730196439e-05, "loss": 1.1248, "step": 2249 }, { "epoch": 0.86, "learning_rate": 1.2836787605909398e-05, "loss": 1.1336, "step": 2250 }, { "epoch": 0.86, "learning_rate": 1.2830758360358427e-05, "loss": 1.0827, "step": 2251 }, { "epoch": 0.87, "learning_rate": 1.2824727995926626e-05, "loss": 1.1039, "step": 2252 }, { "epoch": 0.87, "learning_rate": 1.2818696514997559e-05, "loss": 1.1008, "step": 2253 }, { "epoch": 0.87, "learning_rate": 1.2812663919955209e-05, "loss": 1.0685, "step": 2254 }, { "epoch": 0.87, "learning_rate": 1.2806630213184011e-05, "loss": 1.1235, "step": 2255 }, { "epoch": 0.87, "learning_rate": 1.2800595397068844e-05, "loss": 1.0807, "step": 2256 }, { "epoch": 0.87, "learning_rate": 1.2794559473995011e-05, "loss": 1.1444, "step": 2257 }, { "epoch": 0.87, "learning_rate": 1.2788522446348269e-05, "loss": 1.0743, "step": 2258 }, { "epoch": 0.87, "learning_rate": 1.27824843165148e-05, "loss": 1.1347, "step": 2259 }, { "epoch": 0.87, "learning_rate": 1.2776445086881221e-05, "loss": 1.1306, "step": 2260 }, { "epoch": 0.87, "learning_rate": 1.2770404759834593e-05, "loss": 1.0829, "step": 2261 }, { "epoch": 0.87, "learning_rate": 1.2764363337762404e-05, "loss": 1.1497, "step": 2262 }, { "epoch": 0.87, "learning_rate": 1.2758320823052581e-05, "loss": 1.0917, "step": 2263 }, { "epoch": 0.87, "learning_rate": 1.2752277218093468e-05, "loss": 1.1179, "step": 2264 }, { "epoch": 0.87, "learning_rate": 1.2746232525273861e-05, "loss": 1.1044, "step": 2265 }, { "epoch": 0.87, "learning_rate": 1.2740186746982966e-05, "loss": 1.1125, "step": 2266 }, { "epoch": 0.87, "learning_rate": 1.2734139885610433e-05, "loss": 1.106, "step": 2267 }, { "epoch": 0.87, "learning_rate": 1.2728091943546332e-05, "loss": 1.1008, "step": 2268 }, { "epoch": 0.87, "learning_rate": 1.2722042923181165e-05, "loss": 1.0919, "step": 2269 }, { "epoch": 0.87, "learning_rate": 1.2715992826905853e-05, "loss": 1.1419, "step": 2270 }, { "epoch": 0.87, "learning_rate": 1.270994165711175e-05, "loss": 1.106, "step": 2271 }, { "epoch": 0.87, "learning_rate": 1.2703889416190635e-05, "loss": 1.1009, "step": 2272 }, { "epoch": 0.87, "learning_rate": 1.26978361065347e-05, "loss": 1.0557, "step": 2273 }, { "epoch": 0.87, "learning_rate": 1.269178173053657e-05, "loss": 1.1157, "step": 2274 }, { "epoch": 0.87, "learning_rate": 1.2685726290589287e-05, "loss": 1.1273, "step": 2275 }, { "epoch": 0.87, "learning_rate": 1.2679669789086312e-05, "loss": 1.1363, "step": 2276 }, { "epoch": 0.87, "learning_rate": 1.2673612228421533e-05, "loss": 1.1185, "step": 2277 }, { "epoch": 0.88, "learning_rate": 1.2667553610989243e-05, "loss": 1.1678, "step": 2278 }, { "epoch": 0.88, "learning_rate": 1.2661493939184167e-05, "loss": 1.1409, "step": 2279 }, { "epoch": 0.88, "learning_rate": 1.2655433215401438e-05, "loss": 1.1153, "step": 2280 }, { "epoch": 0.88, "learning_rate": 1.2649371442036614e-05, "loss": 1.0591, "step": 2281 }, { "epoch": 0.88, "learning_rate": 1.2643308621485651e-05, "loss": 1.0748, "step": 2282 }, { "epoch": 0.88, "learning_rate": 1.2637244756144935e-05, "loss": 1.1314, "step": 2283 }, { "epoch": 0.88, "learning_rate": 1.2631179848411259e-05, "loss": 1.0698, "step": 2284 }, { "epoch": 0.88, "learning_rate": 1.2625113900681832e-05, "loss": 1.1246, "step": 2285 }, { "epoch": 0.88, "learning_rate": 1.2619046915354262e-05, "loss": 1.14, "step": 2286 }, { "epoch": 0.88, "learning_rate": 1.2612978894826582e-05, "loss": 1.1166, "step": 2287 }, { "epoch": 0.88, "learning_rate": 1.2606909841497225e-05, "loss": 1.116, "step": 2288 }, { "epoch": 0.88, "learning_rate": 1.2600839757765037e-05, "loss": 1.089, "step": 2289 }, { "epoch": 0.88, "learning_rate": 1.259476864602927e-05, "loss": 1.1306, "step": 2290 }, { "epoch": 0.88, "learning_rate": 1.2588696508689574e-05, "loss": 1.1013, "step": 2291 }, { "epoch": 0.88, "learning_rate": 1.2582623348146022e-05, "loss": 1.0973, "step": 2292 }, { "epoch": 0.88, "learning_rate": 1.2576549166799073e-05, "loss": 1.106, "step": 2293 }, { "epoch": 0.88, "learning_rate": 1.2570473967049602e-05, "loss": 1.1054, "step": 2294 }, { "epoch": 0.88, "learning_rate": 1.2564397751298883e-05, "loss": 1.1033, "step": 2295 }, { "epoch": 0.88, "learning_rate": 1.2558320521948586e-05, "loss": 1.0674, "step": 2296 }, { "epoch": 0.88, "learning_rate": 1.2552242281400794e-05, "loss": 1.144, "step": 2297 }, { "epoch": 0.88, "learning_rate": 1.2546163032057977e-05, "loss": 1.1731, "step": 2298 }, { "epoch": 0.88, "learning_rate": 1.2540082776323009e-05, "loss": 1.1316, "step": 2299 }, { "epoch": 0.88, "learning_rate": 1.2534001516599164e-05, "loss": 1.0564, "step": 2300 }, { "epoch": 0.88, "learning_rate": 1.2527919255290109e-05, "loss": 1.093, "step": 2301 }, { "epoch": 0.88, "learning_rate": 1.2521835994799912e-05, "loss": 1.0894, "step": 2302 }, { "epoch": 0.88, "learning_rate": 1.2515751737533025e-05, "loss": 1.1873, "step": 2303 }, { "epoch": 0.89, "learning_rate": 1.2509666485894304e-05, "loss": 1.122, "step": 2304 }, { "epoch": 0.89, "learning_rate": 1.2503580242289001e-05, "loss": 1.0648, "step": 2305 }, { "epoch": 0.89, "learning_rate": 1.2497493009122753e-05, "loss": 1.076, "step": 2306 }, { "epoch": 0.89, "learning_rate": 1.2491404788801582e-05, "loss": 1.1053, "step": 2307 }, { "epoch": 0.89, "learning_rate": 1.2485315583731914e-05, "loss": 1.1005, "step": 2308 }, { "epoch": 0.89, "learning_rate": 1.2479225396320555e-05, "loss": 1.1047, "step": 2309 }, { "epoch": 0.89, "learning_rate": 1.247313422897471e-05, "loss": 1.1008, "step": 2310 }, { "epoch": 0.89, "learning_rate": 1.2467042084101955e-05, "loss": 1.0888, "step": 2311 }, { "epoch": 0.89, "learning_rate": 1.2460948964110262e-05, "loss": 1.0991, "step": 2312 }, { "epoch": 0.89, "learning_rate": 1.2454854871407993e-05, "loss": 1.0645, "step": 2313 }, { "epoch": 0.89, "learning_rate": 1.2448759808403882e-05, "loss": 1.122, "step": 2314 }, { "epoch": 0.89, "learning_rate": 1.244266377750706e-05, "loss": 1.1039, "step": 2315 }, { "epoch": 0.89, "learning_rate": 1.243656678112703e-05, "loss": 1.1068, "step": 2316 }, { "epoch": 0.89, "learning_rate": 1.243046882167368e-05, "loss": 1.1394, "step": 2317 }, { "epoch": 0.89, "learning_rate": 1.242436990155728e-05, "loss": 1.123, "step": 2318 }, { "epoch": 0.89, "learning_rate": 1.2418270023188481e-05, "loss": 1.1334, "step": 2319 }, { "epoch": 0.89, "learning_rate": 1.2412169188978309e-05, "loss": 1.1198, "step": 2320 }, { "epoch": 0.89, "learning_rate": 1.2406067401338169e-05, "loss": 1.1367, "step": 2321 }, { "epoch": 0.89, "learning_rate": 1.2399964662679845e-05, "loss": 1.0668, "step": 2322 }, { "epoch": 0.89, "learning_rate": 1.2393860975415497e-05, "loss": 1.0819, "step": 2323 }, { "epoch": 0.89, "learning_rate": 1.2387756341957653e-05, "loss": 1.1161, "step": 2324 }, { "epoch": 0.89, "learning_rate": 1.2381650764719224e-05, "loss": 1.0742, "step": 2325 }, { "epoch": 0.89, "learning_rate": 1.2375544246113489e-05, "loss": 1.0824, "step": 2326 }, { "epoch": 0.89, "learning_rate": 1.2369436788554103e-05, "loss": 1.1098, "step": 2327 }, { "epoch": 0.89, "learning_rate": 1.2363328394455084e-05, "loss": 1.1421, "step": 2328 }, { "epoch": 0.89, "learning_rate": 1.2357219066230827e-05, "loss": 1.1627, "step": 2329 }, { "epoch": 0.9, "learning_rate": 1.23511088062961e-05, "loss": 1.0848, "step": 2330 }, { "epoch": 0.9, "learning_rate": 1.2344997617066033e-05, "loss": 1.0868, "step": 2331 }, { "epoch": 0.9, "learning_rate": 1.2338885500956117e-05, "loss": 1.1494, "step": 2332 }, { "epoch": 0.9, "learning_rate": 1.2332772460382226e-05, "loss": 1.1239, "step": 2333 }, { "epoch": 0.9, "learning_rate": 1.2326658497760582e-05, "loss": 1.0705, "step": 2334 }, { "epoch": 0.9, "learning_rate": 1.232054361550779e-05, "loss": 1.1211, "step": 2335 }, { "epoch": 0.9, "learning_rate": 1.2314427816040797e-05, "loss": 1.137, "step": 2336 }, { "epoch": 0.9, "learning_rate": 1.2308311101776933e-05, "loss": 1.169, "step": 2337 }, { "epoch": 0.9, "learning_rate": 1.2302193475133872e-05, "loss": 1.1283, "step": 2338 }, { "epoch": 0.9, "learning_rate": 1.2296074938529663e-05, "loss": 1.1087, "step": 2339 }, { "epoch": 0.9, "learning_rate": 1.2289955494382709e-05, "loss": 1.1452, "step": 2340 }, { "epoch": 0.9, "learning_rate": 1.2283835145111765e-05, "loss": 1.0803, "step": 2341 }, { "epoch": 0.9, "learning_rate": 1.2277713893135953e-05, "loss": 1.141, "step": 2342 }, { "epoch": 0.9, "learning_rate": 1.2271591740874752e-05, "loss": 1.0542, "step": 2343 }, { "epoch": 0.9, "learning_rate": 1.226546869074799e-05, "loss": 1.1051, "step": 2344 }, { "epoch": 0.9, "learning_rate": 1.2259344745175856e-05, "loss": 1.0798, "step": 2345 }, { "epoch": 0.9, "learning_rate": 1.2253219906578888e-05, "loss": 1.1008, "step": 2346 }, { "epoch": 0.9, "learning_rate": 1.2247094177377983e-05, "loss": 1.1287, "step": 2347 }, { "epoch": 0.9, "learning_rate": 1.2240967559994386e-05, "loss": 1.0784, "step": 2348 }, { "epoch": 0.9, "learning_rate": 1.223484005684969e-05, "loss": 1.1149, "step": 2349 }, { "epoch": 0.9, "learning_rate": 1.2228711670365843e-05, "loss": 1.1377, "step": 2350 }, { "epoch": 0.9, "learning_rate": 1.2222582402965142e-05, "loss": 1.0975, "step": 2351 }, { "epoch": 0.9, "learning_rate": 1.2216452257070228e-05, "loss": 1.0805, "step": 2352 }, { "epoch": 0.9, "learning_rate": 1.22103212351041e-05, "loss": 1.1159, "step": 2353 }, { "epoch": 0.9, "learning_rate": 1.2204189339490084e-05, "loss": 1.0776, "step": 2354 }, { "epoch": 0.9, "learning_rate": 1.2198056572651873e-05, "loss": 1.1499, "step": 2355 }, { "epoch": 0.91, "learning_rate": 1.2191922937013489e-05, "loss": 1.1407, "step": 2356 }, { "epoch": 0.91, "learning_rate": 1.2185788434999306e-05, "loss": 1.0641, "step": 2357 }, { "epoch": 0.91, "learning_rate": 1.2179653069034032e-05, "loss": 1.0778, "step": 2358 }, { "epoch": 0.91, "learning_rate": 1.2173516841542725e-05, "loss": 1.0569, "step": 2359 }, { "epoch": 0.91, "learning_rate": 1.2167379754950782e-05, "loss": 1.1214, "step": 2360 }, { "epoch": 0.91, "learning_rate": 1.2161241811683933e-05, "loss": 1.1691, "step": 2361 }, { "epoch": 0.91, "learning_rate": 1.2155103014168252e-05, "loss": 1.0541, "step": 2362 }, { "epoch": 0.91, "learning_rate": 1.2148963364830154e-05, "loss": 1.1153, "step": 2363 }, { "epoch": 0.91, "learning_rate": 1.2142822866096378e-05, "loss": 1.0974, "step": 2364 }, { "epoch": 0.91, "learning_rate": 1.2136681520394019e-05, "loss": 1.0815, "step": 2365 }, { "epoch": 0.91, "learning_rate": 1.2130539330150487e-05, "loss": 1.095, "step": 2366 }, { "epoch": 0.91, "learning_rate": 1.2124396297793533e-05, "loss": 1.1537, "step": 2367 }, { "epoch": 0.91, "learning_rate": 1.2118252425751247e-05, "loss": 1.0896, "step": 2368 }, { "epoch": 0.91, "learning_rate": 1.2112107716452042e-05, "loss": 1.0465, "step": 2369 }, { "epoch": 0.91, "learning_rate": 1.2105962172324667e-05, "loss": 1.0982, "step": 2370 }, { "epoch": 0.91, "learning_rate": 1.20998157957982e-05, "loss": 1.161, "step": 2371 }, { "epoch": 0.91, "learning_rate": 1.2093668589302048e-05, "loss": 1.1363, "step": 2372 }, { "epoch": 0.91, "learning_rate": 1.2087520555265942e-05, "loss": 1.1372, "step": 2373 }, { "epoch": 0.91, "learning_rate": 1.2081371696119945e-05, "loss": 1.1032, "step": 2374 }, { "epoch": 0.91, "learning_rate": 1.2075222014294448e-05, "loss": 1.1146, "step": 2375 }, { "epoch": 0.91, "learning_rate": 1.2069071512220161e-05, "loss": 1.1163, "step": 2376 }, { "epoch": 0.91, "learning_rate": 1.206292019232812e-05, "loss": 1.1382, "step": 2377 }, { "epoch": 0.91, "learning_rate": 1.205676805704969e-05, "loss": 1.1537, "step": 2378 }, { "epoch": 0.91, "learning_rate": 1.205061510881655e-05, "loss": 1.1384, "step": 2379 }, { "epoch": 0.91, "learning_rate": 1.2044461350060705e-05, "loss": 1.1351, "step": 2380 }, { "epoch": 0.91, "learning_rate": 1.2038306783214479e-05, "loss": 1.0879, "step": 2381 }, { "epoch": 0.92, "learning_rate": 1.2032151410710517e-05, "loss": 1.0377, "step": 2382 }, { "epoch": 0.92, "learning_rate": 1.2025995234981782e-05, "loss": 1.0981, "step": 2383 }, { "epoch": 0.92, "learning_rate": 1.2019838258461552e-05, "loss": 1.1193, "step": 2384 }, { "epoch": 0.92, "learning_rate": 1.201368048358342e-05, "loss": 1.0779, "step": 2385 }, { "epoch": 0.92, "learning_rate": 1.2007521912781304e-05, "loss": 1.0877, "step": 2386 }, { "epoch": 0.92, "learning_rate": 1.200136254848943e-05, "loss": 1.0665, "step": 2387 }, { "epoch": 0.92, "learning_rate": 1.1995202393142333e-05, "loss": 1.1399, "step": 2388 }, { "epoch": 0.92, "learning_rate": 1.1989041449174867e-05, "loss": 1.0976, "step": 2389 }, { "epoch": 0.92, "learning_rate": 1.19828797190222e-05, "loss": 1.0911, "step": 2390 }, { "epoch": 0.92, "learning_rate": 1.1976717205119807e-05, "loss": 1.1326, "step": 2391 }, { "epoch": 0.92, "learning_rate": 1.1970553909903471e-05, "loss": 1.0962, "step": 2392 }, { "epoch": 0.92, "learning_rate": 1.1964389835809283e-05, "loss": 1.0801, "step": 2393 }, { "epoch": 0.92, "learning_rate": 1.1958224985273648e-05, "loss": 1.0626, "step": 2394 }, { "epoch": 0.92, "learning_rate": 1.1952059360733277e-05, "loss": 1.0824, "step": 2395 }, { "epoch": 0.92, "learning_rate": 1.1945892964625181e-05, "loss": 1.0729, "step": 2396 }, { "epoch": 0.92, "learning_rate": 1.1939725799386677e-05, "loss": 1.1214, "step": 2397 }, { "epoch": 0.92, "learning_rate": 1.1933557867455394e-05, "loss": 1.0706, "step": 2398 }, { "epoch": 0.92, "learning_rate": 1.1927389171269254e-05, "loss": 1.1349, "step": 2399 }, { "epoch": 0.92, "learning_rate": 1.192121971326649e-05, "loss": 1.119, "step": 2400 }, { "epoch": 0.92, "eval_loss": 1.118867039680481, "eval_runtime": 688.6861, "eval_samples_per_second": 4.938, "eval_steps_per_second": 0.309, "step": 2400 }, { "epoch": 0.92, "learning_rate": 1.1915049495885626e-05, "loss": 1.1259, "step": 2401 }, { "epoch": 0.92, "learning_rate": 1.1908878521565496e-05, "loss": 1.1336, "step": 2402 }, { "epoch": 0.92, "learning_rate": 1.1902706792745232e-05, "loss": 1.1115, "step": 2403 }, { "epoch": 0.92, "learning_rate": 1.1896534311864251e-05, "loss": 1.1122, "step": 2404 }, { "epoch": 0.92, "learning_rate": 1.1890361081362287e-05, "loss": 1.0658, "step": 2405 }, { "epoch": 0.92, "learning_rate": 1.1884187103679357e-05, "loss": 1.1169, "step": 2406 }, { "epoch": 0.92, "learning_rate": 1.1878012381255777e-05, "loss": 1.0749, "step": 2407 }, { "epoch": 0.92, "learning_rate": 1.187183691653216e-05, "loss": 1.123, "step": 2408 }, { "epoch": 0.93, "learning_rate": 1.1865660711949408e-05, "loss": 1.0433, "step": 2409 }, { "epoch": 0.93, "learning_rate": 1.1859483769948715e-05, "loss": 1.1234, "step": 2410 }, { "epoch": 0.93, "learning_rate": 1.185330609297157e-05, "loss": 1.137, "step": 2411 }, { "epoch": 0.93, "learning_rate": 1.1847127683459756e-05, "loss": 1.0465, "step": 2412 }, { "epoch": 0.93, "learning_rate": 1.1840948543855336e-05, "loss": 1.0937, "step": 2413 }, { "epoch": 0.93, "learning_rate": 1.1834768676600665e-05, "loss": 1.1208, "step": 2414 }, { "epoch": 0.93, "learning_rate": 1.1828588084138391e-05, "loss": 1.0938, "step": 2415 }, { "epoch": 0.93, "learning_rate": 1.1822406768911446e-05, "loss": 1.0983, "step": 2416 }, { "epoch": 0.93, "learning_rate": 1.181622473336304e-05, "loss": 1.1251, "step": 2417 }, { "epoch": 0.93, "learning_rate": 1.1810041979936676e-05, "loss": 1.096, "step": 2418 }, { "epoch": 0.93, "learning_rate": 1.1803858511076145e-05, "loss": 1.1438, "step": 2419 }, { "epoch": 0.93, "learning_rate": 1.179767432922551e-05, "loss": 1.1056, "step": 2420 }, { "epoch": 0.93, "learning_rate": 1.179148943682912e-05, "loss": 1.0995, "step": 2421 }, { "epoch": 0.93, "learning_rate": 1.1785303836331603e-05, "loss": 1.1228, "step": 2422 }, { "epoch": 0.93, "learning_rate": 1.1779117530177873e-05, "loss": 1.1006, "step": 2423 }, { "epoch": 0.93, "learning_rate": 1.1772930520813118e-05, "loss": 1.0581, "step": 2424 }, { "epoch": 0.93, "learning_rate": 1.1766742810682805e-05, "loss": 1.1223, "step": 2425 }, { "epoch": 0.93, "learning_rate": 1.1760554402232678e-05, "loss": 1.0941, "step": 2426 }, { "epoch": 0.93, "learning_rate": 1.1754365297908757e-05, "loss": 1.0884, "step": 2427 }, { "epoch": 0.93, "learning_rate": 1.1748175500157335e-05, "loss": 1.1145, "step": 2428 }, { "epoch": 0.93, "learning_rate": 1.1741985011424986e-05, "loss": 1.1269, "step": 2429 }, { "epoch": 0.93, "learning_rate": 1.1735793834158546e-05, "loss": 1.1232, "step": 2430 }, { "epoch": 0.93, "learning_rate": 1.1729601970805133e-05, "loss": 1.092, "step": 2431 }, { "epoch": 0.93, "learning_rate": 1.1723409423812135e-05, "loss": 1.0983, "step": 2432 }, { "epoch": 0.93, "learning_rate": 1.1717216195627207e-05, "loss": 1.0577, "step": 2433 }, { "epoch": 0.93, "learning_rate": 1.1711022288698267e-05, "loss": 1.0677, "step": 2434 }, { "epoch": 0.94, "learning_rate": 1.1704827705473514e-05, "loss": 1.0698, "step": 2435 }, { "epoch": 0.94, "learning_rate": 1.1698632448401406e-05, "loss": 1.0712, "step": 2436 }, { "epoch": 0.94, "learning_rate": 1.1692436519930677e-05, "loss": 1.1321, "step": 2437 }, { "epoch": 0.94, "learning_rate": 1.168623992251031e-05, "loss": 1.14, "step": 2438 }, { "epoch": 0.94, "learning_rate": 1.1680042658589568e-05, "loss": 1.0757, "step": 2439 }, { "epoch": 0.94, "learning_rate": 1.1673844730617967e-05, "loss": 1.1361, "step": 2440 }, { "epoch": 0.94, "learning_rate": 1.1667646141045296e-05, "loss": 1.1315, "step": 2441 }, { "epoch": 0.94, "learning_rate": 1.166144689232159e-05, "loss": 1.0831, "step": 2442 }, { "epoch": 0.94, "learning_rate": 1.1655246986897161e-05, "loss": 1.1379, "step": 2443 }, { "epoch": 0.94, "learning_rate": 1.1649046427222574e-05, "loss": 1.0875, "step": 2444 }, { "epoch": 0.94, "learning_rate": 1.1642845215748646e-05, "loss": 1.1509, "step": 2445 }, { "epoch": 0.94, "learning_rate": 1.1636643354926462e-05, "loss": 1.0766, "step": 2446 }, { "epoch": 0.94, "learning_rate": 1.1630440847207355e-05, "loss": 1.0897, "step": 2447 }, { "epoch": 0.94, "learning_rate": 1.1624237695042921e-05, "loss": 1.1041, "step": 2448 }, { "epoch": 0.94, "learning_rate": 1.1618033900885006e-05, "loss": 1.0847, "step": 2449 }, { "epoch": 0.94, "learning_rate": 1.1611829467185715e-05, "loss": 1.1025, "step": 2450 }, { "epoch": 0.94, "learning_rate": 1.16056243963974e-05, "loss": 1.1448, "step": 2451 }, { "epoch": 0.94, "learning_rate": 1.1599418690972662e-05, "loss": 1.093, "step": 2452 }, { "epoch": 0.94, "learning_rate": 1.1593212353364362e-05, "loss": 1.0841, "step": 2453 }, { "epoch": 0.94, "learning_rate": 1.1587005386025611e-05, "loss": 1.0944, "step": 2454 }, { "epoch": 0.94, "learning_rate": 1.1580797791409757e-05, "loss": 1.1142, "step": 2455 }, { "epoch": 0.94, "learning_rate": 1.157458957197041e-05, "loss": 1.0745, "step": 2456 }, { "epoch": 0.94, "learning_rate": 1.1568380730161414e-05, "loss": 1.1207, "step": 2457 }, { "epoch": 0.94, "learning_rate": 1.156217126843687e-05, "loss": 1.0749, "step": 2458 }, { "epoch": 0.94, "learning_rate": 1.1555961189251119e-05, "loss": 1.1081, "step": 2459 }, { "epoch": 0.94, "learning_rate": 1.1549750495058744e-05, "loss": 1.1288, "step": 2460 }, { "epoch": 0.95, "learning_rate": 1.1543539188314576e-05, "loss": 1.0579, "step": 2461 }, { "epoch": 0.95, "learning_rate": 1.1537327271473685e-05, "loss": 1.1302, "step": 2462 }, { "epoch": 0.95, "learning_rate": 1.1531114746991384e-05, "loss": 1.1455, "step": 2463 }, { "epoch": 0.95, "learning_rate": 1.152490161732322e-05, "loss": 1.1236, "step": 2464 }, { "epoch": 0.95, "learning_rate": 1.1518687884924991e-05, "loss": 1.0787, "step": 2465 }, { "epoch": 0.95, "learning_rate": 1.1512473552252722e-05, "loss": 1.1114, "step": 2466 }, { "epoch": 0.95, "learning_rate": 1.1506258621762685e-05, "loss": 1.1, "step": 2467 }, { "epoch": 0.95, "learning_rate": 1.1500043095911378e-05, "loss": 1.1048, "step": 2468 }, { "epoch": 0.95, "learning_rate": 1.149382697715554e-05, "loss": 1.1087, "step": 2469 }, { "epoch": 0.95, "learning_rate": 1.1487610267952143e-05, "loss": 1.0773, "step": 2470 }, { "epoch": 0.95, "learning_rate": 1.1481392970758399e-05, "loss": 1.1081, "step": 2471 }, { "epoch": 0.95, "learning_rate": 1.1475175088031738e-05, "loss": 1.148, "step": 2472 }, { "epoch": 0.95, "learning_rate": 1.1468956622229837e-05, "loss": 1.1103, "step": 2473 }, { "epoch": 0.95, "learning_rate": 1.146273757581059e-05, "loss": 1.1662, "step": 2474 }, { "epoch": 0.95, "learning_rate": 1.1456517951232136e-05, "loss": 1.1335, "step": 2475 }, { "epoch": 0.95, "learning_rate": 1.1450297750952824e-05, "loss": 1.11, "step": 2476 }, { "epoch": 0.95, "learning_rate": 1.1444076977431242e-05, "loss": 1.0601, "step": 2477 }, { "epoch": 0.95, "learning_rate": 1.1437855633126208e-05, "loss": 1.163, "step": 2478 }, { "epoch": 0.95, "learning_rate": 1.1431633720496755e-05, "loss": 1.1297, "step": 2479 }, { "epoch": 0.95, "learning_rate": 1.142541124200215e-05, "loss": 1.1317, "step": 2480 }, { "epoch": 0.95, "learning_rate": 1.1419188200101875e-05, "loss": 1.1308, "step": 2481 }, { "epoch": 0.95, "learning_rate": 1.141296459725564e-05, "loss": 1.1229, "step": 2482 }, { "epoch": 0.95, "learning_rate": 1.1406740435923378e-05, "loss": 1.1129, "step": 2483 }, { "epoch": 0.95, "learning_rate": 1.1400515718565242e-05, "loss": 1.1179, "step": 2484 }, { "epoch": 0.95, "learning_rate": 1.1394290447641596e-05, "loss": 1.0818, "step": 2485 }, { "epoch": 0.95, "learning_rate": 1.1388064625613038e-05, "loss": 1.0674, "step": 2486 }, { "epoch": 0.96, "learning_rate": 1.138183825494037e-05, "loss": 1.0844, "step": 2487 }, { "epoch": 0.96, "learning_rate": 1.1375611338084624e-05, "loss": 1.0876, "step": 2488 }, { "epoch": 0.96, "learning_rate": 1.1369383877507035e-05, "loss": 1.0901, "step": 2489 }, { "epoch": 0.96, "learning_rate": 1.1363155875669061e-05, "loss": 1.0885, "step": 2490 }, { "epoch": 0.96, "learning_rate": 1.1356927335032368e-05, "loss": 1.1345, "step": 2491 }, { "epoch": 0.96, "learning_rate": 1.1350698258058847e-05, "loss": 1.0868, "step": 2492 }, { "epoch": 0.96, "learning_rate": 1.1344468647210584e-05, "loss": 1.0968, "step": 2493 }, { "epoch": 0.96, "learning_rate": 1.1338238504949887e-05, "loss": 1.1285, "step": 2494 }, { "epoch": 0.96, "learning_rate": 1.1332007833739275e-05, "loss": 1.1578, "step": 2495 }, { "epoch": 0.96, "learning_rate": 1.1325776636041469e-05, "loss": 1.1373, "step": 2496 }, { "epoch": 0.96, "learning_rate": 1.1319544914319402e-05, "loss": 1.1535, "step": 2497 }, { "epoch": 0.96, "learning_rate": 1.1313312671036216e-05, "loss": 1.1302, "step": 2498 }, { "epoch": 0.96, "learning_rate": 1.1307079908655255e-05, "loss": 1.1474, "step": 2499 }, { "epoch": 0.96, "learning_rate": 1.130084662964007e-05, "loss": 1.0849, "step": 2500 }, { "epoch": 0.96, "learning_rate": 1.1294612836454422e-05, "loss": 1.0786, "step": 2501 }, { "epoch": 0.96, "learning_rate": 1.128837853156226e-05, "loss": 1.0889, "step": 2502 }, { "epoch": 0.96, "learning_rate": 1.128214371742775e-05, "loss": 1.0286, "step": 2503 }, { "epoch": 0.96, "learning_rate": 1.127590839651526e-05, "loss": 1.171, "step": 2504 }, { "epoch": 0.96, "learning_rate": 1.1269672571289347e-05, "loss": 1.1269, "step": 2505 }, { "epoch": 0.96, "learning_rate": 1.1263436244214769e-05, "loss": 1.1101, "step": 2506 }, { "epoch": 0.96, "learning_rate": 1.125719941775649e-05, "loss": 1.1244, "step": 2507 }, { "epoch": 0.96, "learning_rate": 1.125096209437967e-05, "loss": 1.1462, "step": 2508 }, { "epoch": 0.96, "learning_rate": 1.124472427654966e-05, "loss": 1.1264, "step": 2509 }, { "epoch": 0.96, "learning_rate": 1.123848596673201e-05, "loss": 1.0648, "step": 2510 }, { "epoch": 0.96, "learning_rate": 1.1232247167392463e-05, "loss": 1.0749, "step": 2511 }, { "epoch": 0.96, "learning_rate": 1.1226007880996956e-05, "loss": 1.1099, "step": 2512 }, { "epoch": 0.97, "learning_rate": 1.1219768110011626e-05, "loss": 1.1081, "step": 2513 }, { "epoch": 0.97, "learning_rate": 1.1213527856902786e-05, "loss": 1.0446, "step": 2514 }, { "epoch": 0.97, "learning_rate": 1.1207287124136949e-05, "loss": 1.1177, "step": 2515 }, { "epoch": 0.97, "learning_rate": 1.1201045914180822e-05, "loss": 1.1171, "step": 2516 }, { "epoch": 0.97, "learning_rate": 1.1194804229501291e-05, "loss": 1.1276, "step": 2517 }, { "epoch": 0.97, "learning_rate": 1.1188562072565434e-05, "loss": 1.1102, "step": 2518 }, { "epoch": 0.97, "learning_rate": 1.1182319445840515e-05, "loss": 1.1278, "step": 2519 }, { "epoch": 0.97, "learning_rate": 1.1176076351793988e-05, "loss": 1.0995, "step": 2520 }, { "epoch": 0.97, "learning_rate": 1.1169832792893483e-05, "loss": 1.0944, "step": 2521 }, { "epoch": 0.97, "learning_rate": 1.1163588771606823e-05, "loss": 1.0833, "step": 2522 }, { "epoch": 0.97, "learning_rate": 1.1157344290402006e-05, "loss": 1.1065, "step": 2523 }, { "epoch": 0.97, "learning_rate": 1.1151099351747216e-05, "loss": 1.1112, "step": 2524 }, { "epoch": 0.97, "learning_rate": 1.1144853958110817e-05, "loss": 1.0537, "step": 2525 }, { "epoch": 0.97, "learning_rate": 1.1138608111961358e-05, "loss": 1.1584, "step": 2526 }, { "epoch": 0.97, "learning_rate": 1.1132361815767554e-05, "loss": 1.1166, "step": 2527 }, { "epoch": 0.97, "learning_rate": 1.1126115071998312e-05, "loss": 1.1211, "step": 2528 }, { "epoch": 0.97, "learning_rate": 1.1119867883122703e-05, "loss": 1.1475, "step": 2529 }, { "epoch": 0.97, "learning_rate": 1.1113620251609985e-05, "loss": 1.1037, "step": 2530 }, { "epoch": 0.97, "learning_rate": 1.1107372179929584e-05, "loss": 1.1281, "step": 2531 }, { "epoch": 0.97, "learning_rate": 1.1101123670551106e-05, "loss": 1.1123, "step": 2532 }, { "epoch": 0.97, "learning_rate": 1.1094874725944322e-05, "loss": 1.174, "step": 2533 }, { "epoch": 0.97, "learning_rate": 1.108862534857918e-05, "loss": 1.2035, "step": 2534 }, { "epoch": 0.97, "learning_rate": 1.1082375540925805e-05, "loss": 1.0774, "step": 2535 }, { "epoch": 0.97, "learning_rate": 1.1076125305454478e-05, "loss": 1.1066, "step": 2536 }, { "epoch": 0.97, "learning_rate": 1.1069874644635657e-05, "loss": 1.0947, "step": 2537 }, { "epoch": 0.97, "learning_rate": 1.1063623560939973e-05, "loss": 1.0627, "step": 2538 }, { "epoch": 0.98, "learning_rate": 1.1057372056838213e-05, "loss": 1.0978, "step": 2539 }, { "epoch": 0.98, "learning_rate": 1.1051120134801343e-05, "loss": 1.1403, "step": 2540 }, { "epoch": 0.98, "learning_rate": 1.104486779730048e-05, "loss": 1.0705, "step": 2541 }, { "epoch": 0.98, "learning_rate": 1.1038615046806916e-05, "loss": 1.0761, "step": 2542 }, { "epoch": 0.98, "learning_rate": 1.1032361885792102e-05, "loss": 1.0725, "step": 2543 }, { "epoch": 0.98, "learning_rate": 1.1026108316727653e-05, "loss": 1.1223, "step": 2544 }, { "epoch": 0.98, "learning_rate": 1.101985434208534e-05, "loss": 1.1064, "step": 2545 }, { "epoch": 0.98, "learning_rate": 1.1013599964337107e-05, "loss": 1.1366, "step": 2546 }, { "epoch": 0.98, "learning_rate": 1.1007345185955044e-05, "loss": 1.153, "step": 2547 }, { "epoch": 0.98, "learning_rate": 1.1001090009411401e-05, "loss": 1.078, "step": 2548 }, { "epoch": 0.98, "learning_rate": 1.0994834437178595e-05, "loss": 1.1026, "step": 2549 }, { "epoch": 0.98, "learning_rate": 1.0988578471729186e-05, "loss": 1.1303, "step": 2550 }, { "epoch": 0.98, "learning_rate": 1.0982322115535904e-05, "loss": 1.0354, "step": 2551 }, { "epoch": 0.98, "learning_rate": 1.0976065371071625e-05, "loss": 1.0821, "step": 2552 }, { "epoch": 0.98, "learning_rate": 1.0969808240809373e-05, "loss": 1.1283, "step": 2553 }, { "epoch": 0.98, "learning_rate": 1.0963550727222332e-05, "loss": 1.1288, "step": 2554 }, { "epoch": 0.98, "learning_rate": 1.0957292832783841e-05, "loss": 1.0459, "step": 2555 }, { "epoch": 0.98, "learning_rate": 1.0951034559967387e-05, "loss": 1.1318, "step": 2556 }, { "epoch": 0.98, "learning_rate": 1.0944775911246595e-05, "loss": 1.0693, "step": 2557 }, { "epoch": 0.98, "learning_rate": 1.0938516889095254e-05, "loss": 1.1592, "step": 2558 }, { "epoch": 0.98, "learning_rate": 1.0932257495987296e-05, "loss": 1.1466, "step": 2559 }, { "epoch": 0.98, "learning_rate": 1.0925997734396797e-05, "loss": 1.1336, "step": 2560 }, { "epoch": 0.98, "learning_rate": 1.0919737606797975e-05, "loss": 1.1096, "step": 2561 }, { "epoch": 0.98, "learning_rate": 1.0913477115665205e-05, "loss": 1.1192, "step": 2562 }, { "epoch": 0.98, "learning_rate": 1.0907216263472995e-05, "loss": 1.1139, "step": 2563 }, { "epoch": 0.98, "learning_rate": 1.0900955052696001e-05, "loss": 1.0609, "step": 2564 }, { "epoch": 0.99, "learning_rate": 1.0894693485809016e-05, "loss": 1.1225, "step": 2565 }, { "epoch": 0.99, "learning_rate": 1.0888431565286974e-05, "loss": 1.1123, "step": 2566 }, { "epoch": 0.99, "learning_rate": 1.0882169293604956e-05, "loss": 1.0804, "step": 2567 }, { "epoch": 0.99, "learning_rate": 1.0875906673238175e-05, "loss": 1.1058, "step": 2568 }, { "epoch": 0.99, "learning_rate": 1.0869643706661988e-05, "loss": 1.0663, "step": 2569 }, { "epoch": 0.99, "learning_rate": 1.0863380396351877e-05, "loss": 1.0662, "step": 2570 }, { "epoch": 0.99, "learning_rate": 1.0857116744783474e-05, "loss": 1.1161, "step": 2571 }, { "epoch": 0.99, "learning_rate": 1.0850852754432537e-05, "loss": 1.1118, "step": 2572 }, { "epoch": 0.99, "learning_rate": 1.084458842777496e-05, "loss": 1.0939, "step": 2573 }, { "epoch": 0.99, "learning_rate": 1.083832376728677e-05, "loss": 1.0809, "step": 2574 }, { "epoch": 0.99, "learning_rate": 1.0832058775444127e-05, "loss": 1.141, "step": 2575 }, { "epoch": 0.99, "learning_rate": 1.0825793454723325e-05, "loss": 1.0319, "step": 2576 }, { "epoch": 0.99, "learning_rate": 1.0819527807600781e-05, "loss": 1.1359, "step": 2577 }, { "epoch": 0.99, "learning_rate": 1.081326183655304e-05, "loss": 1.066, "step": 2578 }, { "epoch": 0.99, "learning_rate": 1.0806995544056782e-05, "loss": 1.0641, "step": 2579 }, { "epoch": 0.99, "learning_rate": 1.0800728932588813e-05, "loss": 1.1229, "step": 2580 }, { "epoch": 0.99, "learning_rate": 1.0794462004626064e-05, "loss": 1.1535, "step": 2581 }, { "epoch": 0.99, "learning_rate": 1.078819476264559e-05, "loss": 1.0972, "step": 2582 }, { "epoch": 0.99, "learning_rate": 1.0781927209124567e-05, "loss": 1.0659, "step": 2583 }, { "epoch": 0.99, "learning_rate": 1.0775659346540303e-05, "loss": 1.1038, "step": 2584 }, { "epoch": 0.99, "learning_rate": 1.0769391177370221e-05, "loss": 1.0321, "step": 2585 }, { "epoch": 0.99, "learning_rate": 1.0763122704091865e-05, "loss": 1.1525, "step": 2586 }, { "epoch": 0.99, "learning_rate": 1.0756853929182902e-05, "loss": 1.1223, "step": 2587 }, { "epoch": 0.99, "learning_rate": 1.0750584855121122e-05, "loss": 1.0829, "step": 2588 }, { "epoch": 0.99, "learning_rate": 1.0744315484384423e-05, "loss": 1.111, "step": 2589 }, { "epoch": 0.99, "learning_rate": 1.0738045819450828e-05, "loss": 1.0538, "step": 2590 }, { "epoch": 1.0, "learning_rate": 1.0731775862798474e-05, "loss": 1.0479, "step": 2591 }, { "epoch": 1.0, "learning_rate": 1.0725505616905613e-05, "loss": 1.0679, "step": 2592 }, { "epoch": 1.0, "learning_rate": 1.0719235084250615e-05, "loss": 1.1336, "step": 2593 }, { "epoch": 1.0, "learning_rate": 1.0712964267311959e-05, "loss": 1.122, "step": 2594 }, { "epoch": 1.0, "learning_rate": 1.0706693168568235e-05, "loss": 1.0915, "step": 2595 }, { "epoch": 1.0, "learning_rate": 1.0700421790498147e-05, "loss": 1.1394, "step": 2596 }, { "epoch": 1.0, "learning_rate": 1.0694150135580512e-05, "loss": 1.0581, "step": 2597 }, { "epoch": 1.0, "learning_rate": 1.0687878206294257e-05, "loss": 1.1065, "step": 2598 }, { "epoch": 1.0, "learning_rate": 1.0681606005118406e-05, "loss": 1.0877, "step": 2599 }, { "epoch": 1.0, "learning_rate": 1.0675333534532106e-05, "loss": 1.1731, "step": 2600 }, { "epoch": 1.0, "learning_rate": 1.0669060797014601e-05, "loss": 1.0822, "step": 2601 }, { "epoch": 1.0, "learning_rate": 1.0662787795045241e-05, "loss": 1.1207, "step": 2602 }, { "epoch": 1.0, "learning_rate": 1.0656514531103484e-05, "loss": 1.0404, "step": 2603 }, { "epoch": 1.0, "learning_rate": 1.0650241007668886e-05, "loss": 1.1412, "step": 2604 }, { "epoch": 1.0, "learning_rate": 1.0643967227221114e-05, "loss": 1.1087, "step": 2605 }, { "epoch": 1.0, "learning_rate": 1.0637693192239935e-05, "loss": 1.0944, "step": 2606 }, { "epoch": 1.0, "learning_rate": 1.0631418905205204e-05, "loss": 1.1443, "step": 2607 }, { "epoch": 1.0, "learning_rate": 1.0625144368596893e-05, "loss": 1.0328, "step": 2608 }, { "epoch": 1.0, "learning_rate": 1.0618869584895062e-05, "loss": 1.1097, "step": 2609 }, { "epoch": 1.0, "learning_rate": 1.0612594556579874e-05, "loss": 1.1094, "step": 2610 }, { "epoch": 1.0, "learning_rate": 1.0606319286131581e-05, "loss": 1.1263, "step": 2611 }, { "epoch": 1.0, "learning_rate": 1.0600043776030542e-05, "loss": 1.123, "step": 2612 }, { "epoch": 1.0, "learning_rate": 1.0593768028757198e-05, "loss": 1.0825, "step": 2613 }, { "epoch": 1.0, "learning_rate": 1.0587492046792095e-05, "loss": 1.1282, "step": 2614 }, { "epoch": 1.0, "learning_rate": 1.058121583261587e-05, "loss": 1.1175, "step": 2615 }, { "epoch": 1.0, "learning_rate": 1.0574939388709243e-05, "loss": 1.104, "step": 2616 }, { "epoch": 1.01, "learning_rate": 1.0568662717553032e-05, "loss": 1.0901, "step": 2617 }, { "epoch": 1.01, "learning_rate": 1.0562385821628144e-05, "loss": 1.0966, "step": 2618 }, { "epoch": 1.01, "learning_rate": 1.0556108703415578e-05, "loss": 1.0694, "step": 2619 }, { "epoch": 1.01, "learning_rate": 1.054983136539641e-05, "loss": 1.0626, "step": 2620 }, { "epoch": 1.01, "learning_rate": 1.0543553810051816e-05, "loss": 1.132, "step": 2621 }, { "epoch": 1.01, "learning_rate": 1.0537276039863049e-05, "loss": 1.1236, "step": 2622 }, { "epoch": 1.01, "learning_rate": 1.0530998057311455e-05, "loss": 1.1778, "step": 2623 }, { "epoch": 1.01, "learning_rate": 1.0524719864878454e-05, "loss": 1.0974, "step": 2624 }, { "epoch": 1.01, "learning_rate": 1.0518441465045555e-05, "loss": 1.1153, "step": 2625 }, { "epoch": 1.01, "learning_rate": 1.0512162860294346e-05, "loss": 1.0351, "step": 2626 }, { "epoch": 1.01, "learning_rate": 1.0505884053106501e-05, "loss": 1.071, "step": 2627 }, { "epoch": 1.01, "learning_rate": 1.0499605045963773e-05, "loss": 1.1453, "step": 2628 }, { "epoch": 1.01, "learning_rate": 1.0493325841347984e-05, "loss": 1.085, "step": 2629 }, { "epoch": 1.01, "learning_rate": 1.0487046441741047e-05, "loss": 1.1325, "step": 2630 }, { "epoch": 1.01, "learning_rate": 1.0480766849624946e-05, "loss": 1.1296, "step": 2631 }, { "epoch": 1.01, "learning_rate": 1.0474487067481745e-05, "loss": 1.071, "step": 2632 }, { "epoch": 1.01, "learning_rate": 1.0468207097793577e-05, "loss": 1.154, "step": 2633 }, { "epoch": 1.01, "learning_rate": 1.0461926943042651e-05, "loss": 1.1255, "step": 2634 }, { "epoch": 1.01, "learning_rate": 1.0455646605711254e-05, "loss": 1.1602, "step": 2635 }, { "epoch": 1.01, "learning_rate": 1.0449366088281742e-05, "loss": 1.0846, "step": 2636 }, { "epoch": 1.01, "learning_rate": 1.0443085393236539e-05, "loss": 1.0818, "step": 2637 }, { "epoch": 1.01, "learning_rate": 1.0436804523058141e-05, "loss": 1.1279, "step": 2638 }, { "epoch": 1.01, "learning_rate": 1.0430523480229118e-05, "loss": 1.084, "step": 2639 }, { "epoch": 1.01, "learning_rate": 1.0424242267232102e-05, "loss": 1.1232, "step": 2640 }, { "epoch": 1.01, "learning_rate": 1.0417960886549799e-05, "loss": 1.0855, "step": 2641 }, { "epoch": 1.01, "learning_rate": 1.0411679340664969e-05, "loss": 1.0659, "step": 2642 }, { "epoch": 1.02, "learning_rate": 1.0405397632060453e-05, "loss": 1.1025, "step": 2643 }, { "epoch": 1.02, "learning_rate": 1.0399115763219145e-05, "loss": 1.0848, "step": 2644 }, { "epoch": 1.02, "learning_rate": 1.0392833736624003e-05, "loss": 1.1059, "step": 2645 }, { "epoch": 1.02, "learning_rate": 1.0386551554758058e-05, "loss": 1.1072, "step": 2646 }, { "epoch": 1.02, "learning_rate": 1.0380269220104391e-05, "loss": 1.1214, "step": 2647 }, { "epoch": 1.02, "learning_rate": 1.0373986735146147e-05, "loss": 1.1557, "step": 2648 }, { "epoch": 1.02, "learning_rate": 1.0367704102366529e-05, "loss": 1.1377, "step": 2649 }, { "epoch": 1.02, "learning_rate": 1.0361421324248801e-05, "loss": 1.1054, "step": 2650 }, { "epoch": 1.02, "learning_rate": 1.0355138403276283e-05, "loss": 1.1055, "step": 2651 }, { "epoch": 1.02, "learning_rate": 1.0348855341932353e-05, "loss": 1.0982, "step": 2652 }, { "epoch": 1.02, "learning_rate": 1.0342572142700447e-05, "loss": 1.107, "step": 2653 }, { "epoch": 1.02, "learning_rate": 1.0336288808064043e-05, "loss": 1.1592, "step": 2654 }, { "epoch": 1.02, "learning_rate": 1.0330005340506685e-05, "loss": 1.1331, "step": 2655 }, { "epoch": 1.02, "learning_rate": 1.032372174251197e-05, "loss": 1.1365, "step": 2656 }, { "epoch": 1.02, "learning_rate": 1.031743801656354e-05, "loss": 1.0975, "step": 2657 }, { "epoch": 1.02, "learning_rate": 1.0311154165145087e-05, "loss": 1.1192, "step": 2658 }, { "epoch": 1.02, "learning_rate": 1.030487019074036e-05, "loss": 1.1277, "step": 2659 }, { "epoch": 1.02, "learning_rate": 1.0298586095833152e-05, "loss": 1.1677, "step": 2660 }, { "epoch": 1.02, "learning_rate": 1.0292301882907299e-05, "loss": 1.1113, "step": 2661 }, { "epoch": 1.02, "learning_rate": 1.0286017554446692e-05, "loss": 1.1255, "step": 2662 }, { "epoch": 1.02, "learning_rate": 1.0279733112935265e-05, "loss": 1.1267, "step": 2663 }, { "epoch": 1.02, "learning_rate": 1.0273448560856993e-05, "loss": 1.1063, "step": 2664 }, { "epoch": 1.02, "learning_rate": 1.02671639006959e-05, "loss": 1.1626, "step": 2665 }, { "epoch": 1.02, "learning_rate": 1.026087913493605e-05, "loss": 1.0729, "step": 2666 }, { "epoch": 1.02, "learning_rate": 1.0254594266061548e-05, "loss": 1.0728, "step": 2667 }, { "epoch": 1.02, "learning_rate": 1.0248309296556543e-05, "loss": 1.1321, "step": 2668 }, { "epoch": 1.03, "learning_rate": 1.0242024228905218e-05, "loss": 1.0818, "step": 2669 }, { "epoch": 1.03, "learning_rate": 1.0235739065591803e-05, "loss": 1.0636, "step": 2670 }, { "epoch": 1.03, "learning_rate": 1.0229453809100556e-05, "loss": 1.1044, "step": 2671 }, { "epoch": 1.03, "learning_rate": 1.022316846191578e-05, "loss": 1.1265, "step": 2672 }, { "epoch": 1.03, "learning_rate": 1.0216883026521812e-05, "loss": 1.0497, "step": 2673 }, { "epoch": 1.03, "learning_rate": 1.021059750540302e-05, "loss": 1.1003, "step": 2674 }, { "epoch": 1.03, "learning_rate": 1.0204311901043807e-05, "loss": 1.0806, "step": 2675 }, { "epoch": 1.03, "learning_rate": 1.0198026215928614e-05, "loss": 1.1022, "step": 2676 }, { "epoch": 1.03, "learning_rate": 1.019174045254191e-05, "loss": 1.093, "step": 2677 }, { "epoch": 1.03, "learning_rate": 1.0185454613368197e-05, "loss": 1.1209, "step": 2678 }, { "epoch": 1.03, "learning_rate": 1.0179168700892001e-05, "loss": 1.0828, "step": 2679 }, { "epoch": 1.03, "learning_rate": 1.0172882717597883e-05, "loss": 1.0893, "step": 2680 }, { "epoch": 1.03, "learning_rate": 1.016659666597043e-05, "loss": 1.0877, "step": 2681 }, { "epoch": 1.03, "learning_rate": 1.016031054849426e-05, "loss": 1.0658, "step": 2682 }, { "epoch": 1.03, "learning_rate": 1.0154024367654007e-05, "loss": 1.087, "step": 2683 }, { "epoch": 1.03, "learning_rate": 1.0147738125934343e-05, "loss": 1.1062, "step": 2684 }, { "epoch": 1.03, "learning_rate": 1.014145182581995e-05, "loss": 1.0524, "step": 2685 }, { "epoch": 1.03, "learning_rate": 1.0135165469795547e-05, "loss": 1.0617, "step": 2686 }, { "epoch": 1.03, "learning_rate": 1.0128879060345866e-05, "loss": 1.1451, "step": 2687 }, { "epoch": 1.03, "learning_rate": 1.0122592599955661e-05, "loss": 1.1352, "step": 2688 }, { "epoch": 1.03, "learning_rate": 1.0116306091109711e-05, "loss": 1.0761, "step": 2689 }, { "epoch": 1.03, "learning_rate": 1.0110019536292807e-05, "loss": 1.1465, "step": 2690 }, { "epoch": 1.03, "learning_rate": 1.010373293798977e-05, "loss": 1.0905, "step": 2691 }, { "epoch": 1.03, "learning_rate": 1.009744629868542e-05, "loss": 1.1452, "step": 2692 }, { "epoch": 1.03, "learning_rate": 1.009115962086461e-05, "loss": 1.1102, "step": 2693 }, { "epoch": 1.03, "learning_rate": 1.00848729070122e-05, "loss": 1.0547, "step": 2694 }, { "epoch": 1.04, "learning_rate": 1.0078586159613071e-05, "loss": 1.1367, "step": 2695 }, { "epoch": 1.04, "learning_rate": 1.0072299381152105e-05, "loss": 1.1186, "step": 2696 }, { "epoch": 1.04, "learning_rate": 1.0066012574114203e-05, "loss": 1.1281, "step": 2697 }, { "epoch": 1.04, "learning_rate": 1.0059725740984285e-05, "loss": 1.0521, "step": 2698 }, { "epoch": 1.04, "learning_rate": 1.0053438884247269e-05, "loss": 1.1145, "step": 2699 }, { "epoch": 1.04, "learning_rate": 1.004715200638809e-05, "loss": 1.1071, "step": 2700 }, { "epoch": 1.04, "learning_rate": 1.0040865109891685e-05, "loss": 1.1878, "step": 2701 }, { "epoch": 1.04, "learning_rate": 1.0034578197243006e-05, "loss": 1.1367, "step": 2702 }, { "epoch": 1.04, "learning_rate": 1.0028291270927006e-05, "loss": 1.1064, "step": 2703 }, { "epoch": 1.04, "learning_rate": 1.0022004333428647e-05, "loss": 1.1113, "step": 2704 }, { "epoch": 1.04, "learning_rate": 1.001571738723289e-05, "loss": 1.0405, "step": 2705 }, { "epoch": 1.04, "learning_rate": 1.0009430434824707e-05, "loss": 1.1025, "step": 2706 }, { "epoch": 1.04, "learning_rate": 1.0003143478689064e-05, "loss": 1.1067, "step": 2707 }, { "epoch": 1.04, "learning_rate": 9.996856521310939e-06, "loss": 1.0839, "step": 2708 }, { "epoch": 1.04, "learning_rate": 9.990569565175294e-06, "loss": 1.1435, "step": 2709 }, { "epoch": 1.04, "learning_rate": 9.984282612767111e-06, "loss": 1.1076, "step": 2710 }, { "epoch": 1.04, "learning_rate": 9.977995666571353e-06, "loss": 1.0999, "step": 2711 }, { "epoch": 1.04, "learning_rate": 9.971708729072998e-06, "loss": 1.0785, "step": 2712 }, { "epoch": 1.04, "learning_rate": 9.965421802756995e-06, "loss": 1.1289, "step": 2713 }, { "epoch": 1.04, "learning_rate": 9.959134890108319e-06, "loss": 1.1182, "step": 2714 }, { "epoch": 1.04, "learning_rate": 9.952847993611917e-06, "loss": 1.0738, "step": 2715 }, { "epoch": 1.04, "learning_rate": 9.946561115752735e-06, "loss": 1.1529, "step": 2716 }, { "epoch": 1.04, "learning_rate": 9.94027425901572e-06, "loss": 1.1581, "step": 2717 }, { "epoch": 1.04, "learning_rate": 9.933987425885799e-06, "loss": 1.1088, "step": 2718 }, { "epoch": 1.04, "learning_rate": 9.9277006188479e-06, "loss": 1.0923, "step": 2719 }, { "epoch": 1.04, "learning_rate": 9.921413840386934e-06, "loss": 1.0692, "step": 2720 }, { "epoch": 1.05, "learning_rate": 9.915127092987802e-06, "loss": 1.0555, "step": 2721 }, { "epoch": 1.05, "learning_rate": 9.90884037913539e-06, "loss": 1.108, "step": 2722 }, { "epoch": 1.05, "learning_rate": 9.902553701314582e-06, "loss": 1.0487, "step": 2723 }, { "epoch": 1.05, "learning_rate": 9.896267062010234e-06, "loss": 1.1324, "step": 2724 }, { "epoch": 1.05, "learning_rate": 9.889980463707195e-06, "loss": 1.1019, "step": 2725 }, { "epoch": 1.05, "learning_rate": 9.883693908890295e-06, "loss": 1.0783, "step": 2726 }, { "epoch": 1.05, "learning_rate": 9.877407400044342e-06, "loss": 1.0745, "step": 2727 }, { "epoch": 1.05, "learning_rate": 9.871120939654139e-06, "loss": 1.0983, "step": 2728 }, { "epoch": 1.05, "learning_rate": 9.864834530204456e-06, "loss": 1.0635, "step": 2729 }, { "epoch": 1.05, "learning_rate": 9.858548174180056e-06, "loss": 1.0976, "step": 2730 }, { "epoch": 1.05, "learning_rate": 9.85226187406566e-06, "loss": 1.1389, "step": 2731 }, { "epoch": 1.05, "learning_rate": 9.845975632345996e-06, "loss": 1.0585, "step": 2732 }, { "epoch": 1.05, "learning_rate": 9.839689451505744e-06, "loss": 1.1185, "step": 2733 }, { "epoch": 1.05, "learning_rate": 9.833403334029573e-06, "loss": 1.0512, "step": 2734 }, { "epoch": 1.05, "learning_rate": 9.827117282402119e-06, "loss": 1.1215, "step": 2735 }, { "epoch": 1.05, "learning_rate": 9.820831299108002e-06, "loss": 1.078, "step": 2736 }, { "epoch": 1.05, "learning_rate": 9.814545386631806e-06, "loss": 1.0879, "step": 2737 }, { "epoch": 1.05, "learning_rate": 9.808259547458092e-06, "loss": 1.1787, "step": 2738 }, { "epoch": 1.05, "learning_rate": 9.80197378407139e-06, "loss": 1.1326, "step": 2739 }, { "epoch": 1.05, "learning_rate": 9.795688098956195e-06, "loss": 1.0554, "step": 2740 }, { "epoch": 1.05, "learning_rate": 9.789402494596985e-06, "loss": 1.0815, "step": 2741 }, { "epoch": 1.05, "learning_rate": 9.783116973478193e-06, "loss": 1.0616, "step": 2742 }, { "epoch": 1.05, "learning_rate": 9.776831538084223e-06, "loss": 1.1564, "step": 2743 }, { "epoch": 1.05, "learning_rate": 9.770546190899446e-06, "loss": 1.0819, "step": 2744 }, { "epoch": 1.05, "learning_rate": 9.7642609344082e-06, "loss": 1.0986, "step": 2745 }, { "epoch": 1.05, "learning_rate": 9.757975771094784e-06, "loss": 1.0828, "step": 2746 }, { "epoch": 1.06, "learning_rate": 9.75169070344346e-06, "loss": 1.0961, "step": 2747 }, { "epoch": 1.06, "learning_rate": 9.745405733938452e-06, "loss": 1.0804, "step": 2748 }, { "epoch": 1.06, "learning_rate": 9.739120865063952e-06, "loss": 1.0998, "step": 2749 }, { "epoch": 1.06, "learning_rate": 9.732836099304103e-06, "loss": 1.096, "step": 2750 }, { "epoch": 1.06, "learning_rate": 9.726551439143009e-06, "loss": 1.0959, "step": 2751 }, { "epoch": 1.06, "learning_rate": 9.72026688706474e-06, "loss": 1.0367, "step": 2752 }, { "epoch": 1.06, "learning_rate": 9.71398244555331e-06, "loss": 1.1138, "step": 2753 }, { "epoch": 1.06, "learning_rate": 9.707698117092704e-06, "loss": 1.1363, "step": 2754 }, { "epoch": 1.06, "learning_rate": 9.701413904166852e-06, "loss": 1.1036, "step": 2755 }, { "epoch": 1.06, "learning_rate": 9.695129809259644e-06, "loss": 1.0677, "step": 2756 }, { "epoch": 1.06, "learning_rate": 9.688845834854912e-06, "loss": 1.0597, "step": 2757 }, { "epoch": 1.06, "learning_rate": 9.682561983436463e-06, "loss": 1.1071, "step": 2758 }, { "epoch": 1.06, "learning_rate": 9.67627825748803e-06, "loss": 1.1057, "step": 2759 }, { "epoch": 1.06, "learning_rate": 9.669994659493316e-06, "loss": 1.0929, "step": 2760 }, { "epoch": 1.06, "learning_rate": 9.663711191935957e-06, "loss": 1.1414, "step": 2761 }, { "epoch": 1.06, "learning_rate": 9.657427857299556e-06, "loss": 1.1224, "step": 2762 }, { "epoch": 1.06, "learning_rate": 9.651144658067649e-06, "loss": 1.0232, "step": 2763 }, { "epoch": 1.06, "learning_rate": 9.644861596723718e-06, "loss": 1.1547, "step": 2764 }, { "epoch": 1.06, "learning_rate": 9.638578675751204e-06, "loss": 1.0924, "step": 2765 }, { "epoch": 1.06, "learning_rate": 9.632295897633473e-06, "loss": 1.0658, "step": 2766 }, { "epoch": 1.06, "learning_rate": 9.626013264853856e-06, "loss": 1.0919, "step": 2767 }, { "epoch": 1.06, "learning_rate": 9.61973077989561e-06, "loss": 1.1088, "step": 2768 }, { "epoch": 1.06, "learning_rate": 9.613448445241944e-06, "loss": 1.0861, "step": 2769 }, { "epoch": 1.06, "learning_rate": 9.607166263375995e-06, "loss": 1.1765, "step": 2770 }, { "epoch": 1.06, "learning_rate": 9.600884236780858e-06, "loss": 1.111, "step": 2771 }, { "epoch": 1.06, "learning_rate": 9.594602367939549e-06, "loss": 1.0734, "step": 2772 }, { "epoch": 1.07, "learning_rate": 9.588320659335034e-06, "loss": 1.1297, "step": 2773 }, { "epoch": 1.07, "learning_rate": 9.582039113450208e-06, "loss": 1.0812, "step": 2774 }, { "epoch": 1.07, "learning_rate": 9.5757577327679e-06, "loss": 1.0966, "step": 2775 }, { "epoch": 1.07, "learning_rate": 9.569476519770887e-06, "loss": 1.1168, "step": 2776 }, { "epoch": 1.07, "learning_rate": 9.563195476941862e-06, "loss": 1.1025, "step": 2777 }, { "epoch": 1.07, "learning_rate": 9.556914606763466e-06, "loss": 1.0411, "step": 2778 }, { "epoch": 1.07, "learning_rate": 9.55063391171826e-06, "loss": 1.0969, "step": 2779 }, { "epoch": 1.07, "learning_rate": 9.54435339428875e-06, "loss": 1.1171, "step": 2780 }, { "epoch": 1.07, "learning_rate": 9.53807305695735e-06, "loss": 1.1303, "step": 2781 }, { "epoch": 1.07, "learning_rate": 9.531792902206428e-06, "loss": 1.0959, "step": 2782 }, { "epoch": 1.07, "learning_rate": 9.525512932518258e-06, "loss": 1.1309, "step": 2783 }, { "epoch": 1.07, "learning_rate": 9.519233150375057e-06, "loss": 1.1032, "step": 2784 }, { "epoch": 1.07, "learning_rate": 9.512953558258955e-06, "loss": 0.9899, "step": 2785 }, { "epoch": 1.07, "learning_rate": 9.50667415865202e-06, "loss": 1.1097, "step": 2786 }, { "epoch": 1.07, "learning_rate": 9.500394954036234e-06, "loss": 1.0812, "step": 2787 }, { "epoch": 1.07, "learning_rate": 9.494115946893502e-06, "loss": 1.1271, "step": 2788 }, { "epoch": 1.07, "learning_rate": 9.487837139705657e-06, "loss": 1.0912, "step": 2789 }, { "epoch": 1.07, "learning_rate": 9.481558534954448e-06, "loss": 1.1144, "step": 2790 }, { "epoch": 1.07, "learning_rate": 9.47528013512155e-06, "loss": 1.0749, "step": 2791 }, { "epoch": 1.07, "learning_rate": 9.469001942688547e-06, "loss": 1.0871, "step": 2792 }, { "epoch": 1.07, "learning_rate": 9.462723960136953e-06, "loss": 1.0736, "step": 2793 }, { "epoch": 1.07, "learning_rate": 9.456446189948186e-06, "loss": 1.0333, "step": 2794 }, { "epoch": 1.07, "learning_rate": 9.450168634603593e-06, "loss": 1.0931, "step": 2795 }, { "epoch": 1.07, "learning_rate": 9.443891296584426e-06, "loss": 1.1225, "step": 2796 }, { "epoch": 1.07, "learning_rate": 9.43761417837186e-06, "loss": 1.0798, "step": 2797 }, { "epoch": 1.07, "learning_rate": 9.431337282446974e-06, "loss": 1.1067, "step": 2798 }, { "epoch": 1.08, "learning_rate": 9.42506061129076e-06, "loss": 1.0738, "step": 2799 }, { "epoch": 1.08, "learning_rate": 9.418784167384136e-06, "loss": 1.1082, "step": 2800 }, { "epoch": 1.08, "eval_loss": 1.113355278968811, "eval_runtime": 688.7565, "eval_samples_per_second": 4.938, "eval_steps_per_second": 0.309, "step": 2800 }, { "epoch": 1.08, "learning_rate": 9.412507953207906e-06, "loss": 1.1496, "step": 2801 }, { "epoch": 1.08, "learning_rate": 9.406231971242807e-06, "loss": 1.1217, "step": 2802 }, { "epoch": 1.08, "learning_rate": 9.399956223969461e-06, "loss": 1.1726, "step": 2803 }, { "epoch": 1.08, "learning_rate": 9.393680713868422e-06, "loss": 1.1234, "step": 2804 }, { "epoch": 1.08, "learning_rate": 9.38740544342013e-06, "loss": 1.1124, "step": 2805 }, { "epoch": 1.08, "learning_rate": 9.381130415104941e-06, "loss": 1.1164, "step": 2806 }, { "epoch": 1.08, "learning_rate": 9.374855631403108e-06, "loss": 1.1257, "step": 2807 }, { "epoch": 1.08, "learning_rate": 9.368581094794798e-06, "loss": 1.1002, "step": 2808 }, { "epoch": 1.08, "learning_rate": 9.362306807760068e-06, "loss": 1.1185, "step": 2809 }, { "epoch": 1.08, "learning_rate": 9.356032772778887e-06, "loss": 1.1196, "step": 2810 }, { "epoch": 1.08, "learning_rate": 9.349758992331119e-06, "loss": 1.0887, "step": 2811 }, { "epoch": 1.08, "learning_rate": 9.34348546889652e-06, "loss": 1.09, "step": 2812 }, { "epoch": 1.08, "learning_rate": 9.337212204954764e-06, "loss": 1.1089, "step": 2813 }, { "epoch": 1.08, "learning_rate": 9.330939202985402e-06, "loss": 1.0943, "step": 2814 }, { "epoch": 1.08, "learning_rate": 9.324666465467897e-06, "loss": 1.037, "step": 2815 }, { "epoch": 1.08, "learning_rate": 9.318393994881594e-06, "loss": 1.085, "step": 2816 }, { "epoch": 1.08, "learning_rate": 9.312121793705746e-06, "loss": 1.1411, "step": 2817 }, { "epoch": 1.08, "learning_rate": 9.305849864419488e-06, "loss": 1.1576, "step": 2818 }, { "epoch": 1.08, "learning_rate": 9.299578209501856e-06, "loss": 1.1097, "step": 2819 }, { "epoch": 1.08, "learning_rate": 9.293306831431767e-06, "loss": 1.1074, "step": 2820 }, { "epoch": 1.08, "learning_rate": 9.287035732688045e-06, "loss": 1.1175, "step": 2821 }, { "epoch": 1.08, "learning_rate": 9.280764915749389e-06, "loss": 1.1279, "step": 2822 }, { "epoch": 1.08, "learning_rate": 9.274494383094389e-06, "loss": 1.1181, "step": 2823 }, { "epoch": 1.08, "learning_rate": 9.26822413720153e-06, "loss": 1.0505, "step": 2824 }, { "epoch": 1.09, "learning_rate": 9.261954180549174e-06, "loss": 1.0701, "step": 2825 }, { "epoch": 1.09, "learning_rate": 9.25568451561558e-06, "loss": 1.149, "step": 2826 }, { "epoch": 1.09, "learning_rate": 9.249415144878881e-06, "loss": 1.0987, "step": 2827 }, { "epoch": 1.09, "learning_rate": 9.2431460708171e-06, "loss": 1.0245, "step": 2828 }, { "epoch": 1.09, "learning_rate": 9.236877295908136e-06, "loss": 1.1234, "step": 2829 }, { "epoch": 1.09, "learning_rate": 9.23060882262978e-06, "loss": 1.098, "step": 2830 }, { "epoch": 1.09, "learning_rate": 9.224340653459698e-06, "loss": 1.0575, "step": 2831 }, { "epoch": 1.09, "learning_rate": 9.218072790875434e-06, "loss": 1.1161, "step": 2832 }, { "epoch": 1.09, "learning_rate": 9.211805237354411e-06, "loss": 1.054, "step": 2833 }, { "epoch": 1.09, "learning_rate": 9.205537995373937e-06, "loss": 1.0969, "step": 2834 }, { "epoch": 1.09, "learning_rate": 9.199271067411189e-06, "loss": 1.129, "step": 2835 }, { "epoch": 1.09, "learning_rate": 9.19300445594322e-06, "loss": 1.1311, "step": 2836 }, { "epoch": 1.09, "learning_rate": 9.186738163446966e-06, "loss": 1.1284, "step": 2837 }, { "epoch": 1.09, "learning_rate": 9.180472192399224e-06, "loss": 1.1349, "step": 2838 }, { "epoch": 1.09, "learning_rate": 9.174206545276678e-06, "loss": 1.1521, "step": 2839 }, { "epoch": 1.09, "learning_rate": 9.167941224555873e-06, "loss": 1.0537, "step": 2840 }, { "epoch": 1.09, "learning_rate": 9.161676232713232e-06, "loss": 1.1064, "step": 2841 }, { "epoch": 1.09, "learning_rate": 9.155411572225041e-06, "loss": 1.0175, "step": 2842 }, { "epoch": 1.09, "learning_rate": 9.149147245567466e-06, "loss": 1.0946, "step": 2843 }, { "epoch": 1.09, "learning_rate": 9.142883255216526e-06, "loss": 1.1629, "step": 2844 }, { "epoch": 1.09, "learning_rate": 9.136619603648125e-06, "loss": 1.1394, "step": 2845 }, { "epoch": 1.09, "learning_rate": 9.130356293338017e-06, "loss": 1.0688, "step": 2846 }, { "epoch": 1.09, "learning_rate": 9.124093326761827e-06, "loss": 1.1879, "step": 2847 }, { "epoch": 1.09, "learning_rate": 9.117830706395049e-06, "loss": 1.1256, "step": 2848 }, { "epoch": 1.09, "learning_rate": 9.111568434713028e-06, "loss": 1.1716, "step": 2849 }, { "epoch": 1.09, "learning_rate": 9.10530651419099e-06, "loss": 1.1172, "step": 2850 }, { "epoch": 1.1, "learning_rate": 9.099044947304004e-06, "loss": 1.0807, "step": 2851 }, { "epoch": 1.1, "learning_rate": 9.092783736527009e-06, "loss": 1.1207, "step": 2852 }, { "epoch": 1.1, "learning_rate": 9.086522884334795e-06, "loss": 1.0641, "step": 2853 }, { "epoch": 1.1, "learning_rate": 9.080262393202026e-06, "loss": 1.0302, "step": 2854 }, { "epoch": 1.1, "learning_rate": 9.074002265603207e-06, "loss": 1.0575, "step": 2855 }, { "epoch": 1.1, "learning_rate": 9.067742504012708e-06, "loss": 1.112, "step": 2856 }, { "epoch": 1.1, "learning_rate": 9.061483110904746e-06, "loss": 1.1037, "step": 2857 }, { "epoch": 1.1, "learning_rate": 9.055224088753408e-06, "loss": 1.1051, "step": 2858 }, { "epoch": 1.1, "learning_rate": 9.04896544003262e-06, "loss": 1.0875, "step": 2859 }, { "epoch": 1.1, "learning_rate": 9.042707167216162e-06, "loss": 1.1136, "step": 2860 }, { "epoch": 1.1, "learning_rate": 9.036449272777673e-06, "loss": 1.0818, "step": 2861 }, { "epoch": 1.1, "learning_rate": 9.030191759190632e-06, "loss": 1.1571, "step": 2862 }, { "epoch": 1.1, "learning_rate": 9.023934628928382e-06, "loss": 1.1099, "step": 2863 }, { "epoch": 1.1, "learning_rate": 9.017677884464098e-06, "loss": 1.0605, "step": 2864 }, { "epoch": 1.1, "learning_rate": 9.011421528270816e-06, "loss": 1.1177, "step": 2865 }, { "epoch": 1.1, "learning_rate": 9.005165562821407e-06, "loss": 1.1514, "step": 2866 }, { "epoch": 1.1, "learning_rate": 8.998909990588602e-06, "loss": 1.0955, "step": 2867 }, { "epoch": 1.1, "learning_rate": 8.992654814044958e-06, "loss": 1.1858, "step": 2868 }, { "epoch": 1.1, "learning_rate": 8.986400035662897e-06, "loss": 1.135, "step": 2869 }, { "epoch": 1.1, "learning_rate": 8.980145657914663e-06, "loss": 1.071, "step": 2870 }, { "epoch": 1.1, "learning_rate": 8.97389168327235e-06, "loss": 1.0743, "step": 2871 }, { "epoch": 1.1, "learning_rate": 8.967638114207901e-06, "loss": 1.0867, "step": 2872 }, { "epoch": 1.1, "learning_rate": 8.961384953193087e-06, "loss": 1.1063, "step": 2873 }, { "epoch": 1.1, "learning_rate": 8.955132202699525e-06, "loss": 1.0858, "step": 2874 }, { "epoch": 1.1, "learning_rate": 8.948879865198659e-06, "loss": 1.0644, "step": 2875 }, { "epoch": 1.1, "learning_rate": 8.942627943161788e-06, "loss": 1.1278, "step": 2876 }, { "epoch": 1.11, "learning_rate": 8.936376439060029e-06, "loss": 1.0487, "step": 2877 }, { "epoch": 1.11, "learning_rate": 8.930125355364344e-06, "loss": 1.0782, "step": 2878 }, { "epoch": 1.11, "learning_rate": 8.923874694545524e-06, "loss": 1.1356, "step": 2879 }, { "epoch": 1.11, "learning_rate": 8.917624459074198e-06, "loss": 1.0569, "step": 2880 }, { "epoch": 1.11, "learning_rate": 8.911374651420818e-06, "loss": 1.0456, "step": 2881 }, { "epoch": 1.11, "learning_rate": 8.905125274055681e-06, "loss": 1.1036, "step": 2882 }, { "epoch": 1.11, "learning_rate": 8.898876329448899e-06, "loss": 1.0821, "step": 2883 }, { "epoch": 1.11, "learning_rate": 8.892627820070417e-06, "loss": 1.1201, "step": 2884 }, { "epoch": 1.11, "learning_rate": 8.886379748390019e-06, "loss": 1.131, "step": 2885 }, { "epoch": 1.11, "learning_rate": 8.8801321168773e-06, "loss": 1.1164, "step": 2886 }, { "epoch": 1.11, "learning_rate": 8.873884928001693e-06, "loss": 1.1145, "step": 2887 }, { "epoch": 1.11, "learning_rate": 8.867638184232446e-06, "loss": 1.0896, "step": 2888 }, { "epoch": 1.11, "learning_rate": 8.861391888038646e-06, "loss": 1.1047, "step": 2889 }, { "epoch": 1.11, "learning_rate": 8.855146041889183e-06, "loss": 1.0716, "step": 2890 }, { "epoch": 1.11, "learning_rate": 8.848900648252788e-06, "loss": 1.0293, "step": 2891 }, { "epoch": 1.11, "learning_rate": 8.842655709597995e-06, "loss": 1.129, "step": 2892 }, { "epoch": 1.11, "learning_rate": 8.83641122839318e-06, "loss": 1.1172, "step": 2893 }, { "epoch": 1.11, "learning_rate": 8.83016720710652e-06, "loss": 1.1289, "step": 2894 }, { "epoch": 1.11, "learning_rate": 8.823923648206017e-06, "loss": 1.062, "step": 2895 }, { "epoch": 1.11, "learning_rate": 8.817680554159489e-06, "loss": 1.1359, "step": 2896 }, { "epoch": 1.11, "learning_rate": 8.811437927434568e-06, "loss": 1.0577, "step": 2897 }, { "epoch": 1.11, "learning_rate": 8.805195770498714e-06, "loss": 1.0859, "step": 2898 }, { "epoch": 1.11, "learning_rate": 8.79895408581918e-06, "loss": 1.0125, "step": 2899 }, { "epoch": 1.11, "learning_rate": 8.792712875863054e-06, "loss": 1.1266, "step": 2900 }, { "epoch": 1.11, "learning_rate": 8.786472143097216e-06, "loss": 1.1034, "step": 2901 }, { "epoch": 1.11, "learning_rate": 8.780231889988378e-06, "loss": 1.0971, "step": 2902 }, { "epoch": 1.12, "learning_rate": 8.773992119003043e-06, "loss": 1.1213, "step": 2903 }, { "epoch": 1.12, "learning_rate": 8.76775283260754e-06, "loss": 1.1205, "step": 2904 }, { "epoch": 1.12, "learning_rate": 8.761514033267992e-06, "loss": 1.0464, "step": 2905 }, { "epoch": 1.12, "learning_rate": 8.755275723450343e-06, "loss": 1.1515, "step": 2906 }, { "epoch": 1.12, "learning_rate": 8.749037905620334e-06, "loss": 1.1506, "step": 2907 }, { "epoch": 1.12, "learning_rate": 8.742800582243513e-06, "loss": 1.0847, "step": 2908 }, { "epoch": 1.12, "learning_rate": 8.736563755785236e-06, "loss": 1.108, "step": 2909 }, { "epoch": 1.12, "learning_rate": 8.730327428710656e-06, "loss": 1.0729, "step": 2910 }, { "epoch": 1.12, "learning_rate": 8.724091603484741e-06, "loss": 1.1154, "step": 2911 }, { "epoch": 1.12, "learning_rate": 8.717856282572247e-06, "loss": 1.0958, "step": 2912 }, { "epoch": 1.12, "learning_rate": 8.711621468437742e-06, "loss": 1.1012, "step": 2913 }, { "epoch": 1.12, "learning_rate": 8.70538716354558e-06, "loss": 1.032, "step": 2914 }, { "epoch": 1.12, "learning_rate": 8.699153370359933e-06, "loss": 1.0661, "step": 2915 }, { "epoch": 1.12, "learning_rate": 8.692920091344747e-06, "loss": 1.1444, "step": 2916 }, { "epoch": 1.12, "learning_rate": 8.686687328963787e-06, "loss": 1.128, "step": 2917 }, { "epoch": 1.12, "learning_rate": 8.680455085680601e-06, "loss": 1.0744, "step": 2918 }, { "epoch": 1.12, "learning_rate": 8.674223363958535e-06, "loss": 1.0677, "step": 2919 }, { "epoch": 1.12, "learning_rate": 8.66799216626073e-06, "loss": 1.1168, "step": 2920 }, { "epoch": 1.12, "learning_rate": 8.661761495050114e-06, "loss": 1.0779, "step": 2921 }, { "epoch": 1.12, "learning_rate": 8.65553135278942e-06, "loss": 1.0806, "step": 2922 }, { "epoch": 1.12, "learning_rate": 8.649301741941156e-06, "loss": 1.0712, "step": 2923 }, { "epoch": 1.12, "learning_rate": 8.643072664967634e-06, "loss": 1.0905, "step": 2924 }, { "epoch": 1.12, "learning_rate": 8.63684412433094e-06, "loss": 1.0943, "step": 2925 }, { "epoch": 1.12, "learning_rate": 8.630616122492967e-06, "loss": 1.1765, "step": 2926 }, { "epoch": 1.12, "learning_rate": 8.624388661915378e-06, "loss": 1.1072, "step": 2927 }, { "epoch": 1.12, "learning_rate": 8.618161745059632e-06, "loss": 1.094, "step": 2928 }, { "epoch": 1.13, "learning_rate": 8.611935374386964e-06, "loss": 1.0679, "step": 2929 }, { "epoch": 1.13, "learning_rate": 8.605709552358407e-06, "loss": 1.1031, "step": 2930 }, { "epoch": 1.13, "learning_rate": 8.599484281434765e-06, "loss": 1.0818, "step": 2931 }, { "epoch": 1.13, "learning_rate": 8.593259564076627e-06, "loss": 1.0202, "step": 2932 }, { "epoch": 1.13, "learning_rate": 8.587035402744365e-06, "loss": 1.103, "step": 2933 }, { "epoch": 1.13, "learning_rate": 8.580811799898128e-06, "loss": 1.1496, "step": 2934 }, { "epoch": 1.13, "learning_rate": 8.574588757997853e-06, "loss": 1.1419, "step": 2935 }, { "epoch": 1.13, "learning_rate": 8.568366279503247e-06, "loss": 1.1262, "step": 2936 }, { "epoch": 1.13, "learning_rate": 8.562144366873796e-06, "loss": 1.0781, "step": 2937 }, { "epoch": 1.13, "learning_rate": 8.555923022568758e-06, "loss": 1.1342, "step": 2938 }, { "epoch": 1.13, "learning_rate": 8.549702249047181e-06, "loss": 1.1014, "step": 2939 }, { "epoch": 1.13, "learning_rate": 8.543482048767869e-06, "loss": 1.1446, "step": 2940 }, { "epoch": 1.13, "learning_rate": 8.537262424189412e-06, "loss": 1.0888, "step": 2941 }, { "epoch": 1.13, "learning_rate": 8.53104337777017e-06, "loss": 1.1253, "step": 2942 }, { "epoch": 1.13, "learning_rate": 8.524824911968264e-06, "loss": 1.1094, "step": 2943 }, { "epoch": 1.13, "learning_rate": 8.518607029241606e-06, "loss": 1.1002, "step": 2944 }, { "epoch": 1.13, "learning_rate": 8.512389732047859e-06, "loss": 1.1267, "step": 2945 }, { "epoch": 1.13, "learning_rate": 8.506173022844466e-06, "loss": 1.1636, "step": 2946 }, { "epoch": 1.13, "learning_rate": 8.499956904088624e-06, "loss": 1.1058, "step": 2947 }, { "epoch": 1.13, "learning_rate": 8.49374137823732e-06, "loss": 1.1373, "step": 2948 }, { "epoch": 1.13, "learning_rate": 8.487526447747278e-06, "loss": 1.0775, "step": 2949 }, { "epoch": 1.13, "learning_rate": 8.481312115075014e-06, "loss": 1.0815, "step": 2950 }, { "epoch": 1.13, "learning_rate": 8.47509838267678e-06, "loss": 1.1575, "step": 2951 }, { "epoch": 1.13, "learning_rate": 8.468885253008621e-06, "loss": 1.0868, "step": 2952 }, { "epoch": 1.13, "learning_rate": 8.462672728526317e-06, "loss": 1.1045, "step": 2953 }, { "epoch": 1.13, "learning_rate": 8.456460811685427e-06, "loss": 1.0372, "step": 2954 }, { "epoch": 1.14, "learning_rate": 8.450249504941261e-06, "loss": 1.0355, "step": 2955 }, { "epoch": 1.14, "learning_rate": 8.444038810748883e-06, "loss": 1.0968, "step": 2956 }, { "epoch": 1.14, "learning_rate": 8.437828731563133e-06, "loss": 1.0537, "step": 2957 }, { "epoch": 1.14, "learning_rate": 8.431619269838587e-06, "loss": 1.059, "step": 2958 }, { "epoch": 1.14, "learning_rate": 8.425410428029595e-06, "loss": 1.0605, "step": 2959 }, { "epoch": 1.14, "learning_rate": 8.419202208590243e-06, "loss": 1.1036, "step": 2960 }, { "epoch": 1.14, "learning_rate": 8.41299461397439e-06, "loss": 1.1188, "step": 2961 }, { "epoch": 1.14, "learning_rate": 8.406787646635636e-06, "loss": 1.0795, "step": 2962 }, { "epoch": 1.14, "learning_rate": 8.400581309027342e-06, "loss": 1.0455, "step": 2963 }, { "epoch": 1.14, "learning_rate": 8.394375603602602e-06, "loss": 1.1082, "step": 2964 }, { "epoch": 1.14, "learning_rate": 8.388170532814286e-06, "loss": 1.0928, "step": 2965 }, { "epoch": 1.14, "learning_rate": 8.381966099114997e-06, "loss": 1.1337, "step": 2966 }, { "epoch": 1.14, "learning_rate": 8.375762304957082e-06, "loss": 1.1181, "step": 2967 }, { "epoch": 1.14, "learning_rate": 8.36955915279265e-06, "loss": 1.1084, "step": 2968 }, { "epoch": 1.14, "learning_rate": 8.363356645073542e-06, "loss": 1.1207, "step": 2969 }, { "epoch": 1.14, "learning_rate": 8.357154784251358e-06, "loss": 1.1668, "step": 2970 }, { "epoch": 1.14, "learning_rate": 8.35095357277743e-06, "loss": 1.1082, "step": 2971 }, { "epoch": 1.14, "learning_rate": 8.34475301310284e-06, "loss": 1.0533, "step": 2972 }, { "epoch": 1.14, "learning_rate": 8.33855310767841e-06, "loss": 1.101, "step": 2973 }, { "epoch": 1.14, "learning_rate": 8.332353858954707e-06, "loss": 1.1131, "step": 2974 }, { "epoch": 1.14, "learning_rate": 8.326155269382034e-06, "loss": 1.2009, "step": 2975 }, { "epoch": 1.14, "learning_rate": 8.319957341410435e-06, "loss": 1.1014, "step": 2976 }, { "epoch": 1.14, "learning_rate": 8.31376007748969e-06, "loss": 1.0725, "step": 2977 }, { "epoch": 1.14, "learning_rate": 8.307563480069327e-06, "loss": 1.1013, "step": 2978 }, { "epoch": 1.14, "learning_rate": 8.301367551598595e-06, "loss": 1.0772, "step": 2979 }, { "epoch": 1.14, "learning_rate": 8.29517229452649e-06, "loss": 1.115, "step": 2980 }, { "epoch": 1.15, "learning_rate": 8.288977711301738e-06, "loss": 1.0963, "step": 2981 }, { "epoch": 1.15, "learning_rate": 8.282783804372797e-06, "loss": 1.0883, "step": 2982 }, { "epoch": 1.15, "learning_rate": 8.27659057618787e-06, "loss": 1.133, "step": 2983 }, { "epoch": 1.15, "learning_rate": 8.270398029194867e-06, "loss": 1.0747, "step": 2984 }, { "epoch": 1.15, "learning_rate": 8.264206165841456e-06, "loss": 1.0992, "step": 2985 }, { "epoch": 1.15, "learning_rate": 8.258014988575015e-06, "loss": 1.094, "step": 2986 }, { "epoch": 1.15, "learning_rate": 8.251824499842669e-06, "loss": 1.0458, "step": 2987 }, { "epoch": 1.15, "learning_rate": 8.245634702091243e-06, "loss": 1.0881, "step": 2988 }, { "epoch": 1.15, "learning_rate": 8.239445597767323e-06, "loss": 1.1351, "step": 2989 }, { "epoch": 1.15, "learning_rate": 8.233257189317199e-06, "loss": 1.0823, "step": 2990 }, { "epoch": 1.15, "learning_rate": 8.227069479186884e-06, "loss": 1.189, "step": 2991 }, { "epoch": 1.15, "learning_rate": 8.220882469822132e-06, "loss": 1.0893, "step": 2992 }, { "epoch": 1.15, "learning_rate": 8.214696163668399e-06, "loss": 1.0554, "step": 2993 }, { "epoch": 1.15, "learning_rate": 8.208510563170885e-06, "loss": 1.1151, "step": 2994 }, { "epoch": 1.15, "learning_rate": 8.202325670774494e-06, "loss": 1.121, "step": 2995 }, { "epoch": 1.15, "learning_rate": 8.196141488923858e-06, "loss": 1.1133, "step": 2996 }, { "epoch": 1.15, "learning_rate": 8.189958020063322e-06, "loss": 1.116, "step": 2997 }, { "epoch": 1.15, "learning_rate": 8.183775266636962e-06, "loss": 1.0402, "step": 2998 }, { "epoch": 1.15, "learning_rate": 8.177593231088557e-06, "loss": 1.0978, "step": 2999 }, { "epoch": 1.15, "learning_rate": 8.171411915861612e-06, "loss": 1.1024, "step": 3000 }, { "epoch": 1.15, "learning_rate": 8.165231323399337e-06, "loss": 1.0697, "step": 3001 }, { "epoch": 1.15, "learning_rate": 8.15905145614467e-06, "loss": 1.1313, "step": 3002 }, { "epoch": 1.15, "learning_rate": 8.152872316540249e-06, "loss": 1.2046, "step": 3003 }, { "epoch": 1.15, "learning_rate": 8.146693907028432e-06, "loss": 1.0727, "step": 3004 }, { "epoch": 1.15, "learning_rate": 8.14051623005129e-06, "loss": 1.1489, "step": 3005 }, { "epoch": 1.15, "learning_rate": 8.134339288050597e-06, "loss": 1.0924, "step": 3006 }, { "epoch": 1.16, "learning_rate": 8.128163083467843e-06, "loss": 1.0816, "step": 3007 }, { "epoch": 1.16, "learning_rate": 8.121987618744224e-06, "loss": 1.0966, "step": 3008 }, { "epoch": 1.16, "learning_rate": 8.115812896320647e-06, "loss": 1.1317, "step": 3009 }, { "epoch": 1.16, "learning_rate": 8.109638918637714e-06, "loss": 1.1145, "step": 3010 }, { "epoch": 1.16, "learning_rate": 8.10346568813575e-06, "loss": 1.1257, "step": 3011 }, { "epoch": 1.16, "learning_rate": 8.097293207254773e-06, "loss": 1.0969, "step": 3012 }, { "epoch": 1.16, "learning_rate": 8.091121478434507e-06, "loss": 1.0623, "step": 3013 }, { "epoch": 1.16, "learning_rate": 8.084950504114379e-06, "loss": 1.1691, "step": 3014 }, { "epoch": 1.16, "learning_rate": 8.078780286733512e-06, "loss": 1.0928, "step": 3015 }, { "epoch": 1.16, "learning_rate": 8.07261082873075e-06, "loss": 1.1503, "step": 3016 }, { "epoch": 1.16, "learning_rate": 8.06644213254461e-06, "loss": 1.0535, "step": 3017 }, { "epoch": 1.16, "learning_rate": 8.060274200613327e-06, "loss": 1.1439, "step": 3018 }, { "epoch": 1.16, "learning_rate": 8.054107035374822e-06, "loss": 1.1115, "step": 3019 }, { "epoch": 1.16, "learning_rate": 8.047940639266727e-06, "loss": 1.1163, "step": 3020 }, { "epoch": 1.16, "learning_rate": 8.041775014726354e-06, "loss": 1.1541, "step": 3021 }, { "epoch": 1.16, "learning_rate": 8.03561016419072e-06, "loss": 1.1043, "step": 3022 }, { "epoch": 1.16, "learning_rate": 8.02944609009653e-06, "loss": 1.1279, "step": 3023 }, { "epoch": 1.16, "learning_rate": 8.023282794880194e-06, "loss": 1.1207, "step": 3024 }, { "epoch": 1.16, "learning_rate": 8.0171202809778e-06, "loss": 1.0999, "step": 3025 }, { "epoch": 1.16, "learning_rate": 8.010958550825134e-06, "loss": 1.1671, "step": 3026 }, { "epoch": 1.16, "learning_rate": 8.004797606857673e-06, "loss": 1.0732, "step": 3027 }, { "epoch": 1.16, "learning_rate": 7.998637451510573e-06, "loss": 1.0925, "step": 3028 }, { "epoch": 1.16, "learning_rate": 7.992478087218698e-06, "loss": 1.131, "step": 3029 }, { "epoch": 1.16, "learning_rate": 7.986319516416581e-06, "loss": 1.1116, "step": 3030 }, { "epoch": 1.16, "learning_rate": 7.980161741538453e-06, "loss": 1.1045, "step": 3031 }, { "epoch": 1.16, "learning_rate": 7.97400476501822e-06, "loss": 1.1089, "step": 3032 }, { "epoch": 1.17, "learning_rate": 7.967848589289485e-06, "loss": 1.0754, "step": 3033 }, { "epoch": 1.17, "learning_rate": 7.961693216785523e-06, "loss": 1.0996, "step": 3034 }, { "epoch": 1.17, "learning_rate": 7.955538649939298e-06, "loss": 1.1851, "step": 3035 }, { "epoch": 1.17, "learning_rate": 7.94938489118345e-06, "loss": 1.1178, "step": 3036 }, { "epoch": 1.17, "learning_rate": 7.943231942950311e-06, "loss": 1.1238, "step": 3037 }, { "epoch": 1.17, "learning_rate": 7.937079807671882e-06, "loss": 1.1006, "step": 3038 }, { "epoch": 1.17, "learning_rate": 7.930928487779842e-06, "loss": 1.0676, "step": 3039 }, { "epoch": 1.17, "learning_rate": 7.924777985705556e-06, "loss": 1.075, "step": 3040 }, { "epoch": 1.17, "learning_rate": 7.918628303880057e-06, "loss": 1.1241, "step": 3041 }, { "epoch": 1.17, "learning_rate": 7.912479444734063e-06, "loss": 1.0521, "step": 3042 }, { "epoch": 1.17, "learning_rate": 7.906331410697955e-06, "loss": 1.0953, "step": 3043 }, { "epoch": 1.17, "learning_rate": 7.900184204201804e-06, "loss": 1.1016, "step": 3044 }, { "epoch": 1.17, "learning_rate": 7.894037827675333e-06, "loss": 1.0878, "step": 3045 }, { "epoch": 1.17, "learning_rate": 7.88789228354796e-06, "loss": 1.0666, "step": 3046 }, { "epoch": 1.17, "learning_rate": 7.881747574248754e-06, "loss": 1.1496, "step": 3047 }, { "epoch": 1.17, "learning_rate": 7.87560370220647e-06, "loss": 1.092, "step": 3048 }, { "epoch": 1.17, "learning_rate": 7.869460669849515e-06, "loss": 1.0823, "step": 3049 }, { "epoch": 1.17, "learning_rate": 7.863318479605983e-06, "loss": 1.0945, "step": 3050 }, { "epoch": 1.17, "learning_rate": 7.857177133903626e-06, "loss": 1.0955, "step": 3051 }, { "epoch": 1.17, "learning_rate": 7.851036635169851e-06, "loss": 1.1317, "step": 3052 }, { "epoch": 1.17, "learning_rate": 7.844896985831753e-06, "loss": 1.1219, "step": 3053 }, { "epoch": 1.17, "learning_rate": 7.838758188316068e-06, "loss": 1.0572, "step": 3054 }, { "epoch": 1.17, "learning_rate": 7.832620245049223e-06, "loss": 1.0479, "step": 3055 }, { "epoch": 1.17, "learning_rate": 7.826483158457275e-06, "loss": 1.1091, "step": 3056 }, { "epoch": 1.17, "learning_rate": 7.82034693096597e-06, "loss": 1.0683, "step": 3057 }, { "epoch": 1.17, "learning_rate": 7.814211565000694e-06, "loss": 1.0408, "step": 3058 }, { "epoch": 1.18, "learning_rate": 7.808077062986515e-06, "loss": 1.0614, "step": 3059 }, { "epoch": 1.18, "learning_rate": 7.801943427348127e-06, "loss": 1.1008, "step": 3060 }, { "epoch": 1.18, "learning_rate": 7.795810660509917e-06, "loss": 1.1127, "step": 3061 }, { "epoch": 1.18, "learning_rate": 7.789678764895906e-06, "loss": 1.0737, "step": 3062 }, { "epoch": 1.18, "learning_rate": 7.783547742929774e-06, "loss": 1.0947, "step": 3063 }, { "epoch": 1.18, "learning_rate": 7.777417597034864e-06, "loss": 1.0947, "step": 3064 }, { "epoch": 1.18, "learning_rate": 7.77128832963416e-06, "loss": 1.071, "step": 3065 }, { "epoch": 1.18, "learning_rate": 7.765159943150315e-06, "loss": 1.1166, "step": 3066 }, { "epoch": 1.18, "learning_rate": 7.759032440005619e-06, "loss": 1.1289, "step": 3067 }, { "epoch": 1.18, "learning_rate": 7.75290582262202e-06, "loss": 1.1238, "step": 3068 }, { "epoch": 1.18, "learning_rate": 7.74678009342111e-06, "loss": 1.0735, "step": 3069 }, { "epoch": 1.18, "learning_rate": 7.740655254824147e-06, "loss": 1.1366, "step": 3070 }, { "epoch": 1.18, "learning_rate": 7.734531309252012e-06, "loss": 1.1487, "step": 3071 }, { "epoch": 1.18, "learning_rate": 7.728408259125252e-06, "loss": 1.0393, "step": 3072 }, { "epoch": 1.18, "learning_rate": 7.722286106864052e-06, "loss": 1.0862, "step": 3073 }, { "epoch": 1.18, "learning_rate": 7.71616485488824e-06, "loss": 1.1078, "step": 3074 }, { "epoch": 1.18, "learning_rate": 7.710044505617298e-06, "loss": 1.0614, "step": 3075 }, { "epoch": 1.18, "learning_rate": 7.70392506147034e-06, "loss": 1.1064, "step": 3076 }, { "epoch": 1.18, "learning_rate": 7.697806524866133e-06, "loss": 1.0662, "step": 3077 }, { "epoch": 1.18, "learning_rate": 7.69168889822307e-06, "loss": 1.0747, "step": 3078 }, { "epoch": 1.18, "learning_rate": 7.685572183959207e-06, "loss": 1.1528, "step": 3079 }, { "epoch": 1.18, "learning_rate": 7.679456384492213e-06, "loss": 1.1031, "step": 3080 }, { "epoch": 1.18, "learning_rate": 7.673341502239421e-06, "loss": 1.0996, "step": 3081 }, { "epoch": 1.18, "learning_rate": 7.667227539617776e-06, "loss": 1.0677, "step": 3082 }, { "epoch": 1.18, "learning_rate": 7.661114499043886e-06, "loss": 1.0672, "step": 3083 }, { "epoch": 1.18, "learning_rate": 7.65500238293397e-06, "loss": 1.1222, "step": 3084 }, { "epoch": 1.19, "learning_rate": 7.648891193703902e-06, "loss": 1.0929, "step": 3085 }, { "epoch": 1.19, "learning_rate": 7.642780933769176e-06, "loss": 1.097, "step": 3086 }, { "epoch": 1.19, "learning_rate": 7.63667160554492e-06, "loss": 1.0944, "step": 3087 }, { "epoch": 1.19, "learning_rate": 7.630563211445903e-06, "loss": 1.113, "step": 3088 }, { "epoch": 1.19, "learning_rate": 7.6244557538865135e-06, "loss": 1.0831, "step": 3089 }, { "epoch": 1.19, "learning_rate": 7.61834923528078e-06, "loss": 1.0955, "step": 3090 }, { "epoch": 1.19, "learning_rate": 7.612243658042349e-06, "loss": 1.0566, "step": 3091 }, { "epoch": 1.19, "learning_rate": 7.606139024584507e-06, "loss": 1.1176, "step": 3092 }, { "epoch": 1.19, "learning_rate": 7.600035337320155e-06, "loss": 1.0467, "step": 3093 }, { "epoch": 1.19, "learning_rate": 7.593932598661833e-06, "loss": 1.0831, "step": 3094 }, { "epoch": 1.19, "learning_rate": 7.587830811021691e-06, "loss": 1.1132, "step": 3095 }, { "epoch": 1.19, "learning_rate": 7.58172997681152e-06, "loss": 1.1096, "step": 3096 }, { "epoch": 1.19, "learning_rate": 7.5756300984427234e-06, "loss": 1.0813, "step": 3097 }, { "epoch": 1.19, "learning_rate": 7.569531178326324e-06, "loss": 1.0153, "step": 3098 }, { "epoch": 1.19, "learning_rate": 7.563433218872975e-06, "loss": 1.0775, "step": 3099 }, { "epoch": 1.19, "learning_rate": 7.557336222492942e-06, "loss": 1.085, "step": 3100 }, { "epoch": 1.19, "learning_rate": 7.551240191596119e-06, "loss": 1.1395, "step": 3101 }, { "epoch": 1.19, "learning_rate": 7.545145128592009e-06, "loss": 1.134, "step": 3102 }, { "epoch": 1.19, "learning_rate": 7.53905103588974e-06, "loss": 1.1166, "step": 3103 }, { "epoch": 1.19, "learning_rate": 7.532957915898047e-06, "loss": 1.1299, "step": 3104 }, { "epoch": 1.19, "learning_rate": 7.526865771025294e-06, "loss": 1.055, "step": 3105 }, { "epoch": 1.19, "learning_rate": 7.520774603679444e-06, "loss": 1.143, "step": 3106 }, { "epoch": 1.19, "learning_rate": 7.51468441626809e-06, "loss": 1.0913, "step": 3107 }, { "epoch": 1.19, "learning_rate": 7.508595211198419e-06, "loss": 1.0429, "step": 3108 }, { "epoch": 1.19, "learning_rate": 7.502506990877252e-06, "loss": 1.1429, "step": 3109 }, { "epoch": 1.19, "learning_rate": 7.496419757711002e-06, "loss": 1.1555, "step": 3110 }, { "epoch": 1.2, "learning_rate": 7.4903335141056965e-06, "loss": 1.0876, "step": 3111 }, { "epoch": 1.2, "learning_rate": 7.48424826246698e-06, "loss": 1.1138, "step": 3112 }, { "epoch": 1.2, "learning_rate": 7.4781640052000925e-06, "loss": 1.1046, "step": 3113 }, { "epoch": 1.2, "learning_rate": 7.472080744709893e-06, "loss": 1.1656, "step": 3114 }, { "epoch": 1.2, "learning_rate": 7.465998483400838e-06, "loss": 1.1008, "step": 3115 }, { "epoch": 1.2, "learning_rate": 7.4599172236769935e-06, "loss": 1.1094, "step": 3116 }, { "epoch": 1.2, "learning_rate": 7.453836967942024e-06, "loss": 1.098, "step": 3117 }, { "epoch": 1.2, "learning_rate": 7.447757718599209e-06, "loss": 1.1053, "step": 3118 }, { "epoch": 1.2, "learning_rate": 7.441679478051414e-06, "loss": 1.0704, "step": 3119 }, { "epoch": 1.2, "learning_rate": 7.4356022487011215e-06, "loss": 1.0914, "step": 3120 }, { "epoch": 1.2, "learning_rate": 7.4295260329504025e-06, "loss": 1.0713, "step": 3121 }, { "epoch": 1.2, "learning_rate": 7.423450833200929e-06, "loss": 1.0984, "step": 3122 }, { "epoch": 1.2, "learning_rate": 7.417376651853985e-06, "loss": 1.1244, "step": 3123 }, { "epoch": 1.2, "learning_rate": 7.411303491310428e-06, "loss": 1.1428, "step": 3124 }, { "epoch": 1.2, "learning_rate": 7.4052313539707366e-06, "loss": 1.0422, "step": 3125 }, { "epoch": 1.2, "learning_rate": 7.399160242234966e-06, "loss": 1.1042, "step": 3126 }, { "epoch": 1.2, "learning_rate": 7.393090158502778e-06, "loss": 1.0963, "step": 3127 }, { "epoch": 1.2, "learning_rate": 7.387021105173419e-06, "loss": 1.0952, "step": 3128 }, { "epoch": 1.2, "learning_rate": 7.380953084645741e-06, "loss": 1.1681, "step": 3129 }, { "epoch": 1.2, "learning_rate": 7.374886099318173e-06, "loss": 1.141, "step": 3130 }, { "epoch": 1.2, "learning_rate": 7.368820151588743e-06, "loss": 1.0524, "step": 3131 }, { "epoch": 1.2, "learning_rate": 7.362755243855067e-06, "loss": 1.1302, "step": 3132 }, { "epoch": 1.2, "learning_rate": 7.356691378514353e-06, "loss": 1.1005, "step": 3133 }, { "epoch": 1.2, "learning_rate": 7.350628557963393e-06, "loss": 1.0987, "step": 3134 }, { "epoch": 1.2, "learning_rate": 7.344566784598565e-06, "loss": 1.0864, "step": 3135 }, { "epoch": 1.2, "learning_rate": 7.338506060815837e-06, "loss": 1.1055, "step": 3136 }, { "epoch": 1.21, "learning_rate": 7.332446389010759e-06, "loss": 1.1238, "step": 3137 }, { "epoch": 1.21, "learning_rate": 7.326387771578472e-06, "loss": 1.1187, "step": 3138 }, { "epoch": 1.21, "learning_rate": 7.3203302109136895e-06, "loss": 1.0928, "step": 3139 }, { "epoch": 1.21, "learning_rate": 7.314273709410718e-06, "loss": 1.1071, "step": 3140 }, { "epoch": 1.21, "learning_rate": 7.308218269463429e-06, "loss": 1.1758, "step": 3141 }, { "epoch": 1.21, "learning_rate": 7.302163893465302e-06, "loss": 1.0873, "step": 3142 }, { "epoch": 1.21, "learning_rate": 7.296110583809366e-06, "loss": 1.0885, "step": 3143 }, { "epoch": 1.21, "learning_rate": 7.290058342888251e-06, "loss": 1.1163, "step": 3144 }, { "epoch": 1.21, "learning_rate": 7.284007173094152e-06, "loss": 1.1163, "step": 3145 }, { "epoch": 1.21, "learning_rate": 7.277957076818838e-06, "loss": 1.1284, "step": 3146 }, { "epoch": 1.21, "learning_rate": 7.271908056453672e-06, "loss": 1.152, "step": 3147 }, { "epoch": 1.21, "learning_rate": 7.265860114389571e-06, "loss": 1.1269, "step": 3148 }, { "epoch": 1.21, "learning_rate": 7.259813253017038e-06, "loss": 1.1415, "step": 3149 }, { "epoch": 1.21, "learning_rate": 7.253767474726143e-06, "loss": 1.1058, "step": 3150 }, { "epoch": 1.21, "learning_rate": 7.247722781906534e-06, "loss": 1.1377, "step": 3151 }, { "epoch": 1.21, "learning_rate": 7.241679176947423e-06, "loss": 1.1368, "step": 3152 }, { "epoch": 1.21, "learning_rate": 7.235636662237597e-06, "loss": 1.0281, "step": 3153 }, { "epoch": 1.21, "learning_rate": 7.229595240165406e-06, "loss": 1.1043, "step": 3154 }, { "epoch": 1.21, "learning_rate": 7.22355491311878e-06, "loss": 1.1009, "step": 3155 }, { "epoch": 1.21, "learning_rate": 7.2175156834852035e-06, "loss": 1.0325, "step": 3156 }, { "epoch": 1.21, "learning_rate": 7.211477553651733e-06, "loss": 1.0872, "step": 3157 }, { "epoch": 1.21, "learning_rate": 7.205440526004992e-06, "loss": 1.0442, "step": 3158 }, { "epoch": 1.21, "learning_rate": 7.199404602931159e-06, "loss": 1.0798, "step": 3159 }, { "epoch": 1.21, "learning_rate": 7.193369786815992e-06, "loss": 1.0805, "step": 3160 }, { "epoch": 1.21, "learning_rate": 7.1873360800447956e-06, "loss": 1.0969, "step": 3161 }, { "epoch": 1.21, "learning_rate": 7.181303485002446e-06, "loss": 1.0429, "step": 3162 }, { "epoch": 1.22, "learning_rate": 7.175272004073374e-06, "loss": 1.1053, "step": 3163 }, { "epoch": 1.22, "learning_rate": 7.169241639641577e-06, "loss": 1.1146, "step": 3164 }, { "epoch": 1.22, "learning_rate": 7.163212394090603e-06, "loss": 1.1276, "step": 3165 }, { "epoch": 1.22, "learning_rate": 7.157184269803565e-06, "loss": 1.0728, "step": 3166 }, { "epoch": 1.22, "learning_rate": 7.151157269163123e-06, "loss": 1.0951, "step": 3167 }, { "epoch": 1.22, "learning_rate": 7.145131394551506e-06, "loss": 1.0582, "step": 3168 }, { "epoch": 1.22, "learning_rate": 7.13910664835049e-06, "loss": 1.0716, "step": 3169 }, { "epoch": 1.22, "learning_rate": 7.133083032941402e-06, "loss": 1.1334, "step": 3170 }, { "epoch": 1.22, "learning_rate": 7.127060550705131e-06, "loss": 1.1013, "step": 3171 }, { "epoch": 1.22, "learning_rate": 7.121039204022106e-06, "loss": 1.0706, "step": 3172 }, { "epoch": 1.22, "learning_rate": 7.115018995272326e-06, "loss": 1.0283, "step": 3173 }, { "epoch": 1.22, "learning_rate": 7.1089999268353185e-06, "loss": 1.1148, "step": 3174 }, { "epoch": 1.22, "learning_rate": 7.102982001090176e-06, "loss": 1.1114, "step": 3175 }, { "epoch": 1.22, "learning_rate": 7.0969652204155286e-06, "loss": 1.0908, "step": 3176 }, { "epoch": 1.22, "learning_rate": 7.090949587189565e-06, "loss": 1.1129, "step": 3177 }, { "epoch": 1.22, "learning_rate": 7.08493510379001e-06, "loss": 1.1268, "step": 3178 }, { "epoch": 1.22, "learning_rate": 7.078921772594143e-06, "loss": 1.1297, "step": 3179 }, { "epoch": 1.22, "learning_rate": 7.072909595978775e-06, "loss": 1.0768, "step": 3180 }, { "epoch": 1.22, "learning_rate": 7.066898576320279e-06, "loss": 1.1445, "step": 3181 }, { "epoch": 1.22, "learning_rate": 7.060888715994555e-06, "loss": 1.1682, "step": 3182 }, { "epoch": 1.22, "learning_rate": 7.054880017377049e-06, "loss": 1.0823, "step": 3183 }, { "epoch": 1.22, "learning_rate": 7.048872482842753e-06, "loss": 1.0872, "step": 3184 }, { "epoch": 1.22, "learning_rate": 7.042866114766189e-06, "loss": 1.059, "step": 3185 }, { "epoch": 1.22, "learning_rate": 7.036860915521433e-06, "loss": 1.0984, "step": 3186 }, { "epoch": 1.22, "learning_rate": 7.030856887482084e-06, "loss": 1.116, "step": 3187 }, { "epoch": 1.22, "learning_rate": 7.024854033021287e-06, "loss": 1.1402, "step": 3188 }, { "epoch": 1.23, "learning_rate": 7.018852354511713e-06, "loss": 1.1246, "step": 3189 }, { "epoch": 1.23, "learning_rate": 7.012851854325586e-06, "loss": 1.0875, "step": 3190 }, { "epoch": 1.23, "learning_rate": 7.006852534834647e-06, "loss": 1.1008, "step": 3191 }, { "epoch": 1.23, "learning_rate": 7.000854398410182e-06, "loss": 1.055, "step": 3192 }, { "epoch": 1.23, "learning_rate": 6.994857447422999e-06, "loss": 1.1447, "step": 3193 }, { "epoch": 1.23, "learning_rate": 6.9888616842434445e-06, "loss": 1.0687, "step": 3194 }, { "epoch": 1.23, "learning_rate": 6.9828671112413946e-06, "loss": 1.0644, "step": 3195 }, { "epoch": 1.23, "learning_rate": 6.976873730786251e-06, "loss": 1.0961, "step": 3196 }, { "epoch": 1.23, "learning_rate": 6.970881545246955e-06, "loss": 1.0859, "step": 3197 }, { "epoch": 1.23, "learning_rate": 6.964890556991959e-06, "loss": 1.0913, "step": 3198 }, { "epoch": 1.23, "learning_rate": 6.958900768389258e-06, "loss": 1.0872, "step": 3199 }, { "epoch": 1.23, "learning_rate": 6.952912181806359e-06, "loss": 1.0806, "step": 3200 }, { "epoch": 1.23, "eval_loss": 1.1093899011611938, "eval_runtime": 688.5318, "eval_samples_per_second": 4.939, "eval_steps_per_second": 0.309, "step": 3200 }, { "epoch": 1.23, "learning_rate": 6.9469247996103095e-06, "loss": 1.1029, "step": 3201 }, { "epoch": 1.23, "learning_rate": 6.9409386241676645e-06, "loss": 1.1056, "step": 3202 }, { "epoch": 1.23, "learning_rate": 6.934953657844514e-06, "loss": 1.1002, "step": 3203 }, { "epoch": 1.23, "learning_rate": 6.928969903006462e-06, "loss": 1.0644, "step": 3204 }, { "epoch": 1.23, "learning_rate": 6.9229873620186425e-06, "loss": 1.0746, "step": 3205 }, { "epoch": 1.23, "learning_rate": 6.917006037245703e-06, "loss": 1.1815, "step": 3206 }, { "epoch": 1.23, "learning_rate": 6.911025931051808e-06, "loss": 1.176, "step": 3207 }, { "epoch": 1.23, "learning_rate": 6.905047045800649e-06, "loss": 1.0819, "step": 3208 }, { "epoch": 1.23, "learning_rate": 6.8990693838554255e-06, "loss": 1.1006, "step": 3209 }, { "epoch": 1.23, "learning_rate": 6.893092947578863e-06, "loss": 1.0572, "step": 3210 }, { "epoch": 1.23, "learning_rate": 6.887117739333195e-06, "loss": 1.1015, "step": 3211 }, { "epoch": 1.23, "learning_rate": 6.881143761480177e-06, "loss": 1.1006, "step": 3212 }, { "epoch": 1.23, "learning_rate": 6.875171016381064e-06, "loss": 1.0575, "step": 3213 }, { "epoch": 1.23, "learning_rate": 6.869199506396642e-06, "loss": 1.1435, "step": 3214 }, { "epoch": 1.23, "learning_rate": 6.863229233887196e-06, "loss": 1.0838, "step": 3215 }, { "epoch": 1.24, "learning_rate": 6.85726020121253e-06, "loss": 1.1, "step": 3216 }, { "epoch": 1.24, "learning_rate": 6.851292410731949e-06, "loss": 1.1072, "step": 3217 }, { "epoch": 1.24, "learning_rate": 6.845325864804271e-06, "loss": 1.0889, "step": 3218 }, { "epoch": 1.24, "learning_rate": 6.839360565787829e-06, "loss": 1.0867, "step": 3219 }, { "epoch": 1.24, "learning_rate": 6.833396516040453e-06, "loss": 1.0733, "step": 3220 }, { "epoch": 1.24, "learning_rate": 6.827433717919485e-06, "loss": 1.1649, "step": 3221 }, { "epoch": 1.24, "learning_rate": 6.821472173781766e-06, "loss": 1.0973, "step": 3222 }, { "epoch": 1.24, "learning_rate": 6.8155118859836535e-06, "loss": 1.0912, "step": 3223 }, { "epoch": 1.24, "learning_rate": 6.809552856880996e-06, "loss": 1.1404, "step": 3224 }, { "epoch": 1.24, "learning_rate": 6.803595088829152e-06, "loss": 1.1121, "step": 3225 }, { "epoch": 1.24, "learning_rate": 6.797638584182974e-06, "loss": 1.1007, "step": 3226 }, { "epoch": 1.24, "learning_rate": 6.791683345296826e-06, "loss": 1.0614, "step": 3227 }, { "epoch": 1.24, "learning_rate": 6.785729374524563e-06, "loss": 1.097, "step": 3228 }, { "epoch": 1.24, "learning_rate": 6.779776674219545e-06, "loss": 1.0911, "step": 3229 }, { "epoch": 1.24, "learning_rate": 6.773825246734622e-06, "loss": 1.0885, "step": 3230 }, { "epoch": 1.24, "learning_rate": 6.767875094422143e-06, "loss": 1.1287, "step": 3231 }, { "epoch": 1.24, "learning_rate": 6.761926219633963e-06, "loss": 1.0975, "step": 3232 }, { "epoch": 1.24, "learning_rate": 6.755978624721418e-06, "loss": 1.1183, "step": 3233 }, { "epoch": 1.24, "learning_rate": 6.7500323120353484e-06, "loss": 1.0932, "step": 3234 }, { "epoch": 1.24, "learning_rate": 6.74408728392608e-06, "loss": 1.1008, "step": 3235 }, { "epoch": 1.24, "learning_rate": 6.738143542743438e-06, "loss": 1.0803, "step": 3236 }, { "epoch": 1.24, "learning_rate": 6.732201090836734e-06, "loss": 1.1455, "step": 3237 }, { "epoch": 1.24, "learning_rate": 6.726259930554775e-06, "loss": 1.1275, "step": 3238 }, { "epoch": 1.24, "learning_rate": 6.720320064245844e-06, "loss": 1.1243, "step": 3239 }, { "epoch": 1.24, "learning_rate": 6.7143814942577365e-06, "loss": 1.1183, "step": 3240 }, { "epoch": 1.24, "learning_rate": 6.708444222937714e-06, "loss": 1.0803, "step": 3241 }, { "epoch": 1.25, "learning_rate": 6.7025082526325325e-06, "loss": 1.0942, "step": 3242 }, { "epoch": 1.25, "learning_rate": 6.696573585688435e-06, "loss": 1.0737, "step": 3243 }, { "epoch": 1.25, "learning_rate": 6.690640224451146e-06, "loss": 1.0701, "step": 3244 }, { "epoch": 1.25, "learning_rate": 6.68470817126588e-06, "loss": 1.0957, "step": 3245 }, { "epoch": 1.25, "learning_rate": 6.678777428477327e-06, "loss": 1.0812, "step": 3246 }, { "epoch": 1.25, "learning_rate": 6.672847998429665e-06, "loss": 1.1096, "step": 3247 }, { "epoch": 1.25, "learning_rate": 6.666919883466545e-06, "loss": 1.1362, "step": 3248 }, { "epoch": 1.25, "learning_rate": 6.660993085931113e-06, "loss": 1.0573, "step": 3249 }, { "epoch": 1.25, "learning_rate": 6.655067608165977e-06, "loss": 1.084, "step": 3250 }, { "epoch": 1.25, "learning_rate": 6.6491434525132385e-06, "loss": 1.1424, "step": 3251 }, { "epoch": 1.25, "learning_rate": 6.6432206213144605e-06, "loss": 1.0739, "step": 3252 }, { "epoch": 1.25, "learning_rate": 6.637299116910701e-06, "loss": 1.0447, "step": 3253 }, { "epoch": 1.25, "learning_rate": 6.6313789416424765e-06, "loss": 1.0723, "step": 3254 }, { "epoch": 1.25, "learning_rate": 6.62546009784979e-06, "loss": 1.061, "step": 3255 }, { "epoch": 1.25, "learning_rate": 6.619542587872112e-06, "loss": 1.0578, "step": 3256 }, { "epoch": 1.25, "learning_rate": 6.613626414048385e-06, "loss": 1.0843, "step": 3257 }, { "epoch": 1.25, "learning_rate": 6.6077115787170334e-06, "loss": 1.0517, "step": 3258 }, { "epoch": 1.25, "learning_rate": 6.601798084215938e-06, "loss": 1.1414, "step": 3259 }, { "epoch": 1.25, "learning_rate": 6.59588593288246e-06, "loss": 1.1089, "step": 3260 }, { "epoch": 1.25, "learning_rate": 6.589975127053423e-06, "loss": 1.1164, "step": 3261 }, { "epoch": 1.25, "learning_rate": 6.584065669065132e-06, "loss": 1.1333, "step": 3262 }, { "epoch": 1.25, "learning_rate": 6.578157561253335e-06, "loss": 1.0974, "step": 3263 }, { "epoch": 1.25, "learning_rate": 6.572250805953273e-06, "loss": 1.0794, "step": 3264 }, { "epoch": 1.25, "learning_rate": 6.566345405499636e-06, "loss": 1.0757, "step": 3265 }, { "epoch": 1.25, "learning_rate": 6.56044136222658e-06, "loss": 1.0366, "step": 3266 }, { "epoch": 1.25, "learning_rate": 6.554538678467732e-06, "loss": 1.1203, "step": 3267 }, { "epoch": 1.26, "learning_rate": 6.548637356556171e-06, "loss": 1.0995, "step": 3268 }, { "epoch": 1.26, "learning_rate": 6.542737398824451e-06, "loss": 1.1197, "step": 3269 }, { "epoch": 1.26, "learning_rate": 6.536838807604573e-06, "loss": 1.097, "step": 3270 }, { "epoch": 1.26, "learning_rate": 6.530941585228008e-06, "loss": 1.1066, "step": 3271 }, { "epoch": 1.26, "learning_rate": 6.525045734025679e-06, "loss": 1.1291, "step": 3272 }, { "epoch": 1.26, "learning_rate": 6.519151256327976e-06, "loss": 1.1349, "step": 3273 }, { "epoch": 1.26, "learning_rate": 6.513258154464735e-06, "loss": 1.1142, "step": 3274 }, { "epoch": 1.26, "learning_rate": 6.507366430765259e-06, "loss": 1.079, "step": 3275 }, { "epoch": 1.26, "learning_rate": 6.501476087558292e-06, "loss": 1.0467, "step": 3276 }, { "epoch": 1.26, "learning_rate": 6.495587127172049e-06, "loss": 1.0894, "step": 3277 }, { "epoch": 1.26, "learning_rate": 6.489699551934192e-06, "loss": 1.0943, "step": 3278 }, { "epoch": 1.26, "learning_rate": 6.483813364171827e-06, "loss": 1.1085, "step": 3279 }, { "epoch": 1.26, "learning_rate": 6.4779285662115235e-06, "loss": 1.1131, "step": 3280 }, { "epoch": 1.26, "learning_rate": 6.472045160379292e-06, "loss": 1.0969, "step": 3281 }, { "epoch": 1.26, "learning_rate": 6.466163149000606e-06, "loss": 1.0636, "step": 3282 }, { "epoch": 1.26, "learning_rate": 6.460282534400373e-06, "loss": 1.1111, "step": 3283 }, { "epoch": 1.26, "learning_rate": 6.454403318902957e-06, "loss": 1.186, "step": 3284 }, { "epoch": 1.26, "learning_rate": 6.448525504832164e-06, "loss": 1.1114, "step": 3285 }, { "epoch": 1.26, "learning_rate": 6.442649094511255e-06, "loss": 1.1271, "step": 3286 }, { "epoch": 1.26, "learning_rate": 6.436774090262925e-06, "loss": 1.1275, "step": 3287 }, { "epoch": 1.26, "learning_rate": 6.430900494409322e-06, "loss": 1.0424, "step": 3288 }, { "epoch": 1.26, "learning_rate": 6.4250283092720335e-06, "loss": 1.1642, "step": 3289 }, { "epoch": 1.26, "learning_rate": 6.419157537172082e-06, "loss": 1.1402, "step": 3290 }, { "epoch": 1.26, "learning_rate": 6.4132881804299505e-06, "loss": 1.1429, "step": 3291 }, { "epoch": 1.26, "learning_rate": 6.407420241365543e-06, "loss": 1.1613, "step": 3292 }, { "epoch": 1.26, "learning_rate": 6.401553722298215e-06, "loss": 1.1285, "step": 3293 }, { "epoch": 1.27, "learning_rate": 6.395688625546753e-06, "loss": 1.0315, "step": 3294 }, { "epoch": 1.27, "learning_rate": 6.389824953429392e-06, "loss": 1.1316, "step": 3295 }, { "epoch": 1.27, "learning_rate": 6.3839627082637915e-06, "loss": 1.1352, "step": 3296 }, { "epoch": 1.27, "learning_rate": 6.378101892367058e-06, "loss": 1.1165, "step": 3297 }, { "epoch": 1.27, "learning_rate": 6.37224250805572e-06, "loss": 1.0972, "step": 3298 }, { "epoch": 1.27, "learning_rate": 6.366384557645755e-06, "loss": 1.1131, "step": 3299 }, { "epoch": 1.27, "learning_rate": 6.360528043452565e-06, "loss": 1.0958, "step": 3300 }, { "epoch": 1.27, "learning_rate": 6.354672967790985e-06, "loss": 1.1567, "step": 3301 }, { "epoch": 1.27, "learning_rate": 6.348819332975284e-06, "loss": 1.0721, "step": 3302 }, { "epoch": 1.27, "learning_rate": 6.342967141319154e-06, "loss": 1.1105, "step": 3303 }, { "epoch": 1.27, "learning_rate": 6.337116395135732e-06, "loss": 1.0971, "step": 3304 }, { "epoch": 1.27, "learning_rate": 6.331267096737567e-06, "loss": 1.1752, "step": 3305 }, { "epoch": 1.27, "learning_rate": 6.325419248436648e-06, "loss": 1.091, "step": 3306 }, { "epoch": 1.27, "learning_rate": 6.319572852544378e-06, "loss": 1.0997, "step": 3307 }, { "epoch": 1.27, "learning_rate": 6.313727911371603e-06, "loss": 1.1014, "step": 3308 }, { "epoch": 1.27, "learning_rate": 6.307884427228579e-06, "loss": 1.1143, "step": 3309 }, { "epoch": 1.27, "learning_rate": 6.302042402424995e-06, "loss": 1.105, "step": 3310 }, { "epoch": 1.27, "learning_rate": 6.296201839269953e-06, "loss": 1.0926, "step": 3311 }, { "epoch": 1.27, "learning_rate": 6.2903627400719935e-06, "loss": 1.0761, "step": 3312 }, { "epoch": 1.27, "learning_rate": 6.284525107139067e-06, "loss": 1.1331, "step": 3313 }, { "epoch": 1.27, "learning_rate": 6.2786889427785415e-06, "loss": 1.137, "step": 3314 }, { "epoch": 1.27, "learning_rate": 6.272854249297215e-06, "loss": 1.1166, "step": 3315 }, { "epoch": 1.27, "learning_rate": 6.267021029001291e-06, "loss": 1.0851, "step": 3316 }, { "epoch": 1.27, "learning_rate": 6.261189284196409e-06, "loss": 1.0525, "step": 3317 }, { "epoch": 1.27, "learning_rate": 6.255359017187608e-06, "loss": 1.1003, "step": 3318 }, { "epoch": 1.27, "learning_rate": 6.249530230279351e-06, "loss": 1.1204, "step": 3319 }, { "epoch": 1.28, "learning_rate": 6.243702925775512e-06, "loss": 1.0661, "step": 3320 }, { "epoch": 1.28, "learning_rate": 6.237877105979389e-06, "loss": 1.1432, "step": 3321 }, { "epoch": 1.28, "learning_rate": 6.232052773193677e-06, "loss": 1.0528, "step": 3322 }, { "epoch": 1.28, "learning_rate": 6.226229929720501e-06, "loss": 1.1179, "step": 3323 }, { "epoch": 1.28, "learning_rate": 6.220408577861379e-06, "loss": 1.1147, "step": 3324 }, { "epoch": 1.28, "learning_rate": 6.214588719917256e-06, "loss": 1.1424, "step": 3325 }, { "epoch": 1.28, "learning_rate": 6.2087703581884775e-06, "loss": 1.0585, "step": 3326 }, { "epoch": 1.28, "learning_rate": 6.202953494974796e-06, "loss": 1.0988, "step": 3327 }, { "epoch": 1.28, "learning_rate": 6.197138132575381e-06, "loss": 1.104, "step": 3328 }, { "epoch": 1.28, "learning_rate": 6.191324273288795e-06, "loss": 1.1087, "step": 3329 }, { "epoch": 1.28, "learning_rate": 6.185511919413025e-06, "loss": 1.0999, "step": 3330 }, { "epoch": 1.28, "learning_rate": 6.179701073245441e-06, "loss": 1.1212, "step": 3331 }, { "epoch": 1.28, "learning_rate": 6.1738917370828376e-06, "loss": 1.0888, "step": 3332 }, { "epoch": 1.28, "learning_rate": 6.168083913221394e-06, "loss": 1.0963, "step": 3333 }, { "epoch": 1.28, "learning_rate": 6.162277603956715e-06, "loss": 1.1331, "step": 3334 }, { "epoch": 1.28, "learning_rate": 6.156472811583776e-06, "loss": 1.1432, "step": 3335 }, { "epoch": 1.28, "learning_rate": 6.150669538396983e-06, "loss": 1.0646, "step": 3336 }, { "epoch": 1.28, "learning_rate": 6.144867786690122e-06, "loss": 1.1487, "step": 3337 }, { "epoch": 1.28, "learning_rate": 6.139067558756382e-06, "loss": 1.1047, "step": 3338 }, { "epoch": 1.28, "learning_rate": 6.133268856888357e-06, "loss": 1.1484, "step": 3339 }, { "epoch": 1.28, "learning_rate": 6.127471683378025e-06, "loss": 1.1123, "step": 3340 }, { "epoch": 1.28, "learning_rate": 6.1216760405167745e-06, "loss": 1.1044, "step": 3341 }, { "epoch": 1.28, "learning_rate": 6.1158819305953766e-06, "loss": 1.1409, "step": 3342 }, { "epoch": 1.28, "learning_rate": 6.110089355904005e-06, "loss": 1.076, "step": 3343 }, { "epoch": 1.28, "learning_rate": 6.104298318732218e-06, "loss": 1.1056, "step": 3344 }, { "epoch": 1.28, "learning_rate": 6.098508821368978e-06, "loss": 1.0939, "step": 3345 }, { "epoch": 1.29, "learning_rate": 6.092720866102626e-06, "loss": 1.1129, "step": 3346 }, { "epoch": 1.29, "learning_rate": 6.086934455220906e-06, "loss": 1.1069, "step": 3347 }, { "epoch": 1.29, "learning_rate": 6.081149591010935e-06, "loss": 1.0985, "step": 3348 }, { "epoch": 1.29, "learning_rate": 6.0753662757592406e-06, "loss": 1.0996, "step": 3349 }, { "epoch": 1.29, "learning_rate": 6.06958451175172e-06, "loss": 1.1177, "step": 3350 }, { "epoch": 1.29, "learning_rate": 6.063804301273663e-06, "loss": 1.0909, "step": 3351 }, { "epoch": 1.29, "learning_rate": 6.0580256466097506e-06, "loss": 1.0473, "step": 3352 }, { "epoch": 1.29, "learning_rate": 6.052248550044035e-06, "loss": 1.0806, "step": 3353 }, { "epoch": 1.29, "learning_rate": 6.046473013859973e-06, "loss": 1.1045, "step": 3354 }, { "epoch": 1.29, "learning_rate": 6.040699040340386e-06, "loss": 1.0942, "step": 3355 }, { "epoch": 1.29, "learning_rate": 6.034926631767488e-06, "loss": 1.0804, "step": 3356 }, { "epoch": 1.29, "learning_rate": 6.029155790422867e-06, "loss": 1.0744, "step": 3357 }, { "epoch": 1.29, "learning_rate": 6.023386518587505e-06, "loss": 1.1001, "step": 3358 }, { "epoch": 1.29, "learning_rate": 6.017618818541747e-06, "loss": 1.0957, "step": 3359 }, { "epoch": 1.29, "learning_rate": 6.01185269256533e-06, "loss": 1.0948, "step": 3360 }, { "epoch": 1.29, "learning_rate": 6.00608814293736e-06, "loss": 1.097, "step": 3361 }, { "epoch": 1.29, "learning_rate": 6.000325171936319e-06, "loss": 1.0509, "step": 3362 }, { "epoch": 1.29, "learning_rate": 5.99456378184008e-06, "loss": 1.1294, "step": 3363 }, { "epoch": 1.29, "learning_rate": 5.98880397492587e-06, "loss": 1.1002, "step": 3364 }, { "epoch": 1.29, "learning_rate": 5.983045753470308e-06, "loss": 1.0621, "step": 3365 }, { "epoch": 1.29, "learning_rate": 5.977289119749372e-06, "loss": 1.1094, "step": 3366 }, { "epoch": 1.29, "learning_rate": 5.971534076038429e-06, "loss": 1.1005, "step": 3367 }, { "epoch": 1.29, "learning_rate": 5.965780624612199e-06, "loss": 1.1597, "step": 3368 }, { "epoch": 1.29, "learning_rate": 5.960028767744788e-06, "loss": 1.0658, "step": 3369 }, { "epoch": 1.29, "learning_rate": 5.9542785077096586e-06, "loss": 1.0618, "step": 3370 }, { "epoch": 1.29, "learning_rate": 5.948529846779656e-06, "loss": 1.0526, "step": 3371 }, { "epoch": 1.3, "learning_rate": 5.9427827872269815e-06, "loss": 1.0934, "step": 3372 }, { "epoch": 1.3, "learning_rate": 5.937037331323212e-06, "loss": 1.1381, "step": 3373 }, { "epoch": 1.3, "learning_rate": 5.9312934813392865e-06, "loss": 1.1681, "step": 3374 }, { "epoch": 1.3, "learning_rate": 5.925551239545501e-06, "loss": 1.0509, "step": 3375 }, { "epoch": 1.3, "learning_rate": 5.919810608211538e-06, "loss": 1.1526, "step": 3376 }, { "epoch": 1.3, "learning_rate": 5.91407158960642e-06, "loss": 1.0845, "step": 3377 }, { "epoch": 1.3, "learning_rate": 5.9083341859985464e-06, "loss": 1.0641, "step": 3378 }, { "epoch": 1.3, "learning_rate": 5.902598399655669e-06, "loss": 1.101, "step": 3379 }, { "epoch": 1.3, "learning_rate": 5.896864232844912e-06, "loss": 1.1351, "step": 3380 }, { "epoch": 1.3, "learning_rate": 5.891131687832747e-06, "loss": 1.0765, "step": 3381 }, { "epoch": 1.3, "learning_rate": 5.885400766885016e-06, "loss": 1.1578, "step": 3382 }, { "epoch": 1.3, "learning_rate": 5.879671472266903e-06, "loss": 1.1335, "step": 3383 }, { "epoch": 1.3, "learning_rate": 5.873943806242972e-06, "loss": 1.1194, "step": 3384 }, { "epoch": 1.3, "learning_rate": 5.868217771077125e-06, "loss": 1.1165, "step": 3385 }, { "epoch": 1.3, "learning_rate": 5.862493369032623e-06, "loss": 1.13, "step": 3386 }, { "epoch": 1.3, "learning_rate": 5.856770602372085e-06, "loss": 1.1126, "step": 3387 }, { "epoch": 1.3, "learning_rate": 5.85104947335748e-06, "loss": 1.0595, "step": 3388 }, { "epoch": 1.3, "learning_rate": 5.845329984250139e-06, "loss": 1.0961, "step": 3389 }, { "epoch": 1.3, "learning_rate": 5.839612137310731e-06, "loss": 1.0813, "step": 3390 }, { "epoch": 1.3, "learning_rate": 5.833895934799285e-06, "loss": 1.0959, "step": 3391 }, { "epoch": 1.3, "learning_rate": 5.8281813789751775e-06, "loss": 1.1135, "step": 3392 }, { "epoch": 1.3, "learning_rate": 5.822468472097133e-06, "loss": 1.0998, "step": 3393 }, { "epoch": 1.3, "learning_rate": 5.816757216423227e-06, "loss": 1.0525, "step": 3394 }, { "epoch": 1.3, "learning_rate": 5.811047614210883e-06, "loss": 1.0759, "step": 3395 }, { "epoch": 1.3, "learning_rate": 5.805339667716865e-06, "loss": 1.0735, "step": 3396 }, { "epoch": 1.3, "learning_rate": 5.799633379197291e-06, "loss": 1.1284, "step": 3397 }, { "epoch": 1.31, "learning_rate": 5.793928750907614e-06, "loss": 1.1351, "step": 3398 }, { "epoch": 1.31, "learning_rate": 5.788225785102637e-06, "loss": 1.1076, "step": 3399 }, { "epoch": 1.31, "learning_rate": 5.782524484036509e-06, "loss": 1.0852, "step": 3400 }, { "epoch": 1.31, "learning_rate": 5.776824849962706e-06, "loss": 1.2142, "step": 3401 }, { "epoch": 1.31, "learning_rate": 5.771126885134071e-06, "loss": 1.1616, "step": 3402 }, { "epoch": 1.31, "learning_rate": 5.765430591802756e-06, "loss": 1.0657, "step": 3403 }, { "epoch": 1.31, "learning_rate": 5.759735972220286e-06, "loss": 1.0676, "step": 3404 }, { "epoch": 1.31, "learning_rate": 5.754043028637492e-06, "loss": 1.132, "step": 3405 }, { "epoch": 1.31, "learning_rate": 5.748351763304565e-06, "loss": 1.0507, "step": 3406 }, { "epoch": 1.31, "learning_rate": 5.742662178471021e-06, "loss": 1.0576, "step": 3407 }, { "epoch": 1.31, "learning_rate": 5.7369742763857224e-06, "loss": 1.0861, "step": 3408 }, { "epoch": 1.31, "learning_rate": 5.731288059296851e-06, "loss": 1.1022, "step": 3409 }, { "epoch": 1.31, "learning_rate": 5.7256035294519356e-06, "loss": 1.0487, "step": 3410 }, { "epoch": 1.31, "learning_rate": 5.719920689097836e-06, "loss": 1.093, "step": 3411 }, { "epoch": 1.31, "learning_rate": 5.714239540480737e-06, "loss": 1.0593, "step": 3412 }, { "epoch": 1.31, "learning_rate": 5.708560085846169e-06, "loss": 1.1313, "step": 3413 }, { "epoch": 1.31, "learning_rate": 5.70288232743897e-06, "loss": 1.0533, "step": 3414 }, { "epoch": 1.31, "learning_rate": 5.697206267503337e-06, "loss": 1.1181, "step": 3415 }, { "epoch": 1.31, "learning_rate": 5.691531908282768e-06, "loss": 1.113, "step": 3416 }, { "epoch": 1.31, "learning_rate": 5.685859252020105e-06, "loss": 1.0911, "step": 3417 }, { "epoch": 1.31, "learning_rate": 5.680188300957515e-06, "loss": 1.0777, "step": 3418 }, { "epoch": 1.31, "learning_rate": 5.674519057336483e-06, "loss": 1.1625, "step": 3419 }, { "epoch": 1.31, "learning_rate": 5.668851523397829e-06, "loss": 1.0892, "step": 3420 }, { "epoch": 1.31, "learning_rate": 5.663185701381697e-06, "loss": 1.1054, "step": 3421 }, { "epoch": 1.31, "learning_rate": 5.657521593527541e-06, "loss": 1.0537, "step": 3422 }, { "epoch": 1.31, "learning_rate": 5.65185920207415e-06, "loss": 1.122, "step": 3423 }, { "epoch": 1.32, "learning_rate": 5.6461985292596345e-06, "loss": 1.0452, "step": 3424 }, { "epoch": 1.32, "learning_rate": 5.64053957732142e-06, "loss": 1.0664, "step": 3425 }, { "epoch": 1.32, "learning_rate": 5.634882348496258e-06, "loss": 1.0976, "step": 3426 }, { "epoch": 1.32, "learning_rate": 5.629226845020206e-06, "loss": 1.1065, "step": 3427 }, { "epoch": 1.32, "learning_rate": 5.623573069128663e-06, "loss": 1.0549, "step": 3428 }, { "epoch": 1.32, "learning_rate": 5.617921023056319e-06, "loss": 1.1052, "step": 3429 }, { "epoch": 1.32, "learning_rate": 5.612270709037197e-06, "loss": 1.0886, "step": 3430 }, { "epoch": 1.32, "learning_rate": 5.606622129304629e-06, "loss": 1.0709, "step": 3431 }, { "epoch": 1.32, "learning_rate": 5.600975286091266e-06, "loss": 1.0693, "step": 3432 }, { "epoch": 1.32, "learning_rate": 5.595330181629072e-06, "loss": 1.1377, "step": 3433 }, { "epoch": 1.32, "learning_rate": 5.5896868181493136e-06, "loss": 1.0826, "step": 3434 }, { "epoch": 1.32, "learning_rate": 5.584045197882582e-06, "loss": 1.1079, "step": 3435 }, { "epoch": 1.32, "learning_rate": 5.578405323058773e-06, "loss": 1.1178, "step": 3436 }, { "epoch": 1.32, "learning_rate": 5.572767195907096e-06, "loss": 1.1373, "step": 3437 }, { "epoch": 1.32, "learning_rate": 5.567130818656067e-06, "loss": 1.0946, "step": 3438 }, { "epoch": 1.32, "learning_rate": 5.561496193533516e-06, "loss": 1.1058, "step": 3439 }, { "epoch": 1.32, "learning_rate": 5.555863322766561e-06, "loss": 1.1587, "step": 3440 }, { "epoch": 1.32, "learning_rate": 5.550232208581658e-06, "loss": 1.1274, "step": 3441 }, { "epoch": 1.32, "learning_rate": 5.544602853204542e-06, "loss": 1.0847, "step": 3442 }, { "epoch": 1.32, "learning_rate": 5.538975258860264e-06, "loss": 1.0684, "step": 3443 }, { "epoch": 1.32, "learning_rate": 5.53334942777318e-06, "loss": 1.1188, "step": 3444 }, { "epoch": 1.32, "learning_rate": 5.527725362166945e-06, "loss": 1.1086, "step": 3445 }, { "epoch": 1.32, "learning_rate": 5.52210306426452e-06, "loss": 1.1095, "step": 3446 }, { "epoch": 1.32, "learning_rate": 5.516482536288161e-06, "loss": 1.1185, "step": 3447 }, { "epoch": 1.32, "learning_rate": 5.51086378045943e-06, "loss": 1.1616, "step": 3448 }, { "epoch": 1.32, "learning_rate": 5.505246798999188e-06, "loss": 1.1216, "step": 3449 }, { "epoch": 1.33, "learning_rate": 5.499631594127593e-06, "loss": 1.062, "step": 3450 }, { "epoch": 1.33, "learning_rate": 5.494018168064101e-06, "loss": 1.0651, "step": 3451 }, { "epoch": 1.33, "learning_rate": 5.488406523027472e-06, "loss": 1.1313, "step": 3452 }, { "epoch": 1.33, "learning_rate": 5.482796661235739e-06, "loss": 1.124, "step": 3453 }, { "epoch": 1.33, "learning_rate": 5.477188584906268e-06, "loss": 1.1226, "step": 3454 }, { "epoch": 1.33, "learning_rate": 5.4715822962556815e-06, "loss": 1.1091, "step": 3455 }, { "epoch": 1.33, "learning_rate": 5.465977797499919e-06, "loss": 1.0763, "step": 3456 }, { "epoch": 1.33, "learning_rate": 5.460375090854206e-06, "loss": 1.1201, "step": 3457 }, { "epoch": 1.33, "learning_rate": 5.454774178533051e-06, "loss": 1.1151, "step": 3458 }, { "epoch": 1.33, "learning_rate": 5.449175062750274e-06, "loss": 1.0779, "step": 3459 }, { "epoch": 1.33, "learning_rate": 5.443577745718961e-06, "loss": 1.0865, "step": 3460 }, { "epoch": 1.33, "learning_rate": 5.437982229651501e-06, "loss": 1.125, "step": 3461 }, { "epoch": 1.33, "learning_rate": 5.432388516759573e-06, "loss": 1.0378, "step": 3462 }, { "epoch": 1.33, "learning_rate": 5.426796609254132e-06, "loss": 1.0363, "step": 3463 }, { "epoch": 1.33, "learning_rate": 5.4212065093454315e-06, "loss": 1.1158, "step": 3464 }, { "epoch": 1.33, "learning_rate": 5.4156182192430065e-06, "loss": 1.0829, "step": 3465 }, { "epoch": 1.33, "learning_rate": 5.410031741155663e-06, "loss": 1.1074, "step": 3466 }, { "epoch": 1.33, "learning_rate": 5.4044470772915215e-06, "loss": 1.0886, "step": 3467 }, { "epoch": 1.33, "learning_rate": 5.398864229857952e-06, "loss": 1.1462, "step": 3468 }, { "epoch": 1.33, "learning_rate": 5.393283201061625e-06, "loss": 1.1197, "step": 3469 }, { "epoch": 1.33, "learning_rate": 5.387703993108496e-06, "loss": 1.0617, "step": 3470 }, { "epoch": 1.33, "learning_rate": 5.38212660820378e-06, "loss": 1.0377, "step": 3471 }, { "epoch": 1.33, "learning_rate": 5.376551048551999e-06, "loss": 1.1535, "step": 3472 }, { "epoch": 1.33, "learning_rate": 5.370977316356929e-06, "loss": 1.0851, "step": 3473 }, { "epoch": 1.33, "learning_rate": 5.3654054138216384e-06, "loss": 1.0468, "step": 3474 }, { "epoch": 1.33, "learning_rate": 5.359835343148467e-06, "loss": 1.1501, "step": 3475 }, { "epoch": 1.34, "learning_rate": 5.354267106539037e-06, "loss": 1.1026, "step": 3476 }, { "epoch": 1.34, "learning_rate": 5.348700706194227e-06, "loss": 1.0507, "step": 3477 }, { "epoch": 1.34, "learning_rate": 5.343136144314221e-06, "loss": 1.0867, "step": 3478 }, { "epoch": 1.34, "learning_rate": 5.337573423098443e-06, "loss": 1.0981, "step": 3479 }, { "epoch": 1.34, "learning_rate": 5.332012544745613e-06, "loss": 1.0791, "step": 3480 }, { "epoch": 1.34, "learning_rate": 5.326453511453715e-06, "loss": 1.0437, "step": 3481 }, { "epoch": 1.34, "learning_rate": 5.320896325419993e-06, "loss": 1.0906, "step": 3482 }, { "epoch": 1.34, "learning_rate": 5.315340988840987e-06, "loss": 1.0544, "step": 3483 }, { "epoch": 1.34, "learning_rate": 5.309787503912478e-06, "loss": 1.1061, "step": 3484 }, { "epoch": 1.34, "learning_rate": 5.30423587282953e-06, "loss": 1.099, "step": 3485 }, { "epoch": 1.34, "learning_rate": 5.298686097786474e-06, "loss": 1.1305, "step": 3486 }, { "epoch": 1.34, "learning_rate": 5.293138180976902e-06, "loss": 1.168, "step": 3487 }, { "epoch": 1.34, "learning_rate": 5.287592124593677e-06, "loss": 1.1216, "step": 3488 }, { "epoch": 1.34, "learning_rate": 5.282047930828926e-06, "loss": 1.0254, "step": 3489 }, { "epoch": 1.34, "learning_rate": 5.276505601874027e-06, "loss": 1.0246, "step": 3490 }, { "epoch": 1.34, "learning_rate": 5.2709651399196456e-06, "loss": 1.164, "step": 3491 }, { "epoch": 1.34, "learning_rate": 5.265426547155685e-06, "loss": 1.1377, "step": 3492 }, { "epoch": 1.34, "learning_rate": 5.259889825771326e-06, "loss": 1.1187, "step": 3493 }, { "epoch": 1.34, "learning_rate": 5.254354977955005e-06, "loss": 1.0916, "step": 3494 }, { "epoch": 1.34, "learning_rate": 5.248822005894405e-06, "loss": 1.0527, "step": 3495 }, { "epoch": 1.34, "learning_rate": 5.243290911776497e-06, "loss": 1.101, "step": 3496 }, { "epoch": 1.34, "learning_rate": 5.237761697787479e-06, "loss": 1.1144, "step": 3497 }, { "epoch": 1.34, "learning_rate": 5.232234366112823e-06, "loss": 1.0704, "step": 3498 }, { "epoch": 1.34, "learning_rate": 5.226708918937252e-06, "loss": 1.1067, "step": 3499 }, { "epoch": 1.34, "learning_rate": 5.221185358444747e-06, "loss": 1.1386, "step": 3500 }, { "epoch": 1.34, "learning_rate": 5.215663686818539e-06, "loss": 1.0891, "step": 3501 }, { "epoch": 1.35, "learning_rate": 5.210143906241121e-06, "loss": 1.0908, "step": 3502 }, { "epoch": 1.35, "learning_rate": 5.204626018894218e-06, "loss": 1.0893, "step": 3503 }, { "epoch": 1.35, "learning_rate": 5.1991100269588376e-06, "loss": 1.0912, "step": 3504 }, { "epoch": 1.35, "learning_rate": 5.193595932615212e-06, "loss": 1.0523, "step": 3505 }, { "epoch": 1.35, "learning_rate": 5.188083738042834e-06, "loss": 1.0843, "step": 3506 }, { "epoch": 1.35, "learning_rate": 5.182573445420449e-06, "loss": 1.0524, "step": 3507 }, { "epoch": 1.35, "learning_rate": 5.177065056926035e-06, "loss": 1.0778, "step": 3508 }, { "epoch": 1.35, "learning_rate": 5.171558574736844e-06, "loss": 1.0989, "step": 3509 }, { "epoch": 1.35, "learning_rate": 5.166054001029346e-06, "loss": 1.1478, "step": 3510 }, { "epoch": 1.35, "learning_rate": 5.160551337979276e-06, "loss": 1.0722, "step": 3511 }, { "epoch": 1.35, "learning_rate": 5.155050587761604e-06, "loss": 1.134, "step": 3512 }, { "epoch": 1.35, "learning_rate": 5.14955175255055e-06, "loss": 1.0392, "step": 3513 }, { "epoch": 1.35, "learning_rate": 5.144054834519574e-06, "loss": 1.082, "step": 3514 }, { "epoch": 1.35, "learning_rate": 5.138559835841381e-06, "loss": 1.0759, "step": 3515 }, { "epoch": 1.35, "learning_rate": 5.133066758687906e-06, "loss": 1.1046, "step": 3516 }, { "epoch": 1.35, "learning_rate": 5.127575605230347e-06, "loss": 1.1369, "step": 3517 }, { "epoch": 1.35, "learning_rate": 5.122086377639117e-06, "loss": 1.0668, "step": 3518 }, { "epoch": 1.35, "learning_rate": 5.116599078083884e-06, "loss": 1.1082, "step": 3519 }, { "epoch": 1.35, "learning_rate": 5.11111370873355e-06, "loss": 1.1232, "step": 3520 }, { "epoch": 1.35, "learning_rate": 5.105630271756243e-06, "loss": 1.0369, "step": 3521 }, { "epoch": 1.35, "learning_rate": 5.100148769319353e-06, "loss": 1.0754, "step": 3522 }, { "epoch": 1.35, "learning_rate": 5.094669203589478e-06, "loss": 1.1077, "step": 3523 }, { "epoch": 1.35, "learning_rate": 5.089191576732465e-06, "loss": 1.1003, "step": 3524 }, { "epoch": 1.35, "learning_rate": 5.083715890913392e-06, "loss": 1.1093, "step": 3525 }, { "epoch": 1.35, "learning_rate": 5.078242148296569e-06, "loss": 1.0898, "step": 3526 }, { "epoch": 1.35, "learning_rate": 5.0727703510455395e-06, "loss": 1.0943, "step": 3527 }, { "epoch": 1.36, "learning_rate": 5.06730050132308e-06, "loss": 1.0702, "step": 3528 }, { "epoch": 1.36, "learning_rate": 5.061832601291186e-06, "loss": 1.1441, "step": 3529 }, { "epoch": 1.36, "learning_rate": 5.056366653111096e-06, "loss": 1.1141, "step": 3530 }, { "epoch": 1.36, "learning_rate": 5.050902658943269e-06, "loss": 1.107, "step": 3531 }, { "epoch": 1.36, "learning_rate": 5.045440620947396e-06, "loss": 1.1627, "step": 3532 }, { "epoch": 1.36, "learning_rate": 5.039980541282396e-06, "loss": 1.0771, "step": 3533 }, { "epoch": 1.36, "learning_rate": 5.034522422106403e-06, "loss": 1.16, "step": 3534 }, { "epoch": 1.36, "learning_rate": 5.0290662655767945e-06, "loss": 1.0984, "step": 3535 }, { "epoch": 1.36, "learning_rate": 5.0236120738501514e-06, "loss": 1.0386, "step": 3536 }, { "epoch": 1.36, "learning_rate": 5.0181598490822935e-06, "loss": 1.1337, "step": 3537 }, { "epoch": 1.36, "learning_rate": 5.012709593428259e-06, "loss": 1.0579, "step": 3538 }, { "epoch": 1.36, "learning_rate": 5.007261309042304e-06, "loss": 1.129, "step": 3539 }, { "epoch": 1.36, "learning_rate": 5.0018149980779115e-06, "loss": 1.1188, "step": 3540 }, { "epoch": 1.36, "learning_rate": 4.996370662687782e-06, "loss": 1.0682, "step": 3541 }, { "epoch": 1.36, "learning_rate": 4.990928305023829e-06, "loss": 1.0827, "step": 3542 }, { "epoch": 1.36, "learning_rate": 4.985487927237193e-06, "loss": 1.1021, "step": 3543 }, { "epoch": 1.36, "learning_rate": 4.9800495314782305e-06, "loss": 1.0611, "step": 3544 }, { "epoch": 1.36, "learning_rate": 4.974613119896504e-06, "loss": 1.113, "step": 3545 }, { "epoch": 1.36, "learning_rate": 4.969178694640815e-06, "loss": 1.0813, "step": 3546 }, { "epoch": 1.36, "learning_rate": 4.963746257859153e-06, "loss": 1.1272, "step": 3547 }, { "epoch": 1.36, "learning_rate": 4.95831581169874e-06, "loss": 1.1068, "step": 3548 }, { "epoch": 1.36, "learning_rate": 4.952887358306001e-06, "loss": 1.1491, "step": 3549 }, { "epoch": 1.36, "learning_rate": 4.9474608998265815e-06, "loss": 1.1872, "step": 3550 }, { "epoch": 1.36, "learning_rate": 4.942036438405332e-06, "loss": 1.0718, "step": 3551 }, { "epoch": 1.36, "learning_rate": 4.936613976186321e-06, "loss": 1.0833, "step": 3552 }, { "epoch": 1.36, "learning_rate": 4.931193515312813e-06, "loss": 1.1096, "step": 3553 }, { "epoch": 1.37, "learning_rate": 4.925775057927297e-06, "loss": 1.1067, "step": 3554 }, { "epoch": 1.37, "learning_rate": 4.920358606171459e-06, "loss": 1.0969, "step": 3555 }, { "epoch": 1.37, "learning_rate": 4.914944162186201e-06, "loss": 1.094, "step": 3556 }, { "epoch": 1.37, "learning_rate": 4.909531728111628e-06, "loss": 1.1296, "step": 3557 }, { "epoch": 1.37, "learning_rate": 4.90412130608704e-06, "loss": 1.0617, "step": 3558 }, { "epoch": 1.37, "learning_rate": 4.898712898250965e-06, "loss": 1.0552, "step": 3559 }, { "epoch": 1.37, "learning_rate": 4.893306506741111e-06, "loss": 1.0785, "step": 3560 }, { "epoch": 1.37, "learning_rate": 4.887902133694402e-06, "loss": 1.1128, "step": 3561 }, { "epoch": 1.37, "learning_rate": 4.882499781246963e-06, "loss": 1.0842, "step": 3562 }, { "epoch": 1.37, "learning_rate": 4.877099451534119e-06, "loss": 1.1023, "step": 3563 }, { "epoch": 1.37, "learning_rate": 4.871701146690393e-06, "loss": 1.1308, "step": 3564 }, { "epoch": 1.37, "learning_rate": 4.866304868849515e-06, "loss": 1.0512, "step": 3565 }, { "epoch": 1.37, "learning_rate": 4.8609106201444025e-06, "loss": 1.0981, "step": 3566 }, { "epoch": 1.37, "learning_rate": 4.8555184027071785e-06, "loss": 1.1002, "step": 3567 }, { "epoch": 1.37, "learning_rate": 4.8501282186691645e-06, "loss": 1.0539, "step": 3568 }, { "epoch": 1.37, "learning_rate": 4.844740070160874e-06, "loss": 1.0727, "step": 3569 }, { "epoch": 1.37, "learning_rate": 4.839353959312021e-06, "loss": 1.1247, "step": 3570 }, { "epoch": 1.37, "learning_rate": 4.8339698882515e-06, "loss": 1.1362, "step": 3571 }, { "epoch": 1.37, "learning_rate": 4.828587859107426e-06, "loss": 1.0942, "step": 3572 }, { "epoch": 1.37, "learning_rate": 4.823207874007077e-06, "loss": 1.0805, "step": 3573 }, { "epoch": 1.37, "learning_rate": 4.817829935076944e-06, "loss": 1.1031, "step": 3574 }, { "epoch": 1.37, "learning_rate": 4.812454044442701e-06, "loss": 1.1074, "step": 3575 }, { "epoch": 1.37, "learning_rate": 4.807080204229211e-06, "loss": 1.132, "step": 3576 }, { "epoch": 1.37, "learning_rate": 4.801708416560537e-06, "loss": 1.1014, "step": 3577 }, { "epoch": 1.37, "learning_rate": 4.796338683559912e-06, "loss": 1.0785, "step": 3578 }, { "epoch": 1.37, "learning_rate": 4.790971007349772e-06, "loss": 1.166, "step": 3579 }, { "epoch": 1.38, "learning_rate": 4.7856053900517374e-06, "loss": 1.131, "step": 3580 }, { "epoch": 1.38, "learning_rate": 4.7802418337866116e-06, "loss": 1.082, "step": 3581 }, { "epoch": 1.38, "learning_rate": 4.774880340674386e-06, "loss": 1.08, "step": 3582 }, { "epoch": 1.38, "learning_rate": 4.769520912834239e-06, "loss": 1.1403, "step": 3583 }, { "epoch": 1.38, "learning_rate": 4.7641635523845165e-06, "loss": 1.1088, "step": 3584 }, { "epoch": 1.38, "learning_rate": 4.758808261442777e-06, "loss": 1.0977, "step": 3585 }, { "epoch": 1.38, "learning_rate": 4.753455042125731e-06, "loss": 1.1198, "step": 3586 }, { "epoch": 1.38, "learning_rate": 4.7481038965492865e-06, "loss": 1.0741, "step": 3587 }, { "epoch": 1.38, "learning_rate": 4.7427548268285295e-06, "loss": 1.1158, "step": 3588 }, { "epoch": 1.38, "learning_rate": 4.737407835077724e-06, "loss": 1.0813, "step": 3589 }, { "epoch": 1.38, "learning_rate": 4.7320629234103166e-06, "loss": 1.0921, "step": 3590 }, { "epoch": 1.38, "learning_rate": 4.7267200939389216e-06, "loss": 1.0631, "step": 3591 }, { "epoch": 1.38, "learning_rate": 4.721379348775339e-06, "loss": 1.1119, "step": 3592 }, { "epoch": 1.38, "learning_rate": 4.716040690030544e-06, "loss": 1.0719, "step": 3593 }, { "epoch": 1.38, "learning_rate": 4.7107041198146855e-06, "loss": 1.1056, "step": 3594 }, { "epoch": 1.38, "learning_rate": 4.705369640237088e-06, "loss": 1.0743, "step": 3595 }, { "epoch": 1.38, "learning_rate": 4.700037253406251e-06, "loss": 1.1039, "step": 3596 }, { "epoch": 1.38, "learning_rate": 4.694706961429835e-06, "loss": 1.1174, "step": 3597 }, { "epoch": 1.38, "learning_rate": 4.689378766414697e-06, "loss": 1.1056, "step": 3598 }, { "epoch": 1.38, "learning_rate": 4.68405267046684e-06, "loss": 1.0954, "step": 3599 }, { "epoch": 1.38, "learning_rate": 4.6787286756914505e-06, "loss": 1.0419, "step": 3600 }, { "epoch": 1.38, "eval_loss": 1.106767177581787, "eval_runtime": 688.2419, "eval_samples_per_second": 4.942, "eval_steps_per_second": 0.309, "step": 3600 }, { "epoch": 1.38, "learning_rate": 4.6734067841928856e-06, "loss": 1.1465, "step": 3601 }, { "epoch": 1.38, "learning_rate": 4.668086998074656e-06, "loss": 1.0892, "step": 3602 }, { "epoch": 1.38, "learning_rate": 4.662769319439468e-06, "loss": 1.154, "step": 3603 }, { "epoch": 1.38, "learning_rate": 4.657453750389163e-06, "loss": 1.111, "step": 3604 }, { "epoch": 1.38, "learning_rate": 4.652140293024774e-06, "loss": 1.076, "step": 3605 }, { "epoch": 1.39, "learning_rate": 4.646828949446484e-06, "loss": 1.1529, "step": 3606 }, { "epoch": 1.39, "learning_rate": 4.641519721753648e-06, "loss": 1.0491, "step": 3607 }, { "epoch": 1.39, "learning_rate": 4.636212612044781e-06, "loss": 1.0967, "step": 3608 }, { "epoch": 1.39, "learning_rate": 4.6309076224175685e-06, "loss": 1.0541, "step": 3609 }, { "epoch": 1.39, "learning_rate": 4.625604754968839e-06, "loss": 1.1365, "step": 3610 }, { "epoch": 1.39, "learning_rate": 4.620304011794611e-06, "loss": 1.1654, "step": 3611 }, { "epoch": 1.39, "learning_rate": 4.615005394990034e-06, "loss": 1.0988, "step": 3612 }, { "epoch": 1.39, "learning_rate": 4.609708906649436e-06, "loss": 1.09, "step": 3613 }, { "epoch": 1.39, "learning_rate": 4.604414548866301e-06, "loss": 1.1278, "step": 3614 }, { "epoch": 1.39, "learning_rate": 4.599122323733262e-06, "loss": 1.0323, "step": 3615 }, { "epoch": 1.39, "learning_rate": 4.5938322333421195e-06, "loss": 1.0873, "step": 3616 }, { "epoch": 1.39, "learning_rate": 4.588544279783822e-06, "loss": 1.1393, "step": 3617 }, { "epoch": 1.39, "learning_rate": 4.583258465148479e-06, "loss": 1.0987, "step": 3618 }, { "epoch": 1.39, "learning_rate": 4.577974791525355e-06, "loss": 1.0846, "step": 3619 }, { "epoch": 1.39, "learning_rate": 4.572693261002867e-06, "loss": 1.0804, "step": 3620 }, { "epoch": 1.39, "learning_rate": 4.5674138756685726e-06, "loss": 1.0718, "step": 3621 }, { "epoch": 1.39, "learning_rate": 4.56213663760921e-06, "loss": 1.0381, "step": 3622 }, { "epoch": 1.39, "learning_rate": 4.5568615489106386e-06, "loss": 1.0804, "step": 3623 }, { "epoch": 1.39, "learning_rate": 4.5515886116578855e-06, "loss": 1.1023, "step": 3624 }, { "epoch": 1.39, "learning_rate": 4.546317827935127e-06, "loss": 1.0678, "step": 3625 }, { "epoch": 1.39, "learning_rate": 4.541049199825672e-06, "loss": 1.102, "step": 3626 }, { "epoch": 1.39, "learning_rate": 4.535782729412006e-06, "loss": 1.0872, "step": 3627 }, { "epoch": 1.39, "learning_rate": 4.530518418775734e-06, "loss": 1.1364, "step": 3628 }, { "epoch": 1.39, "learning_rate": 4.525256269997621e-06, "loss": 1.133, "step": 3629 }, { "epoch": 1.39, "learning_rate": 4.519996285157576e-06, "loss": 1.1248, "step": 3630 }, { "epoch": 1.39, "learning_rate": 4.5147384663346514e-06, "loss": 1.1049, "step": 3631 }, { "epoch": 1.4, "learning_rate": 4.509482815607045e-06, "loss": 1.1063, "step": 3632 }, { "epoch": 1.4, "learning_rate": 4.5042293350520995e-06, "loss": 1.119, "step": 3633 }, { "epoch": 1.4, "learning_rate": 4.498978026746285e-06, "loss": 1.1186, "step": 3634 }, { "epoch": 1.4, "learning_rate": 4.493728892765243e-06, "loss": 1.0707, "step": 3635 }, { "epoch": 1.4, "learning_rate": 4.48848193518372e-06, "loss": 1.1021, "step": 3636 }, { "epoch": 1.4, "learning_rate": 4.483237156075629e-06, "loss": 1.155, "step": 3637 }, { "epoch": 1.4, "learning_rate": 4.477994557514015e-06, "loss": 1.0945, "step": 3638 }, { "epoch": 1.4, "learning_rate": 4.472754141571046e-06, "loss": 1.1134, "step": 3639 }, { "epoch": 1.4, "learning_rate": 4.467515910318058e-06, "loss": 1.1462, "step": 3640 }, { "epoch": 1.4, "learning_rate": 4.462279865825491e-06, "loss": 1.0605, "step": 3641 }, { "epoch": 1.4, "learning_rate": 4.45704601016294e-06, "loss": 1.0616, "step": 3642 }, { "epoch": 1.4, "learning_rate": 4.451814345399131e-06, "loss": 1.0528, "step": 3643 }, { "epoch": 1.4, "learning_rate": 4.446584873601922e-06, "loss": 1.0181, "step": 3644 }, { "epoch": 1.4, "learning_rate": 4.441357596838306e-06, "loss": 1.1491, "step": 3645 }, { "epoch": 1.4, "learning_rate": 4.436132517174412e-06, "loss": 1.1053, "step": 3646 }, { "epoch": 1.4, "learning_rate": 4.430909636675483e-06, "loss": 1.1386, "step": 3647 }, { "epoch": 1.4, "learning_rate": 4.425688957405924e-06, "loss": 1.1165, "step": 3648 }, { "epoch": 1.4, "learning_rate": 4.420470481429238e-06, "loss": 1.0044, "step": 3649 }, { "epoch": 1.4, "learning_rate": 4.415254210808075e-06, "loss": 1.0372, "step": 3650 }, { "epoch": 1.4, "learning_rate": 4.410040147604215e-06, "loss": 1.1406, "step": 3651 }, { "epoch": 1.4, "learning_rate": 4.404828293878547e-06, "loss": 1.0686, "step": 3652 }, { "epoch": 1.4, "learning_rate": 4.399618651691116e-06, "loss": 1.0484, "step": 3653 }, { "epoch": 1.4, "learning_rate": 4.394411223101064e-06, "loss": 1.1269, "step": 3654 }, { "epoch": 1.4, "learning_rate": 4.389206010166676e-06, "loss": 1.0698, "step": 3655 }, { "epoch": 1.4, "learning_rate": 4.384003014945353e-06, "loss": 1.0552, "step": 3656 }, { "epoch": 1.4, "learning_rate": 4.378802239493625e-06, "loss": 1.1291, "step": 3657 }, { "epoch": 1.41, "learning_rate": 4.3736036858671395e-06, "loss": 1.0682, "step": 3658 }, { "epoch": 1.41, "learning_rate": 4.368407356120673e-06, "loss": 1.1064, "step": 3659 }, { "epoch": 1.41, "learning_rate": 4.3632132523081084e-06, "loss": 1.0722, "step": 3660 }, { "epoch": 1.41, "learning_rate": 4.35802137648247e-06, "loss": 1.1167, "step": 3661 }, { "epoch": 1.41, "learning_rate": 4.352831730695882e-06, "loss": 1.1054, "step": 3662 }, { "epoch": 1.41, "learning_rate": 4.347644316999597e-06, "loss": 1.086, "step": 3663 }, { "epoch": 1.41, "learning_rate": 4.342459137443987e-06, "loss": 1.0859, "step": 3664 }, { "epoch": 1.41, "learning_rate": 4.337276194078529e-06, "loss": 1.0562, "step": 3665 }, { "epoch": 1.41, "learning_rate": 4.3320954889518365e-06, "loss": 1.0632, "step": 3666 }, { "epoch": 1.41, "learning_rate": 4.326917024111615e-06, "loss": 1.0321, "step": 3667 }, { "epoch": 1.41, "learning_rate": 4.3217408016047025e-06, "loss": 1.1208, "step": 3668 }, { "epoch": 1.41, "learning_rate": 4.31656682347704e-06, "loss": 1.064, "step": 3669 }, { "epoch": 1.41, "learning_rate": 4.31139509177369e-06, "loss": 1.0345, "step": 3670 }, { "epoch": 1.41, "learning_rate": 4.30622560853882e-06, "loss": 1.0863, "step": 3671 }, { "epoch": 1.41, "learning_rate": 4.301058375815714e-06, "loss": 1.112, "step": 3672 }, { "epoch": 1.41, "learning_rate": 4.2958933956467585e-06, "loss": 1.1061, "step": 3673 }, { "epoch": 1.41, "learning_rate": 4.290730670073459e-06, "loss": 1.0853, "step": 3674 }, { "epoch": 1.41, "learning_rate": 4.285570201136423e-06, "loss": 1.0712, "step": 3675 }, { "epoch": 1.41, "learning_rate": 4.280411990875371e-06, "loss": 1.0454, "step": 3676 }, { "epoch": 1.41, "learning_rate": 4.275256041329131e-06, "loss": 1.1035, "step": 3677 }, { "epoch": 1.41, "learning_rate": 4.270102354535625e-06, "loss": 1.1239, "step": 3678 }, { "epoch": 1.41, "learning_rate": 4.264950932531906e-06, "loss": 1.1527, "step": 3679 }, { "epoch": 1.41, "learning_rate": 4.259801777354103e-06, "loss": 1.0496, "step": 3680 }, { "epoch": 1.41, "learning_rate": 4.2546548910374686e-06, "loss": 1.0749, "step": 3681 }, { "epoch": 1.41, "learning_rate": 4.249510275616352e-06, "loss": 1.1255, "step": 3682 }, { "epoch": 1.41, "learning_rate": 4.244367933124207e-06, "loss": 1.1225, "step": 3683 }, { "epoch": 1.42, "learning_rate": 4.239227865593579e-06, "loss": 1.101, "step": 3684 }, { "epoch": 1.42, "learning_rate": 4.234090075056135e-06, "loss": 1.1738, "step": 3685 }, { "epoch": 1.42, "learning_rate": 4.228954563542621e-06, "loss": 1.1121, "step": 3686 }, { "epoch": 1.42, "learning_rate": 4.223821333082892e-06, "loss": 1.106, "step": 3687 }, { "epoch": 1.42, "learning_rate": 4.218690385705904e-06, "loss": 1.0849, "step": 3688 }, { "epoch": 1.42, "learning_rate": 4.2135617234396964e-06, "loss": 1.1161, "step": 3689 }, { "epoch": 1.42, "learning_rate": 4.208435348311431e-06, "loss": 1.0591, "step": 3690 }, { "epoch": 1.42, "learning_rate": 4.203311262347338e-06, "loss": 1.1024, "step": 3691 }, { "epoch": 1.42, "learning_rate": 4.198189467572759e-06, "loss": 1.1045, "step": 3692 }, { "epoch": 1.42, "learning_rate": 4.193069966012128e-06, "loss": 1.113, "step": 3693 }, { "epoch": 1.42, "learning_rate": 4.187952759688968e-06, "loss": 1.0766, "step": 3694 }, { "epoch": 1.42, "learning_rate": 4.182837850625898e-06, "loss": 1.0627, "step": 3695 }, { "epoch": 1.42, "learning_rate": 4.177725240844633e-06, "loss": 1.0999, "step": 3696 }, { "epoch": 1.42, "learning_rate": 4.172614932365967e-06, "loss": 1.1034, "step": 3697 }, { "epoch": 1.42, "learning_rate": 4.1675069272097945e-06, "loss": 1.0654, "step": 3698 }, { "epoch": 1.42, "learning_rate": 4.162401227395099e-06, "loss": 1.0915, "step": 3699 }, { "epoch": 1.42, "learning_rate": 4.1572978349399495e-06, "loss": 1.077, "step": 3700 }, { "epoch": 1.42, "learning_rate": 4.152196751861508e-06, "loss": 1.0689, "step": 3701 }, { "epoch": 1.42, "learning_rate": 4.1470979801760104e-06, "loss": 1.1115, "step": 3702 }, { "epoch": 1.42, "learning_rate": 4.142001521898802e-06, "loss": 1.0981, "step": 3703 }, { "epoch": 1.42, "learning_rate": 4.13690737904429e-06, "loss": 1.1324, "step": 3704 }, { "epoch": 1.42, "learning_rate": 4.131815553625979e-06, "loss": 1.137, "step": 3705 }, { "epoch": 1.42, "learning_rate": 4.12672604765646e-06, "loss": 1.1123, "step": 3706 }, { "epoch": 1.42, "learning_rate": 4.121638863147397e-06, "loss": 1.103, "step": 3707 }, { "epoch": 1.42, "learning_rate": 4.116554002109545e-06, "loss": 1.0835, "step": 3708 }, { "epoch": 1.42, "learning_rate": 4.11147146655274e-06, "loss": 1.0973, "step": 3709 }, { "epoch": 1.43, "learning_rate": 4.106391258485891e-06, "loss": 1.1792, "step": 3710 }, { "epoch": 1.43, "learning_rate": 4.101313379916995e-06, "loss": 1.055, "step": 3711 }, { "epoch": 1.43, "learning_rate": 4.096237832853126e-06, "loss": 1.019, "step": 3712 }, { "epoch": 1.43, "learning_rate": 4.091164619300436e-06, "loss": 1.1365, "step": 3713 }, { "epoch": 1.43, "learning_rate": 4.08609374126416e-06, "loss": 1.0983, "step": 3714 }, { "epoch": 1.43, "learning_rate": 4.081025200748592e-06, "loss": 1.0865, "step": 3715 }, { "epoch": 1.43, "learning_rate": 4.075958999757129e-06, "loss": 1.0988, "step": 3716 }, { "epoch": 1.43, "learning_rate": 4.070895140292221e-06, "loss": 1.1097, "step": 3717 }, { "epoch": 1.43, "learning_rate": 4.0658336243554016e-06, "loss": 1.1432, "step": 3718 }, { "epoch": 1.43, "learning_rate": 4.060774453947279e-06, "loss": 1.0826, "step": 3719 }, { "epoch": 1.43, "learning_rate": 4.0557176310675315e-06, "loss": 1.0682, "step": 3720 }, { "epoch": 1.43, "learning_rate": 4.0506631577149145e-06, "loss": 1.1282, "step": 3721 }, { "epoch": 1.43, "learning_rate": 4.045611035887243e-06, "loss": 1.031, "step": 3722 }, { "epoch": 1.43, "learning_rate": 4.040561267581413e-06, "loss": 1.0639, "step": 3723 }, { "epoch": 1.43, "learning_rate": 4.035513854793389e-06, "loss": 1.0964, "step": 3724 }, { "epoch": 1.43, "learning_rate": 4.0304687995182035e-06, "loss": 1.1417, "step": 3725 }, { "epoch": 1.43, "learning_rate": 4.025426103749955e-06, "loss": 1.085, "step": 3726 }, { "epoch": 1.43, "learning_rate": 4.020385769481815e-06, "loss": 1.1312, "step": 3727 }, { "epoch": 1.43, "learning_rate": 4.015347798706007e-06, "loss": 1.0947, "step": 3728 }, { "epoch": 1.43, "learning_rate": 4.010312193413847e-06, "loss": 1.0943, "step": 3729 }, { "epoch": 1.43, "learning_rate": 4.005278955595685e-06, "loss": 1.12, "step": 3730 }, { "epoch": 1.43, "learning_rate": 4.000248087240958e-06, "loss": 1.0965, "step": 3731 }, { "epoch": 1.43, "learning_rate": 3.995219590338156e-06, "loss": 1.0897, "step": 3732 }, { "epoch": 1.43, "learning_rate": 3.990193466874835e-06, "loss": 1.0892, "step": 3733 }, { "epoch": 1.43, "learning_rate": 3.985169718837617e-06, "loss": 1.0761, "step": 3734 }, { "epoch": 1.43, "learning_rate": 3.9801483482121714e-06, "loss": 1.1118, "step": 3735 }, { "epoch": 1.44, "learning_rate": 3.975129356983241e-06, "loss": 1.0696, "step": 3736 }, { "epoch": 1.44, "learning_rate": 3.970112747134625e-06, "loss": 1.0893, "step": 3737 }, { "epoch": 1.44, "learning_rate": 3.965098520649178e-06, "loss": 1.1251, "step": 3738 }, { "epoch": 1.44, "learning_rate": 3.960086679508817e-06, "loss": 1.0355, "step": 3739 }, { "epoch": 1.44, "learning_rate": 3.955077225694515e-06, "loss": 1.1401, "step": 3740 }, { "epoch": 1.44, "learning_rate": 3.9500701611862915e-06, "loss": 1.0988, "step": 3741 }, { "epoch": 1.44, "learning_rate": 3.945065487963245e-06, "loss": 1.2135, "step": 3742 }, { "epoch": 1.44, "learning_rate": 3.940063208003503e-06, "loss": 1.0665, "step": 3743 }, { "epoch": 1.44, "learning_rate": 3.935063323284261e-06, "loss": 1.1342, "step": 3744 }, { "epoch": 1.44, "learning_rate": 3.9300658357817705e-06, "loss": 1.1098, "step": 3745 }, { "epoch": 1.44, "learning_rate": 3.925070747471317e-06, "loss": 1.0069, "step": 3746 }, { "epoch": 1.44, "learning_rate": 3.920078060327267e-06, "loss": 1.1227, "step": 3747 }, { "epoch": 1.44, "learning_rate": 3.91508777632301e-06, "loss": 1.0651, "step": 3748 }, { "epoch": 1.44, "learning_rate": 3.9100998974310025e-06, "loss": 1.0772, "step": 3749 }, { "epoch": 1.44, "learning_rate": 3.905114425622742e-06, "loss": 1.1139, "step": 3750 }, { "epoch": 1.44, "learning_rate": 3.900131362868784e-06, "loss": 1.0689, "step": 3751 }, { "epoch": 1.44, "learning_rate": 3.895150711138715e-06, "loss": 1.0852, "step": 3752 }, { "epoch": 1.44, "learning_rate": 3.890172472401193e-06, "loss": 1.0748, "step": 3753 }, { "epoch": 1.44, "learning_rate": 3.885196648623896e-06, "loss": 1.0968, "step": 3754 }, { "epoch": 1.44, "learning_rate": 3.880223241773564e-06, "loss": 1.1221, "step": 3755 }, { "epoch": 1.44, "learning_rate": 3.8752522538159785e-06, "loss": 1.0905, "step": 3756 }, { "epoch": 1.44, "learning_rate": 3.870283686715963e-06, "loss": 1.1053, "step": 3757 }, { "epoch": 1.44, "learning_rate": 3.86531754243739e-06, "loss": 1.1394, "step": 3758 }, { "epoch": 1.44, "learning_rate": 3.86035382294316e-06, "loss": 1.1407, "step": 3759 }, { "epoch": 1.44, "learning_rate": 3.855392530195231e-06, "loss": 1.1154, "step": 3760 }, { "epoch": 1.44, "learning_rate": 3.850433666154592e-06, "loss": 1.0791, "step": 3761 }, { "epoch": 1.45, "learning_rate": 3.845477232781278e-06, "loss": 1.0781, "step": 3762 }, { "epoch": 1.45, "learning_rate": 3.8405232320343585e-06, "loss": 1.1021, "step": 3763 }, { "epoch": 1.45, "learning_rate": 3.835571665871949e-06, "loss": 1.119, "step": 3764 }, { "epoch": 1.45, "learning_rate": 3.830622536251185e-06, "loss": 1.1347, "step": 3765 }, { "epoch": 1.45, "learning_rate": 3.825675845128267e-06, "loss": 1.1074, "step": 3766 }, { "epoch": 1.45, "learning_rate": 3.820731594458405e-06, "loss": 1.1041, "step": 3767 }, { "epoch": 1.45, "learning_rate": 3.815789786195856e-06, "loss": 1.0741, "step": 3768 }, { "epoch": 1.45, "learning_rate": 3.810850422293918e-06, "loss": 1.0667, "step": 3769 }, { "epoch": 1.45, "learning_rate": 3.8059135047049035e-06, "loss": 1.1089, "step": 3770 }, { "epoch": 1.45, "learning_rate": 3.8009790353801844e-06, "loss": 1.1089, "step": 3771 }, { "epoch": 1.45, "learning_rate": 3.7960470162701413e-06, "loss": 1.0696, "step": 3772 }, { "epoch": 1.45, "learning_rate": 3.7911174493241977e-06, "loss": 1.1133, "step": 3773 }, { "epoch": 1.45, "learning_rate": 3.786190336490807e-06, "loss": 1.1444, "step": 3774 }, { "epoch": 1.45, "learning_rate": 3.781265679717452e-06, "loss": 1.1071, "step": 3775 }, { "epoch": 1.45, "learning_rate": 3.7763434809506428e-06, "loss": 1.1313, "step": 3776 }, { "epoch": 1.45, "learning_rate": 3.7714237421359246e-06, "loss": 1.0746, "step": 3777 }, { "epoch": 1.45, "learning_rate": 3.766506465217854e-06, "loss": 1.1043, "step": 3778 }, { "epoch": 1.45, "learning_rate": 3.76159165214004e-06, "loss": 1.1314, "step": 3779 }, { "epoch": 1.45, "learning_rate": 3.7566793048450922e-06, "loss": 1.0626, "step": 3780 }, { "epoch": 1.45, "learning_rate": 3.7517694252746608e-06, "loss": 1.0638, "step": 3781 }, { "epoch": 1.45, "learning_rate": 3.746862015369419e-06, "loss": 1.0949, "step": 3782 }, { "epoch": 1.45, "learning_rate": 3.7419570770690526e-06, "loss": 1.043, "step": 3783 }, { "epoch": 1.45, "learning_rate": 3.737054612312293e-06, "loss": 1.0683, "step": 3784 }, { "epoch": 1.45, "learning_rate": 3.7321546230368687e-06, "loss": 1.1313, "step": 3785 }, { "epoch": 1.45, "learning_rate": 3.7272571111795462e-06, "loss": 1.103, "step": 3786 }, { "epoch": 1.45, "learning_rate": 3.7223620786761072e-06, "loss": 1.0972, "step": 3787 }, { "epoch": 1.46, "learning_rate": 3.7174695274613535e-06, "loss": 1.0926, "step": 3788 }, { "epoch": 1.46, "learning_rate": 3.712579459469108e-06, "loss": 1.1351, "step": 3789 }, { "epoch": 1.46, "learning_rate": 3.707691876632212e-06, "loss": 1.113, "step": 3790 }, { "epoch": 1.46, "learning_rate": 3.702806780882515e-06, "loss": 1.1193, "step": 3791 }, { "epoch": 1.46, "learning_rate": 3.6979241741509066e-06, "loss": 1.0899, "step": 3792 }, { "epoch": 1.46, "learning_rate": 3.6930440583672646e-06, "loss": 1.1112, "step": 3793 }, { "epoch": 1.46, "learning_rate": 3.6881664354604996e-06, "loss": 1.1344, "step": 3794 }, { "epoch": 1.46, "learning_rate": 3.6832913073585375e-06, "loss": 1.1367, "step": 3795 }, { "epoch": 1.46, "learning_rate": 3.6784186759883023e-06, "loss": 1.1126, "step": 3796 }, { "epoch": 1.46, "learning_rate": 3.6735485432757555e-06, "loss": 1.0816, "step": 3797 }, { "epoch": 1.46, "learning_rate": 3.6686809111458477e-06, "loss": 1.0749, "step": 3798 }, { "epoch": 1.46, "learning_rate": 3.663815781522554e-06, "loss": 1.0914, "step": 3799 }, { "epoch": 1.46, "learning_rate": 3.658953156328857e-06, "loss": 1.1066, "step": 3800 }, { "epoch": 1.46, "learning_rate": 3.654093037486752e-06, "loss": 1.0556, "step": 3801 }, { "epoch": 1.46, "learning_rate": 3.6492354269172402e-06, "loss": 1.052, "step": 3802 }, { "epoch": 1.46, "learning_rate": 3.6443803265403356e-06, "loss": 1.1095, "step": 3803 }, { "epoch": 1.46, "learning_rate": 3.6395277382750495e-06, "loss": 1.0486, "step": 3804 }, { "epoch": 1.46, "learning_rate": 3.6346776640394133e-06, "loss": 1.1453, "step": 3805 }, { "epoch": 1.46, "learning_rate": 3.6298301057504583e-06, "loss": 1.0726, "step": 3806 }, { "epoch": 1.46, "learning_rate": 3.6249850653242226e-06, "loss": 1.0699, "step": 3807 }, { "epoch": 1.46, "learning_rate": 3.620142544675751e-06, "loss": 1.0662, "step": 3808 }, { "epoch": 1.46, "learning_rate": 3.6153025457190803e-06, "loss": 1.0685, "step": 3809 }, { "epoch": 1.46, "learning_rate": 3.610465070367276e-06, "loss": 1.1328, "step": 3810 }, { "epoch": 1.46, "learning_rate": 3.6056301205323787e-06, "loss": 1.0666, "step": 3811 }, { "epoch": 1.46, "learning_rate": 3.6007976981254445e-06, "loss": 1.1114, "step": 3812 }, { "epoch": 1.46, "learning_rate": 3.5959678050565318e-06, "loss": 1.0474, "step": 3813 }, { "epoch": 1.47, "learning_rate": 3.5911404432346943e-06, "loss": 1.1031, "step": 3814 }, { "epoch": 1.47, "learning_rate": 3.586315614567987e-06, "loss": 1.021, "step": 3815 }, { "epoch": 1.47, "learning_rate": 3.581493320963466e-06, "loss": 1.0966, "step": 3816 }, { "epoch": 1.47, "learning_rate": 3.576673564327179e-06, "loss": 1.14, "step": 3817 }, { "epoch": 1.47, "learning_rate": 3.5718563465641763e-06, "loss": 1.0951, "step": 3818 }, { "epoch": 1.47, "learning_rate": 3.567041669578507e-06, "loss": 1.1078, "step": 3819 }, { "epoch": 1.47, "learning_rate": 3.562229535273203e-06, "loss": 1.0617, "step": 3820 }, { "epoch": 1.47, "learning_rate": 3.557419945550312e-06, "loss": 1.0817, "step": 3821 }, { "epoch": 1.47, "learning_rate": 3.552612902310857e-06, "loss": 1.1115, "step": 3822 }, { "epoch": 1.47, "learning_rate": 3.5478084074548637e-06, "loss": 1.0647, "step": 3823 }, { "epoch": 1.47, "learning_rate": 3.543006462881349e-06, "loss": 1.1143, "step": 3824 }, { "epoch": 1.47, "learning_rate": 3.53820707048832e-06, "loss": 1.1007, "step": 3825 }, { "epoch": 1.47, "learning_rate": 3.5334102321727792e-06, "loss": 1.0786, "step": 3826 }, { "epoch": 1.47, "learning_rate": 3.5286159498307192e-06, "loss": 1.0962, "step": 3827 }, { "epoch": 1.47, "learning_rate": 3.523824225357112e-06, "loss": 1.1264, "step": 3828 }, { "epoch": 1.47, "learning_rate": 3.5190350606459323e-06, "loss": 1.1123, "step": 3829 }, { "epoch": 1.47, "learning_rate": 3.514248457590136e-06, "loss": 1.1006, "step": 3830 }, { "epoch": 1.47, "learning_rate": 3.5094644180816674e-06, "loss": 1.1216, "step": 3831 }, { "epoch": 1.47, "learning_rate": 3.5046829440114617e-06, "loss": 1.0832, "step": 3832 }, { "epoch": 1.47, "learning_rate": 3.499904037269427e-06, "loss": 1.0744, "step": 3833 }, { "epoch": 1.47, "learning_rate": 3.4951276997444784e-06, "loss": 1.0812, "step": 3834 }, { "epoch": 1.47, "learning_rate": 3.490353933324494e-06, "loss": 1.0924, "step": 3835 }, { "epoch": 1.47, "learning_rate": 3.4855827398963484e-06, "loss": 1.113, "step": 3836 }, { "epoch": 1.47, "learning_rate": 3.4808141213458936e-06, "loss": 1.0883, "step": 3837 }, { "epoch": 1.47, "learning_rate": 3.4760480795579676e-06, "loss": 1.1031, "step": 3838 }, { "epoch": 1.47, "learning_rate": 3.4712846164163872e-06, "loss": 1.1046, "step": 3839 }, { "epoch": 1.48, "learning_rate": 3.4665237338039547e-06, "loss": 1.111, "step": 3840 }, { "epoch": 1.48, "learning_rate": 3.461765433602442e-06, "loss": 1.0734, "step": 3841 }, { "epoch": 1.48, "learning_rate": 3.45700971769261e-06, "loss": 1.0886, "step": 3842 }, { "epoch": 1.48, "learning_rate": 3.4522565879541946e-06, "loss": 1.0872, "step": 3843 }, { "epoch": 1.48, "learning_rate": 3.4475060462659105e-06, "loss": 1.0581, "step": 3844 }, { "epoch": 1.48, "learning_rate": 3.442758094505452e-06, "loss": 1.0827, "step": 3845 }, { "epoch": 1.48, "learning_rate": 3.4380127345494772e-06, "loss": 1.1345, "step": 3846 }, { "epoch": 1.48, "learning_rate": 3.4332699682736413e-06, "loss": 1.1021, "step": 3847 }, { "epoch": 1.48, "learning_rate": 3.4285297975525533e-06, "loss": 1.0601, "step": 3848 }, { "epoch": 1.48, "learning_rate": 3.4237922242598078e-06, "loss": 1.0958, "step": 3849 }, { "epoch": 1.48, "learning_rate": 3.419057250267971e-06, "loss": 1.1784, "step": 3850 }, { "epoch": 1.48, "learning_rate": 3.4143248774485793e-06, "loss": 1.0571, "step": 3851 }, { "epoch": 1.48, "learning_rate": 3.409595107672148e-06, "loss": 1.0623, "step": 3852 }, { "epoch": 1.48, "learning_rate": 3.404867942808149e-06, "loss": 1.1309, "step": 3853 }, { "epoch": 1.48, "learning_rate": 3.4001433847250374e-06, "loss": 1.0716, "step": 3854 }, { "epoch": 1.48, "learning_rate": 3.395421435290236e-06, "loss": 1.0603, "step": 3855 }, { "epoch": 1.48, "learning_rate": 3.390702096370133e-06, "loss": 1.0798, "step": 3856 }, { "epoch": 1.48, "learning_rate": 3.385985369830086e-06, "loss": 1.1185, "step": 3857 }, { "epoch": 1.48, "learning_rate": 3.3812712575344252e-06, "loss": 1.064, "step": 3858 }, { "epoch": 1.48, "learning_rate": 3.376559761346432e-06, "loss": 1.1158, "step": 3859 }, { "epoch": 1.48, "learning_rate": 3.371850883128378e-06, "loss": 1.0717, "step": 3860 }, { "epoch": 1.48, "learning_rate": 3.3671446247414774e-06, "loss": 1.0784, "step": 3861 }, { "epoch": 1.48, "learning_rate": 3.3624409880459196e-06, "loss": 1.041, "step": 3862 }, { "epoch": 1.48, "learning_rate": 3.3577399749008564e-06, "loss": 1.0988, "step": 3863 }, { "epoch": 1.48, "learning_rate": 3.353041587164404e-06, "loss": 1.1254, "step": 3864 }, { "epoch": 1.48, "learning_rate": 3.348345826693642e-06, "loss": 1.0873, "step": 3865 }, { "epoch": 1.49, "learning_rate": 3.3436526953446013e-06, "loss": 1.0852, "step": 3866 }, { "epoch": 1.49, "learning_rate": 3.338962194972285e-06, "loss": 1.1029, "step": 3867 }, { "epoch": 1.49, "learning_rate": 3.3342743274306533e-06, "loss": 1.1339, "step": 3868 }, { "epoch": 1.49, "learning_rate": 3.329589094572623e-06, "loss": 1.0441, "step": 3869 }, { "epoch": 1.49, "learning_rate": 3.3249064982500735e-06, "loss": 1.0698, "step": 3870 }, { "epoch": 1.49, "learning_rate": 3.3202265403138424e-06, "loss": 1.1055, "step": 3871 }, { "epoch": 1.49, "learning_rate": 3.315549222613711e-06, "loss": 1.0888, "step": 3872 }, { "epoch": 1.49, "learning_rate": 3.3108745469984436e-06, "loss": 1.0489, "step": 3873 }, { "epoch": 1.49, "learning_rate": 3.306202515315733e-06, "loss": 1.1287, "step": 3874 }, { "epoch": 1.49, "learning_rate": 3.301533129412243e-06, "loss": 1.0703, "step": 3875 }, { "epoch": 1.49, "learning_rate": 3.2968663911335897e-06, "loss": 1.1197, "step": 3876 }, { "epoch": 1.49, "learning_rate": 3.292202302324331e-06, "loss": 1.1281, "step": 3877 }, { "epoch": 1.49, "learning_rate": 3.2875408648279996e-06, "loss": 1.099, "step": 3878 }, { "epoch": 1.49, "learning_rate": 3.2828820804870577e-06, "loss": 1.1274, "step": 3879 }, { "epoch": 1.49, "learning_rate": 3.2782259511429326e-06, "loss": 1.1118, "step": 3880 }, { "epoch": 1.49, "learning_rate": 3.273572478635998e-06, "loss": 1.1276, "step": 3881 }, { "epoch": 1.49, "learning_rate": 3.2689216648055766e-06, "loss": 1.0455, "step": 3882 }, { "epoch": 1.49, "learning_rate": 3.2642735114899416e-06, "loss": 1.0732, "step": 3883 }, { "epoch": 1.49, "learning_rate": 3.2596280205263197e-06, "loss": 1.1002, "step": 3884 }, { "epoch": 1.49, "learning_rate": 3.254985193750867e-06, "loss": 1.1456, "step": 3885 }, { "epoch": 1.49, "learning_rate": 3.2503450329987153e-06, "loss": 1.173, "step": 3886 }, { "epoch": 1.49, "learning_rate": 3.245707540103914e-06, "loss": 1.0655, "step": 3887 }, { "epoch": 1.49, "learning_rate": 3.2410727168994773e-06, "loss": 1.1017, "step": 3888 }, { "epoch": 1.49, "learning_rate": 3.236440565217358e-06, "loss": 1.0742, "step": 3889 }, { "epoch": 1.49, "learning_rate": 3.231811086888449e-06, "loss": 1.1059, "step": 3890 }, { "epoch": 1.49, "learning_rate": 3.2271842837425917e-06, "loss": 1.082, "step": 3891 }, { "epoch": 1.5, "learning_rate": 3.2225601576085674e-06, "loss": 1.1743, "step": 3892 }, { "epoch": 1.5, "learning_rate": 3.2179387103141037e-06, "loss": 1.0839, "step": 3893 }, { "epoch": 1.5, "learning_rate": 3.213319943685863e-06, "loss": 1.093, "step": 3894 }, { "epoch": 1.5, "learning_rate": 3.2087038595494567e-06, "loss": 1.0946, "step": 3895 }, { "epoch": 1.5, "learning_rate": 3.2040904597294196e-06, "loss": 1.1066, "step": 3896 }, { "epoch": 1.5, "learning_rate": 3.19947974604925e-06, "loss": 1.0756, "step": 3897 }, { "epoch": 1.5, "learning_rate": 3.1948717203313606e-06, "loss": 1.1048, "step": 3898 }, { "epoch": 1.5, "learning_rate": 3.190266384397116e-06, "loss": 1.0757, "step": 3899 }, { "epoch": 1.5, "learning_rate": 3.1856637400668165e-06, "loss": 1.0795, "step": 3900 }, { "epoch": 1.5, "learning_rate": 3.1810637891596854e-06, "loss": 1.0933, "step": 3901 }, { "epoch": 1.5, "learning_rate": 3.176466533493905e-06, "loss": 1.1151, "step": 3902 }, { "epoch": 1.5, "learning_rate": 3.171871974886569e-06, "loss": 1.0317, "step": 3903 }, { "epoch": 1.5, "learning_rate": 3.167280115153718e-06, "loss": 1.114, "step": 3904 }, { "epoch": 1.5, "learning_rate": 3.1626909561103235e-06, "loss": 1.1361, "step": 3905 }, { "epoch": 1.5, "learning_rate": 3.158104499570288e-06, "loss": 1.095, "step": 3906 }, { "epoch": 1.5, "learning_rate": 3.153520747346448e-06, "loss": 1.07, "step": 3907 }, { "epoch": 1.5, "learning_rate": 3.1489397012505718e-06, "loss": 1.1242, "step": 3908 }, { "epoch": 1.5, "learning_rate": 3.144361363093347e-06, "loss": 1.1085, "step": 3909 }, { "epoch": 1.5, "learning_rate": 3.1397857346844116e-06, "loss": 1.0994, "step": 3910 }, { "epoch": 1.5, "learning_rate": 3.1352128178323114e-06, "loss": 1.0701, "step": 3911 }, { "epoch": 1.5, "learning_rate": 3.1306426143445357e-06, "loss": 1.0785, "step": 3912 }, { "epoch": 1.5, "learning_rate": 3.1260751260274957e-06, "loss": 1.0646, "step": 3913 }, { "epoch": 1.5, "learning_rate": 3.12151035468652e-06, "loss": 1.1236, "step": 3914 }, { "epoch": 1.5, "learning_rate": 3.116948302125887e-06, "loss": 1.1201, "step": 3915 }, { "epoch": 1.5, "learning_rate": 3.1123889701487755e-06, "loss": 1.063, "step": 3916 }, { "epoch": 1.5, "learning_rate": 3.1078323605573013e-06, "loss": 1.0477, "step": 3917 }, { "epoch": 1.51, "learning_rate": 3.103278475152505e-06, "loss": 1.0791, "step": 3918 }, { "epoch": 1.51, "learning_rate": 3.0987273157343447e-06, "loss": 1.025, "step": 3919 }, { "epoch": 1.51, "learning_rate": 3.0941788841017062e-06, "loss": 1.0729, "step": 3920 }, { "epoch": 1.51, "learning_rate": 3.0896331820523982e-06, "loss": 1.0679, "step": 3921 }, { "epoch": 1.51, "learning_rate": 3.085090211383136e-06, "loss": 1.138, "step": 3922 }, { "epoch": 1.51, "learning_rate": 3.080549973889582e-06, "loss": 1.0603, "step": 3923 }, { "epoch": 1.51, "learning_rate": 3.07601247136629e-06, "loss": 1.0992, "step": 3924 }, { "epoch": 1.51, "learning_rate": 3.0714777056067514e-06, "loss": 1.1031, "step": 3925 }, { "epoch": 1.51, "learning_rate": 3.0669456784033725e-06, "loss": 1.0355, "step": 3926 }, { "epoch": 1.51, "learning_rate": 3.0624163915474657e-06, "loss": 1.1093, "step": 3927 }, { "epoch": 1.51, "learning_rate": 3.057889846829282e-06, "loss": 1.0944, "step": 3928 }, { "epoch": 1.51, "learning_rate": 3.0533660460379654e-06, "loss": 1.0368, "step": 3929 }, { "epoch": 1.51, "learning_rate": 3.04884499096159e-06, "loss": 1.0824, "step": 3930 }, { "epoch": 1.51, "learning_rate": 3.04432668338714e-06, "loss": 1.1026, "step": 3931 }, { "epoch": 1.51, "learning_rate": 3.0398111251005146e-06, "loss": 1.0694, "step": 3932 }, { "epoch": 1.51, "learning_rate": 3.035298317886526e-06, "loss": 1.0618, "step": 3933 }, { "epoch": 1.51, "learning_rate": 3.030788263528901e-06, "loss": 1.1008, "step": 3934 }, { "epoch": 1.51, "learning_rate": 3.0262809638102664e-06, "loss": 1.0943, "step": 3935 }, { "epoch": 1.51, "learning_rate": 3.0217764205121848e-06, "loss": 1.0875, "step": 3936 }, { "epoch": 1.51, "learning_rate": 3.0172746354151028e-06, "loss": 1.1419, "step": 3937 }, { "epoch": 1.51, "learning_rate": 3.0127756102983928e-06, "loss": 1.0989, "step": 3938 }, { "epoch": 1.51, "learning_rate": 3.008279346940335e-06, "loss": 1.1076, "step": 3939 }, { "epoch": 1.51, "learning_rate": 3.0037858471181047e-06, "loss": 1.0866, "step": 3940 }, { "epoch": 1.51, "learning_rate": 2.9992951126078087e-06, "loss": 1.0921, "step": 3941 }, { "epoch": 1.51, "learning_rate": 2.9948071451844373e-06, "loss": 1.0853, "step": 3942 }, { "epoch": 1.51, "learning_rate": 2.9903219466219002e-06, "loss": 1.0688, "step": 3943 }, { "epoch": 1.52, "learning_rate": 2.98583951869301e-06, "loss": 1.1045, "step": 3944 }, { "epoch": 1.52, "learning_rate": 2.981359863169483e-06, "loss": 1.11, "step": 3945 }, { "epoch": 1.52, "learning_rate": 2.9768829818219403e-06, "loss": 1.0637, "step": 3946 }, { "epoch": 1.52, "learning_rate": 2.97240887641991e-06, "loss": 1.1304, "step": 3947 }, { "epoch": 1.52, "learning_rate": 2.967937548731813e-06, "loss": 1.1034, "step": 3948 }, { "epoch": 1.52, "learning_rate": 2.9634690005249824e-06, "loss": 1.1276, "step": 3949 }, { "epoch": 1.52, "learning_rate": 2.9590032335656483e-06, "loss": 1.0975, "step": 3950 }, { "epoch": 1.52, "learning_rate": 2.9545402496189435e-06, "loss": 1.0725, "step": 3951 }, { "epoch": 1.52, "learning_rate": 2.9500800504489022e-06, "loss": 1.117, "step": 3952 }, { "epoch": 1.52, "learning_rate": 2.945622637818446e-06, "loss": 1.0681, "step": 3953 }, { "epoch": 1.52, "learning_rate": 2.941168013489416e-06, "loss": 1.0751, "step": 3954 }, { "epoch": 1.52, "learning_rate": 2.936716179222531e-06, "loss": 1.0951, "step": 3955 }, { "epoch": 1.52, "learning_rate": 2.932267136777419e-06, "loss": 1.1246, "step": 3956 }, { "epoch": 1.52, "learning_rate": 2.9278208879126003e-06, "loss": 1.0929, "step": 3957 }, { "epoch": 1.52, "learning_rate": 2.923377434385496e-06, "loss": 1.1038, "step": 3958 }, { "epoch": 1.52, "learning_rate": 2.918936777952407e-06, "loss": 1.0664, "step": 3959 }, { "epoch": 1.52, "learning_rate": 2.9144989203685535e-06, "loss": 1.0798, "step": 3960 }, { "epoch": 1.52, "learning_rate": 2.9100638633880264e-06, "loss": 1.084, "step": 3961 }, { "epoch": 1.52, "learning_rate": 2.9056316087638213e-06, "loss": 1.1062, "step": 3962 }, { "epoch": 1.52, "learning_rate": 2.9012021582478277e-06, "loss": 1.0818, "step": 3963 }, { "epoch": 1.52, "learning_rate": 2.896775513590814e-06, "loss": 1.1035, "step": 3964 }, { "epoch": 1.52, "learning_rate": 2.8923516765424586e-06, "loss": 1.0886, "step": 3965 }, { "epoch": 1.52, "learning_rate": 2.8879306488513114e-06, "loss": 1.1258, "step": 3966 }, { "epoch": 1.52, "learning_rate": 2.8835124322648255e-06, "loss": 1.1067, "step": 3967 }, { "epoch": 1.52, "learning_rate": 2.879097028529336e-06, "loss": 1.1266, "step": 3968 }, { "epoch": 1.52, "learning_rate": 2.8746844393900686e-06, "loss": 1.0858, "step": 3969 }, { "epoch": 1.53, "learning_rate": 2.870274666591135e-06, "loss": 1.0913, "step": 3970 }, { "epoch": 1.53, "learning_rate": 2.8658677118755384e-06, "loss": 1.0708, "step": 3971 }, { "epoch": 1.53, "learning_rate": 2.8614635769851595e-06, "loss": 1.0537, "step": 3972 }, { "epoch": 1.53, "learning_rate": 2.85706226366077e-06, "loss": 1.127, "step": 3973 }, { "epoch": 1.53, "learning_rate": 2.8526637736420273e-06, "loss": 1.0824, "step": 3974 }, { "epoch": 1.53, "learning_rate": 2.8482681086674703e-06, "loss": 1.0632, "step": 3975 }, { "epoch": 1.53, "learning_rate": 2.8438752704745255e-06, "loss": 1.1131, "step": 3976 }, { "epoch": 1.53, "learning_rate": 2.8394852607994894e-06, "loss": 1.0702, "step": 3977 }, { "epoch": 1.53, "learning_rate": 2.835098081377564e-06, "loss": 1.1228, "step": 3978 }, { "epoch": 1.53, "learning_rate": 2.830713733942807e-06, "loss": 1.0498, "step": 3979 }, { "epoch": 1.53, "learning_rate": 2.826332220228172e-06, "loss": 1.0891, "step": 3980 }, { "epoch": 1.53, "learning_rate": 2.8219535419654886e-06, "loss": 1.0664, "step": 3981 }, { "epoch": 1.53, "learning_rate": 2.8175777008854666e-06, "loss": 1.1113, "step": 3982 }, { "epoch": 1.53, "learning_rate": 2.8132046987176943e-06, "loss": 1.0913, "step": 3983 }, { "epoch": 1.53, "learning_rate": 2.808834537190638e-06, "loss": 1.0979, "step": 3984 }, { "epoch": 1.53, "learning_rate": 2.8044672180316357e-06, "loss": 1.0549, "step": 3985 }, { "epoch": 1.53, "learning_rate": 2.8001027429669094e-06, "loss": 1.0836, "step": 3986 }, { "epoch": 1.53, "learning_rate": 2.795741113721555e-06, "loss": 1.1458, "step": 3987 }, { "epoch": 1.53, "learning_rate": 2.7913823320195423e-06, "loss": 1.1364, "step": 3988 }, { "epoch": 1.53, "learning_rate": 2.7870263995837187e-06, "loss": 1.1022, "step": 3989 }, { "epoch": 1.53, "learning_rate": 2.7826733181357934e-06, "loss": 1.0432, "step": 3990 }, { "epoch": 1.53, "learning_rate": 2.778323089396372e-06, "loss": 1.1495, "step": 3991 }, { "epoch": 1.53, "learning_rate": 2.773975715084908e-06, "loss": 1.0959, "step": 3992 }, { "epoch": 1.53, "learning_rate": 2.7696311969197408e-06, "loss": 1.0885, "step": 3993 }, { "epoch": 1.53, "learning_rate": 2.7652895366180776e-06, "loss": 1.0905, "step": 3994 }, { "epoch": 1.53, "learning_rate": 2.7609507358959953e-06, "loss": 1.1033, "step": 3995 }, { "epoch": 1.54, "learning_rate": 2.7566147964684443e-06, "loss": 1.0638, "step": 3996 }, { "epoch": 1.54, "learning_rate": 2.7522817200492347e-06, "loss": 1.1079, "step": 3997 }, { "epoch": 1.54, "learning_rate": 2.7479515083510546e-06, "loss": 1.0716, "step": 3998 }, { "epoch": 1.54, "learning_rate": 2.7436241630854554e-06, "loss": 1.1343, "step": 3999 }, { "epoch": 1.54, "learning_rate": 2.7392996859628573e-06, "loss": 1.0868, "step": 4000 }, { "epoch": 1.54, "eval_loss": 1.1051383018493652, "eval_runtime": 688.3006, "eval_samples_per_second": 4.941, "eval_steps_per_second": 0.309, "step": 4000 }, { "epoch": 1.54, "learning_rate": 2.734978078692544e-06, "loss": 1.0993, "step": 4001 }, { "epoch": 1.54, "learning_rate": 2.7306593429826722e-06, "loss": 1.1337, "step": 4002 }, { "epoch": 1.54, "learning_rate": 2.7263434805402476e-06, "loss": 1.1231, "step": 4003 }, { "epoch": 1.54, "learning_rate": 2.722030493071164e-06, "loss": 1.11, "step": 4004 }, { "epoch": 1.54, "learning_rate": 2.7177203822801556e-06, "loss": 1.1294, "step": 4005 }, { "epoch": 1.54, "learning_rate": 2.7134131498708328e-06, "loss": 1.0832, "step": 4006 }, { "epoch": 1.54, "learning_rate": 2.7091087975456655e-06, "loss": 1.1297, "step": 4007 }, { "epoch": 1.54, "learning_rate": 2.7048073270059838e-06, "loss": 1.1248, "step": 4008 }, { "epoch": 1.54, "learning_rate": 2.7005087399519836e-06, "loss": 1.1391, "step": 4009 }, { "epoch": 1.54, "learning_rate": 2.696213038082712e-06, "loss": 1.1234, "step": 4010 }, { "epoch": 1.54, "learning_rate": 2.691920223096083e-06, "loss": 1.082, "step": 4011 }, { "epoch": 1.54, "learning_rate": 2.6876302966888666e-06, "loss": 1.0539, "step": 4012 }, { "epoch": 1.54, "learning_rate": 2.6833432605566934e-06, "loss": 1.0762, "step": 4013 }, { "epoch": 1.54, "learning_rate": 2.6790591163940493e-06, "loss": 1.0177, "step": 4014 }, { "epoch": 1.54, "learning_rate": 2.6747778658942824e-06, "loss": 1.0869, "step": 4015 }, { "epoch": 1.54, "learning_rate": 2.670499510749581e-06, "loss": 1.0528, "step": 4016 }, { "epoch": 1.54, "learning_rate": 2.666224052651015e-06, "loss": 1.1244, "step": 4017 }, { "epoch": 1.54, "learning_rate": 2.661951493288485e-06, "loss": 1.0796, "step": 4018 }, { "epoch": 1.54, "learning_rate": 2.6576818343507592e-06, "loss": 1.108, "step": 4019 }, { "epoch": 1.54, "learning_rate": 2.6534150775254576e-06, "loss": 1.1382, "step": 4020 }, { "epoch": 1.54, "learning_rate": 2.6491512244990435e-06, "loss": 1.1054, "step": 4021 }, { "epoch": 1.54, "learning_rate": 2.6448902769568518e-06, "loss": 1.0371, "step": 4022 }, { "epoch": 1.55, "learning_rate": 2.640632236583048e-06, "loss": 1.1194, "step": 4023 }, { "epoch": 1.55, "learning_rate": 2.6363771050606634e-06, "loss": 1.0959, "step": 4024 }, { "epoch": 1.55, "learning_rate": 2.6321248840715707e-06, "loss": 1.0607, "step": 4025 }, { "epoch": 1.55, "learning_rate": 2.6278755752965012e-06, "loss": 1.0977, "step": 4026 }, { "epoch": 1.55, "learning_rate": 2.6236291804150205e-06, "loss": 1.0426, "step": 4027 }, { "epoch": 1.55, "learning_rate": 2.6193857011055623e-06, "loss": 1.1636, "step": 4028 }, { "epoch": 1.55, "learning_rate": 2.6151451390453854e-06, "loss": 1.068, "step": 4029 }, { "epoch": 1.55, "learning_rate": 2.6109074959106206e-06, "loss": 1.0831, "step": 4030 }, { "epoch": 1.55, "learning_rate": 2.606672773376221e-06, "loss": 1.131, "step": 4031 }, { "epoch": 1.55, "learning_rate": 2.6024409731159996e-06, "loss": 1.0762, "step": 4032 }, { "epoch": 1.55, "learning_rate": 2.5982120968026135e-06, "loss": 1.073, "step": 4033 }, { "epoch": 1.55, "learning_rate": 2.5939861461075554e-06, "loss": 1.117, "step": 4034 }, { "epoch": 1.55, "learning_rate": 2.589763122701171e-06, "loss": 1.0925, "step": 4035 }, { "epoch": 1.55, "learning_rate": 2.585543028252646e-06, "loss": 1.1256, "step": 4036 }, { "epoch": 1.55, "learning_rate": 2.5813258644300056e-06, "loss": 1.0945, "step": 4037 }, { "epoch": 1.55, "learning_rate": 2.5771116329001213e-06, "loss": 1.0186, "step": 4038 }, { "epoch": 1.55, "learning_rate": 2.572900335328705e-06, "loss": 1.0173, "step": 4039 }, { "epoch": 1.55, "learning_rate": 2.5686919733802984e-06, "loss": 1.0758, "step": 4040 }, { "epoch": 1.55, "learning_rate": 2.5644865487183035e-06, "loss": 1.112, "step": 4041 }, { "epoch": 1.55, "learning_rate": 2.560284063004941e-06, "loss": 1.1257, "step": 4042 }, { "epoch": 1.55, "learning_rate": 2.5560845179012804e-06, "loss": 1.1156, "step": 4043 }, { "epoch": 1.55, "learning_rate": 2.5518879150672315e-06, "loss": 1.0848, "step": 4044 }, { "epoch": 1.55, "learning_rate": 2.547694256161526e-06, "loss": 1.1017, "step": 4045 }, { "epoch": 1.55, "learning_rate": 2.5435035428417545e-06, "loss": 1.0856, "step": 4046 }, { "epoch": 1.55, "learning_rate": 2.5393157767643228e-06, "loss": 1.0614, "step": 4047 }, { "epoch": 1.55, "learning_rate": 2.5351309595844843e-06, "loss": 1.094, "step": 4048 }, { "epoch": 1.56, "learning_rate": 2.530949092956321e-06, "loss": 1.0869, "step": 4049 }, { "epoch": 1.56, "learning_rate": 2.5267701785327512e-06, "loss": 1.0969, "step": 4050 }, { "epoch": 1.56, "learning_rate": 2.5225942179655272e-06, "loss": 1.0926, "step": 4051 }, { "epoch": 1.56, "learning_rate": 2.518421212905233e-06, "loss": 1.1072, "step": 4052 }, { "epoch": 1.56, "learning_rate": 2.5142511650012756e-06, "loss": 1.1194, "step": 4053 }, { "epoch": 1.56, "learning_rate": 2.510084075901913e-06, "loss": 1.1193, "step": 4054 }, { "epoch": 1.56, "learning_rate": 2.505919947254213e-06, "loss": 1.0682, "step": 4055 }, { "epoch": 1.56, "learning_rate": 2.5017587807040854e-06, "loss": 1.1027, "step": 4056 }, { "epoch": 1.56, "learning_rate": 2.4976005778962687e-06, "loss": 1.0706, "step": 4057 }, { "epoch": 1.56, "learning_rate": 2.493445340474319e-06, "loss": 1.0886, "step": 4058 }, { "epoch": 1.56, "learning_rate": 2.489293070080637e-06, "loss": 1.1053, "step": 4059 }, { "epoch": 1.56, "learning_rate": 2.4851437683564383e-06, "loss": 1.1376, "step": 4060 }, { "epoch": 1.56, "learning_rate": 2.480997436941769e-06, "loss": 1.0582, "step": 4061 }, { "epoch": 1.56, "learning_rate": 2.476854077475501e-06, "loss": 1.0769, "step": 4062 }, { "epoch": 1.56, "learning_rate": 2.472713691595333e-06, "loss": 1.1302, "step": 4063 }, { "epoch": 1.56, "learning_rate": 2.4685762809377856e-06, "loss": 1.1461, "step": 4064 }, { "epoch": 1.56, "learning_rate": 2.464441847138208e-06, "loss": 1.102, "step": 4065 }, { "epoch": 1.56, "learning_rate": 2.4603103918307624e-06, "loss": 1.0971, "step": 4066 }, { "epoch": 1.56, "learning_rate": 2.4561819166484515e-06, "loss": 1.1183, "step": 4067 }, { "epoch": 1.56, "learning_rate": 2.4520564232230803e-06, "loss": 1.1217, "step": 4068 }, { "epoch": 1.56, "learning_rate": 2.4479339131852876e-06, "loss": 1.133, "step": 4069 }, { "epoch": 1.56, "learning_rate": 2.443814388164534e-06, "loss": 1.1311, "step": 4070 }, { "epoch": 1.56, "learning_rate": 2.4396978497890856e-06, "loss": 1.148, "step": 4071 }, { "epoch": 1.56, "learning_rate": 2.43558429968605e-06, "loss": 1.0835, "step": 4072 }, { "epoch": 1.56, "learning_rate": 2.431473739481335e-06, "loss": 1.1305, "step": 4073 }, { "epoch": 1.56, "learning_rate": 2.4273661707996763e-06, "loss": 1.179, "step": 4074 }, { "epoch": 1.57, "learning_rate": 2.4232615952646244e-06, "loss": 1.0885, "step": 4075 }, { "epoch": 1.57, "learning_rate": 2.4191600144985463e-06, "loss": 1.0283, "step": 4076 }, { "epoch": 1.57, "learning_rate": 2.4150614301226263e-06, "loss": 1.1207, "step": 4077 }, { "epoch": 1.57, "learning_rate": 2.4109658437568673e-06, "loss": 1.0775, "step": 4078 }, { "epoch": 1.57, "learning_rate": 2.4068732570200747e-06, "loss": 1.1379, "step": 4079 }, { "epoch": 1.57, "learning_rate": 2.40278367152989e-06, "loss": 1.1275, "step": 4080 }, { "epoch": 1.57, "learning_rate": 2.3986970889027474e-06, "loss": 1.0612, "step": 4081 }, { "epoch": 1.57, "learning_rate": 2.394613510753904e-06, "loss": 1.1038, "step": 4082 }, { "epoch": 1.57, "learning_rate": 2.3905329386974318e-06, "loss": 1.1102, "step": 4083 }, { "epoch": 1.57, "learning_rate": 2.3864553743462027e-06, "loss": 1.0537, "step": 4084 }, { "epoch": 1.57, "learning_rate": 2.382380819311918e-06, "loss": 1.0622, "step": 4085 }, { "epoch": 1.57, "learning_rate": 2.378309275205073e-06, "loss": 1.0467, "step": 4086 }, { "epoch": 1.57, "learning_rate": 2.3742407436349814e-06, "loss": 1.1059, "step": 4087 }, { "epoch": 1.57, "learning_rate": 2.3701752262097642e-06, "loss": 1.0929, "step": 4088 }, { "epoch": 1.57, "learning_rate": 2.3661127245363515e-06, "loss": 1.0748, "step": 4089 }, { "epoch": 1.57, "learning_rate": 2.362053240220479e-06, "loss": 1.1273, "step": 4090 }, { "epoch": 1.57, "learning_rate": 2.3579967748666965e-06, "loss": 1.0769, "step": 4091 }, { "epoch": 1.57, "learning_rate": 2.35394333007835e-06, "loss": 1.0853, "step": 4092 }, { "epoch": 1.57, "learning_rate": 2.3498929074575993e-06, "loss": 1.1144, "step": 4093 }, { "epoch": 1.57, "learning_rate": 2.3458455086054113e-06, "loss": 1.077, "step": 4094 }, { "epoch": 1.57, "learning_rate": 2.3418011351215443e-06, "loss": 1.112, "step": 4095 }, { "epoch": 1.57, "learning_rate": 2.3377597886045834e-06, "loss": 1.0907, "step": 4096 }, { "epoch": 1.57, "learning_rate": 2.3337214706518918e-06, "loss": 1.1221, "step": 4097 }, { "epoch": 1.57, "learning_rate": 2.3296861828596605e-06, "loss": 1.1262, "step": 4098 }, { "epoch": 1.57, "learning_rate": 2.325653926822863e-06, "loss": 1.1126, "step": 4099 }, { "epoch": 1.57, "learning_rate": 2.3216247041352836e-06, "loss": 1.0683, "step": 4100 }, { "epoch": 1.58, "learning_rate": 2.3175985163895053e-06, "loss": 1.0577, "step": 4101 }, { "epoch": 1.58, "learning_rate": 2.3135753651769177e-06, "loss": 1.068, "step": 4102 }, { "epoch": 1.58, "learning_rate": 2.309555252087694e-06, "loss": 1.0855, "step": 4103 }, { "epoch": 1.58, "learning_rate": 2.305538178710831e-06, "loss": 1.0972, "step": 4104 }, { "epoch": 1.58, "learning_rate": 2.3015241466341e-06, "loss": 1.0555, "step": 4105 }, { "epoch": 1.58, "learning_rate": 2.2975131574440835e-06, "loss": 1.0765, "step": 4106 }, { "epoch": 1.58, "learning_rate": 2.2935052127261626e-06, "loss": 1.121, "step": 4107 }, { "epoch": 1.58, "learning_rate": 2.289500314064501e-06, "loss": 1.114, "step": 4108 }, { "epoch": 1.58, "learning_rate": 2.2854984630420798e-06, "loss": 1.0759, "step": 4109 }, { "epoch": 1.58, "learning_rate": 2.281499661240656e-06, "loss": 1.1039, "step": 4110 }, { "epoch": 1.58, "learning_rate": 2.2775039102407926e-06, "loss": 1.0281, "step": 4111 }, { "epoch": 1.58, "learning_rate": 2.273511211621843e-06, "loss": 1.0717, "step": 4112 }, { "epoch": 1.58, "learning_rate": 2.269521566961953e-06, "loss": 1.0238, "step": 4113 }, { "epoch": 1.58, "learning_rate": 2.2655349778380653e-06, "loss": 1.0824, "step": 4114 }, { "epoch": 1.58, "learning_rate": 2.2615514458259147e-06, "loss": 1.046, "step": 4115 }, { "epoch": 1.58, "learning_rate": 2.2575709725000183e-06, "loss": 1.1141, "step": 4116 }, { "epoch": 1.58, "learning_rate": 2.2535935594336956e-06, "loss": 1.1133, "step": 4117 }, { "epoch": 1.58, "learning_rate": 2.249619208199052e-06, "loss": 1.0799, "step": 4118 }, { "epoch": 1.58, "learning_rate": 2.2456479203669833e-06, "loss": 1.0723, "step": 4119 }, { "epoch": 1.58, "learning_rate": 2.2416796975071754e-06, "loss": 1.0819, "step": 4120 }, { "epoch": 1.58, "learning_rate": 2.237714541188094e-06, "loss": 1.0775, "step": 4121 }, { "epoch": 1.58, "learning_rate": 2.2337524529770127e-06, "loss": 1.0703, "step": 4122 }, { "epoch": 1.58, "learning_rate": 2.2297934344399696e-06, "loss": 1.0825, "step": 4123 }, { "epoch": 1.58, "learning_rate": 2.225837487141802e-06, "loss": 1.0827, "step": 4124 }, { "epoch": 1.58, "learning_rate": 2.2218846126461336e-06, "loss": 1.0598, "step": 4125 }, { "epoch": 1.58, "learning_rate": 2.2179348125153688e-06, "loss": 1.0542, "step": 4126 }, { "epoch": 1.59, "learning_rate": 2.2139880883107e-06, "loss": 1.0879, "step": 4127 }, { "epoch": 1.59, "learning_rate": 2.210044441592105e-06, "loss": 1.0942, "step": 4128 }, { "epoch": 1.59, "learning_rate": 2.206103873918338e-06, "loss": 1.0192, "step": 4129 }, { "epoch": 1.59, "learning_rate": 2.2021663868469445e-06, "loss": 1.0871, "step": 4130 }, { "epoch": 1.59, "learning_rate": 2.198231981934248e-06, "loss": 1.0824, "step": 4131 }, { "epoch": 1.59, "learning_rate": 2.1943006607353544e-06, "loss": 1.0946, "step": 4132 }, { "epoch": 1.59, "learning_rate": 2.190372424804156e-06, "loss": 1.095, "step": 4133 }, { "epoch": 1.59, "learning_rate": 2.1864472756933096e-06, "loss": 1.1056, "step": 4134 }, { "epoch": 1.59, "learning_rate": 2.1825252149542763e-06, "loss": 1.0613, "step": 4135 }, { "epoch": 1.59, "learning_rate": 2.1786062441372756e-06, "loss": 1.0711, "step": 4136 }, { "epoch": 1.59, "learning_rate": 2.1746903647913133e-06, "loss": 1.0905, "step": 4137 }, { "epoch": 1.59, "learning_rate": 2.170777578464176e-06, "loss": 1.0836, "step": 4138 }, { "epoch": 1.59, "learning_rate": 2.166867886702424e-06, "loss": 1.0886, "step": 4139 }, { "epoch": 1.59, "learning_rate": 2.1629612910513977e-06, "loss": 1.0934, "step": 4140 }, { "epoch": 1.59, "learning_rate": 2.159057793055207e-06, "loss": 1.099, "step": 4141 }, { "epoch": 1.59, "learning_rate": 2.155157394256745e-06, "loss": 1.0902, "step": 4142 }, { "epoch": 1.59, "learning_rate": 2.151260096197676e-06, "loss": 1.12, "step": 4143 }, { "epoch": 1.59, "learning_rate": 2.147365900418439e-06, "loss": 1.084, "step": 4144 }, { "epoch": 1.59, "learning_rate": 2.143474808458249e-06, "loss": 1.0523, "step": 4145 }, { "epoch": 1.59, "learning_rate": 2.1395868218550943e-06, "loss": 1.0995, "step": 4146 }, { "epoch": 1.59, "learning_rate": 2.135701942145726e-06, "loss": 1.0785, "step": 4147 }, { "epoch": 1.59, "learning_rate": 2.131820170865685e-06, "loss": 1.1022, "step": 4148 }, { "epoch": 1.59, "learning_rate": 2.127941509549267e-06, "loss": 1.0744, "step": 4149 }, { "epoch": 1.59, "learning_rate": 2.1240659597295486e-06, "loss": 1.0759, "step": 4150 }, { "epoch": 1.59, "learning_rate": 2.1201935229383707e-06, "loss": 1.1077, "step": 4151 }, { "epoch": 1.59, "learning_rate": 2.1163242007063476e-06, "loss": 1.0561, "step": 4152 }, { "epoch": 1.6, "learning_rate": 2.1124579945628644e-06, "loss": 1.0714, "step": 4153 }, { "epoch": 1.6, "learning_rate": 2.1085949060360654e-06, "loss": 1.0977, "step": 4154 }, { "epoch": 1.6, "learning_rate": 2.10473493665287e-06, "loss": 1.0726, "step": 4155 }, { "epoch": 1.6, "learning_rate": 2.1008780879389656e-06, "loss": 1.0504, "step": 4156 }, { "epoch": 1.6, "learning_rate": 2.097024361418801e-06, "loss": 1.0759, "step": 4157 }, { "epoch": 1.6, "learning_rate": 2.0931737586155965e-06, "loss": 1.0751, "step": 4158 }, { "epoch": 1.6, "learning_rate": 2.089326281051336e-06, "loss": 1.0526, "step": 4159 }, { "epoch": 1.6, "learning_rate": 2.08548193024676e-06, "loss": 1.0728, "step": 4160 }, { "epoch": 1.6, "learning_rate": 2.0816407077213897e-06, "loss": 1.103, "step": 4161 }, { "epoch": 1.6, "learning_rate": 2.0778026149934915e-06, "loss": 1.0764, "step": 4162 }, { "epoch": 1.6, "learning_rate": 2.073967653580109e-06, "loss": 1.1582, "step": 4163 }, { "epoch": 1.6, "learning_rate": 2.070135824997044e-06, "loss": 1.128, "step": 4164 }, { "epoch": 1.6, "learning_rate": 2.06630713075885e-06, "loss": 1.122, "step": 4165 }, { "epoch": 1.6, "learning_rate": 2.0624815723788615e-06, "loss": 1.0379, "step": 4166 }, { "epoch": 1.6, "learning_rate": 2.0586591513691535e-06, "loss": 1.1235, "step": 4167 }, { "epoch": 1.6, "learning_rate": 2.0548398692405746e-06, "loss": 1.1018, "step": 4168 }, { "epoch": 1.6, "learning_rate": 2.0510237275027256e-06, "loss": 1.1064, "step": 4169 }, { "epoch": 1.6, "learning_rate": 2.047210727663972e-06, "loss": 1.0842, "step": 4170 }, { "epoch": 1.6, "learning_rate": 2.0434008712314256e-06, "loss": 1.0719, "step": 4171 }, { "epoch": 1.6, "learning_rate": 2.039594159710975e-06, "loss": 1.0721, "step": 4172 }, { "epoch": 1.6, "learning_rate": 2.0357905946072456e-06, "loss": 1.0744, "step": 4173 }, { "epoch": 1.6, "learning_rate": 2.0319901774236317e-06, "loss": 1.0745, "step": 4174 }, { "epoch": 1.6, "learning_rate": 2.0281929096622797e-06, "loss": 1.0982, "step": 4175 }, { "epoch": 1.6, "learning_rate": 2.0243987928240914e-06, "loss": 1.0606, "step": 4176 }, { "epoch": 1.6, "learning_rate": 2.020607828408725e-06, "loss": 1.0924, "step": 4177 }, { "epoch": 1.6, "learning_rate": 2.0168200179145868e-06, "loss": 1.0258, "step": 4178 }, { "epoch": 1.61, "learning_rate": 2.0130353628388423e-06, "loss": 1.1307, "step": 4179 }, { "epoch": 1.61, "learning_rate": 2.0092538646774073e-06, "loss": 1.0976, "step": 4180 }, { "epoch": 1.61, "learning_rate": 2.0054755249249513e-06, "loss": 1.1157, "step": 4181 }, { "epoch": 1.61, "learning_rate": 2.001700345074895e-06, "loss": 1.1064, "step": 4182 }, { "epoch": 1.61, "learning_rate": 1.9979283266194115e-06, "loss": 1.026, "step": 4183 }, { "epoch": 1.61, "learning_rate": 1.9941594710494137e-06, "loss": 1.1145, "step": 4184 }, { "epoch": 1.61, "learning_rate": 1.990393779854587e-06, "loss": 1.11, "step": 4185 }, { "epoch": 1.61, "learning_rate": 1.98663125452334e-06, "loss": 1.1244, "step": 4186 }, { "epoch": 1.61, "learning_rate": 1.9828718965428484e-06, "loss": 1.0775, "step": 4187 }, { "epoch": 1.61, "learning_rate": 1.9791157073990306e-06, "loss": 1.0543, "step": 4188 }, { "epoch": 1.61, "learning_rate": 1.9753626885765443e-06, "loss": 1.0739, "step": 4189 }, { "epoch": 1.61, "learning_rate": 1.971612841558811e-06, "loss": 1.0544, "step": 4190 }, { "epoch": 1.61, "learning_rate": 1.967866167827983e-06, "loss": 1.1065, "step": 4191 }, { "epoch": 1.61, "learning_rate": 1.964122668864965e-06, "loss": 1.0922, "step": 4192 }, { "epoch": 1.61, "learning_rate": 1.9603823461494077e-06, "loss": 1.1062, "step": 4193 }, { "epoch": 1.61, "learning_rate": 1.956645201159704e-06, "loss": 1.1033, "step": 4194 }, { "epoch": 1.61, "learning_rate": 1.9529112353729907e-06, "loss": 1.049, "step": 4195 }, { "epoch": 1.61, "learning_rate": 1.9491804502651534e-06, "loss": 1.0502, "step": 4196 }, { "epoch": 1.61, "learning_rate": 1.945452847310807e-06, "loss": 1.1165, "step": 4197 }, { "epoch": 1.61, "learning_rate": 1.941728427983328e-06, "loss": 1.1022, "step": 4198 }, { "epoch": 1.61, "learning_rate": 1.938007193754816e-06, "loss": 1.1637, "step": 4199 }, { "epoch": 1.61, "learning_rate": 1.9342891460961243e-06, "loss": 1.0516, "step": 4200 }, { "epoch": 1.61, "learning_rate": 1.930574286476843e-06, "loss": 1.0847, "step": 4201 }, { "epoch": 1.61, "learning_rate": 1.926862616365295e-06, "loss": 1.0473, "step": 4202 }, { "epoch": 1.61, "learning_rate": 1.9231541372285566e-06, "loss": 1.0374, "step": 4203 }, { "epoch": 1.61, "learning_rate": 1.9194488505324303e-06, "loss": 1.1048, "step": 4204 }, { "epoch": 1.62, "learning_rate": 1.9157467577414624e-06, "loss": 1.1325, "step": 4205 }, { "epoch": 1.62, "learning_rate": 1.9120478603189363e-06, "loss": 1.1394, "step": 4206 }, { "epoch": 1.62, "learning_rate": 1.908352159726873e-06, "loss": 1.0584, "step": 4207 }, { "epoch": 1.62, "learning_rate": 1.9046596574260267e-06, "loss": 1.0997, "step": 4208 }, { "epoch": 1.62, "learning_rate": 1.900970354875894e-06, "loss": 1.0672, "step": 4209 }, { "epoch": 1.62, "learning_rate": 1.8972842535346946e-06, "loss": 1.0828, "step": 4210 }, { "epoch": 1.62, "learning_rate": 1.8936013548594e-06, "loss": 1.0967, "step": 4211 }, { "epoch": 1.62, "learning_rate": 1.8899216603057013e-06, "loss": 1.0756, "step": 4212 }, { "epoch": 1.62, "learning_rate": 1.8862451713280282e-06, "loss": 1.0523, "step": 4213 }, { "epoch": 1.62, "learning_rate": 1.8825718893795474e-06, "loss": 1.0954, "step": 4214 }, { "epoch": 1.62, "learning_rate": 1.8789018159121463e-06, "loss": 1.0397, "step": 4215 }, { "epoch": 1.62, "learning_rate": 1.8752349523764623e-06, "loss": 1.0512, "step": 4216 }, { "epoch": 1.62, "learning_rate": 1.871571300221846e-06, "loss": 1.0978, "step": 4217 }, { "epoch": 1.62, "learning_rate": 1.8679108608963903e-06, "loss": 1.1331, "step": 4218 }, { "epoch": 1.62, "learning_rate": 1.8642536358469122e-06, "loss": 1.0746, "step": 4219 }, { "epoch": 1.62, "learning_rate": 1.8605996265189618e-06, "loss": 1.0595, "step": 4220 }, { "epoch": 1.62, "learning_rate": 1.8569488343568155e-06, "loss": 1.086, "step": 4221 }, { "epoch": 1.62, "learning_rate": 1.853301260803484e-06, "loss": 1.1138, "step": 4222 }, { "epoch": 1.62, "learning_rate": 1.8496569073006909e-06, "loss": 1.079, "step": 4223 }, { "epoch": 1.62, "learning_rate": 1.8460157752889085e-06, "loss": 1.099, "step": 4224 }, { "epoch": 1.62, "learning_rate": 1.8423778662073178e-06, "loss": 1.1204, "step": 4225 }, { "epoch": 1.62, "learning_rate": 1.838743181493835e-06, "loss": 1.0778, "step": 4226 }, { "epoch": 1.62, "learning_rate": 1.8351117225851011e-06, "loss": 1.1305, "step": 4227 }, { "epoch": 1.62, "learning_rate": 1.8314834909164735e-06, "loss": 1.0591, "step": 4228 }, { "epoch": 1.62, "learning_rate": 1.8278584879220506e-06, "loss": 1.0832, "step": 4229 }, { "epoch": 1.62, "learning_rate": 1.8242367150346385e-06, "loss": 1.1063, "step": 4230 }, { "epoch": 1.63, "learning_rate": 1.8206181736857741e-06, "loss": 1.0728, "step": 4231 }, { "epoch": 1.63, "learning_rate": 1.8170028653057171e-06, "loss": 1.0873, "step": 4232 }, { "epoch": 1.63, "learning_rate": 1.8133907913234482e-06, "loss": 1.0438, "step": 4233 }, { "epoch": 1.63, "learning_rate": 1.8097819531666706e-06, "loss": 1.0766, "step": 4234 }, { "epoch": 1.63, "learning_rate": 1.806176352261808e-06, "loss": 1.0812, "step": 4235 }, { "epoch": 1.63, "learning_rate": 1.8025739900340001e-06, "loss": 1.1073, "step": 4236 }, { "epoch": 1.63, "learning_rate": 1.7989748679071139e-06, "loss": 1.1223, "step": 4237 }, { "epoch": 1.63, "learning_rate": 1.7953789873037343e-06, "loss": 1.0926, "step": 4238 }, { "epoch": 1.63, "learning_rate": 1.7917863496451548e-06, "loss": 1.1229, "step": 4239 }, { "epoch": 1.63, "learning_rate": 1.7881969563514067e-06, "loss": 1.1086, "step": 4240 }, { "epoch": 1.63, "learning_rate": 1.7846108088412196e-06, "loss": 1.0832, "step": 4241 }, { "epoch": 1.63, "learning_rate": 1.7810279085320504e-06, "loss": 1.1285, "step": 4242 }, { "epoch": 1.63, "learning_rate": 1.7774482568400708e-06, "loss": 1.0989, "step": 4243 }, { "epoch": 1.63, "learning_rate": 1.773871855180168e-06, "loss": 1.0808, "step": 4244 }, { "epoch": 1.63, "learning_rate": 1.7702987049659436e-06, "loss": 1.0621, "step": 4245 }, { "epoch": 1.63, "learning_rate": 1.7667288076097178e-06, "loss": 1.08, "step": 4246 }, { "epoch": 1.63, "learning_rate": 1.763162164522516e-06, "loss": 1.069, "step": 4247 }, { "epoch": 1.63, "learning_rate": 1.7595987771140921e-06, "loss": 1.0393, "step": 4248 }, { "epoch": 1.63, "learning_rate": 1.756038646792897e-06, "loss": 1.0994, "step": 4249 }, { "epoch": 1.63, "learning_rate": 1.752481774966106e-06, "loss": 1.0495, "step": 4250 }, { "epoch": 1.63, "learning_rate": 1.7489281630396038e-06, "loss": 1.1102, "step": 4251 }, { "epoch": 1.63, "learning_rate": 1.7453778124179778e-06, "loss": 1.0749, "step": 4252 }, { "epoch": 1.63, "learning_rate": 1.7418307245045441e-06, "loss": 1.1427, "step": 4253 }, { "epoch": 1.63, "learning_rate": 1.7382869007013093e-06, "loss": 1.0595, "step": 4254 }, { "epoch": 1.63, "learning_rate": 1.734746342409004e-06, "loss": 1.0543, "step": 4255 }, { "epoch": 1.63, "learning_rate": 1.7312090510270629e-06, "loss": 1.1716, "step": 4256 }, { "epoch": 1.64, "learning_rate": 1.7276750279536292e-06, "loss": 1.0769, "step": 4257 }, { "epoch": 1.64, "learning_rate": 1.7241442745855552e-06, "loss": 1.0524, "step": 4258 }, { "epoch": 1.64, "learning_rate": 1.7206167923184035e-06, "loss": 1.0949, "step": 4259 }, { "epoch": 1.64, "learning_rate": 1.7170925825464358e-06, "loss": 1.0788, "step": 4260 }, { "epoch": 1.64, "learning_rate": 1.7135716466626274e-06, "loss": 1.0888, "step": 4261 }, { "epoch": 1.64, "learning_rate": 1.7100539860586573e-06, "loss": 1.0993, "step": 4262 }, { "epoch": 1.64, "learning_rate": 1.7065396021249102e-06, "loss": 1.0974, "step": 4263 }, { "epoch": 1.64, "learning_rate": 1.7030284962504796e-06, "loss": 1.1483, "step": 4264 }, { "epoch": 1.64, "learning_rate": 1.6995206698231503e-06, "loss": 1.1688, "step": 4265 }, { "epoch": 1.64, "learning_rate": 1.6960161242294316e-06, "loss": 1.1275, "step": 4266 }, { "epoch": 1.64, "learning_rate": 1.692514860854514e-06, "loss": 1.0938, "step": 4267 }, { "epoch": 1.64, "learning_rate": 1.689016881082306e-06, "loss": 1.0718, "step": 4268 }, { "epoch": 1.64, "learning_rate": 1.685522186295413e-06, "loss": 1.0984, "step": 4269 }, { "epoch": 1.64, "learning_rate": 1.68203077787514e-06, "loss": 1.0799, "step": 4270 }, { "epoch": 1.64, "learning_rate": 1.6785426572014985e-06, "loss": 1.2006, "step": 4271 }, { "epoch": 1.64, "learning_rate": 1.675057825653197e-06, "loss": 1.095, "step": 4272 }, { "epoch": 1.64, "learning_rate": 1.671576284607641e-06, "loss": 1.1344, "step": 4273 }, { "epoch": 1.64, "learning_rate": 1.6680980354409405e-06, "loss": 1.0787, "step": 4274 }, { "epoch": 1.64, "learning_rate": 1.6646230795279028e-06, "loss": 1.087, "step": 4275 }, { "epoch": 1.64, "learning_rate": 1.6611514182420319e-06, "loss": 1.0633, "step": 4276 }, { "epoch": 1.64, "learning_rate": 1.6576830529555344e-06, "loss": 1.0625, "step": 4277 }, { "epoch": 1.64, "learning_rate": 1.6542179850393037e-06, "loss": 1.1001, "step": 4278 }, { "epoch": 1.64, "learning_rate": 1.6507562158629453e-06, "loss": 1.1088, "step": 4279 }, { "epoch": 1.64, "learning_rate": 1.647297746794746e-06, "loss": 1.1184, "step": 4280 }, { "epoch": 1.64, "learning_rate": 1.6438425792016955e-06, "loss": 1.0953, "step": 4281 }, { "epoch": 1.64, "learning_rate": 1.6403907144494801e-06, "loss": 1.1353, "step": 4282 }, { "epoch": 1.65, "learning_rate": 1.6369421539024756e-06, "loss": 1.062, "step": 4283 }, { "epoch": 1.65, "learning_rate": 1.6334968989237577e-06, "loss": 1.0645, "step": 4284 }, { "epoch": 1.65, "learning_rate": 1.6300549508750874e-06, "loss": 1.1205, "step": 4285 }, { "epoch": 1.65, "learning_rate": 1.6266163111169253e-06, "loss": 1.0791, "step": 4286 }, { "epoch": 1.65, "learning_rate": 1.6231809810084232e-06, "loss": 1.1032, "step": 4287 }, { "epoch": 1.65, "learning_rate": 1.6197489619074236e-06, "loss": 1.1331, "step": 4288 }, { "epoch": 1.65, "learning_rate": 1.6163202551704605e-06, "loss": 1.1226, "step": 4289 }, { "epoch": 1.65, "learning_rate": 1.6128948621527607e-06, "loss": 1.0377, "step": 4290 }, { "epoch": 1.65, "learning_rate": 1.6094727842082337e-06, "loss": 1.0402, "step": 4291 }, { "epoch": 1.65, "learning_rate": 1.6060540226894916e-06, "loss": 1.0618, "step": 4292 }, { "epoch": 1.65, "learning_rate": 1.6026385789478238e-06, "loss": 1.0827, "step": 4293 }, { "epoch": 1.65, "learning_rate": 1.5992264543332125e-06, "loss": 1.1372, "step": 4294 }, { "epoch": 1.65, "learning_rate": 1.5958176501943313e-06, "loss": 1.0477, "step": 4295 }, { "epoch": 1.65, "learning_rate": 1.5924121678785353e-06, "loss": 1.143, "step": 4296 }, { "epoch": 1.65, "learning_rate": 1.589010008731875e-06, "loss": 1.1051, "step": 4297 }, { "epoch": 1.65, "learning_rate": 1.5856111740990753e-06, "loss": 1.0428, "step": 4298 }, { "epoch": 1.65, "learning_rate": 1.5822156653235577e-06, "loss": 1.1065, "step": 4299 }, { "epoch": 1.65, "learning_rate": 1.5788234837474248e-06, "loss": 1.066, "step": 4300 }, { "epoch": 1.65, "learning_rate": 1.5754346307114654e-06, "loss": 1.0966, "step": 4301 }, { "epoch": 1.65, "learning_rate": 1.5720491075551504e-06, "loss": 1.0885, "step": 4302 }, { "epoch": 1.65, "learning_rate": 1.568666915616639e-06, "loss": 1.1218, "step": 4303 }, { "epoch": 1.65, "learning_rate": 1.5652880562327633e-06, "loss": 1.1097, "step": 4304 }, { "epoch": 1.65, "learning_rate": 1.561912530739056e-06, "loss": 1.1035, "step": 4305 }, { "epoch": 1.65, "learning_rate": 1.558540340469714e-06, "loss": 1.0868, "step": 4306 }, { "epoch": 1.65, "learning_rate": 1.5551714867576261e-06, "loss": 1.1266, "step": 4307 }, { "epoch": 1.65, "learning_rate": 1.5518059709343624e-06, "loss": 1.0847, "step": 4308 }, { "epoch": 1.66, "learning_rate": 1.5484437943301655e-06, "loss": 1.0789, "step": 4309 }, { "epoch": 1.66, "learning_rate": 1.5450849582739668e-06, "loss": 1.1035, "step": 4310 }, { "epoch": 1.66, "learning_rate": 1.5417294640933745e-06, "loss": 1.0662, "step": 4311 }, { "epoch": 1.66, "learning_rate": 1.538377313114675e-06, "loss": 1.0797, "step": 4312 }, { "epoch": 1.66, "learning_rate": 1.5350285066628346e-06, "loss": 1.0798, "step": 4313 }, { "epoch": 1.66, "learning_rate": 1.5316830460614973e-06, "loss": 1.0979, "step": 4314 }, { "epoch": 1.66, "learning_rate": 1.5283409326329802e-06, "loss": 1.1139, "step": 4315 }, { "epoch": 1.66, "learning_rate": 1.5250021676982884e-06, "loss": 1.1008, "step": 4316 }, { "epoch": 1.66, "learning_rate": 1.5216667525770912e-06, "loss": 1.074, "step": 4317 }, { "epoch": 1.66, "learning_rate": 1.5183346885877415e-06, "loss": 1.1007, "step": 4318 }, { "epoch": 1.66, "learning_rate": 1.5150059770472637e-06, "loss": 1.2039, "step": 4319 }, { "epoch": 1.66, "learning_rate": 1.5116806192713595e-06, "loss": 1.1029, "step": 4320 }, { "epoch": 1.66, "learning_rate": 1.508358616574408e-06, "loss": 1.0234, "step": 4321 }, { "epoch": 1.66, "learning_rate": 1.505039970269452e-06, "loss": 1.1084, "step": 4322 }, { "epoch": 1.66, "learning_rate": 1.5017246816682163e-06, "loss": 1.0789, "step": 4323 }, { "epoch": 1.66, "learning_rate": 1.4984127520810988e-06, "loss": 1.1237, "step": 4324 }, { "epoch": 1.66, "learning_rate": 1.4951041828171643e-06, "loss": 1.082, "step": 4325 }, { "epoch": 1.66, "learning_rate": 1.4917989751841543e-06, "loss": 1.0622, "step": 4326 }, { "epoch": 1.66, "learning_rate": 1.488497130488481e-06, "loss": 1.0798, "step": 4327 }, { "epoch": 1.66, "learning_rate": 1.485198650035219e-06, "loss": 1.1209, "step": 4328 }, { "epoch": 1.66, "learning_rate": 1.4819035351281308e-06, "loss": 1.0963, "step": 4329 }, { "epoch": 1.66, "learning_rate": 1.4786117870696293e-06, "loss": 1.0995, "step": 4330 }, { "epoch": 1.66, "learning_rate": 1.4753234071608092e-06, "loss": 1.0966, "step": 4331 }, { "epoch": 1.66, "learning_rate": 1.4720383967014307e-06, "loss": 1.1027, "step": 4332 }, { "epoch": 1.66, "learning_rate": 1.4687567569899165e-06, "loss": 1.0545, "step": 4333 }, { "epoch": 1.66, "learning_rate": 1.4654784893233699e-06, "loss": 1.006, "step": 4334 }, { "epoch": 1.67, "learning_rate": 1.4622035949975465e-06, "loss": 1.1475, "step": 4335 }, { "epoch": 1.67, "learning_rate": 1.458932075306879e-06, "loss": 1.0557, "step": 4336 }, { "epoch": 1.67, "learning_rate": 1.455663931544462e-06, "loss": 1.1552, "step": 4337 }, { "epoch": 1.67, "learning_rate": 1.4523991650020575e-06, "loss": 1.0854, "step": 4338 }, { "epoch": 1.67, "learning_rate": 1.4491377769700897e-06, "loss": 1.1203, "step": 4339 }, { "epoch": 1.67, "learning_rate": 1.4458797687376536e-06, "loss": 1.1515, "step": 4340 }, { "epoch": 1.67, "learning_rate": 1.4426251415924964e-06, "loss": 1.1531, "step": 4341 }, { "epoch": 1.67, "learning_rate": 1.4393738968210459e-06, "loss": 1.1081, "step": 4342 }, { "epoch": 1.67, "learning_rate": 1.4361260357083762e-06, "loss": 1.1164, "step": 4343 }, { "epoch": 1.67, "learning_rate": 1.432881559538235e-06, "loss": 1.1347, "step": 4344 }, { "epoch": 1.67, "learning_rate": 1.42964046959303e-06, "loss": 1.0603, "step": 4345 }, { "epoch": 1.67, "learning_rate": 1.4264027671538216e-06, "loss": 1.1064, "step": 4346 }, { "epoch": 1.67, "learning_rate": 1.4231684535003498e-06, "loss": 1.1169, "step": 4347 }, { "epoch": 1.67, "learning_rate": 1.4199375299109941e-06, "loss": 1.1124, "step": 4348 }, { "epoch": 1.67, "learning_rate": 1.4167099976628095e-06, "loss": 1.138, "step": 4349 }, { "epoch": 1.67, "learning_rate": 1.4134858580315035e-06, "loss": 1.0816, "step": 4350 }, { "epoch": 1.67, "learning_rate": 1.4102651122914434e-06, "loss": 1.0807, "step": 4351 }, { "epoch": 1.67, "learning_rate": 1.4070477617156564e-06, "loss": 1.0636, "step": 4352 }, { "epoch": 1.67, "learning_rate": 1.4038338075758307e-06, "loss": 1.081, "step": 4353 }, { "epoch": 1.67, "learning_rate": 1.4006232511423002e-06, "loss": 1.1177, "step": 4354 }, { "epoch": 1.67, "learning_rate": 1.3974160936840731e-06, "loss": 1.0819, "step": 4355 }, { "epoch": 1.67, "learning_rate": 1.3942123364687976e-06, "loss": 1.0693, "step": 4356 }, { "epoch": 1.67, "learning_rate": 1.3910119807627897e-06, "loss": 1.0337, "step": 4357 }, { "epoch": 1.67, "learning_rate": 1.3878150278310188e-06, "loss": 1.0438, "step": 4358 }, { "epoch": 1.67, "learning_rate": 1.3846214789370982e-06, "loss": 1.1127, "step": 4359 }, { "epoch": 1.67, "learning_rate": 1.381431335343315e-06, "loss": 1.0958, "step": 4360 }, { "epoch": 1.68, "learning_rate": 1.378244598310594e-06, "loss": 1.0838, "step": 4361 }, { "epoch": 1.68, "learning_rate": 1.375061269098521e-06, "loss": 1.1173, "step": 4362 }, { "epoch": 1.68, "learning_rate": 1.3718813489653338e-06, "loss": 1.1202, "step": 4363 }, { "epoch": 1.68, "learning_rate": 1.3687048391679226e-06, "loss": 1.0533, "step": 4364 }, { "epoch": 1.68, "learning_rate": 1.3655317409618273e-06, "loss": 1.0916, "step": 4365 }, { "epoch": 1.68, "learning_rate": 1.3623620556012462e-06, "loss": 1.0849, "step": 4366 }, { "epoch": 1.68, "learning_rate": 1.3591957843390158e-06, "loss": 1.0641, "step": 4367 }, { "epoch": 1.68, "learning_rate": 1.3560329284266406e-06, "loss": 1.1094, "step": 4368 }, { "epoch": 1.68, "learning_rate": 1.352873489114259e-06, "loss": 1.0889, "step": 4369 }, { "epoch": 1.68, "learning_rate": 1.3497174676506675e-06, "loss": 1.1124, "step": 4370 }, { "epoch": 1.68, "learning_rate": 1.3465648652833118e-06, "loss": 1.1022, "step": 4371 }, { "epoch": 1.68, "learning_rate": 1.343415683258279e-06, "loss": 1.1044, "step": 4372 }, { "epoch": 1.68, "learning_rate": 1.3402699228203176e-06, "loss": 1.1032, "step": 4373 }, { "epoch": 1.68, "learning_rate": 1.3371275852128097e-06, "loss": 1.0748, "step": 4374 }, { "epoch": 1.68, "learning_rate": 1.3339886716777906e-06, "loss": 1.1364, "step": 4375 }, { "epoch": 1.68, "learning_rate": 1.3308531834559446e-06, "loss": 1.0797, "step": 4376 }, { "epoch": 1.68, "learning_rate": 1.3277211217865993e-06, "loss": 1.1353, "step": 4377 }, { "epoch": 1.68, "learning_rate": 1.3245924879077232e-06, "loss": 1.1183, "step": 4378 }, { "epoch": 1.68, "learning_rate": 1.3214672830559427e-06, "loss": 1.0723, "step": 4379 }, { "epoch": 1.68, "learning_rate": 1.3183455084665164e-06, "loss": 1.158, "step": 4380 }, { "epoch": 1.68, "learning_rate": 1.3152271653733517e-06, "loss": 1.0645, "step": 4381 }, { "epoch": 1.68, "learning_rate": 1.3121122550090027e-06, "loss": 1.0894, "step": 4382 }, { "epoch": 1.68, "learning_rate": 1.3090007786046565e-06, "loss": 1.0191, "step": 4383 }, { "epoch": 1.68, "learning_rate": 1.3058927373901608e-06, "loss": 1.1201, "step": 4384 }, { "epoch": 1.68, "learning_rate": 1.302788132593986e-06, "loss": 1.0989, "step": 4385 }, { "epoch": 1.68, "learning_rate": 1.2996869654432554e-06, "loss": 1.1022, "step": 4386 }, { "epoch": 1.69, "learning_rate": 1.2965892371637323e-06, "loss": 1.0997, "step": 4387 }, { "epoch": 1.69, "learning_rate": 1.2934949489798198e-06, "loss": 1.1022, "step": 4388 }, { "epoch": 1.69, "learning_rate": 1.2904041021145597e-06, "loss": 1.0888, "step": 4389 }, { "epoch": 1.69, "learning_rate": 1.2873166977896378e-06, "loss": 1.0705, "step": 4390 }, { "epoch": 1.69, "learning_rate": 1.2842327372253705e-06, "loss": 1.082, "step": 4391 }, { "epoch": 1.69, "learning_rate": 1.2811522216407268e-06, "loss": 1.0955, "step": 4392 }, { "epoch": 1.69, "learning_rate": 1.2780751522532997e-06, "loss": 1.0609, "step": 4393 }, { "epoch": 1.69, "learning_rate": 1.2750015302793283e-06, "loss": 1.0679, "step": 4394 }, { "epoch": 1.69, "learning_rate": 1.27193135693369e-06, "loss": 1.0702, "step": 4395 }, { "epoch": 1.69, "learning_rate": 1.2688646334298904e-06, "loss": 1.0711, "step": 4396 }, { "epoch": 1.69, "learning_rate": 1.2658013609800856e-06, "loss": 1.0575, "step": 4397 }, { "epoch": 1.69, "learning_rate": 1.2627415407950528e-06, "loss": 1.0717, "step": 4398 }, { "epoch": 1.69, "learning_rate": 1.2596851740842141e-06, "loss": 1.0699, "step": 4399 }, { "epoch": 1.69, "learning_rate": 1.2566322620556237e-06, "loss": 1.0281, "step": 4400 }, { "epoch": 1.69, "eval_loss": 1.1042912006378174, "eval_runtime": 688.6762, "eval_samples_per_second": 4.938, "eval_steps_per_second": 0.309, "step": 4400 }, { "epoch": 1.69, "learning_rate": 1.2535828059159693e-06, "loss": 1.0763, "step": 4401 }, { "epoch": 1.69, "learning_rate": 1.250536806870577e-06, "loss": 1.1465, "step": 4402 }, { "epoch": 1.69, "learning_rate": 1.2474942661234024e-06, "loss": 1.0557, "step": 4403 }, { "epoch": 1.69, "learning_rate": 1.2444551848770314e-06, "loss": 1.1413, "step": 4404 }, { "epoch": 1.69, "learning_rate": 1.2414195643326899e-06, "loss": 1.1039, "step": 4405 }, { "epoch": 1.69, "learning_rate": 1.23838740569023e-06, "loss": 1.0727, "step": 4406 }, { "epoch": 1.69, "learning_rate": 1.235358710148139e-06, "loss": 1.1072, "step": 4407 }, { "epoch": 1.69, "learning_rate": 1.2323334789035368e-06, "loss": 1.1296, "step": 4408 }, { "epoch": 1.69, "learning_rate": 1.2293117131521625e-06, "loss": 1.1295, "step": 4409 }, { "epoch": 1.69, "learning_rate": 1.2262934140884053e-06, "loss": 1.0633, "step": 4410 }, { "epoch": 1.69, "learning_rate": 1.2232785829052652e-06, "loss": 1.0935, "step": 4411 }, { "epoch": 1.69, "learning_rate": 1.220267220794382e-06, "loss": 1.1213, "step": 4412 }, { "epoch": 1.7, "learning_rate": 1.2172593289460221e-06, "loss": 1.0734, "step": 4413 }, { "epoch": 1.7, "learning_rate": 1.2142549085490784e-06, "loss": 1.0834, "step": 4414 }, { "epoch": 1.7, "learning_rate": 1.2112539607910733e-06, "loss": 1.0637, "step": 4415 }, { "epoch": 1.7, "learning_rate": 1.2082564868581603e-06, "loss": 1.0964, "step": 4416 }, { "epoch": 1.7, "learning_rate": 1.2052624879351105e-06, "loss": 1.0897, "step": 4417 }, { "epoch": 1.7, "learning_rate": 1.2022719652053284e-06, "loss": 1.1204, "step": 4418 }, { "epoch": 1.7, "learning_rate": 1.1992849198508439e-06, "loss": 1.096, "step": 4419 }, { "epoch": 1.7, "learning_rate": 1.1963013530523115e-06, "loss": 1.0755, "step": 4420 }, { "epoch": 1.7, "learning_rate": 1.1933212659890126e-06, "loss": 1.0434, "step": 4421 }, { "epoch": 1.7, "learning_rate": 1.1903446598388446e-06, "loss": 1.1522, "step": 4422 }, { "epoch": 1.7, "learning_rate": 1.1873715357783456e-06, "loss": 1.0843, "step": 4423 }, { "epoch": 1.7, "learning_rate": 1.1844018949826608e-06, "loss": 1.1187, "step": 4424 }, { "epoch": 1.7, "learning_rate": 1.1814357386255682e-06, "loss": 1.1326, "step": 4425 }, { "epoch": 1.7, "learning_rate": 1.1784730678794642e-06, "loss": 1.0746, "step": 4426 }, { "epoch": 1.7, "learning_rate": 1.1755138839153702e-06, "loss": 1.0647, "step": 4427 }, { "epoch": 1.7, "learning_rate": 1.1725581879029302e-06, "loss": 1.0791, "step": 4428 }, { "epoch": 1.7, "learning_rate": 1.1696059810104032e-06, "loss": 1.17, "step": 4429 }, { "epoch": 1.7, "learning_rate": 1.166657264404677e-06, "loss": 1.0303, "step": 4430 }, { "epoch": 1.7, "learning_rate": 1.1637120392512547e-06, "loss": 1.2008, "step": 4431 }, { "epoch": 1.7, "learning_rate": 1.1607703067142606e-06, "loss": 1.1033, "step": 4432 }, { "epoch": 1.7, "learning_rate": 1.1578320679564414e-06, "loss": 1.0793, "step": 4433 }, { "epoch": 1.7, "learning_rate": 1.1548973241391614e-06, "loss": 1.0766, "step": 4434 }, { "epoch": 1.7, "learning_rate": 1.1519660764223962e-06, "loss": 1.1099, "step": 4435 }, { "epoch": 1.7, "learning_rate": 1.149038325964753e-06, "loss": 1.1242, "step": 4436 }, { "epoch": 1.7, "learning_rate": 1.1461140739234455e-06, "loss": 1.1099, "step": 4437 }, { "epoch": 1.7, "learning_rate": 1.14319332145431e-06, "loss": 1.114, "step": 4438 }, { "epoch": 1.71, "learning_rate": 1.1402760697117976e-06, "loss": 1.1273, "step": 4439 }, { "epoch": 1.71, "learning_rate": 1.1373623198489757e-06, "loss": 1.1652, "step": 4440 }, { "epoch": 1.71, "learning_rate": 1.1344520730175323e-06, "loss": 1.1343, "step": 4441 }, { "epoch": 1.71, "learning_rate": 1.1315453303677616e-06, "loss": 1.1009, "step": 4442 }, { "epoch": 1.71, "learning_rate": 1.1286420930485797e-06, "loss": 1.0928, "step": 4443 }, { "epoch": 1.71, "learning_rate": 1.1257423622075159e-06, "loss": 1.1113, "step": 4444 }, { "epoch": 1.71, "learning_rate": 1.1228461389907142e-06, "loss": 1.1681, "step": 4445 }, { "epoch": 1.71, "learning_rate": 1.1199534245429256e-06, "loss": 1.0906, "step": 4446 }, { "epoch": 1.71, "learning_rate": 1.1170642200075277e-06, "loss": 1.0524, "step": 4447 }, { "epoch": 1.71, "learning_rate": 1.1141785265264959e-06, "loss": 1.1009, "step": 4448 }, { "epoch": 1.71, "learning_rate": 1.111296345240427e-06, "loss": 1.1377, "step": 4449 }, { "epoch": 1.71, "learning_rate": 1.1084176772885269e-06, "loss": 1.0666, "step": 4450 }, { "epoch": 1.71, "learning_rate": 1.1055425238086137e-06, "loss": 1.0006, "step": 4451 }, { "epoch": 1.71, "learning_rate": 1.1026708859371172e-06, "loss": 1.1112, "step": 4452 }, { "epoch": 1.71, "learning_rate": 1.0998027648090726e-06, "loss": 1.1489, "step": 4453 }, { "epoch": 1.71, "learning_rate": 1.0969381615581298e-06, "loss": 1.1295, "step": 4454 }, { "epoch": 1.71, "learning_rate": 1.0940770773165487e-06, "loss": 1.1253, "step": 4455 }, { "epoch": 1.71, "learning_rate": 1.091219513215196e-06, "loss": 1.0444, "step": 4456 }, { "epoch": 1.71, "learning_rate": 1.0883654703835466e-06, "loss": 1.091, "step": 4457 }, { "epoch": 1.71, "learning_rate": 1.0855149499496876e-06, "loss": 1.0725, "step": 4458 }, { "epoch": 1.71, "learning_rate": 1.082667953040305e-06, "loss": 1.0944, "step": 4459 }, { "epoch": 1.71, "learning_rate": 1.0798244807807068e-06, "loss": 1.0576, "step": 4460 }, { "epoch": 1.71, "learning_rate": 1.0769845342947916e-06, "loss": 1.0868, "step": 4461 }, { "epoch": 1.71, "learning_rate": 1.0741481147050737e-06, "loss": 1.1157, "step": 4462 }, { "epoch": 1.71, "learning_rate": 1.071315223132674e-06, "loss": 0.9998, "step": 4463 }, { "epoch": 1.71, "learning_rate": 1.0684858606973137e-06, "loss": 1.0779, "step": 4464 }, { "epoch": 1.72, "learning_rate": 1.0656600285173257e-06, "loss": 1.0731, "step": 4465 }, { "epoch": 1.72, "learning_rate": 1.062837727709639e-06, "loss": 1.1456, "step": 4466 }, { "epoch": 1.72, "learning_rate": 1.0600189593897936e-06, "loss": 1.0556, "step": 4467 }, { "epoch": 1.72, "learning_rate": 1.0572037246719324e-06, "loss": 1.0566, "step": 4468 }, { "epoch": 1.72, "learning_rate": 1.0543920246687978e-06, "loss": 1.0935, "step": 4469 }, { "epoch": 1.72, "learning_rate": 1.0515838604917394e-06, "loss": 1.1021, "step": 4470 }, { "epoch": 1.72, "learning_rate": 1.0487792332507086e-06, "loss": 1.1043, "step": 4471 }, { "epoch": 1.72, "learning_rate": 1.0459781440542526e-06, "loss": 1.0661, "step": 4472 }, { "epoch": 1.72, "learning_rate": 1.043180594009533e-06, "loss": 1.0555, "step": 4473 }, { "epoch": 1.72, "learning_rate": 1.0403865842222981e-06, "loss": 1.1659, "step": 4474 }, { "epoch": 1.72, "learning_rate": 1.0375961157969072e-06, "loss": 1.0625, "step": 4475 }, { "epoch": 1.72, "learning_rate": 1.0348091898363154e-06, "loss": 1.1128, "step": 4476 }, { "epoch": 1.72, "learning_rate": 1.0320258074420753e-06, "loss": 1.0627, "step": 4477 }, { "epoch": 1.72, "learning_rate": 1.0292459697143486e-06, "loss": 1.0934, "step": 4478 }, { "epoch": 1.72, "learning_rate": 1.0264696777518824e-06, "loss": 1.0769, "step": 4479 }, { "epoch": 1.72, "learning_rate": 1.023696932652033e-06, "loss": 1.0374, "step": 4480 }, { "epoch": 1.72, "learning_rate": 1.0209277355107505e-06, "loss": 1.0845, "step": 4481 }, { "epoch": 1.72, "learning_rate": 1.0181620874225829e-06, "loss": 1.0783, "step": 4482 }, { "epoch": 1.72, "learning_rate": 1.0153999894806744e-06, "loss": 1.1694, "step": 4483 }, { "epoch": 1.72, "learning_rate": 1.0126414427767716e-06, "loss": 1.119, "step": 4484 }, { "epoch": 1.72, "learning_rate": 1.009886448401206e-06, "loss": 1.0344, "step": 4485 }, { "epoch": 1.72, "learning_rate": 1.0071350074429198e-06, "loss": 1.1796, "step": 4486 }, { "epoch": 1.72, "learning_rate": 1.0043871209894363e-06, "loss": 1.1065, "step": 4487 }, { "epoch": 1.72, "learning_rate": 1.0016427901268843e-06, "loss": 1.1096, "step": 4488 }, { "epoch": 1.72, "learning_rate": 9.989020159399843e-07, "loss": 1.0962, "step": 4489 }, { "epoch": 1.72, "learning_rate": 9.96164799512045e-07, "loss": 1.1422, "step": 4490 }, { "epoch": 1.73, "learning_rate": 9.934311419249808e-07, "loss": 1.1711, "step": 4491 }, { "epoch": 1.73, "learning_rate": 9.907010442592868e-07, "loss": 1.0548, "step": 4492 }, { "epoch": 1.73, "learning_rate": 9.879745075940594e-07, "loss": 1.073, "step": 4493 }, { "epoch": 1.73, "learning_rate": 9.852515330069846e-07, "loss": 1.122, "step": 4494 }, { "epoch": 1.73, "learning_rate": 9.825321215743411e-07, "loss": 1.0797, "step": 4495 }, { "epoch": 1.73, "learning_rate": 9.798162743709995e-07, "loss": 1.097, "step": 4496 }, { "epoch": 1.73, "learning_rate": 9.771039924704218e-07, "loss": 1.1207, "step": 4497 }, { "epoch": 1.73, "learning_rate": 9.74395276944654e-07, "loss": 1.0958, "step": 4498 }, { "epoch": 1.73, "learning_rate": 9.716901288643477e-07, "loss": 1.1137, "step": 4499 }, { "epoch": 1.73, "learning_rate": 9.689885492987282e-07, "loss": 1.1028, "step": 4500 }, { "epoch": 1.73, "learning_rate": 9.662905393156196e-07, "loss": 1.1528, "step": 4501 }, { "epoch": 1.73, "learning_rate": 9.635960999814343e-07, "loss": 1.0819, "step": 4502 }, { "epoch": 1.73, "learning_rate": 9.609052323611667e-07, "loss": 1.0987, "step": 4503 }, { "epoch": 1.73, "learning_rate": 9.582179375184131e-07, "loss": 1.1154, "step": 4504 }, { "epoch": 1.73, "learning_rate": 9.555342165153425e-07, "loss": 1.087, "step": 4505 }, { "epoch": 1.73, "learning_rate": 9.528540704127187e-07, "loss": 1.0347, "step": 4506 }, { "epoch": 1.73, "learning_rate": 9.501775002698943e-07, "loss": 1.1333, "step": 4507 }, { "epoch": 1.73, "learning_rate": 9.475045071448042e-07, "loss": 1.0899, "step": 4508 }, { "epoch": 1.73, "learning_rate": 9.448350920939719e-07, "loss": 1.1624, "step": 4509 }, { "epoch": 1.73, "learning_rate": 9.421692561725071e-07, "loss": 1.0865, "step": 4510 }, { "epoch": 1.73, "learning_rate": 9.395070004341012e-07, "loss": 1.0967, "step": 4511 }, { "epoch": 1.73, "learning_rate": 9.368483259310335e-07, "loss": 1.1291, "step": 4512 }, { "epoch": 1.73, "learning_rate": 9.341932337141691e-07, "loss": 1.134, "step": 4513 }, { "epoch": 1.73, "learning_rate": 9.315417248329505e-07, "loss": 1.0913, "step": 4514 }, { "epoch": 1.73, "learning_rate": 9.288938003354164e-07, "loss": 1.0711, "step": 4515 }, { "epoch": 1.73, "learning_rate": 9.26249461268175e-07, "loss": 1.0864, "step": 4516 }, { "epoch": 1.74, "learning_rate": 9.236087086764245e-07, "loss": 1.1016, "step": 4517 }, { "epoch": 1.74, "learning_rate": 9.209715436039457e-07, "loss": 1.0938, "step": 4518 }, { "epoch": 1.74, "learning_rate": 9.183379670930992e-07, "loss": 1.1206, "step": 4519 }, { "epoch": 1.74, "learning_rate": 9.157079801848279e-07, "loss": 1.1346, "step": 4520 }, { "epoch": 1.74, "learning_rate": 9.130815839186579e-07, "loss": 1.0904, "step": 4521 }, { "epoch": 1.74, "learning_rate": 9.104587793326903e-07, "loss": 1.0665, "step": 4522 }, { "epoch": 1.74, "learning_rate": 9.078395674636153e-07, "loss": 1.0993, "step": 4523 }, { "epoch": 1.74, "learning_rate": 9.052239493466941e-07, "loss": 1.1053, "step": 4524 }, { "epoch": 1.74, "learning_rate": 9.026119260157728e-07, "loss": 1.0742, "step": 4525 }, { "epoch": 1.74, "learning_rate": 9.000034985032779e-07, "loss": 1.0356, "step": 4526 }, { "epoch": 1.74, "learning_rate": 8.973986678402069e-07, "loss": 1.0983, "step": 4527 }, { "epoch": 1.74, "learning_rate": 8.947974350561473e-07, "loss": 1.1529, "step": 4528 }, { "epoch": 1.74, "learning_rate": 8.921998011792521e-07, "loss": 1.1098, "step": 4529 }, { "epoch": 1.74, "learning_rate": 8.896057672362601e-07, "loss": 1.0619, "step": 4530 }, { "epoch": 1.74, "learning_rate": 8.870153342524857e-07, "loss": 1.0691, "step": 4531 }, { "epoch": 1.74, "learning_rate": 8.844285032518185e-07, "loss": 1.0748, "step": 4532 }, { "epoch": 1.74, "learning_rate": 8.818452752567241e-07, "loss": 1.0186, "step": 4533 }, { "epoch": 1.74, "learning_rate": 8.792656512882481e-07, "loss": 1.0858, "step": 4534 }, { "epoch": 1.74, "learning_rate": 8.766896323660046e-07, "loss": 1.0829, "step": 4535 }, { "epoch": 1.74, "learning_rate": 8.741172195081871e-07, "loss": 1.0486, "step": 4536 }, { "epoch": 1.74, "learning_rate": 8.715484137315633e-07, "loss": 1.1006, "step": 4537 }, { "epoch": 1.74, "learning_rate": 8.689832160514766e-07, "loss": 1.0737, "step": 4538 }, { "epoch": 1.74, "learning_rate": 8.664216274818438e-07, "loss": 1.1249, "step": 4539 }, { "epoch": 1.74, "learning_rate": 8.638636490351493e-07, "loss": 1.1263, "step": 4540 }, { "epoch": 1.74, "learning_rate": 8.613092817224611e-07, "loss": 1.0732, "step": 4541 }, { "epoch": 1.74, "learning_rate": 8.587585265534093e-07, "loss": 1.071, "step": 4542 }, { "epoch": 1.75, "learning_rate": 8.56211384536203e-07, "loss": 1.1117, "step": 4543 }, { "epoch": 1.75, "learning_rate": 8.536678566776224e-07, "loss": 1.0558, "step": 4544 }, { "epoch": 1.75, "learning_rate": 8.511279439830167e-07, "loss": 1.0702, "step": 4545 }, { "epoch": 1.75, "learning_rate": 8.485916474563083e-07, "loss": 1.0917, "step": 4546 }, { "epoch": 1.75, "learning_rate": 8.460589680999909e-07, "loss": 1.1099, "step": 4547 }, { "epoch": 1.75, "learning_rate": 8.435299069151248e-07, "loss": 1.0929, "step": 4548 }, { "epoch": 1.75, "learning_rate": 8.410044649013416e-07, "loss": 1.1064, "step": 4549 }, { "epoch": 1.75, "learning_rate": 8.38482643056846e-07, "loss": 1.1263, "step": 4550 }, { "epoch": 1.75, "learning_rate": 8.359644423784074e-07, "loss": 1.064, "step": 4551 }, { "epoch": 1.75, "learning_rate": 8.334498638613675e-07, "loss": 1.0808, "step": 4552 }, { "epoch": 1.75, "learning_rate": 8.3093890849963e-07, "loss": 1.074, "step": 4553 }, { "epoch": 1.75, "learning_rate": 8.284315772856777e-07, "loss": 1.0497, "step": 4554 }, { "epoch": 1.75, "learning_rate": 8.25927871210549e-07, "loss": 1.1579, "step": 4555 }, { "epoch": 1.75, "learning_rate": 8.234277912638556e-07, "loss": 1.1369, "step": 4556 }, { "epoch": 1.75, "learning_rate": 8.209313384337759e-07, "loss": 1.1164, "step": 4557 }, { "epoch": 1.75, "learning_rate": 8.184385137070527e-07, "loss": 1.0499, "step": 4558 }, { "epoch": 1.75, "learning_rate": 8.159493180689981e-07, "loss": 1.1014, "step": 4559 }, { "epoch": 1.75, "learning_rate": 8.134637525034838e-07, "loss": 1.1046, "step": 4560 }, { "epoch": 1.75, "learning_rate": 8.109818179929518e-07, "loss": 1.1063, "step": 4561 }, { "epoch": 1.75, "learning_rate": 8.085035155184074e-07, "loss": 1.0881, "step": 4562 }, { "epoch": 1.75, "learning_rate": 8.060288460594212e-07, "loss": 1.0816, "step": 4563 }, { "epoch": 1.75, "learning_rate": 8.035578105941255e-07, "loss": 1.1235, "step": 4564 }, { "epoch": 1.75, "learning_rate": 8.0109041009922e-07, "loss": 1.0993, "step": 4565 }, { "epoch": 1.75, "learning_rate": 7.986266455499614e-07, "loss": 1.0605, "step": 4566 }, { "epoch": 1.75, "learning_rate": 7.961665179201772e-07, "loss": 1.0807, "step": 4567 }, { "epoch": 1.75, "learning_rate": 7.937100281822508e-07, "loss": 1.1352, "step": 4568 }, { "epoch": 1.76, "learning_rate": 7.912571773071309e-07, "loss": 1.1212, "step": 4569 }, { "epoch": 1.76, "learning_rate": 7.888079662643277e-07, "loss": 1.0398, "step": 4570 }, { "epoch": 1.76, "learning_rate": 7.863623960219113e-07, "loss": 1.0785, "step": 4571 }, { "epoch": 1.76, "learning_rate": 7.839204675465172e-07, "loss": 1.0759, "step": 4572 }, { "epoch": 1.76, "learning_rate": 7.814821818033325e-07, "loss": 1.0792, "step": 4573 }, { "epoch": 1.76, "learning_rate": 7.790475397561126e-07, "loss": 1.0962, "step": 4574 }, { "epoch": 1.76, "learning_rate": 7.76616542367169e-07, "loss": 1.1053, "step": 4575 }, { "epoch": 1.76, "learning_rate": 7.741891905973764e-07, "loss": 1.0818, "step": 4576 }, { "epoch": 1.76, "learning_rate": 7.717654854061629e-07, "loss": 1.148, "step": 4577 }, { "epoch": 1.76, "learning_rate": 7.693454277515222e-07, "loss": 1.0881, "step": 4578 }, { "epoch": 1.76, "learning_rate": 7.669290185899947e-07, "loss": 1.0702, "step": 4579 }, { "epoch": 1.76, "learning_rate": 7.645162588766952e-07, "loss": 1.1284, "step": 4580 }, { "epoch": 1.76, "learning_rate": 7.62107149565281e-07, "loss": 1.099, "step": 4581 }, { "epoch": 1.76, "learning_rate": 7.597016916079747e-07, "loss": 1.0913, "step": 4582 }, { "epoch": 1.76, "learning_rate": 7.572998859555547e-07, "loss": 1.0764, "step": 4583 }, { "epoch": 1.76, "learning_rate": 7.549017335573516e-07, "loss": 1.0913, "step": 4584 }, { "epoch": 1.76, "learning_rate": 7.525072353612561e-07, "loss": 1.101, "step": 4585 }, { "epoch": 1.76, "learning_rate": 7.501163923137156e-07, "loss": 1.0793, "step": 4586 }, { "epoch": 1.76, "learning_rate": 7.477292053597284e-07, "loss": 1.0107, "step": 4587 }, { "epoch": 1.76, "learning_rate": 7.453456754428523e-07, "loss": 1.0884, "step": 4588 }, { "epoch": 1.76, "learning_rate": 7.429658035051968e-07, "loss": 1.1516, "step": 4589 }, { "epoch": 1.76, "learning_rate": 7.405895904874228e-07, "loss": 1.0731, "step": 4590 }, { "epoch": 1.76, "learning_rate": 7.382170373287533e-07, "loss": 1.1112, "step": 4591 }, { "epoch": 1.76, "learning_rate": 7.358481449669564e-07, "loss": 1.0804, "step": 4592 }, { "epoch": 1.76, "learning_rate": 7.334829143383581e-07, "loss": 1.0848, "step": 4593 }, { "epoch": 1.76, "learning_rate": 7.311213463778332e-07, "loss": 1.0728, "step": 4594 }, { "epoch": 1.77, "learning_rate": 7.287634420188138e-07, "loss": 1.1349, "step": 4595 }, { "epoch": 1.77, "learning_rate": 7.264092021932823e-07, "loss": 1.0759, "step": 4596 }, { "epoch": 1.77, "learning_rate": 7.240586278317674e-07, "loss": 1.0923, "step": 4597 }, { "epoch": 1.77, "learning_rate": 7.217117198633561e-07, "loss": 1.1012, "step": 4598 }, { "epoch": 1.77, "learning_rate": 7.193684792156819e-07, "loss": 1.0899, "step": 4599 }, { "epoch": 1.77, "learning_rate": 7.170289068149305e-07, "loss": 1.0559, "step": 4600 }, { "epoch": 1.77, "learning_rate": 7.146930035858379e-07, "loss": 1.1569, "step": 4601 }, { "epoch": 1.77, "learning_rate": 7.123607704516911e-07, "loss": 1.102, "step": 4602 }, { "epoch": 1.77, "learning_rate": 7.100322083343192e-07, "loss": 1.0418, "step": 4603 }, { "epoch": 1.77, "learning_rate": 7.077073181541128e-07, "loss": 1.0977, "step": 4604 }, { "epoch": 1.77, "learning_rate": 7.053861008299989e-07, "loss": 1.1066, "step": 4605 }, { "epoch": 1.77, "learning_rate": 7.030685572794593e-07, "loss": 1.1368, "step": 4606 }, { "epoch": 1.77, "learning_rate": 7.007546884185246e-07, "loss": 1.0713, "step": 4607 }, { "epoch": 1.77, "learning_rate": 6.984444951617664e-07, "loss": 1.0813, "step": 4608 }, { "epoch": 1.77, "learning_rate": 6.961379784223133e-07, "loss": 1.0548, "step": 4609 }, { "epoch": 1.77, "learning_rate": 6.938351391118303e-07, "loss": 1.0759, "step": 4610 }, { "epoch": 1.77, "learning_rate": 6.915359781405361e-07, "loss": 1.0686, "step": 4611 }, { "epoch": 1.77, "learning_rate": 6.892404964171928e-07, "loss": 1.0757, "step": 4612 }, { "epoch": 1.77, "learning_rate": 6.869486948491089e-07, "loss": 1.0978, "step": 4613 }, { "epoch": 1.77, "learning_rate": 6.846605743421375e-07, "loss": 1.0614, "step": 4614 }, { "epoch": 1.77, "learning_rate": 6.823761358006798e-07, "loss": 1.0839, "step": 4615 }, { "epoch": 1.77, "learning_rate": 6.800953801276744e-07, "loss": 1.0301, "step": 4616 }, { "epoch": 1.77, "learning_rate": 6.778183082246148e-07, "loss": 1.1226, "step": 4617 }, { "epoch": 1.77, "learning_rate": 6.755449209915277e-07, "loss": 1.0973, "step": 4618 }, { "epoch": 1.77, "learning_rate": 6.732752193269898e-07, "loss": 1.1426, "step": 4619 }, { "epoch": 1.77, "learning_rate": 6.710092041281225e-07, "loss": 1.1195, "step": 4620 }, { "epoch": 1.78, "learning_rate": 6.687468762905814e-07, "loss": 1.125, "step": 4621 }, { "epoch": 1.78, "learning_rate": 6.664882367085767e-07, "loss": 1.1075, "step": 4622 }, { "epoch": 1.78, "learning_rate": 6.642332862748502e-07, "loss": 1.1428, "step": 4623 }, { "epoch": 1.78, "learning_rate": 6.619820258806908e-07, "loss": 1.1303, "step": 4624 }, { "epoch": 1.78, "learning_rate": 6.597344564159281e-07, "loss": 1.1545, "step": 4625 }, { "epoch": 1.78, "learning_rate": 6.574905787689334e-07, "loss": 1.1453, "step": 4626 }, { "epoch": 1.78, "learning_rate": 6.552503938266175e-07, "loss": 1.1446, "step": 4627 }, { "epoch": 1.78, "learning_rate": 6.530139024744331e-07, "loss": 1.0816, "step": 4628 }, { "epoch": 1.78, "learning_rate": 6.507811055963686e-07, "loss": 1.0281, "step": 4629 }, { "epoch": 1.78, "learning_rate": 6.485520040749604e-07, "loss": 1.1415, "step": 4630 }, { "epoch": 1.78, "learning_rate": 6.46326598791277e-07, "loss": 1.0729, "step": 4631 }, { "epoch": 1.78, "learning_rate": 6.44104890624927e-07, "loss": 1.1123, "step": 4632 }, { "epoch": 1.78, "learning_rate": 6.418868804540634e-07, "loss": 1.1129, "step": 4633 }, { "epoch": 1.78, "learning_rate": 6.396725691553663e-07, "loss": 1.1013, "step": 4634 }, { "epoch": 1.78, "learning_rate": 6.374619576040675e-07, "loss": 1.0734, "step": 4635 }, { "epoch": 1.78, "learning_rate": 6.352550466739261e-07, "loss": 1.1029, "step": 4636 }, { "epoch": 1.78, "learning_rate": 6.330518372372429e-07, "loss": 1.0609, "step": 4637 }, { "epoch": 1.78, "learning_rate": 6.308523301648539e-07, "loss": 1.0936, "step": 4638 }, { "epoch": 1.78, "learning_rate": 6.286565263261346e-07, "loss": 1.0693, "step": 4639 }, { "epoch": 1.78, "learning_rate": 6.264644265889919e-07, "loss": 1.1239, "step": 4640 }, { "epoch": 1.78, "learning_rate": 6.24276031819876e-07, "loss": 1.0676, "step": 4641 }, { "epoch": 1.78, "learning_rate": 6.220913428837617e-07, "loss": 1.1195, "step": 4642 }, { "epoch": 1.78, "learning_rate": 6.199103606441726e-07, "loss": 1.0785, "step": 4643 }, { "epoch": 1.78, "learning_rate": 6.177330859631536e-07, "loss": 1.0836, "step": 4644 }, { "epoch": 1.78, "learning_rate": 6.155595197012953e-07, "loss": 1.0986, "step": 4645 }, { "epoch": 1.78, "learning_rate": 6.133896627177172e-07, "loss": 1.042, "step": 4646 }, { "epoch": 1.79, "learning_rate": 6.112235158700686e-07, "loss": 1.0691, "step": 4647 }, { "epoch": 1.79, "learning_rate": 6.090610800145447e-07, "loss": 1.0753, "step": 4648 }, { "epoch": 1.79, "learning_rate": 6.069023560058607e-07, "loss": 1.1629, "step": 4649 }, { "epoch": 1.79, "learning_rate": 6.047473446972718e-07, "loss": 1.1324, "step": 4650 }, { "epoch": 1.79, "learning_rate": 6.025960469405645e-07, "loss": 1.0889, "step": 4651 }, { "epoch": 1.79, "learning_rate": 6.004484635860585e-07, "loss": 1.1231, "step": 4652 }, { "epoch": 1.79, "learning_rate": 5.983045954826005e-07, "loss": 1.0622, "step": 4653 }, { "epoch": 1.79, "learning_rate": 5.96164443477576e-07, "loss": 1.0824, "step": 4654 }, { "epoch": 1.79, "learning_rate": 5.940280084168948e-07, "loss": 1.0671, "step": 4655 }, { "epoch": 1.79, "learning_rate": 5.918952911450027e-07, "loss": 1.0896, "step": 4656 }, { "epoch": 1.79, "learning_rate": 5.897662925048742e-07, "loss": 1.0728, "step": 4657 }, { "epoch": 1.79, "learning_rate": 5.876410133380095e-07, "loss": 1.1387, "step": 4658 }, { "epoch": 1.79, "learning_rate": 5.8551945448445e-07, "loss": 1.0845, "step": 4659 }, { "epoch": 1.79, "learning_rate": 5.834016167827539e-07, "loss": 1.1657, "step": 4660 }, { "epoch": 1.79, "learning_rate": 5.812875010700159e-07, "loss": 1.0937, "step": 4661 }, { "epoch": 1.79, "learning_rate": 5.791771081818587e-07, "loss": 1.0959, "step": 4662 }, { "epoch": 1.79, "learning_rate": 5.770704389524306e-07, "loss": 1.0953, "step": 4663 }, { "epoch": 1.79, "learning_rate": 5.749674942144123e-07, "loss": 1.0526, "step": 4664 }, { "epoch": 1.79, "learning_rate": 5.728682747990111e-07, "loss": 1.0999, "step": 4665 }, { "epoch": 1.79, "learning_rate": 5.707727815359565e-07, "loss": 1.0727, "step": 4666 }, { "epoch": 1.79, "learning_rate": 5.68681015253515e-07, "loss": 1.1053, "step": 4667 }, { "epoch": 1.79, "learning_rate": 5.665929767784705e-07, "loss": 1.1175, "step": 4668 }, { "epoch": 1.79, "learning_rate": 5.645086669361399e-07, "loss": 1.0756, "step": 4669 }, { "epoch": 1.79, "learning_rate": 5.624280865503639e-07, "loss": 1.0803, "step": 4670 }, { "epoch": 1.79, "learning_rate": 5.603512364435071e-07, "loss": 1.047, "step": 4671 }, { "epoch": 1.79, "learning_rate": 5.582781174364649e-07, "loss": 1.0777, "step": 4672 }, { "epoch": 1.8, "learning_rate": 5.562087303486529e-07, "loss": 1.1528, "step": 4673 }, { "epoch": 1.8, "learning_rate": 5.541430759980137e-07, "loss": 1.0273, "step": 4674 }, { "epoch": 1.8, "learning_rate": 5.520811552010142e-07, "loss": 1.1185, "step": 4675 }, { "epoch": 1.8, "learning_rate": 5.500229687726466e-07, "loss": 1.0732, "step": 4676 }, { "epoch": 1.8, "learning_rate": 5.47968517526426e-07, "loss": 1.0979, "step": 4677 }, { "epoch": 1.8, "learning_rate": 5.459178022743928e-07, "loss": 1.1012, "step": 4678 }, { "epoch": 1.8, "learning_rate": 5.438708238271062e-07, "loss": 1.126, "step": 4679 }, { "epoch": 1.8, "learning_rate": 5.418275829936537e-07, "loss": 1.0784, "step": 4680 }, { "epoch": 1.8, "learning_rate": 5.397880805816425e-07, "loss": 1.0718, "step": 4681 }, { "epoch": 1.8, "learning_rate": 5.377523173972021e-07, "loss": 1.1076, "step": 4682 }, { "epoch": 1.8, "learning_rate": 5.357202942449879e-07, "loss": 1.0809, "step": 4683 }, { "epoch": 1.8, "learning_rate": 5.336920119281697e-07, "loss": 1.0605, "step": 4684 }, { "epoch": 1.8, "learning_rate": 5.316674712484471e-07, "loss": 1.127, "step": 4685 }, { "epoch": 1.8, "learning_rate": 5.296466730060334e-07, "loss": 1.1224, "step": 4686 }, { "epoch": 1.8, "learning_rate": 5.276296179996677e-07, "loss": 1.1514, "step": 4687 }, { "epoch": 1.8, "learning_rate": 5.256163070266074e-07, "loss": 1.1211, "step": 4688 }, { "epoch": 1.8, "learning_rate": 5.236067408826296e-07, "loss": 1.0867, "step": 4689 }, { "epoch": 1.8, "learning_rate": 5.216009203620332e-07, "loss": 1.1007, "step": 4690 }, { "epoch": 1.8, "learning_rate": 5.195988462576374e-07, "loss": 1.0769, "step": 4691 }, { "epoch": 1.8, "learning_rate": 5.176005193607736e-07, "loss": 1.0944, "step": 4692 }, { "epoch": 1.8, "learning_rate": 5.156059404613012e-07, "loss": 1.0954, "step": 4693 }, { "epoch": 1.8, "learning_rate": 5.136151103475928e-07, "loss": 1.0879, "step": 4694 }, { "epoch": 1.8, "learning_rate": 5.116280298065401e-07, "loss": 1.0631, "step": 4695 }, { "epoch": 1.8, "learning_rate": 5.096446996235548e-07, "loss": 1.1014, "step": 4696 }, { "epoch": 1.8, "learning_rate": 5.076651205825622e-07, "loss": 1.1017, "step": 4697 }, { "epoch": 1.8, "learning_rate": 5.056892934660107e-07, "loss": 1.0932, "step": 4698 }, { "epoch": 1.81, "learning_rate": 5.0371721905486e-07, "loss": 1.1483, "step": 4699 }, { "epoch": 1.81, "learning_rate": 5.01748898128589e-07, "loss": 1.1069, "step": 4700 }, { "epoch": 1.81, "learning_rate": 4.997843314651929e-07, "loss": 1.1508, "step": 4701 }, { "epoch": 1.81, "learning_rate": 4.978235198411851e-07, "loss": 1.0517, "step": 4702 }, { "epoch": 1.81, "learning_rate": 4.958664640315914e-07, "loss": 1.0517, "step": 4703 }, { "epoch": 1.81, "learning_rate": 4.939131648099538e-07, "loss": 1.0649, "step": 4704 }, { "epoch": 1.81, "learning_rate": 4.919636229483304e-07, "loss": 1.1273, "step": 4705 }, { "epoch": 1.81, "learning_rate": 4.900178392172938e-07, "loss": 1.0565, "step": 4706 }, { "epoch": 1.81, "learning_rate": 4.880758143859309e-07, "loss": 1.1077, "step": 4707 }, { "epoch": 1.81, "learning_rate": 4.861375492218445e-07, "loss": 1.0578, "step": 4708 }, { "epoch": 1.81, "learning_rate": 4.842030444911506e-07, "loss": 1.1226, "step": 4709 }, { "epoch": 1.81, "learning_rate": 4.822723009584751e-07, "loss": 1.0997, "step": 4710 }, { "epoch": 1.81, "learning_rate": 4.803453193869645e-07, "loss": 1.0784, "step": 4711 }, { "epoch": 1.81, "learning_rate": 4.784221005382706e-07, "loss": 1.1275, "step": 4712 }, { "epoch": 1.81, "learning_rate": 4.76502645172563e-07, "loss": 1.0989, "step": 4713 }, { "epoch": 1.81, "learning_rate": 4.7458695404852263e-07, "loss": 1.1548, "step": 4714 }, { "epoch": 1.81, "learning_rate": 4.726750279233439e-07, "loss": 1.119, "step": 4715 }, { "epoch": 1.81, "learning_rate": 4.707668675527299e-07, "loss": 1.159, "step": 4716 }, { "epoch": 1.81, "learning_rate": 4.6886247369089645e-07, "loss": 1.1159, "step": 4717 }, { "epoch": 1.81, "learning_rate": 4.6696184709057124e-07, "loss": 1.0775, "step": 4718 }, { "epoch": 1.81, "learning_rate": 4.6506498850299453e-07, "loss": 1.0463, "step": 4719 }, { "epoch": 1.81, "learning_rate": 4.6317189867791434e-07, "loss": 1.1004, "step": 4720 }, { "epoch": 1.81, "learning_rate": 4.612825783635877e-07, "loss": 1.0576, "step": 4721 }, { "epoch": 1.81, "learning_rate": 4.593970283067895e-07, "loss": 1.055, "step": 4722 }, { "epoch": 1.81, "learning_rate": 4.5751524925279344e-07, "loss": 1.1224, "step": 4723 }, { "epoch": 1.81, "learning_rate": 4.556372419453936e-07, "loss": 1.1263, "step": 4724 }, { "epoch": 1.82, "learning_rate": 4.537630071268839e-07, "loss": 1.0745, "step": 4725 }, { "epoch": 1.82, "learning_rate": 4.5189254553807294e-07, "loss": 1.0631, "step": 4726 }, { "epoch": 1.82, "learning_rate": 4.500258579182781e-07, "loss": 1.076, "step": 4727 }, { "epoch": 1.82, "learning_rate": 4.481629450053193e-07, "loss": 1.1343, "step": 4728 }, { "epoch": 1.82, "learning_rate": 4.463038075355308e-07, "loss": 1.1176, "step": 4729 }, { "epoch": 1.82, "learning_rate": 4.444484462437515e-07, "loss": 1.0489, "step": 4730 }, { "epoch": 1.82, "learning_rate": 4.425968618633292e-07, "loss": 1.1448, "step": 4731 }, { "epoch": 1.82, "learning_rate": 4.407490551261173e-07, "loss": 1.1092, "step": 4732 }, { "epoch": 1.82, "learning_rate": 4.3890502676247837e-07, "loss": 1.0643, "step": 4733 }, { "epoch": 1.82, "learning_rate": 4.3706477750127594e-07, "loss": 1.1283, "step": 4734 }, { "epoch": 1.82, "learning_rate": 4.3522830806988934e-07, "loss": 1.0773, "step": 4735 }, { "epoch": 1.82, "learning_rate": 4.3339561919419347e-07, "loss": 1.0384, "step": 4736 }, { "epoch": 1.82, "learning_rate": 4.3156671159857666e-07, "loss": 1.0782, "step": 4737 }, { "epoch": 1.82, "learning_rate": 4.297415860059295e-07, "loss": 1.1127, "step": 4738 }, { "epoch": 1.82, "learning_rate": 4.279202431376483e-07, "loss": 1.0475, "step": 4739 }, { "epoch": 1.82, "learning_rate": 4.2610268371363397e-07, "loss": 1.0766, "step": 4740 }, { "epoch": 1.82, "learning_rate": 4.2428890845229056e-07, "loss": 1.1005, "step": 4741 }, { "epoch": 1.82, "learning_rate": 4.224789180705291e-07, "loss": 1.0634, "step": 4742 }, { "epoch": 1.82, "learning_rate": 4.2067271328376293e-07, "loss": 1.1155, "step": 4743 }, { "epoch": 1.82, "learning_rate": 4.188702948059098e-07, "loss": 1.0721, "step": 4744 }, { "epoch": 1.82, "learning_rate": 4.170716633493899e-07, "loss": 1.1133, "step": 4745 }, { "epoch": 1.82, "learning_rate": 4.1527681962513e-07, "loss": 1.1685, "step": 4746 }, { "epoch": 1.82, "learning_rate": 4.1348576434255274e-07, "loss": 1.0421, "step": 4747 }, { "epoch": 1.82, "learning_rate": 4.1169849820959176e-07, "loss": 1.131, "step": 4748 }, { "epoch": 1.82, "learning_rate": 4.099150219326764e-07, "loss": 1.1067, "step": 4749 }, { "epoch": 1.82, "learning_rate": 4.081353362167406e-07, "loss": 1.0601, "step": 4750 }, { "epoch": 1.83, "learning_rate": 4.063594417652228e-07, "loss": 1.0944, "step": 4751 }, { "epoch": 1.83, "learning_rate": 4.045873392800537e-07, "loss": 1.0872, "step": 4752 }, { "epoch": 1.83, "learning_rate": 4.028190294616796e-07, "loss": 1.1047, "step": 4753 }, { "epoch": 1.83, "learning_rate": 4.0105451300903373e-07, "loss": 1.0606, "step": 4754 }, { "epoch": 1.83, "learning_rate": 3.992937906195582e-07, "loss": 1.0584, "step": 4755 }, { "epoch": 1.83, "learning_rate": 3.975368629891929e-07, "loss": 1.175, "step": 4756 }, { "epoch": 1.83, "learning_rate": 3.9578373081237796e-07, "loss": 1.0623, "step": 4757 }, { "epoch": 1.83, "learning_rate": 3.940343947820546e-07, "loss": 1.1303, "step": 4758 }, { "epoch": 1.83, "learning_rate": 3.92288855589662e-07, "loss": 1.0751, "step": 4759 }, { "epoch": 1.83, "learning_rate": 3.90547113925136e-07, "loss": 1.0375, "step": 4760 }, { "epoch": 1.83, "learning_rate": 3.8880917047692037e-07, "loss": 1.0786, "step": 4761 }, { "epoch": 1.83, "learning_rate": 3.870750259319467e-07, "loss": 1.1017, "step": 4762 }, { "epoch": 1.83, "learning_rate": 3.8534468097565116e-07, "loss": 1.1173, "step": 4763 }, { "epoch": 1.83, "learning_rate": 3.8361813629197e-07, "loss": 1.0516, "step": 4764 }, { "epoch": 1.83, "learning_rate": 3.8189539256332954e-07, "loss": 1.1294, "step": 4765 }, { "epoch": 1.83, "learning_rate": 3.8017645047066285e-07, "loss": 1.079, "step": 4766 }, { "epoch": 1.83, "learning_rate": 3.784613106933932e-07, "loss": 1.0901, "step": 4767 }, { "epoch": 1.83, "learning_rate": 3.7674997390944603e-07, "loss": 1.1145, "step": 4768 }, { "epoch": 1.83, "learning_rate": 3.7504244079524023e-07, "loss": 1.0707, "step": 4769 }, { "epoch": 1.83, "learning_rate": 3.733387120256926e-07, "loss": 1.102, "step": 4770 }, { "epoch": 1.83, "learning_rate": 3.7163878827421565e-07, "loss": 1.1449, "step": 4771 }, { "epoch": 1.83, "learning_rate": 3.6994267021272066e-07, "loss": 1.0535, "step": 4772 }, { "epoch": 1.83, "learning_rate": 3.682503585116093e-07, "loss": 1.1496, "step": 4773 }, { "epoch": 1.83, "learning_rate": 3.6656185383978416e-07, "loss": 1.0731, "step": 4774 }, { "epoch": 1.83, "learning_rate": 3.648771568646403e-07, "loss": 1.0955, "step": 4775 }, { "epoch": 1.83, "learning_rate": 3.6319626825206735e-07, "loss": 1.1199, "step": 4776 }, { "epoch": 1.84, "learning_rate": 3.6151918866645155e-07, "loss": 1.0741, "step": 4777 }, { "epoch": 1.84, "learning_rate": 3.5984591877067044e-07, "loss": 1.0545, "step": 4778 }, { "epoch": 1.84, "learning_rate": 3.5817645922610275e-07, "loss": 1.121, "step": 4779 }, { "epoch": 1.84, "learning_rate": 3.565108106926107e-07, "loss": 1.0441, "step": 4780 }, { "epoch": 1.84, "learning_rate": 3.5484897382855875e-07, "loss": 1.107, "step": 4781 }, { "epoch": 1.84, "learning_rate": 3.5319094929080145e-07, "loss": 1.0774, "step": 4782 }, { "epoch": 1.84, "learning_rate": 3.515367377346857e-07, "loss": 1.0984, "step": 4783 }, { "epoch": 1.84, "learning_rate": 3.4988633981405396e-07, "loss": 1.0939, "step": 4784 }, { "epoch": 1.84, "learning_rate": 3.482397561812401e-07, "loss": 1.1489, "step": 4785 }, { "epoch": 1.84, "learning_rate": 3.465969874870656e-07, "loss": 1.0947, "step": 4786 }, { "epoch": 1.84, "learning_rate": 3.449580343808556e-07, "loss": 1.0639, "step": 4787 }, { "epoch": 1.84, "learning_rate": 3.4332289751041523e-07, "loss": 1.1395, "step": 4788 }, { "epoch": 1.84, "learning_rate": 3.416915775220453e-07, "loss": 1.1333, "step": 4789 }, { "epoch": 1.84, "learning_rate": 3.4006407506054216e-07, "loss": 1.0795, "step": 4790 }, { "epoch": 1.84, "learning_rate": 3.384403907691858e-07, "loss": 1.0449, "step": 4791 }, { "epoch": 1.84, "learning_rate": 3.3682052528975496e-07, "loss": 1.0953, "step": 4792 }, { "epoch": 1.84, "learning_rate": 3.35204479262512e-07, "loss": 1.1076, "step": 4793 }, { "epoch": 1.84, "learning_rate": 3.335922533262137e-07, "loss": 1.093, "step": 4794 }, { "epoch": 1.84, "learning_rate": 3.319838481181048e-07, "loss": 1.0429, "step": 4795 }, { "epoch": 1.84, "learning_rate": 3.303792642739234e-07, "loss": 1.1106, "step": 4796 }, { "epoch": 1.84, "learning_rate": 3.287785024278911e-07, "loss": 1.1061, "step": 4797 }, { "epoch": 1.84, "learning_rate": 3.271815632127262e-07, "loss": 1.0671, "step": 4798 }, { "epoch": 1.84, "learning_rate": 3.255884472596282e-07, "loss": 1.0674, "step": 4799 }, { "epoch": 1.84, "learning_rate": 3.2399915519829237e-07, "loss": 1.0932, "step": 4800 }, { "epoch": 1.84, "eval_loss": 1.1039639711380005, "eval_runtime": 688.4851, "eval_samples_per_second": 4.94, "eval_steps_per_second": 0.309, "step": 4800 }, { "epoch": 1.84, "learning_rate": 3.2241368765689953e-07, "loss": 1.049, "step": 4801 }, { "epoch": 1.84, "learning_rate": 3.208320452621161e-07, "loss": 1.042, "step": 4802 }, { "epoch": 1.85, "learning_rate": 3.192542286391043e-07, "loss": 1.0933, "step": 4803 }, { "epoch": 1.85, "learning_rate": 3.176802384115041e-07, "loss": 1.1655, "step": 4804 }, { "epoch": 1.85, "learning_rate": 3.161100752014512e-07, "loss": 1.0699, "step": 4805 }, { "epoch": 1.85, "learning_rate": 3.145437396295647e-07, "loss": 1.0503, "step": 4806 }, { "epoch": 1.85, "learning_rate": 3.129812323149528e-07, "loss": 1.0814, "step": 4807 }, { "epoch": 1.85, "learning_rate": 3.1142255387520914e-07, "loss": 1.0597, "step": 4808 }, { "epoch": 1.85, "learning_rate": 3.0986770492641427e-07, "loss": 1.0589, "step": 4809 }, { "epoch": 1.85, "learning_rate": 3.0831668608313326e-07, "loss": 1.1315, "step": 4810 }, { "epoch": 1.85, "learning_rate": 3.0676949795842346e-07, "loss": 1.0804, "step": 4811 }, { "epoch": 1.85, "learning_rate": 3.052261411638202e-07, "loss": 1.1063, "step": 4812 }, { "epoch": 1.85, "learning_rate": 3.036866163093488e-07, "loss": 1.1086, "step": 4813 }, { "epoch": 1.85, "learning_rate": 3.0215092400352144e-07, "loss": 1.0973, "step": 4814 }, { "epoch": 1.85, "learning_rate": 3.0061906485332915e-07, "loss": 1.0559, "step": 4815 }, { "epoch": 1.85, "learning_rate": 2.9909103946425767e-07, "loss": 1.1271, "step": 4816 }, { "epoch": 1.85, "learning_rate": 2.9756684844026606e-07, "loss": 1.1319, "step": 4817 }, { "epoch": 1.85, "learning_rate": 2.9604649238380687e-07, "loss": 1.1448, "step": 4818 }, { "epoch": 1.85, "learning_rate": 2.9452997189581276e-07, "loss": 1.1163, "step": 4819 }, { "epoch": 1.85, "learning_rate": 2.930172875757009e-07, "loss": 1.0475, "step": 4820 }, { "epoch": 1.85, "learning_rate": 2.91508440021373e-07, "loss": 1.1274, "step": 4821 }, { "epoch": 1.85, "learning_rate": 2.9000342982921315e-07, "loss": 1.0806, "step": 4822 }, { "epoch": 1.85, "learning_rate": 2.885022575940888e-07, "loss": 1.0899, "step": 4823 }, { "epoch": 1.85, "learning_rate": 2.870049239093509e-07, "loss": 1.0409, "step": 4824 }, { "epoch": 1.85, "learning_rate": 2.8551142936683373e-07, "loss": 1.0835, "step": 4825 }, { "epoch": 1.85, "learning_rate": 2.8402177455685296e-07, "loss": 1.1253, "step": 4826 }, { "epoch": 1.85, "learning_rate": 2.825359600682076e-07, "loss": 1.1079, "step": 4827 }, { "epoch": 1.85, "learning_rate": 2.8105398648817563e-07, "loss": 1.0732, "step": 4828 }, { "epoch": 1.85, "learning_rate": 2.79575854402524e-07, "loss": 1.1078, "step": 4829 }, { "epoch": 1.86, "learning_rate": 2.781015643954932e-07, "loss": 1.1458, "step": 4830 }, { "epoch": 1.86, "learning_rate": 2.766311170498104e-07, "loss": 1.0687, "step": 4831 }, { "epoch": 1.86, "learning_rate": 2.7516451294668065e-07, "loss": 1.0685, "step": 4832 }, { "epoch": 1.86, "learning_rate": 2.737017526657926e-07, "loss": 1.1022, "step": 4833 }, { "epoch": 1.86, "learning_rate": 2.722428367853136e-07, "loss": 1.0726, "step": 4834 }, { "epoch": 1.86, "learning_rate": 2.707877658818947e-07, "loss": 1.0711, "step": 4835 }, { "epoch": 1.86, "learning_rate": 2.693365405306614e-07, "loss": 1.1218, "step": 4836 }, { "epoch": 1.86, "learning_rate": 2.678891613052237e-07, "loss": 1.0996, "step": 4837 }, { "epoch": 1.86, "learning_rate": 2.664456287776718e-07, "loss": 1.1111, "step": 4838 }, { "epoch": 1.86, "learning_rate": 2.6500594351857143e-07, "loss": 1.093, "step": 4839 }, { "epoch": 1.86, "learning_rate": 2.635701060969742e-07, "loss": 1.1095, "step": 4840 }, { "epoch": 1.86, "learning_rate": 2.621381170804005e-07, "loss": 1.1032, "step": 4841 }, { "epoch": 1.86, "learning_rate": 2.607099770348631e-07, "loss": 1.0783, "step": 4842 }, { "epoch": 1.86, "learning_rate": 2.5928568652484163e-07, "loss": 1.0784, "step": 4843 }, { "epoch": 1.86, "learning_rate": 2.578652461132991e-07, "loss": 1.0756, "step": 4844 }, { "epoch": 1.86, "learning_rate": 2.564486563616786e-07, "loss": 1.0766, "step": 4845 }, { "epoch": 1.86, "learning_rate": 2.550359178298967e-07, "loss": 1.108, "step": 4846 }, { "epoch": 1.86, "learning_rate": 2.5362703107635224e-07, "loss": 1.1127, "step": 4847 }, { "epoch": 1.86, "learning_rate": 2.522219966579176e-07, "loss": 1.073, "step": 4848 }, { "epoch": 1.86, "learning_rate": 2.508208151299441e-07, "loss": 1.1115, "step": 4849 }, { "epoch": 1.86, "learning_rate": 2.4942348704626194e-07, "loss": 1.0966, "step": 4850 }, { "epoch": 1.86, "learning_rate": 2.4803001295917616e-07, "loss": 1.1174, "step": 4851 }, { "epoch": 1.86, "learning_rate": 2.4664039341946944e-07, "loss": 1.0554, "step": 4852 }, { "epoch": 1.86, "learning_rate": 2.452546289764002e-07, "loss": 1.1236, "step": 4853 }, { "epoch": 1.86, "learning_rate": 2.4387272017770136e-07, "loss": 1.0702, "step": 4854 }, { "epoch": 1.86, "learning_rate": 2.424946675695872e-07, "loss": 1.0602, "step": 4855 }, { "epoch": 1.87, "learning_rate": 2.4112047169674304e-07, "loss": 1.0581, "step": 4856 }, { "epoch": 1.87, "learning_rate": 2.3975013310232996e-07, "loss": 1.1017, "step": 4857 }, { "epoch": 1.87, "learning_rate": 2.383836523279881e-07, "loss": 1.1252, "step": 4858 }, { "epoch": 1.87, "learning_rate": 2.3702102991382868e-07, "loss": 1.0863, "step": 4859 }, { "epoch": 1.87, "learning_rate": 2.3566226639844093e-07, "loss": 1.0952, "step": 4860 }, { "epoch": 1.87, "learning_rate": 2.343073623188852e-07, "loss": 1.1121, "step": 4861 }, { "epoch": 1.87, "learning_rate": 2.3295631821070085e-07, "loss": 1.0608, "step": 4862 }, { "epoch": 1.87, "learning_rate": 2.3160913460789858e-07, "loss": 1.0827, "step": 4863 }, { "epoch": 1.87, "learning_rate": 2.302658120429635e-07, "loss": 1.0852, "step": 4864 }, { "epoch": 1.87, "learning_rate": 2.2892635104685423e-07, "loss": 1.1315, "step": 4865 }, { "epoch": 1.87, "learning_rate": 2.2759075214900727e-07, "loss": 1.0974, "step": 4866 }, { "epoch": 1.87, "learning_rate": 2.262590158773259e-07, "loss": 1.1271, "step": 4867 }, { "epoch": 1.87, "learning_rate": 2.2493114275819018e-07, "loss": 1.1077, "step": 4868 }, { "epoch": 1.87, "learning_rate": 2.2360713331645466e-07, "loss": 1.071, "step": 4869 }, { "epoch": 1.87, "learning_rate": 2.2228698807544413e-07, "loss": 1.0574, "step": 4870 }, { "epoch": 1.87, "learning_rate": 2.2097070755695894e-07, "loss": 1.0764, "step": 4871 }, { "epoch": 1.87, "learning_rate": 2.1965829228126623e-07, "loss": 1.1049, "step": 4872 }, { "epoch": 1.87, "learning_rate": 2.1834974276711108e-07, "loss": 1.0907, "step": 4873 }, { "epoch": 1.87, "learning_rate": 2.1704505953170863e-07, "loss": 1.1117, "step": 4874 }, { "epoch": 1.87, "learning_rate": 2.1574424309074527e-07, "loss": 1.0754, "step": 4875 }, { "epoch": 1.87, "learning_rate": 2.1444729395838082e-07, "loss": 1.129, "step": 4876 }, { "epoch": 1.87, "learning_rate": 2.1315421264724522e-07, "loss": 1.1169, "step": 4877 }, { "epoch": 1.87, "learning_rate": 2.118649996684363e-07, "loss": 1.0908, "step": 4878 }, { "epoch": 1.87, "learning_rate": 2.1057965553153093e-07, "loss": 1.0417, "step": 4879 }, { "epoch": 1.87, "learning_rate": 2.0929818074456932e-07, "loss": 1.0856, "step": 4880 }, { "epoch": 1.87, "learning_rate": 2.080205758140652e-07, "loss": 1.0939, "step": 4881 }, { "epoch": 1.88, "learning_rate": 2.0674684124500245e-07, "loss": 1.1079, "step": 4882 }, { "epoch": 1.88, "learning_rate": 2.0547697754083607e-07, "loss": 1.1141, "step": 4883 }, { "epoch": 1.88, "learning_rate": 2.0421098520349125e-07, "loss": 1.0826, "step": 4884 }, { "epoch": 1.88, "learning_rate": 2.0294886473335994e-07, "loss": 1.083, "step": 4885 }, { "epoch": 1.88, "learning_rate": 2.0169061662930645e-07, "loss": 1.0718, "step": 4886 }, { "epoch": 1.88, "learning_rate": 2.0043624138866402e-07, "loss": 1.1258, "step": 4887 }, { "epoch": 1.88, "learning_rate": 1.9918573950723385e-07, "loss": 1.1247, "step": 4888 }, { "epoch": 1.88, "learning_rate": 1.9793911147928945e-07, "loss": 1.0764, "step": 4889 }, { "epoch": 1.88, "learning_rate": 1.9669635779757e-07, "loss": 1.0923, "step": 4890 }, { "epoch": 1.88, "learning_rate": 1.954574789532826e-07, "loss": 1.1031, "step": 4891 }, { "epoch": 1.88, "learning_rate": 1.9422247543610662e-07, "loss": 1.1245, "step": 4892 }, { "epoch": 1.88, "learning_rate": 1.9299134773418714e-07, "loss": 1.0807, "step": 4893 }, { "epoch": 1.88, "learning_rate": 1.9176409633413607e-07, "loss": 1.0814, "step": 4894 }, { "epoch": 1.88, "learning_rate": 1.9054072172103754e-07, "loss": 1.1451, "step": 4895 }, { "epoch": 1.88, "learning_rate": 1.8932122437843592e-07, "loss": 1.1133, "step": 4896 }, { "epoch": 1.88, "learning_rate": 1.8810560478835339e-07, "loss": 1.0858, "step": 4897 }, { "epoch": 1.88, "learning_rate": 1.8689386343126893e-07, "loss": 1.0884, "step": 4898 }, { "epoch": 1.88, "learning_rate": 1.856860007861372e-07, "loss": 1.0952, "step": 4899 }, { "epoch": 1.88, "learning_rate": 1.8448201733037295e-07, "loss": 1.1088, "step": 4900 }, { "epoch": 1.88, "learning_rate": 1.8328191353986223e-07, "loss": 1.0709, "step": 4901 }, { "epoch": 1.88, "learning_rate": 1.820856898889556e-07, "loss": 1.0887, "step": 4902 }, { "epoch": 1.88, "learning_rate": 1.8089334685047144e-07, "loss": 1.1108, "step": 4903 }, { "epoch": 1.88, "learning_rate": 1.797048848956906e-07, "loss": 1.1099, "step": 4904 }, { "epoch": 1.88, "learning_rate": 1.785203044943662e-07, "loss": 1.0788, "step": 4905 }, { "epoch": 1.88, "learning_rate": 1.7733960611471145e-07, "loss": 1.0637, "step": 4906 }, { "epoch": 1.88, "learning_rate": 1.7616279022340644e-07, "loss": 1.1117, "step": 4907 }, { "epoch": 1.89, "learning_rate": 1.7498985728560015e-07, "loss": 1.1226, "step": 4908 }, { "epoch": 1.89, "learning_rate": 1.7382080776489952e-07, "loss": 1.1003, "step": 4909 }, { "epoch": 1.89, "learning_rate": 1.7265564212338714e-07, "loss": 1.0459, "step": 4910 }, { "epoch": 1.89, "learning_rate": 1.7149436082159908e-07, "loss": 1.2018, "step": 4911 }, { "epoch": 1.89, "learning_rate": 1.7033696431854485e-07, "loss": 1.1169, "step": 4912 }, { "epoch": 1.89, "learning_rate": 1.6918345307169405e-07, "loss": 1.0747, "step": 4913 }, { "epoch": 1.89, "learning_rate": 1.6803382753697973e-07, "loss": 1.1395, "step": 4914 }, { "epoch": 1.89, "learning_rate": 1.6688808816880399e-07, "loss": 1.0977, "step": 4915 }, { "epoch": 1.89, "learning_rate": 1.65746235420029e-07, "loss": 1.1309, "step": 4916 }, { "epoch": 1.89, "learning_rate": 1.646082697419804e-07, "loss": 1.0831, "step": 4917 }, { "epoch": 1.89, "learning_rate": 1.6347419158444956e-07, "loss": 1.0791, "step": 4918 }, { "epoch": 1.89, "learning_rate": 1.6234400139569006e-07, "loss": 1.0882, "step": 4919 }, { "epoch": 1.89, "learning_rate": 1.6121769962241905e-07, "loss": 1.0298, "step": 4920 }, { "epoch": 1.89, "learning_rate": 1.6009528670981711e-07, "loss": 1.1357, "step": 4921 }, { "epoch": 1.89, "learning_rate": 1.589767631015249e-07, "loss": 1.1314, "step": 4922 }, { "epoch": 1.89, "learning_rate": 1.5786212923965204e-07, "loss": 1.0871, "step": 4923 }, { "epoch": 1.89, "learning_rate": 1.567513855647651e-07, "loss": 1.0447, "step": 4924 }, { "epoch": 1.89, "learning_rate": 1.5564453251589284e-07, "loss": 1.1085, "step": 4925 }, { "epoch": 1.89, "learning_rate": 1.5454157053053086e-07, "loss": 1.0994, "step": 4926 }, { "epoch": 1.89, "learning_rate": 1.5344250004463378e-07, "loss": 1.11, "step": 4927 }, { "epoch": 1.89, "learning_rate": 1.5234732149261632e-07, "loss": 1.0868, "step": 4928 }, { "epoch": 1.89, "learning_rate": 1.5125603530736e-07, "loss": 1.128, "step": 4929 }, { "epoch": 1.89, "learning_rate": 1.5016864192020087e-07, "loss": 1.0767, "step": 4930 }, { "epoch": 1.89, "learning_rate": 1.4908514176094512e-07, "loss": 1.0513, "step": 4931 }, { "epoch": 1.89, "learning_rate": 1.4800553525785132e-07, "loss": 1.0712, "step": 4932 }, { "epoch": 1.89, "learning_rate": 1.469298228376437e-07, "loss": 1.0992, "step": 4933 }, { "epoch": 1.9, "learning_rate": 1.458580049255076e-07, "loss": 1.1273, "step": 4934 }, { "epoch": 1.9, "learning_rate": 1.4479008194508648e-07, "loss": 1.0671, "step": 4935 }, { "epoch": 1.9, "learning_rate": 1.4372605431848707e-07, "loss": 1.0367, "step": 4936 }, { "epoch": 1.9, "learning_rate": 1.426659224662752e-07, "loss": 1.101, "step": 4937 }, { "epoch": 1.9, "learning_rate": 1.4160968680747678e-07, "loss": 1.1268, "step": 4938 }, { "epoch": 1.9, "learning_rate": 1.4055734775957674e-07, "loss": 1.0776, "step": 4939 }, { "epoch": 1.9, "learning_rate": 1.3950890573852129e-07, "loss": 1.0814, "step": 4940 }, { "epoch": 1.9, "learning_rate": 1.3846436115871553e-07, "loss": 1.0901, "step": 4941 }, { "epoch": 1.9, "learning_rate": 1.3742371443302703e-07, "loss": 1.0974, "step": 4942 }, { "epoch": 1.9, "learning_rate": 1.3638696597277678e-07, "loss": 1.1077, "step": 4943 }, { "epoch": 1.9, "learning_rate": 1.3535411618774917e-07, "loss": 1.0819, "step": 4944 }, { "epoch": 1.9, "learning_rate": 1.3432516548618767e-07, "loss": 1.0741, "step": 4945 }, { "epoch": 1.9, "learning_rate": 1.3330011427479138e-07, "loss": 1.1036, "step": 4946 }, { "epoch": 1.9, "learning_rate": 1.32278962958724e-07, "loss": 1.0445, "step": 4947 }, { "epoch": 1.9, "learning_rate": 1.3126171194160043e-07, "loss": 1.0705, "step": 4948 }, { "epoch": 1.9, "learning_rate": 1.3024836162550014e-07, "loss": 1.115, "step": 4949 }, { "epoch": 1.9, "learning_rate": 1.292389124109561e-07, "loss": 1.137, "step": 4950 }, { "epoch": 1.9, "learning_rate": 1.2823336469696357e-07, "loss": 1.092, "step": 4951 }, { "epoch": 1.9, "learning_rate": 1.272317188809724e-07, "loss": 1.1027, "step": 4952 }, { "epoch": 1.9, "learning_rate": 1.262339753588926e-07, "loss": 1.1125, "step": 4953 }, { "epoch": 1.9, "learning_rate": 1.2524013452508865e-07, "loss": 1.049, "step": 4954 }, { "epoch": 1.9, "learning_rate": 1.2425019677238747e-07, "loss": 1.0962, "step": 4955 }, { "epoch": 1.9, "learning_rate": 1.2326416249206718e-07, "loss": 1.1287, "step": 4956 }, { "epoch": 1.9, "learning_rate": 1.2228203207386714e-07, "loss": 1.1189, "step": 4957 }, { "epoch": 1.9, "learning_rate": 1.2130380590598344e-07, "loss": 1.0763, "step": 4958 }, { "epoch": 1.9, "learning_rate": 1.2032948437506574e-07, "loss": 1.0859, "step": 4959 }, { "epoch": 1.91, "learning_rate": 1.1935906786622598e-07, "loss": 1.1093, "step": 4960 }, { "epoch": 1.91, "learning_rate": 1.183925567630273e-07, "loss": 1.155, "step": 4961 }, { "epoch": 1.91, "learning_rate": 1.1742995144749081e-07, "loss": 1.0854, "step": 4962 }, { "epoch": 1.91, "learning_rate": 1.1647125230009549e-07, "loss": 1.0831, "step": 4963 }, { "epoch": 1.91, "learning_rate": 1.1551645969977599e-07, "loss": 1.0714, "step": 4964 }, { "epoch": 1.91, "learning_rate": 1.1456557402391932e-07, "loss": 1.1365, "step": 4965 }, { "epoch": 1.91, "learning_rate": 1.1361859564837374e-07, "loss": 1.06, "step": 4966 }, { "epoch": 1.91, "learning_rate": 1.1267552494743871e-07, "loss": 1.0689, "step": 4967 }, { "epoch": 1.91, "learning_rate": 1.1173636229387164e-07, "loss": 1.0952, "step": 4968 }, { "epoch": 1.91, "learning_rate": 1.1080110805888333e-07, "loss": 1.0865, "step": 4969 }, { "epoch": 1.91, "learning_rate": 1.0986976261214143e-07, "loss": 1.0776, "step": 4970 }, { "epoch": 1.91, "learning_rate": 1.0894232632176926e-07, "loss": 1.0712, "step": 4971 }, { "epoch": 1.91, "learning_rate": 1.0801879955434024e-07, "loss": 1.0955, "step": 4972 }, { "epoch": 1.91, "learning_rate": 1.0709918267489017e-07, "loss": 1.0928, "step": 4973 }, { "epoch": 1.91, "learning_rate": 1.0618347604690271e-07, "loss": 1.0759, "step": 4974 }, { "epoch": 1.91, "learning_rate": 1.0527168003231835e-07, "loss": 1.0624, "step": 4975 }, { "epoch": 1.91, "learning_rate": 1.0436379499153437e-07, "loss": 1.068, "step": 4976 }, { "epoch": 1.91, "learning_rate": 1.0345982128339704e-07, "loss": 1.0804, "step": 4977 }, { "epoch": 1.91, "learning_rate": 1.0255975926521167e-07, "loss": 1.1521, "step": 4978 }, { "epoch": 1.91, "learning_rate": 1.0166360929273477e-07, "loss": 1.0387, "step": 4979 }, { "epoch": 1.91, "learning_rate": 1.0077137172017637e-07, "loss": 1.131, "step": 4980 }, { "epoch": 1.91, "learning_rate": 9.988304690020212e-08, "loss": 1.0653, "step": 4981 }, { "epoch": 1.91, "learning_rate": 9.899863518392783e-08, "loss": 1.0585, "step": 4982 }, { "epoch": 1.91, "learning_rate": 9.811813692092609e-08, "loss": 1.1223, "step": 4983 }, { "epoch": 1.91, "learning_rate": 9.724155245922073e-08, "loss": 1.0899, "step": 4984 }, { "epoch": 1.91, "learning_rate": 9.636888214528905e-08, "loss": 1.0878, "step": 4985 }, { "epoch": 1.92, "learning_rate": 9.55001263240618e-08, "loss": 1.0533, "step": 4986 }, { "epoch": 1.92, "learning_rate": 9.463528533892097e-08, "loss": 1.0494, "step": 4987 }, { "epoch": 1.92, "learning_rate": 9.377435953170199e-08, "loss": 1.0945, "step": 4988 }, { "epoch": 1.92, "learning_rate": 9.29173492426949e-08, "loss": 1.1632, "step": 4989 }, { "epoch": 1.92, "learning_rate": 9.206425481063763e-08, "loss": 1.0967, "step": 4990 }, { "epoch": 1.92, "learning_rate": 9.1215076572726e-08, "loss": 1.0896, "step": 4991 }, { "epoch": 1.92, "learning_rate": 9.03698148646015e-08, "loss": 1.1312, "step": 4992 }, { "epoch": 1.92, "learning_rate": 8.952847002036246e-08, "loss": 1.0898, "step": 4993 }, { "epoch": 1.92, "learning_rate": 8.869104237255732e-08, "loss": 1.0254, "step": 4994 }, { "epoch": 1.92, "learning_rate": 8.785753225218574e-08, "loss": 1.0844, "step": 4995 }, { "epoch": 1.92, "learning_rate": 8.702793998870084e-08, "loss": 1.1133, "step": 4996 }, { "epoch": 1.92, "learning_rate": 8.62022659100048e-08, "loss": 1.0829, "step": 4997 }, { "epoch": 1.92, "learning_rate": 8.53805103424521e-08, "loss": 1.0585, "step": 4998 }, { "epoch": 1.92, "learning_rate": 8.456267361084958e-08, "loss": 1.0841, "step": 4999 }, { "epoch": 1.92, "learning_rate": 8.374875603845312e-08, "loss": 1.1627, "step": 5000 }, { "epoch": 1.92, "learning_rate": 8.293875794696982e-08, "loss": 1.0504, "step": 5001 }, { "epoch": 1.92, "learning_rate": 8.213267965655914e-08, "loss": 1.0825, "step": 5002 }, { "epoch": 1.92, "learning_rate": 8.133052148583176e-08, "loss": 1.0471, "step": 5003 }, { "epoch": 1.92, "learning_rate": 8.053228375184408e-08, "loss": 1.0011, "step": 5004 }, { "epoch": 1.92, "learning_rate": 7.973796677010815e-08, "loss": 1.0757, "step": 5005 }, { "epoch": 1.92, "learning_rate": 7.894757085458282e-08, "loss": 1.1452, "step": 5006 }, { "epoch": 1.92, "learning_rate": 7.816109631768154e-08, "loss": 1.1211, "step": 5007 }, { "epoch": 1.92, "learning_rate": 7.73785434702634e-08, "loss": 1.1184, "step": 5008 }, { "epoch": 1.92, "learning_rate": 7.659991262163658e-08, "loss": 1.0881, "step": 5009 }, { "epoch": 1.92, "learning_rate": 7.582520407956595e-08, "loss": 1.1001, "step": 5010 }, { "epoch": 1.92, "learning_rate": 7.505441815025771e-08, "loss": 1.0617, "step": 5011 }, { "epoch": 1.93, "learning_rate": 7.428755513837371e-08, "loss": 1.1222, "step": 5012 }, { "epoch": 1.93, "learning_rate": 7.352461534702148e-08, "loss": 1.1227, "step": 5013 }, { "epoch": 1.93, "learning_rate": 7.276559907776093e-08, "loss": 1.0898, "step": 5014 }, { "epoch": 1.93, "learning_rate": 7.201050663059761e-08, "loss": 1.0742, "step": 5015 }, { "epoch": 1.93, "learning_rate": 7.125933830398946e-08, "loss": 1.0862, "step": 5016 }, { "epoch": 1.93, "learning_rate": 7.051209439484119e-08, "loss": 1.1288, "step": 5017 }, { "epoch": 1.93, "learning_rate": 6.976877519850877e-08, "loss": 1.0274, "step": 5018 }, { "epoch": 1.93, "learning_rate": 6.902938100879275e-08, "loss": 1.1234, "step": 5019 }, { "epoch": 1.93, "learning_rate": 6.829391211794712e-08, "loss": 1.0606, "step": 5020 }, { "epoch": 1.93, "learning_rate": 6.756236881667156e-08, "loss": 1.0799, "step": 5021 }, { "epoch": 1.93, "learning_rate": 6.683475139411366e-08, "loss": 1.0776, "step": 5022 }, { "epoch": 1.93, "learning_rate": 6.611106013787116e-08, "loss": 1.0882, "step": 5023 }, { "epoch": 1.93, "learning_rate": 6.539129533398858e-08, "loss": 1.0568, "step": 5024 }, { "epoch": 1.93, "learning_rate": 6.467545726695945e-08, "loss": 1.0746, "step": 5025 }, { "epoch": 1.93, "learning_rate": 6.396354621972412e-08, "loss": 1.0936, "step": 5026 }, { "epoch": 1.93, "learning_rate": 6.325556247367192e-08, "loss": 1.0796, "step": 5027 }, { "epoch": 1.93, "learning_rate": 6.255150630864016e-08, "loss": 1.146, "step": 5028 }, { "epoch": 1.93, "learning_rate": 6.185137800291063e-08, "loss": 1.1003, "step": 5029 }, { "epoch": 1.93, "learning_rate": 6.115517783321645e-08, "loss": 1.1107, "step": 5030 }, { "epoch": 1.93, "learning_rate": 6.046290607473637e-08, "loss": 1.0426, "step": 5031 }, { "epoch": 1.93, "learning_rate": 5.977456300109596e-08, "loss": 1.0866, "step": 5032 }, { "epoch": 1.93, "learning_rate": 5.9090148884368706e-08, "loss": 1.1255, "step": 5033 }, { "epoch": 1.93, "learning_rate": 5.840966399507597e-08, "loss": 1.0657, "step": 5034 }, { "epoch": 1.93, "learning_rate": 5.773310860218373e-08, "loss": 1.1523, "step": 5035 }, { "epoch": 1.93, "learning_rate": 5.706048297310807e-08, "loss": 1.0809, "step": 5036 }, { "epoch": 1.93, "learning_rate": 5.639178737370743e-08, "loss": 1.1073, "step": 5037 }, { "epoch": 1.94, "learning_rate": 5.57270220682915e-08, "loss": 1.0872, "step": 5038 }, { "epoch": 1.94, "learning_rate": 5.5066187319612327e-08, "loss": 1.1462, "step": 5039 }, { "epoch": 1.94, "learning_rate": 5.440928338887097e-08, "loss": 1.1067, "step": 5040 }, { "epoch": 1.94, "learning_rate": 5.375631053571639e-08, "loss": 1.1203, "step": 5041 }, { "epoch": 1.94, "learning_rate": 5.310726901823771e-08, "loss": 1.127, "step": 5042 }, { "epoch": 1.94, "learning_rate": 5.2462159092977474e-08, "loss": 1.063, "step": 5043 }, { "epoch": 1.94, "learning_rate": 5.182098101491839e-08, "loss": 1.0822, "step": 5044 }, { "epoch": 1.94, "learning_rate": 5.118373503749219e-08, "loss": 1.0926, "step": 5045 }, { "epoch": 1.94, "learning_rate": 5.055042141257627e-08, "loss": 1.0803, "step": 5046 }, { "epoch": 1.94, "learning_rate": 4.992104039049261e-08, "loss": 1.1309, "step": 5047 }, { "epoch": 1.94, "learning_rate": 4.929559222000779e-08, "loss": 1.0598, "step": 5048 }, { "epoch": 1.94, "learning_rate": 4.8674077148338495e-08, "loss": 1.0983, "step": 5049 }, { "epoch": 1.94, "learning_rate": 4.8056495421142656e-08, "loss": 1.0996, "step": 5050 }, { "epoch": 1.94, "learning_rate": 4.7442847282522795e-08, "loss": 1.0856, "step": 5051 }, { "epoch": 1.94, "learning_rate": 4.683313297503045e-08, "loss": 1.1014, "step": 5052 }, { "epoch": 1.94, "learning_rate": 4.622735273965839e-08, "loss": 1.0723, "step": 5053 }, { "epoch": 1.94, "learning_rate": 4.562550681584954e-08, "loss": 1.1311, "step": 5054 }, { "epoch": 1.94, "learning_rate": 4.502759544148472e-08, "loss": 1.0736, "step": 5055 }, { "epoch": 1.94, "learning_rate": 4.443361885289599e-08, "loss": 1.14, "step": 5056 }, { "epoch": 1.94, "learning_rate": 4.3843577284856664e-08, "loss": 1.0516, "step": 5057 }, { "epoch": 1.94, "learning_rate": 4.325747097058464e-08, "loss": 1.0986, "step": 5058 }, { "epoch": 1.94, "learning_rate": 4.26753001417457e-08, "loss": 1.0675, "step": 5059 }, { "epoch": 1.94, "learning_rate": 4.2097065028445796e-08, "loss": 1.1396, "step": 5060 }, { "epoch": 1.94, "learning_rate": 4.1522765859236537e-08, "loss": 1.076, "step": 5061 }, { "epoch": 1.94, "learning_rate": 4.0952402861117455e-08, "loss": 1.088, "step": 5062 }, { "epoch": 1.94, "learning_rate": 4.038597625952712e-08, "loss": 1.1374, "step": 5063 }, { "epoch": 1.95, "learning_rate": 3.982348627834975e-08, "loss": 1.0861, "step": 5064 }, { "epoch": 1.95, "learning_rate": 3.926493313991531e-08, "loss": 1.1187, "step": 5065 }, { "epoch": 1.95, "learning_rate": 3.871031706499606e-08, "loss": 1.137, "step": 5066 }, { "epoch": 1.95, "learning_rate": 3.815963827281e-08, "loss": 1.1066, "step": 5067 }, { "epoch": 1.95, "learning_rate": 3.761289698101522e-08, "loss": 1.0965, "step": 5068 }, { "epoch": 1.95, "learning_rate": 3.7070093405717764e-08, "loss": 1.134, "step": 5069 }, { "epoch": 1.95, "learning_rate": 3.653122776146489e-08, "loss": 1.0941, "step": 5070 }, { "epoch": 1.95, "learning_rate": 3.5996300261247344e-08, "loss": 1.1323, "step": 5071 }, { "epoch": 1.95, "learning_rate": 3.54653111164982e-08, "loss": 1.0598, "step": 5072 }, { "epoch": 1.95, "learning_rate": 3.493826053709848e-08, "loss": 1.0963, "step": 5073 }, { "epoch": 1.95, "learning_rate": 3.4415148731367086e-08, "loss": 1.0814, "step": 5074 }, { "epoch": 1.95, "learning_rate": 3.389597590606863e-08, "loss": 1.0925, "step": 5075 }, { "epoch": 1.95, "learning_rate": 3.3380742266411195e-08, "loss": 1.1001, "step": 5076 }, { "epoch": 1.95, "learning_rate": 3.286944801604408e-08, "loss": 1.0699, "step": 5077 }, { "epoch": 1.95, "learning_rate": 3.236209335706231e-08, "loss": 1.0892, "step": 5078 }, { "epoch": 1.95, "learning_rate": 3.185867848999991e-08, "loss": 1.0391, "step": 5079 }, { "epoch": 1.95, "learning_rate": 3.135920361383771e-08, "loss": 1.0843, "step": 5080 }, { "epoch": 1.95, "learning_rate": 3.086366892599557e-08, "loss": 1.1235, "step": 5081 }, { "epoch": 1.95, "learning_rate": 3.037207462233904e-08, "loss": 1.0811, "step": 5082 }, { "epoch": 1.95, "learning_rate": 2.9884420897173803e-08, "loss": 1.12, "step": 5083 }, { "epoch": 1.95, "learning_rate": 2.9400707943251227e-08, "loss": 1.0895, "step": 5084 }, { "epoch": 1.95, "learning_rate": 2.8920935951760598e-08, "loss": 1.1074, "step": 5085 }, { "epoch": 1.95, "learning_rate": 2.844510511233578e-08, "loss": 1.0713, "step": 5086 }, { "epoch": 1.95, "learning_rate": 2.7973215613054105e-08, "loss": 1.1337, "step": 5087 }, { "epoch": 1.95, "learning_rate": 2.7505267640434153e-08, "loss": 1.1067, "step": 5088 }, { "epoch": 1.95, "learning_rate": 2.7041261379434637e-08, "loss": 1.0552, "step": 5089 }, { "epoch": 1.96, "learning_rate": 2.658119701345885e-08, "loss": 1.099, "step": 5090 }, { "epoch": 1.96, "learning_rate": 2.612507472435133e-08, "loss": 1.0618, "step": 5091 }, { "epoch": 1.96, "learning_rate": 2.5672894692397865e-08, "loss": 1.1291, "step": 5092 }, { "epoch": 1.96, "learning_rate": 2.522465709632549e-08, "loss": 1.0789, "step": 5093 }, { "epoch": 1.96, "learning_rate": 2.478036211330581e-08, "loss": 1.0942, "step": 5094 }, { "epoch": 1.96, "learning_rate": 2.434000991894836e-08, "loss": 1.1043, "step": 5095 }, { "epoch": 1.96, "learning_rate": 2.3903600687306126e-08, "loss": 1.1141, "step": 5096 }, { "epoch": 1.96, "learning_rate": 2.3471134590874466e-08, "loss": 1.0327, "step": 5097 }, { "epoch": 1.96, "learning_rate": 2.3042611800588865e-08, "loss": 1.0941, "step": 5098 }, { "epoch": 1.96, "learning_rate": 2.2618032485827167e-08, "loss": 1.0594, "step": 5099 }, { "epoch": 1.96, "learning_rate": 2.2197396814406248e-08, "loss": 1.0833, "step": 5100 }, { "epoch": 1.96, "learning_rate": 2.178070495258644e-08, "loss": 1.1408, "step": 5101 }, { "epoch": 1.96, "learning_rate": 2.1367957065070443e-08, "loss": 1.1349, "step": 5102 }, { "epoch": 1.96, "learning_rate": 2.095915331499665e-08, "loss": 1.0903, "step": 5103 }, { "epoch": 1.96, "learning_rate": 2.0554293863952467e-08, "loss": 1.108, "step": 5104 }, { "epoch": 1.96, "learning_rate": 2.0153378871958784e-08, "loss": 1.0472, "step": 5105 }, { "epoch": 1.96, "learning_rate": 1.9756408497482173e-08, "loss": 1.0765, "step": 5106 }, { "epoch": 1.96, "learning_rate": 1.9363382897427118e-08, "loss": 1.0918, "step": 5107 }, { "epoch": 1.96, "learning_rate": 1.8974302227141585e-08, "loss": 1.0534, "step": 5108 }, { "epoch": 1.96, "learning_rate": 1.858916664041255e-08, "loss": 1.0558, "step": 5109 }, { "epoch": 1.96, "learning_rate": 1.8207976289468245e-08, "loss": 1.1142, "step": 5110 }, { "epoch": 1.96, "learning_rate": 1.7830731324977034e-08, "loss": 1.1201, "step": 5111 }, { "epoch": 1.96, "learning_rate": 1.745743189604743e-08, "loss": 1.1059, "step": 5112 }, { "epoch": 1.96, "learning_rate": 1.7088078150230282e-08, "loss": 1.0537, "step": 5113 }, { "epoch": 1.96, "learning_rate": 1.672267023351548e-08, "loss": 1.0677, "step": 5114 }, { "epoch": 1.96, "learning_rate": 1.6361208290333053e-08, "loss": 1.1096, "step": 5115 }, { "epoch": 1.97, "learning_rate": 1.6003692463554265e-08, "loss": 1.105, "step": 5116 }, { "epoch": 1.97, "learning_rate": 1.5650122894490526e-08, "loss": 1.0537, "step": 5117 }, { "epoch": 1.97, "learning_rate": 1.530049972289227e-08, "loss": 1.0997, "step": 5118 }, { "epoch": 1.97, "learning_rate": 1.4954823086952287e-08, "loss": 1.0517, "step": 5119 }, { "epoch": 1.97, "learning_rate": 1.461309312330017e-08, "loss": 1.0827, "step": 5120 }, { "epoch": 1.97, "learning_rate": 1.4275309967010098e-08, "loss": 1.0689, "step": 5121 }, { "epoch": 1.97, "learning_rate": 1.3941473751590829e-08, "loss": 1.0831, "step": 5122 }, { "epoch": 1.97, "learning_rate": 1.3611584608996808e-08, "loss": 1.055, "step": 5123 }, { "epoch": 1.97, "learning_rate": 1.3285642669617071e-08, "loss": 1.1036, "step": 5124 }, { "epoch": 1.97, "learning_rate": 1.2963648062284117e-08, "loss": 1.1426, "step": 5125 }, { "epoch": 1.97, "learning_rate": 1.2645600914268364e-08, "loss": 1.0749, "step": 5126 }, { "epoch": 1.97, "learning_rate": 1.2331501351280361e-08, "loss": 1.1141, "step": 5127 }, { "epoch": 1.97, "learning_rate": 1.2021349497473022e-08, "loss": 1.1101, "step": 5128 }, { "epoch": 1.97, "learning_rate": 1.1715145475432732e-08, "loss": 1.0474, "step": 5129 }, { "epoch": 1.97, "learning_rate": 1.1412889406192673e-08, "loss": 1.092, "step": 5130 }, { "epoch": 1.97, "learning_rate": 1.1114581409219505e-08, "loss": 1.0389, "step": 5131 }, { "epoch": 1.97, "learning_rate": 1.0820221602422243e-08, "loss": 1.0614, "step": 5132 }, { "epoch": 1.97, "learning_rate": 1.052981010215115e-08, "loss": 1.0463, "step": 5133 }, { "epoch": 1.97, "learning_rate": 1.0243347023191075e-08, "loss": 1.0336, "step": 5134 }, { "epoch": 1.97, "learning_rate": 9.960832478771442e-09, "loss": 1.0397, "step": 5135 }, { "epoch": 1.97, "learning_rate": 9.682266580556265e-09, "loss": 1.1146, "step": 5136 }, { "epoch": 1.97, "learning_rate": 9.407649438651911e-09, "loss": 1.1314, "step": 5137 }, { "epoch": 1.97, "learning_rate": 9.136981161603775e-09, "loss": 1.0665, "step": 5138 }, { "epoch": 1.97, "learning_rate": 8.870261856395168e-09, "loss": 1.0808, "step": 5139 }, { "epoch": 1.97, "learning_rate": 8.607491628448427e-09, "loss": 1.1597, "step": 5140 }, { "epoch": 1.97, "learning_rate": 8.348670581627139e-09, "loss": 1.0941, "step": 5141 }, { "epoch": 1.98, "learning_rate": 8.093798818230581e-09, "loss": 1.1333, "step": 5142 }, { "epoch": 1.98, "learning_rate": 7.842876439001501e-09, "loss": 1.0874, "step": 5143 }, { "epoch": 1.98, "learning_rate": 7.595903543116122e-09, "loss": 1.0755, "step": 5144 }, { "epoch": 1.98, "learning_rate": 7.352880228195247e-09, "loss": 1.1249, "step": 5145 }, { "epoch": 1.98, "learning_rate": 7.1138065902942585e-09, "loss": 1.12, "step": 5146 }, { "epoch": 1.98, "learning_rate": 6.878682723909791e-09, "loss": 1.0763, "step": 5147 }, { "epoch": 1.98, "learning_rate": 6.647508721975282e-09, "loss": 1.0833, "step": 5148 }, { "epoch": 1.98, "learning_rate": 6.420284675865418e-09, "loss": 1.0933, "step": 5149 }, { "epoch": 1.98, "learning_rate": 6.19701067539169e-09, "loss": 1.1123, "step": 5150 }, { "epoch": 1.98, "learning_rate": 5.977686808805727e-09, "loss": 1.0966, "step": 5151 }, { "epoch": 1.98, "learning_rate": 5.762313162797073e-09, "loss": 1.0849, "step": 5152 }, { "epoch": 1.98, "learning_rate": 5.550889822492078e-09, "loss": 1.1415, "step": 5153 }, { "epoch": 1.98, "learning_rate": 5.343416871460561e-09, "loss": 1.0754, "step": 5154 }, { "epoch": 1.98, "learning_rate": 5.139894391705813e-09, "loss": 1.1031, "step": 5155 }, { "epoch": 1.98, "learning_rate": 4.940322463672376e-09, "loss": 1.0957, "step": 5156 }, { "epoch": 1.98, "learning_rate": 4.744701166243815e-09, "loss": 1.0353, "step": 5157 }, { "epoch": 1.98, "learning_rate": 4.553030576738282e-09, "loss": 1.1788, "step": 5158 }, { "epoch": 1.98, "learning_rate": 4.365310770918507e-09, "loss": 1.0965, "step": 5159 }, { "epoch": 1.98, "learning_rate": 4.181541822979585e-09, "loss": 1.1232, "step": 5160 }, { "epoch": 1.98, "learning_rate": 4.001723805558966e-09, "loss": 1.1226, "step": 5161 }, { "epoch": 1.98, "learning_rate": 3.825856789732019e-09, "loss": 1.0228, "step": 5162 }, { "epoch": 1.98, "learning_rate": 3.653940845009807e-09, "loss": 1.0112, "step": 5163 }, { "epoch": 1.98, "learning_rate": 3.4859760393446407e-09, "loss": 1.0879, "step": 5164 }, { "epoch": 1.98, "learning_rate": 3.3219624391267468e-09, "loss": 1.0945, "step": 5165 }, { "epoch": 1.98, "learning_rate": 3.1619001091820477e-09, "loss": 1.1301, "step": 5166 }, { "epoch": 1.98, "learning_rate": 3.005789112777713e-09, "loss": 1.1079, "step": 5167 }, { "epoch": 1.99, "learning_rate": 2.853629511617717e-09, "loss": 1.0861, "step": 5168 }, { "epoch": 1.99, "learning_rate": 2.7054213658439522e-09, "loss": 1.096, "step": 5169 }, { "epoch": 1.99, "learning_rate": 2.561164734037336e-09, "loss": 1.0467, "step": 5170 }, { "epoch": 1.99, "learning_rate": 2.420859673216702e-09, "loss": 1.1082, "step": 5171 }, { "epoch": 1.99, "learning_rate": 2.2845062388388016e-09, "loss": 1.0846, "step": 5172 }, { "epoch": 1.99, "learning_rate": 2.1521044847983008e-09, "loss": 1.1262, "step": 5173 }, { "epoch": 1.99, "learning_rate": 2.023654463426672e-09, "loss": 1.0928, "step": 5174 }, { "epoch": 1.99, "learning_rate": 1.8991562254977447e-09, "loss": 1.0565, "step": 5175 }, { "epoch": 1.99, "learning_rate": 1.7786098202177137e-09, "loss": 1.0957, "step": 5176 }, { "epoch": 1.99, "learning_rate": 1.6620152952362412e-09, "loss": 1.1237, "step": 5177 }, { "epoch": 1.99, "learning_rate": 1.549372696635354e-09, "loss": 1.107, "step": 5178 }, { "epoch": 1.99, "learning_rate": 1.4406820689405466e-09, "loss": 1.0615, "step": 5179 }, { "epoch": 1.99, "learning_rate": 1.3359434551107887e-09, "loss": 1.0609, "step": 5180 }, { "epoch": 1.99, "learning_rate": 1.2351568965462967e-09, "loss": 1.0982, "step": 5181 }, { "epoch": 1.99, "learning_rate": 1.1383224330829833e-09, "loss": 1.1361, "step": 5182 }, { "epoch": 1.99, "learning_rate": 1.0454401029957873e-09, "loss": 1.0662, "step": 5183 }, { "epoch": 1.99, "learning_rate": 9.56509942996453e-10, "loss": 1.0634, "step": 5184 }, { "epoch": 1.99, "learning_rate": 8.715319882368622e-10, "loss": 1.0561, "step": 5185 }, { "epoch": 1.99, "learning_rate": 7.905062723034817e-10, "loss": 1.0592, "step": 5186 }, { "epoch": 1.99, "learning_rate": 7.134328272240254e-10, "loss": 1.1308, "step": 5187 }, { "epoch": 1.99, "learning_rate": 6.403116834619027e-10, "loss": 1.0422, "step": 5188 }, { "epoch": 1.99, "learning_rate": 5.711428699184396e-10, "loss": 1.1311, "step": 5189 }, { "epoch": 1.99, "learning_rate": 5.059264139339881e-10, "loss": 1.1092, "step": 5190 }, { "epoch": 1.99, "learning_rate": 4.4466234128459627e-10, "loss": 1.1363, "step": 5191 }, { "epoch": 1.99, "learning_rate": 3.8735067618644875e-10, "loss": 1.1056, "step": 5192 }, { "epoch": 1.99, "learning_rate": 3.339914412914258e-10, "loss": 1.0941, "step": 5193 }, { "epoch": 2.0, "learning_rate": 2.845846576915445e-10, "loss": 1.0823, "step": 5194 }, { "epoch": 2.0, "learning_rate": 2.391303449156279e-10, "loss": 1.1139, "step": 5195 }, { "epoch": 2.0, "learning_rate": 1.976285209270845e-10, "loss": 1.0723, "step": 5196 }, { "epoch": 2.0, "learning_rate": 1.6007920213279017e-10, "loss": 1.0754, "step": 5197 }, { "epoch": 2.0, "learning_rate": 1.2648240337309603e-10, "loss": 1.0497, "step": 5198 }, { "epoch": 2.0, "learning_rate": 9.683813792737973e-11, "loss": 1.09, "step": 5199 }, { "epoch": 2.0, "learning_rate": 7.114641751404528e-11, "loss": 1.0296, "step": 5200 }, { "epoch": 2.0, "eval_loss": 1.1039079427719116, "eval_runtime": 688.5522, "eval_samples_per_second": 4.939, "eval_steps_per_second": 0.309, "step": 5200 }, { "epoch": 2.0, "learning_rate": 4.940725228608223e-11, "loss": 1.1207, "step": 5201 }, { "epoch": 2.0, "learning_rate": 3.162065083661681e-11, "loss": 1.1134, "step": 5202 }, { "epoch": 2.0, "learning_rate": 1.7786620196691418e-11, "loss": 1.0811, "step": 5203 }, { "epoch": 2.0, "learning_rate": 7.90516583415446e-12, "loss": 1.0858, "step": 5204 }, { "epoch": 2.0, "learning_rate": 1.9762916536603118e-12, "loss": 1.1374, "step": 5205 }, { "epoch": 2.0, "learning_rate": 0.0, "loss": 1.1083, "step": 5206 }, { "epoch": 2.0, "step": 5206, "total_flos": 5.264450754214625e+19, "train_loss": 1.143068860665313, "train_runtime": 219037.4015, "train_samples_per_second": 1.521, "train_steps_per_second": 0.024 } ], "max_steps": 5206, "num_train_epochs": 2, "total_flos": 5.264450754214625e+19, "trial_name": null, "trial_params": null }