{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "eval_steps": 500, "global_step": 5862, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 2.1276595744680852e-07, "loss": 0.6677, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.2553191489361704e-07, "loss": 0.764, "step": 2 }, { "epoch": 0.0, "learning_rate": 6.382978723404255e-07, "loss": 0.7506, "step": 3 }, { "epoch": 0.0, "learning_rate": 8.510638297872341e-07, "loss": 0.7243, "step": 4 }, { "epoch": 0.0, "learning_rate": 1.0638297872340427e-06, "loss": 0.768, "step": 5 }, { "epoch": 0.0, "learning_rate": 1.276595744680851e-06, "loss": 0.788, "step": 6 }, { "epoch": 0.0, "learning_rate": 1.4893617021276598e-06, "loss": 0.7433, "step": 7 }, { "epoch": 0.0, "learning_rate": 1.7021276595744682e-06, "loss": 0.7169, "step": 8 }, { "epoch": 0.0, "learning_rate": 1.9148936170212767e-06, "loss": 0.6692, "step": 9 }, { "epoch": 0.0, "learning_rate": 2.1276595744680853e-06, "loss": 0.6926, "step": 10 }, { "epoch": 0.0, "learning_rate": 2.3404255319148935e-06, "loss": 0.6667, "step": 11 }, { "epoch": 0.0, "learning_rate": 2.553191489361702e-06, "loss": 0.5724, "step": 12 }, { "epoch": 0.0, "learning_rate": 2.7659574468085106e-06, "loss": 0.638, "step": 13 }, { "epoch": 0.0, "learning_rate": 2.9787234042553196e-06, "loss": 0.5975, "step": 14 }, { "epoch": 0.01, "learning_rate": 3.1914893617021277e-06, "loss": 0.5579, "step": 15 }, { "epoch": 0.01, "learning_rate": 3.4042553191489363e-06, "loss": 0.5203, "step": 16 }, { "epoch": 0.01, "learning_rate": 3.6170212765957445e-06, "loss": 0.5005, "step": 17 }, { "epoch": 0.01, "learning_rate": 3.8297872340425535e-06, "loss": 0.4405, "step": 18 }, { "epoch": 0.01, "learning_rate": 4.0425531914893625e-06, "loss": 0.5277, "step": 19 }, { "epoch": 0.01, "learning_rate": 4.255319148936171e-06, "loss": 0.4405, "step": 20 }, { "epoch": 0.01, "learning_rate": 4.468085106382979e-06, "loss": 0.4822, "step": 21 }, { "epoch": 0.01, "learning_rate": 4.680851063829787e-06, "loss": 0.5165, "step": 22 }, { "epoch": 0.01, "learning_rate": 4.893617021276596e-06, "loss": 0.4441, "step": 23 }, { "epoch": 0.01, "learning_rate": 5.106382978723404e-06, "loss": 0.4173, "step": 24 }, { "epoch": 0.01, "learning_rate": 5.319148936170213e-06, "loss": 0.4572, "step": 25 }, { "epoch": 0.01, "learning_rate": 5.531914893617021e-06, "loss": 0.4769, "step": 26 }, { "epoch": 0.01, "learning_rate": 5.74468085106383e-06, "loss": 0.4042, "step": 27 }, { "epoch": 0.01, "learning_rate": 5.957446808510639e-06, "loss": 0.4513, "step": 28 }, { "epoch": 0.01, "learning_rate": 6.170212765957447e-06, "loss": 0.3607, "step": 29 }, { "epoch": 0.01, "learning_rate": 6.3829787234042555e-06, "loss": 0.4167, "step": 30 }, { "epoch": 0.01, "learning_rate": 6.595744680851064e-06, "loss": 0.4514, "step": 31 }, { "epoch": 0.01, "learning_rate": 6.808510638297873e-06, "loss": 0.3886, "step": 32 }, { "epoch": 0.01, "learning_rate": 7.021276595744682e-06, "loss": 0.3782, "step": 33 }, { "epoch": 0.01, "learning_rate": 7.234042553191489e-06, "loss": 0.4119, "step": 34 }, { "epoch": 0.01, "learning_rate": 7.446808510638298e-06, "loss": 0.3471, "step": 35 }, { "epoch": 0.01, "learning_rate": 7.659574468085107e-06, "loss": 0.3922, "step": 36 }, { "epoch": 0.01, "learning_rate": 7.872340425531916e-06, "loss": 0.3973, "step": 37 }, { "epoch": 0.01, "learning_rate": 8.085106382978725e-06, "loss": 0.4148, "step": 38 }, { "epoch": 0.01, "learning_rate": 8.297872340425532e-06, "loss": 0.4031, "step": 39 }, { "epoch": 0.01, "learning_rate": 8.510638297872341e-06, "loss": 0.3396, "step": 40 }, { "epoch": 0.01, "learning_rate": 8.72340425531915e-06, "loss": 0.4384, "step": 41 }, { "epoch": 0.01, "learning_rate": 8.936170212765958e-06, "loss": 0.3539, "step": 42 }, { "epoch": 0.01, "learning_rate": 9.148936170212767e-06, "loss": 0.368, "step": 43 }, { "epoch": 0.02, "learning_rate": 9.361702127659574e-06, "loss": 0.3847, "step": 44 }, { "epoch": 0.02, "learning_rate": 9.574468085106383e-06, "loss": 0.4405, "step": 45 }, { "epoch": 0.02, "learning_rate": 9.787234042553192e-06, "loss": 0.3488, "step": 46 }, { "epoch": 0.02, "learning_rate": 1e-05, "loss": 0.3636, "step": 47 }, { "epoch": 0.02, "learning_rate": 1.0212765957446808e-05, "loss": 0.3444, "step": 48 }, { "epoch": 0.02, "learning_rate": 1.0425531914893617e-05, "loss": 0.376, "step": 49 }, { "epoch": 0.02, "learning_rate": 1.0638297872340426e-05, "loss": 0.3506, "step": 50 }, { "epoch": 0.02, "learning_rate": 1.0851063829787235e-05, "loss": 0.3647, "step": 51 }, { "epoch": 0.02, "learning_rate": 1.1063829787234042e-05, "loss": 0.3821, "step": 52 }, { "epoch": 0.02, "learning_rate": 1.1276595744680851e-05, "loss": 0.3962, "step": 53 }, { "epoch": 0.02, "learning_rate": 1.148936170212766e-05, "loss": 0.4277, "step": 54 }, { "epoch": 0.02, "learning_rate": 1.170212765957447e-05, "loss": 0.344, "step": 55 }, { "epoch": 0.02, "learning_rate": 1.1914893617021278e-05, "loss": 0.3481, "step": 56 }, { "epoch": 0.02, "learning_rate": 1.2127659574468086e-05, "loss": 0.3727, "step": 57 }, { "epoch": 0.02, "learning_rate": 1.2340425531914895e-05, "loss": 0.3648, "step": 58 }, { "epoch": 0.02, "learning_rate": 1.2553191489361702e-05, "loss": 0.3576, "step": 59 }, { "epoch": 0.02, "learning_rate": 1.2765957446808511e-05, "loss": 0.4458, "step": 60 }, { "epoch": 0.02, "learning_rate": 1.2978723404255318e-05, "loss": 0.3467, "step": 61 }, { "epoch": 0.02, "learning_rate": 1.3191489361702127e-05, "loss": 0.3488, "step": 62 }, { "epoch": 0.02, "learning_rate": 1.3404255319148936e-05, "loss": 0.3589, "step": 63 }, { "epoch": 0.02, "learning_rate": 1.3617021276595745e-05, "loss": 0.3507, "step": 64 }, { "epoch": 0.02, "learning_rate": 1.3829787234042554e-05, "loss": 0.3901, "step": 65 }, { "epoch": 0.02, "learning_rate": 1.4042553191489363e-05, "loss": 0.3814, "step": 66 }, { "epoch": 0.02, "learning_rate": 1.4255319148936172e-05, "loss": 0.3728, "step": 67 }, { "epoch": 0.02, "learning_rate": 1.4468085106382978e-05, "loss": 0.3987, "step": 68 }, { "epoch": 0.02, "learning_rate": 1.4680851063829787e-05, "loss": 0.3301, "step": 69 }, { "epoch": 0.02, "learning_rate": 1.4893617021276596e-05, "loss": 0.3474, "step": 70 }, { "epoch": 0.02, "learning_rate": 1.5106382978723405e-05, "loss": 0.4003, "step": 71 }, { "epoch": 0.02, "learning_rate": 1.5319148936170214e-05, "loss": 0.4107, "step": 72 }, { "epoch": 0.02, "learning_rate": 1.5531914893617023e-05, "loss": 0.3792, "step": 73 }, { "epoch": 0.03, "learning_rate": 1.5744680851063832e-05, "loss": 0.3952, "step": 74 }, { "epoch": 0.03, "learning_rate": 1.595744680851064e-05, "loss": 0.4233, "step": 75 }, { "epoch": 0.03, "learning_rate": 1.617021276595745e-05, "loss": 0.3341, "step": 76 }, { "epoch": 0.03, "learning_rate": 1.6382978723404255e-05, "loss": 0.3294, "step": 77 }, { "epoch": 0.03, "learning_rate": 1.6595744680851064e-05, "loss": 0.3623, "step": 78 }, { "epoch": 0.03, "learning_rate": 1.6808510638297873e-05, "loss": 0.3712, "step": 79 }, { "epoch": 0.03, "learning_rate": 1.7021276595744682e-05, "loss": 0.3895, "step": 80 }, { "epoch": 0.03, "learning_rate": 1.723404255319149e-05, "loss": 0.3816, "step": 81 }, { "epoch": 0.03, "learning_rate": 1.74468085106383e-05, "loss": 0.384, "step": 82 }, { "epoch": 0.03, "learning_rate": 1.7659574468085106e-05, "loss": 0.3701, "step": 83 }, { "epoch": 0.03, "learning_rate": 1.7872340425531915e-05, "loss": 0.3453, "step": 84 }, { "epoch": 0.03, "learning_rate": 1.8085106382978724e-05, "loss": 0.3908, "step": 85 }, { "epoch": 0.03, "learning_rate": 1.8297872340425533e-05, "loss": 0.3829, "step": 86 }, { "epoch": 0.03, "learning_rate": 1.851063829787234e-05, "loss": 0.4038, "step": 87 }, { "epoch": 0.03, "learning_rate": 1.8723404255319148e-05, "loss": 0.4345, "step": 88 }, { "epoch": 0.03, "learning_rate": 1.8936170212765957e-05, "loss": 0.3702, "step": 89 }, { "epoch": 0.03, "learning_rate": 1.9148936170212766e-05, "loss": 0.4068, "step": 90 }, { "epoch": 0.03, "learning_rate": 1.9361702127659575e-05, "loss": 0.386, "step": 91 }, { "epoch": 0.03, "learning_rate": 1.9574468085106384e-05, "loss": 0.3929, "step": 92 }, { "epoch": 0.03, "learning_rate": 1.9787234042553193e-05, "loss": 0.3897, "step": 93 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 0.3529, "step": 94 }, { "epoch": 0.03, "learning_rate": 2.0212765957446807e-05, "loss": 0.3884, "step": 95 }, { "epoch": 0.03, "learning_rate": 2.0425531914893616e-05, "loss": 0.391, "step": 96 }, { "epoch": 0.03, "learning_rate": 2.0638297872340425e-05, "loss": 0.4112, "step": 97 }, { "epoch": 0.03, "learning_rate": 2.0851063829787234e-05, "loss": 0.4045, "step": 98 }, { "epoch": 0.03, "learning_rate": 2.1063829787234043e-05, "loss": 0.3663, "step": 99 }, { "epoch": 0.03, "learning_rate": 2.1276595744680852e-05, "loss": 0.3779, "step": 100 }, { "epoch": 0.03, "learning_rate": 2.148936170212766e-05, "loss": 0.4283, "step": 101 }, { "epoch": 0.03, "learning_rate": 2.170212765957447e-05, "loss": 0.3949, "step": 102 }, { "epoch": 0.04, "learning_rate": 2.191489361702128e-05, "loss": 0.3788, "step": 103 }, { "epoch": 0.04, "learning_rate": 2.2127659574468085e-05, "loss": 0.3533, "step": 104 }, { "epoch": 0.04, "learning_rate": 2.2340425531914894e-05, "loss": 0.3784, "step": 105 }, { "epoch": 0.04, "learning_rate": 2.2553191489361703e-05, "loss": 0.375, "step": 106 }, { "epoch": 0.04, "learning_rate": 2.2765957446808512e-05, "loss": 0.3847, "step": 107 }, { "epoch": 0.04, "learning_rate": 2.297872340425532e-05, "loss": 0.3854, "step": 108 }, { "epoch": 0.04, "learning_rate": 2.319148936170213e-05, "loss": 0.3837, "step": 109 }, { "epoch": 0.04, "learning_rate": 2.340425531914894e-05, "loss": 0.3928, "step": 110 }, { "epoch": 0.04, "learning_rate": 2.3617021276595748e-05, "loss": 0.3398, "step": 111 }, { "epoch": 0.04, "learning_rate": 2.3829787234042557e-05, "loss": 0.3558, "step": 112 }, { "epoch": 0.04, "learning_rate": 2.4042553191489362e-05, "loss": 0.3972, "step": 113 }, { "epoch": 0.04, "learning_rate": 2.425531914893617e-05, "loss": 0.4176, "step": 114 }, { "epoch": 0.04, "learning_rate": 2.446808510638298e-05, "loss": 0.4012, "step": 115 }, { "epoch": 0.04, "learning_rate": 2.468085106382979e-05, "loss": 0.3635, "step": 116 }, { "epoch": 0.04, "learning_rate": 2.48936170212766e-05, "loss": 0.3842, "step": 117 }, { "epoch": 0.04, "learning_rate": 2.5106382978723404e-05, "loss": 0.3543, "step": 118 }, { "epoch": 0.04, "learning_rate": 2.5319148936170213e-05, "loss": 0.3933, "step": 119 }, { "epoch": 0.04, "learning_rate": 2.5531914893617022e-05, "loss": 0.4029, "step": 120 }, { "epoch": 0.04, "learning_rate": 2.574468085106383e-05, "loss": 0.3621, "step": 121 }, { "epoch": 0.04, "learning_rate": 2.5957446808510637e-05, "loss": 0.3908, "step": 122 }, { "epoch": 0.04, "learning_rate": 2.6170212765957446e-05, "loss": 0.37, "step": 123 }, { "epoch": 0.04, "learning_rate": 2.6382978723404255e-05, "loss": 0.3784, "step": 124 }, { "epoch": 0.04, "learning_rate": 2.6595744680851064e-05, "loss": 0.327, "step": 125 }, { "epoch": 0.04, "learning_rate": 2.6808510638297873e-05, "loss": 0.3395, "step": 126 }, { "epoch": 0.04, "learning_rate": 2.702127659574468e-05, "loss": 0.3359, "step": 127 }, { "epoch": 0.04, "learning_rate": 2.723404255319149e-05, "loss": 0.3664, "step": 128 }, { "epoch": 0.04, "learning_rate": 2.74468085106383e-05, "loss": 0.4259, "step": 129 }, { "epoch": 0.04, "learning_rate": 2.765957446808511e-05, "loss": 0.3788, "step": 130 }, { "epoch": 0.04, "learning_rate": 2.7872340425531918e-05, "loss": 0.3661, "step": 131 }, { "epoch": 0.05, "learning_rate": 2.8085106382978727e-05, "loss": 0.3618, "step": 132 }, { "epoch": 0.05, "learning_rate": 2.8297872340425536e-05, "loss": 0.3453, "step": 133 }, { "epoch": 0.05, "learning_rate": 2.8510638297872345e-05, "loss": 0.4179, "step": 134 }, { "epoch": 0.05, "learning_rate": 2.8723404255319154e-05, "loss": 0.3756, "step": 135 }, { "epoch": 0.05, "learning_rate": 2.8936170212765956e-05, "loss": 0.4077, "step": 136 }, { "epoch": 0.05, "learning_rate": 2.9148936170212765e-05, "loss": 0.4112, "step": 137 }, { "epoch": 0.05, "learning_rate": 2.9361702127659574e-05, "loss": 0.3537, "step": 138 }, { "epoch": 0.05, "learning_rate": 2.9574468085106383e-05, "loss": 0.3797, "step": 139 }, { "epoch": 0.05, "learning_rate": 2.9787234042553192e-05, "loss": 0.3603, "step": 140 }, { "epoch": 0.05, "learning_rate": 3e-05, "loss": 0.3871, "step": 141 }, { "epoch": 0.05, "learning_rate": 3.021276595744681e-05, "loss": 0.419, "step": 142 }, { "epoch": 0.05, "learning_rate": 3.042553191489362e-05, "loss": 0.3761, "step": 143 }, { "epoch": 0.05, "learning_rate": 3.063829787234043e-05, "loss": 0.3628, "step": 144 }, { "epoch": 0.05, "learning_rate": 3.085106382978723e-05, "loss": 0.3537, "step": 145 }, { "epoch": 0.05, "learning_rate": 3.1063829787234046e-05, "loss": 0.3989, "step": 146 }, { "epoch": 0.05, "learning_rate": 3.127659574468085e-05, "loss": 0.3834, "step": 147 }, { "epoch": 0.05, "learning_rate": 3.1489361702127664e-05, "loss": 0.3811, "step": 148 }, { "epoch": 0.05, "learning_rate": 3.170212765957447e-05, "loss": 0.4016, "step": 149 }, { "epoch": 0.05, "learning_rate": 3.191489361702128e-05, "loss": 0.4284, "step": 150 }, { "epoch": 0.05, "learning_rate": 3.212765957446809e-05, "loss": 0.3395, "step": 151 }, { "epoch": 0.05, "learning_rate": 3.23404255319149e-05, "loss": 0.3949, "step": 152 }, { "epoch": 0.05, "learning_rate": 3.2553191489361705e-05, "loss": 0.4735, "step": 153 }, { "epoch": 0.05, "learning_rate": 3.276595744680851e-05, "loss": 0.3722, "step": 154 }, { "epoch": 0.05, "learning_rate": 3.2978723404255317e-05, "loss": 0.3791, "step": 155 }, { "epoch": 0.05, "learning_rate": 3.319148936170213e-05, "loss": 0.4151, "step": 156 }, { "epoch": 0.05, "learning_rate": 3.3404255319148935e-05, "loss": 0.3837, "step": 157 }, { "epoch": 0.05, "learning_rate": 3.361702127659575e-05, "loss": 0.34, "step": 158 }, { "epoch": 0.05, "learning_rate": 3.382978723404255e-05, "loss": 0.4021, "step": 159 }, { "epoch": 0.05, "learning_rate": 3.4042553191489365e-05, "loss": 0.4003, "step": 160 }, { "epoch": 0.05, "learning_rate": 3.425531914893617e-05, "loss": 0.3571, "step": 161 }, { "epoch": 0.06, "learning_rate": 3.446808510638298e-05, "loss": 0.3964, "step": 162 }, { "epoch": 0.06, "learning_rate": 3.468085106382979e-05, "loss": 0.3846, "step": 163 }, { "epoch": 0.06, "learning_rate": 3.48936170212766e-05, "loss": 0.3738, "step": 164 }, { "epoch": 0.06, "learning_rate": 3.5106382978723407e-05, "loss": 0.3824, "step": 165 }, { "epoch": 0.06, "learning_rate": 3.531914893617021e-05, "loss": 0.3824, "step": 166 }, { "epoch": 0.06, "learning_rate": 3.5531914893617025e-05, "loss": 0.4301, "step": 167 }, { "epoch": 0.06, "learning_rate": 3.574468085106383e-05, "loss": 0.3413, "step": 168 }, { "epoch": 0.06, "learning_rate": 3.595744680851064e-05, "loss": 0.3921, "step": 169 }, { "epoch": 0.06, "learning_rate": 3.617021276595745e-05, "loss": 0.4033, "step": 170 }, { "epoch": 0.06, "learning_rate": 3.638297872340426e-05, "loss": 0.4013, "step": 171 }, { "epoch": 0.06, "learning_rate": 3.6595744680851066e-05, "loss": 0.3857, "step": 172 }, { "epoch": 0.06, "learning_rate": 3.680851063829787e-05, "loss": 0.3854, "step": 173 }, { "epoch": 0.06, "learning_rate": 3.702127659574468e-05, "loss": 0.3864, "step": 174 }, { "epoch": 0.06, "learning_rate": 3.723404255319149e-05, "loss": 0.346, "step": 175 }, { "epoch": 0.06, "learning_rate": 3.7446808510638295e-05, "loss": 0.3573, "step": 176 }, { "epoch": 0.06, "learning_rate": 3.765957446808511e-05, "loss": 0.3976, "step": 177 }, { "epoch": 0.06, "learning_rate": 3.787234042553191e-05, "loss": 0.4101, "step": 178 }, { "epoch": 0.06, "learning_rate": 3.8085106382978726e-05, "loss": 0.4003, "step": 179 }, { "epoch": 0.06, "learning_rate": 3.829787234042553e-05, "loss": 0.3662, "step": 180 }, { "epoch": 0.06, "learning_rate": 3.8510638297872344e-05, "loss": 0.3493, "step": 181 }, { "epoch": 0.06, "learning_rate": 3.872340425531915e-05, "loss": 0.3725, "step": 182 }, { "epoch": 0.06, "learning_rate": 3.893617021276596e-05, "loss": 0.6858, "step": 183 }, { "epoch": 0.06, "learning_rate": 3.914893617021277e-05, "loss": 0.3415, "step": 184 }, { "epoch": 0.06, "learning_rate": 3.936170212765958e-05, "loss": 0.3658, "step": 185 }, { "epoch": 0.06, "learning_rate": 3.9574468085106385e-05, "loss": 0.3586, "step": 186 }, { "epoch": 0.06, "learning_rate": 3.97872340425532e-05, "loss": 0.4025, "step": 187 }, { "epoch": 0.06, "learning_rate": 4e-05, "loss": 0.3759, "step": 188 }, { "epoch": 0.06, "learning_rate": 4.0212765957446816e-05, "loss": 0.4257, "step": 189 }, { "epoch": 0.06, "learning_rate": 4.0425531914893614e-05, "loss": 0.4015, "step": 190 }, { "epoch": 0.07, "learning_rate": 4.063829787234043e-05, "loss": 0.4202, "step": 191 }, { "epoch": 0.07, "learning_rate": 4.085106382978723e-05, "loss": 0.3657, "step": 192 }, { "epoch": 0.07, "learning_rate": 4.1063829787234045e-05, "loss": 0.4086, "step": 193 }, { "epoch": 0.07, "learning_rate": 4.127659574468085e-05, "loss": 0.3942, "step": 194 }, { "epoch": 0.07, "learning_rate": 4.148936170212766e-05, "loss": 0.4417, "step": 195 }, { "epoch": 0.07, "learning_rate": 4.170212765957447e-05, "loss": 0.3837, "step": 196 }, { "epoch": 0.07, "learning_rate": 4.191489361702128e-05, "loss": 0.3539, "step": 197 }, { "epoch": 0.07, "learning_rate": 4.2127659574468086e-05, "loss": 0.4167, "step": 198 }, { "epoch": 0.07, "learning_rate": 4.23404255319149e-05, "loss": 0.4009, "step": 199 }, { "epoch": 0.07, "learning_rate": 4.2553191489361704e-05, "loss": 0.3925, "step": 200 }, { "epoch": 0.07, "learning_rate": 4.276595744680851e-05, "loss": 0.4209, "step": 201 }, { "epoch": 0.07, "learning_rate": 4.297872340425532e-05, "loss": 0.3469, "step": 202 }, { "epoch": 0.07, "learning_rate": 4.319148936170213e-05, "loss": 0.3514, "step": 203 }, { "epoch": 0.07, "learning_rate": 4.340425531914894e-05, "loss": 0.3768, "step": 204 }, { "epoch": 0.07, "learning_rate": 4.3617021276595746e-05, "loss": 0.3745, "step": 205 }, { "epoch": 0.07, "learning_rate": 4.382978723404256e-05, "loss": 0.4612, "step": 206 }, { "epoch": 0.07, "learning_rate": 4.4042553191489364e-05, "loss": 0.3555, "step": 207 }, { "epoch": 0.07, "learning_rate": 4.425531914893617e-05, "loss": 0.4696, "step": 208 }, { "epoch": 0.07, "learning_rate": 4.4468085106382975e-05, "loss": 0.3571, "step": 209 }, { "epoch": 0.07, "learning_rate": 4.468085106382979e-05, "loss": 0.4076, "step": 210 }, { "epoch": 0.07, "learning_rate": 4.489361702127659e-05, "loss": 0.4056, "step": 211 }, { "epoch": 0.07, "learning_rate": 4.5106382978723406e-05, "loss": 0.3665, "step": 212 }, { "epoch": 0.07, "learning_rate": 4.531914893617021e-05, "loss": 0.4442, "step": 213 }, { "epoch": 0.07, "learning_rate": 4.5531914893617024e-05, "loss": 0.3823, "step": 214 }, { "epoch": 0.07, "learning_rate": 4.574468085106383e-05, "loss": 0.3484, "step": 215 }, { "epoch": 0.07, "learning_rate": 4.595744680851064e-05, "loss": 0.3711, "step": 216 }, { "epoch": 0.07, "learning_rate": 4.617021276595745e-05, "loss": 0.355, "step": 217 }, { "epoch": 0.07, "learning_rate": 4.638297872340426e-05, "loss": 0.4111, "step": 218 }, { "epoch": 0.07, "learning_rate": 4.6595744680851065e-05, "loss": 0.3628, "step": 219 }, { "epoch": 0.08, "learning_rate": 4.680851063829788e-05, "loss": 0.4505, "step": 220 }, { "epoch": 0.08, "learning_rate": 4.702127659574468e-05, "loss": 0.3718, "step": 221 }, { "epoch": 0.08, "learning_rate": 4.7234042553191496e-05, "loss": 0.4159, "step": 222 }, { "epoch": 0.08, "learning_rate": 4.74468085106383e-05, "loss": 0.3452, "step": 223 }, { "epoch": 0.08, "learning_rate": 4.7659574468085114e-05, "loss": 0.4174, "step": 224 }, { "epoch": 0.08, "learning_rate": 4.787234042553192e-05, "loss": 0.4268, "step": 225 }, { "epoch": 0.08, "learning_rate": 4.8085106382978725e-05, "loss": 0.438, "step": 226 }, { "epoch": 0.08, "learning_rate": 4.829787234042553e-05, "loss": 0.4401, "step": 227 }, { "epoch": 0.08, "learning_rate": 4.851063829787234e-05, "loss": 0.384, "step": 228 }, { "epoch": 0.08, "learning_rate": 4.872340425531915e-05, "loss": 0.3921, "step": 229 }, { "epoch": 0.08, "learning_rate": 4.893617021276596e-05, "loss": 0.3706, "step": 230 }, { "epoch": 0.08, "learning_rate": 4.9148936170212766e-05, "loss": 0.4154, "step": 231 }, { "epoch": 0.08, "learning_rate": 4.936170212765958e-05, "loss": 0.3774, "step": 232 }, { "epoch": 0.08, "learning_rate": 4.9574468085106384e-05, "loss": 0.4357, "step": 233 }, { "epoch": 0.08, "learning_rate": 4.97872340425532e-05, "loss": 0.3946, "step": 234 }, { "epoch": 0.08, "learning_rate": 5e-05, "loss": 0.3855, "step": 235 }, { "epoch": 0.08, "learning_rate": 4.999999610366835e-05, "loss": 0.3769, "step": 236 }, { "epoch": 0.08, "learning_rate": 4.9999984414674593e-05, "loss": 0.3621, "step": 237 }, { "epoch": 0.08, "learning_rate": 4.9999964933022394e-05, "loss": 0.379, "step": 238 }, { "epoch": 0.08, "learning_rate": 4.999993765871782e-05, "loss": 0.3987, "step": 239 }, { "epoch": 0.08, "learning_rate": 4.999990259176936e-05, "loss": 0.4201, "step": 240 }, { "epoch": 0.08, "learning_rate": 4.999985973218796e-05, "loss": 0.4762, "step": 241 }, { "epoch": 0.08, "learning_rate": 4.999980907998697e-05, "loss": 0.4307, "step": 242 }, { "epoch": 0.08, "learning_rate": 4.999975063518218e-05, "loss": 0.4027, "step": 243 }, { "epoch": 0.08, "learning_rate": 4.999968439779181e-05, "loss": 0.3858, "step": 244 }, { "epoch": 0.08, "learning_rate": 4.99996103678365e-05, "loss": 0.5104, "step": 245 }, { "epoch": 0.08, "learning_rate": 4.9999528545339345e-05, "loss": 0.3792, "step": 246 }, { "epoch": 0.08, "learning_rate": 4.999943893032583e-05, "loss": 0.412, "step": 247 }, { "epoch": 0.08, "learning_rate": 4.99993415228239e-05, "loss": 0.4597, "step": 248 }, { "epoch": 0.08, "learning_rate": 4.999923632286389e-05, "loss": 0.413, "step": 249 }, { "epoch": 0.09, "learning_rate": 4.9999123330478636e-05, "loss": 0.4112, "step": 250 }, { "epoch": 0.09, "learning_rate": 4.9999002545703333e-05, "loss": 0.3509, "step": 251 }, { "epoch": 0.09, "learning_rate": 4.9998873968575636e-05, "loss": 0.4245, "step": 252 }, { "epoch": 0.09, "learning_rate": 4.999873759913561e-05, "loss": 0.3676, "step": 253 }, { "epoch": 0.09, "learning_rate": 4.999859343742578e-05, "loss": 0.4154, "step": 254 }, { "epoch": 0.09, "learning_rate": 4.999844148349108e-05, "loss": 0.3947, "step": 255 }, { "epoch": 0.09, "learning_rate": 4.999828173737886e-05, "loss": 0.3872, "step": 256 }, { "epoch": 0.09, "learning_rate": 4.9998114199138926e-05, "loss": 0.3985, "step": 257 }, { "epoch": 0.09, "learning_rate": 4.99979388688235e-05, "loss": 0.4101, "step": 258 }, { "epoch": 0.09, "learning_rate": 4.999775574648724e-05, "loss": 0.4349, "step": 259 }, { "epoch": 0.09, "learning_rate": 4.9997564832187216e-05, "loss": 0.3921, "step": 260 }, { "epoch": 0.09, "learning_rate": 4.9997366125982934e-05, "loss": 0.4459, "step": 261 }, { "epoch": 0.09, "learning_rate": 4.999715962793634e-05, "loss": 0.4204, "step": 262 }, { "epoch": 0.09, "learning_rate": 4.999694533811181e-05, "loss": 0.4179, "step": 263 }, { "epoch": 0.09, "learning_rate": 4.999672325657612e-05, "loss": 0.4148, "step": 264 }, { "epoch": 0.09, "learning_rate": 4.99964933833985e-05, "loss": 0.463, "step": 265 }, { "epoch": 0.09, "learning_rate": 4.999625571865061e-05, "loss": 0.3729, "step": 266 }, { "epoch": 0.09, "learning_rate": 4.9996010262406526e-05, "loss": 0.4077, "step": 267 }, { "epoch": 0.09, "learning_rate": 4.999575701474276e-05, "loss": 0.3674, "step": 268 }, { "epoch": 0.09, "learning_rate": 4.999549597573825e-05, "loss": 0.4214, "step": 269 }, { "epoch": 0.09, "learning_rate": 4.9995227145474364e-05, "loss": 0.4176, "step": 270 }, { "epoch": 0.09, "learning_rate": 4.9994950524034905e-05, "loss": 0.3789, "step": 271 }, { "epoch": 0.09, "learning_rate": 4.999466611150609e-05, "loss": 0.3644, "step": 272 }, { "epoch": 0.09, "learning_rate": 4.9994373907976564e-05, "loss": 0.3681, "step": 273 }, { "epoch": 0.09, "learning_rate": 4.999407391353743e-05, "loss": 0.4424, "step": 274 }, { "epoch": 0.09, "learning_rate": 4.999376612828218e-05, "loss": 0.4185, "step": 275 }, { "epoch": 0.09, "learning_rate": 4.999345055230677e-05, "loss": 0.3608, "step": 276 }, { "epoch": 0.09, "learning_rate": 4.9993127185709536e-05, "loss": 0.4107, "step": 277 }, { "epoch": 0.09, "learning_rate": 4.99927960285913e-05, "loss": 0.4221, "step": 278 }, { "epoch": 0.1, "learning_rate": 4.999245708105529e-05, "loss": 0.4372, "step": 279 }, { "epoch": 0.1, "learning_rate": 4.9992110343207145e-05, "loss": 0.389, "step": 280 }, { "epoch": 0.1, "learning_rate": 4.999175581515494e-05, "loss": 0.3931, "step": 281 }, { "epoch": 0.1, "learning_rate": 4.99913934970092e-05, "loss": 0.3498, "step": 282 }, { "epoch": 0.1, "learning_rate": 4.9991023388882854e-05, "loss": 0.4115, "step": 283 }, { "epoch": 0.1, "learning_rate": 4.999064549089126e-05, "loss": 0.3949, "step": 284 }, { "epoch": 0.1, "learning_rate": 4.9990259803152225e-05, "loss": 0.382, "step": 285 }, { "epoch": 0.1, "learning_rate": 4.998986632578596e-05, "loss": 0.416, "step": 286 }, { "epoch": 0.1, "learning_rate": 4.9989465058915116e-05, "loss": 0.4084, "step": 287 }, { "epoch": 0.1, "learning_rate": 4.998905600266478e-05, "loss": 0.4012, "step": 288 }, { "epoch": 0.1, "learning_rate": 4.998863915716244e-05, "loss": 0.4464, "step": 289 }, { "epoch": 0.1, "learning_rate": 4.998821452253805e-05, "loss": 0.3788, "step": 290 }, { "epoch": 0.1, "learning_rate": 4.9987782098923955e-05, "loss": 0.4288, "step": 291 }, { "epoch": 0.1, "learning_rate": 4.9987341886454945e-05, "loss": 0.4141, "step": 292 }, { "epoch": 0.1, "learning_rate": 4.9986893885268247e-05, "loss": 0.4026, "step": 293 }, { "epoch": 0.1, "learning_rate": 4.9986438095503504e-05, "loss": 0.4128, "step": 294 }, { "epoch": 0.1, "learning_rate": 4.998597451730279e-05, "loss": 0.3682, "step": 295 }, { "epoch": 0.1, "learning_rate": 4.998550315081059e-05, "loss": 0.3991, "step": 296 }, { "epoch": 0.1, "learning_rate": 4.998502399617385e-05, "loss": 0.4281, "step": 297 }, { "epoch": 0.1, "learning_rate": 4.998453705354192e-05, "loss": 0.429, "step": 298 }, { "epoch": 0.1, "learning_rate": 4.998404232306658e-05, "loss": 0.4002, "step": 299 }, { "epoch": 0.1, "learning_rate": 4.9983539804902044e-05, "loss": 0.4029, "step": 300 }, { "epoch": 0.1, "learning_rate": 4.998302949920495e-05, "loss": 0.4472, "step": 301 }, { "epoch": 0.1, "learning_rate": 4.9982511406134356e-05, "loss": 0.3662, "step": 302 }, { "epoch": 0.1, "learning_rate": 4.9981985525851765e-05, "loss": 0.3915, "step": 303 }, { "epoch": 0.1, "learning_rate": 4.9981451858521094e-05, "loss": 0.3335, "step": 304 }, { "epoch": 0.1, "learning_rate": 4.998091040430869e-05, "loss": 0.4067, "step": 305 }, { "epoch": 0.1, "learning_rate": 4.998036116338333e-05, "loss": 0.3585, "step": 306 }, { "epoch": 0.1, "learning_rate": 4.997980413591622e-05, "loss": 0.3905, "step": 307 }, { "epoch": 0.11, "learning_rate": 4.9979239322080976e-05, "loss": 0.4419, "step": 308 }, { "epoch": 0.11, "learning_rate": 4.997866672205366e-05, "loss": 0.4358, "step": 309 }, { "epoch": 0.11, "learning_rate": 4.9978086336012756e-05, "loss": 0.3869, "step": 310 }, { "epoch": 0.11, "learning_rate": 4.997749816413918e-05, "loss": 0.3583, "step": 311 }, { "epoch": 0.11, "learning_rate": 4.9976902206616264e-05, "loss": 0.4075, "step": 312 }, { "epoch": 0.11, "learning_rate": 4.9976298463629764e-05, "loss": 0.3696, "step": 313 }, { "epoch": 0.11, "learning_rate": 4.997568693536789e-05, "loss": 0.402, "step": 314 }, { "epoch": 0.11, "learning_rate": 4.9975067622021245e-05, "loss": 0.3933, "step": 315 }, { "epoch": 0.11, "learning_rate": 4.997444052378287e-05, "loss": 0.3914, "step": 316 }, { "epoch": 0.11, "learning_rate": 4.997380564084824e-05, "loss": 0.3559, "step": 317 }, { "epoch": 0.11, "learning_rate": 4.997316297341526e-05, "loss": 0.4058, "step": 318 }, { "epoch": 0.11, "learning_rate": 4.997251252168425e-05, "loss": 0.3793, "step": 319 }, { "epoch": 0.11, "learning_rate": 4.997185428585794e-05, "loss": 0.3783, "step": 320 }, { "epoch": 0.11, "learning_rate": 4.997118826614155e-05, "loss": 0.4229, "step": 321 }, { "epoch": 0.11, "learning_rate": 4.9970514462742634e-05, "loss": 0.3879, "step": 322 }, { "epoch": 0.11, "learning_rate": 4.9969832875871246e-05, "loss": 0.3654, "step": 323 }, { "epoch": 0.11, "learning_rate": 4.9969143505739846e-05, "loss": 0.3828, "step": 324 }, { "epoch": 0.11, "learning_rate": 4.99684463525633e-05, "loss": 0.3588, "step": 325 }, { "epoch": 0.11, "learning_rate": 4.9967741416558936e-05, "loss": 0.4131, "step": 326 }, { "epoch": 0.11, "learning_rate": 4.996702869794646e-05, "loss": 0.4336, "step": 327 }, { "epoch": 0.11, "learning_rate": 4.9966308196948054e-05, "loss": 0.4052, "step": 328 }, { "epoch": 0.11, "learning_rate": 4.996557991378828e-05, "loss": 0.4317, "step": 329 }, { "epoch": 0.11, "learning_rate": 4.996484384869418e-05, "loss": 0.3655, "step": 330 }, { "epoch": 0.11, "learning_rate": 4.996410000189516e-05, "loss": 0.4058, "step": 331 }, { "epoch": 0.11, "learning_rate": 4.9963348373623104e-05, "loss": 0.4011, "step": 332 }, { "epoch": 0.11, "learning_rate": 4.996258896411229e-05, "loss": 0.3735, "step": 333 }, { "epoch": 0.11, "learning_rate": 4.996182177359942e-05, "loss": 0.3731, "step": 334 }, { "epoch": 0.11, "learning_rate": 4.996104680232365e-05, "loss": 0.4436, "step": 335 }, { "epoch": 0.11, "learning_rate": 4.996026405052654e-05, "loss": 0.3761, "step": 336 }, { "epoch": 0.11, "learning_rate": 4.995947351845207e-05, "loss": 0.3377, "step": 337 }, { "epoch": 0.12, "learning_rate": 4.9958675206346663e-05, "loss": 0.3533, "step": 338 }, { "epoch": 0.12, "learning_rate": 4.9957869114459156e-05, "loss": 0.3534, "step": 339 }, { "epoch": 0.12, "learning_rate": 4.9957055243040803e-05, "loss": 0.4177, "step": 340 }, { "epoch": 0.12, "learning_rate": 4.995623359234531e-05, "loss": 0.4364, "step": 341 }, { "epoch": 0.12, "learning_rate": 4.9955404162628785e-05, "loss": 0.3784, "step": 342 }, { "epoch": 0.12, "learning_rate": 4.9954566954149755e-05, "loss": 0.4643, "step": 343 }, { "epoch": 0.12, "learning_rate": 4.9953721967169195e-05, "loss": 0.3432, "step": 344 }, { "epoch": 0.12, "learning_rate": 4.99528692019505e-05, "loss": 0.3372, "step": 345 }, { "epoch": 0.12, "learning_rate": 4.9952008658759465e-05, "loss": 0.4094, "step": 346 }, { "epoch": 0.12, "learning_rate": 4.995114033786433e-05, "loss": 0.3813, "step": 347 }, { "epoch": 0.12, "learning_rate": 4.995026423953577e-05, "loss": 0.3743, "step": 348 }, { "epoch": 0.12, "learning_rate": 4.9949380364046865e-05, "loss": 0.3592, "step": 349 }, { "epoch": 0.12, "learning_rate": 4.9948488711673117e-05, "loss": 0.3958, "step": 350 }, { "epoch": 0.12, "learning_rate": 4.994758928269246e-05, "loss": 0.3748, "step": 351 }, { "epoch": 0.12, "learning_rate": 4.994668207738527e-05, "loss": 0.3795, "step": 352 }, { "epoch": 0.12, "learning_rate": 4.994576709603431e-05, "loss": 0.3871, "step": 353 }, { "epoch": 0.12, "learning_rate": 4.994484433892479e-05, "loss": 0.3671, "step": 354 }, { "epoch": 0.12, "learning_rate": 4.9943913806344334e-05, "loss": 0.3587, "step": 355 }, { "epoch": 0.12, "learning_rate": 4.9942975498583004e-05, "loss": 0.3794, "step": 356 }, { "epoch": 0.12, "learning_rate": 4.9942029415933276e-05, "loss": 0.3394, "step": 357 }, { "epoch": 0.12, "learning_rate": 4.994107555869005e-05, "loss": 0.369, "step": 358 }, { "epoch": 0.12, "learning_rate": 4.994011392715065e-05, "loss": 0.3716, "step": 359 }, { "epoch": 0.12, "learning_rate": 4.993914452161482e-05, "loss": 0.4321, "step": 360 }, { "epoch": 0.12, "learning_rate": 4.993816734238472e-05, "loss": 0.4041, "step": 361 }, { "epoch": 0.12, "learning_rate": 4.9937182389764966e-05, "loss": 0.3966, "step": 362 }, { "epoch": 0.12, "learning_rate": 4.993618966406256e-05, "loss": 0.3735, "step": 363 }, { "epoch": 0.12, "learning_rate": 4.9935189165586944e-05, "loss": 0.3554, "step": 364 }, { "epoch": 0.12, "learning_rate": 4.9934180894649974e-05, "loss": 0.4128, "step": 365 }, { "epoch": 0.12, "learning_rate": 4.993316485156594e-05, "loss": 0.3662, "step": 366 }, { "epoch": 0.13, "learning_rate": 4.993214103665156e-05, "loss": 0.3913, "step": 367 }, { "epoch": 0.13, "learning_rate": 4.993110945022594e-05, "loss": 0.3825, "step": 368 }, { "epoch": 0.13, "learning_rate": 4.993007009261065e-05, "loss": 0.3352, "step": 369 }, { "epoch": 0.13, "learning_rate": 4.992902296412966e-05, "loss": 0.4566, "step": 370 }, { "epoch": 0.13, "learning_rate": 4.992796806510936e-05, "loss": 0.4114, "step": 371 }, { "epoch": 0.13, "learning_rate": 4.9926905395878586e-05, "loss": 0.4181, "step": 372 }, { "epoch": 0.13, "learning_rate": 4.992583495676856e-05, "loss": 0.4337, "step": 373 }, { "epoch": 0.13, "learning_rate": 4.9924756748112956e-05, "loss": 0.4143, "step": 374 }, { "epoch": 0.13, "learning_rate": 4.9923670770247854e-05, "loss": 0.3993, "step": 375 }, { "epoch": 0.13, "learning_rate": 4.9922577023511763e-05, "loss": 0.3964, "step": 376 }, { "epoch": 0.13, "learning_rate": 4.992147550824561e-05, "loss": 0.3912, "step": 377 }, { "epoch": 0.13, "learning_rate": 4.992036622479275e-05, "loss": 0.3543, "step": 378 }, { "epoch": 0.13, "learning_rate": 4.9919249173498947e-05, "loss": 0.4007, "step": 379 }, { "epoch": 0.13, "learning_rate": 4.991812435471239e-05, "loss": 0.3811, "step": 380 }, { "epoch": 0.13, "learning_rate": 4.99169917687837e-05, "loss": 0.3924, "step": 381 }, { "epoch": 0.13, "learning_rate": 4.9915851416065904e-05, "loss": 0.3991, "step": 382 }, { "epoch": 0.13, "learning_rate": 4.9914703296914464e-05, "loss": 0.4415, "step": 383 }, { "epoch": 0.13, "learning_rate": 4.9913547411687256e-05, "loss": 0.4219, "step": 384 }, { "epoch": 0.13, "learning_rate": 4.9912383760744576e-05, "loss": 0.3864, "step": 385 }, { "epoch": 0.13, "learning_rate": 4.991121234444914e-05, "loss": 0.3748, "step": 386 }, { "epoch": 0.13, "learning_rate": 4.991003316316609e-05, "loss": 0.3996, "step": 387 }, { "epoch": 0.13, "learning_rate": 4.990884621726297e-05, "loss": 0.4588, "step": 388 }, { "epoch": 0.13, "learning_rate": 4.9907651507109785e-05, "loss": 0.3765, "step": 389 }, { "epoch": 0.13, "learning_rate": 4.9906449033078915e-05, "loss": 0.3757, "step": 390 }, { "epoch": 0.13, "learning_rate": 4.990523879554517e-05, "loss": 0.4079, "step": 391 }, { "epoch": 0.13, "learning_rate": 4.990402079488582e-05, "loss": 0.3938, "step": 392 }, { "epoch": 0.13, "learning_rate": 4.99027950314805e-05, "loss": 0.3372, "step": 393 }, { "epoch": 0.13, "learning_rate": 4.9901561505711294e-05, "loss": 0.4265, "step": 394 }, { "epoch": 0.13, "learning_rate": 4.9900320217962694e-05, "loss": 0.3787, "step": 395 }, { "epoch": 0.14, "learning_rate": 4.9899071168621626e-05, "loss": 0.3681, "step": 396 }, { "epoch": 0.14, "learning_rate": 4.989781435807743e-05, "loss": 0.3762, "step": 397 }, { "epoch": 0.14, "learning_rate": 4.989654978672185e-05, "loss": 0.4188, "step": 398 }, { "epoch": 0.14, "learning_rate": 4.9895277454949074e-05, "loss": 0.3678, "step": 399 }, { "epoch": 0.14, "learning_rate": 4.9893997363155685e-05, "loss": 0.3314, "step": 400 }, { "epoch": 0.14, "learning_rate": 4.989270951174071e-05, "loss": 0.4553, "step": 401 }, { "epoch": 0.14, "learning_rate": 4.989141390110557e-05, "loss": 0.382, "step": 402 }, { "epoch": 0.14, "learning_rate": 4.989011053165411e-05, "loss": 0.3993, "step": 403 }, { "epoch": 0.14, "learning_rate": 4.9888799403792606e-05, "loss": 0.4016, "step": 404 }, { "epoch": 0.14, "learning_rate": 4.9887480517929746e-05, "loss": 0.3329, "step": 405 }, { "epoch": 0.14, "learning_rate": 4.988615387447664e-05, "loss": 0.3603, "step": 406 }, { "epoch": 0.14, "learning_rate": 4.98848194738468e-05, "loss": 0.4414, "step": 407 }, { "epoch": 0.14, "learning_rate": 4.988347731645617e-05, "loss": 0.393, "step": 408 }, { "epoch": 0.14, "learning_rate": 4.988212740272312e-05, "loss": 0.3875, "step": 409 }, { "epoch": 0.14, "learning_rate": 4.988076973306841e-05, "loss": 0.3937, "step": 410 }, { "epoch": 0.14, "learning_rate": 4.987940430791525e-05, "loss": 0.3987, "step": 411 }, { "epoch": 0.14, "learning_rate": 4.987803112768925e-05, "loss": 0.4119, "step": 412 }, { "epoch": 0.14, "learning_rate": 4.987665019281843e-05, "loss": 0.4284, "step": 413 }, { "epoch": 0.14, "learning_rate": 4.987526150373324e-05, "loss": 0.3574, "step": 414 }, { "epoch": 0.14, "learning_rate": 4.987386506086654e-05, "loss": 0.3979, "step": 415 }, { "epoch": 0.14, "learning_rate": 4.987246086465362e-05, "loss": 0.3805, "step": 416 }, { "epoch": 0.14, "learning_rate": 4.987104891553218e-05, "loss": 0.4038, "step": 417 }, { "epoch": 0.14, "learning_rate": 4.9869629213942315e-05, "loss": 0.3897, "step": 418 }, { "epoch": 0.14, "learning_rate": 4.986820176032657e-05, "loss": 0.3942, "step": 419 }, { "epoch": 0.14, "learning_rate": 4.98667665551299e-05, "loss": 0.399, "step": 420 }, { "epoch": 0.14, "learning_rate": 4.986532359879964e-05, "loss": 0.3391, "step": 421 }, { "epoch": 0.14, "learning_rate": 4.9863872891785586e-05, "loss": 0.4044, "step": 422 }, { "epoch": 0.14, "learning_rate": 4.9862414434539934e-05, "loss": 0.3521, "step": 423 }, { "epoch": 0.14, "learning_rate": 4.986094822751729e-05, "loss": 0.3755, "step": 424 }, { "epoch": 0.15, "learning_rate": 4.9859474271174686e-05, "loss": 0.4062, "step": 425 }, { "epoch": 0.15, "learning_rate": 4.985799256597155e-05, "loss": 0.3783, "step": 426 }, { "epoch": 0.15, "learning_rate": 4.985650311236977e-05, "loss": 0.3782, "step": 427 }, { "epoch": 0.15, "learning_rate": 4.9855005910833586e-05, "loss": 0.3578, "step": 428 }, { "epoch": 0.15, "learning_rate": 4.9853500961829696e-05, "loss": 0.3793, "step": 429 }, { "epoch": 0.15, "learning_rate": 4.985198826582721e-05, "loss": 0.3509, "step": 430 }, { "epoch": 0.15, "learning_rate": 4.985046782329763e-05, "loss": 0.3299, "step": 431 }, { "epoch": 0.15, "learning_rate": 4.9848939634714906e-05, "loss": 0.3669, "step": 432 }, { "epoch": 0.15, "learning_rate": 4.9847403700555365e-05, "loss": 0.411, "step": 433 }, { "epoch": 0.15, "learning_rate": 4.984586002129778e-05, "loss": 0.4475, "step": 434 }, { "epoch": 0.15, "learning_rate": 4.984430859742333e-05, "loss": 0.3813, "step": 435 }, { "epoch": 0.15, "learning_rate": 4.98427494294156e-05, "loss": 0.3752, "step": 436 }, { "epoch": 0.15, "learning_rate": 4.984118251776059e-05, "loss": 0.3297, "step": 437 }, { "epoch": 0.15, "learning_rate": 4.983960786294671e-05, "loss": 0.3372, "step": 438 }, { "epoch": 0.15, "learning_rate": 4.98380254654648e-05, "loss": 0.3675, "step": 439 }, { "epoch": 0.15, "learning_rate": 4.9836435325808105e-05, "loss": 0.3573, "step": 440 }, { "epoch": 0.15, "learning_rate": 4.983483744447227e-05, "loss": 0.3827, "step": 441 }, { "epoch": 0.15, "learning_rate": 4.9833231821955386e-05, "loss": 0.3423, "step": 442 }, { "epoch": 0.15, "learning_rate": 4.983161845875791e-05, "loss": 0.3774, "step": 443 }, { "epoch": 0.15, "learning_rate": 4.9829997355382764e-05, "loss": 0.3463, "step": 444 }, { "epoch": 0.15, "learning_rate": 4.9828368512335235e-05, "loss": 0.3846, "step": 445 }, { "epoch": 0.15, "learning_rate": 4.982673193012305e-05, "loss": 0.3923, "step": 446 }, { "epoch": 0.15, "learning_rate": 4.982508760925635e-05, "loss": 0.3957, "step": 447 }, { "epoch": 0.15, "learning_rate": 4.982343555024767e-05, "loss": 0.3925, "step": 448 }, { "epoch": 0.15, "learning_rate": 4.982177575361198e-05, "loss": 0.4243, "step": 449 }, { "epoch": 0.15, "learning_rate": 4.982010821986664e-05, "loss": 0.361, "step": 450 }, { "epoch": 0.15, "learning_rate": 4.981843294953143e-05, "loss": 0.394, "step": 451 }, { "epoch": 0.15, "learning_rate": 4.981674994312855e-05, "loss": 0.3591, "step": 452 }, { "epoch": 0.15, "learning_rate": 4.98150592011826e-05, "loss": 0.3955, "step": 453 }, { "epoch": 0.15, "learning_rate": 4.9813360724220594e-05, "loss": 0.3892, "step": 454 }, { "epoch": 0.16, "learning_rate": 4.9811654512771964e-05, "loss": 0.3604, "step": 455 }, { "epoch": 0.16, "learning_rate": 4.980994056736854e-05, "loss": 0.3896, "step": 456 }, { "epoch": 0.16, "learning_rate": 4.980821888854458e-05, "loss": 0.3824, "step": 457 }, { "epoch": 0.16, "learning_rate": 4.9806489476836725e-05, "loss": 0.4012, "step": 458 }, { "epoch": 0.16, "learning_rate": 4.980475233278406e-05, "loss": 0.369, "step": 459 }, { "epoch": 0.16, "learning_rate": 4.980300745692806e-05, "loss": 0.4059, "step": 460 }, { "epoch": 0.16, "learning_rate": 4.9801254849812606e-05, "loss": 0.4166, "step": 461 }, { "epoch": 0.16, "learning_rate": 4.979949451198401e-05, "loss": 0.3894, "step": 462 }, { "epoch": 0.16, "learning_rate": 4.979772644399098e-05, "loss": 0.3954, "step": 463 }, { "epoch": 0.16, "learning_rate": 4.9795950646384624e-05, "loss": 0.334, "step": 464 }, { "epoch": 0.16, "learning_rate": 4.979416711971848e-05, "loss": 0.4614, "step": 465 }, { "epoch": 0.16, "learning_rate": 4.979237586454847e-05, "loss": 0.3654, "step": 466 }, { "epoch": 0.16, "learning_rate": 4.979057688143296e-05, "loss": 0.4303, "step": 467 }, { "epoch": 0.16, "learning_rate": 4.978877017093268e-05, "loss": 0.3988, "step": 468 }, { "epoch": 0.16, "learning_rate": 4.978695573361083e-05, "loss": 0.4, "step": 469 }, { "epoch": 0.16, "learning_rate": 4.978513357003294e-05, "loss": 0.4016, "step": 470 }, { "epoch": 0.16, "learning_rate": 4.9783303680767024e-05, "loss": 0.4093, "step": 471 }, { "epoch": 0.16, "learning_rate": 4.9781466066383454e-05, "loss": 0.3822, "step": 472 }, { "epoch": 0.16, "learning_rate": 4.9779620727455024e-05, "loss": 0.3547, "step": 473 }, { "epoch": 0.16, "learning_rate": 4.977776766455695e-05, "loss": 0.3476, "step": 474 }, { "epoch": 0.16, "learning_rate": 4.977590687826683e-05, "loss": 0.3939, "step": 475 }, { "epoch": 0.16, "learning_rate": 4.97740383691647e-05, "loss": 0.5032, "step": 476 }, { "epoch": 0.16, "learning_rate": 4.977216213783297e-05, "loss": 0.3959, "step": 477 }, { "epoch": 0.16, "learning_rate": 4.977027818485648e-05, "loss": 0.3767, "step": 478 }, { "epoch": 0.16, "learning_rate": 4.9768386510822475e-05, "loss": 0.4035, "step": 479 }, { "epoch": 0.16, "learning_rate": 4.976648711632059e-05, "loss": 0.3627, "step": 480 }, { "epoch": 0.16, "learning_rate": 4.97645800019429e-05, "loss": 0.3727, "step": 481 }, { "epoch": 0.16, "learning_rate": 4.976266516828384e-05, "loss": 0.4056, "step": 482 }, { "epoch": 0.16, "learning_rate": 4.976074261594029e-05, "loss": 0.4138, "step": 483 }, { "epoch": 0.17, "learning_rate": 4.975881234551153e-05, "loss": 0.3884, "step": 484 }, { "epoch": 0.17, "learning_rate": 4.975687435759922e-05, "loss": 0.4172, "step": 485 }, { "epoch": 0.17, "learning_rate": 4.9754928652807456e-05, "loss": 0.4045, "step": 486 }, { "epoch": 0.17, "learning_rate": 4.975297523174271e-05, "loss": 0.3713, "step": 487 }, { "epoch": 0.17, "learning_rate": 4.9751014095013905e-05, "loss": 0.3751, "step": 488 }, { "epoch": 0.17, "learning_rate": 4.974904524323232e-05, "loss": 0.4276, "step": 489 }, { "epoch": 0.17, "learning_rate": 4.974706867701165e-05, "loss": 0.4027, "step": 490 }, { "epoch": 0.17, "learning_rate": 4.974508439696803e-05, "loss": 0.3919, "step": 491 }, { "epoch": 0.17, "learning_rate": 4.974309240371996e-05, "loss": 0.3336, "step": 492 }, { "epoch": 0.17, "learning_rate": 4.974109269788835e-05, "loss": 0.3937, "step": 493 }, { "epoch": 0.17, "learning_rate": 4.973908528009653e-05, "loss": 0.4244, "step": 494 }, { "epoch": 0.17, "learning_rate": 4.973707015097022e-05, "loss": 0.3609, "step": 495 }, { "epoch": 0.17, "learning_rate": 4.9735047311137556e-05, "loss": 0.3636, "step": 496 }, { "epoch": 0.17, "learning_rate": 4.973301676122907e-05, "loss": 0.4057, "step": 497 }, { "epoch": 0.17, "learning_rate": 4.973097850187768e-05, "loss": 0.4317, "step": 498 }, { "epoch": 0.17, "learning_rate": 4.972893253371875e-05, "loss": 0.3745, "step": 499 }, { "epoch": 0.17, "learning_rate": 4.972687885739001e-05, "loss": 0.3917, "step": 500 }, { "epoch": 0.17, "learning_rate": 4.97248174735316e-05, "loss": 0.3655, "step": 501 }, { "epoch": 0.17, "learning_rate": 4.9722748382786077e-05, "loss": 0.4415, "step": 502 }, { "epoch": 0.17, "learning_rate": 4.9720671585798385e-05, "loss": 0.3916, "step": 503 }, { "epoch": 0.17, "learning_rate": 4.971858708321587e-05, "loss": 0.358, "step": 504 }, { "epoch": 0.17, "learning_rate": 4.97164948756883e-05, "loss": 0.3632, "step": 505 }, { "epoch": 0.17, "learning_rate": 4.971439496386782e-05, "loss": 0.3323, "step": 506 }, { "epoch": 0.17, "learning_rate": 4.971228734840897e-05, "loss": 0.3689, "step": 507 }, { "epoch": 0.17, "learning_rate": 4.971017202996874e-05, "loss": 0.3824, "step": 508 }, { "epoch": 0.17, "learning_rate": 4.970804900920646e-05, "loss": 0.3987, "step": 509 }, { "epoch": 0.17, "learning_rate": 4.9705918286783906e-05, "loss": 0.3871, "step": 510 }, { "epoch": 0.17, "learning_rate": 4.970377986336524e-05, "loss": 0.3855, "step": 511 }, { "epoch": 0.17, "learning_rate": 4.9701633739617e-05, "loss": 0.3836, "step": 512 }, { "epoch": 0.18, "learning_rate": 4.969947991620818e-05, "loss": 0.3967, "step": 513 }, { "epoch": 0.18, "learning_rate": 4.969731839381012e-05, "loss": 0.4485, "step": 514 }, { "epoch": 0.18, "learning_rate": 4.9695149173096575e-05, "loss": 0.3904, "step": 515 }, { "epoch": 0.18, "learning_rate": 4.969297225474373e-05, "loss": 0.3631, "step": 516 }, { "epoch": 0.18, "learning_rate": 4.969078763943012e-05, "loss": 0.3748, "step": 517 }, { "epoch": 0.18, "learning_rate": 4.968859532783672e-05, "loss": 0.3873, "step": 518 }, { "epoch": 0.18, "learning_rate": 4.9686395320646876e-05, "loss": 0.4058, "step": 519 }, { "epoch": 0.18, "learning_rate": 4.9684187618546344e-05, "loss": 0.3774, "step": 520 }, { "epoch": 0.18, "learning_rate": 4.9681972222223295e-05, "loss": 0.4021, "step": 521 }, { "epoch": 0.18, "learning_rate": 4.967974913236826e-05, "loss": 0.3549, "step": 522 }, { "epoch": 0.18, "learning_rate": 4.967751834967421e-05, "loss": 0.377, "step": 523 }, { "epoch": 0.18, "learning_rate": 4.9675279874836497e-05, "loss": 0.3911, "step": 524 }, { "epoch": 0.18, "learning_rate": 4.967303370855285e-05, "loss": 0.3561, "step": 525 }, { "epoch": 0.18, "learning_rate": 4.9670779851523416e-05, "loss": 0.3493, "step": 526 }, { "epoch": 0.18, "learning_rate": 4.9668518304450756e-05, "loss": 0.3994, "step": 527 }, { "epoch": 0.18, "learning_rate": 4.9666249068039795e-05, "loss": 0.3487, "step": 528 }, { "epoch": 0.18, "learning_rate": 4.966397214299786e-05, "loss": 0.3662, "step": 529 }, { "epoch": 0.18, "learning_rate": 4.9661687530034706e-05, "loss": 0.3485, "step": 530 }, { "epoch": 0.18, "learning_rate": 4.9659395229862435e-05, "loss": 0.4107, "step": 531 }, { "epoch": 0.18, "learning_rate": 4.96570952431956e-05, "loss": 0.3997, "step": 532 }, { "epoch": 0.18, "learning_rate": 4.9654787570751104e-05, "loss": 0.3654, "step": 533 }, { "epoch": 0.18, "learning_rate": 4.965247221324827e-05, "loss": 0.3678, "step": 534 }, { "epoch": 0.18, "learning_rate": 4.96501491714088e-05, "loss": 0.4096, "step": 535 }, { "epoch": 0.18, "learning_rate": 4.9647818445956814e-05, "loss": 0.3694, "step": 536 }, { "epoch": 0.18, "learning_rate": 4.9645480037618816e-05, "loss": 0.4129, "step": 537 }, { "epoch": 0.18, "learning_rate": 4.964313394712369e-05, "loss": 0.3938, "step": 538 }, { "epoch": 0.18, "learning_rate": 4.9640780175202734e-05, "loss": 0.4155, "step": 539 }, { "epoch": 0.18, "learning_rate": 4.9638418722589636e-05, "loss": 0.395, "step": 540 }, { "epoch": 0.18, "learning_rate": 4.9636049590020475e-05, "loss": 0.3851, "step": 541 }, { "epoch": 0.18, "learning_rate": 4.963367277823373e-05, "loss": 0.3599, "step": 542 }, { "epoch": 0.19, "learning_rate": 4.963128828797026e-05, "loss": 0.4015, "step": 543 }, { "epoch": 0.19, "learning_rate": 4.9628896119973324e-05, "loss": 0.3984, "step": 544 }, { "epoch": 0.19, "learning_rate": 4.962649627498859e-05, "loss": 0.3386, "step": 545 }, { "epoch": 0.19, "learning_rate": 4.962408875376409e-05, "loss": 0.3961, "step": 546 }, { "epoch": 0.19, "learning_rate": 4.9621673557050275e-05, "loss": 0.3297, "step": 547 }, { "epoch": 0.19, "learning_rate": 4.961925068559997e-05, "loss": 0.4011, "step": 548 }, { "epoch": 0.19, "learning_rate": 4.961682014016841e-05, "loss": 0.3785, "step": 549 }, { "epoch": 0.19, "learning_rate": 4.961438192151321e-05, "loss": 0.3827, "step": 550 }, { "epoch": 0.19, "learning_rate": 4.9611936030394365e-05, "loss": 0.4081, "step": 551 }, { "epoch": 0.19, "learning_rate": 4.9609482467574293e-05, "loss": 0.4064, "step": 552 }, { "epoch": 0.19, "learning_rate": 4.960702123381777e-05, "loss": 0.3486, "step": 553 }, { "epoch": 0.19, "learning_rate": 4.9604552329891996e-05, "loss": 0.3769, "step": 554 }, { "epoch": 0.19, "learning_rate": 4.9602075756566524e-05, "loss": 0.355, "step": 555 }, { "epoch": 0.19, "learning_rate": 4.9599591514613336e-05, "loss": 0.4216, "step": 556 }, { "epoch": 0.19, "learning_rate": 4.9597099604806777e-05, "loss": 0.3567, "step": 557 }, { "epoch": 0.19, "learning_rate": 4.9594600027923594e-05, "loss": 0.3474, "step": 558 }, { "epoch": 0.19, "learning_rate": 4.9592092784742924e-05, "loss": 0.3664, "step": 559 }, { "epoch": 0.19, "learning_rate": 4.958957787604628e-05, "loss": 0.3708, "step": 560 }, { "epoch": 0.19, "learning_rate": 4.958705530261759e-05, "loss": 0.358, "step": 561 }, { "epoch": 0.19, "learning_rate": 4.958452506524315e-05, "loss": 0.3594, "step": 562 }, { "epoch": 0.19, "learning_rate": 4.9581987164711654e-05, "loss": 0.3695, "step": 563 }, { "epoch": 0.19, "learning_rate": 4.957944160181417e-05, "loss": 0.4056, "step": 564 }, { "epoch": 0.19, "learning_rate": 4.957688837734419e-05, "loss": 0.4069, "step": 565 }, { "epoch": 0.19, "learning_rate": 4.957432749209755e-05, "loss": 0.3838, "step": 566 }, { "epoch": 0.19, "learning_rate": 4.95717589468725e-05, "loss": 0.3876, "step": 567 }, { "epoch": 0.19, "learning_rate": 4.9569182742469675e-05, "loss": 0.4115, "step": 568 }, { "epoch": 0.19, "learning_rate": 4.9566598879692093e-05, "loss": 0.3657, "step": 569 }, { "epoch": 0.19, "learning_rate": 4.9564007359345164e-05, "loss": 0.3544, "step": 570 }, { "epoch": 0.19, "learning_rate": 4.956140818223669e-05, "loss": 0.3775, "step": 571 }, { "epoch": 0.2, "learning_rate": 4.955880134917683e-05, "loss": 0.3944, "step": 572 }, { "epoch": 0.2, "learning_rate": 4.955618686097816e-05, "loss": 0.3606, "step": 573 }, { "epoch": 0.2, "learning_rate": 4.9553564718455644e-05, "loss": 0.3286, "step": 574 }, { "epoch": 0.2, "learning_rate": 4.9550934922426604e-05, "loss": 0.3551, "step": 575 }, { "epoch": 0.2, "learning_rate": 4.954829747371078e-05, "loss": 0.4077, "step": 576 }, { "epoch": 0.2, "learning_rate": 4.954565237313027e-05, "loss": 0.3858, "step": 577 }, { "epoch": 0.2, "learning_rate": 4.954299962150958e-05, "loss": 0.4152, "step": 578 }, { "epoch": 0.2, "learning_rate": 4.954033921967557e-05, "loss": 0.3483, "step": 579 }, { "epoch": 0.2, "learning_rate": 4.953767116845753e-05, "loss": 0.346, "step": 580 }, { "epoch": 0.2, "learning_rate": 4.953499546868709e-05, "loss": 0.3716, "step": 581 }, { "epoch": 0.2, "learning_rate": 4.9532312121198305e-05, "loss": 0.3548, "step": 582 }, { "epoch": 0.2, "learning_rate": 4.952962112682756e-05, "loss": 0.3629, "step": 583 }, { "epoch": 0.2, "learning_rate": 4.9526922486413686e-05, "loss": 0.3547, "step": 584 }, { "epoch": 0.2, "learning_rate": 4.9524216200797854e-05, "loss": 0.4244, "step": 585 }, { "epoch": 0.2, "learning_rate": 4.952150227082363e-05, "loss": 0.3777, "step": 586 }, { "epoch": 0.2, "learning_rate": 4.951878069733695e-05, "loss": 0.3613, "step": 587 }, { "epoch": 0.2, "learning_rate": 4.9516051481186174e-05, "loss": 0.3475, "step": 588 }, { "epoch": 0.2, "learning_rate": 4.9513314623222005e-05, "loss": 0.4175, "step": 589 }, { "epoch": 0.2, "learning_rate": 4.951057012429753e-05, "loss": 0.3181, "step": 590 }, { "epoch": 0.2, "learning_rate": 4.950781798526824e-05, "loss": 0.3633, "step": 591 }, { "epoch": 0.2, "learning_rate": 4.950505820699199e-05, "loss": 0.3479, "step": 592 }, { "epoch": 0.2, "learning_rate": 4.950229079032902e-05, "loss": 0.3652, "step": 593 }, { "epoch": 0.2, "learning_rate": 4.949951573614195e-05, "loss": 0.3398, "step": 594 }, { "epoch": 0.2, "learning_rate": 4.9496733045295785e-05, "loss": 0.3889, "step": 595 }, { "epoch": 0.2, "learning_rate": 4.9493942718657906e-05, "loss": 0.402, "step": 596 }, { "epoch": 0.2, "learning_rate": 4.9491144757098084e-05, "loss": 0.3798, "step": 597 }, { "epoch": 0.2, "learning_rate": 4.9488339161488456e-05, "loss": 0.4103, "step": 598 }, { "epoch": 0.2, "learning_rate": 4.948552593270354e-05, "loss": 0.3774, "step": 599 }, { "epoch": 0.2, "learning_rate": 4.948270507162025e-05, "loss": 0.3516, "step": 600 }, { "epoch": 0.21, "learning_rate": 4.9479876579117844e-05, "loss": 0.3636, "step": 601 }, { "epoch": 0.21, "learning_rate": 4.9477040456078e-05, "loss": 0.3594, "step": 602 }, { "epoch": 0.21, "learning_rate": 4.947419670338477e-05, "loss": 0.3563, "step": 603 }, { "epoch": 0.21, "learning_rate": 4.947134532192454e-05, "loss": 0.3645, "step": 604 }, { "epoch": 0.21, "learning_rate": 4.9468486312586116e-05, "loss": 0.3743, "step": 605 }, { "epoch": 0.21, "learning_rate": 4.946561967626068e-05, "loss": 0.3667, "step": 606 }, { "epoch": 0.21, "learning_rate": 4.946274541384176e-05, "loss": 0.3402, "step": 607 }, { "epoch": 0.21, "learning_rate": 4.9459863526225306e-05, "loss": 0.3823, "step": 608 }, { "epoch": 0.21, "learning_rate": 4.94569740143096e-05, "loss": 0.3551, "step": 609 }, { "epoch": 0.21, "learning_rate": 4.9454076878995335e-05, "loss": 0.4547, "step": 610 }, { "epoch": 0.21, "learning_rate": 4.945117212118556e-05, "loss": 0.3499, "step": 611 }, { "epoch": 0.21, "learning_rate": 4.9448259741785715e-05, "loss": 0.3629, "step": 612 }, { "epoch": 0.21, "learning_rate": 4.9445339741703607e-05, "loss": 0.4158, "step": 613 }, { "epoch": 0.21, "learning_rate": 4.9442412121849415e-05, "loss": 0.3956, "step": 614 }, { "epoch": 0.21, "learning_rate": 4.9439476883135695e-05, "loss": 0.4046, "step": 615 }, { "epoch": 0.21, "learning_rate": 4.9436534026477376e-05, "loss": 0.3945, "step": 616 }, { "epoch": 0.21, "learning_rate": 4.943358355279178e-05, "loss": 0.3489, "step": 617 }, { "epoch": 0.21, "learning_rate": 4.943062546299858e-05, "loss": 0.4001, "step": 618 }, { "epoch": 0.21, "learning_rate": 4.942765975801983e-05, "loss": 0.3662, "step": 619 }, { "epoch": 0.21, "learning_rate": 4.942468643877996e-05, "loss": 0.3467, "step": 620 }, { "epoch": 0.21, "learning_rate": 4.942170550620579e-05, "loss": 0.378, "step": 621 }, { "epoch": 0.21, "learning_rate": 4.941871696122647e-05, "loss": 0.3878, "step": 622 }, { "epoch": 0.21, "learning_rate": 4.941572080477357e-05, "loss": 0.3219, "step": 623 }, { "epoch": 0.21, "learning_rate": 4.941271703778099e-05, "loss": 0.3944, "step": 624 }, { "epoch": 0.21, "learning_rate": 4.9409705661185044e-05, "loss": 0.371, "step": 625 }, { "epoch": 0.21, "learning_rate": 4.940668667592439e-05, "loss": 0.3981, "step": 626 }, { "epoch": 0.21, "learning_rate": 4.940366008294007e-05, "loss": 0.3622, "step": 627 }, { "epoch": 0.21, "learning_rate": 4.9400625883175475e-05, "loss": 0.3077, "step": 628 }, { "epoch": 0.21, "learning_rate": 4.9397584077576406e-05, "loss": 0.4211, "step": 629 }, { "epoch": 0.21, "learning_rate": 4.9394534667091005e-05, "loss": 0.3031, "step": 630 }, { "epoch": 0.22, "learning_rate": 4.9391477652669806e-05, "loss": 0.3831, "step": 631 }, { "epoch": 0.22, "learning_rate": 4.938841303526567e-05, "loss": 0.337, "step": 632 }, { "epoch": 0.22, "learning_rate": 4.9385340815833884e-05, "loss": 0.3786, "step": 633 }, { "epoch": 0.22, "learning_rate": 4.938226099533207e-05, "loss": 0.3738, "step": 634 }, { "epoch": 0.22, "learning_rate": 4.937917357472024e-05, "loss": 0.3846, "step": 635 }, { "epoch": 0.22, "learning_rate": 4.9376078554960743e-05, "loss": 0.3593, "step": 636 }, { "epoch": 0.22, "learning_rate": 4.937297593701833e-05, "loss": 0.3559, "step": 637 }, { "epoch": 0.22, "learning_rate": 4.93698657218601e-05, "loss": 0.3503, "step": 638 }, { "epoch": 0.22, "learning_rate": 4.936674791045553e-05, "loss": 0.4037, "step": 639 }, { "epoch": 0.22, "learning_rate": 4.936362250377647e-05, "loss": 0.3999, "step": 640 }, { "epoch": 0.22, "learning_rate": 4.936048950279712e-05, "loss": 0.3718, "step": 641 }, { "epoch": 0.22, "learning_rate": 4.935734890849406e-05, "loss": 0.3298, "step": 642 }, { "epoch": 0.22, "learning_rate": 4.935420072184623e-05, "loss": 0.3825, "step": 643 }, { "epoch": 0.22, "learning_rate": 4.935104494383495e-05, "loss": 0.2997, "step": 644 }, { "epoch": 0.22, "learning_rate": 4.934788157544389e-05, "loss": 0.3334, "step": 645 }, { "epoch": 0.22, "learning_rate": 4.93447106176591e-05, "loss": 0.4027, "step": 646 }, { "epoch": 0.22, "learning_rate": 4.934153207146896e-05, "loss": 0.3983, "step": 647 }, { "epoch": 0.22, "learning_rate": 4.9338345937864286e-05, "loss": 0.346, "step": 648 }, { "epoch": 0.22, "learning_rate": 4.9335152217838196e-05, "loss": 0.3757, "step": 649 }, { "epoch": 0.22, "learning_rate": 4.9331950912386185e-05, "loss": 0.3489, "step": 650 }, { "epoch": 0.22, "learning_rate": 4.932874202250613e-05, "loss": 0.3501, "step": 651 }, { "epoch": 0.22, "learning_rate": 4.932552554919826e-05, "loss": 0.343, "step": 652 }, { "epoch": 0.22, "learning_rate": 4.932230149346518e-05, "loss": 0.3962, "step": 653 }, { "epoch": 0.22, "learning_rate": 4.9319069856311836e-05, "loss": 0.3881, "step": 654 }, { "epoch": 0.22, "learning_rate": 4.931583063874556e-05, "loss": 0.3951, "step": 655 }, { "epoch": 0.22, "learning_rate": 4.931258384177603e-05, "loss": 0.3189, "step": 656 }, { "epoch": 0.22, "learning_rate": 4.93093294664153e-05, "loss": 0.3676, "step": 657 }, { "epoch": 0.22, "learning_rate": 4.9306067513677786e-05, "loss": 0.4031, "step": 658 }, { "epoch": 0.22, "learning_rate": 4.930279798458024e-05, "loss": 0.4024, "step": 659 }, { "epoch": 0.23, "learning_rate": 4.929952088014181e-05, "loss": 0.3883, "step": 660 }, { "epoch": 0.23, "learning_rate": 4.9296236201384e-05, "loss": 0.3496, "step": 661 }, { "epoch": 0.23, "learning_rate": 4.9292943949330635e-05, "loss": 0.3396, "step": 662 }, { "epoch": 0.23, "learning_rate": 4.928964412500796e-05, "loss": 0.4452, "step": 663 }, { "epoch": 0.23, "learning_rate": 4.9286336729444535e-05, "loss": 0.3818, "step": 664 }, { "epoch": 0.23, "learning_rate": 4.928302176367131e-05, "loss": 0.3485, "step": 665 }, { "epoch": 0.23, "learning_rate": 4.9279699228721575e-05, "loss": 0.3812, "step": 666 }, { "epoch": 0.23, "learning_rate": 4.927636912563099e-05, "loss": 0.3793, "step": 667 }, { "epoch": 0.23, "learning_rate": 4.927303145543756e-05, "loss": 0.4071, "step": 668 }, { "epoch": 0.23, "learning_rate": 4.926968621918166e-05, "loss": 0.3339, "step": 669 }, { "epoch": 0.23, "learning_rate": 4.926633341790604e-05, "loss": 0.3846, "step": 670 }, { "epoch": 0.23, "learning_rate": 4.926297305265577e-05, "loss": 0.3915, "step": 671 }, { "epoch": 0.23, "learning_rate": 4.925960512447831e-05, "loss": 0.3103, "step": 672 }, { "epoch": 0.23, "learning_rate": 4.925622963442346e-05, "loss": 0.4936, "step": 673 }, { "epoch": 0.23, "learning_rate": 4.925284658354337e-05, "loss": 0.3785, "step": 674 }, { "epoch": 0.23, "learning_rate": 4.924945597289258e-05, "loss": 0.3524, "step": 675 }, { "epoch": 0.23, "learning_rate": 4.924605780352796e-05, "loss": 0.3386, "step": 676 }, { "epoch": 0.23, "learning_rate": 4.9242652076508736e-05, "loss": 0.3945, "step": 677 }, { "epoch": 0.23, "learning_rate": 4.923923879289649e-05, "loss": 0.3744, "step": 678 }, { "epoch": 0.23, "learning_rate": 4.923581795375518e-05, "loss": 0.4101, "step": 679 }, { "epoch": 0.23, "learning_rate": 4.923238956015109e-05, "loss": 0.3886, "step": 680 }, { "epoch": 0.23, "learning_rate": 4.9228953613152884e-05, "loss": 0.3462, "step": 681 }, { "epoch": 0.23, "learning_rate": 4.9225510113831564e-05, "loss": 0.3565, "step": 682 }, { "epoch": 0.23, "learning_rate": 4.922205906326049e-05, "loss": 0.3661, "step": 683 }, { "epoch": 0.23, "learning_rate": 4.9218600462515385e-05, "loss": 0.4009, "step": 684 }, { "epoch": 0.23, "learning_rate": 4.921513431267431e-05, "loss": 0.3459, "step": 685 }, { "epoch": 0.23, "learning_rate": 4.921166061481768e-05, "loss": 0.319, "step": 686 }, { "epoch": 0.23, "learning_rate": 4.920817937002827e-05, "loss": 0.3282, "step": 687 }, { "epoch": 0.23, "learning_rate": 4.920469057939123e-05, "loss": 0.3524, "step": 688 }, { "epoch": 0.24, "learning_rate": 4.9201194243994016e-05, "loss": 0.4298, "step": 689 }, { "epoch": 0.24, "learning_rate": 4.9197690364926464e-05, "loss": 0.2674, "step": 690 }, { "epoch": 0.24, "learning_rate": 4.9194178943280756e-05, "loss": 0.3873, "step": 691 }, { "epoch": 0.24, "learning_rate": 4.919065998015142e-05, "loss": 0.408, "step": 692 }, { "epoch": 0.24, "learning_rate": 4.918713347663536e-05, "loss": 0.3405, "step": 693 }, { "epoch": 0.24, "learning_rate": 4.918359943383178e-05, "loss": 0.3958, "step": 694 }, { "epoch": 0.24, "learning_rate": 4.91800578528423e-05, "loss": 0.4128, "step": 695 }, { "epoch": 0.24, "learning_rate": 4.917650873477082e-05, "loss": 0.4097, "step": 696 }, { "epoch": 0.24, "learning_rate": 4.9172952080723636e-05, "loss": 0.3834, "step": 697 }, { "epoch": 0.24, "learning_rate": 4.916938789180939e-05, "loss": 0.353, "step": 698 }, { "epoch": 0.24, "learning_rate": 4.916581616913904e-05, "loss": 0.3846, "step": 699 }, { "epoch": 0.24, "learning_rate": 4.916223691382595e-05, "loss": 0.4598, "step": 700 }, { "epoch": 0.24, "learning_rate": 4.915865012698577e-05, "loss": 0.347, "step": 701 }, { "epoch": 0.24, "learning_rate": 4.9155055809736525e-05, "loss": 0.3945, "step": 702 }, { "epoch": 0.24, "learning_rate": 4.91514539631986e-05, "loss": 0.3722, "step": 703 }, { "epoch": 0.24, "learning_rate": 4.9147844588494705e-05, "loss": 0.3973, "step": 704 }, { "epoch": 0.24, "learning_rate": 4.914422768674991e-05, "loss": 0.3956, "step": 705 }, { "epoch": 0.24, "learning_rate": 4.9140603259091634e-05, "loss": 0.3346, "step": 706 }, { "epoch": 0.24, "learning_rate": 4.913697130664962e-05, "loss": 0.3781, "step": 707 }, { "epoch": 0.24, "learning_rate": 4.913333183055597e-05, "loss": 0.334, "step": 708 }, { "epoch": 0.24, "learning_rate": 4.912968483194516e-05, "loss": 0.3562, "step": 709 }, { "epoch": 0.24, "learning_rate": 4.912603031195394e-05, "loss": 0.3799, "step": 710 }, { "epoch": 0.24, "learning_rate": 4.912236827172148e-05, "loss": 0.3756, "step": 711 }, { "epoch": 0.24, "learning_rate": 4.911869871238926e-05, "loss": 0.3766, "step": 712 }, { "epoch": 0.24, "learning_rate": 4.91150216351011e-05, "loss": 0.3937, "step": 713 }, { "epoch": 0.24, "learning_rate": 4.9111337041003164e-05, "loss": 0.3901, "step": 714 }, { "epoch": 0.24, "learning_rate": 4.910764493124397e-05, "loss": 0.3636, "step": 715 }, { "epoch": 0.24, "learning_rate": 4.910394530697436e-05, "loss": 0.4095, "step": 716 }, { "epoch": 0.24, "learning_rate": 4.910023816934756e-05, "loss": 0.4292, "step": 717 }, { "epoch": 0.24, "learning_rate": 4.9096523519519076e-05, "loss": 0.3662, "step": 718 }, { "epoch": 0.25, "learning_rate": 4.909280135864681e-05, "loss": 0.3515, "step": 719 }, { "epoch": 0.25, "learning_rate": 4.908907168789097e-05, "loss": 0.4288, "step": 720 }, { "epoch": 0.25, "learning_rate": 4.908533450841414e-05, "loss": 0.3634, "step": 721 }, { "epoch": 0.25, "learning_rate": 4.90815898213812e-05, "loss": 0.3911, "step": 722 }, { "epoch": 0.25, "learning_rate": 4.90778376279594e-05, "loss": 0.3591, "step": 723 }, { "epoch": 0.25, "learning_rate": 4.907407792931833e-05, "loss": 0.4248, "step": 724 }, { "epoch": 0.25, "learning_rate": 4.907031072662991e-05, "loss": 0.3144, "step": 725 }, { "epoch": 0.25, "learning_rate": 4.90665360210684e-05, "loss": 0.4093, "step": 726 }, { "epoch": 0.25, "learning_rate": 4.9062753813810396e-05, "loss": 0.3416, "step": 727 }, { "epoch": 0.25, "learning_rate": 4.9058964106034844e-05, "loss": 0.4166, "step": 728 }, { "epoch": 0.25, "learning_rate": 4.9055166898923024e-05, "loss": 0.3576, "step": 729 }, { "epoch": 0.25, "learning_rate": 4.905136219365854e-05, "loss": 0.3873, "step": 730 }, { "epoch": 0.25, "learning_rate": 4.9047549991427347e-05, "loss": 0.3951, "step": 731 }, { "epoch": 0.25, "learning_rate": 4.904373029341773e-05, "loss": 0.3795, "step": 732 }, { "epoch": 0.25, "learning_rate": 4.903990310082032e-05, "loss": 0.3793, "step": 733 }, { "epoch": 0.25, "learning_rate": 4.903606841482808e-05, "loss": 0.4235, "step": 734 }, { "epoch": 0.25, "learning_rate": 4.90322262366363e-05, "loss": 0.3502, "step": 735 }, { "epoch": 0.25, "learning_rate": 4.902837656744261e-05, "loss": 0.3657, "step": 736 }, { "epoch": 0.25, "learning_rate": 4.902451940844699e-05, "loss": 0.3653, "step": 737 }, { "epoch": 0.25, "learning_rate": 4.9020654760851726e-05, "loss": 0.4012, "step": 738 }, { "epoch": 0.25, "learning_rate": 4.901678262586147e-05, "loss": 0.34, "step": 739 }, { "epoch": 0.25, "learning_rate": 4.901290300468317e-05, "loss": 0.3407, "step": 740 }, { "epoch": 0.25, "learning_rate": 4.900901589852615e-05, "loss": 0.3598, "step": 741 }, { "epoch": 0.25, "learning_rate": 4.900512130860204e-05, "loss": 0.3478, "step": 742 }, { "epoch": 0.25, "learning_rate": 4.9001219236124806e-05, "loss": 0.3805, "step": 743 }, { "epoch": 0.25, "learning_rate": 4.899730968231075e-05, "loss": 0.3347, "step": 744 }, { "epoch": 0.25, "learning_rate": 4.89933926483785e-05, "loss": 0.3942, "step": 745 }, { "epoch": 0.25, "learning_rate": 4.8989468135549044e-05, "loss": 0.3769, "step": 746 }, { "epoch": 0.25, "learning_rate": 4.898553614504565e-05, "loss": 0.3734, "step": 747 }, { "epoch": 0.26, "learning_rate": 4.898159667809396e-05, "loss": 0.3821, "step": 748 }, { "epoch": 0.26, "learning_rate": 4.897764973592193e-05, "loss": 0.4227, "step": 749 }, { "epoch": 0.26, "learning_rate": 4.8973695319759845e-05, "loss": 0.3903, "step": 750 }, { "epoch": 0.26, "learning_rate": 4.8969733430840324e-05, "loss": 0.3854, "step": 751 }, { "epoch": 0.26, "learning_rate": 4.896576407039831e-05, "loss": 0.3495, "step": 752 }, { "epoch": 0.26, "learning_rate": 4.896178723967108e-05, "loss": 0.3402, "step": 753 }, { "epoch": 0.26, "learning_rate": 4.895780293989825e-05, "loss": 0.3594, "step": 754 }, { "epoch": 0.26, "learning_rate": 4.895381117232174e-05, "loss": 0.3063, "step": 755 }, { "epoch": 0.26, "learning_rate": 4.894981193818581e-05, "loss": 0.3899, "step": 756 }, { "epoch": 0.26, "learning_rate": 4.8945805238737044e-05, "loss": 0.3434, "step": 757 }, { "epoch": 0.26, "learning_rate": 4.894179107522437e-05, "loss": 0.3806, "step": 758 }, { "epoch": 0.26, "learning_rate": 4.8937769448899015e-05, "loss": 0.3507, "step": 759 }, { "epoch": 0.26, "learning_rate": 4.8933740361014554e-05, "loss": 0.4326, "step": 760 }, { "epoch": 0.26, "learning_rate": 4.8929703812826885e-05, "loss": 0.374, "step": 761 }, { "epoch": 0.26, "learning_rate": 4.892565980559422e-05, "loss": 0.3908, "step": 762 }, { "epoch": 0.26, "learning_rate": 4.892160834057709e-05, "loss": 0.3865, "step": 763 }, { "epoch": 0.26, "learning_rate": 4.891754941903838e-05, "loss": 0.3934, "step": 764 }, { "epoch": 0.26, "learning_rate": 4.891348304224329e-05, "loss": 0.3372, "step": 765 }, { "epoch": 0.26, "learning_rate": 4.890940921145931e-05, "loss": 0.3752, "step": 766 }, { "epoch": 0.26, "learning_rate": 4.8905327927956294e-05, "loss": 0.3598, "step": 767 }, { "epoch": 0.26, "learning_rate": 4.890123919300641e-05, "loss": 0.3502, "step": 768 }, { "epoch": 0.26, "learning_rate": 4.8897143007884136e-05, "loss": 0.4189, "step": 769 }, { "epoch": 0.26, "learning_rate": 4.8893039373866276e-05, "loss": 0.3706, "step": 770 }, { "epoch": 0.26, "learning_rate": 4.888892829223197e-05, "loss": 0.4151, "step": 771 }, { "epoch": 0.26, "learning_rate": 4.888480976426267e-05, "loss": 0.4136, "step": 772 }, { "epoch": 0.26, "learning_rate": 4.8880683791242134e-05, "loss": 0.3558, "step": 773 }, { "epoch": 0.26, "learning_rate": 4.887655037445646e-05, "loss": 0.3548, "step": 774 }, { "epoch": 0.26, "learning_rate": 4.8872409515194076e-05, "loss": 0.3529, "step": 775 }, { "epoch": 0.26, "learning_rate": 4.886826121474569e-05, "loss": 0.3886, "step": 776 }, { "epoch": 0.27, "learning_rate": 4.886410547440437e-05, "loss": 0.3881, "step": 777 }, { "epoch": 0.27, "learning_rate": 4.8859942295465486e-05, "loss": 0.3713, "step": 778 }, { "epoch": 0.27, "learning_rate": 4.885577167922672e-05, "loss": 0.3643, "step": 779 }, { "epoch": 0.27, "learning_rate": 4.88515936269881e-05, "loss": 0.3476, "step": 780 }, { "epoch": 0.27, "learning_rate": 4.884740814005193e-05, "loss": 0.4322, "step": 781 }, { "epoch": 0.27, "learning_rate": 4.884321521972287e-05, "loss": 0.3403, "step": 782 }, { "epoch": 0.27, "learning_rate": 4.883901486730786e-05, "loss": 0.3116, "step": 783 }, { "epoch": 0.27, "learning_rate": 4.88348070841162e-05, "loss": 0.3855, "step": 784 }, { "epoch": 0.27, "learning_rate": 4.8830591871459475e-05, "loss": 0.426, "step": 785 }, { "epoch": 0.27, "learning_rate": 4.882636923065159e-05, "loss": 0.389, "step": 786 }, { "epoch": 0.27, "learning_rate": 4.8822139163008775e-05, "loss": 0.3947, "step": 787 }, { "epoch": 0.27, "learning_rate": 4.881790166984957e-05, "loss": 0.3327, "step": 788 }, { "epoch": 0.27, "learning_rate": 4.881365675249482e-05, "loss": 0.3728, "step": 789 }, { "epoch": 0.27, "learning_rate": 4.88094044122677e-05, "loss": 0.3582, "step": 790 }, { "epoch": 0.27, "learning_rate": 4.8805144650493695e-05, "loss": 0.3289, "step": 791 }, { "epoch": 0.27, "learning_rate": 4.8800877468500604e-05, "loss": 0.3237, "step": 792 }, { "epoch": 0.27, "learning_rate": 4.879660286761852e-05, "loss": 0.3716, "step": 793 }, { "epoch": 0.27, "learning_rate": 4.879232084917988e-05, "loss": 0.4081, "step": 794 }, { "epoch": 0.27, "learning_rate": 4.878803141451941e-05, "loss": 0.3753, "step": 795 }, { "epoch": 0.27, "learning_rate": 4.878373456497416e-05, "loss": 0.3314, "step": 796 }, { "epoch": 0.27, "learning_rate": 4.8779430301883475e-05, "loss": 0.3725, "step": 797 }, { "epoch": 0.27, "learning_rate": 4.8775118626589034e-05, "loss": 0.3691, "step": 798 }, { "epoch": 0.27, "learning_rate": 4.877079954043481e-05, "loss": 0.364, "step": 799 }, { "epoch": 0.27, "learning_rate": 4.876647304476709e-05, "loss": 0.381, "step": 800 }, { "epoch": 0.27, "learning_rate": 4.8762139140934456e-05, "loss": 0.3521, "step": 801 }, { "epoch": 0.27, "learning_rate": 4.8757797830287846e-05, "loss": 0.3521, "step": 802 }, { "epoch": 0.27, "learning_rate": 4.875344911418045e-05, "loss": 0.3946, "step": 803 }, { "epoch": 0.27, "learning_rate": 4.87490929939678e-05, "loss": 0.3943, "step": 804 }, { "epoch": 0.27, "learning_rate": 4.874472947100772e-05, "loss": 0.3557, "step": 805 }, { "epoch": 0.27, "learning_rate": 4.874035854666036e-05, "loss": 0.3889, "step": 806 }, { "epoch": 0.28, "learning_rate": 4.873598022228817e-05, "loss": 0.3674, "step": 807 }, { "epoch": 0.28, "learning_rate": 4.873159449925588e-05, "loss": 0.4205, "step": 808 }, { "epoch": 0.28, "learning_rate": 4.872720137893056e-05, "loss": 0.3368, "step": 809 }, { "epoch": 0.28, "learning_rate": 4.8722800862681576e-05, "loss": 0.3555, "step": 810 }, { "epoch": 0.28, "learning_rate": 4.8718392951880606e-05, "loss": 0.3359, "step": 811 }, { "epoch": 0.28, "learning_rate": 4.8713977647901605e-05, "loss": 0.3581, "step": 812 }, { "epoch": 0.28, "learning_rate": 4.870955495212087e-05, "loss": 0.3512, "step": 813 }, { "epoch": 0.28, "learning_rate": 4.870512486591697e-05, "loss": 0.3713, "step": 814 }, { "epoch": 0.28, "learning_rate": 4.87006873906708e-05, "loss": 0.3268, "step": 815 }, { "epoch": 0.28, "learning_rate": 4.869624252776555e-05, "loss": 0.3887, "step": 816 }, { "epoch": 0.28, "learning_rate": 4.869179027858671e-05, "loss": 0.3753, "step": 817 }, { "epoch": 0.28, "learning_rate": 4.868733064452208e-05, "loss": 0.3794, "step": 818 }, { "epoch": 0.28, "learning_rate": 4.868286362696175e-05, "loss": 0.3933, "step": 819 }, { "epoch": 0.28, "learning_rate": 4.867838922729813e-05, "loss": 0.3575, "step": 820 }, { "epoch": 0.28, "learning_rate": 4.8673907446925904e-05, "loss": 0.3572, "step": 821 }, { "epoch": 0.28, "learning_rate": 4.866941828724208e-05, "loss": 0.3499, "step": 822 }, { "epoch": 0.28, "learning_rate": 4.866492174964596e-05, "loss": 0.4202, "step": 823 }, { "epoch": 0.28, "learning_rate": 4.866041783553914e-05, "loss": 0.3401, "step": 824 }, { "epoch": 0.28, "learning_rate": 4.865590654632552e-05, "loss": 0.4061, "step": 825 }, { "epoch": 0.28, "learning_rate": 4.8651387883411316e-05, "loss": 0.4104, "step": 826 }, { "epoch": 0.28, "learning_rate": 4.8646861848204994e-05, "loss": 0.3251, "step": 827 }, { "epoch": 0.28, "learning_rate": 4.864232844211737e-05, "loss": 0.3727, "step": 828 }, { "epoch": 0.28, "learning_rate": 4.863778766656152e-05, "loss": 0.3878, "step": 829 }, { "epoch": 0.28, "learning_rate": 4.863323952295286e-05, "loss": 0.392, "step": 830 }, { "epoch": 0.28, "learning_rate": 4.8628684012709045e-05, "loss": 0.3763, "step": 831 }, { "epoch": 0.28, "learning_rate": 4.862412113725008e-05, "loss": 0.429, "step": 832 }, { "epoch": 0.28, "learning_rate": 4.861955089799823e-05, "loss": 0.3342, "step": 833 }, { "epoch": 0.28, "learning_rate": 4.8614973296378076e-05, "loss": 0.3734, "step": 834 }, { "epoch": 0.28, "learning_rate": 4.861038833381648e-05, "loss": 0.3194, "step": 835 }, { "epoch": 0.29, "learning_rate": 4.860579601174261e-05, "loss": 0.3137, "step": 836 }, { "epoch": 0.29, "learning_rate": 4.860119633158793e-05, "loss": 0.3826, "step": 837 }, { "epoch": 0.29, "learning_rate": 4.8596589294786166e-05, "loss": 0.3467, "step": 838 }, { "epoch": 0.29, "learning_rate": 4.859197490277339e-05, "loss": 0.3548, "step": 839 }, { "epoch": 0.29, "learning_rate": 4.858735315698792e-05, "loss": 0.3422, "step": 840 }, { "epoch": 0.29, "learning_rate": 4.8582724058870386e-05, "loss": 0.3329, "step": 841 }, { "epoch": 0.29, "learning_rate": 4.857808760986372e-05, "loss": 0.3639, "step": 842 }, { "epoch": 0.29, "learning_rate": 4.857344381141312e-05, "loss": 0.3232, "step": 843 }, { "epoch": 0.29, "learning_rate": 4.856879266496609e-05, "loss": 0.365, "step": 844 }, { "epoch": 0.29, "learning_rate": 4.856413417197243e-05, "loss": 0.385, "step": 845 }, { "epoch": 0.29, "learning_rate": 4.855946833388422e-05, "loss": 0.3336, "step": 846 }, { "epoch": 0.29, "learning_rate": 4.855479515215583e-05, "loss": 0.3834, "step": 847 }, { "epoch": 0.29, "learning_rate": 4.855011462824392e-05, "loss": 0.3436, "step": 848 }, { "epoch": 0.29, "learning_rate": 4.854542676360745e-05, "loss": 0.3814, "step": 849 }, { "epoch": 0.29, "learning_rate": 4.854073155970765e-05, "loss": 0.3381, "step": 850 }, { "epoch": 0.29, "learning_rate": 4.853602901800803e-05, "loss": 0.3708, "step": 851 }, { "epoch": 0.29, "learning_rate": 4.8531319139974444e-05, "loss": 0.3476, "step": 852 }, { "epoch": 0.29, "learning_rate": 4.8526601927074945e-05, "loss": 0.391, "step": 853 }, { "epoch": 0.29, "learning_rate": 4.852187738077995e-05, "loss": 0.3569, "step": 854 }, { "epoch": 0.29, "learning_rate": 4.8517145502562123e-05, "loss": 0.3687, "step": 855 }, { "epoch": 0.29, "learning_rate": 4.851240629389642e-05, "loss": 0.3563, "step": 856 }, { "epoch": 0.29, "learning_rate": 4.850765975626007e-05, "loss": 0.426, "step": 857 }, { "epoch": 0.29, "learning_rate": 4.8502905891132626e-05, "loss": 0.371, "step": 858 }, { "epoch": 0.29, "learning_rate": 4.8498144699995884e-05, "loss": 0.3479, "step": 859 }, { "epoch": 0.29, "learning_rate": 4.849337618433394e-05, "loss": 0.3489, "step": 860 }, { "epoch": 0.29, "learning_rate": 4.848860034563317e-05, "loss": 0.4309, "step": 861 }, { "epoch": 0.29, "learning_rate": 4.848381718538224e-05, "loss": 0.3706, "step": 862 }, { "epoch": 0.29, "learning_rate": 4.847902670507208e-05, "loss": 0.3817, "step": 863 }, { "epoch": 0.29, "learning_rate": 4.8474228906195936e-05, "loss": 0.3388, "step": 864 }, { "epoch": 0.3, "learning_rate": 4.846942379024929e-05, "loss": 0.3407, "step": 865 }, { "epoch": 0.3, "learning_rate": 4.846461135872994e-05, "loss": 0.3491, "step": 866 }, { "epoch": 0.3, "learning_rate": 4.8459791613137953e-05, "loss": 0.3803, "step": 867 }, { "epoch": 0.3, "learning_rate": 4.845496455497566e-05, "loss": 0.3636, "step": 868 }, { "epoch": 0.3, "learning_rate": 4.845013018574771e-05, "loss": 0.3562, "step": 869 }, { "epoch": 0.3, "learning_rate": 4.8445288506961e-05, "loss": 0.3572, "step": 870 }, { "epoch": 0.3, "learning_rate": 4.844043952012469e-05, "loss": 0.376, "step": 871 }, { "epoch": 0.3, "learning_rate": 4.8435583226750274e-05, "loss": 0.3449, "step": 872 }, { "epoch": 0.3, "learning_rate": 4.843071962835147e-05, "loss": 0.3211, "step": 873 }, { "epoch": 0.3, "learning_rate": 4.842584872644431e-05, "loss": 0.369, "step": 874 }, { "epoch": 0.3, "learning_rate": 4.8420970522547074e-05, "loss": 0.351, "step": 875 }, { "epoch": 0.3, "learning_rate": 4.841608501818033e-05, "loss": 0.3159, "step": 876 }, { "epoch": 0.3, "learning_rate": 4.841119221486691e-05, "loss": 0.3272, "step": 877 }, { "epoch": 0.3, "learning_rate": 4.840629211413196e-05, "loss": 0.4106, "step": 878 }, { "epoch": 0.3, "learning_rate": 4.840138471750285e-05, "loss": 0.5168, "step": 879 }, { "epoch": 0.3, "learning_rate": 4.8396470026509266e-05, "loss": 0.4267, "step": 880 }, { "epoch": 0.3, "learning_rate": 4.839154804268313e-05, "loss": 0.3123, "step": 881 }, { "epoch": 0.3, "learning_rate": 4.8386618767558676e-05, "loss": 0.3444, "step": 882 }, { "epoch": 0.3, "learning_rate": 4.838168220267238e-05, "loss": 0.3526, "step": 883 }, { "epoch": 0.3, "learning_rate": 4.8376738349563e-05, "loss": 0.357, "step": 884 }, { "epoch": 0.3, "learning_rate": 4.837178720977157e-05, "loss": 0.33, "step": 885 }, { "epoch": 0.3, "learning_rate": 4.83668287848414e-05, "loss": 0.3025, "step": 886 }, { "epoch": 0.3, "learning_rate": 4.8361863076318055e-05, "loss": 0.3444, "step": 887 }, { "epoch": 0.3, "learning_rate": 4.835689008574938e-05, "loss": 0.3459, "step": 888 }, { "epoch": 0.3, "learning_rate": 4.835190981468549e-05, "loss": 0.3633, "step": 889 }, { "epoch": 0.3, "learning_rate": 4.8346922264678765e-05, "loss": 0.3391, "step": 890 }, { "epoch": 0.3, "learning_rate": 4.8341927437283855e-05, "loss": 0.3658, "step": 891 }, { "epoch": 0.3, "learning_rate": 4.8336925334057695e-05, "loss": 0.416, "step": 892 }, { "epoch": 0.3, "learning_rate": 4.8331915956559455e-05, "loss": 0.3577, "step": 893 }, { "epoch": 0.31, "learning_rate": 4.8326899306350604e-05, "loss": 0.3336, "step": 894 }, { "epoch": 0.31, "learning_rate": 4.832187538499485e-05, "loss": 0.3629, "step": 895 }, { "epoch": 0.31, "learning_rate": 4.8316844194058194e-05, "loss": 0.3682, "step": 896 }, { "epoch": 0.31, "learning_rate": 4.8311805735108894e-05, "loss": 0.339, "step": 897 }, { "epoch": 0.31, "learning_rate": 4.830676000971747e-05, "loss": 0.4038, "step": 898 }, { "epoch": 0.31, "learning_rate": 4.830170701945669e-05, "loss": 0.3582, "step": 899 }, { "epoch": 0.31, "learning_rate": 4.829664676590162e-05, "loss": 0.379, "step": 900 }, { "epoch": 0.31, "learning_rate": 4.8291579250629574e-05, "loss": 0.354, "step": 901 }, { "epoch": 0.31, "learning_rate": 4.828650447522013e-05, "loss": 0.3983, "step": 902 }, { "epoch": 0.31, "learning_rate": 4.828142244125512e-05, "loss": 0.4093, "step": 903 }, { "epoch": 0.31, "learning_rate": 4.827633315031865e-05, "loss": 0.3365, "step": 904 }, { "epoch": 0.31, "learning_rate": 4.827123660399708e-05, "loss": 0.4063, "step": 905 }, { "epoch": 0.31, "learning_rate": 4.826613280387906e-05, "loss": 0.353, "step": 906 }, { "epoch": 0.31, "learning_rate": 4.8261021751555456e-05, "loss": 0.3662, "step": 907 }, { "epoch": 0.31, "learning_rate": 4.825590344861942e-05, "loss": 0.3807, "step": 908 }, { "epoch": 0.31, "learning_rate": 4.825077789666637e-05, "loss": 0.3588, "step": 909 }, { "epoch": 0.31, "learning_rate": 4.824564509729396e-05, "loss": 0.3714, "step": 910 }, { "epoch": 0.31, "learning_rate": 4.824050505210212e-05, "loss": 0.4295, "step": 911 }, { "epoch": 0.31, "learning_rate": 4.823535776269305e-05, "loss": 0.3331, "step": 912 }, { "epoch": 0.31, "learning_rate": 4.8230203230671174e-05, "loss": 0.3961, "step": 913 }, { "epoch": 0.31, "learning_rate": 4.8225041457643206e-05, "loss": 0.337, "step": 914 }, { "epoch": 0.31, "learning_rate": 4.82198724452181e-05, "loss": 0.4421, "step": 915 }, { "epoch": 0.31, "learning_rate": 4.821469619500707e-05, "loss": 0.3451, "step": 916 }, { "epoch": 0.31, "learning_rate": 4.8209512708623596e-05, "loss": 0.3853, "step": 917 }, { "epoch": 0.31, "learning_rate": 4.820432198768339e-05, "loss": 0.3964, "step": 918 }, { "epoch": 0.31, "learning_rate": 4.8199124033804444e-05, "loss": 0.3503, "step": 919 }, { "epoch": 0.31, "learning_rate": 4.8193918848606994e-05, "loss": 0.3835, "step": 920 }, { "epoch": 0.31, "learning_rate": 4.818870643371353e-05, "loss": 0.4013, "step": 921 }, { "epoch": 0.31, "learning_rate": 4.8183486790748786e-05, "loss": 0.3504, "step": 922 }, { "epoch": 0.31, "learning_rate": 4.817825992133977e-05, "loss": 0.3861, "step": 923 }, { "epoch": 0.32, "learning_rate": 4.817302582711572e-05, "loss": 0.3193, "step": 924 }, { "epoch": 0.32, "learning_rate": 4.816778450970816e-05, "loss": 0.3526, "step": 925 }, { "epoch": 0.32, "learning_rate": 4.816253597075082e-05, "loss": 0.3449, "step": 926 }, { "epoch": 0.32, "learning_rate": 4.815728021187971e-05, "loss": 0.3711, "step": 927 }, { "epoch": 0.32, "learning_rate": 4.815201723473308e-05, "loss": 0.367, "step": 928 }, { "epoch": 0.32, "learning_rate": 4.814674704095145e-05, "loss": 0.4106, "step": 929 }, { "epoch": 0.32, "learning_rate": 4.814146963217756e-05, "loss": 0.3487, "step": 930 }, { "epoch": 0.32, "learning_rate": 4.8136185010056416e-05, "loss": 0.3187, "step": 931 }, { "epoch": 0.32, "learning_rate": 4.813089317623527e-05, "loss": 0.3686, "step": 932 }, { "epoch": 0.32, "learning_rate": 4.8125594132363625e-05, "loss": 0.366, "step": 933 }, { "epoch": 0.32, "learning_rate": 4.8120287880093224e-05, "loss": 0.3671, "step": 934 }, { "epoch": 0.32, "learning_rate": 4.811497442107806e-05, "loss": 0.3703, "step": 935 }, { "epoch": 0.32, "learning_rate": 4.8109653756974374e-05, "loss": 0.412, "step": 936 }, { "epoch": 0.32, "learning_rate": 4.810432588944064e-05, "loss": 0.3442, "step": 937 }, { "epoch": 0.32, "learning_rate": 4.809899082013761e-05, "loss": 0.3454, "step": 938 }, { "epoch": 0.32, "learning_rate": 4.8093648550728253e-05, "loss": 0.3656, "step": 939 }, { "epoch": 0.32, "learning_rate": 4.808829908287778e-05, "loss": 0.4164, "step": 940 }, { "epoch": 0.32, "learning_rate": 4.808294241825366e-05, "loss": 0.3517, "step": 941 }, { "epoch": 0.32, "learning_rate": 4.807757855852561e-05, "loss": 0.3678, "step": 942 }, { "epoch": 0.32, "learning_rate": 4.807220750536557e-05, "loss": 0.3347, "step": 943 }, { "epoch": 0.32, "learning_rate": 4.8066829260447735e-05, "loss": 0.3617, "step": 944 }, { "epoch": 0.32, "learning_rate": 4.8061443825448526e-05, "loss": 0.3611, "step": 945 }, { "epoch": 0.32, "learning_rate": 4.8056051202046646e-05, "loss": 0.3365, "step": 946 }, { "epoch": 0.32, "learning_rate": 4.805065139192299e-05, "loss": 0.3601, "step": 947 }, { "epoch": 0.32, "learning_rate": 4.8045244396760714e-05, "loss": 0.3508, "step": 948 }, { "epoch": 0.32, "learning_rate": 4.803983021824523e-05, "loss": 0.3755, "step": 949 }, { "epoch": 0.32, "learning_rate": 4.803440885806415e-05, "loss": 0.3431, "step": 950 }, { "epoch": 0.32, "learning_rate": 4.802898031790737e-05, "loss": 0.3283, "step": 951 }, { "epoch": 0.32, "learning_rate": 4.802354459946699e-05, "loss": 0.359, "step": 952 }, { "epoch": 0.33, "learning_rate": 4.801810170443735e-05, "loss": 0.3419, "step": 953 }, { "epoch": 0.33, "learning_rate": 4.801265163451505e-05, "loss": 0.3137, "step": 954 }, { "epoch": 0.33, "learning_rate": 4.8007194391398916e-05, "loss": 0.3742, "step": 955 }, { "epoch": 0.33, "learning_rate": 4.8001729976789996e-05, "loss": 0.3464, "step": 956 }, { "epoch": 0.33, "learning_rate": 4.799625839239158e-05, "loss": 0.3714, "step": 957 }, { "epoch": 0.33, "learning_rate": 4.799077963990921e-05, "loss": 0.3663, "step": 958 }, { "epoch": 0.33, "learning_rate": 4.798529372105064e-05, "loss": 0.3485, "step": 959 }, { "epoch": 0.33, "learning_rate": 4.7979800637525864e-05, "loss": 0.3674, "step": 960 }, { "epoch": 0.33, "learning_rate": 4.7974300391047125e-05, "loss": 0.3723, "step": 961 }, { "epoch": 0.33, "learning_rate": 4.796879298332887e-05, "loss": 0.3864, "step": 962 }, { "epoch": 0.33, "learning_rate": 4.796327841608781e-05, "loss": 0.3674, "step": 963 }, { "epoch": 0.33, "learning_rate": 4.795775669104285e-05, "loss": 0.3857, "step": 964 }, { "epoch": 0.33, "learning_rate": 4.795222780991516e-05, "loss": 0.4047, "step": 965 }, { "epoch": 0.33, "learning_rate": 4.7946691774428144e-05, "loss": 0.3159, "step": 966 }, { "epoch": 0.33, "learning_rate": 4.7941148586307396e-05, "loss": 0.3996, "step": 967 }, { "epoch": 0.33, "learning_rate": 4.7935598247280776e-05, "loss": 0.3178, "step": 968 }, { "epoch": 0.33, "learning_rate": 4.7930040759078353e-05, "loss": 0.3911, "step": 969 }, { "epoch": 0.33, "learning_rate": 4.7924476123432446e-05, "loss": 0.4339, "step": 970 }, { "epoch": 0.33, "learning_rate": 4.791890434207757e-05, "loss": 0.3385, "step": 971 }, { "epoch": 0.33, "learning_rate": 4.791332541675051e-05, "loss": 0.3144, "step": 972 }, { "epoch": 0.33, "learning_rate": 4.790773934919023e-05, "loss": 0.3131, "step": 973 }, { "epoch": 0.33, "learning_rate": 4.790214614113796e-05, "loss": 0.3411, "step": 974 }, { "epoch": 0.33, "learning_rate": 4.7896545794337125e-05, "loss": 0.3696, "step": 975 }, { "epoch": 0.33, "learning_rate": 4.7890938310533395e-05, "loss": 0.3334, "step": 976 }, { "epoch": 0.33, "learning_rate": 4.788532369147467e-05, "loss": 0.3149, "step": 977 }, { "epoch": 0.33, "learning_rate": 4.787970193891105e-05, "loss": 0.3743, "step": 978 }, { "epoch": 0.33, "learning_rate": 4.7874073054594874e-05, "loss": 0.3851, "step": 979 }, { "epoch": 0.33, "learning_rate": 4.786843704028071e-05, "loss": 0.3401, "step": 980 }, { "epoch": 0.33, "learning_rate": 4.7862793897725334e-05, "loss": 0.3455, "step": 981 }, { "epoch": 0.34, "learning_rate": 4.785714362868775e-05, "loss": 0.3165, "step": 982 }, { "epoch": 0.34, "learning_rate": 4.7851486234929185e-05, "loss": 0.37, "step": 983 }, { "epoch": 0.34, "learning_rate": 4.784582171821309e-05, "loss": 0.3539, "step": 984 }, { "epoch": 0.34, "learning_rate": 4.784015008030513e-05, "loss": 0.3448, "step": 985 }, { "epoch": 0.34, "learning_rate": 4.783447132297318e-05, "loss": 0.3716, "step": 986 }, { "epoch": 0.34, "learning_rate": 4.7828785447987355e-05, "loss": 0.3881, "step": 987 }, { "epoch": 0.34, "learning_rate": 4.782309245711999e-05, "loss": 0.3687, "step": 988 }, { "epoch": 0.34, "learning_rate": 4.781739235214561e-05, "loss": 0.3369, "step": 989 }, { "epoch": 0.34, "learning_rate": 4.781168513484099e-05, "loss": 0.418, "step": 990 }, { "epoch": 0.34, "learning_rate": 4.780597080698509e-05, "loss": 0.3599, "step": 991 }, { "epoch": 0.34, "learning_rate": 4.7800249370359116e-05, "loss": 0.3806, "step": 992 }, { "epoch": 0.34, "learning_rate": 4.7794520826746474e-05, "loss": 0.3949, "step": 993 }, { "epoch": 0.34, "learning_rate": 4.778878517793279e-05, "loss": 0.3195, "step": 994 }, { "epoch": 0.34, "learning_rate": 4.778304242570589e-05, "loss": 0.4595, "step": 995 }, { "epoch": 0.34, "learning_rate": 4.7777292571855855e-05, "loss": 0.3118, "step": 996 }, { "epoch": 0.34, "learning_rate": 4.777153561817492e-05, "loss": 0.3927, "step": 997 }, { "epoch": 0.34, "learning_rate": 4.7765771566457597e-05, "loss": 0.3494, "step": 998 }, { "epoch": 0.34, "learning_rate": 4.776000041850055e-05, "loss": 0.3758, "step": 999 }, { "epoch": 0.34, "learning_rate": 4.7754222176102694e-05, "loss": 0.396, "step": 1000 }, { "epoch": 0.34, "learning_rate": 4.7748436841065157e-05, "loss": 0.3539, "step": 1001 }, { "epoch": 0.34, "learning_rate": 4.774264441519125e-05, "loss": 0.3674, "step": 1002 }, { "epoch": 0.34, "learning_rate": 4.773684490028651e-05, "loss": 0.3301, "step": 1003 }, { "epoch": 0.34, "learning_rate": 4.77310382981587e-05, "loss": 0.386, "step": 1004 }, { "epoch": 0.34, "learning_rate": 4.772522461061776e-05, "loss": 0.365, "step": 1005 }, { "epoch": 0.34, "learning_rate": 4.771940383947585e-05, "loss": 0.3274, "step": 1006 }, { "epoch": 0.34, "learning_rate": 4.7713575986547366e-05, "loss": 0.4089, "step": 1007 }, { "epoch": 0.34, "learning_rate": 4.770774105364887e-05, "loss": 0.3524, "step": 1008 }, { "epoch": 0.34, "learning_rate": 4.7701899042599146e-05, "loss": 0.3477, "step": 1009 }, { "epoch": 0.34, "learning_rate": 4.769604995521919e-05, "loss": 0.3645, "step": 1010 }, { "epoch": 0.34, "learning_rate": 4.769019379333222e-05, "loss": 0.3434, "step": 1011 }, { "epoch": 0.35, "learning_rate": 4.7684330558763616e-05, "loss": 0.3363, "step": 1012 }, { "epoch": 0.35, "learning_rate": 4.7678460253340987e-05, "loss": 0.3969, "step": 1013 }, { "epoch": 0.35, "learning_rate": 4.767258287889416e-05, "loss": 0.3556, "step": 1014 }, { "epoch": 0.35, "learning_rate": 4.766669843725514e-05, "loss": 0.4047, "step": 1015 }, { "epoch": 0.35, "learning_rate": 4.766080693025815e-05, "loss": 0.3849, "step": 1016 }, { "epoch": 0.35, "learning_rate": 4.7654908359739614e-05, "loss": 0.3765, "step": 1017 }, { "epoch": 0.35, "learning_rate": 4.764900272753815e-05, "loss": 0.3582, "step": 1018 }, { "epoch": 0.35, "learning_rate": 4.764309003549459e-05, "loss": 0.3625, "step": 1019 }, { "epoch": 0.35, "learning_rate": 4.7637170285451936e-05, "loss": 0.3583, "step": 1020 }, { "epoch": 0.35, "learning_rate": 4.763124347925545e-05, "loss": 0.3321, "step": 1021 }, { "epoch": 0.35, "learning_rate": 4.7625309618752524e-05, "loss": 0.3856, "step": 1022 }, { "epoch": 0.35, "learning_rate": 4.76193687057928e-05, "loss": 0.3837, "step": 1023 }, { "epoch": 0.35, "learning_rate": 4.761342074222808e-05, "loss": 0.3268, "step": 1024 }, { "epoch": 0.35, "learning_rate": 4.7607465729912406e-05, "loss": 0.3088, "step": 1025 }, { "epoch": 0.35, "learning_rate": 4.760150367070198e-05, "loss": 0.3884, "step": 1026 }, { "epoch": 0.35, "learning_rate": 4.7595534566455216e-05, "loss": 0.3546, "step": 1027 }, { "epoch": 0.35, "learning_rate": 4.758955841903273e-05, "loss": 0.3562, "step": 1028 }, { "epoch": 0.35, "learning_rate": 4.7583575230297316e-05, "loss": 0.3469, "step": 1029 }, { "epoch": 0.35, "learning_rate": 4.7577585002113986e-05, "loss": 0.361, "step": 1030 }, { "epoch": 0.35, "learning_rate": 4.757158773634992e-05, "loss": 0.3765, "step": 1031 }, { "epoch": 0.35, "learning_rate": 4.756558343487452e-05, "loss": 0.3798, "step": 1032 }, { "epoch": 0.35, "learning_rate": 4.7559572099559346e-05, "loss": 0.3364, "step": 1033 }, { "epoch": 0.35, "learning_rate": 4.755355373227818e-05, "loss": 0.3836, "step": 1034 }, { "epoch": 0.35, "learning_rate": 4.754752833490699e-05, "loss": 0.3717, "step": 1035 }, { "epoch": 0.35, "learning_rate": 4.754149590932393e-05, "loss": 0.3309, "step": 1036 }, { "epoch": 0.35, "learning_rate": 4.7535456457409344e-05, "loss": 0.3238, "step": 1037 }, { "epoch": 0.35, "learning_rate": 4.752940998104577e-05, "loss": 0.316, "step": 1038 }, { "epoch": 0.35, "learning_rate": 4.752335648211794e-05, "loss": 0.3679, "step": 1039 }, { "epoch": 0.35, "learning_rate": 4.751729596251276e-05, "loss": 0.393, "step": 1040 }, { "epoch": 0.36, "learning_rate": 4.7511228424119334e-05, "loss": 0.3478, "step": 1041 }, { "epoch": 0.36, "learning_rate": 4.750515386882896e-05, "loss": 0.3966, "step": 1042 }, { "epoch": 0.36, "learning_rate": 4.749907229853511e-05, "loss": 0.3665, "step": 1043 }, { "epoch": 0.36, "learning_rate": 4.749298371513345e-05, "loss": 0.345, "step": 1044 }, { "epoch": 0.36, "learning_rate": 4.7486888120521846e-05, "loss": 0.3904, "step": 1045 }, { "epoch": 0.36, "learning_rate": 4.748078551660031e-05, "loss": 0.333, "step": 1046 }, { "epoch": 0.36, "learning_rate": 4.7474675905271083e-05, "loss": 0.4079, "step": 1047 }, { "epoch": 0.36, "learning_rate": 4.746855928843856e-05, "loss": 0.3402, "step": 1048 }, { "epoch": 0.36, "learning_rate": 4.746243566800933e-05, "loss": 0.3299, "step": 1049 }, { "epoch": 0.36, "learning_rate": 4.745630504589217e-05, "loss": 0.3126, "step": 1050 }, { "epoch": 0.36, "learning_rate": 4.745016742399804e-05, "loss": 0.3259, "step": 1051 }, { "epoch": 0.36, "learning_rate": 4.7444022804240065e-05, "loss": 0.4203, "step": 1052 }, { "epoch": 0.36, "learning_rate": 4.743787118853358e-05, "loss": 0.3883, "step": 1053 }, { "epoch": 0.36, "learning_rate": 4.743171257879606e-05, "loss": 0.3409, "step": 1054 }, { "epoch": 0.36, "learning_rate": 4.74255469769472e-05, "loss": 0.3872, "step": 1055 }, { "epoch": 0.36, "learning_rate": 4.741937438490885e-05, "loss": 0.3627, "step": 1056 }, { "epoch": 0.36, "learning_rate": 4.7413194804605066e-05, "loss": 0.3762, "step": 1057 }, { "epoch": 0.36, "learning_rate": 4.7407008237962045e-05, "loss": 0.3382, "step": 1058 }, { "epoch": 0.36, "learning_rate": 4.740081468690818e-05, "loss": 0.3977, "step": 1059 }, { "epoch": 0.36, "learning_rate": 4.739461415337405e-05, "loss": 0.4623, "step": 1060 }, { "epoch": 0.36, "learning_rate": 4.73884066392924e-05, "loss": 0.3632, "step": 1061 }, { "epoch": 0.36, "learning_rate": 4.7382192146598154e-05, "loss": 0.3674, "step": 1062 }, { "epoch": 0.36, "learning_rate": 4.73759706772284e-05, "loss": 0.4448, "step": 1063 }, { "epoch": 0.36, "learning_rate": 4.736974223312242e-05, "loss": 0.3566, "step": 1064 }, { "epoch": 0.36, "learning_rate": 4.736350681622166e-05, "loss": 0.3473, "step": 1065 }, { "epoch": 0.36, "learning_rate": 4.7357264428469736e-05, "loss": 0.36, "step": 1066 }, { "epoch": 0.36, "learning_rate": 4.7351015071812445e-05, "loss": 0.4104, "step": 1067 }, { "epoch": 0.36, "learning_rate": 4.7344758748197754e-05, "loss": 0.3794, "step": 1068 }, { "epoch": 0.36, "learning_rate": 4.733849545957579e-05, "loss": 0.3997, "step": 1069 }, { "epoch": 0.37, "learning_rate": 4.733222520789887e-05, "loss": 0.3513, "step": 1070 }, { "epoch": 0.37, "learning_rate": 4.7325947995121475e-05, "loss": 0.3356, "step": 1071 }, { "epoch": 0.37, "learning_rate": 4.7319663823200245e-05, "loss": 0.3688, "step": 1072 }, { "epoch": 0.37, "learning_rate": 4.7313372694094e-05, "loss": 0.3415, "step": 1073 }, { "epoch": 0.37, "learning_rate": 4.730707460976373e-05, "loss": 0.3617, "step": 1074 }, { "epoch": 0.37, "learning_rate": 4.730076957217259e-05, "loss": 0.3814, "step": 1075 }, { "epoch": 0.37, "learning_rate": 4.729445758328589e-05, "loss": 0.3609, "step": 1076 }, { "epoch": 0.37, "learning_rate": 4.7288138645071124e-05, "loss": 0.3601, "step": 1077 }, { "epoch": 0.37, "learning_rate": 4.7281812759497945e-05, "loss": 0.3553, "step": 1078 }, { "epoch": 0.37, "learning_rate": 4.7275479928538185e-05, "loss": 0.3518, "step": 1079 }, { "epoch": 0.37, "learning_rate": 4.726914015416581e-05, "loss": 0.372, "step": 1080 }, { "epoch": 0.37, "learning_rate": 4.726279343835699e-05, "loss": 0.344, "step": 1081 }, { "epoch": 0.37, "learning_rate": 4.725643978309002e-05, "loss": 0.3542, "step": 1082 }, { "epoch": 0.37, "learning_rate": 4.7250079190345386e-05, "loss": 0.3457, "step": 1083 }, { "epoch": 0.37, "learning_rate": 4.724371166210572e-05, "loss": 0.303, "step": 1084 }, { "epoch": 0.37, "learning_rate": 4.723733720035582e-05, "loss": 0.3775, "step": 1085 }, { "epoch": 0.37, "learning_rate": 4.723095580708266e-05, "loss": 0.3448, "step": 1086 }, { "epoch": 0.37, "learning_rate": 4.722456748427535e-05, "loss": 0.3667, "step": 1087 }, { "epoch": 0.37, "learning_rate": 4.721817223392518e-05, "loss": 0.3671, "step": 1088 }, { "epoch": 0.37, "learning_rate": 4.7211770058025584e-05, "loss": 0.3756, "step": 1089 }, { "epoch": 0.37, "learning_rate": 4.720536095857217e-05, "loss": 0.3904, "step": 1090 }, { "epoch": 0.37, "learning_rate": 4.719894493756268e-05, "loss": 0.3256, "step": 1091 }, { "epoch": 0.37, "learning_rate": 4.7192521996997044e-05, "loss": 0.3988, "step": 1092 }, { "epoch": 0.37, "learning_rate": 4.718609213887734e-05, "loss": 0.2986, "step": 1093 }, { "epoch": 0.37, "learning_rate": 4.717965536520778e-05, "loss": 0.3735, "step": 1094 }, { "epoch": 0.37, "learning_rate": 4.7173211677994764e-05, "loss": 0.3433, "step": 1095 }, { "epoch": 0.37, "learning_rate": 4.716676107924682e-05, "loss": 0.4307, "step": 1096 }, { "epoch": 0.37, "learning_rate": 4.716030357097465e-05, "loss": 0.3976, "step": 1097 }, { "epoch": 0.37, "learning_rate": 4.7153839155191095e-05, "loss": 0.365, "step": 1098 }, { "epoch": 0.37, "learning_rate": 4.714736783391115e-05, "loss": 0.3747, "step": 1099 }, { "epoch": 0.38, "learning_rate": 4.714088960915198e-05, "loss": 0.3493, "step": 1100 }, { "epoch": 0.38, "learning_rate": 4.713440448293289e-05, "loss": 0.3375, "step": 1101 }, { "epoch": 0.38, "learning_rate": 4.712791245727534e-05, "loss": 0.3177, "step": 1102 }, { "epoch": 0.38, "learning_rate": 4.712141353420292e-05, "loss": 0.367, "step": 1103 }, { "epoch": 0.38, "learning_rate": 4.7114907715741395e-05, "loss": 0.3575, "step": 1104 }, { "epoch": 0.38, "learning_rate": 4.710839500391868e-05, "loss": 0.3459, "step": 1105 }, { "epoch": 0.38, "learning_rate": 4.710187540076482e-05, "loss": 0.3421, "step": 1106 }, { "epoch": 0.38, "learning_rate": 4.709534890831201e-05, "loss": 0.3894, "step": 1107 }, { "epoch": 0.38, "learning_rate": 4.708881552859462e-05, "loss": 0.3465, "step": 1108 }, { "epoch": 0.38, "learning_rate": 4.7082275263649135e-05, "loss": 0.3717, "step": 1109 }, { "epoch": 0.38, "learning_rate": 4.70757281155142e-05, "loss": 0.3598, "step": 1110 }, { "epoch": 0.38, "learning_rate": 4.70691740862306e-05, "loss": 0.3295, "step": 1111 }, { "epoch": 0.38, "learning_rate": 4.706261317784128e-05, "loss": 0.3354, "step": 1112 }, { "epoch": 0.38, "learning_rate": 4.705604539239131e-05, "loss": 0.3392, "step": 1113 }, { "epoch": 0.38, "learning_rate": 4.704947073192791e-05, "loss": 0.3673, "step": 1114 }, { "epoch": 0.38, "learning_rate": 4.704288919850045e-05, "loss": 0.3308, "step": 1115 }, { "epoch": 0.38, "learning_rate": 4.7036300794160435e-05, "loss": 0.3761, "step": 1116 }, { "epoch": 0.38, "learning_rate": 4.702970552096151e-05, "loss": 0.3442, "step": 1117 }, { "epoch": 0.38, "learning_rate": 4.702310338095947e-05, "loss": 0.3447, "step": 1118 }, { "epoch": 0.38, "learning_rate": 4.701649437621224e-05, "loss": 0.3564, "step": 1119 }, { "epoch": 0.38, "learning_rate": 4.700987850877989e-05, "loss": 0.3279, "step": 1120 }, { "epoch": 0.38, "learning_rate": 4.700325578072464e-05, "loss": 0.3563, "step": 1121 }, { "epoch": 0.38, "learning_rate": 4.699662619411083e-05, "loss": 0.2977, "step": 1122 }, { "epoch": 0.38, "learning_rate": 4.6989989751004936e-05, "loss": 0.4167, "step": 1123 }, { "epoch": 0.38, "learning_rate": 4.6983346453475596e-05, "loss": 0.3771, "step": 1124 }, { "epoch": 0.38, "learning_rate": 4.6976696303593565e-05, "loss": 0.3609, "step": 1125 }, { "epoch": 0.38, "learning_rate": 4.697003930343173e-05, "loss": 0.3347, "step": 1126 }, { "epoch": 0.38, "learning_rate": 4.696337545506514e-05, "loss": 0.363, "step": 1127 }, { "epoch": 0.38, "learning_rate": 4.6956704760570936e-05, "loss": 0.3847, "step": 1128 }, { "epoch": 0.39, "learning_rate": 4.695002722202843e-05, "loss": 0.3273, "step": 1129 }, { "epoch": 0.39, "learning_rate": 4.694334284151906e-05, "loss": 0.3174, "step": 1130 }, { "epoch": 0.39, "learning_rate": 4.6936651621126374e-05, "loss": 0.4282, "step": 1131 }, { "epoch": 0.39, "learning_rate": 4.692995356293609e-05, "loss": 0.3376, "step": 1132 }, { "epoch": 0.39, "learning_rate": 4.692324866903601e-05, "loss": 0.3612, "step": 1133 }, { "epoch": 0.39, "learning_rate": 4.691653694151613e-05, "loss": 0.3558, "step": 1134 }, { "epoch": 0.39, "learning_rate": 4.6909818382468503e-05, "loss": 0.361, "step": 1135 }, { "epoch": 0.39, "learning_rate": 4.690309299398736e-05, "loss": 0.3183, "step": 1136 }, { "epoch": 0.39, "learning_rate": 4.689636077816907e-05, "loss": 0.3874, "step": 1137 }, { "epoch": 0.39, "learning_rate": 4.6889621737112076e-05, "loss": 0.4168, "step": 1138 }, { "epoch": 0.39, "learning_rate": 4.6882875872916996e-05, "loss": 0.432, "step": 1139 }, { "epoch": 0.39, "learning_rate": 4.687612318768656e-05, "loss": 0.3756, "step": 1140 }, { "epoch": 0.39, "learning_rate": 4.686936368352563e-05, "loss": 0.3797, "step": 1141 }, { "epoch": 0.39, "learning_rate": 4.686259736254117e-05, "loss": 0.3856, "step": 1142 }, { "epoch": 0.39, "learning_rate": 4.6855824226842306e-05, "loss": 0.3444, "step": 1143 }, { "epoch": 0.39, "learning_rate": 4.684904427854026e-05, "loss": 0.3696, "step": 1144 }, { "epoch": 0.39, "learning_rate": 4.684225751974838e-05, "loss": 0.3442, "step": 1145 }, { "epoch": 0.39, "learning_rate": 4.683546395258216e-05, "loss": 0.3785, "step": 1146 }, { "epoch": 0.39, "learning_rate": 4.6828663579159184e-05, "loss": 0.3736, "step": 1147 }, { "epoch": 0.39, "learning_rate": 4.682185640159917e-05, "loss": 0.3936, "step": 1148 }, { "epoch": 0.39, "learning_rate": 4.681504242202397e-05, "loss": 0.3965, "step": 1149 }, { "epoch": 0.39, "learning_rate": 4.680822164255755e-05, "loss": 0.3555, "step": 1150 }, { "epoch": 0.39, "learning_rate": 4.680139406532598e-05, "loss": 0.3907, "step": 1151 }, { "epoch": 0.39, "learning_rate": 4.679455969245746e-05, "loss": 0.3139, "step": 1152 }, { "epoch": 0.39, "learning_rate": 4.6787718526082326e-05, "loss": 0.3643, "step": 1153 }, { "epoch": 0.39, "learning_rate": 4.678087056833299e-05, "loss": 0.3467, "step": 1154 }, { "epoch": 0.39, "learning_rate": 4.677401582134402e-05, "loss": 0.3722, "step": 1155 }, { "epoch": 0.39, "learning_rate": 4.6767154287252084e-05, "loss": 0.3498, "step": 1156 }, { "epoch": 0.39, "learning_rate": 4.676028596819597e-05, "loss": 0.3471, "step": 1157 }, { "epoch": 0.4, "learning_rate": 4.675341086631656e-05, "loss": 0.3416, "step": 1158 }, { "epoch": 0.4, "learning_rate": 4.674652898375689e-05, "loss": 0.3749, "step": 1159 }, { "epoch": 0.4, "learning_rate": 4.673964032266208e-05, "loss": 0.4216, "step": 1160 }, { "epoch": 0.4, "learning_rate": 4.673274488517937e-05, "loss": 0.3825, "step": 1161 }, { "epoch": 0.4, "learning_rate": 4.6725842673458116e-05, "loss": 0.3836, "step": 1162 }, { "epoch": 0.4, "learning_rate": 4.6718933689649775e-05, "loss": 0.3802, "step": 1163 }, { "epoch": 0.4, "learning_rate": 4.671201793590792e-05, "loss": 0.3279, "step": 1164 }, { "epoch": 0.4, "learning_rate": 4.670509541438826e-05, "loss": 0.3362, "step": 1165 }, { "epoch": 0.4, "learning_rate": 4.669816612724856e-05, "loss": 0.427, "step": 1166 }, { "epoch": 0.4, "learning_rate": 4.669123007664874e-05, "loss": 0.3516, "step": 1167 }, { "epoch": 0.4, "learning_rate": 4.6684287264750814e-05, "loss": 0.3347, "step": 1168 }, { "epoch": 0.4, "learning_rate": 4.6677337693718895e-05, "loss": 0.5573, "step": 1169 }, { "epoch": 0.4, "learning_rate": 4.6670381365719215e-05, "loss": 0.3515, "step": 1170 }, { "epoch": 0.4, "learning_rate": 4.6663418282920104e-05, "loss": 0.3665, "step": 1171 }, { "epoch": 0.4, "learning_rate": 4.6656448447492e-05, "loss": 0.3696, "step": 1172 }, { "epoch": 0.4, "learning_rate": 4.664947186160744e-05, "loss": 0.3694, "step": 1173 }, { "epoch": 0.4, "learning_rate": 4.664248852744109e-05, "loss": 0.3649, "step": 1174 }, { "epoch": 0.4, "learning_rate": 4.663549844716968e-05, "loss": 0.3753, "step": 1175 }, { "epoch": 0.4, "learning_rate": 4.662850162297208e-05, "loss": 0.3789, "step": 1176 }, { "epoch": 0.4, "learning_rate": 4.662149805702922e-05, "loss": 0.3759, "step": 1177 }, { "epoch": 0.4, "learning_rate": 4.661448775152419e-05, "loss": 0.3571, "step": 1178 }, { "epoch": 0.4, "learning_rate": 4.6607470708642145e-05, "loss": 0.3231, "step": 1179 }, { "epoch": 0.4, "learning_rate": 4.660044693057032e-05, "loss": 0.3458, "step": 1180 }, { "epoch": 0.4, "learning_rate": 4.659341641949808e-05, "loss": 0.3274, "step": 1181 }, { "epoch": 0.4, "learning_rate": 4.6586379177616894e-05, "loss": 0.326, "step": 1182 }, { "epoch": 0.4, "learning_rate": 4.65793352071203e-05, "loss": 0.3857, "step": 1183 }, { "epoch": 0.4, "learning_rate": 4.6572284510203964e-05, "loss": 0.3376, "step": 1184 }, { "epoch": 0.4, "learning_rate": 4.656522708906563e-05, "loss": 0.3206, "step": 1185 }, { "epoch": 0.4, "learning_rate": 4.655816294590514e-05, "loss": 0.3834, "step": 1186 }, { "epoch": 0.4, "learning_rate": 4.655109208292443e-05, "loss": 0.3492, "step": 1187 }, { "epoch": 0.41, "learning_rate": 4.6544014502327535e-05, "loss": 0.348, "step": 1188 }, { "epoch": 0.41, "learning_rate": 4.65369302063206e-05, "loss": 0.3661, "step": 1189 }, { "epoch": 0.41, "learning_rate": 4.6529839197111814e-05, "loss": 0.3846, "step": 1190 }, { "epoch": 0.41, "learning_rate": 4.6522741476911526e-05, "loss": 0.3209, "step": 1191 }, { "epoch": 0.41, "learning_rate": 4.651563704793212e-05, "loss": 0.3147, "step": 1192 }, { "epoch": 0.41, "learning_rate": 4.650852591238809e-05, "loss": 0.3291, "step": 1193 }, { "epoch": 0.41, "learning_rate": 4.6501408072496036e-05, "loss": 0.3526, "step": 1194 }, { "epoch": 0.41, "learning_rate": 4.6494283530474624e-05, "loss": 0.4177, "step": 1195 }, { "epoch": 0.41, "learning_rate": 4.648715228854463e-05, "loss": 0.4025, "step": 1196 }, { "epoch": 0.41, "learning_rate": 4.6480014348928914e-05, "loss": 0.3895, "step": 1197 }, { "epoch": 0.41, "learning_rate": 4.647286971385241e-05, "loss": 0.3792, "step": 1198 }, { "epoch": 0.41, "learning_rate": 4.646571838554214e-05, "loss": 0.3407, "step": 1199 }, { "epoch": 0.41, "learning_rate": 4.6458560366227224e-05, "loss": 0.4844, "step": 1200 }, { "epoch": 0.41, "learning_rate": 4.6451395658138874e-05, "loss": 0.3485, "step": 1201 }, { "epoch": 0.41, "learning_rate": 4.6444224263510375e-05, "loss": 0.4483, "step": 1202 }, { "epoch": 0.41, "learning_rate": 4.6437046184577073e-05, "loss": 0.3221, "step": 1203 }, { "epoch": 0.41, "learning_rate": 4.642986142357645e-05, "loss": 0.4153, "step": 1204 }, { "epoch": 0.41, "learning_rate": 4.6422669982748036e-05, "loss": 0.3795, "step": 1205 }, { "epoch": 0.41, "learning_rate": 4.641547186433345e-05, "loss": 0.3537, "step": 1206 }, { "epoch": 0.41, "learning_rate": 4.640826707057638e-05, "loss": 0.3704, "step": 1207 }, { "epoch": 0.41, "learning_rate": 4.640105560372263e-05, "loss": 0.3233, "step": 1208 }, { "epoch": 0.41, "learning_rate": 4.6393837466020035e-05, "loss": 0.3271, "step": 1209 }, { "epoch": 0.41, "learning_rate": 4.638661265971856e-05, "loss": 0.4121, "step": 1210 }, { "epoch": 0.41, "learning_rate": 4.63793811870702e-05, "loss": 0.3606, "step": 1211 }, { "epoch": 0.41, "learning_rate": 4.637214305032908e-05, "loss": 0.3999, "step": 1212 }, { "epoch": 0.41, "learning_rate": 4.636489825175135e-05, "loss": 0.3039, "step": 1213 }, { "epoch": 0.41, "learning_rate": 4.635764679359528e-05, "loss": 0.3978, "step": 1214 }, { "epoch": 0.41, "learning_rate": 4.6350388678121184e-05, "loss": 0.3253, "step": 1215 }, { "epoch": 0.41, "learning_rate": 4.6343123907591476e-05, "loss": 0.4326, "step": 1216 }, { "epoch": 0.42, "learning_rate": 4.6335852484270617e-05, "loss": 0.3516, "step": 1217 }, { "epoch": 0.42, "learning_rate": 4.6328574410425164e-05, "loss": 0.4009, "step": 1218 }, { "epoch": 0.42, "learning_rate": 4.632128968832374e-05, "loss": 0.4642, "step": 1219 }, { "epoch": 0.42, "learning_rate": 4.6313998320237054e-05, "loss": 0.411, "step": 1220 }, { "epoch": 0.42, "learning_rate": 4.630670030843785e-05, "loss": 0.3414, "step": 1221 }, { "epoch": 0.42, "learning_rate": 4.629939565520097e-05, "loss": 0.3474, "step": 1222 }, { "epoch": 0.42, "learning_rate": 4.629208436280334e-05, "loss": 0.32, "step": 1223 }, { "epoch": 0.42, "learning_rate": 4.628476643352392e-05, "loss": 0.3521, "step": 1224 }, { "epoch": 0.42, "learning_rate": 4.6277441869643764e-05, "loss": 0.3898, "step": 1225 }, { "epoch": 0.42, "learning_rate": 4.627011067344599e-05, "loss": 0.3183, "step": 1226 }, { "epoch": 0.42, "learning_rate": 4.626277284721577e-05, "loss": 0.3529, "step": 1227 }, { "epoch": 0.42, "learning_rate": 4.625542839324036e-05, "loss": 0.3722, "step": 1228 }, { "epoch": 0.42, "learning_rate": 4.624807731380907e-05, "loss": 0.3176, "step": 1229 }, { "epoch": 0.42, "learning_rate": 4.624071961121328e-05, "loss": 0.3139, "step": 1230 }, { "epoch": 0.42, "learning_rate": 4.623335528774643e-05, "loss": 0.3183, "step": 1231 }, { "epoch": 0.42, "learning_rate": 4.622598434570404e-05, "loss": 0.3716, "step": 1232 }, { "epoch": 0.42, "learning_rate": 4.621860678738367e-05, "loss": 0.3707, "step": 1233 }, { "epoch": 0.42, "learning_rate": 4.621122261508496e-05, "loss": 0.3447, "step": 1234 }, { "epoch": 0.42, "learning_rate": 4.620383183110959e-05, "loss": 0.3824, "step": 1235 }, { "epoch": 0.42, "learning_rate": 4.6196434437761335e-05, "loss": 0.3746, "step": 1236 }, { "epoch": 0.42, "learning_rate": 4.618903043734601e-05, "loss": 0.325, "step": 1237 }, { "epoch": 0.42, "learning_rate": 4.6181619832171473e-05, "loss": 0.3985, "step": 1238 }, { "epoch": 0.42, "learning_rate": 4.6174202624547676e-05, "loss": 0.3478, "step": 1239 }, { "epoch": 0.42, "learning_rate": 4.616677881678659e-05, "loss": 0.393, "step": 1240 }, { "epoch": 0.42, "learning_rate": 4.615934841120229e-05, "loss": 0.376, "step": 1241 }, { "epoch": 0.42, "learning_rate": 4.6151911410110874e-05, "loss": 0.3479, "step": 1242 }, { "epoch": 0.42, "learning_rate": 4.6144467815830484e-05, "loss": 0.3732, "step": 1243 }, { "epoch": 0.42, "learning_rate": 4.6137017630681355e-05, "loss": 0.3125, "step": 1244 }, { "epoch": 0.42, "learning_rate": 4.6129560856985764e-05, "loss": 0.362, "step": 1245 }, { "epoch": 0.43, "learning_rate": 4.6122097497068026e-05, "loss": 0.3791, "step": 1246 }, { "epoch": 0.43, "learning_rate": 4.611462755325451e-05, "loss": 0.3403, "step": 1247 }, { "epoch": 0.43, "learning_rate": 4.610715102787367e-05, "loss": 0.3537, "step": 1248 }, { "epoch": 0.43, "learning_rate": 4.609966792325597e-05, "loss": 0.3791, "step": 1249 }, { "epoch": 0.43, "learning_rate": 4.6092178241733946e-05, "loss": 0.3715, "step": 1250 }, { "epoch": 0.43, "learning_rate": 4.608468198564218e-05, "loss": 0.3834, "step": 1251 }, { "epoch": 0.43, "learning_rate": 4.6077179157317304e-05, "loss": 0.3643, "step": 1252 }, { "epoch": 0.43, "learning_rate": 4.606966975909801e-05, "loss": 0.3429, "step": 1253 }, { "epoch": 0.43, "learning_rate": 4.606215379332501e-05, "loss": 0.3888, "step": 1254 }, { "epoch": 0.43, "learning_rate": 4.6054631262341096e-05, "loss": 0.3736, "step": 1255 }, { "epoch": 0.43, "learning_rate": 4.6047102168491076e-05, "loss": 0.332, "step": 1256 }, { "epoch": 0.43, "learning_rate": 4.603956651412182e-05, "loss": 0.3499, "step": 1257 }, { "epoch": 0.43, "learning_rate": 4.603202430158225e-05, "loss": 0.3175, "step": 1258 }, { "epoch": 0.43, "learning_rate": 4.60244755332233e-05, "loss": 0.3505, "step": 1259 }, { "epoch": 0.43, "learning_rate": 4.6016920211397995e-05, "loss": 0.355, "step": 1260 }, { "epoch": 0.43, "learning_rate": 4.6009358338461374e-05, "loss": 0.364, "step": 1261 }, { "epoch": 0.43, "learning_rate": 4.600178991677052e-05, "loss": 0.3264, "step": 1262 }, { "epoch": 0.43, "learning_rate": 4.599421494868455e-05, "loss": 0.3491, "step": 1263 }, { "epoch": 0.43, "learning_rate": 4.598663343656464e-05, "loss": 0.3608, "step": 1264 }, { "epoch": 0.43, "learning_rate": 4.5979045382774e-05, "loss": 0.3699, "step": 1265 }, { "epoch": 0.43, "learning_rate": 4.597145078967786e-05, "loss": 0.3494, "step": 1266 }, { "epoch": 0.43, "learning_rate": 4.5963849659643526e-05, "loss": 0.3625, "step": 1267 }, { "epoch": 0.43, "learning_rate": 4.5956241995040304e-05, "loss": 0.3511, "step": 1268 }, { "epoch": 0.43, "learning_rate": 4.594862779823956e-05, "loss": 0.3909, "step": 1269 }, { "epoch": 0.43, "learning_rate": 4.5941007071614696e-05, "loss": 0.3593, "step": 1270 }, { "epoch": 0.43, "learning_rate": 4.593337981754112e-05, "loss": 0.3624, "step": 1271 }, { "epoch": 0.43, "learning_rate": 4.5925746038396314e-05, "loss": 0.3442, "step": 1272 }, { "epoch": 0.43, "learning_rate": 4.591810573655978e-05, "loss": 0.3365, "step": 1273 }, { "epoch": 0.43, "learning_rate": 4.5910458914413035e-05, "loss": 0.3185, "step": 1274 }, { "epoch": 0.44, "learning_rate": 4.590280557433965e-05, "loss": 0.3701, "step": 1275 }, { "epoch": 0.44, "learning_rate": 4.589514571872523e-05, "loss": 0.3802, "step": 1276 }, { "epoch": 0.44, "learning_rate": 4.588747934995739e-05, "loss": 0.3635, "step": 1277 }, { "epoch": 0.44, "learning_rate": 4.587980647042579e-05, "loss": 0.3763, "step": 1278 }, { "epoch": 0.44, "learning_rate": 4.587212708252212e-05, "loss": 0.4003, "step": 1279 }, { "epoch": 0.44, "learning_rate": 4.586444118864009e-05, "loss": 0.3556, "step": 1280 }, { "epoch": 0.44, "learning_rate": 4.585674879117546e-05, "loss": 0.3706, "step": 1281 }, { "epoch": 0.44, "learning_rate": 4.584904989252598e-05, "loss": 0.3535, "step": 1282 }, { "epoch": 0.44, "learning_rate": 4.5841344495091454e-05, "loss": 0.3755, "step": 1283 }, { "epoch": 0.44, "learning_rate": 4.58336326012737e-05, "loss": 0.3422, "step": 1284 }, { "epoch": 0.44, "learning_rate": 4.582591421347658e-05, "loss": 0.3767, "step": 1285 }, { "epoch": 0.44, "learning_rate": 4.581818933410596e-05, "loss": 0.343, "step": 1286 }, { "epoch": 0.44, "learning_rate": 4.581045796556972e-05, "loss": 0.3394, "step": 1287 }, { "epoch": 0.44, "learning_rate": 4.58027201102778e-05, "loss": 0.3373, "step": 1288 }, { "epoch": 0.44, "learning_rate": 4.5794975770642125e-05, "loss": 0.3696, "step": 1289 }, { "epoch": 0.44, "learning_rate": 4.578722494907666e-05, "loss": 0.3349, "step": 1290 }, { "epoch": 0.44, "learning_rate": 4.577946764799739e-05, "loss": 0.3706, "step": 1291 }, { "epoch": 0.44, "learning_rate": 4.5771703869822324e-05, "loss": 0.3497, "step": 1292 }, { "epoch": 0.44, "learning_rate": 4.576393361697146e-05, "loss": 0.3182, "step": 1293 }, { "epoch": 0.44, "learning_rate": 4.5756156891866856e-05, "loss": 0.393, "step": 1294 }, { "epoch": 0.44, "learning_rate": 4.5748373696932566e-05, "loss": 0.3896, "step": 1295 }, { "epoch": 0.44, "learning_rate": 4.574058403459465e-05, "loss": 0.3648, "step": 1296 }, { "epoch": 0.44, "learning_rate": 4.573278790728121e-05, "loss": 0.37, "step": 1297 }, { "epoch": 0.44, "learning_rate": 4.5724985317422343e-05, "loss": 0.4031, "step": 1298 }, { "epoch": 0.44, "learning_rate": 4.5717176267450176e-05, "loss": 0.4519, "step": 1299 }, { "epoch": 0.44, "learning_rate": 4.5709360759798824e-05, "loss": 0.3217, "step": 1300 }, { "epoch": 0.44, "learning_rate": 4.5701538796904447e-05, "loss": 0.3393, "step": 1301 }, { "epoch": 0.44, "learning_rate": 4.56937103812052e-05, "loss": 0.4134, "step": 1302 }, { "epoch": 0.44, "learning_rate": 4.568587551514124e-05, "loss": 0.3087, "step": 1303 }, { "epoch": 0.44, "learning_rate": 4.5678034201154765e-05, "loss": 0.343, "step": 1304 }, { "epoch": 0.45, "learning_rate": 4.567018644168994e-05, "loss": 0.4063, "step": 1305 }, { "epoch": 0.45, "learning_rate": 4.566233223919298e-05, "loss": 0.4115, "step": 1306 }, { "epoch": 0.45, "learning_rate": 4.565447159611209e-05, "loss": 0.3322, "step": 1307 }, { "epoch": 0.45, "learning_rate": 4.564660451489749e-05, "loss": 0.3032, "step": 1308 }, { "epoch": 0.45, "learning_rate": 4.5638730998001374e-05, "loss": 0.3336, "step": 1309 }, { "epoch": 0.45, "learning_rate": 4.563085104787799e-05, "loss": 0.3876, "step": 1310 }, { "epoch": 0.45, "learning_rate": 4.562296466698357e-05, "loss": 0.3449, "step": 1311 }, { "epoch": 0.45, "learning_rate": 4.561507185777635e-05, "loss": 0.3347, "step": 1312 }, { "epoch": 0.45, "learning_rate": 4.5607172622716555e-05, "loss": 0.3764, "step": 1313 }, { "epoch": 0.45, "learning_rate": 4.559926696426645e-05, "loss": 0.389, "step": 1314 }, { "epoch": 0.45, "learning_rate": 4.5591354884890264e-05, "loss": 0.3249, "step": 1315 }, { "epoch": 0.45, "learning_rate": 4.558343638705426e-05, "loss": 0.3791, "step": 1316 }, { "epoch": 0.45, "learning_rate": 4.557551147322666e-05, "loss": 0.3398, "step": 1317 }, { "epoch": 0.45, "learning_rate": 4.556758014587774e-05, "loss": 0.3086, "step": 1318 }, { "epoch": 0.45, "learning_rate": 4.5559642407479726e-05, "loss": 0.4698, "step": 1319 }, { "epoch": 0.45, "learning_rate": 4.555169826050687e-05, "loss": 0.3371, "step": 1320 }, { "epoch": 0.45, "learning_rate": 4.554374770743542e-05, "loss": 0.3264, "step": 1321 }, { "epoch": 0.45, "learning_rate": 4.553579075074361e-05, "loss": 0.3442, "step": 1322 }, { "epoch": 0.45, "learning_rate": 4.552782739291167e-05, "loss": 0.4239, "step": 1323 }, { "epoch": 0.45, "learning_rate": 4.551985763642184e-05, "loss": 0.3945, "step": 1324 }, { "epoch": 0.45, "learning_rate": 4.551188148375834e-05, "loss": 0.4069, "step": 1325 }, { "epoch": 0.45, "learning_rate": 4.550389893740739e-05, "loss": 0.3729, "step": 1326 }, { "epoch": 0.45, "learning_rate": 4.5495909999857204e-05, "loss": 0.3857, "step": 1327 }, { "epoch": 0.45, "learning_rate": 4.5487914673597985e-05, "loss": 0.3759, "step": 1328 }, { "epoch": 0.45, "learning_rate": 4.547991296112193e-05, "loss": 0.325, "step": 1329 }, { "epoch": 0.45, "learning_rate": 4.547190486492321e-05, "loss": 0.3326, "step": 1330 }, { "epoch": 0.45, "learning_rate": 4.546389038749802e-05, "loss": 0.3233, "step": 1331 }, { "epoch": 0.45, "learning_rate": 4.545586953134452e-05, "loss": 0.3834, "step": 1332 }, { "epoch": 0.45, "learning_rate": 4.5447842298962864e-05, "loss": 0.3589, "step": 1333 }, { "epoch": 0.46, "learning_rate": 4.543980869285519e-05, "loss": 0.3646, "step": 1334 }, { "epoch": 0.46, "learning_rate": 4.543176871552563e-05, "loss": 0.3593, "step": 1335 }, { "epoch": 0.46, "learning_rate": 4.542372236948028e-05, "loss": 0.3832, "step": 1336 }, { "epoch": 0.46, "learning_rate": 4.541566965722726e-05, "loss": 0.3214, "step": 1337 }, { "epoch": 0.46, "learning_rate": 4.540761058127664e-05, "loss": 0.3561, "step": 1338 }, { "epoch": 0.46, "learning_rate": 4.53995451441405e-05, "loss": 0.3328, "step": 1339 }, { "epoch": 0.46, "learning_rate": 4.539147334833288e-05, "loss": 0.3604, "step": 1340 }, { "epoch": 0.46, "learning_rate": 4.5383395196369806e-05, "loss": 0.3656, "step": 1341 }, { "epoch": 0.46, "learning_rate": 4.53753106907693e-05, "loss": 0.3048, "step": 1342 }, { "epoch": 0.46, "learning_rate": 4.5367219834051354e-05, "loss": 0.3515, "step": 1343 }, { "epoch": 0.46, "learning_rate": 4.535912262873794e-05, "loss": 0.3567, "step": 1344 }, { "epoch": 0.46, "learning_rate": 4.5351019077353006e-05, "loss": 0.3619, "step": 1345 }, { "epoch": 0.46, "learning_rate": 4.534290918242249e-05, "loss": 0.3787, "step": 1346 }, { "epoch": 0.46, "learning_rate": 4.533479294647429e-05, "loss": 0.341, "step": 1347 }, { "epoch": 0.46, "learning_rate": 4.5326670372038296e-05, "loss": 0.3134, "step": 1348 }, { "epoch": 0.46, "learning_rate": 4.531854146164637e-05, "loss": 0.3385, "step": 1349 }, { "epoch": 0.46, "learning_rate": 4.531040621783233e-05, "loss": 0.3745, "step": 1350 }, { "epoch": 0.46, "learning_rate": 4.530226464313201e-05, "loss": 0.3826, "step": 1351 }, { "epoch": 0.46, "learning_rate": 4.5294116740083173e-05, "loss": 0.3747, "step": 1352 }, { "epoch": 0.46, "learning_rate": 4.528596251122558e-05, "loss": 0.3884, "step": 1353 }, { "epoch": 0.46, "learning_rate": 4.5277801959100954e-05, "loss": 0.3285, "step": 1354 }, { "epoch": 0.46, "learning_rate": 4.5269635086253e-05, "loss": 0.3403, "step": 1355 }, { "epoch": 0.46, "learning_rate": 4.5261461895227374e-05, "loss": 0.3755, "step": 1356 }, { "epoch": 0.46, "learning_rate": 4.525328238857173e-05, "loss": 0.3581, "step": 1357 }, { "epoch": 0.46, "learning_rate": 4.5245096568835656e-05, "loss": 0.3367, "step": 1358 }, { "epoch": 0.46, "learning_rate": 4.5236904438570734e-05, "loss": 0.3459, "step": 1359 }, { "epoch": 0.46, "learning_rate": 4.52287060003305e-05, "loss": 0.3876, "step": 1360 }, { "epoch": 0.46, "learning_rate": 4.5220501256670466e-05, "loss": 0.3075, "step": 1361 }, { "epoch": 0.46, "learning_rate": 4.521229021014811e-05, "loss": 0.4039, "step": 1362 }, { "epoch": 0.47, "learning_rate": 4.520407286332285e-05, "loss": 0.3766, "step": 1363 }, { "epoch": 0.47, "learning_rate": 4.5195849218756104e-05, "loss": 0.3743, "step": 1364 }, { "epoch": 0.47, "learning_rate": 4.518761927901123e-05, "loss": 0.4024, "step": 1365 }, { "epoch": 0.47, "learning_rate": 4.517938304665355e-05, "loss": 0.3614, "step": 1366 }, { "epoch": 0.47, "learning_rate": 4.5171140524250356e-05, "loss": 0.3423, "step": 1367 }, { "epoch": 0.47, "learning_rate": 4.51628917143709e-05, "loss": 0.3368, "step": 1368 }, { "epoch": 0.47, "learning_rate": 4.515463661958638e-05, "loss": 0.3375, "step": 1369 }, { "epoch": 0.47, "learning_rate": 4.5146375242469975e-05, "loss": 0.345, "step": 1370 }, { "epoch": 0.47, "learning_rate": 4.51381075855968e-05, "loss": 0.3569, "step": 1371 }, { "epoch": 0.47, "learning_rate": 4.5129833651543936e-05, "loss": 0.3914, "step": 1372 }, { "epoch": 0.47, "learning_rate": 4.512155344289043e-05, "loss": 0.3757, "step": 1373 }, { "epoch": 0.47, "learning_rate": 4.511326696221727e-05, "loss": 0.3409, "step": 1374 }, { "epoch": 0.47, "learning_rate": 4.510497421210742e-05, "loss": 0.376, "step": 1375 }, { "epoch": 0.47, "learning_rate": 4.509667519514577e-05, "loss": 0.3474, "step": 1376 }, { "epoch": 0.47, "learning_rate": 4.5088369913919184e-05, "loss": 0.3754, "step": 1377 }, { "epoch": 0.47, "learning_rate": 4.5080058371016464e-05, "loss": 0.3683, "step": 1378 }, { "epoch": 0.47, "learning_rate": 4.5071740569028386e-05, "loss": 0.3682, "step": 1379 }, { "epoch": 0.47, "learning_rate": 4.5063416510547655e-05, "loss": 0.3474, "step": 1380 }, { "epoch": 0.47, "learning_rate": 4.505508619816894e-05, "loss": 0.3266, "step": 1381 }, { "epoch": 0.47, "learning_rate": 4.5046749634488844e-05, "loss": 0.3463, "step": 1382 }, { "epoch": 0.47, "learning_rate": 4.5038406822105935e-05, "loss": 0.39, "step": 1383 }, { "epoch": 0.47, "learning_rate": 4.5030057763620724e-05, "loss": 0.3505, "step": 1384 }, { "epoch": 0.47, "learning_rate": 4.502170246163566e-05, "loss": 0.4029, "step": 1385 }, { "epoch": 0.47, "learning_rate": 4.501334091875515e-05, "loss": 0.3718, "step": 1386 }, { "epoch": 0.47, "learning_rate": 4.500497313758555e-05, "loss": 0.3494, "step": 1387 }, { "epoch": 0.47, "learning_rate": 4.4996599120735134e-05, "loss": 0.3477, "step": 1388 }, { "epoch": 0.47, "learning_rate": 4.498821887081415e-05, "loss": 0.3732, "step": 1389 }, { "epoch": 0.47, "learning_rate": 4.4979832390434775e-05, "loss": 0.3492, "step": 1390 }, { "epoch": 0.47, "learning_rate": 4.4971439682211125e-05, "loss": 0.3297, "step": 1391 }, { "epoch": 0.47, "learning_rate": 4.496304074875926e-05, "loss": 0.3744, "step": 1392 }, { "epoch": 0.48, "learning_rate": 4.495463559269719e-05, "loss": 0.3399, "step": 1393 }, { "epoch": 0.48, "learning_rate": 4.4946224216644864e-05, "loss": 0.3623, "step": 1394 }, { "epoch": 0.48, "learning_rate": 4.493780662322414e-05, "loss": 0.3963, "step": 1395 }, { "epoch": 0.48, "learning_rate": 4.4929382815058864e-05, "loss": 0.3138, "step": 1396 }, { "epoch": 0.48, "learning_rate": 4.4920952794774776e-05, "loss": 0.3846, "step": 1397 }, { "epoch": 0.48, "learning_rate": 4.491251656499956e-05, "loss": 0.392, "step": 1398 }, { "epoch": 0.48, "learning_rate": 4.490407412836286e-05, "loss": 0.3155, "step": 1399 }, { "epoch": 0.48, "learning_rate": 4.489562548749624e-05, "loss": 0.3732, "step": 1400 }, { "epoch": 0.48, "learning_rate": 4.4887170645033185e-05, "loss": 0.348, "step": 1401 }, { "epoch": 0.48, "learning_rate": 4.487870960360913e-05, "loss": 0.3664, "step": 1402 }, { "epoch": 0.48, "learning_rate": 4.487024236586143e-05, "loss": 0.3491, "step": 1403 }, { "epoch": 0.48, "learning_rate": 4.486176893442939e-05, "loss": 0.3589, "step": 1404 }, { "epoch": 0.48, "learning_rate": 4.485328931195423e-05, "loss": 0.3259, "step": 1405 }, { "epoch": 0.48, "learning_rate": 4.48448035010791e-05, "loss": 0.3843, "step": 1406 }, { "epoch": 0.48, "learning_rate": 4.483631150444909e-05, "loss": 0.388, "step": 1407 }, { "epoch": 0.48, "learning_rate": 4.48278133247112e-05, "loss": 0.3444, "step": 1408 }, { "epoch": 0.48, "learning_rate": 4.481930896451437e-05, "loss": 0.348, "step": 1409 }, { "epoch": 0.48, "learning_rate": 4.4810798426509484e-05, "loss": 0.3238, "step": 1410 }, { "epoch": 0.48, "learning_rate": 4.4802281713349306e-05, "loss": 0.389, "step": 1411 }, { "epoch": 0.48, "learning_rate": 4.479375882768856e-05, "loss": 0.3862, "step": 1412 }, { "epoch": 0.48, "learning_rate": 4.478522977218389e-05, "loss": 0.342, "step": 1413 }, { "epoch": 0.48, "learning_rate": 4.477669454949386e-05, "loss": 0.3816, "step": 1414 }, { "epoch": 0.48, "learning_rate": 4.4768153162278944e-05, "loss": 0.3251, "step": 1415 }, { "epoch": 0.48, "learning_rate": 4.475960561320156e-05, "loss": 0.3605, "step": 1416 }, { "epoch": 0.48, "learning_rate": 4.475105190492602e-05, "loss": 0.3333, "step": 1417 }, { "epoch": 0.48, "learning_rate": 4.474249204011859e-05, "loss": 0.3305, "step": 1418 }, { "epoch": 0.48, "learning_rate": 4.4733926021447416e-05, "loss": 0.3462, "step": 1419 }, { "epoch": 0.48, "learning_rate": 4.4725353851582593e-05, "loss": 0.3989, "step": 1420 }, { "epoch": 0.48, "learning_rate": 4.471677553319613e-05, "loss": 0.3584, "step": 1421 }, { "epoch": 0.49, "learning_rate": 4.470819106896193e-05, "loss": 0.3686, "step": 1422 }, { "epoch": 0.49, "learning_rate": 4.469960046155584e-05, "loss": 0.3344, "step": 1423 }, { "epoch": 0.49, "learning_rate": 4.4691003713655586e-05, "loss": 0.3808, "step": 1424 }, { "epoch": 0.49, "learning_rate": 4.468240082794086e-05, "loss": 0.3406, "step": 1425 }, { "epoch": 0.49, "learning_rate": 4.467379180709321e-05, "loss": 0.3408, "step": 1426 }, { "epoch": 0.49, "learning_rate": 4.466517665379615e-05, "loss": 0.3473, "step": 1427 }, { "epoch": 0.49, "learning_rate": 4.4656555370735065e-05, "loss": 0.3398, "step": 1428 }, { "epoch": 0.49, "learning_rate": 4.464792796059726e-05, "loss": 0.3811, "step": 1429 }, { "epoch": 0.49, "learning_rate": 4.463929442607197e-05, "loss": 0.3369, "step": 1430 }, { "epoch": 0.49, "learning_rate": 4.46306547698503e-05, "loss": 0.3154, "step": 1431 }, { "epoch": 0.49, "learning_rate": 4.462200899462532e-05, "loss": 0.3425, "step": 1432 }, { "epoch": 0.49, "learning_rate": 4.461335710309195e-05, "loss": 0.3457, "step": 1433 }, { "epoch": 0.49, "learning_rate": 4.4604699097947055e-05, "loss": 0.3602, "step": 1434 }, { "epoch": 0.49, "learning_rate": 4.459603498188938e-05, "loss": 0.3298, "step": 1435 }, { "epoch": 0.49, "learning_rate": 4.45873647576196e-05, "loss": 0.5398, "step": 1436 }, { "epoch": 0.49, "learning_rate": 4.4578688427840266e-05, "loss": 0.3266, "step": 1437 }, { "epoch": 0.49, "learning_rate": 4.457000599525586e-05, "loss": 0.3363, "step": 1438 }, { "epoch": 0.49, "learning_rate": 4.456131746257274e-05, "loss": 0.3668, "step": 1439 }, { "epoch": 0.49, "learning_rate": 4.455262283249919e-05, "loss": 0.3462, "step": 1440 }, { "epoch": 0.49, "learning_rate": 4.4543922107745375e-05, "loss": 0.3685, "step": 1441 }, { "epoch": 0.49, "learning_rate": 4.453521529102337e-05, "loss": 0.3295, "step": 1442 }, { "epoch": 0.49, "learning_rate": 4.4526502385047144e-05, "loss": 0.3249, "step": 1443 }, { "epoch": 0.49, "learning_rate": 4.4517783392532574e-05, "loss": 0.3957, "step": 1444 }, { "epoch": 0.49, "learning_rate": 4.450905831619743e-05, "loss": 0.3979, "step": 1445 }, { "epoch": 0.49, "learning_rate": 4.450032715876136e-05, "loss": 0.3448, "step": 1446 }, { "epoch": 0.49, "learning_rate": 4.449158992294593e-05, "loss": 0.3681, "step": 1447 }, { "epoch": 0.49, "learning_rate": 4.44828466114746e-05, "loss": 0.3927, "step": 1448 }, { "epoch": 0.49, "learning_rate": 4.4474097227072715e-05, "loss": 0.352, "step": 1449 }, { "epoch": 0.49, "learning_rate": 4.446534177246751e-05, "loss": 0.3488, "step": 1450 }, { "epoch": 0.5, "learning_rate": 4.445658025038811e-05, "loss": 0.3408, "step": 1451 }, { "epoch": 0.5, "learning_rate": 4.444781266356556e-05, "loss": 0.3645, "step": 1452 }, { "epoch": 0.5, "learning_rate": 4.4439039014732755e-05, "loss": 0.3847, "step": 1453 }, { "epoch": 0.5, "learning_rate": 4.443025930662451e-05, "loss": 0.3667, "step": 1454 }, { "epoch": 0.5, "learning_rate": 4.4421473541977515e-05, "loss": 0.3276, "step": 1455 }, { "epoch": 0.5, "learning_rate": 4.441268172353035e-05, "loss": 0.3521, "step": 1456 }, { "epoch": 0.5, "learning_rate": 4.440388385402347e-05, "loss": 0.3365, "step": 1457 }, { "epoch": 0.5, "learning_rate": 4.4395079936199245e-05, "loss": 0.3147, "step": 1458 }, { "epoch": 0.5, "learning_rate": 4.438626997280191e-05, "loss": 0.3224, "step": 1459 }, { "epoch": 0.5, "learning_rate": 4.4377453966577585e-05, "loss": 0.3394, "step": 1460 }, { "epoch": 0.5, "learning_rate": 4.436863192027428e-05, "loss": 0.3494, "step": 1461 }, { "epoch": 0.5, "learning_rate": 4.4359803836641876e-05, "loss": 0.3532, "step": 1462 }, { "epoch": 0.5, "learning_rate": 4.4350969718432154e-05, "loss": 0.3713, "step": 1463 }, { "epoch": 0.5, "learning_rate": 4.434212956839877e-05, "loss": 0.4061, "step": 1464 }, { "epoch": 0.5, "learning_rate": 4.433328338929724e-05, "loss": 0.3462, "step": 1465 }, { "epoch": 0.5, "learning_rate": 4.432443118388499e-05, "loss": 0.3584, "step": 1466 }, { "epoch": 0.5, "learning_rate": 4.4315572954921306e-05, "loss": 0.3691, "step": 1467 }, { "epoch": 0.5, "learning_rate": 4.430670870516735e-05, "loss": 0.3593, "step": 1468 }, { "epoch": 0.5, "learning_rate": 4.4297838437386176e-05, "loss": 0.3886, "step": 1469 }, { "epoch": 0.5, "learning_rate": 4.42889621543427e-05, "loss": 0.3357, "step": 1470 }, { "epoch": 0.5, "learning_rate": 4.4280079858803716e-05, "loss": 0.3475, "step": 1471 }, { "epoch": 0.5, "learning_rate": 4.42711915535379e-05, "loss": 0.3437, "step": 1472 }, { "epoch": 0.5, "learning_rate": 4.426229724131579e-05, "loss": 0.3728, "step": 1473 }, { "epoch": 0.5, "learning_rate": 4.4253396924909795e-05, "loss": 0.3177, "step": 1474 }, { "epoch": 0.5, "learning_rate": 4.4244490607094215e-05, "loss": 0.394, "step": 1475 }, { "epoch": 0.5, "learning_rate": 4.4235578290645194e-05, "loss": 0.377, "step": 1476 }, { "epoch": 0.5, "learning_rate": 4.422665997834077e-05, "loss": 0.3536, "step": 1477 }, { "epoch": 0.5, "learning_rate": 4.421773567296084e-05, "loss": 0.4041, "step": 1478 }, { "epoch": 0.5, "learning_rate": 4.420880537728715e-05, "loss": 0.3707, "step": 1479 }, { "epoch": 0.5, "learning_rate": 4.419986909410335e-05, "loss": 0.3472, "step": 1480 }, { "epoch": 0.51, "learning_rate": 4.4190926826194936e-05, "loss": 0.3849, "step": 1481 }, { "epoch": 0.51, "learning_rate": 4.418197857634927e-05, "loss": 0.2987, "step": 1482 }, { "epoch": 0.51, "learning_rate": 4.417302434735557e-05, "loss": 0.3686, "step": 1483 }, { "epoch": 0.51, "learning_rate": 4.4164064142004946e-05, "loss": 0.3504, "step": 1484 }, { "epoch": 0.51, "learning_rate": 4.415509796309033e-05, "loss": 0.2969, "step": 1485 }, { "epoch": 0.51, "learning_rate": 4.4146125813406555e-05, "loss": 0.3268, "step": 1486 }, { "epoch": 0.51, "learning_rate": 4.4137147695750305e-05, "loss": 0.3653, "step": 1487 }, { "epoch": 0.51, "learning_rate": 4.4128163612920096e-05, "loss": 0.3559, "step": 1488 }, { "epoch": 0.51, "learning_rate": 4.4119173567716334e-05, "loss": 0.4042, "step": 1489 }, { "epoch": 0.51, "learning_rate": 4.411017756294128e-05, "loss": 0.3758, "step": 1490 }, { "epoch": 0.51, "learning_rate": 4.4101175601399046e-05, "loss": 0.3732, "step": 1491 }, { "epoch": 0.51, "learning_rate": 4.40921676858956e-05, "loss": 0.3399, "step": 1492 }, { "epoch": 0.51, "learning_rate": 4.408315381923878e-05, "loss": 0.3145, "step": 1493 }, { "epoch": 0.51, "learning_rate": 4.4074134004238246e-05, "loss": 0.2858, "step": 1494 }, { "epoch": 0.51, "learning_rate": 4.406510824370554e-05, "loss": 0.3538, "step": 1495 }, { "epoch": 0.51, "learning_rate": 4.405607654045406e-05, "loss": 0.3745, "step": 1496 }, { "epoch": 0.51, "learning_rate": 4.4047038897299044e-05, "loss": 0.3325, "step": 1497 }, { "epoch": 0.51, "learning_rate": 4.403799531705757e-05, "loss": 0.3501, "step": 1498 }, { "epoch": 0.51, "learning_rate": 4.4028945802548604e-05, "loss": 0.4226, "step": 1499 }, { "epoch": 0.51, "learning_rate": 4.401989035659292e-05, "loss": 0.3657, "step": 1500 }, { "epoch": 0.51, "learning_rate": 4.4010828982013176e-05, "loss": 0.3823, "step": 1501 }, { "epoch": 0.51, "learning_rate": 4.400176168163384e-05, "loss": 0.3345, "step": 1502 }, { "epoch": 0.51, "learning_rate": 4.399268845828127e-05, "loss": 0.3628, "step": 1503 }, { "epoch": 0.51, "learning_rate": 4.3983609314783635e-05, "loss": 0.4098, "step": 1504 }, { "epoch": 0.51, "learning_rate": 4.3974524253970974e-05, "loss": 0.3427, "step": 1505 }, { "epoch": 0.51, "learning_rate": 4.396543327867515e-05, "loss": 0.3921, "step": 1506 }, { "epoch": 0.51, "learning_rate": 4.395633639172988e-05, "loss": 0.385, "step": 1507 }, { "epoch": 0.51, "learning_rate": 4.394723359597074e-05, "loss": 0.342, "step": 1508 }, { "epoch": 0.51, "learning_rate": 4.39381248942351e-05, "loss": 0.3198, "step": 1509 }, { "epoch": 0.52, "learning_rate": 4.392901028936223e-05, "loss": 0.3442, "step": 1510 }, { "epoch": 0.52, "learning_rate": 4.39198897841932e-05, "loss": 0.3344, "step": 1511 }, { "epoch": 0.52, "learning_rate": 4.391076338157093e-05, "loss": 0.3615, "step": 1512 }, { "epoch": 0.52, "learning_rate": 4.390163108434019e-05, "loss": 0.375, "step": 1513 }, { "epoch": 0.52, "learning_rate": 4.3892492895347556e-05, "loss": 0.3486, "step": 1514 }, { "epoch": 0.52, "learning_rate": 4.388334881744149e-05, "loss": 0.3376, "step": 1515 }, { "epoch": 0.52, "learning_rate": 4.387419885347223e-05, "loss": 0.3603, "step": 1516 }, { "epoch": 0.52, "learning_rate": 4.38650430062919e-05, "loss": 0.3299, "step": 1517 }, { "epoch": 0.52, "learning_rate": 4.385588127875443e-05, "loss": 0.3399, "step": 1518 }, { "epoch": 0.52, "learning_rate": 4.38467136737156e-05, "loss": 0.3453, "step": 1519 }, { "epoch": 0.52, "learning_rate": 4.3837540194032986e-05, "loss": 0.3634, "step": 1520 }, { "epoch": 0.52, "learning_rate": 4.3828360842566054e-05, "loss": 0.3433, "step": 1521 }, { "epoch": 0.52, "learning_rate": 4.381917562217605e-05, "loss": 0.3812, "step": 1522 }, { "epoch": 0.52, "learning_rate": 4.3809984535726064e-05, "loss": 0.3225, "step": 1523 }, { "epoch": 0.52, "learning_rate": 4.3800787586081036e-05, "loss": 0.3392, "step": 1524 }, { "epoch": 0.52, "learning_rate": 4.37915847761077e-05, "loss": 0.3199, "step": 1525 }, { "epoch": 0.52, "learning_rate": 4.3782376108674636e-05, "loss": 0.3484, "step": 1526 }, { "epoch": 0.52, "learning_rate": 4.377316158665224e-05, "loss": 0.3253, "step": 1527 }, { "epoch": 0.52, "learning_rate": 4.376394121291275e-05, "loss": 0.4147, "step": 1528 }, { "epoch": 0.52, "learning_rate": 4.37547149903302e-05, "loss": 0.3649, "step": 1529 }, { "epoch": 0.52, "learning_rate": 4.374548292178049e-05, "loss": 0.322, "step": 1530 }, { "epoch": 0.52, "learning_rate": 4.3736245010141295e-05, "loss": 0.311, "step": 1531 }, { "epoch": 0.52, "learning_rate": 4.372700125829213e-05, "loss": 0.3577, "step": 1532 }, { "epoch": 0.52, "learning_rate": 4.371775166911435e-05, "loss": 0.3841, "step": 1533 }, { "epoch": 0.52, "learning_rate": 4.3708496245491106e-05, "loss": 0.3583, "step": 1534 }, { "epoch": 0.52, "learning_rate": 4.369923499030737e-05, "loss": 0.3334, "step": 1535 }, { "epoch": 0.52, "learning_rate": 4.368996790644992e-05, "loss": 0.3593, "step": 1536 }, { "epoch": 0.52, "learning_rate": 4.36806949968074e-05, "loss": 0.3276, "step": 1537 }, { "epoch": 0.52, "learning_rate": 4.367141626427022e-05, "loss": 0.3725, "step": 1538 }, { "epoch": 0.53, "learning_rate": 4.3662131711730616e-05, "loss": 0.3489, "step": 1539 }, { "epoch": 0.53, "learning_rate": 4.3652841342082654e-05, "loss": 0.3813, "step": 1540 }, { "epoch": 0.53, "learning_rate": 4.3643545158222197e-05, "loss": 0.3584, "step": 1541 }, { "epoch": 0.53, "learning_rate": 4.363424316304693e-05, "loss": 0.3035, "step": 1542 }, { "epoch": 0.53, "learning_rate": 4.362493535945634e-05, "loss": 0.3461, "step": 1543 }, { "epoch": 0.53, "learning_rate": 4.361562175035173e-05, "loss": 0.3911, "step": 1544 }, { "epoch": 0.53, "learning_rate": 4.360630233863623e-05, "loss": 0.429, "step": 1545 }, { "epoch": 0.53, "learning_rate": 4.359697712721473e-05, "loss": 0.3445, "step": 1546 }, { "epoch": 0.53, "learning_rate": 4.358764611899399e-05, "loss": 0.3465, "step": 1547 }, { "epoch": 0.53, "learning_rate": 4.357830931688253e-05, "loss": 0.3277, "step": 1548 }, { "epoch": 0.53, "learning_rate": 4.3568966723790686e-05, "loss": 0.3615, "step": 1549 }, { "epoch": 0.53, "learning_rate": 4.3559618342630626e-05, "loss": 0.3543, "step": 1550 }, { "epoch": 0.53, "learning_rate": 4.355026417631629e-05, "loss": 0.3262, "step": 1551 }, { "epoch": 0.53, "learning_rate": 4.3540904227763425e-05, "loss": 0.3738, "step": 1552 }, { "epoch": 0.53, "learning_rate": 4.3531538499889605e-05, "loss": 0.396, "step": 1553 }, { "epoch": 0.53, "learning_rate": 4.352216699561417e-05, "loss": 0.3117, "step": 1554 }, { "epoch": 0.53, "learning_rate": 4.35127897178583e-05, "loss": 0.3343, "step": 1555 }, { "epoch": 0.53, "learning_rate": 4.3503406669544936e-05, "loss": 0.3716, "step": 1556 }, { "epoch": 0.53, "learning_rate": 4.349401785359884e-05, "loss": 0.3365, "step": 1557 }, { "epoch": 0.53, "learning_rate": 4.348462327294657e-05, "loss": 0.3893, "step": 1558 }, { "epoch": 0.53, "learning_rate": 4.347522293051648e-05, "loss": 0.3779, "step": 1559 }, { "epoch": 0.53, "learning_rate": 4.346581682923871e-05, "loss": 0.3223, "step": 1560 }, { "epoch": 0.53, "learning_rate": 4.3456404972045216e-05, "loss": 0.4029, "step": 1561 }, { "epoch": 0.53, "learning_rate": 4.344698736186973e-05, "loss": 0.3546, "step": 1562 }, { "epoch": 0.53, "learning_rate": 4.343756400164777e-05, "loss": 0.4174, "step": 1563 }, { "epoch": 0.53, "learning_rate": 4.3428134894316676e-05, "loss": 0.3611, "step": 1564 }, { "epoch": 0.53, "learning_rate": 4.3418700042815555e-05, "loss": 0.345, "step": 1565 }, { "epoch": 0.53, "learning_rate": 4.3409259450085315e-05, "loss": 0.3724, "step": 1566 }, { "epoch": 0.53, "learning_rate": 4.339981311906865e-05, "loss": 0.3582, "step": 1567 }, { "epoch": 0.53, "learning_rate": 4.339036105271004e-05, "loss": 0.3125, "step": 1568 }, { "epoch": 0.54, "learning_rate": 4.338090325395575e-05, "loss": 0.3446, "step": 1569 }, { "epoch": 0.54, "learning_rate": 4.337143972575385e-05, "loss": 0.3919, "step": 1570 }, { "epoch": 0.54, "learning_rate": 4.336197047105418e-05, "loss": 0.3197, "step": 1571 }, { "epoch": 0.54, "learning_rate": 4.3352495492808365e-05, "loss": 0.3715, "step": 1572 }, { "epoch": 0.54, "learning_rate": 4.3343014793969816e-05, "loss": 0.3466, "step": 1573 }, { "epoch": 0.54, "learning_rate": 4.3333528377493736e-05, "loss": 0.3692, "step": 1574 }, { "epoch": 0.54, "learning_rate": 4.3324036246337096e-05, "loss": 0.353, "step": 1575 }, { "epoch": 0.54, "learning_rate": 4.331453840345867e-05, "loss": 0.3253, "step": 1576 }, { "epoch": 0.54, "learning_rate": 4.3305034851818974e-05, "loss": 0.3605, "step": 1577 }, { "epoch": 0.54, "learning_rate": 4.329552559438034e-05, "loss": 0.3113, "step": 1578 }, { "epoch": 0.54, "learning_rate": 4.3286010634106875e-05, "loss": 0.2882, "step": 1579 }, { "epoch": 0.54, "learning_rate": 4.3276489973964434e-05, "loss": 0.3358, "step": 1580 }, { "epoch": 0.54, "learning_rate": 4.326696361692069e-05, "loss": 0.3528, "step": 1581 }, { "epoch": 0.54, "learning_rate": 4.3257431565945045e-05, "loss": 0.3269, "step": 1582 }, { "epoch": 0.54, "learning_rate": 4.3247893824008725e-05, "loss": 0.3309, "step": 1583 }, { "epoch": 0.54, "learning_rate": 4.32383503940847e-05, "loss": 0.3722, "step": 1584 }, { "epoch": 0.54, "learning_rate": 4.3228801279147715e-05, "loss": 0.4254, "step": 1585 }, { "epoch": 0.54, "learning_rate": 4.321924648217429e-05, "loss": 0.3217, "step": 1586 }, { "epoch": 0.54, "learning_rate": 4.3209686006142726e-05, "loss": 0.3389, "step": 1587 }, { "epoch": 0.54, "learning_rate": 4.3200119854033084e-05, "loss": 0.3284, "step": 1588 }, { "epoch": 0.54, "learning_rate": 4.3190548028827185e-05, "loss": 0.323, "step": 1589 }, { "epoch": 0.54, "learning_rate": 4.3180970533508645e-05, "loss": 0.3196, "step": 1590 }, { "epoch": 0.54, "learning_rate": 4.317138737106282e-05, "loss": 0.3439, "step": 1591 }, { "epoch": 0.54, "learning_rate": 4.316179854447685e-05, "loss": 0.3586, "step": 1592 }, { "epoch": 0.54, "learning_rate": 4.315220405673964e-05, "loss": 0.3157, "step": 1593 }, { "epoch": 0.54, "learning_rate": 4.314260391084184e-05, "loss": 0.3283, "step": 1594 }, { "epoch": 0.54, "learning_rate": 4.313299810977589e-05, "loss": 0.3382, "step": 1595 }, { "epoch": 0.54, "learning_rate": 4.312338665653598e-05, "loss": 0.363, "step": 1596 }, { "epoch": 0.54, "learning_rate": 4.311376955411806e-05, "loss": 0.4139, "step": 1597 }, { "epoch": 0.55, "learning_rate": 4.310414680551985e-05, "loss": 0.3327, "step": 1598 }, { "epoch": 0.55, "learning_rate": 4.30945184137408e-05, "loss": 0.34, "step": 1599 }, { "epoch": 0.55, "learning_rate": 4.308488438178218e-05, "loss": 0.3818, "step": 1600 }, { "epoch": 0.55, "learning_rate": 4.307524471264695e-05, "loss": 0.357, "step": 1601 }, { "epoch": 0.55, "learning_rate": 4.3065599409339867e-05, "loss": 0.3788, "step": 1602 }, { "epoch": 0.55, "learning_rate": 4.3055948474867446e-05, "loss": 0.3873, "step": 1603 }, { "epoch": 0.55, "learning_rate": 4.304629191223793e-05, "loss": 0.3432, "step": 1604 }, { "epoch": 0.55, "learning_rate": 4.3036629724461344e-05, "loss": 0.3434, "step": 1605 }, { "epoch": 0.55, "learning_rate": 4.302696191454945e-05, "loss": 0.3318, "step": 1606 }, { "epoch": 0.55, "learning_rate": 4.301728848551576e-05, "loss": 0.3389, "step": 1607 }, { "epoch": 0.55, "learning_rate": 4.300760944037556e-05, "loss": 0.3557, "step": 1608 }, { "epoch": 0.55, "learning_rate": 4.299792478214587e-05, "loss": 0.2988, "step": 1609 }, { "epoch": 0.55, "learning_rate": 4.298823451384545e-05, "loss": 0.3513, "step": 1610 }, { "epoch": 0.55, "learning_rate": 4.2978538638494824e-05, "loss": 0.3569, "step": 1611 }, { "epoch": 0.55, "learning_rate": 4.296883715911627e-05, "loss": 0.3475, "step": 1612 }, { "epoch": 0.55, "learning_rate": 4.295913007873379e-05, "loss": 0.3861, "step": 1613 }, { "epoch": 0.55, "learning_rate": 4.294941740037315e-05, "loss": 0.359, "step": 1614 }, { "epoch": 0.55, "learning_rate": 4.293969912706185e-05, "loss": 0.3674, "step": 1615 }, { "epoch": 0.55, "learning_rate": 4.292997526182915e-05, "loss": 0.3195, "step": 1616 }, { "epoch": 0.55, "learning_rate": 4.292024580770603e-05, "loss": 0.3315, "step": 1617 }, { "epoch": 0.55, "learning_rate": 4.2910510767725235e-05, "loss": 0.4013, "step": 1618 }, { "epoch": 0.55, "learning_rate": 4.2900770144921224e-05, "loss": 0.3449, "step": 1619 }, { "epoch": 0.55, "learning_rate": 4.2891023942330235e-05, "loss": 0.3796, "step": 1620 }, { "epoch": 0.55, "learning_rate": 4.28812721629902e-05, "loss": 0.3773, "step": 1621 }, { "epoch": 0.55, "learning_rate": 4.287151480994084e-05, "loss": 0.3421, "step": 1622 }, { "epoch": 0.55, "learning_rate": 4.2861751886223565e-05, "loss": 0.3517, "step": 1623 }, { "epoch": 0.55, "learning_rate": 4.2851983394881545e-05, "loss": 0.3237, "step": 1624 }, { "epoch": 0.55, "learning_rate": 4.2842209338959694e-05, "loss": 0.3634, "step": 1625 }, { "epoch": 0.55, "learning_rate": 4.283242972150463e-05, "loss": 0.321, "step": 1626 }, { "epoch": 0.56, "learning_rate": 4.282264454556473e-05, "loss": 0.3727, "step": 1627 }, { "epoch": 0.56, "learning_rate": 4.2812853814190115e-05, "loss": 0.3283, "step": 1628 }, { "epoch": 0.56, "learning_rate": 4.28030575304326e-05, "loss": 0.3508, "step": 1629 }, { "epoch": 0.56, "learning_rate": 4.2793255697345745e-05, "loss": 0.378, "step": 1630 }, { "epoch": 0.56, "learning_rate": 4.2783448317984875e-05, "loss": 0.3429, "step": 1631 }, { "epoch": 0.56, "learning_rate": 4.277363539540698e-05, "loss": 0.3166, "step": 1632 }, { "epoch": 0.56, "learning_rate": 4.276381693267084e-05, "loss": 0.3204, "step": 1633 }, { "epoch": 0.56, "learning_rate": 4.275399293283692e-05, "loss": 0.3975, "step": 1634 }, { "epoch": 0.56, "learning_rate": 4.274416339896742e-05, "loss": 0.3335, "step": 1635 }, { "epoch": 0.56, "learning_rate": 4.2734328334126286e-05, "loss": 0.3635, "step": 1636 }, { "epoch": 0.56, "learning_rate": 4.272448774137916e-05, "loss": 0.3529, "step": 1637 }, { "epoch": 0.56, "learning_rate": 4.271464162379342e-05, "loss": 0.3408, "step": 1638 }, { "epoch": 0.56, "learning_rate": 4.270478998443817e-05, "loss": 0.352, "step": 1639 }, { "epoch": 0.56, "learning_rate": 4.2694932826384226e-05, "loss": 0.3793, "step": 1640 }, { "epoch": 0.56, "learning_rate": 4.268507015270413e-05, "loss": 0.313, "step": 1641 }, { "epoch": 0.56, "learning_rate": 4.2675201966472145e-05, "loss": 0.3513, "step": 1642 }, { "epoch": 0.56, "learning_rate": 4.2665328270764235e-05, "loss": 0.3324, "step": 1643 }, { "epoch": 0.56, "learning_rate": 4.265544906865812e-05, "loss": 0.3867, "step": 1644 }, { "epoch": 0.56, "learning_rate": 4.264556436323319e-05, "loss": 0.3431, "step": 1645 }, { "epoch": 0.56, "learning_rate": 4.263567415757058e-05, "loss": 0.3895, "step": 1646 }, { "epoch": 0.56, "learning_rate": 4.262577845475314e-05, "loss": 0.33, "step": 1647 }, { "epoch": 0.56, "learning_rate": 4.2615877257865415e-05, "loss": 0.3657, "step": 1648 }, { "epoch": 0.56, "learning_rate": 4.260597056999367e-05, "loss": 0.3531, "step": 1649 }, { "epoch": 0.56, "learning_rate": 4.2596058394225893e-05, "loss": 0.384, "step": 1650 }, { "epoch": 0.56, "learning_rate": 4.258614073365177e-05, "loss": 0.3438, "step": 1651 }, { "epoch": 0.56, "learning_rate": 4.257621759136271e-05, "loss": 0.3454, "step": 1652 }, { "epoch": 0.56, "learning_rate": 4.2566288970451794e-05, "loss": 0.3907, "step": 1653 }, { "epoch": 0.56, "learning_rate": 4.255635487401387e-05, "loss": 0.3386, "step": 1654 }, { "epoch": 0.56, "learning_rate": 4.254641530514544e-05, "loss": 0.345, "step": 1655 }, { "epoch": 0.56, "learning_rate": 4.253647026694475e-05, "loss": 0.357, "step": 1656 }, { "epoch": 0.57, "learning_rate": 4.2526519762511716e-05, "loss": 0.3547, "step": 1657 }, { "epoch": 0.57, "learning_rate": 4.251656379494798e-05, "loss": 0.2969, "step": 1658 }, { "epoch": 0.57, "learning_rate": 4.2506602367356884e-05, "loss": 0.3459, "step": 1659 }, { "epoch": 0.57, "learning_rate": 4.249663548284347e-05, "loss": 0.3326, "step": 1660 }, { "epoch": 0.57, "learning_rate": 4.2486663144514486e-05, "loss": 0.3829, "step": 1661 }, { "epoch": 0.57, "learning_rate": 4.247668535547837e-05, "loss": 0.3645, "step": 1662 }, { "epoch": 0.57, "learning_rate": 4.2466702118845265e-05, "loss": 0.3224, "step": 1663 }, { "epoch": 0.57, "learning_rate": 4.2456713437727e-05, "loss": 0.3465, "step": 1664 }, { "epoch": 0.57, "learning_rate": 4.2446719315237135e-05, "loss": 0.366, "step": 1665 }, { "epoch": 0.57, "learning_rate": 4.2436719754490896e-05, "loss": 0.3734, "step": 1666 }, { "epoch": 0.57, "learning_rate": 4.242671475860519e-05, "loss": 0.3447, "step": 1667 }, { "epoch": 0.57, "learning_rate": 4.2416704330698665e-05, "loss": 0.3176, "step": 1668 }, { "epoch": 0.57, "learning_rate": 4.2406688473891626e-05, "loss": 0.3468, "step": 1669 }, { "epoch": 0.57, "learning_rate": 4.239666719130608e-05, "loss": 0.3515, "step": 1670 }, { "epoch": 0.57, "learning_rate": 4.238664048606573e-05, "loss": 0.3732, "step": 1671 }, { "epoch": 0.57, "learning_rate": 4.237660836129597e-05, "loss": 0.3358, "step": 1672 }, { "epoch": 0.57, "learning_rate": 4.236657082012387e-05, "loss": 0.3472, "step": 1673 }, { "epoch": 0.57, "learning_rate": 4.2356527865678196e-05, "loss": 0.3629, "step": 1674 }, { "epoch": 0.57, "learning_rate": 4.234647950108941e-05, "loss": 0.3888, "step": 1675 }, { "epoch": 0.57, "learning_rate": 4.233642572948966e-05, "loss": 0.3201, "step": 1676 }, { "epoch": 0.57, "learning_rate": 4.232636655401275e-05, "loss": 0.3273, "step": 1677 }, { "epoch": 0.57, "learning_rate": 4.2316301977794203e-05, "loss": 0.3665, "step": 1678 }, { "epoch": 0.57, "learning_rate": 4.2306232003971215e-05, "loss": 0.3501, "step": 1679 }, { "epoch": 0.57, "learning_rate": 4.229615663568266e-05, "loss": 0.3101, "step": 1680 }, { "epoch": 0.57, "learning_rate": 4.22860758760691e-05, "loss": 0.3885, "step": 1681 }, { "epoch": 0.57, "learning_rate": 4.227598972827277e-05, "loss": 0.3322, "step": 1682 }, { "epoch": 0.57, "learning_rate": 4.226589819543758e-05, "loss": 0.3988, "step": 1683 }, { "epoch": 0.57, "learning_rate": 4.2255801280709144e-05, "loss": 0.3371, "step": 1684 }, { "epoch": 0.57, "learning_rate": 4.224569898723473e-05, "loss": 0.3069, "step": 1685 }, { "epoch": 0.58, "learning_rate": 4.223559131816328e-05, "loss": 0.3375, "step": 1686 }, { "epoch": 0.58, "learning_rate": 4.2225478276645423e-05, "loss": 0.2907, "step": 1687 }, { "epoch": 0.58, "learning_rate": 4.221535986583347e-05, "loss": 0.362, "step": 1688 }, { "epoch": 0.58, "learning_rate": 4.220523608888139e-05, "loss": 0.3448, "step": 1689 }, { "epoch": 0.58, "learning_rate": 4.219510694894483e-05, "loss": 0.4547, "step": 1690 }, { "epoch": 0.58, "learning_rate": 4.2184972449181105e-05, "loss": 0.3015, "step": 1691 }, { "epoch": 0.58, "learning_rate": 4.2174832592749214e-05, "loss": 0.3628, "step": 1692 }, { "epoch": 0.58, "learning_rate": 4.216468738280981e-05, "loss": 0.3431, "step": 1693 }, { "epoch": 0.58, "learning_rate": 4.215453682252522e-05, "loss": 0.327, "step": 1694 }, { "epoch": 0.58, "learning_rate": 4.2144380915059445e-05, "loss": 0.3749, "step": 1695 }, { "epoch": 0.58, "learning_rate": 4.2134219663578146e-05, "loss": 0.3149, "step": 1696 }, { "epoch": 0.58, "learning_rate": 4.212405307124865e-05, "loss": 0.3633, "step": 1697 }, { "epoch": 0.58, "learning_rate": 4.211388114123995e-05, "loss": 0.3065, "step": 1698 }, { "epoch": 0.58, "learning_rate": 4.21037038767227e-05, "loss": 0.3755, "step": 1699 }, { "epoch": 0.58, "learning_rate": 4.209352128086923e-05, "loss": 0.4098, "step": 1700 }, { "epoch": 0.58, "learning_rate": 4.208333335685352e-05, "loss": 0.3522, "step": 1701 }, { "epoch": 0.58, "learning_rate": 4.207314010785119e-05, "loss": 0.366, "step": 1702 }, { "epoch": 0.58, "learning_rate": 4.206294153703958e-05, "loss": 0.3263, "step": 1703 }, { "epoch": 0.58, "learning_rate": 4.2052737647597616e-05, "loss": 0.3367, "step": 1704 }, { "epoch": 0.58, "learning_rate": 4.204252844270594e-05, "loss": 0.3041, "step": 1705 }, { "epoch": 0.58, "learning_rate": 4.2032313925546826e-05, "loss": 0.3564, "step": 1706 }, { "epoch": 0.58, "learning_rate": 4.2022094099304194e-05, "loss": 0.3432, "step": 1707 }, { "epoch": 0.58, "learning_rate": 4.201186896716363e-05, "loss": 0.3273, "step": 1708 }, { "epoch": 0.58, "learning_rate": 4.200163853231239e-05, "loss": 0.3365, "step": 1709 }, { "epoch": 0.58, "learning_rate": 4.1991402797939356e-05, "loss": 0.3551, "step": 1710 }, { "epoch": 0.58, "learning_rate": 4.1981161767235075e-05, "loss": 0.4036, "step": 1711 }, { "epoch": 0.58, "learning_rate": 4.1970915443391745e-05, "loss": 0.3735, "step": 1712 }, { "epoch": 0.58, "learning_rate": 4.196066382960321e-05, "loss": 0.3368, "step": 1713 }, { "epoch": 0.58, "learning_rate": 4.1950406929064965e-05, "loss": 0.3372, "step": 1714 }, { "epoch": 0.59, "learning_rate": 4.194014474497415e-05, "loss": 0.3303, "step": 1715 }, { "epoch": 0.59, "learning_rate": 4.1929877280529564e-05, "loss": 0.3246, "step": 1716 }, { "epoch": 0.59, "learning_rate": 4.191960453893165e-05, "loss": 0.3366, "step": 1717 }, { "epoch": 0.59, "learning_rate": 4.190932652338245e-05, "loss": 0.3361, "step": 1718 }, { "epoch": 0.59, "learning_rate": 4.189904323708573e-05, "loss": 0.3684, "step": 1719 }, { "epoch": 0.59, "learning_rate": 4.188875468324685e-05, "loss": 0.3263, "step": 1720 }, { "epoch": 0.59, "learning_rate": 4.18784608650728e-05, "loss": 0.3371, "step": 1721 }, { "epoch": 0.59, "learning_rate": 4.186816178577223e-05, "loss": 0.3286, "step": 1722 }, { "epoch": 0.59, "learning_rate": 4.185785744855546e-05, "loss": 0.4653, "step": 1723 }, { "epoch": 0.59, "learning_rate": 4.1847547856634395e-05, "loss": 0.4089, "step": 1724 }, { "epoch": 0.59, "learning_rate": 4.183723301322261e-05, "loss": 0.3597, "step": 1725 }, { "epoch": 0.59, "learning_rate": 4.182691292153531e-05, "loss": 0.3581, "step": 1726 }, { "epoch": 0.59, "learning_rate": 4.181658758478934e-05, "loss": 0.3718, "step": 1727 }, { "epoch": 0.59, "learning_rate": 4.1806257006203157e-05, "loss": 0.3301, "step": 1728 }, { "epoch": 0.59, "learning_rate": 4.1795921188996864e-05, "loss": 0.3489, "step": 1729 }, { "epoch": 0.59, "learning_rate": 4.178558013639224e-05, "loss": 0.3693, "step": 1730 }, { "epoch": 0.59, "learning_rate": 4.177523385161264e-05, "loss": 0.3153, "step": 1731 }, { "epoch": 0.59, "learning_rate": 4.176488233788306e-05, "loss": 0.3489, "step": 1732 }, { "epoch": 0.59, "learning_rate": 4.175452559843014e-05, "loss": 0.3853, "step": 1733 }, { "epoch": 0.59, "learning_rate": 4.174416363648215e-05, "loss": 0.3767, "step": 1734 }, { "epoch": 0.59, "learning_rate": 4.1733796455268966e-05, "loss": 0.3677, "step": 1735 }, { "epoch": 0.59, "learning_rate": 4.172342405802212e-05, "loss": 0.3516, "step": 1736 }, { "epoch": 0.59, "learning_rate": 4.171304644797476e-05, "loss": 0.375, "step": 1737 }, { "epoch": 0.59, "learning_rate": 4.1702663628361636e-05, "loss": 0.3405, "step": 1738 }, { "epoch": 0.59, "learning_rate": 4.169227560241915e-05, "loss": 0.3683, "step": 1739 }, { "epoch": 0.59, "learning_rate": 4.1681882373385326e-05, "loss": 0.4032, "step": 1740 }, { "epoch": 0.59, "learning_rate": 4.167148394449979e-05, "loss": 0.3569, "step": 1741 }, { "epoch": 0.59, "learning_rate": 4.16610803190038e-05, "loss": 0.3325, "step": 1742 }, { "epoch": 0.59, "learning_rate": 4.1650671500140244e-05, "loss": 0.3425, "step": 1743 }, { "epoch": 0.6, "learning_rate": 4.164025749115361e-05, "loss": 0.3125, "step": 1744 }, { "epoch": 0.6, "learning_rate": 4.1629838295290014e-05, "loss": 0.3477, "step": 1745 }, { "epoch": 0.6, "learning_rate": 4.1619413915797195e-05, "loss": 0.4626, "step": 1746 }, { "epoch": 0.6, "learning_rate": 4.160898435592449e-05, "loss": 0.3109, "step": 1747 }, { "epoch": 0.6, "learning_rate": 4.159854961892287e-05, "loss": 0.3455, "step": 1748 }, { "epoch": 0.6, "learning_rate": 4.15881097080449e-05, "loss": 0.3475, "step": 1749 }, { "epoch": 0.6, "learning_rate": 4.157766462654478e-05, "loss": 0.3495, "step": 1750 }, { "epoch": 0.6, "learning_rate": 4.156721437767829e-05, "loss": 0.3629, "step": 1751 }, { "epoch": 0.6, "learning_rate": 4.1556758964702866e-05, "loss": 0.3568, "step": 1752 }, { "epoch": 0.6, "learning_rate": 4.154629839087752e-05, "loss": 0.3337, "step": 1753 }, { "epoch": 0.6, "learning_rate": 4.153583265946288e-05, "loss": 0.3206, "step": 1754 }, { "epoch": 0.6, "learning_rate": 4.1525361773721176e-05, "loss": 0.3544, "step": 1755 }, { "epoch": 0.6, "learning_rate": 4.151488573691626e-05, "loss": 0.3701, "step": 1756 }, { "epoch": 0.6, "learning_rate": 4.150440455231357e-05, "loss": 0.4084, "step": 1757 }, { "epoch": 0.6, "learning_rate": 4.1493918223180187e-05, "loss": 0.548, "step": 1758 }, { "epoch": 0.6, "learning_rate": 4.148342675278474e-05, "loss": 0.3565, "step": 1759 }, { "epoch": 0.6, "learning_rate": 4.147293014439749e-05, "loss": 0.337, "step": 1760 }, { "epoch": 0.6, "learning_rate": 4.146242840129031e-05, "loss": 0.2923, "step": 1761 }, { "epoch": 0.6, "learning_rate": 4.145192152673667e-05, "loss": 0.3551, "step": 1762 }, { "epoch": 0.6, "learning_rate": 4.144140952401161e-05, "loss": 0.3446, "step": 1763 }, { "epoch": 0.6, "learning_rate": 4.14308923963918e-05, "loss": 0.3477, "step": 1764 }, { "epoch": 0.6, "learning_rate": 4.142037014715549e-05, "loss": 0.3604, "step": 1765 }, { "epoch": 0.6, "learning_rate": 4.140984277958256e-05, "loss": 0.3753, "step": 1766 }, { "epoch": 0.6, "learning_rate": 4.139931029695443e-05, "loss": 0.3256, "step": 1767 }, { "epoch": 0.6, "learning_rate": 4.1388772702554154e-05, "loss": 0.3353, "step": 1768 }, { "epoch": 0.6, "learning_rate": 4.1378229999666376e-05, "loss": 0.3429, "step": 1769 }, { "epoch": 0.6, "learning_rate": 4.136768219157731e-05, "loss": 0.3389, "step": 1770 }, { "epoch": 0.6, "learning_rate": 4.135712928157478e-05, "loss": 0.3319, "step": 1771 }, { "epoch": 0.6, "learning_rate": 4.134657127294821e-05, "loss": 0.3258, "step": 1772 }, { "epoch": 0.6, "learning_rate": 4.1336008168988596e-05, "loss": 0.4302, "step": 1773 }, { "epoch": 0.61, "learning_rate": 4.132543997298851e-05, "loss": 0.3146, "step": 1774 }, { "epoch": 0.61, "learning_rate": 4.131486668824215e-05, "loss": 0.3126, "step": 1775 }, { "epoch": 0.61, "learning_rate": 4.130428831804527e-05, "loss": 0.3476, "step": 1776 }, { "epoch": 0.61, "learning_rate": 4.129370486569521e-05, "loss": 0.3492, "step": 1777 }, { "epoch": 0.61, "learning_rate": 4.1283116334490914e-05, "loss": 0.3163, "step": 1778 }, { "epoch": 0.61, "learning_rate": 4.127252272773288e-05, "loss": 0.3532, "step": 1779 }, { "epoch": 0.61, "learning_rate": 4.126192404872322e-05, "loss": 0.3182, "step": 1780 }, { "epoch": 0.61, "learning_rate": 4.12513203007656e-05, "loss": 0.3743, "step": 1781 }, { "epoch": 0.61, "learning_rate": 4.1240711487165284e-05, "loss": 0.3423, "step": 1782 }, { "epoch": 0.61, "learning_rate": 4.1230097611229114e-05, "loss": 0.3466, "step": 1783 }, { "epoch": 0.61, "learning_rate": 4.121947867626549e-05, "loss": 0.3435, "step": 1784 }, { "epoch": 0.61, "learning_rate": 4.120885468558441e-05, "loss": 0.409, "step": 1785 }, { "epoch": 0.61, "learning_rate": 4.1198225642497445e-05, "loss": 0.3036, "step": 1786 }, { "epoch": 0.61, "learning_rate": 4.118759155031774e-05, "loss": 0.3846, "step": 1787 }, { "epoch": 0.61, "learning_rate": 4.117695241236e-05, "loss": 0.3111, "step": 1788 }, { "epoch": 0.61, "learning_rate": 4.116630823194051e-05, "loss": 0.3199, "step": 1789 }, { "epoch": 0.61, "learning_rate": 4.1155659012377155e-05, "loss": 0.3575, "step": 1790 }, { "epoch": 0.61, "learning_rate": 4.1145004756989344e-05, "loss": 0.4029, "step": 1791 }, { "epoch": 0.61, "learning_rate": 4.113434546909808e-05, "loss": 0.3862, "step": 1792 }, { "epoch": 0.61, "learning_rate": 4.112368115202595e-05, "loss": 0.353, "step": 1793 }, { "epoch": 0.61, "learning_rate": 4.111301180909707e-05, "loss": 0.3607, "step": 1794 }, { "epoch": 0.61, "learning_rate": 4.110233744363716e-05, "loss": 0.3355, "step": 1795 }, { "epoch": 0.61, "learning_rate": 4.109165805897347e-05, "loss": 0.3507, "step": 1796 }, { "epoch": 0.61, "learning_rate": 4.108097365843486e-05, "loss": 0.3294, "step": 1797 }, { "epoch": 0.61, "learning_rate": 4.107028424535171e-05, "loss": 0.3349, "step": 1798 }, { "epoch": 0.61, "learning_rate": 4.105958982305598e-05, "loss": 0.3591, "step": 1799 }, { "epoch": 0.61, "learning_rate": 4.104889039488119e-05, "loss": 0.3801, "step": 1800 }, { "epoch": 0.61, "learning_rate": 4.1038185964162434e-05, "loss": 0.3464, "step": 1801 }, { "epoch": 0.61, "learning_rate": 4.1027476534236345e-05, "loss": 0.3489, "step": 1802 }, { "epoch": 0.62, "learning_rate": 4.101676210844113e-05, "loss": 0.3842, "step": 1803 }, { "epoch": 0.62, "learning_rate": 4.100604269011653e-05, "loss": 0.2983, "step": 1804 }, { "epoch": 0.62, "learning_rate": 4.0995318282603855e-05, "loss": 0.3093, "step": 1805 }, { "epoch": 0.62, "learning_rate": 4.0984588889246e-05, "loss": 0.3336, "step": 1806 }, { "epoch": 0.62, "learning_rate": 4.0973854513387366e-05, "loss": 0.3486, "step": 1807 }, { "epoch": 0.62, "learning_rate": 4.096311515837393e-05, "loss": 0.3464, "step": 1808 }, { "epoch": 0.62, "learning_rate": 4.095237082755323e-05, "loss": 0.3302, "step": 1809 }, { "epoch": 0.62, "learning_rate": 4.0941621524274335e-05, "loss": 0.2946, "step": 1810 }, { "epoch": 0.62, "learning_rate": 4.0930867251887874e-05, "loss": 0.3079, "step": 1811 }, { "epoch": 0.62, "learning_rate": 4.092010801374602e-05, "loss": 0.3564, "step": 1812 }, { "epoch": 0.62, "learning_rate": 4.0909343813202494e-05, "loss": 0.3269, "step": 1813 }, { "epoch": 0.62, "learning_rate": 4.0898574653612585e-05, "loss": 0.375, "step": 1814 }, { "epoch": 0.62, "learning_rate": 4.0887800538333096e-05, "loss": 0.3518, "step": 1815 }, { "epoch": 0.62, "learning_rate": 4.087702147072241e-05, "loss": 0.3628, "step": 1816 }, { "epoch": 0.62, "learning_rate": 4.08662374541404e-05, "loss": 0.3218, "step": 1817 }, { "epoch": 0.62, "learning_rate": 4.0855448491948536e-05, "loss": 0.3295, "step": 1818 }, { "epoch": 0.62, "learning_rate": 4.0844654587509804e-05, "loss": 0.3375, "step": 1819 }, { "epoch": 0.62, "learning_rate": 4.083385574418873e-05, "loss": 0.2979, "step": 1820 }, { "epoch": 0.62, "learning_rate": 4.08230519653514e-05, "loss": 0.3226, "step": 1821 }, { "epoch": 0.62, "learning_rate": 4.08122432543654e-05, "loss": 0.366, "step": 1822 }, { "epoch": 0.62, "learning_rate": 4.080142961459989e-05, "loss": 0.338, "step": 1823 }, { "epoch": 0.62, "learning_rate": 4.079061104942555e-05, "loss": 0.3134, "step": 1824 }, { "epoch": 0.62, "learning_rate": 4.077978756221459e-05, "loss": 0.3042, "step": 1825 }, { "epoch": 0.62, "learning_rate": 4.076895915634077e-05, "loss": 0.3507, "step": 1826 }, { "epoch": 0.62, "learning_rate": 4.075812583517937e-05, "loss": 0.3606, "step": 1827 }, { "epoch": 0.62, "learning_rate": 4.074728760210721e-05, "loss": 0.3869, "step": 1828 }, { "epoch": 0.62, "learning_rate": 4.073644446050264e-05, "loss": 0.3239, "step": 1829 }, { "epoch": 0.62, "learning_rate": 4.072559641374554e-05, "loss": 0.3423, "step": 1830 }, { "epoch": 0.62, "learning_rate": 4.07147434652173e-05, "loss": 0.4036, "step": 1831 }, { "epoch": 0.63, "learning_rate": 4.0703885618300876e-05, "loss": 0.3262, "step": 1832 }, { "epoch": 0.63, "learning_rate": 4.069302287638071e-05, "loss": 0.3171, "step": 1833 }, { "epoch": 0.63, "learning_rate": 4.068215524284281e-05, "loss": 0.3347, "step": 1834 }, { "epoch": 0.63, "learning_rate": 4.067128272107468e-05, "loss": 0.3188, "step": 1835 }, { "epoch": 0.63, "learning_rate": 4.0660405314465354e-05, "loss": 0.3305, "step": 1836 }, { "epoch": 0.63, "learning_rate": 4.064952302640539e-05, "loss": 0.3293, "step": 1837 }, { "epoch": 0.63, "learning_rate": 4.0638635860286875e-05, "loss": 0.342, "step": 1838 }, { "epoch": 0.63, "learning_rate": 4.06277438195034e-05, "loss": 0.3251, "step": 1839 }, { "epoch": 0.63, "learning_rate": 4.061684690745009e-05, "loss": 0.3348, "step": 1840 }, { "epoch": 0.63, "learning_rate": 4.060594512752359e-05, "loss": 0.315, "step": 1841 }, { "epoch": 0.63, "learning_rate": 4.059503848312204e-05, "loss": 0.3551, "step": 1842 }, { "epoch": 0.63, "learning_rate": 4.058412697764513e-05, "loss": 0.3225, "step": 1843 }, { "epoch": 0.63, "learning_rate": 4.057321061449404e-05, "loss": 0.3642, "step": 1844 }, { "epoch": 0.63, "learning_rate": 4.056228939707147e-05, "loss": 0.3192, "step": 1845 }, { "epoch": 0.63, "learning_rate": 4.055136332878163e-05, "loss": 0.3988, "step": 1846 }, { "epoch": 0.63, "learning_rate": 4.054043241303026e-05, "loss": 0.3182, "step": 1847 }, { "epoch": 0.63, "learning_rate": 4.052949665322459e-05, "loss": 0.3427, "step": 1848 }, { "epoch": 0.63, "learning_rate": 4.0518556052773366e-05, "loss": 0.288, "step": 1849 }, { "epoch": 0.63, "learning_rate": 4.050761061508685e-05, "loss": 0.3293, "step": 1850 }, { "epoch": 0.63, "learning_rate": 4.0496660343576796e-05, "loss": 0.3673, "step": 1851 }, { "epoch": 0.63, "learning_rate": 4.04857052416565e-05, "loss": 0.3552, "step": 1852 }, { "epoch": 0.63, "learning_rate": 4.04747453127407e-05, "loss": 0.3318, "step": 1853 }, { "epoch": 0.63, "learning_rate": 4.0463780560245713e-05, "loss": 0.3585, "step": 1854 }, { "epoch": 0.63, "learning_rate": 4.0452810987589304e-05, "loss": 0.3517, "step": 1855 }, { "epoch": 0.63, "learning_rate": 4.0441836598190776e-05, "loss": 0.311, "step": 1856 }, { "epoch": 0.63, "learning_rate": 4.04308573954709e-05, "loss": 0.3412, "step": 1857 }, { "epoch": 0.63, "learning_rate": 4.041987338285198e-05, "loss": 0.3599, "step": 1858 }, { "epoch": 0.63, "learning_rate": 4.0408884563757796e-05, "loss": 0.3327, "step": 1859 }, { "epoch": 0.63, "learning_rate": 4.0397890941613636e-05, "loss": 0.3191, "step": 1860 }, { "epoch": 0.63, "learning_rate": 4.0386892519846274e-05, "loss": 0.3756, "step": 1861 }, { "epoch": 0.64, "learning_rate": 4.037588930188402e-05, "loss": 0.3165, "step": 1862 }, { "epoch": 0.64, "learning_rate": 4.036488129115662e-05, "loss": 0.372, "step": 1863 }, { "epoch": 0.64, "learning_rate": 4.035386849109535e-05, "loss": 0.3607, "step": 1864 }, { "epoch": 0.64, "learning_rate": 4.0342850905132976e-05, "loss": 0.3455, "step": 1865 }, { "epoch": 0.64, "learning_rate": 4.0331828536703754e-05, "loss": 0.3349, "step": 1866 }, { "epoch": 0.64, "learning_rate": 4.032080138924341e-05, "loss": 0.3387, "step": 1867 }, { "epoch": 0.64, "learning_rate": 4.030976946618921e-05, "loss": 0.3307, "step": 1868 }, { "epoch": 0.64, "learning_rate": 4.029873277097985e-05, "loss": 0.3365, "step": 1869 }, { "epoch": 0.64, "learning_rate": 4.028769130705555e-05, "loss": 0.3279, "step": 1870 }, { "epoch": 0.64, "learning_rate": 4.0276645077858e-05, "loss": 0.3744, "step": 1871 }, { "epoch": 0.64, "learning_rate": 4.026559408683039e-05, "loss": 0.3119, "step": 1872 }, { "epoch": 0.64, "learning_rate": 4.025453833741738e-05, "loss": 0.3447, "step": 1873 }, { "epoch": 0.64, "learning_rate": 4.024347783306512e-05, "loss": 0.3046, "step": 1874 }, { "epoch": 0.64, "learning_rate": 4.023241257722126e-05, "loss": 0.3477, "step": 1875 }, { "epoch": 0.64, "learning_rate": 4.022134257333487e-05, "loss": 0.3114, "step": 1876 }, { "epoch": 0.64, "learning_rate": 4.021026782485659e-05, "loss": 0.3573, "step": 1877 }, { "epoch": 0.64, "learning_rate": 4.0199188335238456e-05, "loss": 0.3292, "step": 1878 }, { "epoch": 0.64, "learning_rate": 4.018810410793404e-05, "loss": 0.3374, "step": 1879 }, { "epoch": 0.64, "learning_rate": 4.0177015146398355e-05, "loss": 0.3376, "step": 1880 }, { "epoch": 0.64, "learning_rate": 4.0165921454087904e-05, "loss": 0.3274, "step": 1881 }, { "epoch": 0.64, "learning_rate": 4.0154823034460675e-05, "loss": 0.289, "step": 1882 }, { "epoch": 0.64, "learning_rate": 4.014371989097611e-05, "loss": 0.3655, "step": 1883 }, { "epoch": 0.64, "learning_rate": 4.0132612027095116e-05, "loss": 0.37, "step": 1884 }, { "epoch": 0.64, "learning_rate": 4.0121499446280114e-05, "loss": 0.3256, "step": 1885 }, { "epoch": 0.64, "learning_rate": 4.011038215199495e-05, "loss": 0.3564, "step": 1886 }, { "epoch": 0.64, "learning_rate": 4.009926014770496e-05, "loss": 0.3432, "step": 1887 }, { "epoch": 0.64, "learning_rate": 4.008813343687695e-05, "loss": 0.3677, "step": 1888 }, { "epoch": 0.64, "learning_rate": 4.007700202297919e-05, "loss": 0.3168, "step": 1889 }, { "epoch": 0.64, "learning_rate": 4.0065865909481417e-05, "loss": 0.328, "step": 1890 }, { "epoch": 0.65, "learning_rate": 4.005472509985481e-05, "loss": 0.2907, "step": 1891 }, { "epoch": 0.65, "learning_rate": 4.0043579597572054e-05, "loss": 0.3311, "step": 1892 }, { "epoch": 0.65, "learning_rate": 4.003242940610726e-05, "loss": 0.2887, "step": 1893 }, { "epoch": 0.65, "learning_rate": 4.002127452893604e-05, "loss": 0.3408, "step": 1894 }, { "epoch": 0.65, "learning_rate": 4.001011496953541e-05, "loss": 0.3712, "step": 1895 }, { "epoch": 0.65, "learning_rate": 3.999895073138389e-05, "loss": 0.3368, "step": 1896 }, { "epoch": 0.65, "learning_rate": 3.998778181796145e-05, "loss": 0.4231, "step": 1897 }, { "epoch": 0.65, "learning_rate": 3.997660823274952e-05, "loss": 0.3775, "step": 1898 }, { "epoch": 0.65, "learning_rate": 3.996542997923096e-05, "loss": 0.3006, "step": 1899 }, { "epoch": 0.65, "learning_rate": 3.995424706089013e-05, "loss": 0.3339, "step": 1900 }, { "epoch": 0.65, "learning_rate": 3.9943059481212795e-05, "loss": 0.3631, "step": 1901 }, { "epoch": 0.65, "learning_rate": 3.993186724368621e-05, "loss": 0.3271, "step": 1902 }, { "epoch": 0.65, "learning_rate": 3.992067035179906e-05, "loss": 0.3671, "step": 1903 }, { "epoch": 0.65, "learning_rate": 3.9909468809041494e-05, "loss": 0.3464, "step": 1904 }, { "epoch": 0.65, "learning_rate": 3.98982626189051e-05, "loss": 0.3196, "step": 1905 }, { "epoch": 0.65, "learning_rate": 3.988705178488294e-05, "loss": 0.3665, "step": 1906 }, { "epoch": 0.65, "learning_rate": 3.9875836310469485e-05, "loss": 0.3753, "step": 1907 }, { "epoch": 0.65, "learning_rate": 3.9864616199160677e-05, "loss": 0.3822, "step": 1908 }, { "epoch": 0.65, "learning_rate": 3.9853391454453894e-05, "loss": 0.3061, "step": 1909 }, { "epoch": 0.65, "learning_rate": 3.984216207984798e-05, "loss": 0.3414, "step": 1910 }, { "epoch": 0.65, "learning_rate": 3.983092807884318e-05, "loss": 0.3369, "step": 1911 }, { "epoch": 0.65, "learning_rate": 3.981968945494122e-05, "loss": 0.388, "step": 1912 }, { "epoch": 0.65, "learning_rate": 3.9808446211645245e-05, "loss": 0.3637, "step": 1913 }, { "epoch": 0.65, "learning_rate": 3.979719835245986e-05, "loss": 0.3859, "step": 1914 }, { "epoch": 0.65, "learning_rate": 3.978594588089108e-05, "loss": 0.3547, "step": 1915 }, { "epoch": 0.65, "learning_rate": 3.9774688800446384e-05, "loss": 0.3424, "step": 1916 }, { "epoch": 0.65, "learning_rate": 3.9763427114634674e-05, "loss": 0.3406, "step": 1917 }, { "epoch": 0.65, "learning_rate": 3.9752160826966295e-05, "loss": 0.347, "step": 1918 }, { "epoch": 0.65, "learning_rate": 3.974088994095302e-05, "loss": 0.3663, "step": 1919 }, { "epoch": 0.66, "learning_rate": 3.972961446010806e-05, "loss": 0.3702, "step": 1920 }, { "epoch": 0.66, "learning_rate": 3.9718334387946045e-05, "loss": 0.3767, "step": 1921 }, { "epoch": 0.66, "learning_rate": 3.970704972798306e-05, "loss": 0.3369, "step": 1922 }, { "epoch": 0.66, "learning_rate": 3.96957604837366e-05, "loss": 0.3045, "step": 1923 }, { "epoch": 0.66, "learning_rate": 3.968446665872561e-05, "loss": 0.3841, "step": 1924 }, { "epoch": 0.66, "learning_rate": 3.967316825647042e-05, "loss": 0.3444, "step": 1925 }, { "epoch": 0.66, "learning_rate": 3.966186528049285e-05, "loss": 0.3261, "step": 1926 }, { "epoch": 0.66, "learning_rate": 3.965055773431608e-05, "loss": 0.3205, "step": 1927 }, { "epoch": 0.66, "learning_rate": 3.963924562146477e-05, "loss": 0.3292, "step": 1928 }, { "epoch": 0.66, "learning_rate": 3.9627928945464976e-05, "loss": 0.314, "step": 1929 }, { "epoch": 0.66, "learning_rate": 3.9616607709844166e-05, "loss": 0.3405, "step": 1930 }, { "epoch": 0.66, "learning_rate": 3.960528191813125e-05, "loss": 0.3517, "step": 1931 }, { "epoch": 0.66, "learning_rate": 3.959395157385656e-05, "loss": 0.3274, "step": 1932 }, { "epoch": 0.66, "learning_rate": 3.958261668055183e-05, "loss": 0.3681, "step": 1933 }, { "epoch": 0.66, "learning_rate": 3.9571277241750226e-05, "loss": 0.3433, "step": 1934 }, { "epoch": 0.66, "learning_rate": 3.955993326098631e-05, "loss": 0.3405, "step": 1935 }, { "epoch": 0.66, "learning_rate": 3.95485847417961e-05, "loss": 0.3345, "step": 1936 }, { "epoch": 0.66, "learning_rate": 3.9537231687716976e-05, "loss": 0.336, "step": 1937 }, { "epoch": 0.66, "learning_rate": 3.9525874102287776e-05, "loss": 0.3423, "step": 1938 }, { "epoch": 0.66, "learning_rate": 3.9514511989048733e-05, "loss": 0.3829, "step": 1939 }, { "epoch": 0.66, "learning_rate": 3.950314535154148e-05, "loss": 0.3275, "step": 1940 }, { "epoch": 0.66, "learning_rate": 3.9491774193309096e-05, "loss": 0.3717, "step": 1941 }, { "epoch": 0.66, "learning_rate": 3.948039851789602e-05, "loss": 0.3391, "step": 1942 }, { "epoch": 0.66, "learning_rate": 3.946901832884813e-05, "loss": 0.3753, "step": 1943 }, { "epoch": 0.66, "learning_rate": 3.9457633629712716e-05, "loss": 0.3829, "step": 1944 }, { "epoch": 0.66, "learning_rate": 3.944624442403846e-05, "loss": 0.3513, "step": 1945 }, { "epoch": 0.66, "learning_rate": 3.9434850715375446e-05, "loss": 0.3225, "step": 1946 }, { "epoch": 0.66, "learning_rate": 3.9423452507275166e-05, "loss": 0.3616, "step": 1947 }, { "epoch": 0.66, "learning_rate": 3.941204980329053e-05, "loss": 0.3659, "step": 1948 }, { "epoch": 0.66, "learning_rate": 3.940064260697581e-05, "loss": 0.324, "step": 1949 }, { "epoch": 0.67, "learning_rate": 3.9389230921886734e-05, "loss": 0.3254, "step": 1950 }, { "epoch": 0.67, "learning_rate": 3.937781475158038e-05, "loss": 0.3636, "step": 1951 }, { "epoch": 0.67, "learning_rate": 3.936639409961524e-05, "loss": 0.3487, "step": 1952 }, { "epoch": 0.67, "learning_rate": 3.935496896955122e-05, "loss": 0.3155, "step": 1953 }, { "epoch": 0.67, "learning_rate": 3.9343539364949594e-05, "loss": 0.3267, "step": 1954 }, { "epoch": 0.67, "learning_rate": 3.9332105289373054e-05, "loss": 0.3279, "step": 1955 }, { "epoch": 0.67, "learning_rate": 3.932066674638567e-05, "loss": 0.3125, "step": 1956 }, { "epoch": 0.67, "learning_rate": 3.9309223739552926e-05, "loss": 0.2901, "step": 1957 }, { "epoch": 0.67, "learning_rate": 3.929777627244165e-05, "loss": 0.3665, "step": 1958 }, { "epoch": 0.67, "learning_rate": 3.9286324348620115e-05, "loss": 0.3042, "step": 1959 }, { "epoch": 0.67, "learning_rate": 3.927486797165797e-05, "loss": 0.3187, "step": 1960 }, { "epoch": 0.67, "learning_rate": 3.926340714512622e-05, "loss": 0.2991, "step": 1961 }, { "epoch": 0.67, "learning_rate": 3.925194187259729e-05, "loss": 0.3591, "step": 1962 }, { "epoch": 0.67, "learning_rate": 3.924047215764498e-05, "loss": 0.3436, "step": 1963 }, { "epoch": 0.67, "learning_rate": 3.9228998003844474e-05, "loss": 0.3214, "step": 1964 }, { "epoch": 0.67, "learning_rate": 3.921751941477234e-05, "loss": 0.355, "step": 1965 }, { "epoch": 0.67, "learning_rate": 3.920603639400653e-05, "loss": 0.3811, "step": 1966 }, { "epoch": 0.67, "learning_rate": 3.9194548945126386e-05, "loss": 0.34, "step": 1967 }, { "epoch": 0.67, "learning_rate": 3.9183057071712605e-05, "loss": 0.3662, "step": 1968 }, { "epoch": 0.67, "learning_rate": 3.917156077734728e-05, "loss": 0.3034, "step": 1969 }, { "epoch": 0.67, "learning_rate": 3.916006006561389e-05, "loss": 0.3571, "step": 1970 }, { "epoch": 0.67, "learning_rate": 3.9148554940097285e-05, "loss": 0.326, "step": 1971 }, { "epoch": 0.67, "learning_rate": 3.913704540438368e-05, "loss": 0.2981, "step": 1972 }, { "epoch": 0.67, "learning_rate": 3.912553146206066e-05, "loss": 0.361, "step": 1973 }, { "epoch": 0.67, "learning_rate": 3.9114013116717224e-05, "loss": 0.3411, "step": 1974 }, { "epoch": 0.67, "learning_rate": 3.91024903719437e-05, "loss": 0.3016, "step": 1975 }, { "epoch": 0.67, "learning_rate": 3.90909632313318e-05, "loss": 0.3286, "step": 1976 }, { "epoch": 0.67, "learning_rate": 3.907943169847462e-05, "loss": 0.3152, "step": 1977 }, { "epoch": 0.67, "learning_rate": 3.9067895776966604e-05, "loss": 0.3682, "step": 1978 }, { "epoch": 0.68, "learning_rate": 3.9056355470403574e-05, "loss": 0.4004, "step": 1979 }, { "epoch": 0.68, "learning_rate": 3.904481078238272e-05, "loss": 0.3526, "step": 1980 }, { "epoch": 0.68, "learning_rate": 3.903326171650261e-05, "loss": 0.3602, "step": 1981 }, { "epoch": 0.68, "learning_rate": 3.902170827636314e-05, "loss": 0.3388, "step": 1982 }, { "epoch": 0.68, "learning_rate": 3.901015046556562e-05, "loss": 0.3662, "step": 1983 }, { "epoch": 0.68, "learning_rate": 3.8998588287712674e-05, "loss": 0.3654, "step": 1984 }, { "epoch": 0.68, "learning_rate": 3.8987021746408315e-05, "loss": 0.3434, "step": 1985 }, { "epoch": 0.68, "learning_rate": 3.89754508452579e-05, "loss": 0.3523, "step": 1986 }, { "epoch": 0.68, "learning_rate": 3.8963875587868173e-05, "loss": 0.3761, "step": 1987 }, { "epoch": 0.68, "learning_rate": 3.895229597784721e-05, "loss": 0.3642, "step": 1988 }, { "epoch": 0.68, "learning_rate": 3.894071201880444e-05, "loss": 0.3523, "step": 1989 }, { "epoch": 0.68, "learning_rate": 3.892912371435068e-05, "loss": 0.3106, "step": 1990 }, { "epoch": 0.68, "learning_rate": 3.8917531068098054e-05, "loss": 0.3141, "step": 1991 }, { "epoch": 0.68, "learning_rate": 3.890593408366009e-05, "loss": 0.3445, "step": 1992 }, { "epoch": 0.68, "learning_rate": 3.889433276465163e-05, "loss": 0.3688, "step": 1993 }, { "epoch": 0.68, "learning_rate": 3.8882727114688885e-05, "loss": 0.3733, "step": 1994 }, { "epoch": 0.68, "learning_rate": 3.887111713738941e-05, "loss": 0.366, "step": 1995 }, { "epoch": 0.68, "learning_rate": 3.885950283637211e-05, "loss": 0.4771, "step": 1996 }, { "epoch": 0.68, "learning_rate": 3.884788421525725e-05, "loss": 0.3583, "step": 1997 }, { "epoch": 0.68, "learning_rate": 3.8836261277666405e-05, "loss": 0.2864, "step": 1998 }, { "epoch": 0.68, "learning_rate": 3.882463402722254e-05, "loss": 0.3454, "step": 1999 }, { "epoch": 0.68, "learning_rate": 3.881300246754995e-05, "loss": 0.3024, "step": 2000 }, { "epoch": 0.68, "learning_rate": 3.880136660227426e-05, "loss": 0.3587, "step": 2001 }, { "epoch": 0.68, "learning_rate": 3.878972643502243e-05, "loss": 0.3661, "step": 2002 }, { "epoch": 0.68, "learning_rate": 3.8778081969422807e-05, "loss": 0.3696, "step": 2003 }, { "epoch": 0.68, "learning_rate": 3.876643320910502e-05, "loss": 0.3282, "step": 2004 }, { "epoch": 0.68, "learning_rate": 3.8754780157700075e-05, "loss": 0.3301, "step": 2005 }, { "epoch": 0.68, "learning_rate": 3.874312281884031e-05, "loss": 0.3684, "step": 2006 }, { "epoch": 0.68, "learning_rate": 3.873146119615938e-05, "loss": 0.3648, "step": 2007 }, { "epoch": 0.69, "learning_rate": 3.8719795293292295e-05, "loss": 0.3275, "step": 2008 }, { "epoch": 0.69, "learning_rate": 3.8708125113875406e-05, "loss": 0.4067, "step": 2009 }, { "epoch": 0.69, "learning_rate": 3.869645066154636e-05, "loss": 0.2998, "step": 2010 }, { "epoch": 0.69, "learning_rate": 3.8684771939944174e-05, "loss": 0.3566, "step": 2011 }, { "epoch": 0.69, "learning_rate": 3.867308895270919e-05, "loss": 0.3761, "step": 2012 }, { "epoch": 0.69, "learning_rate": 3.8661401703483046e-05, "loss": 0.3366, "step": 2013 }, { "epoch": 0.69, "learning_rate": 3.8649710195908764e-05, "loss": 0.3485, "step": 2014 }, { "epoch": 0.69, "learning_rate": 3.863801443363064e-05, "loss": 0.3184, "step": 2015 }, { "epoch": 0.69, "learning_rate": 3.862631442029434e-05, "loss": 0.3933, "step": 2016 }, { "epoch": 0.69, "learning_rate": 3.861461015954681e-05, "loss": 0.3452, "step": 2017 }, { "epoch": 0.69, "learning_rate": 3.860290165503636e-05, "loss": 0.3548, "step": 2018 }, { "epoch": 0.69, "learning_rate": 3.859118891041261e-05, "loss": 0.3583, "step": 2019 }, { "epoch": 0.69, "learning_rate": 3.8579471929326494e-05, "loss": 0.3529, "step": 2020 }, { "epoch": 0.69, "learning_rate": 3.856775071543026e-05, "loss": 0.3081, "step": 2021 }, { "epoch": 0.69, "learning_rate": 3.8556025272377515e-05, "loss": 0.4169, "step": 2022 }, { "epoch": 0.69, "learning_rate": 3.854429560382313e-05, "loss": 0.3666, "step": 2023 }, { "epoch": 0.69, "learning_rate": 3.853256171342333e-05, "loss": 0.3751, "step": 2024 }, { "epoch": 0.69, "learning_rate": 3.852082360483564e-05, "loss": 0.3109, "step": 2025 }, { "epoch": 0.69, "learning_rate": 3.850908128171892e-05, "loss": 0.3096, "step": 2026 }, { "epoch": 0.69, "learning_rate": 3.849733474773331e-05, "loss": 0.2982, "step": 2027 }, { "epoch": 0.69, "learning_rate": 3.848558400654029e-05, "loss": 0.3361, "step": 2028 }, { "epoch": 0.69, "learning_rate": 3.8473829061802644e-05, "loss": 0.359, "step": 2029 }, { "epoch": 0.69, "learning_rate": 3.846206991718446e-05, "loss": 0.3308, "step": 2030 }, { "epoch": 0.69, "learning_rate": 3.8450306576351145e-05, "loss": 0.3309, "step": 2031 }, { "epoch": 0.69, "learning_rate": 3.843853904296942e-05, "loss": 0.3365, "step": 2032 }, { "epoch": 0.69, "learning_rate": 3.842676732070728e-05, "loss": 0.3483, "step": 2033 }, { "epoch": 0.69, "learning_rate": 3.841499141323405e-05, "loss": 0.3599, "step": 2034 }, { "epoch": 0.69, "learning_rate": 3.8403211324220375e-05, "loss": 0.3503, "step": 2035 }, { "epoch": 0.69, "learning_rate": 3.839142705733817e-05, "loss": 0.3595, "step": 2036 }, { "epoch": 0.69, "learning_rate": 3.837963861626067e-05, "loss": 0.3714, "step": 2037 }, { "epoch": 0.7, "learning_rate": 3.836784600466242e-05, "loss": 0.3804, "step": 2038 }, { "epoch": 0.7, "learning_rate": 3.835604922621925e-05, "loss": 0.3849, "step": 2039 }, { "epoch": 0.7, "learning_rate": 3.8344248284608284e-05, "loss": 0.307, "step": 2040 }, { "epoch": 0.7, "learning_rate": 3.833244318350796e-05, "loss": 0.3268, "step": 2041 }, { "epoch": 0.7, "learning_rate": 3.8320633926597995e-05, "loss": 0.348, "step": 2042 }, { "epoch": 0.7, "learning_rate": 3.830882051755942e-05, "loss": 0.3664, "step": 2043 }, { "epoch": 0.7, "learning_rate": 3.829700296007456e-05, "loss": 0.3177, "step": 2044 }, { "epoch": 0.7, "learning_rate": 3.8285181257827016e-05, "loss": 0.3401, "step": 2045 }, { "epoch": 0.7, "learning_rate": 3.827335541450169e-05, "loss": 0.3103, "step": 2046 }, { "epoch": 0.7, "learning_rate": 3.8261525433784764e-05, "loss": 0.3251, "step": 2047 }, { "epoch": 0.7, "learning_rate": 3.824969131936375e-05, "loss": 0.349, "step": 2048 }, { "epoch": 0.7, "learning_rate": 3.8237853074927376e-05, "loss": 0.3374, "step": 2049 }, { "epoch": 0.7, "learning_rate": 3.822601070416574e-05, "loss": 0.3055, "step": 2050 }, { "epoch": 0.7, "learning_rate": 3.8214164210770175e-05, "loss": 0.3594, "step": 2051 }, { "epoch": 0.7, "learning_rate": 3.820231359843329e-05, "loss": 0.3079, "step": 2052 }, { "epoch": 0.7, "learning_rate": 3.819045887084903e-05, "loss": 0.3968, "step": 2053 }, { "epoch": 0.7, "learning_rate": 3.817860003171256e-05, "loss": 0.351, "step": 2054 }, { "epoch": 0.7, "learning_rate": 3.816673708472038e-05, "loss": 0.3506, "step": 2055 }, { "epoch": 0.7, "learning_rate": 3.815487003357025e-05, "loss": 0.3597, "step": 2056 }, { "epoch": 0.7, "learning_rate": 3.8142998881961176e-05, "loss": 0.3526, "step": 2057 }, { "epoch": 0.7, "learning_rate": 3.8131123633593514e-05, "loss": 0.3267, "step": 2058 }, { "epoch": 0.7, "learning_rate": 3.811924429216883e-05, "loss": 0.3823, "step": 2059 }, { "epoch": 0.7, "learning_rate": 3.8107360861390004e-05, "loss": 0.3695, "step": 2060 }, { "epoch": 0.7, "learning_rate": 3.8095473344961175e-05, "loss": 0.3494, "step": 2061 }, { "epoch": 0.7, "learning_rate": 3.8083581746587746e-05, "loss": 0.32, "step": 2062 }, { "epoch": 0.7, "learning_rate": 3.807168606997644e-05, "loss": 0.3436, "step": 2063 }, { "epoch": 0.7, "learning_rate": 3.805978631883519e-05, "loss": 0.3409, "step": 2064 }, { "epoch": 0.7, "learning_rate": 3.804788249687323e-05, "loss": 0.3246, "step": 2065 }, { "epoch": 0.7, "learning_rate": 3.803597460780107e-05, "loss": 0.3212, "step": 2066 }, { "epoch": 0.71, "learning_rate": 3.8024062655330455e-05, "loss": 0.3291, "step": 2067 }, { "epoch": 0.71, "learning_rate": 3.801214664317444e-05, "loss": 0.3162, "step": 2068 }, { "epoch": 0.71, "learning_rate": 3.8000226575047314e-05, "loss": 0.3427, "step": 2069 }, { "epoch": 0.71, "learning_rate": 3.798830245466465e-05, "loss": 0.3582, "step": 2070 }, { "epoch": 0.71, "learning_rate": 3.797637428574326e-05, "loss": 0.3649, "step": 2071 }, { "epoch": 0.71, "learning_rate": 3.7964442072001236e-05, "loss": 0.3345, "step": 2072 }, { "epoch": 0.71, "learning_rate": 3.795250581715793e-05, "loss": 0.3004, "step": 2073 }, { "epoch": 0.71, "learning_rate": 3.7940565524933956e-05, "loss": 0.3106, "step": 2074 }, { "epoch": 0.71, "learning_rate": 3.792862119905117e-05, "loss": 0.3602, "step": 2075 }, { "epoch": 0.71, "learning_rate": 3.79166728432327e-05, "loss": 0.3233, "step": 2076 }, { "epoch": 0.71, "learning_rate": 3.790472046120294e-05, "loss": 0.3783, "step": 2077 }, { "epoch": 0.71, "learning_rate": 3.78927640566875e-05, "loss": 0.3494, "step": 2078 }, { "epoch": 0.71, "learning_rate": 3.78808036334133e-05, "loss": 0.3352, "step": 2079 }, { "epoch": 0.71, "learning_rate": 3.786883919510844e-05, "loss": 0.3543, "step": 2080 }, { "epoch": 0.71, "learning_rate": 3.785687074550236e-05, "loss": 0.3377, "step": 2081 }, { "epoch": 0.71, "learning_rate": 3.784489828832567e-05, "loss": 0.3835, "step": 2082 }, { "epoch": 0.71, "learning_rate": 3.7832921827310286e-05, "loss": 0.3483, "step": 2083 }, { "epoch": 0.71, "learning_rate": 3.782094136618933e-05, "loss": 0.3414, "step": 2084 }, { "epoch": 0.71, "learning_rate": 3.780895690869721e-05, "loss": 0.3665, "step": 2085 }, { "epoch": 0.71, "learning_rate": 3.779696845856954e-05, "loss": 0.307, "step": 2086 }, { "epoch": 0.71, "learning_rate": 3.7784976019543204e-05, "loss": 0.3742, "step": 2087 }, { "epoch": 0.71, "learning_rate": 3.777297959535634e-05, "loss": 0.3328, "step": 2088 }, { "epoch": 0.71, "learning_rate": 3.776097918974828e-05, "loss": 0.3789, "step": 2089 }, { "epoch": 0.71, "learning_rate": 3.774897480645966e-05, "loss": 0.3132, "step": 2090 }, { "epoch": 0.71, "learning_rate": 3.7736966449232314e-05, "loss": 0.3319, "step": 2091 }, { "epoch": 0.71, "learning_rate": 3.772495412180932e-05, "loss": 0.3278, "step": 2092 }, { "epoch": 0.71, "learning_rate": 3.7712937827934994e-05, "loss": 0.3737, "step": 2093 }, { "epoch": 0.71, "learning_rate": 3.770091757135491e-05, "loss": 0.3952, "step": 2094 }, { "epoch": 0.71, "learning_rate": 3.7688893355815846e-05, "loss": 0.3708, "step": 2095 }, { "epoch": 0.72, "learning_rate": 3.767686518506583e-05, "loss": 0.3105, "step": 2096 }, { "epoch": 0.72, "learning_rate": 3.7664833062854143e-05, "loss": 0.3864, "step": 2097 }, { "epoch": 0.72, "learning_rate": 3.765279699293125e-05, "loss": 0.389, "step": 2098 }, { "epoch": 0.72, "learning_rate": 3.764075697904887e-05, "loss": 0.3433, "step": 2099 }, { "epoch": 0.72, "learning_rate": 3.7628713024959984e-05, "loss": 0.3258, "step": 2100 }, { "epoch": 0.72, "learning_rate": 3.761666513441874e-05, "loss": 0.3748, "step": 2101 }, { "epoch": 0.72, "learning_rate": 3.760461331118057e-05, "loss": 0.3152, "step": 2102 }, { "epoch": 0.72, "learning_rate": 3.7592557559002076e-05, "loss": 0.3554, "step": 2103 }, { "epoch": 0.72, "learning_rate": 3.7580497881641144e-05, "loss": 0.3374, "step": 2104 }, { "epoch": 0.72, "learning_rate": 3.756843428285684e-05, "loss": 0.3274, "step": 2105 }, { "epoch": 0.72, "learning_rate": 3.7556366766409463e-05, "loss": 0.354, "step": 2106 }, { "epoch": 0.72, "learning_rate": 3.7544295336060544e-05, "loss": 0.3729, "step": 2107 }, { "epoch": 0.72, "learning_rate": 3.7532219995572824e-05, "loss": 0.3859, "step": 2108 }, { "epoch": 0.72, "learning_rate": 3.7520140748710274e-05, "loss": 0.321, "step": 2109 }, { "epoch": 0.72, "learning_rate": 3.7508057599238056e-05, "loss": 0.3339, "step": 2110 }, { "epoch": 0.72, "learning_rate": 3.749597055092258e-05, "loss": 0.3383, "step": 2111 }, { "epoch": 0.72, "learning_rate": 3.748387960753145e-05, "loss": 0.3113, "step": 2112 }, { "epoch": 0.72, "learning_rate": 3.7471784772833504e-05, "loss": 0.3236, "step": 2113 }, { "epoch": 0.72, "learning_rate": 3.745968605059878e-05, "loss": 0.3459, "step": 2114 }, { "epoch": 0.72, "learning_rate": 3.744758344459852e-05, "loss": 0.4025, "step": 2115 }, { "epoch": 0.72, "learning_rate": 3.743547695860518e-05, "loss": 0.3421, "step": 2116 }, { "epoch": 0.72, "learning_rate": 3.742336659639244e-05, "loss": 0.3515, "step": 2117 }, { "epoch": 0.72, "learning_rate": 3.7411252361735186e-05, "loss": 0.3631, "step": 2118 }, { "epoch": 0.72, "learning_rate": 3.739913425840948e-05, "loss": 0.3478, "step": 2119 }, { "epoch": 0.72, "learning_rate": 3.738701229019265e-05, "loss": 0.3487, "step": 2120 }, { "epoch": 0.72, "learning_rate": 3.7374886460863154e-05, "loss": 0.3463, "step": 2121 }, { "epoch": 0.72, "learning_rate": 3.7362756774200725e-05, "loss": 0.3123, "step": 2122 }, { "epoch": 0.72, "learning_rate": 3.735062323398625e-05, "loss": 0.3619, "step": 2123 }, { "epoch": 0.72, "learning_rate": 3.7338485844001824e-05, "loss": 0.3219, "step": 2124 }, { "epoch": 0.73, "learning_rate": 3.732634460803076e-05, "loss": 0.3594, "step": 2125 }, { "epoch": 0.73, "learning_rate": 3.731419952985758e-05, "loss": 0.3018, "step": 2126 }, { "epoch": 0.73, "learning_rate": 3.730205061326795e-05, "loss": 0.3614, "step": 2127 }, { "epoch": 0.73, "learning_rate": 3.7289897862048783e-05, "loss": 0.3142, "step": 2128 }, { "epoch": 0.73, "learning_rate": 3.727774127998817e-05, "loss": 0.3757, "step": 2129 }, { "epoch": 0.73, "learning_rate": 3.726558087087541e-05, "loss": 0.3775, "step": 2130 }, { "epoch": 0.73, "learning_rate": 3.725341663850095e-05, "loss": 0.3821, "step": 2131 }, { "epoch": 0.73, "learning_rate": 3.724124858665649e-05, "loss": 0.3809, "step": 2132 }, { "epoch": 0.73, "learning_rate": 3.722907671913488e-05, "loss": 0.3592, "step": 2133 }, { "epoch": 0.73, "learning_rate": 3.721690103973018e-05, "loss": 0.3607, "step": 2134 }, { "epoch": 0.73, "learning_rate": 3.720472155223761e-05, "loss": 0.3375, "step": 2135 }, { "epoch": 0.73, "learning_rate": 3.7192538260453616e-05, "loss": 0.3307, "step": 2136 }, { "epoch": 0.73, "learning_rate": 3.718035116817579e-05, "loss": 0.3363, "step": 2137 }, { "epoch": 0.73, "learning_rate": 3.7168160279202935e-05, "loss": 0.3302, "step": 2138 }, { "epoch": 0.73, "learning_rate": 3.715596559733504e-05, "loss": 0.3047, "step": 2139 }, { "epoch": 0.73, "learning_rate": 3.714376712637326e-05, "loss": 0.3677, "step": 2140 }, { "epoch": 0.73, "learning_rate": 3.7131564870119945e-05, "loss": 0.3119, "step": 2141 }, { "epoch": 0.73, "learning_rate": 3.71193588323786e-05, "loss": 0.3472, "step": 2142 }, { "epoch": 0.73, "learning_rate": 3.710714901695396e-05, "loss": 0.3729, "step": 2143 }, { "epoch": 0.73, "learning_rate": 3.709493542765186e-05, "loss": 0.2948, "step": 2144 }, { "epoch": 0.73, "learning_rate": 3.7082718068279386e-05, "loss": 0.3968, "step": 2145 }, { "epoch": 0.73, "learning_rate": 3.707049694264476e-05, "loss": 0.3731, "step": 2146 }, { "epoch": 0.73, "learning_rate": 3.70582720545574e-05, "loss": 0.3589, "step": 2147 }, { "epoch": 0.73, "learning_rate": 3.704604340782786e-05, "loss": 0.3628, "step": 2148 }, { "epoch": 0.73, "learning_rate": 3.70338110062679e-05, "loss": 0.3394, "step": 2149 }, { "epoch": 0.73, "learning_rate": 3.7021574853690444e-05, "loss": 0.3661, "step": 2150 }, { "epoch": 0.73, "learning_rate": 3.700933495390958e-05, "loss": 0.3085, "step": 2151 }, { "epoch": 0.73, "learning_rate": 3.699709131074055e-05, "loss": 0.3334, "step": 2152 }, { "epoch": 0.73, "learning_rate": 3.698484392799979e-05, "loss": 0.3313, "step": 2153 }, { "epoch": 0.73, "learning_rate": 3.6972592809504906e-05, "loss": 0.3727, "step": 2154 }, { "epoch": 0.74, "learning_rate": 3.696033795907462e-05, "loss": 0.2809, "step": 2155 }, { "epoch": 0.74, "learning_rate": 3.694807938052887e-05, "loss": 0.3374, "step": 2156 }, { "epoch": 0.74, "learning_rate": 3.6935817077688716e-05, "loss": 0.3638, "step": 2157 }, { "epoch": 0.74, "learning_rate": 3.692355105437642e-05, "loss": 0.3716, "step": 2158 }, { "epoch": 0.74, "learning_rate": 3.691128131441536e-05, "loss": 0.3451, "step": 2159 }, { "epoch": 0.74, "learning_rate": 3.6899007861630116e-05, "loss": 0.275, "step": 2160 }, { "epoch": 0.74, "learning_rate": 3.6886730699846386e-05, "loss": 0.326, "step": 2161 }, { "epoch": 0.74, "learning_rate": 3.6874449832891056e-05, "loss": 0.328, "step": 2162 }, { "epoch": 0.74, "learning_rate": 3.686216526459214e-05, "loss": 0.3717, "step": 2163 }, { "epoch": 0.74, "learning_rate": 3.684987699877882e-05, "loss": 0.324, "step": 2164 }, { "epoch": 0.74, "learning_rate": 3.683758503928143e-05, "loss": 0.3203, "step": 2165 }, { "epoch": 0.74, "learning_rate": 3.682528938993146e-05, "loss": 0.3382, "step": 2166 }, { "epoch": 0.74, "learning_rate": 3.681299005456155e-05, "loss": 0.3318, "step": 2167 }, { "epoch": 0.74, "learning_rate": 3.680068703700546e-05, "loss": 0.3314, "step": 2168 }, { "epoch": 0.74, "learning_rate": 3.678838034109814e-05, "loss": 0.3671, "step": 2169 }, { "epoch": 0.74, "learning_rate": 3.677606997067566e-05, "loss": 0.3209, "step": 2170 }, { "epoch": 0.74, "learning_rate": 3.6763755929575236e-05, "loss": 0.3336, "step": 2171 }, { "epoch": 0.74, "learning_rate": 3.675143822163526e-05, "loss": 0.3471, "step": 2172 }, { "epoch": 0.74, "learning_rate": 3.673911685069521e-05, "loss": 0.3584, "step": 2173 }, { "epoch": 0.74, "learning_rate": 3.6726791820595766e-05, "loss": 0.3474, "step": 2174 }, { "epoch": 0.74, "learning_rate": 3.67144631351787e-05, "loss": 0.3674, "step": 2175 }, { "epoch": 0.74, "learning_rate": 3.670213079828695e-05, "loss": 0.3524, "step": 2176 }, { "epoch": 0.74, "learning_rate": 3.6689794813764596e-05, "loss": 0.4197, "step": 2177 }, { "epoch": 0.74, "learning_rate": 3.667745518545683e-05, "loss": 0.3026, "step": 2178 }, { "epoch": 0.74, "learning_rate": 3.666511191721e-05, "loss": 0.3248, "step": 2179 }, { "epoch": 0.74, "learning_rate": 3.665276501287159e-05, "loss": 0.2784, "step": 2180 }, { "epoch": 0.74, "learning_rate": 3.6640414476290205e-05, "loss": 0.3923, "step": 2181 }, { "epoch": 0.74, "learning_rate": 3.662806031131558e-05, "loss": 0.2961, "step": 2182 }, { "epoch": 0.74, "learning_rate": 3.66157025217986e-05, "loss": 0.3187, "step": 2183 }, { "epoch": 0.75, "learning_rate": 3.660334111159127e-05, "loss": 0.359, "step": 2184 }, { "epoch": 0.75, "learning_rate": 3.659097608454672e-05, "loss": 0.3143, "step": 2185 }, { "epoch": 0.75, "learning_rate": 3.65786074445192e-05, "loss": 0.2923, "step": 2186 }, { "epoch": 0.75, "learning_rate": 3.656623519536411e-05, "loss": 0.3926, "step": 2187 }, { "epoch": 0.75, "learning_rate": 3.6553859340937955e-05, "loss": 0.3007, "step": 2188 }, { "epoch": 0.75, "learning_rate": 3.654147988509836e-05, "loss": 0.3296, "step": 2189 }, { "epoch": 0.75, "learning_rate": 3.65290968317041e-05, "loss": 0.3266, "step": 2190 }, { "epoch": 0.75, "learning_rate": 3.651671018461504e-05, "loss": 0.3225, "step": 2191 }, { "epoch": 0.75, "learning_rate": 3.6504319947692195e-05, "loss": 0.3014, "step": 2192 }, { "epoch": 0.75, "learning_rate": 3.649192612479766e-05, "loss": 0.3786, "step": 2193 }, { "epoch": 0.75, "learning_rate": 3.647952871979469e-05, "loss": 0.3254, "step": 2194 }, { "epoch": 0.75, "learning_rate": 3.646712773654763e-05, "loss": 0.313, "step": 2195 }, { "epoch": 0.75, "learning_rate": 3.645472317892193e-05, "loss": 0.3186, "step": 2196 }, { "epoch": 0.75, "learning_rate": 3.644231505078421e-05, "loss": 0.34, "step": 2197 }, { "epoch": 0.75, "learning_rate": 3.642990335600214e-05, "loss": 0.3323, "step": 2198 }, { "epoch": 0.75, "learning_rate": 3.641748809844452e-05, "loss": 0.3021, "step": 2199 }, { "epoch": 0.75, "learning_rate": 3.640506928198128e-05, "loss": 0.2975, "step": 2200 }, { "epoch": 0.75, "learning_rate": 3.639264691048345e-05, "loss": 0.3293, "step": 2201 }, { "epoch": 0.75, "learning_rate": 3.638022098782314e-05, "loss": 0.3768, "step": 2202 }, { "epoch": 0.75, "learning_rate": 3.636779151787362e-05, "loss": 0.3483, "step": 2203 }, { "epoch": 0.75, "learning_rate": 3.635535850450923e-05, "loss": 0.3402, "step": 2204 }, { "epoch": 0.75, "learning_rate": 3.634292195160541e-05, "loss": 0.3241, "step": 2205 }, { "epoch": 0.75, "learning_rate": 3.633048186303872e-05, "loss": 0.3354, "step": 2206 }, { "epoch": 0.75, "learning_rate": 3.631803824268682e-05, "loss": 0.298, "step": 2207 }, { "epoch": 0.75, "learning_rate": 3.630559109442847e-05, "loss": 0.3344, "step": 2208 }, { "epoch": 0.75, "learning_rate": 3.629314042214351e-05, "loss": 0.375, "step": 2209 }, { "epoch": 0.75, "learning_rate": 3.6280686229712924e-05, "loss": 0.3278, "step": 2210 }, { "epoch": 0.75, "learning_rate": 3.626822852101875e-05, "loss": 0.3292, "step": 2211 }, { "epoch": 0.75, "learning_rate": 3.6255767299944134e-05, "loss": 0.3328, "step": 2212 }, { "epoch": 0.76, "learning_rate": 3.624330257037333e-05, "loss": 0.3296, "step": 2213 }, { "epoch": 0.76, "learning_rate": 3.623083433619166e-05, "loss": 0.3522, "step": 2214 }, { "epoch": 0.76, "learning_rate": 3.621836260128556e-05, "loss": 0.2695, "step": 2215 }, { "epoch": 0.76, "learning_rate": 3.6205887369542555e-05, "loss": 0.3719, "step": 2216 }, { "epoch": 0.76, "learning_rate": 3.6193408644851265e-05, "loss": 0.3334, "step": 2217 }, { "epoch": 0.76, "learning_rate": 3.618092643110138e-05, "loss": 0.3394, "step": 2218 }, { "epoch": 0.76, "learning_rate": 3.616844073218369e-05, "loss": 0.3362, "step": 2219 }, { "epoch": 0.76, "learning_rate": 3.6155951551990064e-05, "loss": 0.3185, "step": 2220 }, { "epoch": 0.76, "learning_rate": 3.6143458894413465e-05, "loss": 0.3336, "step": 2221 }, { "epoch": 0.76, "learning_rate": 3.613096276334793e-05, "loss": 0.3274, "step": 2222 }, { "epoch": 0.76, "learning_rate": 3.611846316268859e-05, "loss": 0.3575, "step": 2223 }, { "epoch": 0.76, "learning_rate": 3.610596009633166e-05, "loss": 0.4068, "step": 2224 }, { "epoch": 0.76, "learning_rate": 3.609345356817442e-05, "loss": 0.3701, "step": 2225 }, { "epoch": 0.76, "learning_rate": 3.608094358211523e-05, "loss": 0.3292, "step": 2226 }, { "epoch": 0.76, "learning_rate": 3.606843014205354e-05, "loss": 0.3689, "step": 2227 }, { "epoch": 0.76, "learning_rate": 3.6055913251889876e-05, "loss": 0.3326, "step": 2228 }, { "epoch": 0.76, "learning_rate": 3.6043392915525826e-05, "loss": 0.3157, "step": 2229 }, { "epoch": 0.76, "learning_rate": 3.603086913686407e-05, "loss": 0.3228, "step": 2230 }, { "epoch": 0.76, "learning_rate": 3.6018341919808344e-05, "loss": 0.3536, "step": 2231 }, { "epoch": 0.76, "learning_rate": 3.600581126826346e-05, "loss": 0.3548, "step": 2232 }, { "epoch": 0.76, "learning_rate": 3.5993277186135324e-05, "loss": 0.3242, "step": 2233 }, { "epoch": 0.76, "learning_rate": 3.598073967733086e-05, "loss": 0.2965, "step": 2234 }, { "epoch": 0.76, "learning_rate": 3.596819874575811e-05, "loss": 0.3076, "step": 2235 }, { "epoch": 0.76, "learning_rate": 3.595565439532617e-05, "loss": 0.3006, "step": 2236 }, { "epoch": 0.76, "learning_rate": 3.594310662994519e-05, "loss": 0.3491, "step": 2237 }, { "epoch": 0.76, "learning_rate": 3.5930555453526385e-05, "loss": 0.3262, "step": 2238 }, { "epoch": 0.76, "learning_rate": 3.591800086998205e-05, "loss": 0.3731, "step": 2239 }, { "epoch": 0.76, "learning_rate": 3.590544288322551e-05, "loss": 0.3551, "step": 2240 }, { "epoch": 0.76, "learning_rate": 3.589288149717119e-05, "loss": 0.3117, "step": 2241 }, { "epoch": 0.76, "learning_rate": 3.588031671573455e-05, "loss": 0.3237, "step": 2242 }, { "epoch": 0.77, "learning_rate": 3.5867748542832116e-05, "loss": 0.3272, "step": 2243 }, { "epoch": 0.77, "learning_rate": 3.5855176982381476e-05, "loss": 0.3242, "step": 2244 }, { "epoch": 0.77, "learning_rate": 3.584260203830126e-05, "loss": 0.3413, "step": 2245 }, { "epoch": 0.77, "learning_rate": 3.583002371451116e-05, "loss": 0.3522, "step": 2246 }, { "epoch": 0.77, "learning_rate": 3.581744201493191e-05, "loss": 0.3228, "step": 2247 }, { "epoch": 0.77, "learning_rate": 3.5804856943485334e-05, "loss": 0.3037, "step": 2248 }, { "epoch": 0.77, "learning_rate": 3.5792268504094256e-05, "loss": 0.3407, "step": 2249 }, { "epoch": 0.77, "learning_rate": 3.5779676700682595e-05, "loss": 0.3598, "step": 2250 }, { "epoch": 0.77, "learning_rate": 3.5767081537175285e-05, "loss": 0.348, "step": 2251 }, { "epoch": 0.77, "learning_rate": 3.575448301749833e-05, "loss": 0.3469, "step": 2252 }, { "epoch": 0.77, "learning_rate": 3.574188114557876e-05, "loss": 0.2799, "step": 2253 }, { "epoch": 0.77, "learning_rate": 3.5729275925344676e-05, "loss": 0.3655, "step": 2254 }, { "epoch": 0.77, "learning_rate": 3.571666736072519e-05, "loss": 0.2945, "step": 2255 }, { "epoch": 0.77, "learning_rate": 3.5704055455650484e-05, "loss": 0.3237, "step": 2256 }, { "epoch": 0.77, "learning_rate": 3.569144021405177e-05, "loss": 0.3584, "step": 2257 }, { "epoch": 0.77, "learning_rate": 3.5678821639861305e-05, "loss": 0.3335, "step": 2258 }, { "epoch": 0.77, "learning_rate": 3.566619973701237e-05, "loss": 0.3527, "step": 2259 }, { "epoch": 0.77, "learning_rate": 3.56535745094393e-05, "loss": 0.3798, "step": 2260 }, { "epoch": 0.77, "learning_rate": 3.564094596107747e-05, "loss": 0.3145, "step": 2261 }, { "epoch": 0.77, "learning_rate": 3.562831409586327e-05, "loss": 0.356, "step": 2262 }, { "epoch": 0.77, "learning_rate": 3.561567891773414e-05, "loss": 0.3584, "step": 2263 }, { "epoch": 0.77, "learning_rate": 3.560304043062854e-05, "loss": 0.3244, "step": 2264 }, { "epoch": 0.77, "learning_rate": 3.559039863848598e-05, "loss": 0.303, "step": 2265 }, { "epoch": 0.77, "learning_rate": 3.557775354524698e-05, "loss": 0.3663, "step": 2266 }, { "epoch": 0.77, "learning_rate": 3.556510515485311e-05, "loss": 0.3153, "step": 2267 }, { "epoch": 0.77, "learning_rate": 3.555245347124694e-05, "loss": 0.3328, "step": 2268 }, { "epoch": 0.77, "learning_rate": 3.55397984983721e-05, "loss": 0.281, "step": 2269 }, { "epoch": 0.77, "learning_rate": 3.552714024017321e-05, "loss": 0.2885, "step": 2270 }, { "epoch": 0.77, "learning_rate": 3.551447870059594e-05, "loss": 0.3839, "step": 2271 }, { "epoch": 0.78, "learning_rate": 3.5501813883586974e-05, "loss": 0.3421, "step": 2272 }, { "epoch": 0.78, "learning_rate": 3.548914579309402e-05, "loss": 0.2969, "step": 2273 }, { "epoch": 0.78, "learning_rate": 3.54764744330658e-05, "loss": 0.3679, "step": 2274 }, { "epoch": 0.78, "learning_rate": 3.5463799807452075e-05, "loss": 0.334, "step": 2275 }, { "epoch": 0.78, "learning_rate": 3.5451121920203584e-05, "loss": 0.3139, "step": 2276 }, { "epoch": 0.78, "learning_rate": 3.5438440775272125e-05, "loss": 0.3383, "step": 2277 }, { "epoch": 0.78, "learning_rate": 3.542575637661049e-05, "loss": 0.3173, "step": 2278 }, { "epoch": 0.78, "learning_rate": 3.541306872817248e-05, "loss": 0.3449, "step": 2279 }, { "epoch": 0.78, "learning_rate": 3.540037783391293e-05, "loss": 0.3164, "step": 2280 }, { "epoch": 0.78, "learning_rate": 3.538768369778768e-05, "loss": 0.2896, "step": 2281 }, { "epoch": 0.78, "learning_rate": 3.537498632375355e-05, "loss": 0.3279, "step": 2282 }, { "epoch": 0.78, "learning_rate": 3.536228571576841e-05, "loss": 0.3119, "step": 2283 }, { "epoch": 0.78, "learning_rate": 3.534958187779113e-05, "loss": 0.3569, "step": 2284 }, { "epoch": 0.78, "learning_rate": 3.533687481378157e-05, "loss": 0.3836, "step": 2285 }, { "epoch": 0.78, "learning_rate": 3.53241645277006e-05, "loss": 0.3295, "step": 2286 }, { "epoch": 0.78, "learning_rate": 3.531145102351012e-05, "loss": 0.3856, "step": 2287 }, { "epoch": 0.78, "learning_rate": 3.529873430517299e-05, "loss": 0.2707, "step": 2288 }, { "epoch": 0.78, "learning_rate": 3.52860143766531e-05, "loss": 0.3567, "step": 2289 }, { "epoch": 0.78, "learning_rate": 3.527329124191533e-05, "loss": 0.2622, "step": 2290 }, { "epoch": 0.78, "learning_rate": 3.5260564904925576e-05, "loss": 0.3002, "step": 2291 }, { "epoch": 0.78, "learning_rate": 3.5247835369650715e-05, "loss": 0.3283, "step": 2292 }, { "epoch": 0.78, "learning_rate": 3.5235102640058634e-05, "loss": 0.3556, "step": 2293 }, { "epoch": 0.78, "learning_rate": 3.5222366720118196e-05, "loss": 0.32, "step": 2294 }, { "epoch": 0.78, "learning_rate": 3.520962761379927e-05, "loss": 0.3686, "step": 2295 }, { "epoch": 0.78, "learning_rate": 3.519688532507273e-05, "loss": 0.2927, "step": 2296 }, { "epoch": 0.78, "learning_rate": 3.5184139857910416e-05, "loss": 0.319, "step": 2297 }, { "epoch": 0.78, "learning_rate": 3.5171391216285196e-05, "loss": 0.3505, "step": 2298 }, { "epoch": 0.78, "learning_rate": 3.515863940417087e-05, "loss": 0.3126, "step": 2299 }, { "epoch": 0.78, "learning_rate": 3.51458844255423e-05, "loss": 0.355, "step": 2300 }, { "epoch": 0.79, "learning_rate": 3.5133126284375275e-05, "loss": 0.295, "step": 2301 }, { "epoch": 0.79, "learning_rate": 3.5120364984646594e-05, "loss": 0.3707, "step": 2302 }, { "epoch": 0.79, "learning_rate": 3.5107600530334026e-05, "loss": 0.3644, "step": 2303 }, { "epoch": 0.79, "learning_rate": 3.5094832925416355e-05, "loss": 0.3274, "step": 2304 }, { "epoch": 0.79, "learning_rate": 3.5082062173873316e-05, "loss": 0.3241, "step": 2305 }, { "epoch": 0.79, "learning_rate": 3.506928827968564e-05, "loss": 0.3192, "step": 2306 }, { "epoch": 0.79, "learning_rate": 3.5056511246835025e-05, "loss": 0.328, "step": 2307 }, { "epoch": 0.79, "learning_rate": 3.5043731079304174e-05, "loss": 0.33, "step": 2308 }, { "epoch": 0.79, "learning_rate": 3.503094778107673e-05, "loss": 0.368, "step": 2309 }, { "epoch": 0.79, "learning_rate": 3.501816135613733e-05, "loss": 0.3417, "step": 2310 }, { "epoch": 0.79, "learning_rate": 3.500537180847159e-05, "loss": 0.3119, "step": 2311 }, { "epoch": 0.79, "learning_rate": 3.4992579142066104e-05, "loss": 0.356, "step": 2312 }, { "epoch": 0.79, "learning_rate": 3.497978336090843e-05, "loss": 0.373, "step": 2313 }, { "epoch": 0.79, "learning_rate": 3.496698446898709e-05, "loss": 0.3771, "step": 2314 }, { "epoch": 0.79, "learning_rate": 3.495418247029158e-05, "loss": 0.3233, "step": 2315 }, { "epoch": 0.79, "learning_rate": 3.494137736881237e-05, "loss": 0.3339, "step": 2316 }, { "epoch": 0.79, "learning_rate": 3.492856916854089e-05, "loss": 0.341, "step": 2317 }, { "epoch": 0.79, "learning_rate": 3.491575787346954e-05, "loss": 0.3383, "step": 2318 }, { "epoch": 0.79, "learning_rate": 3.4902943487591696e-05, "loss": 0.3593, "step": 2319 }, { "epoch": 0.79, "learning_rate": 3.489012601490167e-05, "loss": 0.3579, "step": 2320 }, { "epoch": 0.79, "learning_rate": 3.487730545939477e-05, "loss": 0.4045, "step": 2321 }, { "epoch": 0.79, "learning_rate": 3.486448182506723e-05, "loss": 0.3094, "step": 2322 }, { "epoch": 0.79, "learning_rate": 3.485165511591627e-05, "loss": 0.3625, "step": 2323 }, { "epoch": 0.79, "learning_rate": 3.483882533594005e-05, "loss": 0.3524, "step": 2324 }, { "epoch": 0.79, "learning_rate": 3.48259924891377e-05, "loss": 0.3285, "step": 2325 }, { "epoch": 0.79, "learning_rate": 3.481315657950931e-05, "loss": 0.3239, "step": 2326 }, { "epoch": 0.79, "learning_rate": 3.48003176110559e-05, "loss": 0.3651, "step": 2327 }, { "epoch": 0.79, "learning_rate": 3.478747558777949e-05, "loss": 0.3451, "step": 2328 }, { "epoch": 0.79, "learning_rate": 3.4774630513682995e-05, "loss": 0.3242, "step": 2329 }, { "epoch": 0.79, "learning_rate": 3.476178239277031e-05, "loss": 0.3192, "step": 2330 }, { "epoch": 0.8, "learning_rate": 3.4748931229046295e-05, "loss": 0.3222, "step": 2331 }, { "epoch": 0.8, "learning_rate": 3.4736077026516724e-05, "loss": 0.3731, "step": 2332 }, { "epoch": 0.8, "learning_rate": 3.472321978918835e-05, "loss": 0.3818, "step": 2333 }, { "epoch": 0.8, "learning_rate": 3.471035952106885e-05, "loss": 0.3541, "step": 2334 }, { "epoch": 0.8, "learning_rate": 3.469749622616686e-05, "loss": 0.3525, "step": 2335 }, { "epoch": 0.8, "learning_rate": 3.468462990849195e-05, "loss": 0.3241, "step": 2336 }, { "epoch": 0.8, "learning_rate": 3.4671760572054624e-05, "loss": 0.3313, "step": 2337 }, { "epoch": 0.8, "learning_rate": 3.465888822086636e-05, "loss": 0.3377, "step": 2338 }, { "epoch": 0.8, "learning_rate": 3.4646012858939526e-05, "loss": 0.306, "step": 2339 }, { "epoch": 0.8, "learning_rate": 3.463313449028748e-05, "loss": 0.3329, "step": 2340 }, { "epoch": 0.8, "learning_rate": 3.462025311892449e-05, "loss": 0.3022, "step": 2341 }, { "epoch": 0.8, "learning_rate": 3.460736874886576e-05, "loss": 0.3242, "step": 2342 }, { "epoch": 0.8, "learning_rate": 3.4594481384127426e-05, "loss": 0.3285, "step": 2343 }, { "epoch": 0.8, "learning_rate": 3.458159102872658e-05, "loss": 0.3784, "step": 2344 }, { "epoch": 0.8, "learning_rate": 3.456869768668121e-05, "loss": 0.3378, "step": 2345 }, { "epoch": 0.8, "learning_rate": 3.455580136201026e-05, "loss": 0.3229, "step": 2346 }, { "epoch": 0.8, "learning_rate": 3.454290205873362e-05, "loss": 0.3602, "step": 2347 }, { "epoch": 0.8, "learning_rate": 3.452999978087206e-05, "loss": 0.3658, "step": 2348 }, { "epoch": 0.8, "learning_rate": 3.451709453244732e-05, "loss": 0.3349, "step": 2349 }, { "epoch": 0.8, "learning_rate": 3.450418631748203e-05, "loss": 0.3026, "step": 2350 }, { "epoch": 0.8, "learning_rate": 3.44912751399998e-05, "loss": 0.3471, "step": 2351 }, { "epoch": 0.8, "learning_rate": 3.44783610040251e-05, "loss": 0.3114, "step": 2352 }, { "epoch": 0.8, "learning_rate": 3.446544391358336e-05, "loss": 0.3059, "step": 2353 }, { "epoch": 0.8, "learning_rate": 3.445252387270092e-05, "loss": 0.36, "step": 2354 }, { "epoch": 0.8, "learning_rate": 3.443960088540504e-05, "loss": 0.3533, "step": 2355 }, { "epoch": 0.8, "learning_rate": 3.44266749557239e-05, "loss": 0.2955, "step": 2356 }, { "epoch": 0.8, "learning_rate": 3.44137460876866e-05, "loss": 0.3348, "step": 2357 }, { "epoch": 0.8, "learning_rate": 3.4400814285323146e-05, "loss": 0.3277, "step": 2358 }, { "epoch": 0.8, "learning_rate": 3.438787955266447e-05, "loss": 0.3425, "step": 2359 }, { "epoch": 0.81, "learning_rate": 3.437494189374242e-05, "loss": 0.3535, "step": 2360 }, { "epoch": 0.81, "learning_rate": 3.4362001312589733e-05, "loss": 0.3169, "step": 2361 }, { "epoch": 0.81, "learning_rate": 3.4349057813240095e-05, "loss": 0.3216, "step": 2362 }, { "epoch": 0.81, "learning_rate": 3.433611139972804e-05, "loss": 0.3685, "step": 2363 }, { "epoch": 0.81, "learning_rate": 3.43231620760891e-05, "loss": 0.3281, "step": 2364 }, { "epoch": 0.81, "learning_rate": 3.4310209846359624e-05, "loss": 0.338, "step": 2365 }, { "epoch": 0.81, "learning_rate": 3.429725471457692e-05, "loss": 0.3174, "step": 2366 }, { "epoch": 0.81, "learning_rate": 3.42842966847792e-05, "loss": 0.31, "step": 2367 }, { "epoch": 0.81, "learning_rate": 3.4271335761005544e-05, "loss": 0.3218, "step": 2368 }, { "epoch": 0.81, "learning_rate": 3.425837194729597e-05, "loss": 0.3316, "step": 2369 }, { "epoch": 0.81, "learning_rate": 3.424540524769139e-05, "loss": 0.3462, "step": 2370 }, { "epoch": 0.81, "learning_rate": 3.4232435666233585e-05, "loss": 0.3229, "step": 2371 }, { "epoch": 0.81, "learning_rate": 3.421946320696528e-05, "loss": 0.3114, "step": 2372 }, { "epoch": 0.81, "learning_rate": 3.420648787393007e-05, "loss": 0.2831, "step": 2373 }, { "epoch": 0.81, "learning_rate": 3.419350967117245e-05, "loss": 0.3115, "step": 2374 }, { "epoch": 0.81, "learning_rate": 3.418052860273781e-05, "loss": 0.3927, "step": 2375 }, { "epoch": 0.81, "learning_rate": 3.416754467267242e-05, "loss": 0.3468, "step": 2376 }, { "epoch": 0.81, "learning_rate": 3.415455788502349e-05, "loss": 0.317, "step": 2377 }, { "epoch": 0.81, "learning_rate": 3.414156824383906e-05, "loss": 0.2958, "step": 2378 }, { "epoch": 0.81, "learning_rate": 3.412857575316809e-05, "loss": 0.4387, "step": 2379 }, { "epoch": 0.81, "learning_rate": 3.4115580417060423e-05, "loss": 0.2938, "step": 2380 }, { "epoch": 0.81, "learning_rate": 3.41025822395668e-05, "loss": 0.3542, "step": 2381 }, { "epoch": 0.81, "learning_rate": 3.408958122473883e-05, "loss": 0.3658, "step": 2382 }, { "epoch": 0.81, "learning_rate": 3.4076577376629014e-05, "loss": 0.3804, "step": 2383 }, { "epoch": 0.81, "learning_rate": 3.406357069929074e-05, "loss": 0.2952, "step": 2384 }, { "epoch": 0.81, "learning_rate": 3.405056119677827e-05, "loss": 0.3159, "step": 2385 }, { "epoch": 0.81, "learning_rate": 3.4037548873146755e-05, "loss": 0.3347, "step": 2386 }, { "epoch": 0.81, "learning_rate": 3.402453373245222e-05, "loss": 0.3441, "step": 2387 }, { "epoch": 0.81, "learning_rate": 3.401151577875157e-05, "loss": 0.35, "step": 2388 }, { "epoch": 0.82, "learning_rate": 3.3998495016102574e-05, "loss": 0.3159, "step": 2389 }, { "epoch": 0.82, "learning_rate": 3.398547144856391e-05, "loss": 0.3628, "step": 2390 }, { "epoch": 0.82, "learning_rate": 3.3972445080195104e-05, "loss": 0.3371, "step": 2391 }, { "epoch": 0.82, "learning_rate": 3.395941591505655e-05, "loss": 0.3967, "step": 2392 }, { "epoch": 0.82, "learning_rate": 3.394638395720953e-05, "loss": 0.3442, "step": 2393 }, { "epoch": 0.82, "learning_rate": 3.393334921071618e-05, "loss": 0.3422, "step": 2394 }, { "epoch": 0.82, "learning_rate": 3.392031167963953e-05, "loss": 0.3124, "step": 2395 }, { "epoch": 0.82, "learning_rate": 3.3907271368043466e-05, "loss": 0.3706, "step": 2396 }, { "epoch": 0.82, "learning_rate": 3.3894228279992726e-05, "loss": 0.3624, "step": 2397 }, { "epoch": 0.82, "learning_rate": 3.388118241955293e-05, "loss": 0.3233, "step": 2398 }, { "epoch": 0.82, "learning_rate": 3.386813379079056e-05, "loss": 0.3375, "step": 2399 }, { "epoch": 0.82, "learning_rate": 3.385508239777295e-05, "loss": 0.3056, "step": 2400 }, { "epoch": 0.82, "learning_rate": 3.3842028244568324e-05, "loss": 0.335, "step": 2401 }, { "epoch": 0.82, "learning_rate": 3.3828971335245725e-05, "loss": 0.3207, "step": 2402 }, { "epoch": 0.82, "learning_rate": 3.381591167387509e-05, "loss": 0.362, "step": 2403 }, { "epoch": 0.82, "learning_rate": 3.3802849264527205e-05, "loss": 0.3676, "step": 2404 }, { "epoch": 0.82, "learning_rate": 3.378978411127369e-05, "loss": 0.3744, "step": 2405 }, { "epoch": 0.82, "learning_rate": 3.377671621818706e-05, "loss": 0.3508, "step": 2406 }, { "epoch": 0.82, "learning_rate": 3.376364558934064e-05, "loss": 0.3862, "step": 2407 }, { "epoch": 0.82, "learning_rate": 3.375057222880865e-05, "loss": 0.3436, "step": 2408 }, { "epoch": 0.82, "learning_rate": 3.373749614066613e-05, "loss": 0.3004, "step": 2409 }, { "epoch": 0.82, "learning_rate": 3.372441732898899e-05, "loss": 0.3113, "step": 2410 }, { "epoch": 0.82, "learning_rate": 3.3711335797853977e-05, "loss": 0.3483, "step": 2411 }, { "epoch": 0.82, "learning_rate": 3.369825155133869e-05, "loss": 0.2809, "step": 2412 }, { "epoch": 0.82, "learning_rate": 3.3685164593521565e-05, "loss": 0.3241, "step": 2413 }, { "epoch": 0.82, "learning_rate": 3.3672074928481916e-05, "loss": 0.4024, "step": 2414 }, { "epoch": 0.82, "learning_rate": 3.365898256029985e-05, "loss": 0.311, "step": 2415 }, { "epoch": 0.82, "learning_rate": 3.364588749305637e-05, "loss": 0.3243, "step": 2416 }, { "epoch": 0.82, "learning_rate": 3.363278973083327e-05, "loss": 0.3912, "step": 2417 }, { "epoch": 0.82, "learning_rate": 3.3619689277713226e-05, "loss": 0.3064, "step": 2418 }, { "epoch": 0.83, "learning_rate": 3.360658613777972e-05, "loss": 0.3441, "step": 2419 }, { "epoch": 0.83, "learning_rate": 3.3593480315117095e-05, "loss": 0.363, "step": 2420 }, { "epoch": 0.83, "learning_rate": 3.358037181381052e-05, "loss": 0.3244, "step": 2421 }, { "epoch": 0.83, "learning_rate": 3.3567260637946e-05, "loss": 0.3094, "step": 2422 }, { "epoch": 0.83, "learning_rate": 3.355414679161037e-05, "loss": 0.3072, "step": 2423 }, { "epoch": 0.83, "learning_rate": 3.3541030278891315e-05, "loss": 0.3098, "step": 2424 }, { "epoch": 0.83, "learning_rate": 3.352791110387733e-05, "loss": 0.3047, "step": 2425 }, { "epoch": 0.83, "learning_rate": 3.351478927065774e-05, "loss": 0.3906, "step": 2426 }, { "epoch": 0.83, "learning_rate": 3.350166478332272e-05, "loss": 0.2923, "step": 2427 }, { "epoch": 0.83, "learning_rate": 3.348853764596324e-05, "loss": 0.348, "step": 2428 }, { "epoch": 0.83, "learning_rate": 3.347540786267113e-05, "loss": 0.3156, "step": 2429 }, { "epoch": 0.83, "learning_rate": 3.346227543753903e-05, "loss": 0.3241, "step": 2430 }, { "epoch": 0.83, "learning_rate": 3.344914037466039e-05, "loss": 0.347, "step": 2431 }, { "epoch": 0.83, "learning_rate": 3.343600267812951e-05, "loss": 0.2728, "step": 2432 }, { "epoch": 0.83, "learning_rate": 3.342286235204147e-05, "loss": 0.3629, "step": 2433 }, { "epoch": 0.83, "learning_rate": 3.340971940049223e-05, "loss": 0.347, "step": 2434 }, { "epoch": 0.83, "learning_rate": 3.339657382757852e-05, "loss": 0.3655, "step": 2435 }, { "epoch": 0.83, "learning_rate": 3.3383425637397895e-05, "loss": 0.2947, "step": 2436 }, { "epoch": 0.83, "learning_rate": 3.337027483404873e-05, "loss": 0.3316, "step": 2437 }, { "epoch": 0.83, "learning_rate": 3.335712142163022e-05, "loss": 0.3272, "step": 2438 }, { "epoch": 0.83, "learning_rate": 3.334396540424238e-05, "loss": 0.2951, "step": 2439 }, { "epoch": 0.83, "learning_rate": 3.3330806785986015e-05, "loss": 0.3167, "step": 2440 }, { "epoch": 0.83, "learning_rate": 3.3317645570962755e-05, "loss": 0.2966, "step": 2441 }, { "epoch": 0.83, "learning_rate": 3.330448176327503e-05, "loss": 0.3821, "step": 2442 }, { "epoch": 0.83, "learning_rate": 3.32913153670261e-05, "loss": 0.3387, "step": 2443 }, { "epoch": 0.83, "learning_rate": 3.327814638632e-05, "loss": 0.3253, "step": 2444 }, { "epoch": 0.83, "learning_rate": 3.3264974825261595e-05, "loss": 0.3018, "step": 2445 }, { "epoch": 0.83, "learning_rate": 3.3251800687956554e-05, "loss": 0.3385, "step": 2446 }, { "epoch": 0.83, "learning_rate": 3.323862397851133e-05, "loss": 0.3261, "step": 2447 }, { "epoch": 0.84, "learning_rate": 3.32254447010332e-05, "loss": 0.3297, "step": 2448 }, { "epoch": 0.84, "learning_rate": 3.321226285963022e-05, "loss": 0.3122, "step": 2449 }, { "epoch": 0.84, "learning_rate": 3.319907845841126e-05, "loss": 0.3403, "step": 2450 }, { "epoch": 0.84, "learning_rate": 3.318589150148599e-05, "loss": 0.3369, "step": 2451 }, { "epoch": 0.84, "learning_rate": 3.317270199296486e-05, "loss": 0.3463, "step": 2452 }, { "epoch": 0.84, "learning_rate": 3.315950993695913e-05, "loss": 0.3533, "step": 2453 }, { "epoch": 0.84, "learning_rate": 3.314631533758086e-05, "loss": 0.2919, "step": 2454 }, { "epoch": 0.84, "learning_rate": 3.3133118198942885e-05, "loss": 0.3271, "step": 2455 }, { "epoch": 0.84, "learning_rate": 3.311991852515883e-05, "loss": 0.3338, "step": 2456 }, { "epoch": 0.84, "learning_rate": 3.3106716320343134e-05, "loss": 0.3612, "step": 2457 }, { "epoch": 0.84, "learning_rate": 3.3093511588611e-05, "loss": 0.3367, "step": 2458 }, { "epoch": 0.84, "learning_rate": 3.308030433407844e-05, "loss": 0.3518, "step": 2459 }, { "epoch": 0.84, "learning_rate": 3.306709456086223e-05, "loss": 0.3515, "step": 2460 }, { "epoch": 0.84, "learning_rate": 3.305388227307996e-05, "loss": 0.3571, "step": 2461 }, { "epoch": 0.84, "learning_rate": 3.304066747484997e-05, "loss": 0.3413, "step": 2462 }, { "epoch": 0.84, "learning_rate": 3.302745017029139e-05, "loss": 0.3144, "step": 2463 }, { "epoch": 0.84, "learning_rate": 3.301423036352417e-05, "loss": 0.3849, "step": 2464 }, { "epoch": 0.84, "learning_rate": 3.300100805866898e-05, "loss": 0.2997, "step": 2465 }, { "epoch": 0.84, "learning_rate": 3.298778325984732e-05, "loss": 0.3604, "step": 2466 }, { "epoch": 0.84, "learning_rate": 3.297455597118144e-05, "loss": 0.3408, "step": 2467 }, { "epoch": 0.84, "learning_rate": 3.2961326196794367e-05, "loss": 0.3416, "step": 2468 }, { "epoch": 0.84, "learning_rate": 3.294809394080992e-05, "loss": 0.3121, "step": 2469 }, { "epoch": 0.84, "learning_rate": 3.293485920735266e-05, "loss": 0.286, "step": 2470 }, { "epoch": 0.84, "learning_rate": 3.2921622000547956e-05, "loss": 0.326, "step": 2471 }, { "epoch": 0.84, "learning_rate": 3.2908382324521926e-05, "loss": 0.3383, "step": 2472 }, { "epoch": 0.84, "learning_rate": 3.289514018340148e-05, "loss": 0.3163, "step": 2473 }, { "epoch": 0.84, "learning_rate": 3.288189558131425e-05, "loss": 0.3483, "step": 2474 }, { "epoch": 0.84, "learning_rate": 3.286864852238868e-05, "loss": 0.3274, "step": 2475 }, { "epoch": 0.84, "learning_rate": 3.2855399010753965e-05, "loss": 0.3063, "step": 2476 }, { "epoch": 0.85, "learning_rate": 3.284214705054006e-05, "loss": 0.3557, "step": 2477 }, { "epoch": 0.85, "learning_rate": 3.28288926458777e-05, "loss": 0.3595, "step": 2478 }, { "epoch": 0.85, "learning_rate": 3.2815635800898356e-05, "loss": 0.326, "step": 2479 }, { "epoch": 0.85, "learning_rate": 3.2802376519734284e-05, "loss": 0.336, "step": 2480 }, { "epoch": 0.85, "learning_rate": 3.2789114806518486e-05, "loss": 0.3314, "step": 2481 }, { "epoch": 0.85, "learning_rate": 3.277585066538471e-05, "loss": 0.3388, "step": 2482 }, { "epoch": 0.85, "learning_rate": 3.27625841004675e-05, "loss": 0.3005, "step": 2483 }, { "epoch": 0.85, "learning_rate": 3.27493151159021e-05, "loss": 0.2823, "step": 2484 }, { "epoch": 0.85, "learning_rate": 3.273604371582456e-05, "loss": 0.3331, "step": 2485 }, { "epoch": 0.85, "learning_rate": 3.272276990437167e-05, "loss": 0.3432, "step": 2486 }, { "epoch": 0.85, "learning_rate": 3.2709493685680947e-05, "loss": 0.2664, "step": 2487 }, { "epoch": 0.85, "learning_rate": 3.269621506389069e-05, "loss": 0.3105, "step": 2488 }, { "epoch": 0.85, "learning_rate": 3.2682934043139915e-05, "loss": 0.3146, "step": 2489 }, { "epoch": 0.85, "learning_rate": 3.2669650627568405e-05, "loss": 0.3486, "step": 2490 }, { "epoch": 0.85, "learning_rate": 3.26563648213167e-05, "loss": 0.2999, "step": 2491 }, { "epoch": 0.85, "learning_rate": 3.264307662852606e-05, "loss": 0.3232, "step": 2492 }, { "epoch": 0.85, "learning_rate": 3.262978605333852e-05, "loss": 0.3146, "step": 2493 }, { "epoch": 0.85, "learning_rate": 3.261649309989682e-05, "loss": 0.3246, "step": 2494 }, { "epoch": 0.85, "learning_rate": 3.260319777234446e-05, "loss": 0.3733, "step": 2495 }, { "epoch": 0.85, "learning_rate": 3.2589900074825696e-05, "loss": 0.3411, "step": 2496 }, { "epoch": 0.85, "learning_rate": 3.25766000114855e-05, "loss": 0.3399, "step": 2497 }, { "epoch": 0.85, "learning_rate": 3.256329758646958e-05, "loss": 0.2965, "step": 2498 }, { "epoch": 0.85, "learning_rate": 3.2549992803924394e-05, "loss": 0.3338, "step": 2499 }, { "epoch": 0.85, "learning_rate": 3.253668566799714e-05, "loss": 0.4007, "step": 2500 }, { "epoch": 0.85, "learning_rate": 3.252337618283573e-05, "loss": 0.3264, "step": 2501 }, { "epoch": 0.85, "learning_rate": 3.251006435258882e-05, "loss": 0.3374, "step": 2502 }, { "epoch": 0.85, "learning_rate": 3.249675018140578e-05, "loss": 0.3395, "step": 2503 }, { "epoch": 0.85, "learning_rate": 3.2483433673436755e-05, "loss": 0.3551, "step": 2504 }, { "epoch": 0.85, "learning_rate": 3.247011483283255e-05, "loss": 0.3364, "step": 2505 }, { "epoch": 0.85, "learning_rate": 3.245679366374477e-05, "loss": 0.3158, "step": 2506 }, { "epoch": 0.86, "learning_rate": 3.244347017032569e-05, "loss": 0.3424, "step": 2507 }, { "epoch": 0.86, "learning_rate": 3.243014435672833e-05, "loss": 0.298, "step": 2508 }, { "epoch": 0.86, "learning_rate": 3.241681622710644e-05, "loss": 0.3437, "step": 2509 }, { "epoch": 0.86, "learning_rate": 3.240348578561449e-05, "loss": 0.326, "step": 2510 }, { "epoch": 0.86, "learning_rate": 3.239015303640765e-05, "loss": 0.3219, "step": 2511 }, { "epoch": 0.86, "learning_rate": 3.237681798364183e-05, "loss": 0.3406, "step": 2512 }, { "epoch": 0.86, "learning_rate": 3.236348063147366e-05, "loss": 0.3287, "step": 2513 }, { "epoch": 0.86, "learning_rate": 3.2350140984060486e-05, "loss": 0.3456, "step": 2514 }, { "epoch": 0.86, "learning_rate": 3.233679904556034e-05, "loss": 0.3549, "step": 2515 }, { "epoch": 0.86, "learning_rate": 3.2323454820132005e-05, "loss": 0.3302, "step": 2516 }, { "epoch": 0.86, "learning_rate": 3.231010831193497e-05, "loss": 0.3247, "step": 2517 }, { "epoch": 0.86, "learning_rate": 3.229675952512942e-05, "loss": 0.366, "step": 2518 }, { "epoch": 0.86, "learning_rate": 3.228340846387626e-05, "loss": 0.3068, "step": 2519 }, { "epoch": 0.86, "learning_rate": 3.2270055132337105e-05, "loss": 0.2877, "step": 2520 }, { "epoch": 0.86, "learning_rate": 3.225669953467427e-05, "loss": 0.3339, "step": 2521 }, { "epoch": 0.86, "learning_rate": 3.22433416750508e-05, "loss": 0.3223, "step": 2522 }, { "epoch": 0.86, "learning_rate": 3.2229981557630395e-05, "loss": 0.3371, "step": 2523 }, { "epoch": 0.86, "learning_rate": 3.2216619186577524e-05, "loss": 0.3354, "step": 2524 }, { "epoch": 0.86, "learning_rate": 3.2203254566057304e-05, "loss": 0.3292, "step": 2525 }, { "epoch": 0.86, "learning_rate": 3.218988770023558e-05, "loss": 0.2931, "step": 2526 }, { "epoch": 0.86, "learning_rate": 3.217651859327889e-05, "loss": 0.3547, "step": 2527 }, { "epoch": 0.86, "learning_rate": 3.2163147249354484e-05, "loss": 0.2943, "step": 2528 }, { "epoch": 0.86, "learning_rate": 3.214977367263029e-05, "loss": 0.3219, "step": 2529 }, { "epoch": 0.86, "learning_rate": 3.2136397867274925e-05, "loss": 0.3779, "step": 2530 }, { "epoch": 0.86, "learning_rate": 3.212301983745774e-05, "loss": 0.3275, "step": 2531 }, { "epoch": 0.86, "learning_rate": 3.2109639587348734e-05, "loss": 0.3785, "step": 2532 }, { "epoch": 0.86, "learning_rate": 3.2096257121118626e-05, "loss": 0.3111, "step": 2533 }, { "epoch": 0.86, "learning_rate": 3.208287244293882e-05, "loss": 0.3575, "step": 2534 }, { "epoch": 0.86, "learning_rate": 3.20694855569814e-05, "loss": 0.3208, "step": 2535 }, { "epoch": 0.87, "learning_rate": 3.2056096467419156e-05, "loss": 0.3246, "step": 2536 }, { "epoch": 0.87, "learning_rate": 3.204270517842555e-05, "loss": 0.3424, "step": 2537 }, { "epoch": 0.87, "learning_rate": 3.202931169417474e-05, "loss": 0.3536, "step": 2538 }, { "epoch": 0.87, "learning_rate": 3.2015916018841547e-05, "loss": 0.3487, "step": 2539 }, { "epoch": 0.87, "learning_rate": 3.2002518156601504e-05, "loss": 0.3126, "step": 2540 }, { "epoch": 0.87, "learning_rate": 3.198911811163081e-05, "loss": 0.3041, "step": 2541 }, { "epoch": 0.87, "learning_rate": 3.197571588810634e-05, "loss": 0.3466, "step": 2542 }, { "epoch": 0.87, "learning_rate": 3.196231149020566e-05, "loss": 0.333, "step": 2543 }, { "epoch": 0.87, "learning_rate": 3.1948904922107015e-05, "loss": 0.362, "step": 2544 }, { "epoch": 0.87, "learning_rate": 3.193549618798931e-05, "loss": 0.3174, "step": 2545 }, { "epoch": 0.87, "learning_rate": 3.192208529203214e-05, "loss": 0.3694, "step": 2546 }, { "epoch": 0.87, "learning_rate": 3.190867223841576e-05, "loss": 0.3291, "step": 2547 }, { "epoch": 0.87, "learning_rate": 3.189525703132112e-05, "loss": 0.3488, "step": 2548 }, { "epoch": 0.87, "learning_rate": 3.1881839674929815e-05, "loss": 0.3251, "step": 2549 }, { "epoch": 0.87, "learning_rate": 3.186842017342413e-05, "loss": 0.3422, "step": 2550 }, { "epoch": 0.87, "learning_rate": 3.185499853098701e-05, "loss": 0.3471, "step": 2551 }, { "epoch": 0.87, "learning_rate": 3.1841574751802076e-05, "loss": 0.3071, "step": 2552 }, { "epoch": 0.87, "learning_rate": 3.18281488400536e-05, "loss": 0.3242, "step": 2553 }, { "epoch": 0.87, "learning_rate": 3.181472079992651e-05, "loss": 0.2937, "step": 2554 }, { "epoch": 0.87, "learning_rate": 3.180129063560644e-05, "loss": 0.3341, "step": 2555 }, { "epoch": 0.87, "learning_rate": 3.1787858351279653e-05, "loss": 0.3255, "step": 2556 }, { "epoch": 0.87, "learning_rate": 3.177442395113308e-05, "loss": 0.3252, "step": 2557 }, { "epoch": 0.87, "learning_rate": 3.17609874393543e-05, "loss": 0.3051, "step": 2558 }, { "epoch": 0.87, "learning_rate": 3.174754882013158e-05, "loss": 0.37, "step": 2559 }, { "epoch": 0.87, "learning_rate": 3.1734108097653805e-05, "loss": 0.3032, "step": 2560 }, { "epoch": 0.87, "learning_rate": 3.172066527611055e-05, "loss": 0.3378, "step": 2561 }, { "epoch": 0.87, "learning_rate": 3.170722035969203e-05, "loss": 0.2864, "step": 2562 }, { "epoch": 0.87, "learning_rate": 3.16937733525891e-05, "loss": 0.3531, "step": 2563 }, { "epoch": 0.87, "learning_rate": 3.16803242589933e-05, "loss": 0.3099, "step": 2564 }, { "epoch": 0.88, "learning_rate": 3.1666873083096784e-05, "loss": 0.3104, "step": 2565 }, { "epoch": 0.88, "learning_rate": 3.165341982909239e-05, "loss": 0.2914, "step": 2566 }, { "epoch": 0.88, "learning_rate": 3.1639964501173556e-05, "loss": 0.3499, "step": 2567 }, { "epoch": 0.88, "learning_rate": 3.1626507103534415e-05, "loss": 0.3349, "step": 2568 }, { "epoch": 0.88, "learning_rate": 3.161304764036973e-05, "loss": 0.3913, "step": 2569 }, { "epoch": 0.88, "learning_rate": 3.159958611587489e-05, "loss": 0.32, "step": 2570 }, { "epoch": 0.88, "learning_rate": 3.158612253424595e-05, "loss": 0.336, "step": 2571 }, { "epoch": 0.88, "learning_rate": 3.15726568996796e-05, "loss": 0.3859, "step": 2572 }, { "epoch": 0.88, "learning_rate": 3.1559189216373154e-05, "loss": 0.3541, "step": 2573 }, { "epoch": 0.88, "learning_rate": 3.1545719488524576e-05, "loss": 0.3436, "step": 2574 }, { "epoch": 0.88, "learning_rate": 3.1532247720332476e-05, "loss": 0.3394, "step": 2575 }, { "epoch": 0.88, "learning_rate": 3.15187739159961e-05, "loss": 0.3689, "step": 2576 }, { "epoch": 0.88, "learning_rate": 3.15052980797153e-05, "loss": 0.3324, "step": 2577 }, { "epoch": 0.88, "learning_rate": 3.14918202156906e-05, "loss": 0.327, "step": 2578 }, { "epoch": 0.88, "learning_rate": 3.147834032812314e-05, "loss": 0.3707, "step": 2579 }, { "epoch": 0.88, "learning_rate": 3.146485842121466e-05, "loss": 0.3532, "step": 2580 }, { "epoch": 0.88, "learning_rate": 3.1451374499167586e-05, "loss": 0.3106, "step": 2581 }, { "epoch": 0.88, "learning_rate": 3.143788856618495e-05, "loss": 0.3437, "step": 2582 }, { "epoch": 0.88, "learning_rate": 3.142440062647038e-05, "loss": 0.3458, "step": 2583 }, { "epoch": 0.88, "learning_rate": 3.141091068422818e-05, "loss": 0.3583, "step": 2584 }, { "epoch": 0.88, "learning_rate": 3.139741874366323e-05, "loss": 0.3567, "step": 2585 }, { "epoch": 0.88, "learning_rate": 3.138392480898107e-05, "loss": 0.2971, "step": 2586 }, { "epoch": 0.88, "learning_rate": 3.137042888438785e-05, "loss": 0.358, "step": 2587 }, { "epoch": 0.88, "learning_rate": 3.135693097409033e-05, "loss": 0.3674, "step": 2588 }, { "epoch": 0.88, "learning_rate": 3.134343108229589e-05, "loss": 0.3379, "step": 2589 }, { "epoch": 0.88, "learning_rate": 3.1329929213212564e-05, "loss": 0.3111, "step": 2590 }, { "epoch": 0.88, "learning_rate": 3.131642537104894e-05, "loss": 0.3186, "step": 2591 }, { "epoch": 0.88, "learning_rate": 3.130291956001427e-05, "loss": 0.3483, "step": 2592 }, { "epoch": 0.88, "learning_rate": 3.128941178431839e-05, "loss": 0.3157, "step": 2593 }, { "epoch": 0.89, "learning_rate": 3.127590204817178e-05, "loss": 0.3163, "step": 2594 }, { "epoch": 0.89, "learning_rate": 3.126239035578551e-05, "loss": 0.354, "step": 2595 }, { "epoch": 0.89, "learning_rate": 3.124887671137125e-05, "loss": 0.3248, "step": 2596 }, { "epoch": 0.89, "learning_rate": 3.1235361119141305e-05, "loss": 0.3112, "step": 2597 }, { "epoch": 0.89, "learning_rate": 3.1221843583308566e-05, "loss": 0.33, "step": 2598 }, { "epoch": 0.89, "learning_rate": 3.120832410808654e-05, "loss": 0.2996, "step": 2599 }, { "epoch": 0.89, "learning_rate": 3.119480269768933e-05, "loss": 0.3169, "step": 2600 }, { "epoch": 0.89, "learning_rate": 3.118127935633166e-05, "loss": 0.3376, "step": 2601 }, { "epoch": 0.89, "learning_rate": 3.116775408822882e-05, "loss": 0.325, "step": 2602 }, { "epoch": 0.89, "learning_rate": 3.1154226897596754e-05, "loss": 0.3249, "step": 2603 }, { "epoch": 0.89, "learning_rate": 3.1140697788651964e-05, "loss": 0.2811, "step": 2604 }, { "epoch": 0.89, "learning_rate": 3.1127166765611556e-05, "loss": 0.3094, "step": 2605 }, { "epoch": 0.89, "learning_rate": 3.111363383269323e-05, "loss": 0.3301, "step": 2606 }, { "epoch": 0.89, "learning_rate": 3.110009899411532e-05, "loss": 0.3442, "step": 2607 }, { "epoch": 0.89, "learning_rate": 3.10865622540967e-05, "loss": 0.3159, "step": 2608 }, { "epoch": 0.89, "learning_rate": 3.107302361685686e-05, "loss": 0.2932, "step": 2609 }, { "epoch": 0.89, "learning_rate": 3.105948308661589e-05, "loss": 0.3102, "step": 2610 }, { "epoch": 0.89, "learning_rate": 3.104594066759446e-05, "loss": 0.2946, "step": 2611 }, { "epoch": 0.89, "learning_rate": 3.103239636401384e-05, "loss": 0.3043, "step": 2612 }, { "epoch": 0.89, "learning_rate": 3.101885018009585e-05, "loss": 0.2836, "step": 2613 }, { "epoch": 0.89, "learning_rate": 3.1005302120062954e-05, "loss": 0.2664, "step": 2614 }, { "epoch": 0.89, "learning_rate": 3.099175218813816e-05, "loss": 0.3021, "step": 2615 }, { "epoch": 0.89, "learning_rate": 3.097820038854505e-05, "loss": 0.3188, "step": 2616 }, { "epoch": 0.89, "learning_rate": 3.096464672550785e-05, "loss": 0.2533, "step": 2617 }, { "epoch": 0.89, "learning_rate": 3.0951091203251286e-05, "loss": 0.3333, "step": 2618 }, { "epoch": 0.89, "learning_rate": 3.093753382600072e-05, "loss": 0.3874, "step": 2619 }, { "epoch": 0.89, "learning_rate": 3.092397459798208e-05, "loss": 0.2593, "step": 2620 }, { "epoch": 0.89, "learning_rate": 3.0910413523421865e-05, "loss": 0.3162, "step": 2621 }, { "epoch": 0.89, "learning_rate": 3.089685060654714e-05, "loss": 0.316, "step": 2622 }, { "epoch": 0.89, "learning_rate": 3.088328585158556e-05, "loss": 0.3508, "step": 2623 }, { "epoch": 0.9, "learning_rate": 3.086971926276535e-05, "loss": 0.3264, "step": 2624 }, { "epoch": 0.9, "learning_rate": 3.08561508443153e-05, "loss": 0.3169, "step": 2625 }, { "epoch": 0.9, "learning_rate": 3.0842580600464774e-05, "loss": 0.3316, "step": 2626 }, { "epoch": 0.9, "learning_rate": 3.082900853544371e-05, "loss": 0.3187, "step": 2627 }, { "epoch": 0.9, "learning_rate": 3.081543465348262e-05, "loss": 0.3181, "step": 2628 }, { "epoch": 0.9, "learning_rate": 3.080185895881255e-05, "loss": 0.3086, "step": 2629 }, { "epoch": 0.9, "learning_rate": 3.078828145566514e-05, "loss": 0.3527, "step": 2630 }, { "epoch": 0.9, "learning_rate": 3.0774702148272596e-05, "loss": 0.312, "step": 2631 }, { "epoch": 0.9, "learning_rate": 3.076112104086766e-05, "loss": 0.3075, "step": 2632 }, { "epoch": 0.9, "learning_rate": 3.0747538137683675e-05, "loss": 0.348, "step": 2633 }, { "epoch": 0.9, "learning_rate": 3.07339534429545e-05, "loss": 0.3511, "step": 2634 }, { "epoch": 0.9, "learning_rate": 3.072036696091459e-05, "loss": 0.4192, "step": 2635 }, { "epoch": 0.9, "learning_rate": 3.070677869579892e-05, "loss": 0.3311, "step": 2636 }, { "epoch": 0.9, "learning_rate": 3.0693188651843054e-05, "loss": 0.4121, "step": 2637 }, { "epoch": 0.9, "learning_rate": 3.0679596833283094e-05, "loss": 0.3121, "step": 2638 }, { "epoch": 0.9, "learning_rate": 3.066600324435569e-05, "loss": 0.3767, "step": 2639 }, { "epoch": 0.9, "learning_rate": 3.0652407889298086e-05, "loss": 0.3189, "step": 2640 }, { "epoch": 0.9, "learning_rate": 3.0638810772348e-05, "loss": 0.3694, "step": 2641 }, { "epoch": 0.9, "learning_rate": 3.0625211897743776e-05, "loss": 0.3035, "step": 2642 }, { "epoch": 0.9, "learning_rate": 3.061161126972425e-05, "loss": 0.3372, "step": 2643 }, { "epoch": 0.9, "learning_rate": 3.059800889252883e-05, "loss": 0.3337, "step": 2644 }, { "epoch": 0.9, "learning_rate": 3.058440477039747e-05, "loss": 0.3291, "step": 2645 }, { "epoch": 0.9, "learning_rate": 3.0570798907570666e-05, "loss": 0.3098, "step": 2646 }, { "epoch": 0.9, "learning_rate": 3.0557191308289454e-05, "loss": 0.3073, "step": 2647 }, { "epoch": 0.9, "learning_rate": 3.054358197679542e-05, "loss": 0.3155, "step": 2648 }, { "epoch": 0.9, "learning_rate": 3.052997091733065e-05, "loss": 0.326, "step": 2649 }, { "epoch": 0.9, "learning_rate": 3.051635813413784e-05, "loss": 0.2786, "step": 2650 }, { "epoch": 0.9, "learning_rate": 3.050274363146015e-05, "loss": 0.3401, "step": 2651 }, { "epoch": 0.9, "learning_rate": 3.0489127413541325e-05, "loss": 0.3476, "step": 2652 }, { "epoch": 0.91, "learning_rate": 3.0475509484625636e-05, "loss": 0.3024, "step": 2653 }, { "epoch": 0.91, "learning_rate": 3.046188984895787e-05, "loss": 0.3453, "step": 2654 }, { "epoch": 0.91, "learning_rate": 3.044826851078336e-05, "loss": 0.3822, "step": 2655 }, { "epoch": 0.91, "learning_rate": 3.0434645474347966e-05, "loss": 0.335, "step": 2656 }, { "epoch": 0.91, "learning_rate": 3.042102074389807e-05, "loss": 0.3301, "step": 2657 }, { "epoch": 0.91, "learning_rate": 3.04073943236806e-05, "loss": 0.2975, "step": 2658 }, { "epoch": 0.91, "learning_rate": 3.0393766217942994e-05, "loss": 0.3509, "step": 2659 }, { "epoch": 0.91, "learning_rate": 3.0380136430933225e-05, "loss": 0.3159, "step": 2660 }, { "epoch": 0.91, "learning_rate": 3.036650496689979e-05, "loss": 0.3375, "step": 2661 }, { "epoch": 0.91, "learning_rate": 3.03528718300917e-05, "loss": 0.3372, "step": 2662 }, { "epoch": 0.91, "learning_rate": 3.0339237024758492e-05, "loss": 0.3024, "step": 2663 }, { "epoch": 0.91, "learning_rate": 3.0325600555150225e-05, "loss": 0.3567, "step": 2664 }, { "epoch": 0.91, "learning_rate": 3.031196242551747e-05, "loss": 0.3274, "step": 2665 }, { "epoch": 0.91, "learning_rate": 3.029832264011133e-05, "loss": 0.3018, "step": 2666 }, { "epoch": 0.91, "learning_rate": 3.0284681203183417e-05, "loss": 0.3183, "step": 2667 }, { "epoch": 0.91, "learning_rate": 3.027103811898585e-05, "loss": 0.3112, "step": 2668 }, { "epoch": 0.91, "learning_rate": 3.025739339177127e-05, "loss": 0.3287, "step": 2669 }, { "epoch": 0.91, "learning_rate": 3.0243747025792822e-05, "loss": 0.2826, "step": 2670 }, { "epoch": 0.91, "learning_rate": 3.0230099025304175e-05, "loss": 0.3669, "step": 2671 }, { "epoch": 0.91, "learning_rate": 3.02164493945595e-05, "loss": 0.3488, "step": 2672 }, { "epoch": 0.91, "learning_rate": 3.020279813781347e-05, "loss": 0.2474, "step": 2673 }, { "epoch": 0.91, "learning_rate": 3.0189145259321273e-05, "loss": 0.2929, "step": 2674 }, { "epoch": 0.91, "learning_rate": 3.0175490763338603e-05, "loss": 0.2875, "step": 2675 }, { "epoch": 0.91, "learning_rate": 3.0161834654121657e-05, "loss": 0.3454, "step": 2676 }, { "epoch": 0.91, "learning_rate": 3.014817693592712e-05, "loss": 0.3113, "step": 2677 }, { "epoch": 0.91, "learning_rate": 3.0134517613012202e-05, "loss": 0.301, "step": 2678 }, { "epoch": 0.91, "learning_rate": 3.0120856689634613e-05, "loss": 0.323, "step": 2679 }, { "epoch": 0.91, "learning_rate": 3.0107194170052542e-05, "loss": 0.315, "step": 2680 }, { "epoch": 0.91, "learning_rate": 3.0093530058524682e-05, "loss": 0.3267, "step": 2681 }, { "epoch": 0.92, "learning_rate": 3.0079864359310232e-05, "loss": 0.3469, "step": 2682 }, { "epoch": 0.92, "learning_rate": 3.0066197076668878e-05, "loss": 0.3501, "step": 2683 }, { "epoch": 0.92, "learning_rate": 3.0052528214860797e-05, "loss": 0.3166, "step": 2684 }, { "epoch": 0.92, "learning_rate": 3.003885777814667e-05, "loss": 0.3374, "step": 2685 }, { "epoch": 0.92, "learning_rate": 3.0025185770787657e-05, "loss": 0.3247, "step": 2686 }, { "epoch": 0.92, "learning_rate": 3.0011512197045412e-05, "loss": 0.3644, "step": 2687 }, { "epoch": 0.92, "learning_rate": 2.999783706118208e-05, "loss": 0.3192, "step": 2688 }, { "epoch": 0.92, "learning_rate": 2.998416036746029e-05, "loss": 0.2881, "step": 2689 }, { "epoch": 0.92, "learning_rate": 2.997048212014315e-05, "loss": 0.2663, "step": 2690 }, { "epoch": 0.92, "learning_rate": 2.9956802323494266e-05, "loss": 0.3175, "step": 2691 }, { "epoch": 0.92, "learning_rate": 2.9943120981777718e-05, "loss": 0.3412, "step": 2692 }, { "epoch": 0.92, "learning_rate": 2.9929438099258068e-05, "loss": 0.3461, "step": 2693 }, { "epoch": 0.92, "learning_rate": 2.9915753680200366e-05, "loss": 0.3148, "step": 2694 }, { "epoch": 0.92, "learning_rate": 2.9902067728870133e-05, "loss": 0.3129, "step": 2695 }, { "epoch": 0.92, "learning_rate": 2.9888380249533354e-05, "loss": 0.3733, "step": 2696 }, { "epoch": 0.92, "learning_rate": 2.9874691246456527e-05, "loss": 0.3156, "step": 2697 }, { "epoch": 0.92, "learning_rate": 2.9861000723906596e-05, "loss": 0.336, "step": 2698 }, { "epoch": 0.92, "learning_rate": 2.9847308686150982e-05, "loss": 0.321, "step": 2699 }, { "epoch": 0.92, "learning_rate": 2.9833615137457587e-05, "loss": 0.3333, "step": 2700 }, { "epoch": 0.92, "learning_rate": 2.9819920082094783e-05, "loss": 0.3766, "step": 2701 }, { "epoch": 0.92, "learning_rate": 2.98062235243314e-05, "loss": 0.2803, "step": 2702 }, { "epoch": 0.92, "learning_rate": 2.979252546843675e-05, "loss": 0.3199, "step": 2703 }, { "epoch": 0.92, "learning_rate": 2.9778825918680603e-05, "loss": 0.3171, "step": 2704 }, { "epoch": 0.92, "learning_rate": 2.97651248793332e-05, "loss": 0.3223, "step": 2705 }, { "epoch": 0.92, "learning_rate": 2.9751422354665243e-05, "loss": 0.3216, "step": 2706 }, { "epoch": 0.92, "learning_rate": 2.9737718348947902e-05, "loss": 0.3018, "step": 2707 }, { "epoch": 0.92, "learning_rate": 2.9724012866452805e-05, "loss": 0.3085, "step": 2708 }, { "epoch": 0.92, "learning_rate": 2.9710305911452035e-05, "loss": 0.3405, "step": 2709 }, { "epoch": 0.92, "learning_rate": 2.9696597488218147e-05, "loss": 0.3298, "step": 2710 }, { "epoch": 0.92, "learning_rate": 2.9682887601024138e-05, "loss": 0.3123, "step": 2711 }, { "epoch": 0.93, "learning_rate": 2.9669176254143476e-05, "loss": 0.3635, "step": 2712 }, { "epoch": 0.93, "learning_rate": 2.9655463451850074e-05, "loss": 0.3085, "step": 2713 }, { "epoch": 0.93, "learning_rate": 2.9641749198418307e-05, "loss": 0.3573, "step": 2714 }, { "epoch": 0.93, "learning_rate": 2.962803349812299e-05, "loss": 0.3528, "step": 2715 }, { "epoch": 0.93, "learning_rate": 2.961431635523939e-05, "loss": 0.2993, "step": 2716 }, { "epoch": 0.93, "learning_rate": 2.9600597774043255e-05, "loss": 0.3031, "step": 2717 }, { "epoch": 0.93, "learning_rate": 2.9586877758810732e-05, "loss": 0.3602, "step": 2718 }, { "epoch": 0.93, "learning_rate": 2.957315631381845e-05, "loss": 0.3329, "step": 2719 }, { "epoch": 0.93, "learning_rate": 2.955943344334347e-05, "loss": 0.2966, "step": 2720 }, { "epoch": 0.93, "learning_rate": 2.9545709151663308e-05, "loss": 0.2866, "step": 2721 }, { "epoch": 0.93, "learning_rate": 2.95319834430559e-05, "loss": 0.2969, "step": 2722 }, { "epoch": 0.93, "learning_rate": 2.9518256321799653e-05, "loss": 0.3189, "step": 2723 }, { "epoch": 0.93, "learning_rate": 2.9504527792173397e-05, "loss": 0.3444, "step": 2724 }, { "epoch": 0.93, "learning_rate": 2.9490797858456404e-05, "loss": 0.3359, "step": 2725 }, { "epoch": 0.93, "learning_rate": 2.947706652492838e-05, "loss": 0.3217, "step": 2726 }, { "epoch": 0.93, "learning_rate": 2.9463333795869474e-05, "loss": 0.277, "step": 2727 }, { "epoch": 0.93, "learning_rate": 2.944959967556027e-05, "loss": 0.3268, "step": 2728 }, { "epoch": 0.93, "learning_rate": 2.9435864168281772e-05, "loss": 0.3133, "step": 2729 }, { "epoch": 0.93, "learning_rate": 2.9422127278315448e-05, "loss": 0.2643, "step": 2730 }, { "epoch": 0.93, "learning_rate": 2.940838900994316e-05, "loss": 0.321, "step": 2731 }, { "epoch": 0.93, "learning_rate": 2.9394649367447225e-05, "loss": 0.3048, "step": 2732 }, { "epoch": 0.93, "learning_rate": 2.938090835511037e-05, "loss": 0.3106, "step": 2733 }, { "epoch": 0.93, "learning_rate": 2.936716597721576e-05, "loss": 0.3078, "step": 2734 }, { "epoch": 0.93, "learning_rate": 2.9353422238046985e-05, "loss": 0.3894, "step": 2735 }, { "epoch": 0.93, "learning_rate": 2.9339677141888068e-05, "loss": 0.2992, "step": 2736 }, { "epoch": 0.93, "learning_rate": 2.9325930693023433e-05, "loss": 0.3939, "step": 2737 }, { "epoch": 0.93, "learning_rate": 2.9312182895737943e-05, "loss": 0.3387, "step": 2738 }, { "epoch": 0.93, "learning_rate": 2.9298433754316878e-05, "loss": 0.3343, "step": 2739 }, { "epoch": 0.93, "learning_rate": 2.928468327304593e-05, "loss": 0.3214, "step": 2740 }, { "epoch": 0.94, "learning_rate": 2.927093145621122e-05, "loss": 0.3621, "step": 2741 }, { "epoch": 0.94, "learning_rate": 2.925717830809927e-05, "loss": 0.2799, "step": 2742 }, { "epoch": 0.94, "learning_rate": 2.9243423832997037e-05, "loss": 0.3814, "step": 2743 }, { "epoch": 0.94, "learning_rate": 2.922966803519187e-05, "loss": 0.2912, "step": 2744 }, { "epoch": 0.94, "learning_rate": 2.9215910918971557e-05, "loss": 0.328, "step": 2745 }, { "epoch": 0.94, "learning_rate": 2.9202152488624263e-05, "loss": 0.3346, "step": 2746 }, { "epoch": 0.94, "learning_rate": 2.9188392748438588e-05, "loss": 0.3287, "step": 2747 }, { "epoch": 0.94, "learning_rate": 2.9174631702703532e-05, "loss": 0.3473, "step": 2748 }, { "epoch": 0.94, "learning_rate": 2.916086935570851e-05, "loss": 0.3062, "step": 2749 }, { "epoch": 0.94, "learning_rate": 2.914710571174332e-05, "loss": 0.3088, "step": 2750 }, { "epoch": 0.94, "learning_rate": 2.91333407750982e-05, "loss": 0.31, "step": 2751 }, { "epoch": 0.94, "learning_rate": 2.9119574550063755e-05, "loss": 0.3297, "step": 2752 }, { "epoch": 0.94, "learning_rate": 2.910580704093101e-05, "loss": 0.3222, "step": 2753 }, { "epoch": 0.94, "learning_rate": 2.9092038251991393e-05, "loss": 0.3403, "step": 2754 }, { "epoch": 0.94, "learning_rate": 2.9078268187536718e-05, "loss": 0.3813, "step": 2755 }, { "epoch": 0.94, "learning_rate": 2.9064496851859214e-05, "loss": 0.3618, "step": 2756 }, { "epoch": 0.94, "learning_rate": 2.9050724249251488e-05, "loss": 0.308, "step": 2757 }, { "epoch": 0.94, "learning_rate": 2.9036950384006557e-05, "loss": 0.3473, "step": 2758 }, { "epoch": 0.94, "learning_rate": 2.902317526041782e-05, "loss": 0.3248, "step": 2759 }, { "epoch": 0.94, "learning_rate": 2.900939888277906e-05, "loss": 0.3277, "step": 2760 }, { "epoch": 0.94, "learning_rate": 2.8995621255384487e-05, "loss": 0.2908, "step": 2761 }, { "epoch": 0.94, "learning_rate": 2.8981842382528667e-05, "loss": 0.2865, "step": 2762 }, { "epoch": 0.94, "learning_rate": 2.896806226850657e-05, "loss": 0.3604, "step": 2763 }, { "epoch": 0.94, "learning_rate": 2.8954280917613536e-05, "loss": 0.3523, "step": 2764 }, { "epoch": 0.94, "learning_rate": 2.8940498334145316e-05, "loss": 0.3022, "step": 2765 }, { "epoch": 0.94, "learning_rate": 2.8926714522398013e-05, "loss": 0.3214, "step": 2766 }, { "epoch": 0.94, "learning_rate": 2.8912929486668155e-05, "loss": 0.2935, "step": 2767 }, { "epoch": 0.94, "learning_rate": 2.8899143231252605e-05, "loss": 0.302, "step": 2768 }, { "epoch": 0.94, "learning_rate": 2.8885355760448646e-05, "loss": 0.3452, "step": 2769 }, { "epoch": 0.95, "learning_rate": 2.887156707855392e-05, "loss": 0.3295, "step": 2770 }, { "epoch": 0.95, "learning_rate": 2.8857777189866437e-05, "loss": 0.3108, "step": 2771 }, { "epoch": 0.95, "learning_rate": 2.8843986098684606e-05, "loss": 0.3041, "step": 2772 }, { "epoch": 0.95, "learning_rate": 2.8830193809307192e-05, "loss": 0.3344, "step": 2773 }, { "epoch": 0.95, "learning_rate": 2.881640032603336e-05, "loss": 0.3297, "step": 2774 }, { "epoch": 0.95, "learning_rate": 2.8802605653162608e-05, "loss": 0.3574, "step": 2775 }, { "epoch": 0.95, "learning_rate": 2.878880979499483e-05, "loss": 0.3674, "step": 2776 }, { "epoch": 0.95, "learning_rate": 2.87750127558303e-05, "loss": 0.3182, "step": 2777 }, { "epoch": 0.95, "learning_rate": 2.8761214539969626e-05, "loss": 0.3057, "step": 2778 }, { "epoch": 0.95, "learning_rate": 2.8747415151713812e-05, "loss": 0.335, "step": 2779 }, { "epoch": 0.95, "learning_rate": 2.8733614595364216e-05, "loss": 0.359, "step": 2780 }, { "epoch": 0.95, "learning_rate": 2.871981287522256e-05, "loss": 0.3138, "step": 2781 }, { "epoch": 0.95, "learning_rate": 2.8706009995590933e-05, "loss": 0.303, "step": 2782 }, { "epoch": 0.95, "learning_rate": 2.8692205960771782e-05, "loss": 0.2763, "step": 2783 }, { "epoch": 0.95, "learning_rate": 2.8678400775067916e-05, "loss": 0.288, "step": 2784 }, { "epoch": 0.95, "learning_rate": 2.86645944427825e-05, "loss": 0.2932, "step": 2785 }, { "epoch": 0.95, "learning_rate": 2.865078696821905e-05, "loss": 0.3342, "step": 2786 }, { "epoch": 0.95, "learning_rate": 2.863697835568146e-05, "loss": 0.3794, "step": 2787 }, { "epoch": 0.95, "learning_rate": 2.862316860947396e-05, "loss": 0.3289, "step": 2788 }, { "epoch": 0.95, "learning_rate": 2.8609357733901133e-05, "loss": 0.3126, "step": 2789 }, { "epoch": 0.95, "learning_rate": 2.859554573326792e-05, "loss": 0.2998, "step": 2790 }, { "epoch": 0.95, "learning_rate": 2.858173261187962e-05, "loss": 0.2901, "step": 2791 }, { "epoch": 0.95, "learning_rate": 2.8567918374041863e-05, "loss": 0.3665, "step": 2792 }, { "epoch": 0.95, "learning_rate": 2.8554103024060648e-05, "loss": 0.2948, "step": 2793 }, { "epoch": 0.95, "learning_rate": 2.8540286566242298e-05, "loss": 0.3486, "step": 2794 }, { "epoch": 0.95, "learning_rate": 2.8526469004893503e-05, "loss": 0.3342, "step": 2795 }, { "epoch": 0.95, "learning_rate": 2.851265034432128e-05, "loss": 0.2766, "step": 2796 }, { "epoch": 0.95, "learning_rate": 2.8498830588832996e-05, "loss": 0.334, "step": 2797 }, { "epoch": 0.95, "learning_rate": 2.8485009742736368e-05, "loss": 0.3009, "step": 2798 }, { "epoch": 0.95, "learning_rate": 2.8471187810339433e-05, "loss": 0.3246, "step": 2799 }, { "epoch": 0.96, "learning_rate": 2.8457364795950587e-05, "loss": 0.2941, "step": 2800 }, { "epoch": 0.96, "learning_rate": 2.8443540703878547e-05, "loss": 0.3598, "step": 2801 }, { "epoch": 0.96, "learning_rate": 2.8429715538432377e-05, "loss": 0.2653, "step": 2802 }, { "epoch": 0.96, "learning_rate": 2.8415889303921463e-05, "loss": 0.3012, "step": 2803 }, { "epoch": 0.96, "learning_rate": 2.8402062004655544e-05, "loss": 0.3377, "step": 2804 }, { "epoch": 0.96, "learning_rate": 2.8388233644944673e-05, "loss": 0.3209, "step": 2805 }, { "epoch": 0.96, "learning_rate": 2.8374404229099242e-05, "loss": 0.337, "step": 2806 }, { "epoch": 0.96, "learning_rate": 2.8360573761429972e-05, "loss": 0.3181, "step": 2807 }, { "epoch": 0.96, "learning_rate": 2.834674224624791e-05, "loss": 0.3071, "step": 2808 }, { "epoch": 0.96, "learning_rate": 2.8332909687864424e-05, "loss": 0.3316, "step": 2809 }, { "epoch": 0.96, "learning_rate": 2.8319076090591213e-05, "loss": 0.3326, "step": 2810 }, { "epoch": 0.96, "learning_rate": 2.8305241458740305e-05, "loss": 0.3219, "step": 2811 }, { "epoch": 0.96, "learning_rate": 2.8291405796624043e-05, "loss": 0.3186, "step": 2812 }, { "epoch": 0.96, "learning_rate": 2.8277569108555102e-05, "loss": 0.2997, "step": 2813 }, { "epoch": 0.96, "learning_rate": 2.826373139884645e-05, "loss": 0.3026, "step": 2814 }, { "epoch": 0.96, "learning_rate": 2.8249892671811406e-05, "loss": 0.3053, "step": 2815 }, { "epoch": 0.96, "learning_rate": 2.823605293176359e-05, "loss": 0.2969, "step": 2816 }, { "epoch": 0.96, "learning_rate": 2.8222212183016927e-05, "loss": 0.3225, "step": 2817 }, { "epoch": 0.96, "learning_rate": 2.8208370429885677e-05, "loss": 0.3209, "step": 2818 }, { "epoch": 0.96, "learning_rate": 2.8194527676684413e-05, "loss": 0.3105, "step": 2819 }, { "epoch": 0.96, "learning_rate": 2.8180683927728004e-05, "loss": 0.2878, "step": 2820 }, { "epoch": 0.96, "learning_rate": 2.8166839187331636e-05, "loss": 0.3214, "step": 2821 }, { "epoch": 0.96, "learning_rate": 2.8152993459810806e-05, "loss": 0.3378, "step": 2822 }, { "epoch": 0.96, "learning_rate": 2.813914674948132e-05, "loss": 0.3041, "step": 2823 }, { "epoch": 0.96, "learning_rate": 2.8125299060659276e-05, "loss": 0.2817, "step": 2824 }, { "epoch": 0.96, "learning_rate": 2.8111450397661105e-05, "loss": 0.3293, "step": 2825 }, { "epoch": 0.96, "learning_rate": 2.809760076480352e-05, "loss": 0.3317, "step": 2826 }, { "epoch": 0.96, "learning_rate": 2.808375016640355e-05, "loss": 0.3244, "step": 2827 }, { "epoch": 0.96, "learning_rate": 2.8069898606778504e-05, "loss": 0.2874, "step": 2828 }, { "epoch": 0.97, "learning_rate": 2.8056046090246007e-05, "loss": 0.3016, "step": 2829 }, { "epoch": 0.97, "learning_rate": 2.804219262112398e-05, "loss": 0.2883, "step": 2830 }, { "epoch": 0.97, "learning_rate": 2.8028338203730636e-05, "loss": 0.3162, "step": 2831 }, { "epoch": 0.97, "learning_rate": 2.80144828423845e-05, "loss": 0.3126, "step": 2832 }, { "epoch": 0.97, "learning_rate": 2.8000626541404366e-05, "loss": 0.3394, "step": 2833 }, { "epoch": 0.97, "learning_rate": 2.7986769305109338e-05, "loss": 0.3025, "step": 2834 }, { "epoch": 0.97, "learning_rate": 2.7972911137818814e-05, "loss": 0.3203, "step": 2835 }, { "epoch": 0.97, "learning_rate": 2.7959052043852464e-05, "loss": 0.322, "step": 2836 }, { "epoch": 0.97, "learning_rate": 2.7945192027530254e-05, "loss": 0.3578, "step": 2837 }, { "epoch": 0.97, "learning_rate": 2.793133109317245e-05, "loss": 0.3276, "step": 2838 }, { "epoch": 0.97, "learning_rate": 2.7917469245099602e-05, "loss": 0.3193, "step": 2839 }, { "epoch": 0.97, "learning_rate": 2.7903606487632532e-05, "loss": 0.3534, "step": 2840 }, { "epoch": 0.97, "learning_rate": 2.7889742825092358e-05, "loss": 0.374, "step": 2841 }, { "epoch": 0.97, "learning_rate": 2.7875878261800464e-05, "loss": 0.329, "step": 2842 }, { "epoch": 0.97, "learning_rate": 2.7862012802078525e-05, "loss": 0.337, "step": 2843 }, { "epoch": 0.97, "learning_rate": 2.7848146450248502e-05, "loss": 0.3612, "step": 2844 }, { "epoch": 0.97, "learning_rate": 2.7834279210632623e-05, "loss": 0.3118, "step": 2845 }, { "epoch": 0.97, "learning_rate": 2.7820411087553404e-05, "loss": 0.3733, "step": 2846 }, { "epoch": 0.97, "learning_rate": 2.7806542085333627e-05, "loss": 0.2939, "step": 2847 }, { "epoch": 0.97, "learning_rate": 2.779267220829635e-05, "loss": 0.292, "step": 2848 }, { "epoch": 0.97, "learning_rate": 2.77788014607649e-05, "loss": 0.3048, "step": 2849 }, { "epoch": 0.97, "learning_rate": 2.7764929847062876e-05, "loss": 0.3016, "step": 2850 }, { "epoch": 0.97, "learning_rate": 2.775105737151416e-05, "loss": 0.3519, "step": 2851 }, { "epoch": 0.97, "learning_rate": 2.7737184038442894e-05, "loss": 0.325, "step": 2852 }, { "epoch": 0.97, "learning_rate": 2.772330985217348e-05, "loss": 0.3268, "step": 2853 }, { "epoch": 0.97, "learning_rate": 2.77094348170306e-05, "loss": 0.3394, "step": 2854 }, { "epoch": 0.97, "learning_rate": 2.7695558937339183e-05, "loss": 0.3076, "step": 2855 }, { "epoch": 0.97, "learning_rate": 2.7681682217424437e-05, "loss": 0.3005, "step": 2856 }, { "epoch": 0.97, "learning_rate": 2.766780466161183e-05, "loss": 0.2886, "step": 2857 }, { "epoch": 0.98, "learning_rate": 2.7653926274227077e-05, "loss": 0.3065, "step": 2858 }, { "epoch": 0.98, "learning_rate": 2.7640047059596176e-05, "loss": 0.2826, "step": 2859 }, { "epoch": 0.98, "learning_rate": 2.7626167022045357e-05, "loss": 0.3298, "step": 2860 }, { "epoch": 0.98, "learning_rate": 2.7612286165901125e-05, "loss": 0.2817, "step": 2861 }, { "epoch": 0.98, "learning_rate": 2.7598404495490226e-05, "loss": 0.3745, "step": 2862 }, { "epoch": 0.98, "learning_rate": 2.758452201513967e-05, "loss": 0.2956, "step": 2863 }, { "epoch": 0.98, "learning_rate": 2.7570638729176723e-05, "loss": 0.3011, "step": 2864 }, { "epoch": 0.98, "learning_rate": 2.755675464192889e-05, "loss": 0.321, "step": 2865 }, { "epoch": 0.98, "learning_rate": 2.7542869757723933e-05, "loss": 0.2936, "step": 2866 }, { "epoch": 0.98, "learning_rate": 2.7528984080889864e-05, "loss": 0.2859, "step": 2867 }, { "epoch": 0.98, "learning_rate": 2.7515097615754936e-05, "loss": 0.3188, "step": 2868 }, { "epoch": 0.98, "learning_rate": 2.750121036664765e-05, "loss": 0.2878, "step": 2869 }, { "epoch": 0.98, "learning_rate": 2.7487322337896754e-05, "loss": 0.3279, "step": 2870 }, { "epoch": 0.98, "learning_rate": 2.7473433533831243e-05, "loss": 0.2676, "step": 2871 }, { "epoch": 0.98, "learning_rate": 2.745954395878033e-05, "loss": 0.2977, "step": 2872 }, { "epoch": 0.98, "learning_rate": 2.744565361707351e-05, "loss": 0.309, "step": 2873 }, { "epoch": 0.98, "learning_rate": 2.7431762513040477e-05, "loss": 0.3097, "step": 2874 }, { "epoch": 0.98, "learning_rate": 2.7417870651011185e-05, "loss": 0.3512, "step": 2875 }, { "epoch": 0.98, "learning_rate": 2.7403978035315813e-05, "loss": 0.3231, "step": 2876 }, { "epoch": 0.98, "learning_rate": 2.7390084670284788e-05, "loss": 0.3212, "step": 2877 }, { "epoch": 0.98, "learning_rate": 2.737619056024876e-05, "loss": 0.3314, "step": 2878 }, { "epoch": 0.98, "learning_rate": 2.7362295709538606e-05, "loss": 0.2878, "step": 2879 }, { "epoch": 0.98, "learning_rate": 2.734840012248545e-05, "loss": 0.2851, "step": 2880 }, { "epoch": 0.98, "learning_rate": 2.7334503803420636e-05, "loss": 0.3421, "step": 2881 }, { "epoch": 0.98, "learning_rate": 2.732060675667574e-05, "loss": 0.3129, "step": 2882 }, { "epoch": 0.98, "learning_rate": 2.7306708986582553e-05, "loss": 0.3211, "step": 2883 }, { "epoch": 0.98, "learning_rate": 2.729281049747311e-05, "loss": 0.3462, "step": 2884 }, { "epoch": 0.98, "learning_rate": 2.727891129367966e-05, "loss": 0.3069, "step": 2885 }, { "epoch": 0.98, "learning_rate": 2.726501137953466e-05, "loss": 0.3142, "step": 2886 }, { "epoch": 0.98, "learning_rate": 2.725111075937083e-05, "loss": 0.358, "step": 2887 }, { "epoch": 0.99, "learning_rate": 2.723720943752107e-05, "loss": 0.319, "step": 2888 }, { "epoch": 0.99, "learning_rate": 2.722330741831851e-05, "loss": 0.3147, "step": 2889 }, { "epoch": 0.99, "learning_rate": 2.720940470609651e-05, "loss": 0.3243, "step": 2890 }, { "epoch": 0.99, "learning_rate": 2.7195501305188624e-05, "loss": 0.3191, "step": 2891 }, { "epoch": 0.99, "learning_rate": 2.7181597219928645e-05, "loss": 0.3478, "step": 2892 }, { "epoch": 0.99, "learning_rate": 2.716769245465055e-05, "loss": 0.3504, "step": 2893 }, { "epoch": 0.99, "learning_rate": 2.715378701368857e-05, "loss": 0.3426, "step": 2894 }, { "epoch": 0.99, "learning_rate": 2.71398809013771e-05, "loss": 0.3443, "step": 2895 }, { "epoch": 0.99, "learning_rate": 2.712597412205078e-05, "loss": 0.3409, "step": 2896 }, { "epoch": 0.99, "learning_rate": 2.711206668004444e-05, "loss": 0.3396, "step": 2897 }, { "epoch": 0.99, "learning_rate": 2.7098158579693116e-05, "loss": 0.3113, "step": 2898 }, { "epoch": 0.99, "learning_rate": 2.708424982533206e-05, "loss": 0.3362, "step": 2899 }, { "epoch": 0.99, "learning_rate": 2.707034042129671e-05, "loss": 0.3454, "step": 2900 }, { "epoch": 0.99, "learning_rate": 2.705643037192273e-05, "loss": 0.3573, "step": 2901 }, { "epoch": 0.99, "learning_rate": 2.7042519681545968e-05, "loss": 0.3144, "step": 2902 }, { "epoch": 0.99, "learning_rate": 2.702860835450248e-05, "loss": 0.3302, "step": 2903 }, { "epoch": 0.99, "learning_rate": 2.7014696395128523e-05, "loss": 0.3212, "step": 2904 }, { "epoch": 0.99, "learning_rate": 2.7000783807760533e-05, "loss": 0.3048, "step": 2905 }, { "epoch": 0.99, "learning_rate": 2.6986870596735164e-05, "loss": 0.3072, "step": 2906 }, { "epoch": 0.99, "learning_rate": 2.6972956766389245e-05, "loss": 0.3235, "step": 2907 }, { "epoch": 0.99, "learning_rate": 2.6959042321059814e-05, "loss": 0.3386, "step": 2908 }, { "epoch": 0.99, "learning_rate": 2.69451272650841e-05, "loss": 0.3045, "step": 2909 }, { "epoch": 0.99, "learning_rate": 2.6931211602799505e-05, "loss": 0.3135, "step": 2910 }, { "epoch": 0.99, "learning_rate": 2.6917295338543647e-05, "loss": 0.2985, "step": 2911 }, { "epoch": 0.99, "learning_rate": 2.69033784766543e-05, "loss": 0.3195, "step": 2912 }, { "epoch": 0.99, "learning_rate": 2.6889461021469453e-05, "loss": 0.2911, "step": 2913 }, { "epoch": 0.99, "learning_rate": 2.6875542977327257e-05, "loss": 0.2812, "step": 2914 }, { "epoch": 0.99, "learning_rate": 2.686162434856606e-05, "loss": 0.3033, "step": 2915 }, { "epoch": 0.99, "learning_rate": 2.6847705139524403e-05, "loss": 0.273, "step": 2916 }, { "epoch": 1.0, "learning_rate": 2.683378535454098e-05, "loss": 0.284, "step": 2917 }, { "epoch": 1.0, "learning_rate": 2.6819864997954685e-05, "loss": 0.338, "step": 2918 }, { "epoch": 1.0, "learning_rate": 2.6805944074104582e-05, "loss": 0.3286, "step": 2919 }, { "epoch": 1.0, "learning_rate": 2.6792022587329912e-05, "loss": 0.2931, "step": 2920 }, { "epoch": 1.0, "learning_rate": 2.677810054197009e-05, "loss": 0.3398, "step": 2921 }, { "epoch": 1.0, "learning_rate": 2.676417794236472e-05, "loss": 0.3692, "step": 2922 }, { "epoch": 1.0, "learning_rate": 2.6750254792853563e-05, "loss": 0.3583, "step": 2923 }, { "epoch": 1.0, "learning_rate": 2.6736331097776552e-05, "loss": 0.4662, "step": 2924 }, { "epoch": 1.0, "learning_rate": 2.67224068614738e-05, "loss": 0.3385, "step": 2925 }, { "epoch": 1.0, "learning_rate": 2.6708482088285567e-05, "loss": 0.3338, "step": 2926 }, { "epoch": 1.0, "learning_rate": 2.6694556782552317e-05, "loss": 0.3011, "step": 2927 }, { "epoch": 1.0, "learning_rate": 2.668063094861464e-05, "loss": 0.3895, "step": 2928 }, { "epoch": 1.0, "learning_rate": 2.666670459081332e-05, "loss": 0.3016, "step": 2929 }, { "epoch": 1.0, "learning_rate": 2.6652777713489296e-05, "loss": 0.2765, "step": 2930 }, { "epoch": 1.0, "learning_rate": 2.663885032098366e-05, "loss": 0.2915, "step": 2931 }, { "epoch": 1.0, "eval_loss": 0.32550960779190063, "eval_runtime": 82.7705, "eval_samples_per_second": 9.206, "eval_steps_per_second": 0.58, "step": 2931 }, { "epoch": 1.0, "learning_rate": 2.6624922417637672e-05, "loss": 0.2319, "step": 2932 }, { "epoch": 1.0, "learning_rate": 2.661099400779275e-05, "loss": 0.2446, "step": 2933 }, { "epoch": 1.0, "learning_rate": 2.6597065095790476e-05, "loss": 0.2427, "step": 2934 }, { "epoch": 1.0, "learning_rate": 2.6583135685972572e-05, "loss": 0.2668, "step": 2935 }, { "epoch": 1.0, "learning_rate": 2.656920578268094e-05, "loss": 0.255, "step": 2936 }, { "epoch": 1.0, "learning_rate": 2.6555275390257618e-05, "loss": 0.2442, "step": 2937 }, { "epoch": 1.0, "learning_rate": 2.654134451304479e-05, "loss": 0.2198, "step": 2938 }, { "epoch": 1.0, "learning_rate": 2.6527413155384818e-05, "loss": 0.2289, "step": 2939 }, { "epoch": 1.0, "learning_rate": 2.6513481321620176e-05, "loss": 0.23, "step": 2940 }, { "epoch": 1.0, "learning_rate": 2.649954901609352e-05, "loss": 0.2252, "step": 2941 }, { "epoch": 1.0, "learning_rate": 2.6485616243147645e-05, "loss": 0.2442, "step": 2942 }, { "epoch": 1.0, "learning_rate": 2.6471683007125485e-05, "loss": 0.2256, "step": 2943 }, { "epoch": 1.0, "learning_rate": 2.6457749312370118e-05, "loss": 0.2073, "step": 2944 }, { "epoch": 1.0, "learning_rate": 2.644381516322476e-05, "loss": 0.2695, "step": 2945 }, { "epoch": 1.01, "learning_rate": 2.642988056403279e-05, "loss": 0.2388, "step": 2946 }, { "epoch": 1.01, "learning_rate": 2.6415945519137707e-05, "loss": 0.2431, "step": 2947 }, { "epoch": 1.01, "learning_rate": 2.6402010032883157e-05, "loss": 0.2139, "step": 2948 }, { "epoch": 1.01, "learning_rate": 2.6388074109612915e-05, "loss": 0.226, "step": 2949 }, { "epoch": 1.01, "learning_rate": 2.6374137753670912e-05, "loss": 0.2369, "step": 2950 }, { "epoch": 1.01, "learning_rate": 2.63602009694012e-05, "loss": 0.2409, "step": 2951 }, { "epoch": 1.01, "learning_rate": 2.6346263761147954e-05, "loss": 0.2362, "step": 2952 }, { "epoch": 1.01, "learning_rate": 2.6332326133255497e-05, "loss": 0.2214, "step": 2953 }, { "epoch": 1.01, "learning_rate": 2.631838809006828e-05, "loss": 0.2033, "step": 2954 }, { "epoch": 1.01, "learning_rate": 2.6304449635930882e-05, "loss": 0.2518, "step": 2955 }, { "epoch": 1.01, "learning_rate": 2.6290510775188022e-05, "loss": 0.2995, "step": 2956 }, { "epoch": 1.01, "learning_rate": 2.6276571512184515e-05, "loss": 0.2312, "step": 2957 }, { "epoch": 1.01, "learning_rate": 2.6262631851265333e-05, "loss": 0.2397, "step": 2958 }, { "epoch": 1.01, "learning_rate": 2.624869179677555e-05, "loss": 0.1944, "step": 2959 }, { "epoch": 1.01, "learning_rate": 2.6234751353060382e-05, "loss": 0.233, "step": 2960 }, { "epoch": 1.01, "learning_rate": 2.622081052446515e-05, "loss": 0.2417, "step": 2961 }, { "epoch": 1.01, "learning_rate": 2.6206869315335296e-05, "loss": 0.237, "step": 2962 }, { "epoch": 1.01, "learning_rate": 2.6192927730016404e-05, "loss": 0.216, "step": 2963 }, { "epoch": 1.01, "learning_rate": 2.6178985772854142e-05, "loss": 0.2545, "step": 2964 }, { "epoch": 1.01, "learning_rate": 2.616504344819431e-05, "loss": 0.2245, "step": 2965 }, { "epoch": 1.01, "learning_rate": 2.6151100760382823e-05, "loss": 0.2566, "step": 2966 }, { "epoch": 1.01, "learning_rate": 2.6137157713765713e-05, "loss": 0.2468, "step": 2967 }, { "epoch": 1.01, "learning_rate": 2.612321431268912e-05, "loss": 0.2272, "step": 2968 }, { "epoch": 1.01, "learning_rate": 2.6109270561499278e-05, "loss": 0.2335, "step": 2969 }, { "epoch": 1.01, "learning_rate": 2.6095326464542564e-05, "loss": 0.2468, "step": 2970 }, { "epoch": 1.01, "learning_rate": 2.6081382026165435e-05, "loss": 0.2266, "step": 2971 }, { "epoch": 1.01, "learning_rate": 2.606743725071446e-05, "loss": 0.2636, "step": 2972 }, { "epoch": 1.01, "learning_rate": 2.6053492142536324e-05, "loss": 0.2435, "step": 2973 }, { "epoch": 1.01, "learning_rate": 2.6039546705977807e-05, "loss": 0.2223, "step": 2974 }, { "epoch": 1.02, "learning_rate": 2.6025600945385792e-05, "loss": 0.2497, "step": 2975 }, { "epoch": 1.02, "learning_rate": 2.6011654865107256e-05, "loss": 0.2497, "step": 2976 }, { "epoch": 1.02, "learning_rate": 2.5997708469489297e-05, "loss": 0.2239, "step": 2977 }, { "epoch": 1.02, "learning_rate": 2.5983761762879082e-05, "loss": 0.203, "step": 2978 }, { "epoch": 1.02, "learning_rate": 2.5969814749623906e-05, "loss": 0.2354, "step": 2979 }, { "epoch": 1.02, "learning_rate": 2.5955867434071134e-05, "loss": 0.2691, "step": 2980 }, { "epoch": 1.02, "learning_rate": 2.5941919820568245e-05, "loss": 0.2536, "step": 2981 }, { "epoch": 1.02, "learning_rate": 2.592797191346279e-05, "loss": 0.2443, "step": 2982 }, { "epoch": 1.02, "learning_rate": 2.5914023717102426e-05, "loss": 0.2289, "step": 2983 }, { "epoch": 1.02, "learning_rate": 2.5900075235834902e-05, "loss": 0.2559, "step": 2984 }, { "epoch": 1.02, "learning_rate": 2.5886126474008044e-05, "loss": 0.2239, "step": 2985 }, { "epoch": 1.02, "learning_rate": 2.587217743596978e-05, "loss": 0.2363, "step": 2986 }, { "epoch": 1.02, "learning_rate": 2.5858228126068114e-05, "loss": 0.2383, "step": 2987 }, { "epoch": 1.02, "learning_rate": 2.584427854865113e-05, "loss": 0.2332, "step": 2988 }, { "epoch": 1.02, "learning_rate": 2.583032870806701e-05, "loss": 0.2231, "step": 2989 }, { "epoch": 1.02, "learning_rate": 2.5816378608664e-05, "loss": 0.235, "step": 2990 }, { "epoch": 1.02, "learning_rate": 2.580242825479045e-05, "loss": 0.204, "step": 2991 }, { "epoch": 1.02, "learning_rate": 2.578847765079477e-05, "loss": 0.2194, "step": 2992 }, { "epoch": 1.02, "learning_rate": 2.577452680102546e-05, "loss": 0.2616, "step": 2993 }, { "epoch": 1.02, "learning_rate": 2.5760575709831086e-05, "loss": 0.215, "step": 2994 }, { "epoch": 1.02, "learning_rate": 2.574662438156029e-05, "loss": 0.2366, "step": 2995 }, { "epoch": 1.02, "learning_rate": 2.5732672820561804e-05, "loss": 0.2304, "step": 2996 }, { "epoch": 1.02, "learning_rate": 2.5718721031184405e-05, "loss": 0.2136, "step": 2997 }, { "epoch": 1.02, "learning_rate": 2.5704769017776965e-05, "loss": 0.217, "step": 2998 }, { "epoch": 1.02, "learning_rate": 2.5690816784688427e-05, "loss": 0.2306, "step": 2999 }, { "epoch": 1.02, "learning_rate": 2.567686433626778e-05, "loss": 0.2238, "step": 3000 }, { "epoch": 1.02, "learning_rate": 2.5662911676864104e-05, "loss": 0.258, "step": 3001 }, { "epoch": 1.02, "learning_rate": 2.5648958810826524e-05, "loss": 0.2433, "step": 3002 }, { "epoch": 1.02, "learning_rate": 2.5635005742504238e-05, "loss": 0.2205, "step": 3003 }, { "epoch": 1.02, "learning_rate": 2.562105247624652e-05, "loss": 0.2464, "step": 3004 }, { "epoch": 1.03, "learning_rate": 2.560709901640268e-05, "loss": 0.202, "step": 3005 }, { "epoch": 1.03, "learning_rate": 2.559314536732212e-05, "loss": 0.2308, "step": 3006 }, { "epoch": 1.03, "learning_rate": 2.5579191533354274e-05, "loss": 0.2077, "step": 3007 }, { "epoch": 1.03, "learning_rate": 2.5565237518848644e-05, "loss": 0.2574, "step": 3008 }, { "epoch": 1.03, "learning_rate": 2.5551283328154784e-05, "loss": 0.2194, "step": 3009 }, { "epoch": 1.03, "learning_rate": 2.5537328965622303e-05, "loss": 0.2204, "step": 3010 }, { "epoch": 1.03, "learning_rate": 2.552337443560088e-05, "loss": 0.214, "step": 3011 }, { "epoch": 1.03, "learning_rate": 2.5509419742440227e-05, "loss": 0.2199, "step": 3012 }, { "epoch": 1.03, "learning_rate": 2.5495464890490113e-05, "loss": 0.2224, "step": 3013 }, { "epoch": 1.03, "learning_rate": 2.5481509884100357e-05, "loss": 0.224, "step": 3014 }, { "epoch": 1.03, "learning_rate": 2.5467554727620825e-05, "loss": 0.2144, "step": 3015 }, { "epoch": 1.03, "learning_rate": 2.5453599425401424e-05, "loss": 0.2113, "step": 3016 }, { "epoch": 1.03, "learning_rate": 2.543964398179213e-05, "loss": 0.2462, "step": 3017 }, { "epoch": 1.03, "learning_rate": 2.5425688401142915e-05, "loss": 0.2521, "step": 3018 }, { "epoch": 1.03, "learning_rate": 2.5411732687803862e-05, "loss": 0.2222, "step": 3019 }, { "epoch": 1.03, "learning_rate": 2.5397776846125037e-05, "loss": 0.255, "step": 3020 }, { "epoch": 1.03, "learning_rate": 2.538382088045657e-05, "loss": 0.2543, "step": 3021 }, { "epoch": 1.03, "learning_rate": 2.536986479514863e-05, "loss": 0.2183, "step": 3022 }, { "epoch": 1.03, "learning_rate": 2.5355908594551414e-05, "loss": 0.2323, "step": 3023 }, { "epoch": 1.03, "learning_rate": 2.5341952283015162e-05, "loss": 0.2147, "step": 3024 }, { "epoch": 1.03, "learning_rate": 2.5327995864890152e-05, "loss": 0.2153, "step": 3025 }, { "epoch": 1.03, "learning_rate": 2.5314039344526693e-05, "loss": 0.2214, "step": 3026 }, { "epoch": 1.03, "learning_rate": 2.5300082726275108e-05, "loss": 0.2065, "step": 3027 }, { "epoch": 1.03, "learning_rate": 2.5286126014485788e-05, "loss": 0.2457, "step": 3028 }, { "epoch": 1.03, "learning_rate": 2.5272169213509112e-05, "loss": 0.2473, "step": 3029 }, { "epoch": 1.03, "learning_rate": 2.525821232769552e-05, "loss": 0.2508, "step": 3030 }, { "epoch": 1.03, "learning_rate": 2.524425536139546e-05, "loss": 0.225, "step": 3031 }, { "epoch": 1.03, "learning_rate": 2.52302983189594e-05, "loss": 0.2038, "step": 3032 }, { "epoch": 1.03, "learning_rate": 2.5216341204737855e-05, "loss": 0.2413, "step": 3033 }, { "epoch": 1.04, "learning_rate": 2.5202384023081344e-05, "loss": 0.2182, "step": 3034 }, { "epoch": 1.04, "learning_rate": 2.5188426778340414e-05, "loss": 0.2234, "step": 3035 }, { "epoch": 1.04, "learning_rate": 2.5174469474865613e-05, "loss": 0.2098, "step": 3036 }, { "epoch": 1.04, "learning_rate": 2.516051211700755e-05, "loss": 0.2639, "step": 3037 }, { "epoch": 1.04, "learning_rate": 2.5146554709116808e-05, "loss": 0.251, "step": 3038 }, { "epoch": 1.04, "learning_rate": 2.5132597255543998e-05, "loss": 0.2324, "step": 3039 }, { "epoch": 1.04, "learning_rate": 2.511863976063977e-05, "loss": 0.2237, "step": 3040 }, { "epoch": 1.04, "learning_rate": 2.510468222875475e-05, "loss": 0.2273, "step": 3041 }, { "epoch": 1.04, "learning_rate": 2.509072466423959e-05, "loss": 0.2269, "step": 3042 }, { "epoch": 1.04, "learning_rate": 2.5076767071444962e-05, "loss": 0.2629, "step": 3043 }, { "epoch": 1.04, "learning_rate": 2.506280945472154e-05, "loss": 0.2287, "step": 3044 }, { "epoch": 1.04, "learning_rate": 2.504885181842e-05, "loss": 0.2158, "step": 3045 }, { "epoch": 1.04, "learning_rate": 2.5034894166891027e-05, "loss": 0.2285, "step": 3046 }, { "epoch": 1.04, "learning_rate": 2.5020936504485313e-05, "loss": 0.2011, "step": 3047 }, { "epoch": 1.04, "learning_rate": 2.5006978835553556e-05, "loss": 0.2302, "step": 3048 }, { "epoch": 1.04, "learning_rate": 2.499302116444645e-05, "loss": 0.2686, "step": 3049 }, { "epoch": 1.04, "learning_rate": 2.497906349551469e-05, "loss": 0.2664, "step": 3050 }, { "epoch": 1.04, "learning_rate": 2.4965105833108982e-05, "loss": 0.2482, "step": 3051 }, { "epoch": 1.04, "learning_rate": 2.495114818158001e-05, "loss": 0.2505, "step": 3052 }, { "epoch": 1.04, "learning_rate": 2.4937190545278468e-05, "loss": 0.2388, "step": 3053 }, { "epoch": 1.04, "learning_rate": 2.492323292855504e-05, "loss": 0.2442, "step": 3054 }, { "epoch": 1.04, "learning_rate": 2.4909275335760415e-05, "loss": 0.2052, "step": 3055 }, { "epoch": 1.04, "learning_rate": 2.4895317771245256e-05, "loss": 0.2212, "step": 3056 }, { "epoch": 1.04, "learning_rate": 2.488136023936023e-05, "loss": 0.2201, "step": 3057 }, { "epoch": 1.04, "learning_rate": 2.4867402744455998e-05, "loss": 0.2054, "step": 3058 }, { "epoch": 1.04, "learning_rate": 2.4853445290883194e-05, "loss": 0.2429, "step": 3059 }, { "epoch": 1.04, "learning_rate": 2.4839487882992456e-05, "loss": 0.233, "step": 3060 }, { "epoch": 1.04, "learning_rate": 2.482553052513439e-05, "loss": 0.2279, "step": 3061 }, { "epoch": 1.04, "learning_rate": 2.4811573221659595e-05, "loss": 0.2291, "step": 3062 }, { "epoch": 1.05, "learning_rate": 2.4797615976918662e-05, "loss": 0.2008, "step": 3063 }, { "epoch": 1.05, "learning_rate": 2.478365879526215e-05, "loss": 0.2435, "step": 3064 }, { "epoch": 1.05, "learning_rate": 2.476970168104061e-05, "loss": 0.235, "step": 3065 }, { "epoch": 1.05, "learning_rate": 2.4755744638604554e-05, "loss": 0.2256, "step": 3066 }, { "epoch": 1.05, "learning_rate": 2.4741787672304486e-05, "loss": 0.2273, "step": 3067 }, { "epoch": 1.05, "learning_rate": 2.4727830786490894e-05, "loss": 0.2341, "step": 3068 }, { "epoch": 1.05, "learning_rate": 2.471387398551422e-05, "loss": 0.2425, "step": 3069 }, { "epoch": 1.05, "learning_rate": 2.4699917273724887e-05, "loss": 0.2113, "step": 3070 }, { "epoch": 1.05, "learning_rate": 2.468596065547331e-05, "loss": 0.2194, "step": 3071 }, { "epoch": 1.05, "learning_rate": 2.467200413510985e-05, "loss": 0.2506, "step": 3072 }, { "epoch": 1.05, "learning_rate": 2.465804771698484e-05, "loss": 0.2046, "step": 3073 }, { "epoch": 1.05, "learning_rate": 2.4644091405448592e-05, "loss": 0.2341, "step": 3074 }, { "epoch": 1.05, "learning_rate": 2.4630135204851373e-05, "loss": 0.2098, "step": 3075 }, { "epoch": 1.05, "learning_rate": 2.4616179119543433e-05, "loss": 0.2209, "step": 3076 }, { "epoch": 1.05, "learning_rate": 2.460222315387497e-05, "loss": 0.2071, "step": 3077 }, { "epoch": 1.05, "learning_rate": 2.4588267312196144e-05, "loss": 0.2278, "step": 3078 }, { "epoch": 1.05, "learning_rate": 2.4574311598857088e-05, "loss": 0.2397, "step": 3079 }, { "epoch": 1.05, "learning_rate": 2.456035601820788e-05, "loss": 0.2639, "step": 3080 }, { "epoch": 1.05, "learning_rate": 2.4546400574598582e-05, "loss": 0.2122, "step": 3081 }, { "epoch": 1.05, "learning_rate": 2.4532445272379188e-05, "loss": 0.1979, "step": 3082 }, { "epoch": 1.05, "learning_rate": 2.4518490115899652e-05, "loss": 0.1969, "step": 3083 }, { "epoch": 1.05, "learning_rate": 2.4504535109509886e-05, "loss": 0.1964, "step": 3084 }, { "epoch": 1.05, "learning_rate": 2.4490580257559772e-05, "loss": 0.2453, "step": 3085 }, { "epoch": 1.05, "learning_rate": 2.447662556439912e-05, "loss": 0.229, "step": 3086 }, { "epoch": 1.05, "learning_rate": 2.44626710343777e-05, "loss": 0.235, "step": 3087 }, { "epoch": 1.05, "learning_rate": 2.444871667184522e-05, "loss": 0.2237, "step": 3088 }, { "epoch": 1.05, "learning_rate": 2.4434762481151365e-05, "loss": 0.2331, "step": 3089 }, { "epoch": 1.05, "learning_rate": 2.442080846664573e-05, "loss": 0.2312, "step": 3090 }, { "epoch": 1.05, "learning_rate": 2.4406854632677883e-05, "loss": 0.2135, "step": 3091 }, { "epoch": 1.05, "learning_rate": 2.4392900983597326e-05, "loss": 0.1992, "step": 3092 }, { "epoch": 1.06, "learning_rate": 2.437894752375349e-05, "loss": 0.241, "step": 3093 }, { "epoch": 1.06, "learning_rate": 2.4364994257495768e-05, "loss": 0.1956, "step": 3094 }, { "epoch": 1.06, "learning_rate": 2.435104118917349e-05, "loss": 0.2035, "step": 3095 }, { "epoch": 1.06, "learning_rate": 2.433708832313591e-05, "loss": 0.2251, "step": 3096 }, { "epoch": 1.06, "learning_rate": 2.4323135663732226e-05, "loss": 0.2493, "step": 3097 }, { "epoch": 1.06, "learning_rate": 2.4309183215311575e-05, "loss": 0.2115, "step": 3098 }, { "epoch": 1.06, "learning_rate": 2.4295230982223034e-05, "loss": 0.2055, "step": 3099 }, { "epoch": 1.06, "learning_rate": 2.4281278968815598e-05, "loss": 0.2887, "step": 3100 }, { "epoch": 1.06, "learning_rate": 2.42673271794382e-05, "loss": 0.2015, "step": 3101 }, { "epoch": 1.06, "learning_rate": 2.4253375618439713e-05, "loss": 0.2177, "step": 3102 }, { "epoch": 1.06, "learning_rate": 2.4239424290168923e-05, "loss": 0.2458, "step": 3103 }, { "epoch": 1.06, "learning_rate": 2.4225473198974548e-05, "loss": 0.214, "step": 3104 }, { "epoch": 1.06, "learning_rate": 2.4211522349205233e-05, "loss": 0.2249, "step": 3105 }, { "epoch": 1.06, "learning_rate": 2.4197571745209552e-05, "loss": 0.2382, "step": 3106 }, { "epoch": 1.06, "learning_rate": 2.4183621391336005e-05, "loss": 0.2161, "step": 3107 }, { "epoch": 1.06, "learning_rate": 2.4169671291933003e-05, "loss": 0.2138, "step": 3108 }, { "epoch": 1.06, "learning_rate": 2.4155721451348883e-05, "loss": 0.2393, "step": 3109 }, { "epoch": 1.06, "learning_rate": 2.4141771873931902e-05, "loss": 0.2391, "step": 3110 }, { "epoch": 1.06, "learning_rate": 2.412782256403022e-05, "loss": 0.2264, "step": 3111 }, { "epoch": 1.06, "learning_rate": 2.4113873525991955e-05, "loss": 0.2267, "step": 3112 }, { "epoch": 1.06, "learning_rate": 2.4099924764165104e-05, "loss": 0.2346, "step": 3113 }, { "epoch": 1.06, "learning_rate": 2.4085976282897573e-05, "loss": 0.2527, "step": 3114 }, { "epoch": 1.06, "learning_rate": 2.4072028086537212e-05, "loss": 0.2151, "step": 3115 }, { "epoch": 1.06, "learning_rate": 2.405808017943176e-05, "loss": 0.2766, "step": 3116 }, { "epoch": 1.06, "learning_rate": 2.4044132565928872e-05, "loss": 0.2058, "step": 3117 }, { "epoch": 1.06, "learning_rate": 2.40301852503761e-05, "loss": 0.2063, "step": 3118 }, { "epoch": 1.06, "learning_rate": 2.401623823712092e-05, "loss": 0.2459, "step": 3119 }, { "epoch": 1.06, "learning_rate": 2.4002291530510712e-05, "loss": 0.1991, "step": 3120 }, { "epoch": 1.06, "learning_rate": 2.398834513489275e-05, "loss": 0.2683, "step": 3121 }, { "epoch": 1.07, "learning_rate": 2.397439905461422e-05, "loss": 0.2189, "step": 3122 }, { "epoch": 1.07, "learning_rate": 2.3960453294022206e-05, "loss": 0.2135, "step": 3123 }, { "epoch": 1.07, "learning_rate": 2.3946507857463688e-05, "loss": 0.2375, "step": 3124 }, { "epoch": 1.07, "learning_rate": 2.3932562749285543e-05, "loss": 0.2304, "step": 3125 }, { "epoch": 1.07, "learning_rate": 2.391861797383457e-05, "loss": 0.2168, "step": 3126 }, { "epoch": 1.07, "learning_rate": 2.3904673535457435e-05, "loss": 0.2444, "step": 3127 }, { "epoch": 1.07, "learning_rate": 2.389072943850072e-05, "loss": 0.2183, "step": 3128 }, { "epoch": 1.07, "learning_rate": 2.387678568731089e-05, "loss": 0.2293, "step": 3129 }, { "epoch": 1.07, "learning_rate": 2.3862842286234293e-05, "loss": 0.2695, "step": 3130 }, { "epoch": 1.07, "learning_rate": 2.3848899239617183e-05, "loss": 0.2071, "step": 3131 }, { "epoch": 1.07, "learning_rate": 2.3834956551805693e-05, "loss": 0.2313, "step": 3132 }, { "epoch": 1.07, "learning_rate": 2.3821014227145867e-05, "loss": 0.2132, "step": 3133 }, { "epoch": 1.07, "learning_rate": 2.38070722699836e-05, "loss": 0.2322, "step": 3134 }, { "epoch": 1.07, "learning_rate": 2.3793130684664707e-05, "loss": 0.2427, "step": 3135 }, { "epoch": 1.07, "learning_rate": 2.377918947553486e-05, "loss": 0.2394, "step": 3136 }, { "epoch": 1.07, "learning_rate": 2.3765248646939627e-05, "loss": 0.21, "step": 3137 }, { "epoch": 1.07, "learning_rate": 2.3751308203224454e-05, "loss": 0.2251, "step": 3138 }, { "epoch": 1.07, "learning_rate": 2.373736814873467e-05, "loss": 0.2167, "step": 3139 }, { "epoch": 1.07, "learning_rate": 2.3723428487815484e-05, "loss": 0.2186, "step": 3140 }, { "epoch": 1.07, "learning_rate": 2.3709489224811977e-05, "loss": 0.2518, "step": 3141 }, { "epoch": 1.07, "learning_rate": 2.3695550364069113e-05, "loss": 0.2262, "step": 3142 }, { "epoch": 1.07, "learning_rate": 2.3681611909931724e-05, "loss": 0.2002, "step": 3143 }, { "epoch": 1.07, "learning_rate": 2.3667673866744512e-05, "loss": 0.2228, "step": 3144 }, { "epoch": 1.07, "learning_rate": 2.3653736238852055e-05, "loss": 0.2331, "step": 3145 }, { "epoch": 1.07, "learning_rate": 2.363979903059881e-05, "loss": 0.2416, "step": 3146 }, { "epoch": 1.07, "learning_rate": 2.3625862246329093e-05, "loss": 0.2419, "step": 3147 }, { "epoch": 1.07, "learning_rate": 2.3611925890387087e-05, "loss": 0.2534, "step": 3148 }, { "epoch": 1.07, "learning_rate": 2.3597989967116856e-05, "loss": 0.2229, "step": 3149 }, { "epoch": 1.07, "learning_rate": 2.3584054480862303e-05, "loss": 0.1696, "step": 3150 }, { "epoch": 1.08, "learning_rate": 2.3570119435967216e-05, "loss": 0.2292, "step": 3151 }, { "epoch": 1.08, "learning_rate": 2.3556184836775246e-05, "loss": 0.2532, "step": 3152 }, { "epoch": 1.08, "learning_rate": 2.3542250687629884e-05, "loss": 0.2261, "step": 3153 }, { "epoch": 1.08, "learning_rate": 2.3528316992874518e-05, "loss": 0.2432, "step": 3154 }, { "epoch": 1.08, "learning_rate": 2.3514383756852354e-05, "loss": 0.2159, "step": 3155 }, { "epoch": 1.08, "learning_rate": 2.350045098390648e-05, "loss": 0.2172, "step": 3156 }, { "epoch": 1.08, "learning_rate": 2.3486518678379833e-05, "loss": 0.2186, "step": 3157 }, { "epoch": 1.08, "learning_rate": 2.347258684461519e-05, "loss": 0.2395, "step": 3158 }, { "epoch": 1.08, "learning_rate": 2.3458655486955213e-05, "loss": 0.2081, "step": 3159 }, { "epoch": 1.08, "learning_rate": 2.344472460974239e-05, "loss": 0.2172, "step": 3160 }, { "epoch": 1.08, "learning_rate": 2.3430794217319065e-05, "loss": 0.2207, "step": 3161 }, { "epoch": 1.08, "learning_rate": 2.3416864314027434e-05, "loss": 0.2137, "step": 3162 }, { "epoch": 1.08, "learning_rate": 2.3402934904209537e-05, "loss": 0.2131, "step": 3163 }, { "epoch": 1.08, "learning_rate": 2.338900599220726e-05, "loss": 0.1979, "step": 3164 }, { "epoch": 1.08, "learning_rate": 2.337507758236234e-05, "loss": 0.2042, "step": 3165 }, { "epoch": 1.08, "learning_rate": 2.3361149679016352e-05, "loss": 0.2411, "step": 3166 }, { "epoch": 1.08, "learning_rate": 2.3347222286510706e-05, "loss": 0.2176, "step": 3167 }, { "epoch": 1.08, "learning_rate": 2.333329540918668e-05, "loss": 0.214, "step": 3168 }, { "epoch": 1.08, "learning_rate": 2.3319369051385363e-05, "loss": 0.2365, "step": 3169 }, { "epoch": 1.08, "learning_rate": 2.3305443217447693e-05, "loss": 0.2464, "step": 3170 }, { "epoch": 1.08, "learning_rate": 2.3291517911714435e-05, "loss": 0.2395, "step": 3171 }, { "epoch": 1.08, "learning_rate": 2.3277593138526207e-05, "loss": 0.2244, "step": 3172 }, { "epoch": 1.08, "learning_rate": 2.326366890222345e-05, "loss": 0.2375, "step": 3173 }, { "epoch": 1.08, "learning_rate": 2.3249745207146442e-05, "loss": 0.2305, "step": 3174 }, { "epoch": 1.08, "learning_rate": 2.3235822057635285e-05, "loss": 0.212, "step": 3175 }, { "epoch": 1.08, "learning_rate": 2.3221899458029918e-05, "loss": 0.2143, "step": 3176 }, { "epoch": 1.08, "learning_rate": 2.3207977412670094e-05, "loss": 0.208, "step": 3177 }, { "epoch": 1.08, "learning_rate": 2.3194055925895427e-05, "loss": 0.2498, "step": 3178 }, { "epoch": 1.08, "learning_rate": 2.3180135002045325e-05, "loss": 0.1996, "step": 3179 }, { "epoch": 1.08, "learning_rate": 2.3166214645459028e-05, "loss": 0.2209, "step": 3180 }, { "epoch": 1.09, "learning_rate": 2.3152294860475596e-05, "loss": 0.2934, "step": 3181 }, { "epoch": 1.09, "learning_rate": 2.3138375651433934e-05, "loss": 0.2476, "step": 3182 }, { "epoch": 1.09, "learning_rate": 2.312445702267275e-05, "loss": 0.204, "step": 3183 }, { "epoch": 1.09, "learning_rate": 2.3110538978530552e-05, "loss": 0.2157, "step": 3184 }, { "epoch": 1.09, "learning_rate": 2.3096621523345702e-05, "loss": 0.2318, "step": 3185 }, { "epoch": 1.09, "learning_rate": 2.308270466145636e-05, "loss": 0.2313, "step": 3186 }, { "epoch": 1.09, "learning_rate": 2.3068788397200497e-05, "loss": 0.2078, "step": 3187 }, { "epoch": 1.09, "learning_rate": 2.3054872734915907e-05, "loss": 0.2437, "step": 3188 }, { "epoch": 1.09, "learning_rate": 2.304095767894019e-05, "loss": 0.2099, "step": 3189 }, { "epoch": 1.09, "learning_rate": 2.302704323361076e-05, "loss": 0.2429, "step": 3190 }, { "epoch": 1.09, "learning_rate": 2.3013129403264845e-05, "loss": 0.227, "step": 3191 }, { "epoch": 1.09, "learning_rate": 2.2999216192239476e-05, "loss": 0.2093, "step": 3192 }, { "epoch": 1.09, "learning_rate": 2.298530360487149e-05, "loss": 0.2311, "step": 3193 }, { "epoch": 1.09, "learning_rate": 2.2971391645497528e-05, "loss": 0.2285, "step": 3194 }, { "epoch": 1.09, "learning_rate": 2.2957480318454034e-05, "loss": 0.2172, "step": 3195 }, { "epoch": 1.09, "learning_rate": 2.2943569628077275e-05, "loss": 0.2261, "step": 3196 }, { "epoch": 1.09, "learning_rate": 2.292965957870329e-05, "loss": 0.2469, "step": 3197 }, { "epoch": 1.09, "learning_rate": 2.291575017466795e-05, "loss": 0.2248, "step": 3198 }, { "epoch": 1.09, "learning_rate": 2.290184142030689e-05, "loss": 0.2335, "step": 3199 }, { "epoch": 1.09, "learning_rate": 2.2887933319955568e-05, "loss": 0.1989, "step": 3200 }, { "epoch": 1.09, "learning_rate": 2.2874025877949226e-05, "loss": 0.2347, "step": 3201 }, { "epoch": 1.09, "learning_rate": 2.2860119098622906e-05, "loss": 0.2099, "step": 3202 }, { "epoch": 1.09, "learning_rate": 2.2846212986311436e-05, "loss": 0.2109, "step": 3203 }, { "epoch": 1.09, "learning_rate": 2.283230754534945e-05, "loss": 0.2467, "step": 3204 }, { "epoch": 1.09, "learning_rate": 2.2818402780071368e-05, "loss": 0.2084, "step": 3205 }, { "epoch": 1.09, "learning_rate": 2.2804498694811385e-05, "loss": 0.2251, "step": 3206 }, { "epoch": 1.09, "learning_rate": 2.2790595293903504e-05, "loss": 0.2164, "step": 3207 }, { "epoch": 1.09, "learning_rate": 2.2776692581681492e-05, "loss": 0.225, "step": 3208 }, { "epoch": 1.09, "learning_rate": 2.2762790562478935e-05, "loss": 0.211, "step": 3209 }, { "epoch": 1.1, "learning_rate": 2.2748889240629167e-05, "loss": 0.2281, "step": 3210 }, { "epoch": 1.1, "learning_rate": 2.2734988620465338e-05, "loss": 0.2073, "step": 3211 }, { "epoch": 1.1, "learning_rate": 2.2721088706320347e-05, "loss": 0.2494, "step": 3212 }, { "epoch": 1.1, "learning_rate": 2.2707189502526895e-05, "loss": 0.207, "step": 3213 }, { "epoch": 1.1, "learning_rate": 2.2693291013417453e-05, "loss": 0.2163, "step": 3214 }, { "epoch": 1.1, "learning_rate": 2.267939324332427e-05, "loss": 0.223, "step": 3215 }, { "epoch": 1.1, "learning_rate": 2.2665496196579367e-05, "loss": 0.2518, "step": 3216 }, { "epoch": 1.1, "learning_rate": 2.2651599877514555e-05, "loss": 0.2127, "step": 3217 }, { "epoch": 1.1, "learning_rate": 2.26377042904614e-05, "loss": 0.2366, "step": 3218 }, { "epoch": 1.1, "learning_rate": 2.2623809439751252e-05, "loss": 0.243, "step": 3219 }, { "epoch": 1.1, "learning_rate": 2.260991532971522e-05, "loss": 0.2417, "step": 3220 }, { "epoch": 1.1, "learning_rate": 2.259602196468419e-05, "loss": 0.2262, "step": 3221 }, { "epoch": 1.1, "learning_rate": 2.2582129348988817e-05, "loss": 0.2212, "step": 3222 }, { "epoch": 1.1, "learning_rate": 2.256823748695952e-05, "loss": 0.1948, "step": 3223 }, { "epoch": 1.1, "learning_rate": 2.255434638292649e-05, "loss": 0.2336, "step": 3224 }, { "epoch": 1.1, "learning_rate": 2.2540456041219668e-05, "loss": 0.2305, "step": 3225 }, { "epoch": 1.1, "learning_rate": 2.2526566466168763e-05, "loss": 0.2132, "step": 3226 }, { "epoch": 1.1, "learning_rate": 2.251267766210325e-05, "loss": 0.2159, "step": 3227 }, { "epoch": 1.1, "learning_rate": 2.2498789633352358e-05, "loss": 0.274, "step": 3228 }, { "epoch": 1.1, "learning_rate": 2.248490238424507e-05, "loss": 0.1872, "step": 3229 }, { "epoch": 1.1, "learning_rate": 2.2471015919110142e-05, "loss": 0.219, "step": 3230 }, { "epoch": 1.1, "learning_rate": 2.2457130242276073e-05, "loss": 0.2415, "step": 3231 }, { "epoch": 1.1, "learning_rate": 2.244324535807112e-05, "loss": 0.247, "step": 3232 }, { "epoch": 1.1, "learning_rate": 2.2429361270823287e-05, "loss": 0.2555, "step": 3233 }, { "epoch": 1.1, "learning_rate": 2.2415477984860335e-05, "loss": 0.2128, "step": 3234 }, { "epoch": 1.1, "learning_rate": 2.2401595504509783e-05, "loss": 0.2667, "step": 3235 }, { "epoch": 1.1, "learning_rate": 2.2387713834098878e-05, "loss": 0.2409, "step": 3236 }, { "epoch": 1.1, "learning_rate": 2.2373832977954646e-05, "loss": 0.2449, "step": 3237 }, { "epoch": 1.1, "learning_rate": 2.2359952940403827e-05, "loss": 0.205, "step": 3238 }, { "epoch": 1.11, "learning_rate": 2.2346073725772922e-05, "loss": 0.2129, "step": 3239 }, { "epoch": 1.11, "learning_rate": 2.2332195338388177e-05, "loss": 0.228, "step": 3240 }, { "epoch": 1.11, "learning_rate": 2.231831778257557e-05, "loss": 0.2148, "step": 3241 }, { "epoch": 1.11, "learning_rate": 2.2304441062660822e-05, "loss": 0.2073, "step": 3242 }, { "epoch": 1.11, "learning_rate": 2.2290565182969407e-05, "loss": 0.2371, "step": 3243 }, { "epoch": 1.11, "learning_rate": 2.2276690147826527e-05, "loss": 0.226, "step": 3244 }, { "epoch": 1.11, "learning_rate": 2.2262815961557115e-05, "loss": 0.2138, "step": 3245 }, { "epoch": 1.11, "learning_rate": 2.224894262848585e-05, "loss": 0.2122, "step": 3246 }, { "epoch": 1.11, "learning_rate": 2.2235070152937133e-05, "loss": 0.2357, "step": 3247 }, { "epoch": 1.11, "learning_rate": 2.2221198539235114e-05, "loss": 0.2174, "step": 3248 }, { "epoch": 1.11, "learning_rate": 2.2207327791703665e-05, "loss": 0.2373, "step": 3249 }, { "epoch": 1.11, "learning_rate": 2.2193457914666372e-05, "loss": 0.2309, "step": 3250 }, { "epoch": 1.11, "learning_rate": 2.2179588912446595e-05, "loss": 0.2207, "step": 3251 }, { "epoch": 1.11, "learning_rate": 2.2165720789367376e-05, "loss": 0.1978, "step": 3252 }, { "epoch": 1.11, "learning_rate": 2.2151853549751503e-05, "loss": 0.216, "step": 3253 }, { "epoch": 1.11, "learning_rate": 2.213798719792148e-05, "loss": 0.216, "step": 3254 }, { "epoch": 1.11, "learning_rate": 2.2124121738199542e-05, "loss": 0.21, "step": 3255 }, { "epoch": 1.11, "learning_rate": 2.2110257174907648e-05, "loss": 0.2123, "step": 3256 }, { "epoch": 1.11, "learning_rate": 2.209639351236747e-05, "loss": 0.2534, "step": 3257 }, { "epoch": 1.11, "learning_rate": 2.20825307549004e-05, "loss": 0.1969, "step": 3258 }, { "epoch": 1.11, "learning_rate": 2.2068668906827554e-05, "loss": 0.2198, "step": 3259 }, { "epoch": 1.11, "learning_rate": 2.2054807972469755e-05, "loss": 0.2066, "step": 3260 }, { "epoch": 1.11, "learning_rate": 2.204094795614755e-05, "loss": 0.2182, "step": 3261 }, { "epoch": 1.11, "learning_rate": 2.20270888621812e-05, "loss": 0.2077, "step": 3262 }, { "epoch": 1.11, "learning_rate": 2.201323069489067e-05, "loss": 0.2384, "step": 3263 }, { "epoch": 1.11, "learning_rate": 2.1999373458595636e-05, "loss": 0.228, "step": 3264 }, { "epoch": 1.11, "learning_rate": 2.1985517157615503e-05, "loss": 0.2119, "step": 3265 }, { "epoch": 1.11, "learning_rate": 2.1971661796269367e-05, "loss": 0.2369, "step": 3266 }, { "epoch": 1.11, "learning_rate": 2.195780737887603e-05, "loss": 0.2239, "step": 3267 }, { "epoch": 1.11, "learning_rate": 2.1943953909754e-05, "loss": 0.223, "step": 3268 }, { "epoch": 1.12, "learning_rate": 2.19301013932215e-05, "loss": 0.2242, "step": 3269 }, { "epoch": 1.12, "learning_rate": 2.1916249833596456e-05, "loss": 0.2247, "step": 3270 }, { "epoch": 1.12, "learning_rate": 2.190239923519648e-05, "loss": 0.2517, "step": 3271 }, { "epoch": 1.12, "learning_rate": 2.1888549602338898e-05, "loss": 0.252, "step": 3272 }, { "epoch": 1.12, "learning_rate": 2.1874700939340727e-05, "loss": 0.1928, "step": 3273 }, { "epoch": 1.12, "learning_rate": 2.1860853250518692e-05, "loss": 0.1987, "step": 3274 }, { "epoch": 1.12, "learning_rate": 2.1847006540189203e-05, "loss": 0.2064, "step": 3275 }, { "epoch": 1.12, "learning_rate": 2.1833160812668377e-05, "loss": 0.2042, "step": 3276 }, { "epoch": 1.12, "learning_rate": 2.181931607227201e-05, "loss": 0.2213, "step": 3277 }, { "epoch": 1.12, "learning_rate": 2.180547232331559e-05, "loss": 0.1994, "step": 3278 }, { "epoch": 1.12, "learning_rate": 2.1791629570114325e-05, "loss": 0.1944, "step": 3279 }, { "epoch": 1.12, "learning_rate": 2.177778781698308e-05, "loss": 0.2393, "step": 3280 }, { "epoch": 1.12, "learning_rate": 2.1763947068236417e-05, "loss": 0.2139, "step": 3281 }, { "epoch": 1.12, "learning_rate": 2.17501073281886e-05, "loss": 0.2365, "step": 3282 }, { "epoch": 1.12, "learning_rate": 2.1736268601153553e-05, "loss": 0.2351, "step": 3283 }, { "epoch": 1.12, "learning_rate": 2.1722430891444907e-05, "loss": 0.2105, "step": 3284 }, { "epoch": 1.12, "learning_rate": 2.170859420337596e-05, "loss": 0.2137, "step": 3285 }, { "epoch": 1.12, "learning_rate": 2.1694758541259697e-05, "loss": 0.2164, "step": 3286 }, { "epoch": 1.12, "learning_rate": 2.1680923909408793e-05, "loss": 0.2316, "step": 3287 }, { "epoch": 1.12, "learning_rate": 2.1667090312135585e-05, "loss": 0.2268, "step": 3288 }, { "epoch": 1.12, "learning_rate": 2.1653257753752103e-05, "loss": 0.2215, "step": 3289 }, { "epoch": 1.12, "learning_rate": 2.1639426238570037e-05, "loss": 0.203, "step": 3290 }, { "epoch": 1.12, "learning_rate": 2.1625595770900763e-05, "loss": 0.2339, "step": 3291 }, { "epoch": 1.12, "learning_rate": 2.161176635505533e-05, "loss": 0.2353, "step": 3292 }, { "epoch": 1.12, "learning_rate": 2.1597937995344462e-05, "loss": 0.2056, "step": 3293 }, { "epoch": 1.12, "learning_rate": 2.158411069607854e-05, "loss": 0.2527, "step": 3294 }, { "epoch": 1.12, "learning_rate": 2.1570284461567632e-05, "loss": 0.2242, "step": 3295 }, { "epoch": 1.12, "learning_rate": 2.155645929612146e-05, "loss": 0.2372, "step": 3296 }, { "epoch": 1.12, "learning_rate": 2.1542635204049422e-05, "loss": 0.2397, "step": 3297 }, { "epoch": 1.13, "learning_rate": 2.1528812189660573e-05, "loss": 0.2163, "step": 3298 }, { "epoch": 1.13, "learning_rate": 2.1514990257263638e-05, "loss": 0.202, "step": 3299 }, { "epoch": 1.13, "learning_rate": 2.1501169411167006e-05, "loss": 0.2449, "step": 3300 }, { "epoch": 1.13, "learning_rate": 2.148734965567873e-05, "loss": 0.2393, "step": 3301 }, { "epoch": 1.13, "learning_rate": 2.147353099510651e-05, "loss": 0.2149, "step": 3302 }, { "epoch": 1.13, "learning_rate": 2.145971343375771e-05, "loss": 0.2114, "step": 3303 }, { "epoch": 1.13, "learning_rate": 2.144589697593936e-05, "loss": 0.2222, "step": 3304 }, { "epoch": 1.13, "learning_rate": 2.143208162595814e-05, "loss": 0.2163, "step": 3305 }, { "epoch": 1.13, "learning_rate": 2.1418267388120384e-05, "loss": 0.2358, "step": 3306 }, { "epoch": 1.13, "learning_rate": 2.1404454266732077e-05, "loss": 0.2153, "step": 3307 }, { "epoch": 1.13, "learning_rate": 2.139064226609887e-05, "loss": 0.2364, "step": 3308 }, { "epoch": 1.13, "learning_rate": 2.1376831390526045e-05, "loss": 0.2042, "step": 3309 }, { "epoch": 1.13, "learning_rate": 2.136302164431854e-05, "loss": 0.2125, "step": 3310 }, { "epoch": 1.13, "learning_rate": 2.1349213031780955e-05, "loss": 0.2504, "step": 3311 }, { "epoch": 1.13, "learning_rate": 2.1335405557217507e-05, "loss": 0.1996, "step": 3312 }, { "epoch": 1.13, "learning_rate": 2.1321599224932086e-05, "loss": 0.225, "step": 3313 }, { "epoch": 1.13, "learning_rate": 2.130779403922822e-05, "loss": 0.1989, "step": 3314 }, { "epoch": 1.13, "learning_rate": 2.129399000440907e-05, "loss": 0.2364, "step": 3315 }, { "epoch": 1.13, "learning_rate": 2.1280187124777444e-05, "loss": 0.204, "step": 3316 }, { "epoch": 1.13, "learning_rate": 2.126638540463579e-05, "loss": 0.2365, "step": 3317 }, { "epoch": 1.13, "learning_rate": 2.1252584848286193e-05, "loss": 0.2247, "step": 3318 }, { "epoch": 1.13, "learning_rate": 2.1238785460030376e-05, "loss": 0.2789, "step": 3319 }, { "epoch": 1.13, "learning_rate": 2.12249872441697e-05, "loss": 0.2158, "step": 3320 }, { "epoch": 1.13, "learning_rate": 2.1211190205005165e-05, "loss": 0.2411, "step": 3321 }, { "epoch": 1.13, "learning_rate": 2.1197394346837395e-05, "loss": 0.2177, "step": 3322 }, { "epoch": 1.13, "learning_rate": 2.118359967396665e-05, "loss": 0.2247, "step": 3323 }, { "epoch": 1.13, "learning_rate": 2.116980619069281e-05, "loss": 0.2088, "step": 3324 }, { "epoch": 1.13, "learning_rate": 2.1156013901315397e-05, "loss": 0.2397, "step": 3325 }, { "epoch": 1.13, "learning_rate": 2.1142222810133565e-05, "loss": 0.205, "step": 3326 }, { "epoch": 1.14, "learning_rate": 2.112843292144609e-05, "loss": 0.1994, "step": 3327 }, { "epoch": 1.14, "learning_rate": 2.111464423955136e-05, "loss": 0.2172, "step": 3328 }, { "epoch": 1.14, "learning_rate": 2.11008567687474e-05, "loss": 0.2075, "step": 3329 }, { "epoch": 1.14, "learning_rate": 2.1087070513331854e-05, "loss": 0.1855, "step": 3330 }, { "epoch": 1.14, "learning_rate": 2.107328547760199e-05, "loss": 0.2346, "step": 3331 }, { "epoch": 1.14, "learning_rate": 2.1059501665854697e-05, "loss": 0.2203, "step": 3332 }, { "epoch": 1.14, "learning_rate": 2.1045719082386463e-05, "loss": 0.2013, "step": 3333 }, { "epoch": 1.14, "learning_rate": 2.1031937731493434e-05, "loss": 0.2256, "step": 3334 }, { "epoch": 1.14, "learning_rate": 2.1018157617471335e-05, "loss": 0.2203, "step": 3335 }, { "epoch": 1.14, "learning_rate": 2.1004378744615515e-05, "loss": 0.2073, "step": 3336 }, { "epoch": 1.14, "learning_rate": 2.0990601117220942e-05, "loss": 0.2363, "step": 3337 }, { "epoch": 1.14, "learning_rate": 2.097682473958219e-05, "loss": 0.1997, "step": 3338 }, { "epoch": 1.14, "learning_rate": 2.096304961599345e-05, "loss": 0.2461, "step": 3339 }, { "epoch": 1.14, "learning_rate": 2.0949275750748514e-05, "loss": 0.2289, "step": 3340 }, { "epoch": 1.14, "learning_rate": 2.0935503148140792e-05, "loss": 0.2196, "step": 3341 }, { "epoch": 1.14, "learning_rate": 2.0921731812463284e-05, "loss": 0.2463, "step": 3342 }, { "epoch": 1.14, "learning_rate": 2.0907961748008613e-05, "loss": 0.2169, "step": 3343 }, { "epoch": 1.14, "learning_rate": 2.0894192959068994e-05, "loss": 0.2209, "step": 3344 }, { "epoch": 1.14, "learning_rate": 2.0880425449936254e-05, "loss": 0.2247, "step": 3345 }, { "epoch": 1.14, "learning_rate": 2.0866659224901812e-05, "loss": 0.2255, "step": 3346 }, { "epoch": 1.14, "learning_rate": 2.0852894288256674e-05, "loss": 0.2455, "step": 3347 }, { "epoch": 1.14, "learning_rate": 2.0839130644291492e-05, "loss": 0.2313, "step": 3348 }, { "epoch": 1.14, "learning_rate": 2.082536829729647e-05, "loss": 0.1746, "step": 3349 }, { "epoch": 1.14, "learning_rate": 2.0811607251561418e-05, "loss": 0.2206, "step": 3350 }, { "epoch": 1.14, "learning_rate": 2.079784751137574e-05, "loss": 0.2014, "step": 3351 }, { "epoch": 1.14, "learning_rate": 2.078408908102845e-05, "loss": 0.2044, "step": 3352 }, { "epoch": 1.14, "learning_rate": 2.0770331964808133e-05, "loss": 0.2335, "step": 3353 }, { "epoch": 1.14, "learning_rate": 2.075657616700297e-05, "loss": 0.2197, "step": 3354 }, { "epoch": 1.14, "learning_rate": 2.0742821691900738e-05, "loss": 0.2029, "step": 3355 }, { "epoch": 1.15, "learning_rate": 2.0729068543788786e-05, "loss": 0.2486, "step": 3356 }, { "epoch": 1.15, "learning_rate": 2.0715316726954075e-05, "loss": 0.2481, "step": 3357 }, { "epoch": 1.15, "learning_rate": 2.0701566245683128e-05, "loss": 0.1967, "step": 3358 }, { "epoch": 1.15, "learning_rate": 2.0687817104262063e-05, "loss": 0.2223, "step": 3359 }, { "epoch": 1.15, "learning_rate": 2.067406930697658e-05, "loss": 0.173, "step": 3360 }, { "epoch": 1.15, "learning_rate": 2.0660322858111934e-05, "loss": 0.1889, "step": 3361 }, { "epoch": 1.15, "learning_rate": 2.0646577761953014e-05, "loss": 0.2334, "step": 3362 }, { "epoch": 1.15, "learning_rate": 2.0632834022784247e-05, "loss": 0.2502, "step": 3363 }, { "epoch": 1.15, "learning_rate": 2.0619091644889635e-05, "loss": 0.2216, "step": 3364 }, { "epoch": 1.15, "learning_rate": 2.0605350632552784e-05, "loss": 0.2329, "step": 3365 }, { "epoch": 1.15, "learning_rate": 2.059161099005684e-05, "loss": 0.2112, "step": 3366 }, { "epoch": 1.15, "learning_rate": 2.0577872721684558e-05, "loss": 0.2192, "step": 3367 }, { "epoch": 1.15, "learning_rate": 2.056413583171823e-05, "loss": 0.2105, "step": 3368 }, { "epoch": 1.15, "learning_rate": 2.0550400324439733e-05, "loss": 0.2135, "step": 3369 }, { "epoch": 1.15, "learning_rate": 2.0536666204130532e-05, "loss": 0.2349, "step": 3370 }, { "epoch": 1.15, "learning_rate": 2.0522933475071625e-05, "loss": 0.2078, "step": 3371 }, { "epoch": 1.15, "learning_rate": 2.0509202141543606e-05, "loss": 0.2168, "step": 3372 }, { "epoch": 1.15, "learning_rate": 2.0495472207826612e-05, "loss": 0.2435, "step": 3373 }, { "epoch": 1.15, "learning_rate": 2.0481743678200353e-05, "loss": 0.2402, "step": 3374 }, { "epoch": 1.15, "learning_rate": 2.04680165569441e-05, "loss": 0.2286, "step": 3375 }, { "epoch": 1.15, "learning_rate": 2.0454290848336698e-05, "loss": 0.2378, "step": 3376 }, { "epoch": 1.15, "learning_rate": 2.044056655665653e-05, "loss": 0.2307, "step": 3377 }, { "epoch": 1.15, "learning_rate": 2.0426843686181553e-05, "loss": 0.2176, "step": 3378 }, { "epoch": 1.15, "learning_rate": 2.0413122241189274e-05, "loss": 0.2138, "step": 3379 }, { "epoch": 1.15, "learning_rate": 2.0399402225956754e-05, "loss": 0.2104, "step": 3380 }, { "epoch": 1.15, "learning_rate": 2.0385683644760613e-05, "loss": 0.2595, "step": 3381 }, { "epoch": 1.15, "learning_rate": 2.0371966501877017e-05, "loss": 0.2369, "step": 3382 }, { "epoch": 1.15, "learning_rate": 2.0358250801581703e-05, "loss": 0.2044, "step": 3383 }, { "epoch": 1.15, "learning_rate": 2.034453654814993e-05, "loss": 0.207, "step": 3384 }, { "epoch": 1.15, "learning_rate": 2.033082374585653e-05, "loss": 0.2065, "step": 3385 }, { "epoch": 1.16, "learning_rate": 2.031711239897587e-05, "loss": 0.2642, "step": 3386 }, { "epoch": 1.16, "learning_rate": 2.0303402511781865e-05, "loss": 0.2165, "step": 3387 }, { "epoch": 1.16, "learning_rate": 2.028969408854797e-05, "loss": 0.2126, "step": 3388 }, { "epoch": 1.16, "learning_rate": 2.0275987133547197e-05, "loss": 0.2108, "step": 3389 }, { "epoch": 1.16, "learning_rate": 2.0262281651052094e-05, "loss": 0.2191, "step": 3390 }, { "epoch": 1.16, "learning_rate": 2.0248577645334756e-05, "loss": 0.2082, "step": 3391 }, { "epoch": 1.16, "learning_rate": 2.0234875120666806e-05, "loss": 0.1964, "step": 3392 }, { "epoch": 1.16, "learning_rate": 2.0221174081319406e-05, "loss": 0.2211, "step": 3393 }, { "epoch": 1.16, "learning_rate": 2.020747453156326e-05, "loss": 0.2235, "step": 3394 }, { "epoch": 1.16, "learning_rate": 2.0193776475668605e-05, "loss": 0.1884, "step": 3395 }, { "epoch": 1.16, "learning_rate": 2.0180079917905223e-05, "loss": 0.2209, "step": 3396 }, { "epoch": 1.16, "learning_rate": 2.016638486254242e-05, "loss": 0.1937, "step": 3397 }, { "epoch": 1.16, "learning_rate": 2.0152691313849027e-05, "loss": 0.2345, "step": 3398 }, { "epoch": 1.16, "learning_rate": 2.0138999276093413e-05, "loss": 0.1822, "step": 3399 }, { "epoch": 1.16, "learning_rate": 2.0125308753543482e-05, "loss": 0.2013, "step": 3400 }, { "epoch": 1.16, "learning_rate": 2.0111619750466652e-05, "loss": 0.2158, "step": 3401 }, { "epoch": 1.16, "learning_rate": 2.0097932271129877e-05, "loss": 0.2366, "step": 3402 }, { "epoch": 1.16, "learning_rate": 2.0084246319799636e-05, "loss": 0.1824, "step": 3403 }, { "epoch": 1.16, "learning_rate": 2.007056190074193e-05, "loss": 0.2494, "step": 3404 }, { "epoch": 1.16, "learning_rate": 2.0056879018222284e-05, "loss": 0.2374, "step": 3405 }, { "epoch": 1.16, "learning_rate": 2.0043197676505737e-05, "loss": 0.2099, "step": 3406 }, { "epoch": 1.16, "learning_rate": 2.0029517879856858e-05, "loss": 0.215, "step": 3407 }, { "epoch": 1.16, "learning_rate": 2.0015839632539715e-05, "loss": 0.2102, "step": 3408 }, { "epoch": 1.16, "learning_rate": 2.0002162938817924e-05, "loss": 0.2548, "step": 3409 }, { "epoch": 1.16, "learning_rate": 1.998848780295459e-05, "loss": 0.2489, "step": 3410 }, { "epoch": 1.16, "learning_rate": 1.997481422921235e-05, "loss": 0.2428, "step": 3411 }, { "epoch": 1.16, "learning_rate": 1.9961142221853337e-05, "loss": 0.2188, "step": 3412 }, { "epoch": 1.16, "learning_rate": 1.994747178513921e-05, "loss": 0.214, "step": 3413 }, { "epoch": 1.16, "learning_rate": 1.9933802923331128e-05, "loss": 0.2196, "step": 3414 }, { "epoch": 1.17, "learning_rate": 1.9920135640689773e-05, "loss": 0.2306, "step": 3415 }, { "epoch": 1.17, "learning_rate": 1.9906469941475313e-05, "loss": 0.2302, "step": 3416 }, { "epoch": 1.17, "learning_rate": 1.989280582994746e-05, "loss": 0.1938, "step": 3417 }, { "epoch": 1.17, "learning_rate": 1.9879143310365386e-05, "loss": 0.2185, "step": 3418 }, { "epoch": 1.17, "learning_rate": 1.9865482386987794e-05, "loss": 0.1971, "step": 3419 }, { "epoch": 1.17, "learning_rate": 1.9851823064072887e-05, "loss": 0.2384, "step": 3420 }, { "epoch": 1.17, "learning_rate": 1.9838165345878352e-05, "loss": 0.2445, "step": 3421 }, { "epoch": 1.17, "learning_rate": 1.98245092366614e-05, "loss": 0.2193, "step": 3422 }, { "epoch": 1.17, "learning_rate": 1.981085474067873e-05, "loss": 0.2127, "step": 3423 }, { "epoch": 1.17, "learning_rate": 1.9797201862186536e-05, "loss": 0.236, "step": 3424 }, { "epoch": 1.17, "learning_rate": 1.978355060544051e-05, "loss": 0.1892, "step": 3425 }, { "epoch": 1.17, "learning_rate": 1.976990097469583e-05, "loss": 0.2054, "step": 3426 }, { "epoch": 1.17, "learning_rate": 1.9756252974207183e-05, "loss": 0.2171, "step": 3427 }, { "epoch": 1.17, "learning_rate": 1.9742606608228738e-05, "loss": 0.226, "step": 3428 }, { "epoch": 1.17, "learning_rate": 1.9728961881014158e-05, "loss": 0.2384, "step": 3429 }, { "epoch": 1.17, "learning_rate": 1.9715318796816582e-05, "loss": 0.2084, "step": 3430 }, { "epoch": 1.17, "learning_rate": 1.970167735988867e-05, "loss": 0.2015, "step": 3431 }, { "epoch": 1.17, "learning_rate": 1.9688037574482532e-05, "loss": 0.2322, "step": 3432 }, { "epoch": 1.17, "learning_rate": 1.9674399444849784e-05, "loss": 0.2043, "step": 3433 }, { "epoch": 1.17, "learning_rate": 1.966076297524151e-05, "loss": 0.2032, "step": 3434 }, { "epoch": 1.17, "learning_rate": 1.9647128169908304e-05, "loss": 0.2258, "step": 3435 }, { "epoch": 1.17, "learning_rate": 1.9633495033100212e-05, "loss": 0.2475, "step": 3436 }, { "epoch": 1.17, "learning_rate": 1.9619863569066777e-05, "loss": 0.2619, "step": 3437 }, { "epoch": 1.17, "learning_rate": 1.9606233782057008e-05, "loss": 0.1935, "step": 3438 }, { "epoch": 1.17, "learning_rate": 1.959260567631941e-05, "loss": 0.2355, "step": 3439 }, { "epoch": 1.17, "learning_rate": 1.9578979256101936e-05, "loss": 0.2409, "step": 3440 }, { "epoch": 1.17, "learning_rate": 1.9565354525652043e-05, "loss": 0.2235, "step": 3441 }, { "epoch": 1.17, "learning_rate": 1.955173148921665e-05, "loss": 0.2327, "step": 3442 }, { "epoch": 1.17, "learning_rate": 1.9538110151042142e-05, "loss": 0.2251, "step": 3443 }, { "epoch": 1.18, "learning_rate": 1.9524490515374363e-05, "loss": 0.2191, "step": 3444 }, { "epoch": 1.18, "learning_rate": 1.9510872586458674e-05, "loss": 0.2071, "step": 3445 }, { "epoch": 1.18, "learning_rate": 1.9497256368539858e-05, "loss": 0.248, "step": 3446 }, { "epoch": 1.18, "learning_rate": 1.9483641865862168e-05, "loss": 0.2401, "step": 3447 }, { "epoch": 1.18, "learning_rate": 1.947002908266935e-05, "loss": 0.24, "step": 3448 }, { "epoch": 1.18, "learning_rate": 1.945641802320459e-05, "loss": 0.2356, "step": 3449 }, { "epoch": 1.18, "learning_rate": 1.944280869171055e-05, "loss": 0.2203, "step": 3450 }, { "epoch": 1.18, "learning_rate": 1.9429201092429336e-05, "loss": 0.2345, "step": 3451 }, { "epoch": 1.18, "learning_rate": 1.9415595229602535e-05, "loss": 0.2126, "step": 3452 }, { "epoch": 1.18, "learning_rate": 1.9401991107471175e-05, "loss": 0.2489, "step": 3453 }, { "epoch": 1.18, "learning_rate": 1.938838873027576e-05, "loss": 0.1952, "step": 3454 }, { "epoch": 1.18, "learning_rate": 1.9374788102256237e-05, "loss": 0.2378, "step": 3455 }, { "epoch": 1.18, "learning_rate": 1.9361189227652004e-05, "loss": 0.255, "step": 3456 }, { "epoch": 1.18, "learning_rate": 1.934759211070193e-05, "loss": 0.2181, "step": 3457 }, { "epoch": 1.18, "learning_rate": 1.93339967556443e-05, "loss": 0.2181, "step": 3458 }, { "epoch": 1.18, "learning_rate": 1.932040316671691e-05, "loss": 0.2165, "step": 3459 }, { "epoch": 1.18, "learning_rate": 1.930681134815695e-05, "loss": 0.1885, "step": 3460 }, { "epoch": 1.18, "learning_rate": 1.9293221304201083e-05, "loss": 0.2354, "step": 3461 }, { "epoch": 1.18, "learning_rate": 1.927963303908542e-05, "loss": 0.2422, "step": 3462 }, { "epoch": 1.18, "learning_rate": 1.9266046557045503e-05, "loss": 0.2097, "step": 3463 }, { "epoch": 1.18, "learning_rate": 1.9252461862316328e-05, "loss": 0.2215, "step": 3464 }, { "epoch": 1.18, "learning_rate": 1.923887895913234e-05, "loss": 0.2101, "step": 3465 }, { "epoch": 1.18, "learning_rate": 1.922529785172741e-05, "loss": 0.1893, "step": 3466 }, { "epoch": 1.18, "learning_rate": 1.921171854433486e-05, "loss": 0.2233, "step": 3467 }, { "epoch": 1.18, "learning_rate": 1.919814104118746e-05, "loss": 0.2385, "step": 3468 }, { "epoch": 1.18, "learning_rate": 1.918456534651739e-05, "loss": 0.2419, "step": 3469 }, { "epoch": 1.18, "learning_rate": 1.9170991464556296e-05, "loss": 0.2191, "step": 3470 }, { "epoch": 1.18, "learning_rate": 1.9157419399535232e-05, "loss": 0.1935, "step": 3471 }, { "epoch": 1.18, "learning_rate": 1.9143849155684705e-05, "loss": 0.2043, "step": 3472 }, { "epoch": 1.18, "learning_rate": 1.9130280737234652e-05, "loss": 0.2287, "step": 3473 }, { "epoch": 1.19, "learning_rate": 1.9116714148414444e-05, "loss": 0.1927, "step": 3474 }, { "epoch": 1.19, "learning_rate": 1.9103149393452866e-05, "loss": 0.2146, "step": 3475 }, { "epoch": 1.19, "learning_rate": 1.908958647657814e-05, "loss": 0.2334, "step": 3476 }, { "epoch": 1.19, "learning_rate": 1.907602540201792e-05, "loss": 0.26, "step": 3477 }, { "epoch": 1.19, "learning_rate": 1.9062466173999283e-05, "loss": 0.2331, "step": 3478 }, { "epoch": 1.19, "learning_rate": 1.904890879674872e-05, "loss": 0.2737, "step": 3479 }, { "epoch": 1.19, "learning_rate": 1.9035353274492162e-05, "loss": 0.1937, "step": 3480 }, { "epoch": 1.19, "learning_rate": 1.9021799611454953e-05, "loss": 0.2307, "step": 3481 }, { "epoch": 1.19, "learning_rate": 1.9008247811861856e-05, "loss": 0.2047, "step": 3482 }, { "epoch": 1.19, "learning_rate": 1.8994697879937058e-05, "loss": 0.2039, "step": 3483 }, { "epoch": 1.19, "learning_rate": 1.8981149819904153e-05, "loss": 0.2296, "step": 3484 }, { "epoch": 1.19, "learning_rate": 1.8967603635986173e-05, "loss": 0.2364, "step": 3485 }, { "epoch": 1.19, "learning_rate": 1.8954059332405534e-05, "loss": 0.2034, "step": 3486 }, { "epoch": 1.19, "learning_rate": 1.8940516913384104e-05, "loss": 0.2345, "step": 3487 }, { "epoch": 1.19, "learning_rate": 1.8926976383143137e-05, "loss": 0.2418, "step": 3488 }, { "epoch": 1.19, "learning_rate": 1.8913437745903307e-05, "loss": 0.2032, "step": 3489 }, { "epoch": 1.19, "learning_rate": 1.8899901005884685e-05, "loss": 0.1885, "step": 3490 }, { "epoch": 1.19, "learning_rate": 1.888636616730677e-05, "loss": 0.2082, "step": 3491 }, { "epoch": 1.19, "learning_rate": 1.887283323438845e-05, "loss": 0.2332, "step": 3492 }, { "epoch": 1.19, "learning_rate": 1.8859302211348045e-05, "loss": 0.265, "step": 3493 }, { "epoch": 1.19, "learning_rate": 1.8845773102403252e-05, "loss": 0.2072, "step": 3494 }, { "epoch": 1.19, "learning_rate": 1.8832245911771186e-05, "loss": 0.2207, "step": 3495 }, { "epoch": 1.19, "learning_rate": 1.8818720643668358e-05, "loss": 0.2335, "step": 3496 }, { "epoch": 1.19, "learning_rate": 1.880519730231068e-05, "loss": 0.2238, "step": 3497 }, { "epoch": 1.19, "learning_rate": 1.8791675891913472e-05, "loss": 0.2454, "step": 3498 }, { "epoch": 1.19, "learning_rate": 1.8778156416691433e-05, "loss": 0.2261, "step": 3499 }, { "epoch": 1.19, "learning_rate": 1.8764638880858694e-05, "loss": 0.2191, "step": 3500 }, { "epoch": 1.19, "learning_rate": 1.8751123288628747e-05, "loss": 0.2387, "step": 3501 }, { "epoch": 1.19, "learning_rate": 1.8737609644214492e-05, "loss": 0.23, "step": 3502 }, { "epoch": 1.2, "learning_rate": 1.872409795182822e-05, "loss": 0.2046, "step": 3503 }, { "epoch": 1.2, "learning_rate": 1.8710588215681613e-05, "loss": 0.2219, "step": 3504 }, { "epoch": 1.2, "learning_rate": 1.8697080439985738e-05, "loss": 0.2335, "step": 3505 }, { "epoch": 1.2, "learning_rate": 1.8683574628951068e-05, "loss": 0.1937, "step": 3506 }, { "epoch": 1.2, "learning_rate": 1.8670070786787448e-05, "loss": 0.2226, "step": 3507 }, { "epoch": 1.2, "learning_rate": 1.865656891770411e-05, "loss": 0.2254, "step": 3508 }, { "epoch": 1.2, "learning_rate": 1.864306902590968e-05, "loss": 0.2048, "step": 3509 }, { "epoch": 1.2, "learning_rate": 1.8629571115612153e-05, "loss": 0.2349, "step": 3510 }, { "epoch": 1.2, "learning_rate": 1.8616075191018934e-05, "loss": 0.2195, "step": 3511 }, { "epoch": 1.2, "learning_rate": 1.8602581256336776e-05, "loss": 0.2348, "step": 3512 }, { "epoch": 1.2, "learning_rate": 1.8589089315771824e-05, "loss": 0.2333, "step": 3513 }, { "epoch": 1.2, "learning_rate": 1.8575599373529616e-05, "loss": 0.182, "step": 3514 }, { "epoch": 1.2, "learning_rate": 1.8562111433815056e-05, "loss": 0.2447, "step": 3515 }, { "epoch": 1.2, "learning_rate": 1.854862550083241e-05, "loss": 0.2185, "step": 3516 }, { "epoch": 1.2, "learning_rate": 1.8535141578785345e-05, "loss": 0.2155, "step": 3517 }, { "epoch": 1.2, "learning_rate": 1.8521659671876875e-05, "loss": 0.2328, "step": 3518 }, { "epoch": 1.2, "learning_rate": 1.8508179784309405e-05, "loss": 0.2071, "step": 3519 }, { "epoch": 1.2, "learning_rate": 1.8494701920284703e-05, "loss": 0.2055, "step": 3520 }, { "epoch": 1.2, "learning_rate": 1.848122608400391e-05, "loss": 0.2082, "step": 3521 }, { "epoch": 1.2, "learning_rate": 1.8467752279667526e-05, "loss": 0.2516, "step": 3522 }, { "epoch": 1.2, "learning_rate": 1.845428051147543e-05, "loss": 0.2275, "step": 3523 }, { "epoch": 1.2, "learning_rate": 1.8440810783626855e-05, "loss": 0.2445, "step": 3524 }, { "epoch": 1.2, "learning_rate": 1.842734310032041e-05, "loss": 0.2176, "step": 3525 }, { "epoch": 1.2, "learning_rate": 1.8413877465754053e-05, "loss": 0.2316, "step": 3526 }, { "epoch": 1.2, "learning_rate": 1.8400413884125106e-05, "loss": 0.2204, "step": 3527 }, { "epoch": 1.2, "learning_rate": 1.838695235963027e-05, "loss": 0.2013, "step": 3528 }, { "epoch": 1.2, "learning_rate": 1.8373492896465584e-05, "loss": 0.2265, "step": 3529 }, { "epoch": 1.2, "learning_rate": 1.836003549882645e-05, "loss": 0.2117, "step": 3530 }, { "epoch": 1.2, "learning_rate": 1.8346580170907617e-05, "loss": 0.2299, "step": 3531 }, { "epoch": 1.21, "learning_rate": 1.8333126916903215e-05, "loss": 0.2233, "step": 3532 }, { "epoch": 1.21, "learning_rate": 1.8319675741006704e-05, "loss": 0.2184, "step": 3533 }, { "epoch": 1.21, "learning_rate": 1.8306226647410902e-05, "loss": 0.2071, "step": 3534 }, { "epoch": 1.21, "learning_rate": 1.829277964030798e-05, "loss": 0.2408, "step": 3535 }, { "epoch": 1.21, "learning_rate": 1.8279334723889454e-05, "loss": 0.2391, "step": 3536 }, { "epoch": 1.21, "learning_rate": 1.82658919023462e-05, "loss": 0.2126, "step": 3537 }, { "epoch": 1.21, "learning_rate": 1.825245117986843e-05, "loss": 0.225, "step": 3538 }, { "epoch": 1.21, "learning_rate": 1.8239012560645706e-05, "loss": 0.2021, "step": 3539 }, { "epoch": 1.21, "learning_rate": 1.822557604886693e-05, "loss": 0.2437, "step": 3540 }, { "epoch": 1.21, "learning_rate": 1.8212141648720346e-05, "loss": 0.1965, "step": 3541 }, { "epoch": 1.21, "learning_rate": 1.8198709364393557e-05, "loss": 0.225, "step": 3542 }, { "epoch": 1.21, "learning_rate": 1.818527920007349e-05, "loss": 0.2136, "step": 3543 }, { "epoch": 1.21, "learning_rate": 1.8171851159946408e-05, "loss": 0.2166, "step": 3544 }, { "epoch": 1.21, "learning_rate": 1.815842524819793e-05, "loss": 0.2764, "step": 3545 }, { "epoch": 1.21, "learning_rate": 1.814500146901299e-05, "loss": 0.1969, "step": 3546 }, { "epoch": 1.21, "learning_rate": 1.8131579826575872e-05, "loss": 0.2003, "step": 3547 }, { "epoch": 1.21, "learning_rate": 1.8118160325070194e-05, "loss": 0.2384, "step": 3548 }, { "epoch": 1.21, "learning_rate": 1.8104742968678887e-05, "loss": 0.2278, "step": 3549 }, { "epoch": 1.21, "learning_rate": 1.8091327761584244e-05, "loss": 0.2281, "step": 3550 }, { "epoch": 1.21, "learning_rate": 1.807791470796787e-05, "loss": 0.2204, "step": 3551 }, { "epoch": 1.21, "learning_rate": 1.80645038120107e-05, "loss": 0.2073, "step": 3552 }, { "epoch": 1.21, "learning_rate": 1.8051095077892998e-05, "loss": 0.2269, "step": 3553 }, { "epoch": 1.21, "learning_rate": 1.8037688509794347e-05, "loss": 0.2198, "step": 3554 }, { "epoch": 1.21, "learning_rate": 1.8024284111893665e-05, "loss": 0.23, "step": 3555 }, { "epoch": 1.21, "learning_rate": 1.80108818883692e-05, "loss": 0.2267, "step": 3556 }, { "epoch": 1.21, "learning_rate": 1.79974818433985e-05, "loss": 0.2142, "step": 3557 }, { "epoch": 1.21, "learning_rate": 1.798408398115846e-05, "loss": 0.2508, "step": 3558 }, { "epoch": 1.21, "learning_rate": 1.7970688305825272e-05, "loss": 0.225, "step": 3559 }, { "epoch": 1.21, "learning_rate": 1.7957294821574456e-05, "loss": 0.2021, "step": 3560 }, { "epoch": 1.21, "learning_rate": 1.7943903532580854e-05, "loss": 0.2195, "step": 3561 }, { "epoch": 1.22, "learning_rate": 1.7930514443018603e-05, "loss": 0.2432, "step": 3562 }, { "epoch": 1.22, "learning_rate": 1.7917127557061188e-05, "loss": 0.2296, "step": 3563 }, { "epoch": 1.22, "learning_rate": 1.7903742878881387e-05, "loss": 0.2082, "step": 3564 }, { "epoch": 1.22, "learning_rate": 1.789036041265128e-05, "loss": 0.2289, "step": 3565 }, { "epoch": 1.22, "learning_rate": 1.7876980162542272e-05, "loss": 0.2318, "step": 3566 }, { "epoch": 1.22, "learning_rate": 1.7863602132725084e-05, "loss": 0.2017, "step": 3567 }, { "epoch": 1.22, "learning_rate": 1.7850226327369725e-05, "loss": 0.2247, "step": 3568 }, { "epoch": 1.22, "learning_rate": 1.7836852750645518e-05, "loss": 0.228, "step": 3569 }, { "epoch": 1.22, "learning_rate": 1.7823481406721106e-05, "loss": 0.216, "step": 3570 }, { "epoch": 1.22, "learning_rate": 1.7810112299764418e-05, "loss": 0.2089, "step": 3571 }, { "epoch": 1.22, "learning_rate": 1.7796745433942702e-05, "loss": 0.2155, "step": 3572 }, { "epoch": 1.22, "learning_rate": 1.7783380813422485e-05, "loss": 0.2168, "step": 3573 }, { "epoch": 1.22, "learning_rate": 1.777001844236961e-05, "loss": 0.1927, "step": 3574 }, { "epoch": 1.22, "learning_rate": 1.775665832494921e-05, "loss": 0.2291, "step": 3575 }, { "epoch": 1.22, "learning_rate": 1.774330046532573e-05, "loss": 0.1973, "step": 3576 }, { "epoch": 1.22, "learning_rate": 1.77299448676629e-05, "loss": 0.2042, "step": 3577 }, { "epoch": 1.22, "learning_rate": 1.771659153612375e-05, "loss": 0.2051, "step": 3578 }, { "epoch": 1.22, "learning_rate": 1.770324047487059e-05, "loss": 0.2187, "step": 3579 }, { "epoch": 1.22, "learning_rate": 1.768989168806504e-05, "loss": 0.2038, "step": 3580 }, { "epoch": 1.22, "learning_rate": 1.7676545179867998e-05, "loss": 0.2423, "step": 3581 }, { "epoch": 1.22, "learning_rate": 1.766320095443967e-05, "loss": 0.2269, "step": 3582 }, { "epoch": 1.22, "learning_rate": 1.764985901593952e-05, "loss": 0.2201, "step": 3583 }, { "epoch": 1.22, "learning_rate": 1.7636519368526337e-05, "loss": 0.2369, "step": 3584 }, { "epoch": 1.22, "learning_rate": 1.762318201635817e-05, "loss": 0.2246, "step": 3585 }, { "epoch": 1.22, "learning_rate": 1.760984696359236e-05, "loss": 0.2166, "step": 3586 }, { "epoch": 1.22, "learning_rate": 1.7596514214385518e-05, "loss": 0.2076, "step": 3587 }, { "epoch": 1.22, "learning_rate": 1.7583183772893558e-05, "loss": 0.2368, "step": 3588 }, { "epoch": 1.22, "learning_rate": 1.756985564327167e-05, "loss": 0.2272, "step": 3589 }, { "epoch": 1.22, "learning_rate": 1.7556529829674318e-05, "loss": 0.2319, "step": 3590 }, { "epoch": 1.23, "learning_rate": 1.7543206336255236e-05, "loss": 0.2059, "step": 3591 }, { "epoch": 1.23, "learning_rate": 1.7529885167167453e-05, "loss": 0.1967, "step": 3592 }, { "epoch": 1.23, "learning_rate": 1.751656632656326e-05, "loss": 0.1975, "step": 3593 }, { "epoch": 1.23, "learning_rate": 1.750324981859422e-05, "loss": 0.2276, "step": 3594 }, { "epoch": 1.23, "learning_rate": 1.7489935647411194e-05, "loss": 0.2052, "step": 3595 }, { "epoch": 1.23, "learning_rate": 1.7476623817164266e-05, "loss": 0.2193, "step": 3596 }, { "epoch": 1.23, "learning_rate": 1.746331433200286e-05, "loss": 0.2099, "step": 3597 }, { "epoch": 1.23, "learning_rate": 1.74500071960756e-05, "loss": 0.2385, "step": 3598 }, { "epoch": 1.23, "learning_rate": 1.7436702413530427e-05, "loss": 0.2291, "step": 3599 }, { "epoch": 1.23, "learning_rate": 1.742339998851451e-05, "loss": 0.2197, "step": 3600 }, { "epoch": 1.23, "learning_rate": 1.7410099925174307e-05, "loss": 0.2211, "step": 3601 }, { "epoch": 1.23, "learning_rate": 1.739680222765554e-05, "loss": 0.2326, "step": 3602 }, { "epoch": 1.23, "learning_rate": 1.7383506900103187e-05, "loss": 0.2009, "step": 3603 }, { "epoch": 1.23, "learning_rate": 1.737021394666149e-05, "loss": 0.2151, "step": 3604 }, { "epoch": 1.23, "learning_rate": 1.7356923371473942e-05, "loss": 0.2273, "step": 3605 }, { "epoch": 1.23, "learning_rate": 1.7343635178683308e-05, "loss": 0.2216, "step": 3606 }, { "epoch": 1.23, "learning_rate": 1.73303493724316e-05, "loss": 0.2136, "step": 3607 }, { "epoch": 1.23, "learning_rate": 1.7317065956860098e-05, "loss": 0.1768, "step": 3608 }, { "epoch": 1.23, "learning_rate": 1.7303784936109326e-05, "loss": 0.2115, "step": 3609 }, { "epoch": 1.23, "learning_rate": 1.729050631431905e-05, "loss": 0.2451, "step": 3610 }, { "epoch": 1.23, "learning_rate": 1.727723009562833e-05, "loss": 0.2003, "step": 3611 }, { "epoch": 1.23, "learning_rate": 1.7263956284175436e-05, "loss": 0.2059, "step": 3612 }, { "epoch": 1.23, "learning_rate": 1.7250684884097906e-05, "loss": 0.2283, "step": 3613 }, { "epoch": 1.23, "learning_rate": 1.723741589953251e-05, "loss": 0.2275, "step": 3614 }, { "epoch": 1.23, "learning_rate": 1.7224149334615294e-05, "loss": 0.2294, "step": 3615 }, { "epoch": 1.23, "learning_rate": 1.7210885193481523e-05, "loss": 0.2501, "step": 3616 }, { "epoch": 1.23, "learning_rate": 1.7197623480265722e-05, "loss": 0.2076, "step": 3617 }, { "epoch": 1.23, "learning_rate": 1.718436419910165e-05, "loss": 0.195, "step": 3618 }, { "epoch": 1.23, "learning_rate": 1.7171107354122308e-05, "loss": 0.2197, "step": 3619 }, { "epoch": 1.24, "learning_rate": 1.715785294945994e-05, "loss": 0.2199, "step": 3620 }, { "epoch": 1.24, "learning_rate": 1.7144600989246044e-05, "loss": 0.2029, "step": 3621 }, { "epoch": 1.24, "learning_rate": 1.713135147761133e-05, "loss": 0.2052, "step": 3622 }, { "epoch": 1.24, "learning_rate": 1.7118104418685766e-05, "loss": 0.217, "step": 3623 }, { "epoch": 1.24, "learning_rate": 1.710485981659853e-05, "loss": 0.2257, "step": 3624 }, { "epoch": 1.24, "learning_rate": 1.7091617675478073e-05, "loss": 0.2256, "step": 3625 }, { "epoch": 1.24, "learning_rate": 1.7078377999452043e-05, "loss": 0.2128, "step": 3626 }, { "epoch": 1.24, "learning_rate": 1.706514079264734e-05, "loss": 0.2488, "step": 3627 }, { "epoch": 1.24, "learning_rate": 1.7051906059190088e-05, "loss": 0.2345, "step": 3628 }, { "epoch": 1.24, "learning_rate": 1.7038673803205636e-05, "loss": 0.2379, "step": 3629 }, { "epoch": 1.24, "learning_rate": 1.7025444028818566e-05, "loss": 0.205, "step": 3630 }, { "epoch": 1.24, "learning_rate": 1.7012216740152687e-05, "loss": 0.2171, "step": 3631 }, { "epoch": 1.24, "learning_rate": 1.6998991941331028e-05, "loss": 0.2073, "step": 3632 }, { "epoch": 1.24, "learning_rate": 1.698576963647584e-05, "loss": 0.2515, "step": 3633 }, { "epoch": 1.24, "learning_rate": 1.697254982970861e-05, "loss": 0.2474, "step": 3634 }, { "epoch": 1.24, "learning_rate": 1.6959332525150046e-05, "loss": 0.203, "step": 3635 }, { "epoch": 1.24, "learning_rate": 1.6946117726920054e-05, "loss": 0.2282, "step": 3636 }, { "epoch": 1.24, "learning_rate": 1.6932905439137776e-05, "loss": 0.253, "step": 3637 }, { "epoch": 1.24, "learning_rate": 1.6919695665921558e-05, "loss": 0.2347, "step": 3638 }, { "epoch": 1.24, "learning_rate": 1.6906488411389e-05, "loss": 0.2397, "step": 3639 }, { "epoch": 1.24, "learning_rate": 1.689328367965686e-05, "loss": 0.1672, "step": 3640 }, { "epoch": 1.24, "learning_rate": 1.688008147484117e-05, "loss": 0.2351, "step": 3641 }, { "epoch": 1.24, "learning_rate": 1.686688180105712e-05, "loss": 0.2069, "step": 3642 }, { "epoch": 1.24, "learning_rate": 1.6853684662419142e-05, "loss": 0.2254, "step": 3643 }, { "epoch": 1.24, "learning_rate": 1.684049006304087e-05, "loss": 0.2039, "step": 3644 }, { "epoch": 1.24, "learning_rate": 1.6827298007035146e-05, "loss": 0.2278, "step": 3645 }, { "epoch": 1.24, "learning_rate": 1.6814108498514015e-05, "loss": 0.2134, "step": 3646 }, { "epoch": 1.24, "learning_rate": 1.6800921541588744e-05, "loss": 0.1957, "step": 3647 }, { "epoch": 1.24, "learning_rate": 1.6787737140369788e-05, "loss": 0.2221, "step": 3648 }, { "epoch": 1.24, "learning_rate": 1.6774555298966808e-05, "loss": 0.2486, "step": 3649 }, { "epoch": 1.25, "learning_rate": 1.676137602148868e-05, "loss": 0.2328, "step": 3650 }, { "epoch": 1.25, "learning_rate": 1.6748199312043452e-05, "loss": 0.233, "step": 3651 }, { "epoch": 1.25, "learning_rate": 1.6735025174738404e-05, "loss": 0.2425, "step": 3652 }, { "epoch": 1.25, "learning_rate": 1.672185361368e-05, "loss": 0.2202, "step": 3653 }, { "epoch": 1.25, "learning_rate": 1.670868463297391e-05, "loss": 0.209, "step": 3654 }, { "epoch": 1.25, "learning_rate": 1.6695518236724976e-05, "loss": 0.212, "step": 3655 }, { "epoch": 1.25, "learning_rate": 1.6682354429037257e-05, "loss": 0.2138, "step": 3656 }, { "epoch": 1.25, "learning_rate": 1.6669193214013994e-05, "loss": 0.2086, "step": 3657 }, { "epoch": 1.25, "learning_rate": 1.665603459575763e-05, "loss": 0.2368, "step": 3658 }, { "epoch": 1.25, "learning_rate": 1.664287857836978e-05, "loss": 0.3149, "step": 3659 }, { "epoch": 1.25, "learning_rate": 1.6629725165951277e-05, "loss": 0.2442, "step": 3660 }, { "epoch": 1.25, "learning_rate": 1.6616574362602118e-05, "loss": 0.2208, "step": 3661 }, { "epoch": 1.25, "learning_rate": 1.6603426172421487e-05, "loss": 0.2031, "step": 3662 }, { "epoch": 1.25, "learning_rate": 1.6590280599507773e-05, "loss": 0.2123, "step": 3663 }, { "epoch": 1.25, "learning_rate": 1.657713764795853e-05, "loss": 0.216, "step": 3664 }, { "epoch": 1.25, "learning_rate": 1.6563997321870502e-05, "loss": 0.1914, "step": 3665 }, { "epoch": 1.25, "learning_rate": 1.655085962533961e-05, "loss": 0.2499, "step": 3666 }, { "epoch": 1.25, "learning_rate": 1.6537724562460975e-05, "loss": 0.1992, "step": 3667 }, { "epoch": 1.25, "learning_rate": 1.652459213732887e-05, "loss": 0.2262, "step": 3668 }, { "epoch": 1.25, "learning_rate": 1.6511462354036762e-05, "loss": 0.2332, "step": 3669 }, { "epoch": 1.25, "learning_rate": 1.6498335216677293e-05, "loss": 0.2334, "step": 3670 }, { "epoch": 1.25, "learning_rate": 1.6485210729342266e-05, "loss": 0.1947, "step": 3671 }, { "epoch": 1.25, "learning_rate": 1.6472088896122676e-05, "loss": 0.2077, "step": 3672 }, { "epoch": 1.25, "learning_rate": 1.6458969721108687e-05, "loss": 0.2316, "step": 3673 }, { "epoch": 1.25, "learning_rate": 1.644585320838963e-05, "loss": 0.2232, "step": 3674 }, { "epoch": 1.25, "learning_rate": 1.6432739362054007e-05, "loss": 0.2123, "step": 3675 }, { "epoch": 1.25, "learning_rate": 1.641962818618949e-05, "loss": 0.2061, "step": 3676 }, { "epoch": 1.25, "learning_rate": 1.640651968488291e-05, "loss": 0.2112, "step": 3677 }, { "epoch": 1.25, "learning_rate": 1.639341386222029e-05, "loss": 0.2006, "step": 3678 }, { "epoch": 1.26, "learning_rate": 1.6380310722286784e-05, "loss": 0.2062, "step": 3679 }, { "epoch": 1.26, "learning_rate": 1.6367210269166727e-05, "loss": 0.221, "step": 3680 }, { "epoch": 1.26, "learning_rate": 1.6354112506943635e-05, "loss": 0.1921, "step": 3681 }, { "epoch": 1.26, "learning_rate": 1.6341017439700147e-05, "loss": 0.2318, "step": 3682 }, { "epoch": 1.26, "learning_rate": 1.632792507151809e-05, "loss": 0.2253, "step": 3683 }, { "epoch": 1.26, "learning_rate": 1.6314835406478434e-05, "loss": 0.2255, "step": 3684 }, { "epoch": 1.26, "learning_rate": 1.6301748448661312e-05, "loss": 0.2274, "step": 3685 }, { "epoch": 1.26, "learning_rate": 1.628866420214603e-05, "loss": 0.1994, "step": 3686 }, { "epoch": 1.26, "learning_rate": 1.6275582671011018e-05, "loss": 0.2198, "step": 3687 }, { "epoch": 1.26, "learning_rate": 1.626250385933388e-05, "loss": 0.2135, "step": 3688 }, { "epoch": 1.26, "learning_rate": 1.624942777119136e-05, "loss": 0.2193, "step": 3689 }, { "epoch": 1.26, "learning_rate": 1.6236354410659364e-05, "loss": 0.2331, "step": 3690 }, { "epoch": 1.26, "learning_rate": 1.622328378181295e-05, "loss": 0.2184, "step": 3691 }, { "epoch": 1.26, "learning_rate": 1.6210215888726315e-05, "loss": 0.2253, "step": 3692 }, { "epoch": 1.26, "learning_rate": 1.6197150735472797e-05, "loss": 0.2196, "step": 3693 }, { "epoch": 1.26, "learning_rate": 1.6184088326124907e-05, "loss": 0.2226, "step": 3694 }, { "epoch": 1.26, "learning_rate": 1.6171028664754274e-05, "loss": 0.2098, "step": 3695 }, { "epoch": 1.26, "learning_rate": 1.615797175543168e-05, "loss": 0.2196, "step": 3696 }, { "epoch": 1.26, "learning_rate": 1.614491760222705e-05, "loss": 0.2164, "step": 3697 }, { "epoch": 1.26, "learning_rate": 1.6131866209209444e-05, "loss": 0.2343, "step": 3698 }, { "epoch": 1.26, "learning_rate": 1.6118817580447076e-05, "loss": 0.219, "step": 3699 }, { "epoch": 1.26, "learning_rate": 1.610577172000728e-05, "loss": 0.2352, "step": 3700 }, { "epoch": 1.26, "learning_rate": 1.6092728631956543e-05, "loss": 0.1883, "step": 3701 }, { "epoch": 1.26, "learning_rate": 1.6079688320360476e-05, "loss": 0.2247, "step": 3702 }, { "epoch": 1.26, "learning_rate": 1.6066650789283823e-05, "loss": 0.2093, "step": 3703 }, { "epoch": 1.26, "learning_rate": 1.6053616042790484e-05, "loss": 0.2143, "step": 3704 }, { "epoch": 1.26, "learning_rate": 1.6040584084943462e-05, "loss": 0.2241, "step": 3705 }, { "epoch": 1.26, "learning_rate": 1.6027554919804912e-05, "loss": 0.226, "step": 3706 }, { "epoch": 1.26, "learning_rate": 1.6014528551436088e-05, "loss": 0.214, "step": 3707 }, { "epoch": 1.27, "learning_rate": 1.6001504983897425e-05, "loss": 0.2634, "step": 3708 }, { "epoch": 1.27, "learning_rate": 1.598848422124844e-05, "loss": 0.222, "step": 3709 }, { "epoch": 1.27, "learning_rate": 1.597546626754779e-05, "loss": 0.2073, "step": 3710 }, { "epoch": 1.27, "learning_rate": 1.596245112685325e-05, "loss": 0.1945, "step": 3711 }, { "epoch": 1.27, "learning_rate": 1.5949438803221734e-05, "loss": 0.203, "step": 3712 }, { "epoch": 1.27, "learning_rate": 1.5936429300709264e-05, "loss": 0.2126, "step": 3713 }, { "epoch": 1.27, "learning_rate": 1.592342262337099e-05, "loss": 0.2113, "step": 3714 }, { "epoch": 1.27, "learning_rate": 1.5910418775261178e-05, "loss": 0.2106, "step": 3715 }, { "epoch": 1.27, "learning_rate": 1.5897417760433205e-05, "loss": 0.2264, "step": 3716 }, { "epoch": 1.27, "learning_rate": 1.588441958293958e-05, "loss": 0.2054, "step": 3717 }, { "epoch": 1.27, "learning_rate": 1.5871424246831917e-05, "loss": 0.2043, "step": 3718 }, { "epoch": 1.27, "learning_rate": 1.585843175616095e-05, "loss": 0.2333, "step": 3719 }, { "epoch": 1.27, "learning_rate": 1.584544211497652e-05, "loss": 0.2289, "step": 3720 }, { "epoch": 1.27, "learning_rate": 1.5832455327327573e-05, "loss": 0.1879, "step": 3721 }, { "epoch": 1.27, "learning_rate": 1.5819471397262193e-05, "loss": 0.2564, "step": 3722 }, { "epoch": 1.27, "learning_rate": 1.5806490328827556e-05, "loss": 0.2387, "step": 3723 }, { "epoch": 1.27, "learning_rate": 1.579351212606993e-05, "loss": 0.3985, "step": 3724 }, { "epoch": 1.27, "learning_rate": 1.578053679303472e-05, "loss": 0.2095, "step": 3725 }, { "epoch": 1.27, "learning_rate": 1.5767564333766417e-05, "loss": 0.2244, "step": 3726 }, { "epoch": 1.27, "learning_rate": 1.5754594752308623e-05, "loss": 0.2252, "step": 3727 }, { "epoch": 1.27, "learning_rate": 1.574162805270403e-05, "loss": 0.2167, "step": 3728 }, { "epoch": 1.27, "learning_rate": 1.5728664238994455e-05, "loss": 0.2147, "step": 3729 }, { "epoch": 1.27, "learning_rate": 1.571570331522081e-05, "loss": 0.185, "step": 3730 }, { "epoch": 1.27, "learning_rate": 1.570274528542308e-05, "loss": 0.2443, "step": 3731 }, { "epoch": 1.27, "learning_rate": 1.5689790153640382e-05, "loss": 0.2418, "step": 3732 }, { "epoch": 1.27, "learning_rate": 1.5676837923910914e-05, "loss": 0.2119, "step": 3733 }, { "epoch": 1.27, "learning_rate": 1.566388860027196e-05, "loss": 0.2034, "step": 3734 }, { "epoch": 1.27, "learning_rate": 1.5650942186759914e-05, "loss": 0.2296, "step": 3735 }, { "epoch": 1.27, "learning_rate": 1.5637998687410266e-05, "loss": 0.2108, "step": 3736 }, { "epoch": 1.27, "learning_rate": 1.562505810625758e-05, "loss": 0.2266, "step": 3737 }, { "epoch": 1.28, "learning_rate": 1.5612120447335525e-05, "loss": 0.2106, "step": 3738 }, { "epoch": 1.28, "learning_rate": 1.5599185714676857e-05, "loss": 0.2061, "step": 3739 }, { "epoch": 1.28, "learning_rate": 1.558625391231341e-05, "loss": 0.2108, "step": 3740 }, { "epoch": 1.28, "learning_rate": 1.5573325044276107e-05, "loss": 0.2017, "step": 3741 }, { "epoch": 1.28, "learning_rate": 1.5560399114594964e-05, "loss": 0.2145, "step": 3742 }, { "epoch": 1.28, "learning_rate": 1.5547476127299088e-05, "loss": 0.2098, "step": 3743 }, { "epoch": 1.28, "learning_rate": 1.5534556086416648e-05, "loss": 0.224, "step": 3744 }, { "epoch": 1.28, "learning_rate": 1.5521638995974907e-05, "loss": 0.2151, "step": 3745 }, { "epoch": 1.28, "learning_rate": 1.5508724860000214e-05, "loss": 0.1961, "step": 3746 }, { "epoch": 1.28, "learning_rate": 1.5495813682517967e-05, "loss": 0.1908, "step": 3747 }, { "epoch": 1.28, "learning_rate": 1.548290546755269e-05, "loss": 0.2068, "step": 3748 }, { "epoch": 1.28, "learning_rate": 1.5470000219127944e-05, "loss": 0.2079, "step": 3749 }, { "epoch": 1.28, "learning_rate": 1.5457097941266382e-05, "loss": 0.2183, "step": 3750 }, { "epoch": 1.28, "learning_rate": 1.5444198637989737e-05, "loss": 0.2012, "step": 3751 }, { "epoch": 1.28, "learning_rate": 1.5431302313318796e-05, "loss": 0.2142, "step": 3752 }, { "epoch": 1.28, "learning_rate": 1.541840897127343e-05, "loss": 0.2369, "step": 3753 }, { "epoch": 1.28, "learning_rate": 1.5405518615872577e-05, "loss": 0.2294, "step": 3754 }, { "epoch": 1.28, "learning_rate": 1.5392631251134244e-05, "loss": 0.2036, "step": 3755 }, { "epoch": 1.28, "learning_rate": 1.5379746881075513e-05, "loss": 0.2218, "step": 3756 }, { "epoch": 1.28, "learning_rate": 1.5366865509712526e-05, "loss": 0.2164, "step": 3757 }, { "epoch": 1.28, "learning_rate": 1.535398714106048e-05, "loss": 0.1849, "step": 3758 }, { "epoch": 1.28, "learning_rate": 1.5341111779133656e-05, "loss": 0.2555, "step": 3759 }, { "epoch": 1.28, "learning_rate": 1.5328239427945385e-05, "loss": 0.2162, "step": 3760 }, { "epoch": 1.28, "learning_rate": 1.5315370091508063e-05, "loss": 0.2105, "step": 3761 }, { "epoch": 1.28, "learning_rate": 1.530250377383315e-05, "loss": 0.225, "step": 3762 }, { "epoch": 1.28, "learning_rate": 1.528964047893115e-05, "loss": 0.2122, "step": 3763 }, { "epoch": 1.28, "learning_rate": 1.527678021081165e-05, "loss": 0.2164, "step": 3764 }, { "epoch": 1.28, "learning_rate": 1.526392297348328e-05, "loss": 0.2165, "step": 3765 }, { "epoch": 1.28, "learning_rate": 1.5251068770953713e-05, "loss": 0.2351, "step": 3766 }, { "epoch": 1.29, "learning_rate": 1.5238217607229693e-05, "loss": 0.241, "step": 3767 }, { "epoch": 1.29, "learning_rate": 1.5225369486317013e-05, "loss": 0.2042, "step": 3768 }, { "epoch": 1.29, "learning_rate": 1.521252441222052e-05, "loss": 0.2103, "step": 3769 }, { "epoch": 1.29, "learning_rate": 1.5199682388944097e-05, "loss": 0.2111, "step": 3770 }, { "epoch": 1.29, "learning_rate": 1.5186843420490698e-05, "loss": 0.2105, "step": 3771 }, { "epoch": 1.29, "learning_rate": 1.517400751086231e-05, "loss": 0.2354, "step": 3772 }, { "epoch": 1.29, "learning_rate": 1.5161174664059957e-05, "loss": 0.2308, "step": 3773 }, { "epoch": 1.29, "learning_rate": 1.5148344884083743e-05, "loss": 0.2149, "step": 3774 }, { "epoch": 1.29, "learning_rate": 1.5135518174932778e-05, "loss": 0.2494, "step": 3775 }, { "epoch": 1.29, "learning_rate": 1.512269454060524e-05, "loss": 0.2396, "step": 3776 }, { "epoch": 1.29, "learning_rate": 1.5109873985098325e-05, "loss": 0.2226, "step": 3777 }, { "epoch": 1.29, "learning_rate": 1.5097056512408303e-05, "loss": 0.1964, "step": 3778 }, { "epoch": 1.29, "learning_rate": 1.5084242126530456e-05, "loss": 0.223, "step": 3779 }, { "epoch": 1.29, "learning_rate": 1.5071430831459116e-05, "loss": 0.2309, "step": 3780 }, { "epoch": 1.29, "learning_rate": 1.5058622631187636e-05, "loss": 0.216, "step": 3781 }, { "epoch": 1.29, "learning_rate": 1.5045817529708426e-05, "loss": 0.2299, "step": 3782 }, { "epoch": 1.29, "learning_rate": 1.5033015531012918e-05, "loss": 0.2284, "step": 3783 }, { "epoch": 1.29, "learning_rate": 1.5020216639091574e-05, "loss": 0.223, "step": 3784 }, { "epoch": 1.29, "learning_rate": 1.5007420857933898e-05, "loss": 0.2145, "step": 3785 }, { "epoch": 1.29, "learning_rate": 1.4994628191528409e-05, "loss": 0.1884, "step": 3786 }, { "epoch": 1.29, "learning_rate": 1.4981838643862678e-05, "loss": 0.2282, "step": 3787 }, { "epoch": 1.29, "learning_rate": 1.4969052218923286e-05, "loss": 0.2122, "step": 3788 }, { "epoch": 1.29, "learning_rate": 1.495626892069584e-05, "loss": 0.1979, "step": 3789 }, { "epoch": 1.29, "learning_rate": 1.4943488753164972e-05, "loss": 0.2284, "step": 3790 }, { "epoch": 1.29, "learning_rate": 1.4930711720314363e-05, "loss": 0.2126, "step": 3791 }, { "epoch": 1.29, "learning_rate": 1.4917937826126688e-05, "loss": 0.2133, "step": 3792 }, { "epoch": 1.29, "learning_rate": 1.4905167074583649e-05, "loss": 0.2012, "step": 3793 }, { "epoch": 1.29, "learning_rate": 1.4892399469665973e-05, "loss": 0.2165, "step": 3794 }, { "epoch": 1.29, "learning_rate": 1.4879635015353415e-05, "loss": 0.2556, "step": 3795 }, { "epoch": 1.3, "learning_rate": 1.4866873715624733e-05, "loss": 0.2292, "step": 3796 }, { "epoch": 1.3, "learning_rate": 1.4854115574457706e-05, "loss": 0.2266, "step": 3797 }, { "epoch": 1.3, "learning_rate": 1.4841360595829129e-05, "loss": 0.2251, "step": 3798 }, { "epoch": 1.3, "learning_rate": 1.4828608783714814e-05, "loss": 0.2104, "step": 3799 }, { "epoch": 1.3, "learning_rate": 1.4815860142089588e-05, "loss": 0.2172, "step": 3800 }, { "epoch": 1.3, "learning_rate": 1.480311467492728e-05, "loss": 0.2402, "step": 3801 }, { "epoch": 1.3, "learning_rate": 1.479037238620074e-05, "loss": 0.2224, "step": 3802 }, { "epoch": 1.3, "learning_rate": 1.477763327988182e-05, "loss": 0.2259, "step": 3803 }, { "epoch": 1.3, "learning_rate": 1.476489735994137e-05, "loss": 0.2156, "step": 3804 }, { "epoch": 1.3, "learning_rate": 1.4752164630349286e-05, "loss": 0.2097, "step": 3805 }, { "epoch": 1.3, "learning_rate": 1.4739435095074428e-05, "loss": 0.2238, "step": 3806 }, { "epoch": 1.3, "learning_rate": 1.4726708758084673e-05, "loss": 0.2217, "step": 3807 }, { "epoch": 1.3, "learning_rate": 1.471398562334691e-05, "loss": 0.2431, "step": 3808 }, { "epoch": 1.3, "learning_rate": 1.470126569482702e-05, "loss": 0.2007, "step": 3809 }, { "epoch": 1.3, "learning_rate": 1.4688548976489892e-05, "loss": 0.2346, "step": 3810 }, { "epoch": 1.3, "learning_rate": 1.4675835472299403e-05, "loss": 0.2069, "step": 3811 }, { "epoch": 1.3, "learning_rate": 1.4663125186218435e-05, "loss": 0.2361, "step": 3812 }, { "epoch": 1.3, "learning_rate": 1.4650418122208876e-05, "loss": 0.2047, "step": 3813 }, { "epoch": 1.3, "learning_rate": 1.4637714284231591e-05, "loss": 0.2137, "step": 3814 }, { "epoch": 1.3, "learning_rate": 1.4625013676246457e-05, "loss": 0.1989, "step": 3815 }, { "epoch": 1.3, "learning_rate": 1.4612316302212336e-05, "loss": 0.1949, "step": 3816 }, { "epoch": 1.3, "learning_rate": 1.459962216608708e-05, "loss": 0.2077, "step": 3817 }, { "epoch": 1.3, "learning_rate": 1.4586931271827514e-05, "loss": 0.2144, "step": 3818 }, { "epoch": 1.3, "learning_rate": 1.4574243623389516e-05, "loss": 0.2217, "step": 3819 }, { "epoch": 1.3, "learning_rate": 1.4561559224727877e-05, "loss": 0.2585, "step": 3820 }, { "epoch": 1.3, "learning_rate": 1.4548878079796425e-05, "loss": 0.2251, "step": 3821 }, { "epoch": 1.3, "learning_rate": 1.4536200192547933e-05, "loss": 0.2038, "step": 3822 }, { "epoch": 1.3, "learning_rate": 1.4523525566934196e-05, "loss": 0.1931, "step": 3823 }, { "epoch": 1.3, "learning_rate": 1.4510854206905986e-05, "loss": 0.1995, "step": 3824 }, { "epoch": 1.31, "learning_rate": 1.449818611641303e-05, "loss": 0.2072, "step": 3825 }, { "epoch": 1.31, "learning_rate": 1.448552129940407e-05, "loss": 0.2208, "step": 3826 }, { "epoch": 1.31, "learning_rate": 1.4472859759826798e-05, "loss": 0.2177, "step": 3827 }, { "epoch": 1.31, "learning_rate": 1.4460201501627908e-05, "loss": 0.2155, "step": 3828 }, { "epoch": 1.31, "learning_rate": 1.4447546528753068e-05, "loss": 0.2257, "step": 3829 }, { "epoch": 1.31, "learning_rate": 1.4434894845146896e-05, "loss": 0.2248, "step": 3830 }, { "epoch": 1.31, "learning_rate": 1.4422246454753027e-05, "loss": 0.2428, "step": 3831 }, { "epoch": 1.31, "learning_rate": 1.4409601361514017e-05, "loss": 0.1855, "step": 3832 }, { "epoch": 1.31, "learning_rate": 1.439695956937146e-05, "loss": 0.227, "step": 3833 }, { "epoch": 1.31, "learning_rate": 1.4384321082265862e-05, "loss": 0.2117, "step": 3834 }, { "epoch": 1.31, "learning_rate": 1.4371685904136736e-05, "loss": 0.3071, "step": 3835 }, { "epoch": 1.31, "learning_rate": 1.4359054038922531e-05, "loss": 0.1867, "step": 3836 }, { "epoch": 1.31, "learning_rate": 1.4346425490560697e-05, "loss": 0.2184, "step": 3837 }, { "epoch": 1.31, "learning_rate": 1.4333800262987638e-05, "loss": 0.2195, "step": 3838 }, { "epoch": 1.31, "learning_rate": 1.4321178360138705e-05, "loss": 0.1978, "step": 3839 }, { "epoch": 1.31, "learning_rate": 1.430855978594824e-05, "loss": 0.2029, "step": 3840 }, { "epoch": 1.31, "learning_rate": 1.4295944544349523e-05, "loss": 0.2427, "step": 3841 }, { "epoch": 1.31, "learning_rate": 1.4283332639274815e-05, "loss": 0.219, "step": 3842 }, { "epoch": 1.31, "learning_rate": 1.4270724074655339e-05, "loss": 0.1888, "step": 3843 }, { "epoch": 1.31, "learning_rate": 1.4258118854421244e-05, "loss": 0.2183, "step": 3844 }, { "epoch": 1.31, "learning_rate": 1.4245516982501683e-05, "loss": 0.2555, "step": 3845 }, { "epoch": 1.31, "learning_rate": 1.4232918462824713e-05, "loss": 0.2232, "step": 3846 }, { "epoch": 1.31, "learning_rate": 1.4220323299317411e-05, "loss": 0.218, "step": 3847 }, { "epoch": 1.31, "learning_rate": 1.4207731495905747e-05, "loss": 0.2079, "step": 3848 }, { "epoch": 1.31, "learning_rate": 1.419514305651467e-05, "loss": 0.2421, "step": 3849 }, { "epoch": 1.31, "learning_rate": 1.4182557985068095e-05, "loss": 0.2112, "step": 3850 }, { "epoch": 1.31, "learning_rate": 1.416997628548885e-05, "loss": 0.2263, "step": 3851 }, { "epoch": 1.31, "learning_rate": 1.4157397961698755e-05, "loss": 0.2329, "step": 3852 }, { "epoch": 1.31, "learning_rate": 1.4144823017618528e-05, "loss": 0.2397, "step": 3853 }, { "epoch": 1.31, "learning_rate": 1.4132251457167877e-05, "loss": 0.2096, "step": 3854 }, { "epoch": 1.32, "learning_rate": 1.4119683284265456e-05, "loss": 0.2445, "step": 3855 }, { "epoch": 1.32, "learning_rate": 1.4107118502828814e-05, "loss": 0.2625, "step": 3856 }, { "epoch": 1.32, "learning_rate": 1.4094557116774501e-05, "loss": 0.1911, "step": 3857 }, { "epoch": 1.32, "learning_rate": 1.4081999130017964e-05, "loss": 0.1976, "step": 3858 }, { "epoch": 1.32, "learning_rate": 1.4069444546473621e-05, "loss": 0.2247, "step": 3859 }, { "epoch": 1.32, "learning_rate": 1.4056893370054814e-05, "loss": 0.2148, "step": 3860 }, { "epoch": 1.32, "learning_rate": 1.4044345604673836e-05, "loss": 0.2108, "step": 3861 }, { "epoch": 1.32, "learning_rate": 1.4031801254241889e-05, "loss": 0.2206, "step": 3862 }, { "epoch": 1.32, "learning_rate": 1.4019260322669143e-05, "loss": 0.2214, "step": 3863 }, { "epoch": 1.32, "learning_rate": 1.4006722813864692e-05, "loss": 0.222, "step": 3864 }, { "epoch": 1.32, "learning_rate": 1.399418873173654e-05, "loss": 0.2124, "step": 3865 }, { "epoch": 1.32, "learning_rate": 1.398165808019167e-05, "loss": 0.257, "step": 3866 }, { "epoch": 1.32, "learning_rate": 1.3969130863135938e-05, "loss": 0.1928, "step": 3867 }, { "epoch": 1.32, "learning_rate": 1.3956607084474178e-05, "loss": 0.2491, "step": 3868 }, { "epoch": 1.32, "learning_rate": 1.3944086748110135e-05, "loss": 0.1977, "step": 3869 }, { "epoch": 1.32, "learning_rate": 1.3931569857946466e-05, "loss": 0.4025, "step": 3870 }, { "epoch": 1.32, "learning_rate": 1.3919056417884785e-05, "loss": 0.2342, "step": 3871 }, { "epoch": 1.32, "learning_rate": 1.3906546431825596e-05, "loss": 0.2318, "step": 3872 }, { "epoch": 1.32, "learning_rate": 1.3894039903668346e-05, "loss": 0.2514, "step": 3873 }, { "epoch": 1.32, "learning_rate": 1.3881536837311409e-05, "loss": 0.1932, "step": 3874 }, { "epoch": 1.32, "learning_rate": 1.3869037236652071e-05, "loss": 0.1881, "step": 3875 }, { "epoch": 1.32, "learning_rate": 1.3856541105586545e-05, "loss": 0.2364, "step": 3876 }, { "epoch": 1.32, "learning_rate": 1.3844048448009942e-05, "loss": 0.2505, "step": 3877 }, { "epoch": 1.32, "learning_rate": 1.3831559267816322e-05, "loss": 0.2396, "step": 3878 }, { "epoch": 1.32, "learning_rate": 1.3819073568898621e-05, "loss": 0.1922, "step": 3879 }, { "epoch": 1.32, "learning_rate": 1.3806591355148732e-05, "loss": 0.1825, "step": 3880 }, { "epoch": 1.32, "learning_rate": 1.3794112630457444e-05, "loss": 0.2364, "step": 3881 }, { "epoch": 1.32, "learning_rate": 1.3781637398714442e-05, "loss": 0.2548, "step": 3882 }, { "epoch": 1.32, "learning_rate": 1.3769165663808353e-05, "loss": 0.1998, "step": 3883 }, { "epoch": 1.33, "learning_rate": 1.3756697429626686e-05, "loss": 0.2069, "step": 3884 }, { "epoch": 1.33, "learning_rate": 1.374423270005587e-05, "loss": 0.2152, "step": 3885 }, { "epoch": 1.33, "learning_rate": 1.3731771478981259e-05, "loss": 0.2211, "step": 3886 }, { "epoch": 1.33, "learning_rate": 1.371931377028708e-05, "loss": 0.2306, "step": 3887 }, { "epoch": 1.33, "learning_rate": 1.3706859577856485e-05, "loss": 0.2126, "step": 3888 }, { "epoch": 1.33, "learning_rate": 1.3694408905571532e-05, "loss": 0.2152, "step": 3889 }, { "epoch": 1.33, "learning_rate": 1.3681961757313183e-05, "loss": 0.1926, "step": 3890 }, { "epoch": 1.33, "learning_rate": 1.366951813696128e-05, "loss": 0.203, "step": 3891 }, { "epoch": 1.33, "learning_rate": 1.36570780483946e-05, "loss": 0.1777, "step": 3892 }, { "epoch": 1.33, "learning_rate": 1.364464149549078e-05, "loss": 0.2178, "step": 3893 }, { "epoch": 1.33, "learning_rate": 1.3632208482126376e-05, "loss": 0.2227, "step": 3894 }, { "epoch": 1.33, "learning_rate": 1.3619779012176861e-05, "loss": 0.2068, "step": 3895 }, { "epoch": 1.33, "learning_rate": 1.3607353089516559e-05, "loss": 0.2204, "step": 3896 }, { "epoch": 1.33, "learning_rate": 1.3594930718018728e-05, "loss": 0.2344, "step": 3897 }, { "epoch": 1.33, "learning_rate": 1.3582511901555487e-05, "loss": 0.2009, "step": 3898 }, { "epoch": 1.33, "learning_rate": 1.3570096643997867e-05, "loss": 0.2117, "step": 3899 }, { "epoch": 1.33, "learning_rate": 1.3557684949215799e-05, "loss": 0.244, "step": 3900 }, { "epoch": 1.33, "learning_rate": 1.3545276821078057e-05, "loss": 0.1967, "step": 3901 }, { "epoch": 1.33, "learning_rate": 1.3532872263452379e-05, "loss": 0.2136, "step": 3902 }, { "epoch": 1.33, "learning_rate": 1.352047128020531e-05, "loss": 0.202, "step": 3903 }, { "epoch": 1.33, "learning_rate": 1.3508073875202342e-05, "loss": 0.1823, "step": 3904 }, { "epoch": 1.33, "learning_rate": 1.3495680052307807e-05, "loss": 0.2288, "step": 3905 }, { "epoch": 1.33, "learning_rate": 1.3483289815384954e-05, "loss": 0.2367, "step": 3906 }, { "epoch": 1.33, "learning_rate": 1.3470903168295904e-05, "loss": 0.2232, "step": 3907 }, { "epoch": 1.33, "learning_rate": 1.3458520114901639e-05, "loss": 0.2236, "step": 3908 }, { "epoch": 1.33, "learning_rate": 1.3446140659062056e-05, "loss": 0.2543, "step": 3909 }, { "epoch": 1.33, "learning_rate": 1.3433764804635895e-05, "loss": 0.2175, "step": 3910 }, { "epoch": 1.33, "learning_rate": 1.34213925554808e-05, "loss": 0.2017, "step": 3911 }, { "epoch": 1.33, "learning_rate": 1.340902391545329e-05, "loss": 0.198, "step": 3912 }, { "epoch": 1.34, "learning_rate": 1.3396658888408733e-05, "loss": 0.2378, "step": 3913 }, { "epoch": 1.34, "learning_rate": 1.3384297478201407e-05, "loss": 0.2259, "step": 3914 }, { "epoch": 1.34, "learning_rate": 1.3371939688684419e-05, "loss": 0.2122, "step": 3915 }, { "epoch": 1.34, "learning_rate": 1.3359585523709806e-05, "loss": 0.2133, "step": 3916 }, { "epoch": 1.34, "learning_rate": 1.3347234987128415e-05, "loss": 0.2093, "step": 3917 }, { "epoch": 1.34, "learning_rate": 1.3334888082790007e-05, "loss": 0.2569, "step": 3918 }, { "epoch": 1.34, "learning_rate": 1.3322544814543176e-05, "loss": 0.2493, "step": 3919 }, { "epoch": 1.34, "learning_rate": 1.3310205186235408e-05, "loss": 0.2346, "step": 3920 }, { "epoch": 1.34, "learning_rate": 1.3297869201713053e-05, "loss": 0.1942, "step": 3921 }, { "epoch": 1.34, "learning_rate": 1.3285536864821305e-05, "loss": 0.2053, "step": 3922 }, { "epoch": 1.34, "learning_rate": 1.3273208179404245e-05, "loss": 0.2152, "step": 3923 }, { "epoch": 1.34, "learning_rate": 1.3260883149304792e-05, "loss": 0.1924, "step": 3924 }, { "epoch": 1.34, "learning_rate": 1.3248561778364746e-05, "loss": 0.1947, "step": 3925 }, { "epoch": 1.34, "learning_rate": 1.323624407042477e-05, "loss": 0.227, "step": 3926 }, { "epoch": 1.34, "learning_rate": 1.322393002932435e-05, "loss": 0.2185, "step": 3927 }, { "epoch": 1.34, "learning_rate": 1.3211619658901876e-05, "loss": 0.239, "step": 3928 }, { "epoch": 1.34, "learning_rate": 1.3199312962994536e-05, "loss": 0.199, "step": 3929 }, { "epoch": 1.34, "learning_rate": 1.3187009945438459e-05, "loss": 0.2166, "step": 3930 }, { "epoch": 1.34, "learning_rate": 1.3174710610068536e-05, "loss": 0.2198, "step": 3931 }, { "epoch": 1.34, "learning_rate": 1.3162414960718561e-05, "loss": 0.2173, "step": 3932 }, { "epoch": 1.34, "learning_rate": 1.3150123001221185e-05, "loss": 0.1957, "step": 3933 }, { "epoch": 1.34, "learning_rate": 1.3137834735407863e-05, "loss": 0.2131, "step": 3934 }, { "epoch": 1.34, "learning_rate": 1.3125550167108955e-05, "loss": 0.2261, "step": 3935 }, { "epoch": 1.34, "learning_rate": 1.3113269300153618e-05, "loss": 0.2308, "step": 3936 }, { "epoch": 1.34, "learning_rate": 1.3100992138369888e-05, "loss": 0.1867, "step": 3937 }, { "epoch": 1.34, "learning_rate": 1.3088718685584644e-05, "loss": 0.2146, "step": 3938 }, { "epoch": 1.34, "learning_rate": 1.307644894562359e-05, "loss": 0.1942, "step": 3939 }, { "epoch": 1.34, "learning_rate": 1.3064182922311296e-05, "loss": 0.2208, "step": 3940 }, { "epoch": 1.34, "learning_rate": 1.3051920619471145e-05, "loss": 0.185, "step": 3941 }, { "epoch": 1.34, "learning_rate": 1.3039662040925396e-05, "loss": 0.2309, "step": 3942 }, { "epoch": 1.35, "learning_rate": 1.30274071904951e-05, "loss": 0.2138, "step": 3943 }, { "epoch": 1.35, "learning_rate": 1.301515607200021e-05, "loss": 0.2107, "step": 3944 }, { "epoch": 1.35, "learning_rate": 1.3002908689259451e-05, "loss": 0.2046, "step": 3945 }, { "epoch": 1.35, "learning_rate": 1.2990665046090425e-05, "loss": 0.2017, "step": 3946 }, { "epoch": 1.35, "learning_rate": 1.2978425146309564e-05, "loss": 0.1899, "step": 3947 }, { "epoch": 1.35, "learning_rate": 1.2966188993732103e-05, "loss": 0.2103, "step": 3948 }, { "epoch": 1.35, "learning_rate": 1.2953956592172151e-05, "loss": 0.2291, "step": 3949 }, { "epoch": 1.35, "learning_rate": 1.2941727945442611e-05, "loss": 0.2221, "step": 3950 }, { "epoch": 1.35, "learning_rate": 1.2929503057355235e-05, "loss": 0.202, "step": 3951 }, { "epoch": 1.35, "learning_rate": 1.291728193172062e-05, "loss": 0.2023, "step": 3952 }, { "epoch": 1.35, "learning_rate": 1.2905064572348146e-05, "loss": 0.236, "step": 3953 }, { "epoch": 1.35, "learning_rate": 1.2892850983046062e-05, "loss": 0.2156, "step": 3954 }, { "epoch": 1.35, "learning_rate": 1.2880641167621405e-05, "loss": 0.2231, "step": 3955 }, { "epoch": 1.35, "learning_rate": 1.2868435129880064e-05, "loss": 0.2223, "step": 3956 }, { "epoch": 1.35, "learning_rate": 1.2856232873626739e-05, "loss": 0.1925, "step": 3957 }, { "epoch": 1.35, "learning_rate": 1.2844034402664956e-05, "loss": 0.2089, "step": 3958 }, { "epoch": 1.35, "learning_rate": 1.2831839720797067e-05, "loss": 0.2051, "step": 3959 }, { "epoch": 1.35, "learning_rate": 1.2819648831824216e-05, "loss": 0.1919, "step": 3960 }, { "epoch": 1.35, "learning_rate": 1.28074617395464e-05, "loss": 0.2268, "step": 3961 }, { "epoch": 1.35, "learning_rate": 1.2795278447762393e-05, "loss": 0.2405, "step": 3962 }, { "epoch": 1.35, "learning_rate": 1.2783098960269823e-05, "loss": 0.219, "step": 3963 }, { "epoch": 1.35, "learning_rate": 1.2770923280865122e-05, "loss": 0.1935, "step": 3964 }, { "epoch": 1.35, "learning_rate": 1.2758751413343511e-05, "loss": 0.2242, "step": 3965 }, { "epoch": 1.35, "learning_rate": 1.2746583361499053e-05, "loss": 0.2179, "step": 3966 }, { "epoch": 1.35, "learning_rate": 1.2734419129124603e-05, "loss": 0.209, "step": 3967 }, { "epoch": 1.35, "learning_rate": 1.2722258720011838e-05, "loss": 0.2171, "step": 3968 }, { "epoch": 1.35, "learning_rate": 1.2710102137951224e-05, "loss": 0.2253, "step": 3969 }, { "epoch": 1.35, "learning_rate": 1.2697949386732055e-05, "loss": 0.2376, "step": 3970 }, { "epoch": 1.35, "learning_rate": 1.2685800470142429e-05, "loss": 0.2257, "step": 3971 }, { "epoch": 1.36, "learning_rate": 1.267365539196923e-05, "loss": 0.2275, "step": 3972 }, { "epoch": 1.36, "learning_rate": 1.266151415599818e-05, "loss": 0.2031, "step": 3973 }, { "epoch": 1.36, "learning_rate": 1.2649376766013757e-05, "loss": 0.2029, "step": 3974 }, { "epoch": 1.36, "learning_rate": 1.2637243225799283e-05, "loss": 0.2054, "step": 3975 }, { "epoch": 1.36, "learning_rate": 1.2625113539136846e-05, "loss": 0.2137, "step": 3976 }, { "epoch": 1.36, "learning_rate": 1.2612987709807356e-05, "loss": 0.2146, "step": 3977 }, { "epoch": 1.36, "learning_rate": 1.2600865741590523e-05, "loss": 0.1926, "step": 3978 }, { "epoch": 1.36, "learning_rate": 1.2588747638264822e-05, "loss": 0.2167, "step": 3979 }, { "epoch": 1.36, "learning_rate": 1.2576633403607568e-05, "loss": 0.2102, "step": 3980 }, { "epoch": 1.36, "learning_rate": 1.2564523041394826e-05, "loss": 0.2067, "step": 3981 }, { "epoch": 1.36, "learning_rate": 1.255241655540149e-05, "loss": 0.1933, "step": 3982 }, { "epoch": 1.36, "learning_rate": 1.254031394940123e-05, "loss": 0.2274, "step": 3983 }, { "epoch": 1.36, "learning_rate": 1.2528215227166495e-05, "loss": 0.2089, "step": 3984 }, { "epoch": 1.36, "learning_rate": 1.2516120392468544e-05, "loss": 0.229, "step": 3985 }, { "epoch": 1.36, "learning_rate": 1.2504029449077421e-05, "loss": 0.2255, "step": 3986 }, { "epoch": 1.36, "learning_rate": 1.249194240076195e-05, "loss": 0.2107, "step": 3987 }, { "epoch": 1.36, "learning_rate": 1.2479859251289732e-05, "loss": 0.2263, "step": 3988 }, { "epoch": 1.36, "learning_rate": 1.2467780004427173e-05, "loss": 0.1895, "step": 3989 }, { "epoch": 1.36, "learning_rate": 1.245570466393946e-05, "loss": 0.2222, "step": 3990 }, { "epoch": 1.36, "learning_rate": 1.244363323359054e-05, "loss": 0.2083, "step": 3991 }, { "epoch": 1.36, "learning_rate": 1.2431565717143171e-05, "loss": 0.2014, "step": 3992 }, { "epoch": 1.36, "learning_rate": 1.241950211835886e-05, "loss": 0.2029, "step": 3993 }, { "epoch": 1.36, "learning_rate": 1.2407442440997932e-05, "loss": 0.2306, "step": 3994 }, { "epoch": 1.36, "learning_rate": 1.2395386688819444e-05, "loss": 0.2065, "step": 3995 }, { "epoch": 1.36, "learning_rate": 1.2383334865581261e-05, "loss": 0.1666, "step": 3996 }, { "epoch": 1.36, "learning_rate": 1.2371286975040029e-05, "loss": 0.2153, "step": 3997 }, { "epoch": 1.36, "learning_rate": 1.2359243020951119e-05, "loss": 0.2259, "step": 3998 }, { "epoch": 1.36, "learning_rate": 1.2347203007068758e-05, "loss": 0.2119, "step": 3999 }, { "epoch": 1.36, "learning_rate": 1.2335166937145859e-05, "loss": 0.2116, "step": 4000 }, { "epoch": 1.37, "learning_rate": 1.232313481493417e-05, "loss": 0.2369, "step": 4001 }, { "epoch": 1.37, "learning_rate": 1.2311106644184157e-05, "loss": 0.206, "step": 4002 }, { "epoch": 1.37, "learning_rate": 1.2299082428645092e-05, "loss": 0.1658, "step": 4003 }, { "epoch": 1.37, "learning_rate": 1.2287062172065012e-05, "loss": 0.2183, "step": 4004 }, { "epoch": 1.37, "learning_rate": 1.2275045878190688e-05, "loss": 0.2121, "step": 4005 }, { "epoch": 1.37, "learning_rate": 1.2263033550767697e-05, "loss": 0.1875, "step": 4006 }, { "epoch": 1.37, "learning_rate": 1.2251025193540344e-05, "loss": 0.2037, "step": 4007 }, { "epoch": 1.37, "learning_rate": 1.2239020810251716e-05, "loss": 0.2158, "step": 4008 }, { "epoch": 1.37, "learning_rate": 1.2227020404643671e-05, "loss": 0.2166, "step": 4009 }, { "epoch": 1.37, "learning_rate": 1.2215023980456797e-05, "loss": 0.2252, "step": 4010 }, { "epoch": 1.37, "learning_rate": 1.2203031541430471e-05, "loss": 0.2264, "step": 4011 }, { "epoch": 1.37, "learning_rate": 1.219104309130279e-05, "loss": 0.2007, "step": 4012 }, { "epoch": 1.37, "learning_rate": 1.217905863381067e-05, "loss": 0.199, "step": 4013 }, { "epoch": 1.37, "learning_rate": 1.2167078172689717e-05, "loss": 0.2382, "step": 4014 }, { "epoch": 1.37, "learning_rate": 1.2155101711674324e-05, "loss": 0.2134, "step": 4015 }, { "epoch": 1.37, "learning_rate": 1.2143129254497646e-05, "loss": 0.2164, "step": 4016 }, { "epoch": 1.37, "learning_rate": 1.2131160804891556e-05, "loss": 0.2177, "step": 4017 }, { "epoch": 1.37, "learning_rate": 1.2119196366586717e-05, "loss": 0.2095, "step": 4018 }, { "epoch": 1.37, "learning_rate": 1.21072359433125e-05, "loss": 0.2099, "step": 4019 }, { "epoch": 1.37, "learning_rate": 1.2095279538797072e-05, "loss": 0.2256, "step": 4020 }, { "epoch": 1.37, "learning_rate": 1.2083327156767302e-05, "loss": 0.2339, "step": 4021 }, { "epoch": 1.37, "learning_rate": 1.2071378800948831e-05, "loss": 0.2084, "step": 4022 }, { "epoch": 1.37, "learning_rate": 1.2059434475066053e-05, "loss": 0.1948, "step": 4023 }, { "epoch": 1.37, "learning_rate": 1.2047494182842073e-05, "loss": 0.2364, "step": 4024 }, { "epoch": 1.37, "learning_rate": 1.2035557927998778e-05, "loss": 0.2275, "step": 4025 }, { "epoch": 1.37, "learning_rate": 1.2023625714256743e-05, "loss": 0.2117, "step": 4026 }, { "epoch": 1.37, "learning_rate": 1.201169754533536e-05, "loss": 0.2136, "step": 4027 }, { "epoch": 1.37, "learning_rate": 1.1999773424952687e-05, "loss": 0.213, "step": 4028 }, { "epoch": 1.37, "learning_rate": 1.198785335682556e-05, "loss": 0.2231, "step": 4029 }, { "epoch": 1.37, "learning_rate": 1.1975937344669549e-05, "loss": 0.2164, "step": 4030 }, { "epoch": 1.38, "learning_rate": 1.1964025392198937e-05, "loss": 0.2292, "step": 4031 }, { "epoch": 1.38, "learning_rate": 1.1952117503126778e-05, "loss": 0.2216, "step": 4032 }, { "epoch": 1.38, "learning_rate": 1.1940213681164816e-05, "loss": 0.235, "step": 4033 }, { "epoch": 1.38, "learning_rate": 1.1928313930023563e-05, "loss": 0.2031, "step": 4034 }, { "epoch": 1.38, "learning_rate": 1.1916418253412253e-05, "loss": 0.2157, "step": 4035 }, { "epoch": 1.38, "learning_rate": 1.1904526655038834e-05, "loss": 0.2002, "step": 4036 }, { "epoch": 1.38, "learning_rate": 1.189263913861001e-05, "loss": 0.2044, "step": 4037 }, { "epoch": 1.38, "learning_rate": 1.1880755707831178e-05, "loss": 0.196, "step": 4038 }, { "epoch": 1.38, "learning_rate": 1.1868876366406492e-05, "loss": 0.2202, "step": 4039 }, { "epoch": 1.38, "learning_rate": 1.1857001118038821e-05, "loss": 0.2218, "step": 4040 }, { "epoch": 1.38, "learning_rate": 1.1845129966429755e-05, "loss": 0.2332, "step": 4041 }, { "epoch": 1.38, "learning_rate": 1.1833262915279622e-05, "loss": 0.2107, "step": 4042 }, { "epoch": 1.38, "learning_rate": 1.1821399968287435e-05, "loss": 0.1997, "step": 4043 }, { "epoch": 1.38, "learning_rate": 1.1809541129150981e-05, "loss": 0.213, "step": 4044 }, { "epoch": 1.38, "learning_rate": 1.179768640156671e-05, "loss": 0.2105, "step": 4045 }, { "epoch": 1.38, "learning_rate": 1.1785835789229837e-05, "loss": 0.1787, "step": 4046 }, { "epoch": 1.38, "learning_rate": 1.1773989295834262e-05, "loss": 0.2239, "step": 4047 }, { "epoch": 1.38, "learning_rate": 1.176214692507262e-05, "loss": 0.2153, "step": 4048 }, { "epoch": 1.38, "learning_rate": 1.1750308680636266e-05, "loss": 0.2205, "step": 4049 }, { "epoch": 1.38, "learning_rate": 1.1738474566215238e-05, "loss": 0.2014, "step": 4050 }, { "epoch": 1.38, "learning_rate": 1.1726644585498326e-05, "loss": 0.2094, "step": 4051 }, { "epoch": 1.38, "learning_rate": 1.1714818742172995e-05, "loss": 0.2178, "step": 4052 }, { "epoch": 1.38, "learning_rate": 1.1702997039925445e-05, "loss": 0.2223, "step": 4053 }, { "epoch": 1.38, "learning_rate": 1.1691179482440578e-05, "loss": 0.22, "step": 4054 }, { "epoch": 1.38, "learning_rate": 1.1679366073402004e-05, "loss": 0.2011, "step": 4055 }, { "epoch": 1.38, "learning_rate": 1.166755681649205e-05, "loss": 0.2338, "step": 4056 }, { "epoch": 1.38, "learning_rate": 1.1655751715391722e-05, "loss": 0.2126, "step": 4057 }, { "epoch": 1.38, "learning_rate": 1.1643950773780763e-05, "loss": 0.2109, "step": 4058 }, { "epoch": 1.38, "learning_rate": 1.1632153995337583e-05, "loss": 0.1969, "step": 4059 }, { "epoch": 1.39, "learning_rate": 1.1620361383739328e-05, "loss": 0.2028, "step": 4060 }, { "epoch": 1.39, "learning_rate": 1.1608572942661838e-05, "loss": 0.2499, "step": 4061 }, { "epoch": 1.39, "learning_rate": 1.1596788675779632e-05, "loss": 0.2421, "step": 4062 }, { "epoch": 1.39, "learning_rate": 1.1585008586765959e-05, "loss": 0.2045, "step": 4063 }, { "epoch": 1.39, "learning_rate": 1.1573232679292733e-05, "loss": 0.2115, "step": 4064 }, { "epoch": 1.39, "learning_rate": 1.1561460957030587e-05, "loss": 0.2056, "step": 4065 }, { "epoch": 1.39, "learning_rate": 1.1549693423648858e-05, "loss": 0.2423, "step": 4066 }, { "epoch": 1.39, "learning_rate": 1.1537930082815543e-05, "loss": 0.2108, "step": 4067 }, { "epoch": 1.39, "learning_rate": 1.1526170938197358e-05, "loss": 0.243, "step": 4068 }, { "epoch": 1.39, "learning_rate": 1.1514415993459706e-05, "loss": 0.2121, "step": 4069 }, { "epoch": 1.39, "learning_rate": 1.1502665252266695e-05, "loss": 0.2344, "step": 4070 }, { "epoch": 1.39, "learning_rate": 1.1490918718281084e-05, "loss": 0.2368, "step": 4071 }, { "epoch": 1.39, "learning_rate": 1.1479176395164364e-05, "loss": 0.1929, "step": 4072 }, { "epoch": 1.39, "learning_rate": 1.1467438286576676e-05, "loss": 0.2314, "step": 4073 }, { "epoch": 1.39, "learning_rate": 1.1455704396176873e-05, "loss": 0.2244, "step": 4074 }, { "epoch": 1.39, "learning_rate": 1.1443974727622494e-05, "loss": 0.2022, "step": 4075 }, { "epoch": 1.39, "learning_rate": 1.1432249284569737e-05, "loss": 0.1995, "step": 4076 }, { "epoch": 1.39, "learning_rate": 1.1420528070673519e-05, "loss": 0.222, "step": 4077 }, { "epoch": 1.39, "learning_rate": 1.1408811089587399e-05, "loss": 0.2074, "step": 4078 }, { "epoch": 1.39, "learning_rate": 1.1397098344963642e-05, "loss": 0.1964, "step": 4079 }, { "epoch": 1.39, "learning_rate": 1.13853898404532e-05, "loss": 0.2144, "step": 4080 }, { "epoch": 1.39, "learning_rate": 1.1373685579705673e-05, "loss": 0.2177, "step": 4081 }, { "epoch": 1.39, "learning_rate": 1.1361985566369359e-05, "loss": 0.2086, "step": 4082 }, { "epoch": 1.39, "learning_rate": 1.1350289804091235e-05, "loss": 0.1965, "step": 4083 }, { "epoch": 1.39, "learning_rate": 1.1338598296516956e-05, "loss": 0.1948, "step": 4084 }, { "epoch": 1.39, "learning_rate": 1.1326911047290817e-05, "loss": 0.2014, "step": 4085 }, { "epoch": 1.39, "learning_rate": 1.1315228060055821e-05, "loss": 0.1986, "step": 4086 }, { "epoch": 1.39, "learning_rate": 1.1303549338453646e-05, "loss": 0.211, "step": 4087 }, { "epoch": 1.39, "learning_rate": 1.12918748861246e-05, "loss": 0.241, "step": 4088 }, { "epoch": 1.4, "learning_rate": 1.128020470670771e-05, "loss": 0.2194, "step": 4089 }, { "epoch": 1.4, "learning_rate": 1.1268538803840626e-05, "loss": 0.193, "step": 4090 }, { "epoch": 1.4, "learning_rate": 1.1256877181159694e-05, "loss": 0.2231, "step": 4091 }, { "epoch": 1.4, "learning_rate": 1.1245219842299931e-05, "loss": 0.2312, "step": 4092 }, { "epoch": 1.4, "learning_rate": 1.1233566790894987e-05, "loss": 0.2291, "step": 4093 }, { "epoch": 1.4, "learning_rate": 1.1221918030577208e-05, "loss": 0.2452, "step": 4094 }, { "epoch": 1.4, "learning_rate": 1.1210273564977564e-05, "loss": 0.2206, "step": 4095 }, { "epoch": 1.4, "learning_rate": 1.119863339772575e-05, "loss": 0.1998, "step": 4096 }, { "epoch": 1.4, "learning_rate": 1.118699753245005e-05, "loss": 0.2244, "step": 4097 }, { "epoch": 1.4, "learning_rate": 1.1175365972777462e-05, "loss": 0.2212, "step": 4098 }, { "epoch": 1.4, "learning_rate": 1.1163738722333597e-05, "loss": 0.2025, "step": 4099 }, { "epoch": 1.4, "learning_rate": 1.1152115784742758e-05, "loss": 0.211, "step": 4100 }, { "epoch": 1.4, "learning_rate": 1.1140497163627897e-05, "loss": 0.2284, "step": 4101 }, { "epoch": 1.4, "learning_rate": 1.1128882862610595e-05, "loss": 0.2189, "step": 4102 }, { "epoch": 1.4, "learning_rate": 1.1117272885311128e-05, "loss": 0.2079, "step": 4103 }, { "epoch": 1.4, "learning_rate": 1.1105667235348376e-05, "loss": 0.21, "step": 4104 }, { "epoch": 1.4, "learning_rate": 1.1094065916339912e-05, "loss": 0.2151, "step": 4105 }, { "epoch": 1.4, "learning_rate": 1.1082468931901952e-05, "loss": 0.2081, "step": 4106 }, { "epoch": 1.4, "learning_rate": 1.107087628564933e-05, "loss": 0.2113, "step": 4107 }, { "epoch": 1.4, "learning_rate": 1.1059287981195569e-05, "loss": 0.2193, "step": 4108 }, { "epoch": 1.4, "learning_rate": 1.1047704022152792e-05, "loss": 0.1895, "step": 4109 }, { "epoch": 1.4, "learning_rate": 1.1036124412131832e-05, "loss": 0.1883, "step": 4110 }, { "epoch": 1.4, "learning_rate": 1.1024549154742098e-05, "loss": 0.2156, "step": 4111 }, { "epoch": 1.4, "learning_rate": 1.1012978253591687e-05, "loss": 0.2139, "step": 4112 }, { "epoch": 1.4, "learning_rate": 1.1001411712287332e-05, "loss": 0.2204, "step": 4113 }, { "epoch": 1.4, "learning_rate": 1.0989849534434383e-05, "loss": 0.2254, "step": 4114 }, { "epoch": 1.4, "learning_rate": 1.097829172363686e-05, "loss": 0.2142, "step": 4115 }, { "epoch": 1.4, "learning_rate": 1.0966738283497394e-05, "loss": 0.2167, "step": 4116 }, { "epoch": 1.4, "learning_rate": 1.0955189217617277e-05, "loss": 0.2117, "step": 4117 }, { "epoch": 1.4, "learning_rate": 1.0943644529596434e-05, "loss": 0.197, "step": 4118 }, { "epoch": 1.41, "learning_rate": 1.0932104223033402e-05, "loss": 0.2144, "step": 4119 }, { "epoch": 1.41, "learning_rate": 1.0920568301525392e-05, "loss": 0.1995, "step": 4120 }, { "epoch": 1.41, "learning_rate": 1.0909036768668204e-05, "loss": 0.2053, "step": 4121 }, { "epoch": 1.41, "learning_rate": 1.0897509628056304e-05, "loss": 0.2266, "step": 4122 }, { "epoch": 1.41, "learning_rate": 1.0885986883282775e-05, "loss": 0.1984, "step": 4123 }, { "epoch": 1.41, "learning_rate": 1.0874468537939342e-05, "loss": 0.2126, "step": 4124 }, { "epoch": 1.41, "learning_rate": 1.0862954595616328e-05, "loss": 0.2195, "step": 4125 }, { "epoch": 1.41, "learning_rate": 1.0851445059902718e-05, "loss": 0.2063, "step": 4126 }, { "epoch": 1.41, "learning_rate": 1.0839939934386114e-05, "loss": 0.2934, "step": 4127 }, { "epoch": 1.41, "learning_rate": 1.0828439222652723e-05, "loss": 0.2183, "step": 4128 }, { "epoch": 1.41, "learning_rate": 1.0816942928287408e-05, "loss": 0.2066, "step": 4129 }, { "epoch": 1.41, "learning_rate": 1.0805451054873623e-05, "loss": 0.2028, "step": 4130 }, { "epoch": 1.41, "learning_rate": 1.0793963605993468e-05, "loss": 0.2136, "step": 4131 }, { "epoch": 1.41, "learning_rate": 1.0782480585227666e-05, "loss": 0.2286, "step": 4132 }, { "epoch": 1.41, "learning_rate": 1.077100199615553e-05, "loss": 0.2328, "step": 4133 }, { "epoch": 1.41, "learning_rate": 1.075952784235503e-05, "loss": 0.2145, "step": 4134 }, { "epoch": 1.41, "learning_rate": 1.0748058127402717e-05, "loss": 0.1868, "step": 4135 }, { "epoch": 1.41, "learning_rate": 1.0736592854873783e-05, "loss": 0.2257, "step": 4136 }, { "epoch": 1.41, "learning_rate": 1.0725132028342033e-05, "loss": 0.2094, "step": 4137 }, { "epoch": 1.41, "learning_rate": 1.0713675651379875e-05, "loss": 0.2282, "step": 4138 }, { "epoch": 1.41, "learning_rate": 1.0702223727558352e-05, "loss": 0.2405, "step": 4139 }, { "epoch": 1.41, "learning_rate": 1.0690776260447084e-05, "loss": 0.2077, "step": 4140 }, { "epoch": 1.41, "learning_rate": 1.0679333253614332e-05, "loss": 0.2184, "step": 4141 }, { "epoch": 1.41, "learning_rate": 1.066789471062695e-05, "loss": 0.2106, "step": 4142 }, { "epoch": 1.41, "learning_rate": 1.0656460635050405e-05, "loss": 0.2052, "step": 4143 }, { "epoch": 1.41, "learning_rate": 1.0645031030448788e-05, "loss": 0.1872, "step": 4144 }, { "epoch": 1.41, "learning_rate": 1.0633605900384764e-05, "loss": 0.2255, "step": 4145 }, { "epoch": 1.41, "learning_rate": 1.0622185248419634e-05, "loss": 0.2226, "step": 4146 }, { "epoch": 1.41, "learning_rate": 1.0610769078113272e-05, "loss": 0.2314, "step": 4147 }, { "epoch": 1.42, "learning_rate": 1.059935739302419e-05, "loss": 0.1993, "step": 4148 }, { "epoch": 1.42, "learning_rate": 1.0587950196709484e-05, "loss": 0.2309, "step": 4149 }, { "epoch": 1.42, "learning_rate": 1.0576547492724838e-05, "loss": 0.2026, "step": 4150 }, { "epoch": 1.42, "learning_rate": 1.0565149284624558e-05, "loss": 0.2016, "step": 4151 }, { "epoch": 1.42, "learning_rate": 1.055375557596154e-05, "loss": 0.2057, "step": 4152 }, { "epoch": 1.42, "learning_rate": 1.0542366370287288e-05, "loss": 0.2157, "step": 4153 }, { "epoch": 1.42, "learning_rate": 1.0530981671151871e-05, "loss": 0.2173, "step": 4154 }, { "epoch": 1.42, "learning_rate": 1.0519601482103992e-05, "loss": 0.1787, "step": 4155 }, { "epoch": 1.42, "learning_rate": 1.0508225806690913e-05, "loss": 0.2324, "step": 4156 }, { "epoch": 1.42, "learning_rate": 1.0496854648458518e-05, "loss": 0.1948, "step": 4157 }, { "epoch": 1.42, "learning_rate": 1.0485488010951277e-05, "loss": 0.2143, "step": 4158 }, { "epoch": 1.42, "learning_rate": 1.0474125897712228e-05, "loss": 0.1942, "step": 4159 }, { "epoch": 1.42, "learning_rate": 1.0462768312283033e-05, "loss": 0.197, "step": 4160 }, { "epoch": 1.42, "learning_rate": 1.0451415258203912e-05, "loss": 0.2152, "step": 4161 }, { "epoch": 1.42, "learning_rate": 1.044006673901369e-05, "loss": 0.207, "step": 4162 }, { "epoch": 1.42, "learning_rate": 1.0428722758249784e-05, "loss": 0.2468, "step": 4163 }, { "epoch": 1.42, "learning_rate": 1.0417383319448174e-05, "loss": 0.2039, "step": 4164 }, { "epoch": 1.42, "learning_rate": 1.040604842614344e-05, "loss": 0.2083, "step": 4165 }, { "epoch": 1.42, "learning_rate": 1.0394718081868743e-05, "loss": 0.2175, "step": 4166 }, { "epoch": 1.42, "learning_rate": 1.0383392290155838e-05, "loss": 0.2727, "step": 4167 }, { "epoch": 1.42, "learning_rate": 1.037207105453503e-05, "loss": 0.1683, "step": 4168 }, { "epoch": 1.42, "learning_rate": 1.0360754378535226e-05, "loss": 0.2011, "step": 4169 }, { "epoch": 1.42, "learning_rate": 1.034944226568392e-05, "loss": 0.2224, "step": 4170 }, { "epoch": 1.42, "learning_rate": 1.0338134719507156e-05, "loss": 0.2081, "step": 4171 }, { "epoch": 1.42, "learning_rate": 1.0326831743529586e-05, "loss": 0.1954, "step": 4172 }, { "epoch": 1.42, "learning_rate": 1.0315533341274403e-05, "loss": 0.2033, "step": 4173 }, { "epoch": 1.42, "learning_rate": 1.03042395162634e-05, "loss": 0.1954, "step": 4174 }, { "epoch": 1.42, "learning_rate": 1.029295027201695e-05, "loss": 0.2598, "step": 4175 }, { "epoch": 1.42, "learning_rate": 1.0281665612053962e-05, "loss": 0.1743, "step": 4176 }, { "epoch": 1.43, "learning_rate": 1.0270385539891955e-05, "loss": 0.213, "step": 4177 }, { "epoch": 1.43, "learning_rate": 1.0259110059046989e-05, "loss": 0.1963, "step": 4178 }, { "epoch": 1.43, "learning_rate": 1.0247839173033707e-05, "loss": 0.2064, "step": 4179 }, { "epoch": 1.43, "learning_rate": 1.0236572885365325e-05, "loss": 0.2494, "step": 4180 }, { "epoch": 1.43, "learning_rate": 1.0225311199553622e-05, "loss": 0.2055, "step": 4181 }, { "epoch": 1.43, "learning_rate": 1.0214054119108923e-05, "loss": 0.1929, "step": 4182 }, { "epoch": 1.43, "learning_rate": 1.0202801647540142e-05, "loss": 0.2171, "step": 4183 }, { "epoch": 1.43, "learning_rate": 1.019155378835476e-05, "loss": 0.1944, "step": 4184 }, { "epoch": 1.43, "learning_rate": 1.0180310545058785e-05, "loss": 0.2345, "step": 4185 }, { "epoch": 1.43, "learning_rate": 1.016907192115683e-05, "loss": 0.2135, "step": 4186 }, { "epoch": 1.43, "learning_rate": 1.015783792015203e-05, "loss": 0.2174, "step": 4187 }, { "epoch": 1.43, "learning_rate": 1.0146608545546105e-05, "loss": 0.2378, "step": 4188 }, { "epoch": 1.43, "learning_rate": 1.0135383800839333e-05, "loss": 0.1846, "step": 4189 }, { "epoch": 1.43, "learning_rate": 1.0124163689530522e-05, "loss": 0.1904, "step": 4190 }, { "epoch": 1.43, "learning_rate": 1.011294821511707e-05, "loss": 0.2576, "step": 4191 }, { "epoch": 1.43, "learning_rate": 1.0101737381094895e-05, "loss": 0.221, "step": 4192 }, { "epoch": 1.43, "learning_rate": 1.0090531190958514e-05, "loss": 0.2381, "step": 4193 }, { "epoch": 1.43, "learning_rate": 1.0079329648200948e-05, "loss": 0.2136, "step": 4194 }, { "epoch": 1.43, "learning_rate": 1.0068132756313794e-05, "loss": 0.222, "step": 4195 }, { "epoch": 1.43, "learning_rate": 1.005694051878721e-05, "loss": 0.2034, "step": 4196 }, { "epoch": 1.43, "learning_rate": 1.0045752939109876e-05, "loss": 0.2164, "step": 4197 }, { "epoch": 1.43, "learning_rate": 1.0034570020769043e-05, "loss": 0.1969, "step": 4198 }, { "epoch": 1.43, "learning_rate": 1.0023391767250484e-05, "loss": 0.2273, "step": 4199 }, { "epoch": 1.43, "learning_rate": 1.0012218182038546e-05, "loss": 0.1719, "step": 4200 }, { "epoch": 1.43, "learning_rate": 1.0001049268616117e-05, "loss": 0.2072, "step": 4201 }, { "epoch": 1.43, "learning_rate": 9.989885030464598e-06, "loss": 0.2191, "step": 4202 }, { "epoch": 1.43, "learning_rate": 9.978725471063977e-06, "loss": 0.2009, "step": 4203 }, { "epoch": 1.43, "learning_rate": 9.967570593892742e-06, "loss": 0.2127, "step": 4204 }, { "epoch": 1.43, "learning_rate": 9.956420402427957e-06, "loss": 0.2192, "step": 4205 }, { "epoch": 1.44, "learning_rate": 9.945274900145188e-06, "loss": 0.1987, "step": 4206 }, { "epoch": 1.44, "learning_rate": 9.934134090518593e-06, "loss": 0.1953, "step": 4207 }, { "epoch": 1.44, "learning_rate": 9.922997977020809e-06, "loss": 0.1929, "step": 4208 }, { "epoch": 1.44, "learning_rate": 9.911866563123042e-06, "loss": 0.2259, "step": 4209 }, { "epoch": 1.44, "learning_rate": 9.900739852295041e-06, "loss": 0.2324, "step": 4210 }, { "epoch": 1.44, "learning_rate": 9.889617848005054e-06, "loss": 0.1879, "step": 4211 }, { "epoch": 1.44, "learning_rate": 9.878500553719893e-06, "loss": 0.2042, "step": 4212 }, { "epoch": 1.44, "learning_rate": 9.867387972904885e-06, "loss": 0.213, "step": 4213 }, { "epoch": 1.44, "learning_rate": 9.856280109023897e-06, "loss": 0.1981, "step": 4214 }, { "epoch": 1.44, "learning_rate": 9.845176965539333e-06, "loss": 0.2321, "step": 4215 }, { "epoch": 1.44, "learning_rate": 9.834078545912095e-06, "loss": 0.2361, "step": 4216 }, { "epoch": 1.44, "learning_rate": 9.822984853601652e-06, "loss": 0.2291, "step": 4217 }, { "epoch": 1.44, "learning_rate": 9.811895892065967e-06, "loss": 0.2255, "step": 4218 }, { "epoch": 1.44, "learning_rate": 9.800811664761545e-06, "loss": 0.1958, "step": 4219 }, { "epoch": 1.44, "learning_rate": 9.789732175143413e-06, "loss": 0.224, "step": 4220 }, { "epoch": 1.44, "learning_rate": 9.77865742666512e-06, "loss": 0.2042, "step": 4221 }, { "epoch": 1.44, "learning_rate": 9.76758742277875e-06, "loss": 0.1992, "step": 4222 }, { "epoch": 1.44, "learning_rate": 9.756522166934873e-06, "loss": 0.1756, "step": 4223 }, { "epoch": 1.44, "learning_rate": 9.745461662582622e-06, "loss": 0.2292, "step": 4224 }, { "epoch": 1.44, "learning_rate": 9.734405913169612e-06, "loss": 0.2378, "step": 4225 }, { "epoch": 1.44, "learning_rate": 9.723354922141998e-06, "loss": 0.2, "step": 4226 }, { "epoch": 1.44, "learning_rate": 9.712308692944458e-06, "loss": 0.2213, "step": 4227 }, { "epoch": 1.44, "learning_rate": 9.701267229020153e-06, "loss": 0.2293, "step": 4228 }, { "epoch": 1.44, "learning_rate": 9.690230533810798e-06, "loss": 0.2094, "step": 4229 }, { "epoch": 1.44, "learning_rate": 9.679198610756588e-06, "loss": 0.2135, "step": 4230 }, { "epoch": 1.44, "learning_rate": 9.66817146329626e-06, "loss": 0.2345, "step": 4231 }, { "epoch": 1.44, "learning_rate": 9.65714909486703e-06, "loss": 0.219, "step": 4232 }, { "epoch": 1.44, "learning_rate": 9.646131508904654e-06, "loss": 0.1884, "step": 4233 }, { "epoch": 1.44, "learning_rate": 9.635118708843386e-06, "loss": 0.2437, "step": 4234 }, { "epoch": 1.44, "learning_rate": 9.624110698115982e-06, "loss": 0.2031, "step": 4235 }, { "epoch": 1.45, "learning_rate": 9.613107480153723e-06, "loss": 0.209, "step": 4236 }, { "epoch": 1.45, "learning_rate": 9.60210905838637e-06, "loss": 0.2075, "step": 4237 }, { "epoch": 1.45, "learning_rate": 9.591115436242215e-06, "loss": 0.2134, "step": 4238 }, { "epoch": 1.45, "learning_rate": 9.580126617148025e-06, "loss": 0.1841, "step": 4239 }, { "epoch": 1.45, "learning_rate": 9.5691426045291e-06, "loss": 0.1943, "step": 4240 }, { "epoch": 1.45, "learning_rate": 9.558163401809231e-06, "loss": 0.2129, "step": 4241 }, { "epoch": 1.45, "learning_rate": 9.547189012410693e-06, "loss": 0.2228, "step": 4242 }, { "epoch": 1.45, "learning_rate": 9.536219439754292e-06, "loss": 0.2271, "step": 4243 }, { "epoch": 1.45, "learning_rate": 9.525254687259302e-06, "loss": 0.2238, "step": 4244 }, { "epoch": 1.45, "learning_rate": 9.514294758343509e-06, "loss": 0.2207, "step": 4245 }, { "epoch": 1.45, "learning_rate": 9.50333965642321e-06, "loss": 0.2038, "step": 4246 }, { "epoch": 1.45, "learning_rate": 9.492389384913158e-06, "loss": 0.1983, "step": 4247 }, { "epoch": 1.45, "learning_rate": 9.481443947226636e-06, "loss": 0.2237, "step": 4248 }, { "epoch": 1.45, "learning_rate": 9.470503346775414e-06, "loss": 0.2403, "step": 4249 }, { "epoch": 1.45, "learning_rate": 9.459567586969748e-06, "loss": 0.2191, "step": 4250 }, { "epoch": 1.45, "learning_rate": 9.448636671218372e-06, "loss": 0.2344, "step": 4251 }, { "epoch": 1.45, "learning_rate": 9.437710602928535e-06, "loss": 0.2031, "step": 4252 }, { "epoch": 1.45, "learning_rate": 9.42678938550597e-06, "loss": 0.2448, "step": 4253 }, { "epoch": 1.45, "learning_rate": 9.41587302235487e-06, "loss": 0.2102, "step": 4254 }, { "epoch": 1.45, "learning_rate": 9.404961516877964e-06, "loss": 0.1956, "step": 4255 }, { "epoch": 1.45, "learning_rate": 9.394054872476419e-06, "loss": 0.1964, "step": 4256 }, { "epoch": 1.45, "learning_rate": 9.383153092549918e-06, "loss": 0.2275, "step": 4257 }, { "epoch": 1.45, "learning_rate": 9.372256180496605e-06, "loss": 0.213, "step": 4258 }, { "epoch": 1.45, "learning_rate": 9.361364139713131e-06, "loss": 0.207, "step": 4259 }, { "epoch": 1.45, "learning_rate": 9.350476973594615e-06, "loss": 0.1993, "step": 4260 }, { "epoch": 1.45, "learning_rate": 9.33959468553465e-06, "loss": 0.2028, "step": 4261 }, { "epoch": 1.45, "learning_rate": 9.328717278925323e-06, "loss": 0.2403, "step": 4262 }, { "epoch": 1.45, "learning_rate": 9.317844757157188e-06, "loss": 0.1885, "step": 4263 }, { "epoch": 1.45, "learning_rate": 9.306977123619293e-06, "loss": 0.2206, "step": 4264 }, { "epoch": 1.46, "learning_rate": 9.29611438169913e-06, "loss": 0.2195, "step": 4265 }, { "epoch": 1.46, "learning_rate": 9.285256534782702e-06, "loss": 0.2464, "step": 4266 }, { "epoch": 1.46, "learning_rate": 9.274403586254473e-06, "loss": 0.2054, "step": 4267 }, { "epoch": 1.46, "learning_rate": 9.263555539497362e-06, "loss": 0.2477, "step": 4268 }, { "epoch": 1.46, "learning_rate": 9.252712397892798e-06, "loss": 0.2426, "step": 4269 }, { "epoch": 1.46, "learning_rate": 9.241874164820635e-06, "loss": 0.2027, "step": 4270 }, { "epoch": 1.46, "learning_rate": 9.231040843659233e-06, "loss": 0.1856, "step": 4271 }, { "epoch": 1.46, "learning_rate": 9.22021243778542e-06, "loss": 0.2365, "step": 4272 }, { "epoch": 1.46, "learning_rate": 9.20938895057446e-06, "loss": 0.1934, "step": 4273 }, { "epoch": 1.46, "learning_rate": 9.198570385400121e-06, "loss": 0.1878, "step": 4274 }, { "epoch": 1.46, "learning_rate": 9.187756745634608e-06, "loss": 0.2061, "step": 4275 }, { "epoch": 1.46, "learning_rate": 9.176948034648606e-06, "loss": 0.2226, "step": 4276 }, { "epoch": 1.46, "learning_rate": 9.166144255811266e-06, "loss": 0.2053, "step": 4277 }, { "epoch": 1.46, "learning_rate": 9.155345412490193e-06, "loss": 0.1882, "step": 4278 }, { "epoch": 1.46, "learning_rate": 9.144551508051466e-06, "loss": 0.2161, "step": 4279 }, { "epoch": 1.46, "learning_rate": 9.1337625458596e-06, "loss": 0.1814, "step": 4280 }, { "epoch": 1.46, "learning_rate": 9.1229785292776e-06, "loss": 0.2035, "step": 4281 }, { "epoch": 1.46, "learning_rate": 9.112199461666901e-06, "loss": 0.2127, "step": 4282 }, { "epoch": 1.46, "learning_rate": 9.101425346387421e-06, "loss": 0.2322, "step": 4283 }, { "epoch": 1.46, "learning_rate": 9.090656186797508e-06, "loss": 0.2364, "step": 4284 }, { "epoch": 1.46, "learning_rate": 9.079891986253988e-06, "loss": 0.2016, "step": 4285 }, { "epoch": 1.46, "learning_rate": 9.069132748112142e-06, "loss": 0.2087, "step": 4286 }, { "epoch": 1.46, "learning_rate": 9.058378475725673e-06, "loss": 0.1974, "step": 4287 }, { "epoch": 1.46, "learning_rate": 9.047629172446781e-06, "loss": 0.2118, "step": 4288 }, { "epoch": 1.46, "learning_rate": 9.036884841626064e-06, "loss": 0.2187, "step": 4289 }, { "epoch": 1.46, "learning_rate": 9.026145486612639e-06, "loss": 0.218, "step": 4290 }, { "epoch": 1.46, "learning_rate": 9.015411110754002e-06, "loss": 0.202, "step": 4291 }, { "epoch": 1.46, "learning_rate": 9.004681717396139e-06, "loss": 0.1965, "step": 4292 }, { "epoch": 1.46, "learning_rate": 8.99395730988348e-06, "loss": 0.1891, "step": 4293 }, { "epoch": 1.47, "learning_rate": 8.983237891558876e-06, "loss": 0.1897, "step": 4294 }, { "epoch": 1.47, "learning_rate": 8.97252346576366e-06, "loss": 0.1818, "step": 4295 }, { "epoch": 1.47, "learning_rate": 8.961814035837569e-06, "loss": 0.2195, "step": 4296 }, { "epoch": 1.47, "learning_rate": 8.951109605118807e-06, "loss": 0.2197, "step": 4297 }, { "epoch": 1.47, "learning_rate": 8.940410176944028e-06, "loss": 0.2407, "step": 4298 }, { "epoch": 1.47, "learning_rate": 8.929715754648297e-06, "loss": 0.2329, "step": 4299 }, { "epoch": 1.47, "learning_rate": 8.919026341565152e-06, "loss": 0.1951, "step": 4300 }, { "epoch": 1.47, "learning_rate": 8.908341941026533e-06, "loss": 0.21, "step": 4301 }, { "epoch": 1.47, "learning_rate": 8.897662556362847e-06, "loss": 0.2263, "step": 4302 }, { "epoch": 1.47, "learning_rate": 8.886988190902929e-06, "loss": 0.1822, "step": 4303 }, { "epoch": 1.47, "learning_rate": 8.876318847974058e-06, "loss": 0.1796, "step": 4304 }, { "epoch": 1.47, "learning_rate": 8.865654530901918e-06, "loss": 0.2254, "step": 4305 }, { "epoch": 1.47, "learning_rate": 8.854995243010655e-06, "loss": 0.2202, "step": 4306 }, { "epoch": 1.47, "learning_rate": 8.844340987622851e-06, "loss": 0.1917, "step": 4307 }, { "epoch": 1.47, "learning_rate": 8.833691768059488e-06, "loss": 0.1789, "step": 4308 }, { "epoch": 1.47, "learning_rate": 8.823047587640012e-06, "loss": 0.1896, "step": 4309 }, { "epoch": 1.47, "learning_rate": 8.812408449682269e-06, "loss": 0.2185, "step": 4310 }, { "epoch": 1.47, "learning_rate": 8.801774357502554e-06, "loss": 0.217, "step": 4311 }, { "epoch": 1.47, "learning_rate": 8.791145314415594e-06, "loss": 0.2088, "step": 4312 }, { "epoch": 1.47, "learning_rate": 8.780521323734516e-06, "loss": 0.2223, "step": 4313 }, { "epoch": 1.47, "learning_rate": 8.769902388770899e-06, "loss": 0.2108, "step": 4314 }, { "epoch": 1.47, "learning_rate": 8.75928851283472e-06, "loss": 0.1977, "step": 4315 }, { "epoch": 1.47, "learning_rate": 8.748679699234403e-06, "loss": 0.2228, "step": 4316 }, { "epoch": 1.47, "learning_rate": 8.738075951276784e-06, "loss": 0.2167, "step": 4317 }, { "epoch": 1.47, "learning_rate": 8.727477272267117e-06, "loss": 0.223, "step": 4318 }, { "epoch": 1.47, "learning_rate": 8.716883665509094e-06, "loss": 0.192, "step": 4319 }, { "epoch": 1.47, "learning_rate": 8.70629513430479e-06, "loss": 0.2149, "step": 4320 }, { "epoch": 1.47, "learning_rate": 8.695711681954737e-06, "loss": 0.2242, "step": 4321 }, { "epoch": 1.47, "learning_rate": 8.685133311757853e-06, "loss": 0.2196, "step": 4322 }, { "epoch": 1.47, "learning_rate": 8.674560027011488e-06, "loss": 0.2306, "step": 4323 }, { "epoch": 1.48, "learning_rate": 8.663991831011415e-06, "loss": 0.2188, "step": 4324 }, { "epoch": 1.48, "learning_rate": 8.653428727051793e-06, "loss": 0.1828, "step": 4325 }, { "epoch": 1.48, "learning_rate": 8.642870718425227e-06, "loss": 0.1966, "step": 4326 }, { "epoch": 1.48, "learning_rate": 8.632317808422702e-06, "loss": 0.23, "step": 4327 }, { "epoch": 1.48, "learning_rate": 8.621770000333635e-06, "loss": 0.2117, "step": 4328 }, { "epoch": 1.48, "learning_rate": 8.611227297445857e-06, "loss": 0.2095, "step": 4329 }, { "epoch": 1.48, "learning_rate": 8.600689703045578e-06, "loss": 0.2356, "step": 4330 }, { "epoch": 1.48, "learning_rate": 8.590157220417445e-06, "loss": 0.2319, "step": 4331 }, { "epoch": 1.48, "learning_rate": 8.5796298528445e-06, "loss": 0.2535, "step": 4332 }, { "epoch": 1.48, "learning_rate": 8.569107603608206e-06, "loss": 0.1928, "step": 4333 }, { "epoch": 1.48, "learning_rate": 8.558590475988395e-06, "loss": 0.2453, "step": 4334 }, { "epoch": 1.48, "learning_rate": 8.548078473263344e-06, "loss": 0.1915, "step": 4335 }, { "epoch": 1.48, "learning_rate": 8.53757159870969e-06, "loss": 0.362, "step": 4336 }, { "epoch": 1.48, "learning_rate": 8.527069855602512e-06, "loss": 0.1975, "step": 4337 }, { "epoch": 1.48, "learning_rate": 8.516573247215273e-06, "loss": 0.1986, "step": 4338 }, { "epoch": 1.48, "learning_rate": 8.506081776819821e-06, "loss": 0.229, "step": 4339 }, { "epoch": 1.48, "learning_rate": 8.495595447686434e-06, "loss": 0.2313, "step": 4340 }, { "epoch": 1.48, "learning_rate": 8.485114263083749e-06, "loss": 0.203, "step": 4341 }, { "epoch": 1.48, "learning_rate": 8.474638226278827e-06, "loss": 0.2237, "step": 4342 }, { "epoch": 1.48, "learning_rate": 8.464167340537132e-06, "loss": 0.2111, "step": 4343 }, { "epoch": 1.48, "learning_rate": 8.453701609122486e-06, "loss": 0.2063, "step": 4344 }, { "epoch": 1.48, "learning_rate": 8.443241035297133e-06, "loss": 0.1943, "step": 4345 }, { "epoch": 1.48, "learning_rate": 8.432785622321707e-06, "loss": 0.2011, "step": 4346 }, { "epoch": 1.48, "learning_rate": 8.422335373455232e-06, "loss": 0.2046, "step": 4347 }, { "epoch": 1.48, "learning_rate": 8.411890291955105e-06, "loss": 0.1798, "step": 4348 }, { "epoch": 1.48, "learning_rate": 8.401450381077134e-06, "loss": 0.2125, "step": 4349 }, { "epoch": 1.48, "learning_rate": 8.391015644075514e-06, "loss": 0.2107, "step": 4350 }, { "epoch": 1.48, "learning_rate": 8.380586084202807e-06, "loss": 0.2288, "step": 4351 }, { "epoch": 1.48, "learning_rate": 8.370161704709992e-06, "loss": 0.228, "step": 4352 }, { "epoch": 1.49, "learning_rate": 8.359742508846394e-06, "loss": 0.26, "step": 4353 }, { "epoch": 1.49, "learning_rate": 8.34932849985976e-06, "loss": 0.2003, "step": 4354 }, { "epoch": 1.49, "learning_rate": 8.338919680996207e-06, "loss": 0.197, "step": 4355 }, { "epoch": 1.49, "learning_rate": 8.328516055500218e-06, "loss": 0.2137, "step": 4356 }, { "epoch": 1.49, "learning_rate": 8.318117626614685e-06, "loss": 0.2305, "step": 4357 }, { "epoch": 1.49, "learning_rate": 8.307724397580854e-06, "loss": 0.1997, "step": 4358 }, { "epoch": 1.49, "learning_rate": 8.29733637163837e-06, "loss": 0.1766, "step": 4359 }, { "epoch": 1.49, "learning_rate": 8.286953552025245e-06, "loss": 0.1736, "step": 4360 }, { "epoch": 1.49, "learning_rate": 8.276575941977881e-06, "loss": 0.2095, "step": 4361 }, { "epoch": 1.49, "learning_rate": 8.266203544731033e-06, "loss": 0.2144, "step": 4362 }, { "epoch": 1.49, "learning_rate": 8.255836363517852e-06, "loss": 0.2415, "step": 4363 }, { "epoch": 1.49, "learning_rate": 8.245474401569863e-06, "loss": 0.2024, "step": 4364 }, { "epoch": 1.49, "learning_rate": 8.235117662116942e-06, "loss": 0.2101, "step": 4365 }, { "epoch": 1.49, "learning_rate": 8.22476614838737e-06, "loss": 0.2067, "step": 4366 }, { "epoch": 1.49, "learning_rate": 8.214419863607758e-06, "loss": 0.178, "step": 4367 }, { "epoch": 1.49, "learning_rate": 8.20407881100313e-06, "loss": 0.2185, "step": 4368 }, { "epoch": 1.49, "learning_rate": 8.19374299379686e-06, "loss": 0.2375, "step": 4369 }, { "epoch": 1.49, "learning_rate": 8.183412415210675e-06, "loss": 0.2196, "step": 4370 }, { "epoch": 1.49, "learning_rate": 8.1730870784647e-06, "loss": 0.2081, "step": 4371 }, { "epoch": 1.49, "learning_rate": 8.162766986777396e-06, "loss": 0.2331, "step": 4372 }, { "epoch": 1.49, "learning_rate": 8.152452143365607e-06, "loss": 0.1825, "step": 4373 }, { "epoch": 1.49, "learning_rate": 8.142142551444543e-06, "loss": 0.2025, "step": 4374 }, { "epoch": 1.49, "learning_rate": 8.131838214227763e-06, "loss": 0.2188, "step": 4375 }, { "epoch": 1.49, "learning_rate": 8.12153913492721e-06, "loss": 0.2125, "step": 4376 }, { "epoch": 1.49, "learning_rate": 8.11124531675316e-06, "loss": 0.2244, "step": 4377 }, { "epoch": 1.49, "learning_rate": 8.100956762914275e-06, "loss": 0.2103, "step": 4378 }, { "epoch": 1.49, "learning_rate": 8.09067347661755e-06, "loss": 0.2032, "step": 4379 }, { "epoch": 1.49, "learning_rate": 8.08039546106836e-06, "loss": 0.199, "step": 4380 }, { "epoch": 1.49, "learning_rate": 8.070122719470439e-06, "loss": 0.2091, "step": 4381 }, { "epoch": 1.5, "learning_rate": 8.059855255025848e-06, "loss": 0.2231, "step": 4382 }, { "epoch": 1.5, "learning_rate": 8.049593070935044e-06, "loss": 0.2242, "step": 4383 }, { "epoch": 1.5, "learning_rate": 8.039336170396796e-06, "loss": 0.2162, "step": 4384 }, { "epoch": 1.5, "learning_rate": 8.029084556608266e-06, "loss": 0.2277, "step": 4385 }, { "epoch": 1.5, "learning_rate": 8.018838232764924e-06, "loss": 0.2022, "step": 4386 }, { "epoch": 1.5, "learning_rate": 8.00859720206065e-06, "loss": 0.2159, "step": 4387 }, { "epoch": 1.5, "learning_rate": 7.998361467687612e-06, "loss": 0.1998, "step": 4388 }, { "epoch": 1.5, "learning_rate": 7.988131032836366e-06, "loss": 0.2328, "step": 4389 }, { "epoch": 1.5, "learning_rate": 7.977905900695814e-06, "loss": 0.1936, "step": 4390 }, { "epoch": 1.5, "learning_rate": 7.96768607445318e-06, "loss": 0.1904, "step": 4391 }, { "epoch": 1.5, "learning_rate": 7.957471557294064e-06, "loss": 0.1927, "step": 4392 }, { "epoch": 1.5, "learning_rate": 7.947262352402385e-06, "loss": 0.216, "step": 4393 }, { "epoch": 1.5, "learning_rate": 7.937058462960425e-06, "loss": 0.1965, "step": 4394 }, { "epoch": 1.5, "learning_rate": 7.92685989214881e-06, "loss": 0.188, "step": 4395 }, { "epoch": 1.5, "learning_rate": 7.91666664314649e-06, "loss": 0.2158, "step": 4396 }, { "epoch": 1.5, "learning_rate": 7.90647871913078e-06, "loss": 0.1947, "step": 4397 }, { "epoch": 1.5, "learning_rate": 7.896296123277306e-06, "loss": 0.2007, "step": 4398 }, { "epoch": 1.5, "learning_rate": 7.886118858760056e-06, "loss": 0.2188, "step": 4399 }, { "epoch": 1.5, "learning_rate": 7.875946928751355e-06, "loss": 0.2259, "step": 4400 }, { "epoch": 1.5, "learning_rate": 7.865780336421855e-06, "loss": 0.2101, "step": 4401 }, { "epoch": 1.5, "learning_rate": 7.855619084940557e-06, "loss": 0.2016, "step": 4402 }, { "epoch": 1.5, "learning_rate": 7.84546317747478e-06, "loss": 0.1818, "step": 4403 }, { "epoch": 1.5, "learning_rate": 7.835312617190197e-06, "loss": 0.2236, "step": 4404 }, { "epoch": 1.5, "learning_rate": 7.825167407250789e-06, "loss": 0.2048, "step": 4405 }, { "epoch": 1.5, "learning_rate": 7.815027550818893e-06, "loss": 0.2274, "step": 4406 }, { "epoch": 1.5, "learning_rate": 7.804893051055176e-06, "loss": 0.1923, "step": 4407 }, { "epoch": 1.5, "learning_rate": 7.794763911118614e-06, "loss": 0.2041, "step": 4408 }, { "epoch": 1.5, "learning_rate": 7.784640134166537e-06, "loss": 0.1982, "step": 4409 }, { "epoch": 1.5, "learning_rate": 7.77452172335458e-06, "loss": 0.2153, "step": 4410 }, { "epoch": 1.5, "learning_rate": 7.764408681836732e-06, "loss": 0.2455, "step": 4411 }, { "epoch": 1.51, "learning_rate": 7.75430101276528e-06, "loss": 0.2105, "step": 4412 }, { "epoch": 1.51, "learning_rate": 7.74419871929086e-06, "loss": 0.2258, "step": 4413 }, { "epoch": 1.51, "learning_rate": 7.73410180456242e-06, "loss": 0.1856, "step": 4414 }, { "epoch": 1.51, "learning_rate": 7.724010271727233e-06, "loss": 0.2298, "step": 4415 }, { "epoch": 1.51, "learning_rate": 7.713924123930905e-06, "loss": 0.2076, "step": 4416 }, { "epoch": 1.51, "learning_rate": 7.70384336431734e-06, "loss": 0.2201, "step": 4417 }, { "epoch": 1.51, "learning_rate": 7.693767996028791e-06, "loss": 0.1724, "step": 4418 }, { "epoch": 1.51, "learning_rate": 7.6836980222058e-06, "loss": 0.2161, "step": 4419 }, { "epoch": 1.51, "learning_rate": 7.673633445987252e-06, "loss": 0.204, "step": 4420 }, { "epoch": 1.51, "learning_rate": 7.66357427051035e-06, "loss": 0.2247, "step": 4421 }, { "epoch": 1.51, "learning_rate": 7.653520498910588e-06, "loss": 0.2022, "step": 4422 }, { "epoch": 1.51, "learning_rate": 7.643472134321808e-06, "loss": 0.2348, "step": 4423 }, { "epoch": 1.51, "learning_rate": 7.633429179876137e-06, "loss": 0.2288, "step": 4424 }, { "epoch": 1.51, "learning_rate": 7.6233916387040335e-06, "loss": 0.2327, "step": 4425 }, { "epoch": 1.51, "learning_rate": 7.613359513934274e-06, "loss": 0.2101, "step": 4426 }, { "epoch": 1.51, "learning_rate": 7.603332808693922e-06, "loss": 0.1969, "step": 4427 }, { "epoch": 1.51, "learning_rate": 7.5933115261083765e-06, "loss": 0.1912, "step": 4428 }, { "epoch": 1.51, "learning_rate": 7.5832956693013355e-06, "loss": 0.1782, "step": 4429 }, { "epoch": 1.51, "learning_rate": 7.573285241394815e-06, "loss": 0.2215, "step": 4430 }, { "epoch": 1.51, "learning_rate": 7.563280245509114e-06, "loss": 0.2069, "step": 4431 }, { "epoch": 1.51, "learning_rate": 7.5532806847628616e-06, "loss": 0.1971, "step": 4432 }, { "epoch": 1.51, "learning_rate": 7.543286562272997e-06, "loss": 0.2196, "step": 4433 }, { "epoch": 1.51, "learning_rate": 7.533297881154738e-06, "loss": 0.2356, "step": 4434 }, { "epoch": 1.51, "learning_rate": 7.5233146445216366e-06, "loss": 0.1617, "step": 4435 }, { "epoch": 1.51, "learning_rate": 7.513336855485517e-06, "loss": 0.2172, "step": 4436 }, { "epoch": 1.51, "learning_rate": 7.503364517156536e-06, "loss": 0.2122, "step": 4437 }, { "epoch": 1.51, "learning_rate": 7.493397632643123e-06, "loss": 0.1836, "step": 4438 }, { "epoch": 1.51, "learning_rate": 7.4834362050520275e-06, "loss": 0.2115, "step": 4439 }, { "epoch": 1.51, "learning_rate": 7.4734802374882974e-06, "loss": 0.2056, "step": 4440 }, { "epoch": 1.52, "learning_rate": 7.46352973305526e-06, "loss": 0.1927, "step": 4441 }, { "epoch": 1.52, "learning_rate": 7.4535846948545595e-06, "loss": 0.2057, "step": 4442 }, { "epoch": 1.52, "learning_rate": 7.443645125986132e-06, "loss": 0.2267, "step": 4443 }, { "epoch": 1.52, "learning_rate": 7.433711029548207e-06, "loss": 0.2022, "step": 4444 }, { "epoch": 1.52, "learning_rate": 7.423782408637298e-06, "loss": 0.1978, "step": 4445 }, { "epoch": 1.52, "learning_rate": 7.413859266348228e-06, "loss": 0.2057, "step": 4446 }, { "epoch": 1.52, "learning_rate": 7.403941605774109e-06, "loss": 0.1967, "step": 4447 }, { "epoch": 1.52, "learning_rate": 7.39402943000633e-06, "loss": 0.2091, "step": 4448 }, { "epoch": 1.52, "learning_rate": 7.384122742134594e-06, "loss": 0.1968, "step": 4449 }, { "epoch": 1.52, "learning_rate": 7.3742215452468645e-06, "loss": 0.2333, "step": 4450 }, { "epoch": 1.52, "learning_rate": 7.364325842429417e-06, "loss": 0.1713, "step": 4451 }, { "epoch": 1.52, "learning_rate": 7.354435636766815e-06, "loss": 0.207, "step": 4452 }, { "epoch": 1.52, "learning_rate": 7.3445509313418845e-06, "loss": 0.1795, "step": 4453 }, { "epoch": 1.52, "learning_rate": 7.3346717292357704e-06, "loss": 0.2385, "step": 4454 }, { "epoch": 1.52, "learning_rate": 7.324798033527866e-06, "loss": 0.1988, "step": 4455 }, { "epoch": 1.52, "learning_rate": 7.314929847295873e-06, "loss": 0.2357, "step": 4456 }, { "epoch": 1.52, "learning_rate": 7.305067173615776e-06, "loss": 0.1998, "step": 4457 }, { "epoch": 1.52, "learning_rate": 7.295210015561829e-06, "loss": 0.1982, "step": 4458 }, { "epoch": 1.52, "learning_rate": 7.285358376206583e-06, "loss": 0.2165, "step": 4459 }, { "epoch": 1.52, "learning_rate": 7.275512258620842e-06, "loss": 0.2134, "step": 4460 }, { "epoch": 1.52, "learning_rate": 7.265671665873719e-06, "loss": 0.1991, "step": 4461 }, { "epoch": 1.52, "learning_rate": 7.255836601032579e-06, "loss": 0.2046, "step": 4462 }, { "epoch": 1.52, "learning_rate": 7.246007067163091e-06, "loss": 0.2267, "step": 4463 }, { "epoch": 1.52, "learning_rate": 7.236183067329166e-06, "loss": 0.2057, "step": 4464 }, { "epoch": 1.52, "learning_rate": 7.2263646045930195e-06, "loss": 0.2323, "step": 4465 }, { "epoch": 1.52, "learning_rate": 7.216551682015138e-06, "loss": 0.2055, "step": 4466 }, { "epoch": 1.52, "learning_rate": 7.206744302654256e-06, "loss": 0.2052, "step": 4467 }, { "epoch": 1.52, "learning_rate": 7.196942469567417e-06, "loss": 0.2093, "step": 4468 }, { "epoch": 1.52, "learning_rate": 7.187146185809895e-06, "loss": 0.266, "step": 4469 }, { "epoch": 1.53, "learning_rate": 7.177355454435272e-06, "loss": 0.2042, "step": 4470 }, { "epoch": 1.53, "learning_rate": 7.167570278495375e-06, "loss": 0.1891, "step": 4471 }, { "epoch": 1.53, "learning_rate": 7.157790661040311e-06, "loss": 0.2312, "step": 4472 }, { "epoch": 1.53, "learning_rate": 7.1480166051184575e-06, "loss": 0.202, "step": 4473 }, { "epoch": 1.53, "learning_rate": 7.138248113776435e-06, "loss": 0.2314, "step": 4474 }, { "epoch": 1.53, "learning_rate": 7.128485190059167e-06, "loss": 0.2382, "step": 4475 }, { "epoch": 1.53, "learning_rate": 7.118727837009798e-06, "loss": 0.2031, "step": 4476 }, { "epoch": 1.53, "learning_rate": 7.108976057669769e-06, "loss": 0.2109, "step": 4477 }, { "epoch": 1.53, "learning_rate": 7.099229855078782e-06, "loss": 0.1945, "step": 4478 }, { "epoch": 1.53, "learning_rate": 7.089489232274776e-06, "loss": 0.1873, "step": 4479 }, { "epoch": 1.53, "learning_rate": 7.079754192293983e-06, "loss": 0.2289, "step": 4480 }, { "epoch": 1.53, "learning_rate": 7.070024738170861e-06, "loss": 0.2285, "step": 4481 }, { "epoch": 1.53, "learning_rate": 7.060300872938156e-06, "loss": 0.2115, "step": 4482 }, { "epoch": 1.53, "learning_rate": 7.0505825996268535e-06, "loss": 0.2261, "step": 4483 }, { "epoch": 1.53, "learning_rate": 7.040869921266211e-06, "loss": 0.2343, "step": 4484 }, { "epoch": 1.53, "learning_rate": 7.0311628408837345e-06, "loss": 0.2566, "step": 4485 }, { "epoch": 1.53, "learning_rate": 7.021461361505174e-06, "loss": 0.2008, "step": 4486 }, { "epoch": 1.53, "learning_rate": 7.011765486154556e-06, "loss": 0.2014, "step": 4487 }, { "epoch": 1.53, "learning_rate": 7.002075217854134e-06, "loss": 0.1825, "step": 4488 }, { "epoch": 1.53, "learning_rate": 6.992390559624446e-06, "loss": 0.2139, "step": 4489 }, { "epoch": 1.53, "learning_rate": 6.982711514484241e-06, "loss": 0.21, "step": 4490 }, { "epoch": 1.53, "learning_rate": 6.973038085450556e-06, "loss": 0.232, "step": 4491 }, { "epoch": 1.53, "learning_rate": 6.963370275538666e-06, "loss": 0.2157, "step": 4492 }, { "epoch": 1.53, "learning_rate": 6.953708087762076e-06, "loss": 0.2072, "step": 4493 }, { "epoch": 1.53, "learning_rate": 6.944051525132567e-06, "loss": 0.231, "step": 4494 }, { "epoch": 1.53, "learning_rate": 6.934400590660137e-06, "loss": 0.2105, "step": 4495 }, { "epoch": 1.53, "learning_rate": 6.924755287353055e-06, "loss": 0.2445, "step": 4496 }, { "epoch": 1.53, "learning_rate": 6.915115618217824e-06, "loss": 0.2173, "step": 4497 }, { "epoch": 1.53, "learning_rate": 6.905481586259194e-06, "loss": 0.202, "step": 4498 }, { "epoch": 1.53, "learning_rate": 6.895853194480159e-06, "loss": 0.1948, "step": 4499 }, { "epoch": 1.54, "learning_rate": 6.886230445881939e-06, "loss": 0.2257, "step": 4500 }, { "epoch": 1.54, "learning_rate": 6.876613343464023e-06, "loss": 0.1916, "step": 4501 }, { "epoch": 1.54, "learning_rate": 6.867001890224109e-06, "loss": 0.2089, "step": 4502 }, { "epoch": 1.54, "learning_rate": 6.857396089158158e-06, "loss": 0.2115, "step": 4503 }, { "epoch": 1.54, "learning_rate": 6.847795943260365e-06, "loss": 0.1848, "step": 4504 }, { "epoch": 1.54, "learning_rate": 6.8382014555231495e-06, "loss": 0.1977, "step": 4505 }, { "epoch": 1.54, "learning_rate": 6.828612628937187e-06, "loss": 0.2047, "step": 4506 }, { "epoch": 1.54, "learning_rate": 6.8190294664913615e-06, "loss": 0.2257, "step": 4507 }, { "epoch": 1.54, "learning_rate": 6.809451971172817e-06, "loss": 0.2067, "step": 4508 }, { "epoch": 1.54, "learning_rate": 6.799880145966927e-06, "loss": 0.2119, "step": 4509 }, { "epoch": 1.54, "learning_rate": 6.790313993857281e-06, "loss": 0.2238, "step": 4510 }, { "epoch": 1.54, "learning_rate": 6.780753517825711e-06, "loss": 0.2413, "step": 4511 }, { "epoch": 1.54, "learning_rate": 6.771198720852287e-06, "loss": 0.211, "step": 4512 }, { "epoch": 1.54, "learning_rate": 6.761649605915307e-06, "loss": 0.2097, "step": 4513 }, { "epoch": 1.54, "learning_rate": 6.752106175991277e-06, "loss": 0.1938, "step": 4514 }, { "epoch": 1.54, "learning_rate": 6.742568434054961e-06, "loss": 0.1526, "step": 4515 }, { "epoch": 1.54, "learning_rate": 6.733036383079322e-06, "loss": 0.2444, "step": 4516 }, { "epoch": 1.54, "learning_rate": 6.723510026035568e-06, "loss": 0.1931, "step": 4517 }, { "epoch": 1.54, "learning_rate": 6.7139893658931334e-06, "loss": 0.2001, "step": 4518 }, { "epoch": 1.54, "learning_rate": 6.70447440561966e-06, "loss": 0.2274, "step": 4519 }, { "epoch": 1.54, "learning_rate": 6.694965148181034e-06, "loss": 0.1891, "step": 4520 }, { "epoch": 1.54, "learning_rate": 6.685461596541339e-06, "loss": 0.2177, "step": 4521 }, { "epoch": 1.54, "learning_rate": 6.6759637536629035e-06, "loss": 0.2241, "step": 4522 }, { "epoch": 1.54, "learning_rate": 6.666471622506271e-06, "loss": 0.2167, "step": 4523 }, { "epoch": 1.54, "learning_rate": 6.6569852060301864e-06, "loss": 0.2501, "step": 4524 }, { "epoch": 1.54, "learning_rate": 6.64750450719164e-06, "loss": 0.1978, "step": 4525 }, { "epoch": 1.54, "learning_rate": 6.638029528945822e-06, "loss": 0.2104, "step": 4526 }, { "epoch": 1.54, "learning_rate": 6.628560274246154e-06, "loss": 0.2085, "step": 4527 }, { "epoch": 1.54, "learning_rate": 6.619096746044251e-06, "loss": 0.1965, "step": 4528 }, { "epoch": 1.55, "learning_rate": 6.609638947289964e-06, "loss": 0.1838, "step": 4529 }, { "epoch": 1.55, "learning_rate": 6.600186880931358e-06, "loss": 0.2141, "step": 4530 }, { "epoch": 1.55, "learning_rate": 6.590740549914687e-06, "loss": 0.1948, "step": 4531 }, { "epoch": 1.55, "learning_rate": 6.5812999571844505e-06, "loss": 0.2071, "step": 4532 }, { "epoch": 1.55, "learning_rate": 6.571865105683325e-06, "loss": 0.2174, "step": 4533 }, { "epoch": 1.55, "learning_rate": 6.562435998352229e-06, "loss": 0.2361, "step": 4534 }, { "epoch": 1.55, "learning_rate": 6.553012638130279e-06, "loss": 0.202, "step": 4535 }, { "epoch": 1.55, "learning_rate": 6.543595027954785e-06, "loss": 0.1965, "step": 4536 }, { "epoch": 1.55, "learning_rate": 6.5341831707612935e-06, "loss": 0.2034, "step": 4537 }, { "epoch": 1.55, "learning_rate": 6.524777069483526e-06, "loss": 0.2015, "step": 4538 }, { "epoch": 1.55, "learning_rate": 6.515376727053432e-06, "loss": 0.2032, "step": 4539 }, { "epoch": 1.55, "learning_rate": 6.505982146401163e-06, "loss": 0.1871, "step": 4540 }, { "epoch": 1.55, "learning_rate": 6.496593330455073e-06, "loss": 0.201, "step": 4541 }, { "epoch": 1.55, "learning_rate": 6.487210282141706e-06, "loss": 0.1832, "step": 4542 }, { "epoch": 1.55, "learning_rate": 6.477833004385828e-06, "loss": 0.2056, "step": 4543 }, { "epoch": 1.55, "learning_rate": 6.468461500110404e-06, "loss": 0.1989, "step": 4544 }, { "epoch": 1.55, "learning_rate": 6.4590957722365755e-06, "loss": 0.2092, "step": 4545 }, { "epoch": 1.55, "learning_rate": 6.449735823683723e-06, "loss": 0.1931, "step": 4546 }, { "epoch": 1.55, "learning_rate": 6.440381657369379e-06, "loss": 0.1967, "step": 4547 }, { "epoch": 1.55, "learning_rate": 6.431033276209314e-06, "loss": 0.1791, "step": 4548 }, { "epoch": 1.55, "learning_rate": 6.421690683117482e-06, "loss": 0.2163, "step": 4549 }, { "epoch": 1.55, "learning_rate": 6.412353881006017e-06, "loss": 0.2085, "step": 4550 }, { "epoch": 1.55, "learning_rate": 6.403022872785278e-06, "loss": 0.2188, "step": 4551 }, { "epoch": 1.55, "learning_rate": 6.393697661363781e-06, "loss": 0.1812, "step": 4552 }, { "epoch": 1.55, "learning_rate": 6.38437824964827e-06, "loss": 0.203, "step": 4553 }, { "epoch": 1.55, "learning_rate": 6.375064640543663e-06, "loss": 0.2094, "step": 4554 }, { "epoch": 1.55, "learning_rate": 6.365756836953071e-06, "loss": 0.2048, "step": 4555 }, { "epoch": 1.55, "learning_rate": 6.356454841777807e-06, "loss": 0.2258, "step": 4556 }, { "epoch": 1.55, "learning_rate": 6.347158657917346e-06, "loss": 0.1728, "step": 4557 }, { "epoch": 1.56, "learning_rate": 6.337868288269389e-06, "loss": 0.1935, "step": 4558 }, { "epoch": 1.56, "learning_rate": 6.328583735729785e-06, "loss": 0.2145, "step": 4559 }, { "epoch": 1.56, "learning_rate": 6.3193050031925995e-06, "loss": 0.2175, "step": 4560 }, { "epoch": 1.56, "learning_rate": 6.310032093550081e-06, "loss": 0.195, "step": 4561 }, { "epoch": 1.56, "learning_rate": 6.300765009692641e-06, "loss": 0.1903, "step": 4562 }, { "epoch": 1.56, "learning_rate": 6.291503754508906e-06, "loss": 0.1768, "step": 4563 }, { "epoch": 1.56, "learning_rate": 6.282248330885656e-06, "loss": 0.209, "step": 4564 }, { "epoch": 1.56, "learning_rate": 6.272998741707869e-06, "loss": 0.2177, "step": 4565 }, { "epoch": 1.56, "learning_rate": 6.263754989858717e-06, "loss": 0.1752, "step": 4566 }, { "epoch": 1.56, "learning_rate": 6.254517078219518e-06, "loss": 0.1914, "step": 4567 }, { "epoch": 1.56, "learning_rate": 6.245285009669799e-06, "loss": 0.2303, "step": 4568 }, { "epoch": 1.56, "learning_rate": 6.236058787087254e-06, "loss": 0.2626, "step": 4569 }, { "epoch": 1.56, "learning_rate": 6.226838413347766e-06, "loss": 0.2153, "step": 4570 }, { "epoch": 1.56, "learning_rate": 6.217623891325369e-06, "loss": 0.1811, "step": 4571 }, { "epoch": 1.56, "learning_rate": 6.208415223892308e-06, "loss": 0.1781, "step": 4572 }, { "epoch": 1.56, "learning_rate": 6.199212413918967e-06, "loss": 0.203, "step": 4573 }, { "epoch": 1.56, "learning_rate": 6.190015464273932e-06, "loss": 0.1934, "step": 4574 }, { "epoch": 1.56, "learning_rate": 6.180824377823957e-06, "loss": 0.1917, "step": 4575 }, { "epoch": 1.56, "learning_rate": 6.17163915743395e-06, "loss": 0.211, "step": 4576 }, { "epoch": 1.56, "learning_rate": 6.162459805967019e-06, "loss": 0.2391, "step": 4577 }, { "epoch": 1.56, "learning_rate": 6.153286326284413e-06, "loss": 0.2045, "step": 4578 }, { "epoch": 1.56, "learning_rate": 6.144118721245573e-06, "loss": 0.2171, "step": 4579 }, { "epoch": 1.56, "learning_rate": 6.134956993708099e-06, "loss": 0.1989, "step": 4580 }, { "epoch": 1.56, "learning_rate": 6.1258011465277655e-06, "loss": 0.1926, "step": 4581 }, { "epoch": 1.56, "learning_rate": 6.116651182558516e-06, "loss": 0.2158, "step": 4582 }, { "epoch": 1.56, "learning_rate": 6.107507104652438e-06, "loss": 0.182, "step": 4583 }, { "epoch": 1.56, "learning_rate": 6.0983689156598166e-06, "loss": 0.1958, "step": 4584 }, { "epoch": 1.56, "learning_rate": 6.089236618429068e-06, "loss": 0.3798, "step": 4585 }, { "epoch": 1.56, "learning_rate": 6.080110215806797e-06, "loss": 0.2264, "step": 4586 }, { "epoch": 1.56, "learning_rate": 6.070989710637773e-06, "loss": 0.1877, "step": 4587 }, { "epoch": 1.57, "learning_rate": 6.0618751057649e-06, "loss": 0.2178, "step": 4588 }, { "epoch": 1.57, "learning_rate": 6.052766404029272e-06, "loss": 0.212, "step": 4589 }, { "epoch": 1.57, "learning_rate": 6.043663608270123e-06, "loss": 0.2058, "step": 4590 }, { "epoch": 1.57, "learning_rate": 6.0345667213248554e-06, "loss": 0.2186, "step": 4591 }, { "epoch": 1.57, "learning_rate": 6.0254757460290355e-06, "loss": 0.213, "step": 4592 }, { "epoch": 1.57, "learning_rate": 6.016390685216369e-06, "loss": 0.2129, "step": 4593 }, { "epoch": 1.57, "learning_rate": 6.007311541718732e-06, "loss": 0.1823, "step": 4594 }, { "epoch": 1.57, "learning_rate": 5.998238318366159e-06, "loss": 0.2109, "step": 4595 }, { "epoch": 1.57, "learning_rate": 5.989171017986834e-06, "loss": 0.1848, "step": 4596 }, { "epoch": 1.57, "learning_rate": 5.980109643407078e-06, "loss": 0.2089, "step": 4597 }, { "epoch": 1.57, "learning_rate": 5.971054197451401e-06, "loss": 0.1943, "step": 4598 }, { "epoch": 1.57, "learning_rate": 5.962004682942429e-06, "loss": 0.1901, "step": 4599 }, { "epoch": 1.57, "learning_rate": 5.9529611027009606e-06, "loss": 0.2236, "step": 4600 }, { "epoch": 1.57, "learning_rate": 5.9439234595459445e-06, "loss": 0.2116, "step": 4601 }, { "epoch": 1.57, "learning_rate": 5.934891756294458e-06, "loss": 0.2213, "step": 4602 }, { "epoch": 1.57, "learning_rate": 5.925865995761762e-06, "loss": 0.2105, "step": 4603 }, { "epoch": 1.57, "learning_rate": 5.916846180761229e-06, "loss": 0.2324, "step": 4604 }, { "epoch": 1.57, "learning_rate": 5.907832314104397e-06, "loss": 0.2132, "step": 4605 }, { "epoch": 1.57, "learning_rate": 5.898824398600958e-06, "loss": 0.2025, "step": 4606 }, { "epoch": 1.57, "learning_rate": 5.8898224370587204e-06, "loss": 0.2008, "step": 4607 }, { "epoch": 1.57, "learning_rate": 5.880826432283668e-06, "loss": 0.1974, "step": 4608 }, { "epoch": 1.57, "learning_rate": 5.871836387079907e-06, "loss": 0.2128, "step": 4609 }, { "epoch": 1.57, "learning_rate": 5.862852304249705e-06, "loss": 0.2054, "step": 4610 }, { "epoch": 1.57, "learning_rate": 5.853874186593444e-06, "loss": 0.1821, "step": 4611 }, { "epoch": 1.57, "learning_rate": 5.844902036909667e-06, "loss": 0.3183, "step": 4612 }, { "epoch": 1.57, "learning_rate": 5.8359358579950615e-06, "loss": 0.1832, "step": 4613 }, { "epoch": 1.57, "learning_rate": 5.826975652644428e-06, "loss": 0.1743, "step": 4614 }, { "epoch": 1.57, "learning_rate": 5.818021423650738e-06, "loss": 0.1994, "step": 4615 }, { "epoch": 1.57, "learning_rate": 5.809073173805066e-06, "loss": 0.184, "step": 4616 }, { "epoch": 1.58, "learning_rate": 5.80013090589665e-06, "loss": 0.1929, "step": 4617 }, { "epoch": 1.58, "learning_rate": 5.791194622712856e-06, "loss": 0.2053, "step": 4618 }, { "epoch": 1.58, "learning_rate": 5.782264327039169e-06, "loss": 0.1901, "step": 4619 }, { "epoch": 1.58, "learning_rate": 5.773340021659238e-06, "loss": 0.1933, "step": 4620 }, { "epoch": 1.58, "learning_rate": 5.764421709354811e-06, "loss": 0.1946, "step": 4621 }, { "epoch": 1.58, "learning_rate": 5.755509392905792e-06, "loss": 0.2215, "step": 4622 }, { "epoch": 1.58, "learning_rate": 5.7466030750902035e-06, "loss": 0.1981, "step": 4623 }, { "epoch": 1.58, "learning_rate": 5.737702758684218e-06, "loss": 0.2069, "step": 4624 }, { "epoch": 1.58, "learning_rate": 5.728808446462103e-06, "loss": 0.181, "step": 4625 }, { "epoch": 1.58, "learning_rate": 5.719920141196281e-06, "loss": 0.2071, "step": 4626 }, { "epoch": 1.58, "learning_rate": 5.711037845657302e-06, "loss": 0.231, "step": 4627 }, { "epoch": 1.58, "learning_rate": 5.702161562613826e-06, "loss": 0.1995, "step": 4628 }, { "epoch": 1.58, "learning_rate": 5.693291294832656e-06, "loss": 0.1849, "step": 4629 }, { "epoch": 1.58, "learning_rate": 5.6844270450786994e-06, "loss": 0.1891, "step": 4630 }, { "epoch": 1.58, "learning_rate": 5.6755688161150126e-06, "loss": 0.2152, "step": 4631 }, { "epoch": 1.58, "learning_rate": 5.666716610702766e-06, "loss": 0.1831, "step": 4632 }, { "epoch": 1.58, "learning_rate": 5.657870431601236e-06, "loss": 0.2391, "step": 4633 }, { "epoch": 1.58, "learning_rate": 5.649030281567849e-06, "loss": 0.2374, "step": 4634 }, { "epoch": 1.58, "learning_rate": 5.640196163358128e-06, "loss": 0.2343, "step": 4635 }, { "epoch": 1.58, "learning_rate": 5.631368079725724e-06, "loss": 0.2085, "step": 4636 }, { "epoch": 1.58, "learning_rate": 5.622546033422416e-06, "loss": 0.2139, "step": 4637 }, { "epoch": 1.58, "learning_rate": 5.613730027198089e-06, "loss": 0.2235, "step": 4638 }, { "epoch": 1.58, "learning_rate": 5.604920063800756e-06, "loss": 0.1787, "step": 4639 }, { "epoch": 1.58, "learning_rate": 5.596116145976529e-06, "loss": 0.1991, "step": 4640 }, { "epoch": 1.58, "learning_rate": 5.587318276469661e-06, "loss": 0.2136, "step": 4641 }, { "epoch": 1.58, "learning_rate": 5.57852645802249e-06, "loss": 0.2057, "step": 4642 }, { "epoch": 1.58, "learning_rate": 5.569740693375488e-06, "loss": 0.1877, "step": 4643 }, { "epoch": 1.58, "learning_rate": 5.56096098526725e-06, "loss": 0.2146, "step": 4644 }, { "epoch": 1.58, "learning_rate": 5.552187336434445e-06, "loss": 0.1815, "step": 4645 }, { "epoch": 1.59, "learning_rate": 5.543419749611895e-06, "loss": 0.2239, "step": 4646 }, { "epoch": 1.59, "learning_rate": 5.534658227532499e-06, "loss": 0.1758, "step": 4647 }, { "epoch": 1.59, "learning_rate": 5.525902772927296e-06, "loss": 0.2243, "step": 4648 }, { "epoch": 1.59, "learning_rate": 5.517153388525404e-06, "loss": 0.2233, "step": 4649 }, { "epoch": 1.59, "learning_rate": 5.508410077054068e-06, "loss": 0.2336, "step": 4650 }, { "epoch": 1.59, "learning_rate": 5.49967284123864e-06, "loss": 0.1994, "step": 4651 }, { "epoch": 1.59, "learning_rate": 5.490941683802572e-06, "loss": 0.1945, "step": 4652 }, { "epoch": 1.59, "learning_rate": 5.482216607467427e-06, "loss": 0.2182, "step": 4653 }, { "epoch": 1.59, "learning_rate": 5.473497614952855e-06, "loss": 0.2172, "step": 4654 }, { "epoch": 1.59, "learning_rate": 5.464784708976639e-06, "loss": 0.2008, "step": 4655 }, { "epoch": 1.59, "learning_rate": 5.4560778922546325e-06, "loss": 0.2215, "step": 4656 }, { "epoch": 1.59, "learning_rate": 5.447377167500817e-06, "loss": 0.2069, "step": 4657 }, { "epoch": 1.59, "learning_rate": 5.4386825374272694e-06, "loss": 0.2166, "step": 4658 }, { "epoch": 1.59, "learning_rate": 5.4299940047441485e-06, "loss": 0.1909, "step": 4659 }, { "epoch": 1.59, "learning_rate": 5.421311572159743e-06, "loss": 0.2216, "step": 4660 }, { "epoch": 1.59, "learning_rate": 5.4126352423804095e-06, "loss": 0.1769, "step": 4661 }, { "epoch": 1.59, "learning_rate": 5.403965018110621e-06, "loss": 0.22, "step": 4662 }, { "epoch": 1.59, "learning_rate": 5.395300902052955e-06, "loss": 0.2138, "step": 4663 }, { "epoch": 1.59, "learning_rate": 5.386642896908048e-06, "loss": 0.2437, "step": 4664 }, { "epoch": 1.59, "learning_rate": 5.377991005374683e-06, "loss": 0.1924, "step": 4665 }, { "epoch": 1.59, "learning_rate": 5.369345230149697e-06, "loss": 0.2169, "step": 4666 }, { "epoch": 1.59, "learning_rate": 5.360705573928043e-06, "loss": 0.2021, "step": 4667 }, { "epoch": 1.59, "learning_rate": 5.3520720394027446e-06, "loss": 0.2276, "step": 4668 }, { "epoch": 1.59, "learning_rate": 5.34344462926494e-06, "loss": 0.222, "step": 4669 }, { "epoch": 1.59, "learning_rate": 5.3348233462038545e-06, "loss": 0.203, "step": 4670 }, { "epoch": 1.59, "learning_rate": 5.326208192906787e-06, "loss": 0.1946, "step": 4671 }, { "epoch": 1.59, "learning_rate": 5.317599172059148e-06, "loss": 0.198, "step": 4672 }, { "epoch": 1.59, "learning_rate": 5.308996286344414e-06, "loss": 0.1733, "step": 4673 }, { "epoch": 1.59, "learning_rate": 5.300399538444173e-06, "loss": 0.1952, "step": 4674 }, { "epoch": 1.6, "learning_rate": 5.291808931038075e-06, "loss": 0.1971, "step": 4675 }, { "epoch": 1.6, "learning_rate": 5.283224466803874e-06, "loss": 0.1831, "step": 4676 }, { "epoch": 1.6, "learning_rate": 5.2746461484174115e-06, "loss": 0.2306, "step": 4677 }, { "epoch": 1.6, "learning_rate": 5.266073978552583e-06, "loss": 0.1927, "step": 4678 }, { "epoch": 1.6, "learning_rate": 5.257507959881419e-06, "loss": 0.219, "step": 4679 }, { "epoch": 1.6, "learning_rate": 5.24894809507398e-06, "loss": 0.2079, "step": 4680 }, { "epoch": 1.6, "learning_rate": 5.24039438679845e-06, "loss": 0.178, "step": 4681 }, { "epoch": 1.6, "learning_rate": 5.2318468377210605e-06, "loss": 0.2212, "step": 4682 }, { "epoch": 1.6, "learning_rate": 5.223305450506144e-06, "loss": 0.2071, "step": 4683 }, { "epoch": 1.6, "learning_rate": 5.2147702278161155e-06, "loss": 0.2021, "step": 4684 }, { "epoch": 1.6, "learning_rate": 5.206241172311443e-06, "loss": 0.2162, "step": 4685 }, { "epoch": 1.6, "learning_rate": 5.197718286650705e-06, "loss": 0.1972, "step": 4686 }, { "epoch": 1.6, "learning_rate": 5.189201573490526e-06, "loss": 0.2363, "step": 4687 }, { "epoch": 1.6, "learning_rate": 5.180691035485629e-06, "loss": 0.2063, "step": 4688 }, { "epoch": 1.6, "learning_rate": 5.17218667528881e-06, "loss": 0.2091, "step": 4689 }, { "epoch": 1.6, "learning_rate": 5.163688495550917e-06, "loss": 0.1901, "step": 4690 }, { "epoch": 1.6, "learning_rate": 5.155196498920903e-06, "loss": 0.2253, "step": 4691 }, { "epoch": 1.6, "learning_rate": 5.1467106880457694e-06, "loss": 0.1782, "step": 4692 }, { "epoch": 1.6, "learning_rate": 5.1382310655706145e-06, "loss": 0.2091, "step": 4693 }, { "epoch": 1.6, "learning_rate": 5.129757634138571e-06, "loss": 0.2389, "step": 4694 }, { "epoch": 1.6, "learning_rate": 5.121290396390873e-06, "loss": 0.1905, "step": 4695 }, { "epoch": 1.6, "learning_rate": 5.112829354966822e-06, "loss": 0.2176, "step": 4696 }, { "epoch": 1.6, "learning_rate": 5.104374512503765e-06, "loss": 0.202, "step": 4697 }, { "epoch": 1.6, "learning_rate": 5.095925871637144e-06, "loss": 0.2185, "step": 4698 }, { "epoch": 1.6, "learning_rate": 5.087483435000442e-06, "loss": 0.1889, "step": 4699 }, { "epoch": 1.6, "learning_rate": 5.079047205225237e-06, "loss": 0.1836, "step": 4700 }, { "epoch": 1.6, "learning_rate": 5.070617184941142e-06, "loss": 0.2145, "step": 4701 }, { "epoch": 1.6, "learning_rate": 5.0621933767758586e-06, "loss": 0.196, "step": 4702 }, { "epoch": 1.6, "learning_rate": 5.0537757833551445e-06, "loss": 0.1719, "step": 4703 }, { "epoch": 1.6, "learning_rate": 5.04536440730281e-06, "loss": 0.2024, "step": 4704 }, { "epoch": 1.61, "learning_rate": 5.036959251240742e-06, "loss": 0.2154, "step": 4705 }, { "epoch": 1.61, "learning_rate": 5.02856031778888e-06, "loss": 0.2059, "step": 4706 }, { "epoch": 1.61, "learning_rate": 5.020167609565235e-06, "loss": 0.2096, "step": 4707 }, { "epoch": 1.61, "learning_rate": 5.011781129185855e-06, "loss": 0.1884, "step": 4708 }, { "epoch": 1.61, "learning_rate": 5.003400879264866e-06, "loss": 0.2261, "step": 4709 }, { "epoch": 1.61, "learning_rate": 4.9950268624144565e-06, "loss": 0.2378, "step": 4710 }, { "epoch": 1.61, "learning_rate": 4.986659081244849e-06, "loss": 0.2065, "step": 4711 }, { "epoch": 1.61, "learning_rate": 4.978297538364343e-06, "loss": 0.1961, "step": 4712 }, { "epoch": 1.61, "learning_rate": 4.969942236379279e-06, "loss": 0.215, "step": 4713 }, { "epoch": 1.61, "learning_rate": 4.961593177894064e-06, "loss": 0.2911, "step": 4714 }, { "epoch": 1.61, "learning_rate": 4.9532503655111614e-06, "loss": 0.2291, "step": 4715 }, { "epoch": 1.61, "learning_rate": 4.944913801831067e-06, "loss": 0.1852, "step": 4716 }, { "epoch": 1.61, "learning_rate": 4.936583489452351e-06, "loss": 0.1745, "step": 4717 }, { "epoch": 1.61, "learning_rate": 4.928259430971618e-06, "loss": 0.209, "step": 4718 }, { "epoch": 1.61, "learning_rate": 4.919941628983537e-06, "loss": 0.1935, "step": 4719 }, { "epoch": 1.61, "learning_rate": 4.91163008608082e-06, "loss": 0.1882, "step": 4720 }, { "epoch": 1.61, "learning_rate": 4.903324804854231e-06, "loss": 0.1918, "step": 4721 }, { "epoch": 1.61, "learning_rate": 4.895025787892585e-06, "loss": 0.2097, "step": 4722 }, { "epoch": 1.61, "learning_rate": 4.886733037782728e-06, "loss": 0.1972, "step": 4723 }, { "epoch": 1.61, "learning_rate": 4.8784465571095755e-06, "loss": 0.2094, "step": 4724 }, { "epoch": 1.61, "learning_rate": 4.870166348456068e-06, "loss": 0.208, "step": 4725 }, { "epoch": 1.61, "learning_rate": 4.861892414403213e-06, "loss": 0.2078, "step": 4726 }, { "epoch": 1.61, "learning_rate": 4.853624757530035e-06, "loss": 0.2343, "step": 4727 }, { "epoch": 1.61, "learning_rate": 4.84536338041362e-06, "loss": 0.1804, "step": 4728 }, { "epoch": 1.61, "learning_rate": 4.837108285629108e-06, "loss": 0.1944, "step": 4729 }, { "epoch": 1.61, "learning_rate": 4.828859475749645e-06, "loss": 0.1774, "step": 4730 }, { "epoch": 1.61, "learning_rate": 4.820616953346458e-06, "loss": 0.2175, "step": 4731 }, { "epoch": 1.61, "learning_rate": 4.8123807209887765e-06, "loss": 0.1945, "step": 4732 }, { "epoch": 1.61, "learning_rate": 4.804150781243902e-06, "loss": 0.1961, "step": 4733 }, { "epoch": 1.62, "learning_rate": 4.795927136677153e-06, "loss": 0.2028, "step": 4734 }, { "epoch": 1.62, "learning_rate": 4.787709789851894e-06, "loss": 0.1868, "step": 4735 }, { "epoch": 1.62, "learning_rate": 4.779498743329538e-06, "loss": 0.1942, "step": 4736 }, { "epoch": 1.62, "learning_rate": 4.771293999669502e-06, "loss": 0.2212, "step": 4737 }, { "epoch": 1.62, "learning_rate": 4.763095561429276e-06, "loss": 0.1758, "step": 4738 }, { "epoch": 1.62, "learning_rate": 4.7549034311643535e-06, "loss": 0.2266, "step": 4739 }, { "epoch": 1.62, "learning_rate": 4.746717611428278e-06, "loss": 0.1893, "step": 4740 }, { "epoch": 1.62, "learning_rate": 4.73853810477263e-06, "loss": 0.2023, "step": 4741 }, { "epoch": 1.62, "learning_rate": 4.7303649137470066e-06, "loss": 0.1897, "step": 4742 }, { "epoch": 1.62, "learning_rate": 4.7221980408990534e-06, "loss": 0.2233, "step": 4743 }, { "epoch": 1.62, "learning_rate": 4.714037488774428e-06, "loss": 0.1879, "step": 4744 }, { "epoch": 1.62, "learning_rate": 4.705883259916832e-06, "loss": 0.2124, "step": 4745 }, { "epoch": 1.62, "learning_rate": 4.697735356867999e-06, "loss": 0.2198, "step": 4746 }, { "epoch": 1.62, "learning_rate": 4.689593782167665e-06, "loss": 0.2044, "step": 4747 }, { "epoch": 1.62, "learning_rate": 4.681458538353636e-06, "loss": 0.2066, "step": 4748 }, { "epoch": 1.62, "learning_rate": 4.673329627961703e-06, "loss": 0.3547, "step": 4749 }, { "epoch": 1.62, "learning_rate": 4.6652070535257134e-06, "loss": 0.2164, "step": 4750 }, { "epoch": 1.62, "learning_rate": 4.657090817577511e-06, "loss": 0.1854, "step": 4751 }, { "epoch": 1.62, "learning_rate": 4.6489809226469976e-06, "loss": 0.2151, "step": 4752 }, { "epoch": 1.62, "learning_rate": 4.6408773712620645e-06, "loss": 0.2252, "step": 4753 }, { "epoch": 1.62, "learning_rate": 4.632780165948647e-06, "loss": 0.2138, "step": 4754 }, { "epoch": 1.62, "learning_rate": 4.624689309230704e-06, "loss": 0.1811, "step": 4755 }, { "epoch": 1.62, "learning_rate": 4.616604803630198e-06, "loss": 0.1858, "step": 4756 }, { "epoch": 1.62, "learning_rate": 4.60852665166713e-06, "loss": 0.2088, "step": 4757 }, { "epoch": 1.62, "learning_rate": 4.600454855859507e-06, "loss": 0.2012, "step": 4758 }, { "epoch": 1.62, "learning_rate": 4.592389418723361e-06, "loss": 0.1852, "step": 4759 }, { "epoch": 1.62, "learning_rate": 4.5843303427727495e-06, "loss": 0.1889, "step": 4760 }, { "epoch": 1.62, "learning_rate": 4.576277630519718e-06, "loss": 0.1941, "step": 4761 }, { "epoch": 1.62, "learning_rate": 4.568231284474381e-06, "loss": 0.194, "step": 4762 }, { "epoch": 1.63, "learning_rate": 4.560191307144812e-06, "loss": 0.2162, "step": 4763 }, { "epoch": 1.63, "learning_rate": 4.552157701037141e-06, "loss": 0.21, "step": 4764 }, { "epoch": 1.63, "learning_rate": 4.54413046865548e-06, "loss": 0.2153, "step": 4765 }, { "epoch": 1.63, "learning_rate": 4.5361096125019765e-06, "loss": 0.1881, "step": 4766 }, { "epoch": 1.63, "learning_rate": 4.528095135076793e-06, "loss": 0.2057, "step": 4767 }, { "epoch": 1.63, "learning_rate": 4.520087038878079e-06, "loss": 0.2063, "step": 4768 }, { "epoch": 1.63, "learning_rate": 4.512085326402024e-06, "loss": 0.2218, "step": 4769 }, { "epoch": 1.63, "learning_rate": 4.5040900001427995e-06, "loss": 0.2099, "step": 4770 }, { "epoch": 1.63, "learning_rate": 4.496101062592611e-06, "loss": 0.2001, "step": 4771 }, { "epoch": 1.63, "learning_rate": 4.488118516241665e-06, "loss": 0.1831, "step": 4772 }, { "epoch": 1.63, "learning_rate": 4.480142363578163e-06, "loss": 0.201, "step": 4773 }, { "epoch": 1.63, "learning_rate": 4.4721726070883375e-06, "loss": 0.174, "step": 4774 }, { "epoch": 1.63, "learning_rate": 4.464209249256393e-06, "loss": 0.2065, "step": 4775 }, { "epoch": 1.63, "learning_rate": 4.456252292564583e-06, "loss": 0.1857, "step": 4776 }, { "epoch": 1.63, "learning_rate": 4.44830173949313e-06, "loss": 0.2284, "step": 4777 }, { "epoch": 1.63, "learning_rate": 4.440357592520281e-06, "loss": 0.1849, "step": 4778 }, { "epoch": 1.63, "learning_rate": 4.432419854122266e-06, "loss": 0.2132, "step": 4779 }, { "epoch": 1.63, "learning_rate": 4.4244885267733375e-06, "loss": 0.2159, "step": 4780 }, { "epoch": 1.63, "learning_rate": 4.416563612945751e-06, "loss": 0.184, "step": 4781 }, { "epoch": 1.63, "learning_rate": 4.408645115109739e-06, "loss": 0.2064, "step": 4782 }, { "epoch": 1.63, "learning_rate": 4.400733035733559e-06, "loss": 0.1927, "step": 4783 }, { "epoch": 1.63, "learning_rate": 4.392827377283448e-06, "loss": 0.2076, "step": 4784 }, { "epoch": 1.63, "learning_rate": 4.384928142223657e-06, "loss": 0.1917, "step": 4785 }, { "epoch": 1.63, "learning_rate": 4.377035333016436e-06, "loss": 0.1873, "step": 4786 }, { "epoch": 1.63, "learning_rate": 4.3691489521220125e-06, "loss": 0.1767, "step": 4787 }, { "epoch": 1.63, "learning_rate": 4.36126900199863e-06, "loss": 0.2162, "step": 4788 }, { "epoch": 1.63, "learning_rate": 4.353395485102518e-06, "loss": 0.2058, "step": 4789 }, { "epoch": 1.63, "learning_rate": 4.345528403887913e-06, "loss": 0.2052, "step": 4790 }, { "epoch": 1.63, "learning_rate": 4.337667760807018e-06, "loss": 0.2078, "step": 4791 }, { "epoch": 1.63, "learning_rate": 4.329813558310059e-06, "loss": 0.2102, "step": 4792 }, { "epoch": 1.64, "learning_rate": 4.321965798845246e-06, "loss": 0.1896, "step": 4793 }, { "epoch": 1.64, "learning_rate": 4.314124484858761e-06, "loss": 0.2071, "step": 4794 }, { "epoch": 1.64, "learning_rate": 4.306289618794809e-06, "loss": 0.1916, "step": 4795 }, { "epoch": 1.64, "learning_rate": 4.298461203095555e-06, "loss": 0.1865, "step": 4796 }, { "epoch": 1.64, "learning_rate": 4.290639240201177e-06, "loss": 0.2025, "step": 4797 }, { "epoch": 1.64, "learning_rate": 4.2828237325498345e-06, "loss": 0.1981, "step": 4798 }, { "epoch": 1.64, "learning_rate": 4.275014682577658e-06, "loss": 0.1935, "step": 4799 }, { "epoch": 1.64, "learning_rate": 4.267212092718798e-06, "loss": 0.1998, "step": 4800 }, { "epoch": 1.64, "learning_rate": 4.259415965405356e-06, "loss": 0.1814, "step": 4801 }, { "epoch": 1.64, "learning_rate": 4.251626303067441e-06, "loss": 0.1888, "step": 4802 }, { "epoch": 1.64, "learning_rate": 4.243843108133144e-06, "loss": 0.2162, "step": 4803 }, { "epoch": 1.64, "learning_rate": 4.236066383028545e-06, "loss": 0.2145, "step": 4804 }, { "epoch": 1.64, "learning_rate": 4.228296130177684e-06, "loss": 0.1822, "step": 4805 }, { "epoch": 1.64, "learning_rate": 4.22053235200261e-06, "loss": 0.186, "step": 4806 }, { "epoch": 1.64, "learning_rate": 4.212775050923345e-06, "loss": 0.1785, "step": 4807 }, { "epoch": 1.64, "learning_rate": 4.205024229357882e-06, "loss": 0.1993, "step": 4808 }, { "epoch": 1.64, "learning_rate": 4.197279889722211e-06, "loss": 0.2211, "step": 4809 }, { "epoch": 1.64, "learning_rate": 4.189542034430285e-06, "loss": 0.2048, "step": 4810 }, { "epoch": 1.64, "learning_rate": 4.181810665894051e-06, "loss": 0.1951, "step": 4811 }, { "epoch": 1.64, "learning_rate": 4.174085786523427e-06, "loss": 0.2221, "step": 4812 }, { "epoch": 1.64, "learning_rate": 4.1663673987263e-06, "loss": 0.2092, "step": 4813 }, { "epoch": 1.64, "learning_rate": 4.158655504908557e-06, "loss": 0.1734, "step": 4814 }, { "epoch": 1.64, "learning_rate": 4.15095010747403e-06, "loss": 0.2233, "step": 4815 }, { "epoch": 1.64, "learning_rate": 4.1432512088245474e-06, "loss": 0.2165, "step": 4816 }, { "epoch": 1.64, "learning_rate": 4.135558811359907e-06, "loss": 0.2132, "step": 4817 }, { "epoch": 1.64, "learning_rate": 4.127872917477879e-06, "loss": 0.2031, "step": 4818 }, { "epoch": 1.64, "learning_rate": 4.120193529574215e-06, "loss": 0.1854, "step": 4819 }, { "epoch": 1.64, "learning_rate": 4.112520650042614e-06, "loss": 0.2084, "step": 4820 }, { "epoch": 1.64, "learning_rate": 4.104854281274778e-06, "loss": 0.2223, "step": 4821 }, { "epoch": 1.65, "learning_rate": 4.097194425660353e-06, "loss": 0.1793, "step": 4822 }, { "epoch": 1.65, "learning_rate": 4.089541085586968e-06, "loss": 0.2024, "step": 4823 }, { "epoch": 1.65, "learning_rate": 4.081894263440231e-06, "loss": 0.2017, "step": 4824 }, { "epoch": 1.65, "learning_rate": 4.074253961603688e-06, "loss": 0.2116, "step": 4825 }, { "epoch": 1.65, "learning_rate": 4.066620182458886e-06, "loss": 0.1969, "step": 4826 }, { "epoch": 1.65, "learning_rate": 4.058992928385314e-06, "loss": 0.2021, "step": 4827 }, { "epoch": 1.65, "learning_rate": 4.051372201760439e-06, "loss": 0.2009, "step": 4828 }, { "epoch": 1.65, "learning_rate": 4.043758004959699e-06, "loss": 0.2322, "step": 4829 }, { "epoch": 1.65, "learning_rate": 4.036150340356479e-06, "loss": 0.1876, "step": 4830 }, { "epoch": 1.65, "learning_rate": 4.0285492103221385e-06, "loss": 0.2026, "step": 4831 }, { "epoch": 1.65, "learning_rate": 4.020954617226003e-06, "loss": 0.2079, "step": 4832 }, { "epoch": 1.65, "learning_rate": 4.013366563435364e-06, "loss": 0.1769, "step": 4833 }, { "epoch": 1.65, "learning_rate": 4.005785051315453e-06, "loss": 0.2056, "step": 4834 }, { "epoch": 1.65, "learning_rate": 3.9982100832294895e-06, "loss": 0.1868, "step": 4835 }, { "epoch": 1.65, "learning_rate": 3.99064166153863e-06, "loss": 0.1833, "step": 4836 }, { "epoch": 1.65, "learning_rate": 3.983079788602004e-06, "loss": 0.2108, "step": 4837 }, { "epoch": 1.65, "learning_rate": 3.975524466776704e-06, "loss": 0.2402, "step": 4838 }, { "epoch": 1.65, "learning_rate": 3.967975698417761e-06, "loss": 0.1996, "step": 4839 }, { "epoch": 1.65, "learning_rate": 3.96043348587819e-06, "loss": 0.1803, "step": 4840 }, { "epoch": 1.65, "learning_rate": 3.952897831508934e-06, "loss": 0.1821, "step": 4841 }, { "epoch": 1.65, "learning_rate": 3.9453687376589086e-06, "loss": 0.2138, "step": 4842 }, { "epoch": 1.65, "learning_rate": 3.937846206674992e-06, "loss": 0.2124, "step": 4843 }, { "epoch": 1.65, "learning_rate": 3.930330240901986e-06, "loss": 0.1884, "step": 4844 }, { "epoch": 1.65, "learning_rate": 3.922820842682692e-06, "loss": 0.2233, "step": 4845 }, { "epoch": 1.65, "learning_rate": 3.915318014357819e-06, "loss": 0.2115, "step": 4846 }, { "epoch": 1.65, "learning_rate": 3.907821758266059e-06, "loss": 0.2004, "step": 4847 }, { "epoch": 1.65, "learning_rate": 3.900332076744034e-06, "loss": 0.1955, "step": 4848 }, { "epoch": 1.65, "learning_rate": 3.892848972126331e-06, "loss": 0.2161, "step": 4849 }, { "epoch": 1.65, "learning_rate": 3.885372446745492e-06, "loss": 0.1881, "step": 4850 }, { "epoch": 1.66, "learning_rate": 3.877902502931979e-06, "loss": 0.2116, "step": 4851 }, { "epoch": 1.66, "learning_rate": 3.8704391430142425e-06, "loss": 0.2078, "step": 4852 }, { "epoch": 1.66, "learning_rate": 3.862982369318646e-06, "loss": 0.2266, "step": 4853 }, { "epoch": 1.66, "learning_rate": 3.855532184169519e-06, "loss": 0.2008, "step": 4854 }, { "epoch": 1.66, "learning_rate": 3.84808858988914e-06, "loss": 0.2033, "step": 4855 }, { "epoch": 1.66, "learning_rate": 3.8406515887977116e-06, "loss": 0.196, "step": 4856 }, { "epoch": 1.66, "learning_rate": 3.833221183213412e-06, "loss": 0.1925, "step": 4857 }, { "epoch": 1.66, "learning_rate": 3.825797375452325e-06, "loss": 0.222, "step": 4858 }, { "epoch": 1.66, "learning_rate": 3.818380167828528e-06, "loss": 0.2329, "step": 4859 }, { "epoch": 1.66, "learning_rate": 3.8109695626539932e-06, "loss": 0.2118, "step": 4860 }, { "epoch": 1.66, "learning_rate": 3.8035655622386644e-06, "loss": 0.1924, "step": 4861 }, { "epoch": 1.66, "learning_rate": 3.7961681688904073e-06, "loss": 0.1948, "step": 4862 }, { "epoch": 1.66, "learning_rate": 3.788777384915046e-06, "loss": 0.208, "step": 4863 }, { "epoch": 1.66, "learning_rate": 3.7813932126163353e-06, "loss": 0.18, "step": 4864 }, { "epoch": 1.66, "learning_rate": 3.7740156542959648e-06, "loss": 0.2228, "step": 4865 }, { "epoch": 1.66, "learning_rate": 3.7666447122535755e-06, "loss": 0.2393, "step": 4866 }, { "epoch": 1.66, "learning_rate": 3.759280388786729e-06, "loss": 0.2161, "step": 4867 }, { "epoch": 1.66, "learning_rate": 3.751922686190937e-06, "loss": 0.2179, "step": 4868 }, { "epoch": 1.66, "learning_rate": 3.7445716067596503e-06, "loss": 0.2125, "step": 4869 }, { "epoch": 1.66, "learning_rate": 3.7372271527842367e-06, "loss": 0.1925, "step": 4870 }, { "epoch": 1.66, "learning_rate": 3.72988932655402e-06, "loss": 0.2304, "step": 4871 }, { "epoch": 1.66, "learning_rate": 3.722558130356235e-06, "loss": 0.1691, "step": 4872 }, { "epoch": 1.66, "learning_rate": 3.715233566476084e-06, "loss": 0.1878, "step": 4873 }, { "epoch": 1.66, "learning_rate": 3.707915637196663e-06, "loss": 0.1967, "step": 4874 }, { "epoch": 1.66, "learning_rate": 3.7006043447990272e-06, "loss": 0.2023, "step": 4875 }, { "epoch": 1.66, "learning_rate": 3.69329969156216e-06, "loss": 0.2268, "step": 4876 }, { "epoch": 1.66, "learning_rate": 3.686001679762954e-06, "loss": 0.2046, "step": 4877 }, { "epoch": 1.66, "learning_rate": 3.6787103116762616e-06, "loss": 0.233, "step": 4878 }, { "epoch": 1.66, "learning_rate": 3.6714255895748394e-06, "loss": 0.1911, "step": 4879 }, { "epoch": 1.66, "learning_rate": 3.6641475157293855e-06, "loss": 0.2124, "step": 4880 }, { "epoch": 1.67, "learning_rate": 3.6568760924085314e-06, "loss": 0.1766, "step": 4881 }, { "epoch": 1.67, "learning_rate": 3.6496113218788154e-06, "loss": 0.1984, "step": 4882 }, { "epoch": 1.67, "learning_rate": 3.642353206404725e-06, "loss": 0.1997, "step": 4883 }, { "epoch": 1.67, "learning_rate": 3.6351017482486483e-06, "loss": 0.2026, "step": 4884 }, { "epoch": 1.67, "learning_rate": 3.6278569496709213e-06, "loss": 0.2067, "step": 4885 }, { "epoch": 1.67, "learning_rate": 3.6206188129297963e-06, "loss": 0.2011, "step": 4886 }, { "epoch": 1.67, "learning_rate": 3.613387340281449e-06, "loss": 0.1895, "step": 4887 }, { "epoch": 1.67, "learning_rate": 3.606162533979968e-06, "loss": 0.2233, "step": 4888 }, { "epoch": 1.67, "learning_rate": 3.5989443962773763e-06, "loss": 0.1954, "step": 4889 }, { "epoch": 1.67, "learning_rate": 3.5917329294236217e-06, "loss": 0.1735, "step": 4890 }, { "epoch": 1.67, "learning_rate": 3.5845281356665567e-06, "loss": 0.1979, "step": 4891 }, { "epoch": 1.67, "learning_rate": 3.5773300172519673e-06, "loss": 0.2164, "step": 4892 }, { "epoch": 1.67, "learning_rate": 3.5701385764235513e-06, "loss": 0.2029, "step": 4893 }, { "epoch": 1.67, "learning_rate": 3.562953815422926e-06, "loss": 0.1996, "step": 4894 }, { "epoch": 1.67, "learning_rate": 3.555775736489639e-06, "loss": 0.2375, "step": 4895 }, { "epoch": 1.67, "learning_rate": 3.548604341861128e-06, "loss": 0.2261, "step": 4896 }, { "epoch": 1.67, "learning_rate": 3.5414396337727805e-06, "loss": 0.1966, "step": 4897 }, { "epoch": 1.67, "learning_rate": 3.5342816144578685e-06, "loss": 0.1958, "step": 4898 }, { "epoch": 1.67, "learning_rate": 3.527130286147598e-06, "loss": 0.2453, "step": 4899 }, { "epoch": 1.67, "learning_rate": 3.5199856510710867e-06, "loss": 0.1817, "step": 4900 }, { "epoch": 1.67, "learning_rate": 3.5128477114553637e-06, "loss": 0.2165, "step": 4901 }, { "epoch": 1.67, "learning_rate": 3.5057164695253774e-06, "loss": 0.198, "step": 4902 }, { "epoch": 1.67, "learning_rate": 3.4985919275039692e-06, "loss": 0.1812, "step": 4903 }, { "epoch": 1.67, "learning_rate": 3.491474087611918e-06, "loss": 0.2166, "step": 4904 }, { "epoch": 1.67, "learning_rate": 3.4843629520678904e-06, "loss": 0.233, "step": 4905 }, { "epoch": 1.67, "learning_rate": 3.4772585230884808e-06, "loss": 0.2024, "step": 4906 }, { "epoch": 1.67, "learning_rate": 3.4701608028881866e-06, "loss": 0.1955, "step": 4907 }, { "epoch": 1.67, "learning_rate": 3.4630697936794076e-06, "loss": 0.2197, "step": 4908 }, { "epoch": 1.67, "learning_rate": 3.4559854976724678e-06, "loss": 0.2048, "step": 4909 }, { "epoch": 1.68, "learning_rate": 3.448907917075575e-06, "loss": 0.1961, "step": 4910 }, { "epoch": 1.68, "learning_rate": 3.441837054094868e-06, "loss": 0.2093, "step": 4911 }, { "epoch": 1.68, "learning_rate": 3.4347729109343725e-06, "loss": 0.1801, "step": 4912 }, { "epoch": 1.68, "learning_rate": 3.427715489796035e-06, "loss": 0.1943, "step": 4913 }, { "epoch": 1.68, "learning_rate": 3.4206647928796965e-06, "loss": 0.1681, "step": 4914 }, { "epoch": 1.68, "learning_rate": 3.413620822383107e-06, "loss": 0.2033, "step": 4915 }, { "epoch": 1.68, "learning_rate": 3.4065835805019207e-06, "loss": 0.1988, "step": 4916 }, { "epoch": 1.68, "learning_rate": 3.399553069429684e-06, "loss": 0.2189, "step": 4917 }, { "epoch": 1.68, "learning_rate": 3.3925292913578653e-06, "loss": 0.2019, "step": 4918 }, { "epoch": 1.68, "learning_rate": 3.3855122484758055e-06, "loss": 0.1999, "step": 4919 }, { "epoch": 1.68, "learning_rate": 3.3785019429707743e-06, "loss": 0.1971, "step": 4920 }, { "epoch": 1.68, "learning_rate": 3.371498377027932e-06, "loss": 0.173, "step": 4921 }, { "epoch": 1.68, "learning_rate": 3.3645015528303263e-06, "loss": 0.1608, "step": 4922 }, { "epoch": 1.68, "learning_rate": 3.357511472558922e-06, "loss": 0.2057, "step": 4923 }, { "epoch": 1.68, "learning_rate": 3.350528138392564e-06, "loss": 0.1789, "step": 4924 }, { "epoch": 1.68, "learning_rate": 3.343551552508009e-06, "loss": 0.2022, "step": 4925 }, { "epoch": 1.68, "learning_rate": 3.336581717079906e-06, "loss": 0.1942, "step": 4926 }, { "epoch": 1.68, "learning_rate": 3.3296186342807845e-06, "loss": 0.211, "step": 4927 }, { "epoch": 1.68, "learning_rate": 3.3226623062811076e-06, "loss": 0.1791, "step": 4928 }, { "epoch": 1.68, "learning_rate": 3.3157127352491852e-06, "loss": 0.2121, "step": 4929 }, { "epoch": 1.68, "learning_rate": 3.3087699233512627e-06, "loss": 0.2051, "step": 4930 }, { "epoch": 1.68, "learning_rate": 3.301833872751442e-06, "loss": 0.2279, "step": 4931 }, { "epoch": 1.68, "learning_rate": 3.2949045856117437e-06, "loss": 0.1687, "step": 4932 }, { "epoch": 1.68, "learning_rate": 3.2879820640920766e-06, "loss": 0.2072, "step": 4933 }, { "epoch": 1.68, "learning_rate": 3.281066310350228e-06, "loss": 0.213, "step": 4934 }, { "epoch": 1.68, "learning_rate": 3.2741573265418905e-06, "loss": 0.2003, "step": 4935 }, { "epoch": 1.68, "learning_rate": 3.2672551148206306e-06, "loss": 0.2123, "step": 4936 }, { "epoch": 1.68, "learning_rate": 3.2603596773379237e-06, "loss": 0.1762, "step": 4937 }, { "epoch": 1.68, "learning_rate": 3.253471016243112e-06, "loss": 0.1947, "step": 4938 }, { "epoch": 1.69, "learning_rate": 3.2465891336834394e-06, "loss": 0.193, "step": 4939 }, { "epoch": 1.69, "learning_rate": 3.239714031804042e-06, "loss": 0.1995, "step": 4940 }, { "epoch": 1.69, "learning_rate": 3.2328457127479144e-06, "loss": 0.2237, "step": 4941 }, { "epoch": 1.69, "learning_rate": 3.225984178655983e-06, "loss": 0.234, "step": 4942 }, { "epoch": 1.69, "learning_rate": 3.2191294316670117e-06, "loss": 0.2023, "step": 4943 }, { "epoch": 1.69, "learning_rate": 3.2122814739176833e-06, "loss": 0.208, "step": 4944 }, { "epoch": 1.69, "learning_rate": 3.20544030754254e-06, "loss": 0.1889, "step": 4945 }, { "epoch": 1.69, "learning_rate": 3.1986059346740234e-06, "loss": 0.1815, "step": 4946 }, { "epoch": 1.69, "learning_rate": 3.1917783574424572e-06, "loss": 0.2047, "step": 4947 }, { "epoch": 1.69, "learning_rate": 3.1849575779760303e-06, "loss": 0.1949, "step": 4948 }, { "epoch": 1.69, "learning_rate": 3.1781435984008347e-06, "loss": 0.1985, "step": 4949 }, { "epoch": 1.69, "learning_rate": 3.171336420840826e-06, "loss": 0.185, "step": 4950 }, { "epoch": 1.69, "learning_rate": 3.164536047417846e-06, "loss": 0.2099, "step": 4951 }, { "epoch": 1.69, "learning_rate": 3.1577424802516215e-06, "loss": 0.1961, "step": 4952 }, { "epoch": 1.69, "learning_rate": 3.1509557214597447e-06, "loss": 0.2058, "step": 4953 }, { "epoch": 1.69, "learning_rate": 3.144175773157698e-06, "loss": 0.1775, "step": 4954 }, { "epoch": 1.69, "learning_rate": 3.137402637458825e-06, "loss": 0.1957, "step": 4955 }, { "epoch": 1.69, "learning_rate": 3.130636316474375e-06, "loss": 0.187, "step": 4956 }, { "epoch": 1.69, "learning_rate": 3.1238768123134395e-06, "loss": 0.2198, "step": 4957 }, { "epoch": 1.69, "learning_rate": 3.1171241270830013e-06, "loss": 0.2152, "step": 4958 }, { "epoch": 1.69, "learning_rate": 3.110378262887928e-06, "loss": 0.2012, "step": 4959 }, { "epoch": 1.69, "learning_rate": 3.103639221830937e-06, "loss": 0.2351, "step": 4960 }, { "epoch": 1.69, "learning_rate": 3.096907006012639e-06, "loss": 0.1881, "step": 4961 }, { "epoch": 1.69, "learning_rate": 3.090181617531501e-06, "loss": 0.2219, "step": 4962 }, { "epoch": 1.69, "learning_rate": 3.083463058483882e-06, "loss": 0.2233, "step": 4963 }, { "epoch": 1.69, "learning_rate": 3.0767513309639893e-06, "loss": 0.1946, "step": 4964 }, { "epoch": 1.69, "learning_rate": 3.070046437063917e-06, "loss": 0.1989, "step": 4965 }, { "epoch": 1.69, "learning_rate": 3.063348378873632e-06, "loss": 0.1858, "step": 4966 }, { "epoch": 1.69, "learning_rate": 3.0566571584809497e-06, "loss": 0.1876, "step": 4967 }, { "epoch": 1.69, "learning_rate": 3.049972777971577e-06, "loss": 0.1954, "step": 4968 }, { "epoch": 1.7, "learning_rate": 3.043295239429064e-06, "loss": 0.1896, "step": 4969 }, { "epoch": 1.7, "learning_rate": 3.0366245449348685e-06, "loss": 0.1985, "step": 4970 }, { "epoch": 1.7, "learning_rate": 3.029960696568268e-06, "loss": 0.2068, "step": 4971 }, { "epoch": 1.7, "learning_rate": 3.023303696406435e-06, "loss": 0.1868, "step": 4972 }, { "epoch": 1.7, "learning_rate": 3.0166535465244063e-06, "loss": 0.1831, "step": 4973 }, { "epoch": 1.7, "learning_rate": 3.010010248995063e-06, "loss": 0.1914, "step": 4974 }, { "epoch": 1.7, "learning_rate": 3.0033738058891784e-06, "loss": 0.1814, "step": 4975 }, { "epoch": 1.7, "learning_rate": 2.9967442192753632e-06, "loss": 0.2057, "step": 4976 }, { "epoch": 1.7, "learning_rate": 2.9901214912201075e-06, "loss": 0.2002, "step": 4977 }, { "epoch": 1.7, "learning_rate": 2.983505623787766e-06, "loss": 0.19, "step": 4978 }, { "epoch": 1.7, "learning_rate": 2.976896619040537e-06, "loss": 0.22, "step": 4979 }, { "epoch": 1.7, "learning_rate": 2.970294479038499e-06, "loss": 0.1934, "step": 4980 }, { "epoch": 1.7, "learning_rate": 2.963699205839576e-06, "loss": 0.197, "step": 4981 }, { "epoch": 1.7, "learning_rate": 2.9571108014995573e-06, "loss": 0.1641, "step": 4982 }, { "epoch": 1.7, "learning_rate": 2.950529268072094e-06, "loss": 0.2041, "step": 4983 }, { "epoch": 1.7, "learning_rate": 2.9439546076086926e-06, "loss": 0.2088, "step": 4984 }, { "epoch": 1.7, "learning_rate": 2.9373868221587245e-06, "loss": 0.1884, "step": 4985 }, { "epoch": 1.7, "learning_rate": 2.9308259137694e-06, "loss": 0.1774, "step": 4986 }, { "epoch": 1.7, "learning_rate": 2.9242718844858075e-06, "loss": 0.2409, "step": 4987 }, { "epoch": 1.7, "learning_rate": 2.9177247363508715e-06, "loss": 0.2049, "step": 4988 }, { "epoch": 1.7, "learning_rate": 2.91118447140539e-06, "loss": 0.2209, "step": 4989 }, { "epoch": 1.7, "learning_rate": 2.9046510916879932e-06, "loss": 0.2072, "step": 4990 }, { "epoch": 1.7, "learning_rate": 2.8981245992351887e-06, "loss": 0.2123, "step": 4991 }, { "epoch": 1.7, "learning_rate": 2.891604996081329e-06, "loss": 0.1878, "step": 4992 }, { "epoch": 1.7, "learning_rate": 2.885092284258606e-06, "loss": 0.186, "step": 4993 }, { "epoch": 1.7, "learning_rate": 2.878586465797087e-06, "loss": 0.1983, "step": 4994 }, { "epoch": 1.7, "learning_rate": 2.872087542724669e-06, "loss": 0.2065, "step": 4995 }, { "epoch": 1.7, "learning_rate": 2.8655955170671073e-06, "loss": 0.1874, "step": 4996 }, { "epoch": 1.7, "learning_rate": 2.859110390848016e-06, "loss": 0.1902, "step": 4997 }, { "epoch": 1.71, "learning_rate": 2.852632166088848e-06, "loss": 0.2098, "step": 4998 }, { "epoch": 1.71, "learning_rate": 2.846160844808915e-06, "loss": 0.1922, "step": 4999 }, { "epoch": 1.71, "learning_rate": 2.839696429025357e-06, "loss": 0.1825, "step": 5000 }, { "epoch": 1.71, "learning_rate": 2.8332389207531868e-06, "loss": 0.1825, "step": 5001 }, { "epoch": 1.71, "learning_rate": 2.8267883220052414e-06, "loss": 0.2136, "step": 5002 }, { "epoch": 1.71, "learning_rate": 2.8203446347922213e-06, "loss": 0.187, "step": 5003 }, { "epoch": 1.71, "learning_rate": 2.8139078611226668e-06, "loss": 0.2003, "step": 5004 }, { "epoch": 1.71, "learning_rate": 2.8074780030029584e-06, "loss": 0.1928, "step": 5005 }, { "epoch": 1.71, "learning_rate": 2.8010550624373273e-06, "loss": 0.2119, "step": 5006 }, { "epoch": 1.71, "learning_rate": 2.7946390414278435e-06, "loss": 0.1885, "step": 5007 }, { "epoch": 1.71, "learning_rate": 2.788229941974421e-06, "loss": 0.1803, "step": 5008 }, { "epoch": 1.71, "learning_rate": 2.7818277660748293e-06, "loss": 0.1862, "step": 5009 }, { "epoch": 1.71, "learning_rate": 2.775432515724649e-06, "loss": 0.2046, "step": 5010 }, { "epoch": 1.71, "learning_rate": 2.7690441929173416e-06, "loss": 0.2231, "step": 5011 }, { "epoch": 1.71, "learning_rate": 2.762662799644178e-06, "loss": 0.2011, "step": 5012 }, { "epoch": 1.71, "learning_rate": 2.756288337894283e-06, "loss": 0.2122, "step": 5013 }, { "epoch": 1.71, "learning_rate": 2.749920809654616e-06, "loss": 0.2099, "step": 5014 }, { "epoch": 1.71, "learning_rate": 2.7435602169099823e-06, "loss": 0.1766, "step": 5015 }, { "epoch": 1.71, "learning_rate": 2.737206561643013e-06, "loss": 0.1985, "step": 5016 }, { "epoch": 1.71, "learning_rate": 2.7308598458341855e-06, "loss": 0.2042, "step": 5017 }, { "epoch": 1.71, "learning_rate": 2.7245200714618197e-06, "loss": 0.188, "step": 5018 }, { "epoch": 1.71, "learning_rate": 2.718187240502054e-06, "loss": 0.208, "step": 5019 }, { "epoch": 1.71, "learning_rate": 2.711861354928885e-06, "loss": 0.2143, "step": 5020 }, { "epoch": 1.71, "learning_rate": 2.7055424167141185e-06, "loss": 0.1828, "step": 5021 }, { "epoch": 1.71, "learning_rate": 2.6992304278274206e-06, "loss": 0.2053, "step": 5022 }, { "epoch": 1.71, "learning_rate": 2.6929253902362757e-06, "loss": 0.2038, "step": 5023 }, { "epoch": 1.71, "learning_rate": 2.6866273059059975e-06, "loss": 0.2367, "step": 5024 }, { "epoch": 1.71, "learning_rate": 2.6803361767997583e-06, "loss": 0.203, "step": 5025 }, { "epoch": 1.71, "learning_rate": 2.6740520048785272e-06, "loss": 0.2124, "step": 5026 }, { "epoch": 1.72, "learning_rate": 2.66777479210113e-06, "loss": 0.1891, "step": 5027 }, { "epoch": 1.72, "learning_rate": 2.6615045404242117e-06, "loss": 0.2017, "step": 5028 }, { "epoch": 1.72, "learning_rate": 2.6552412518022484e-06, "loss": 0.2095, "step": 5029 }, { "epoch": 1.72, "learning_rate": 2.648984928187559e-06, "loss": 0.209, "step": 5030 }, { "epoch": 1.72, "learning_rate": 2.6427355715302664e-06, "loss": 0.1802, "step": 5031 }, { "epoch": 1.72, "learning_rate": 2.6364931837783462e-06, "loss": 0.2226, "step": 5032 }, { "epoch": 1.72, "learning_rate": 2.6302577668775837e-06, "loss": 0.1919, "step": 5033 }, { "epoch": 1.72, "learning_rate": 2.6240293227716013e-06, "loss": 0.1735, "step": 5034 }, { "epoch": 1.72, "learning_rate": 2.6178078534018546e-06, "loss": 0.1827, "step": 5035 }, { "epoch": 1.72, "learning_rate": 2.611593360707601e-06, "loss": 0.238, "step": 5036 }, { "epoch": 1.72, "learning_rate": 2.6053858466259546e-06, "loss": 0.1969, "step": 5037 }, { "epoch": 1.72, "learning_rate": 2.5991853130918175e-06, "loss": 0.2012, "step": 5038 }, { "epoch": 1.72, "learning_rate": 2.592991762037958e-06, "loss": 0.1876, "step": 5039 }, { "epoch": 1.72, "learning_rate": 2.586805195394934e-06, "loss": 0.2164, "step": 5040 }, { "epoch": 1.72, "learning_rate": 2.5806256150911477e-06, "loss": 0.2099, "step": 5041 }, { "epoch": 1.72, "learning_rate": 2.574453023052803e-06, "loss": 0.2268, "step": 5042 }, { "epoch": 1.72, "learning_rate": 2.568287421203941e-06, "loss": 0.1988, "step": 5043 }, { "epoch": 1.72, "learning_rate": 2.5621288114664297e-06, "loss": 0.2119, "step": 5044 }, { "epoch": 1.72, "learning_rate": 2.5559771957599344e-06, "loss": 0.1744, "step": 5045 }, { "epoch": 1.72, "learning_rate": 2.549832576001965e-06, "loss": 0.1894, "step": 5046 }, { "epoch": 1.72, "learning_rate": 2.5436949541078307e-06, "loss": 0.192, "step": 5047 }, { "epoch": 1.72, "learning_rate": 2.5375643319906715e-06, "loss": 0.2261, "step": 5048 }, { "epoch": 1.72, "learning_rate": 2.5314407115614473e-06, "loss": 0.1849, "step": 5049 }, { "epoch": 1.72, "learning_rate": 2.5253240947289232e-06, "loss": 0.1859, "step": 5050 }, { "epoch": 1.72, "learning_rate": 2.5192144833996954e-06, "loss": 0.2286, "step": 5051 }, { "epoch": 1.72, "learning_rate": 2.5131118794781553e-06, "loss": 0.1969, "step": 5052 }, { "epoch": 1.72, "learning_rate": 2.507016284866548e-06, "loss": 0.1976, "step": 5053 }, { "epoch": 1.72, "learning_rate": 2.500927701464889e-06, "loss": 0.187, "step": 5054 }, { "epoch": 1.72, "learning_rate": 2.4948461311710413e-06, "loss": 0.2119, "step": 5055 }, { "epoch": 1.73, "learning_rate": 2.488771575880669e-06, "loss": 0.1912, "step": 5056 }, { "epoch": 1.73, "learning_rate": 2.4827040374872435e-06, "loss": 0.2118, "step": 5057 }, { "epoch": 1.73, "learning_rate": 2.4766435178820657e-06, "loss": 0.2116, "step": 5058 }, { "epoch": 1.73, "learning_rate": 2.47059001895423e-06, "loss": 0.2093, "step": 5059 }, { "epoch": 1.73, "learning_rate": 2.4645435425906556e-06, "loss": 0.1812, "step": 5060 }, { "epoch": 1.73, "learning_rate": 2.4585040906760755e-06, "loss": 0.2151, "step": 5061 }, { "epoch": 1.73, "learning_rate": 2.4524716650930117e-06, "loss": 0.1912, "step": 5062 }, { "epoch": 1.73, "learning_rate": 2.4464462677218273e-06, "loss": 0.1985, "step": 5063 }, { "epoch": 1.73, "learning_rate": 2.4404279004406616e-06, "loss": 0.2142, "step": 5064 }, { "epoch": 1.73, "learning_rate": 2.434416565125494e-06, "loss": 0.1864, "step": 5065 }, { "epoch": 1.73, "learning_rate": 2.4284122636500788e-06, "loss": 0.2069, "step": 5066 }, { "epoch": 1.73, "learning_rate": 2.422414997886019e-06, "loss": 0.2049, "step": 5067 }, { "epoch": 1.73, "learning_rate": 2.4164247697026827e-06, "loss": 0.216, "step": 5068 }, { "epoch": 1.73, "learning_rate": 2.410441580967271e-06, "loss": 0.1891, "step": 5069 }, { "epoch": 1.73, "learning_rate": 2.4044654335447865e-06, "loss": 0.2227, "step": 5070 }, { "epoch": 1.73, "learning_rate": 2.3984963292980223e-06, "loss": 0.2075, "step": 5071 }, { "epoch": 1.73, "learning_rate": 2.3925342700876014e-06, "loss": 0.1708, "step": 5072 }, { "epoch": 1.73, "learning_rate": 2.3865792577719204e-06, "loss": 0.1981, "step": 5073 }, { "epoch": 1.73, "learning_rate": 2.3806312942072083e-06, "loss": 0.1997, "step": 5074 }, { "epoch": 1.73, "learning_rate": 2.3746903812474823e-06, "loss": 0.1741, "step": 5075 }, { "epoch": 1.73, "learning_rate": 2.3687565207445584e-06, "loss": 0.2142, "step": 5076 }, { "epoch": 1.73, "learning_rate": 2.362829714548065e-06, "loss": 0.1865, "step": 5077 }, { "epoch": 1.73, "learning_rate": 2.3569099645054216e-06, "loss": 0.2072, "step": 5078 }, { "epoch": 1.73, "learning_rate": 2.350997272461855e-06, "loss": 0.1928, "step": 5079 }, { "epoch": 1.73, "learning_rate": 2.3450916402603885e-06, "loss": 0.1974, "step": 5080 }, { "epoch": 1.73, "learning_rate": 2.339193069741849e-06, "loss": 0.1846, "step": 5081 }, { "epoch": 1.73, "learning_rate": 2.3333015627448635e-06, "loss": 0.2041, "step": 5082 }, { "epoch": 1.73, "learning_rate": 2.3274171211058458e-06, "loss": 0.2205, "step": 5083 }, { "epoch": 1.73, "learning_rate": 2.3215397466590184e-06, "loss": 0.1974, "step": 5084 }, { "epoch": 1.73, "learning_rate": 2.3156694412363926e-06, "loss": 0.1912, "step": 5085 }, { "epoch": 1.74, "learning_rate": 2.309806206667786e-06, "loss": 0.2296, "step": 5086 }, { "epoch": 1.74, "learning_rate": 2.303950044780809e-06, "loss": 0.2157, "step": 5087 }, { "epoch": 1.74, "learning_rate": 2.298100957400859e-06, "loss": 0.1921, "step": 5088 }, { "epoch": 1.74, "learning_rate": 2.292258946351139e-06, "loss": 0.2208, "step": 5089 }, { "epoch": 1.74, "learning_rate": 2.28642401345264e-06, "loss": 0.2128, "step": 5090 }, { "epoch": 1.74, "learning_rate": 2.280596160524154e-06, "loss": 0.2089, "step": 5091 }, { "epoch": 1.74, "learning_rate": 2.27477538938225e-06, "loss": 0.2277, "step": 5092 }, { "epoch": 1.74, "learning_rate": 2.268961701841307e-06, "loss": 0.1874, "step": 5093 }, { "epoch": 1.74, "learning_rate": 2.2631550997134877e-06, "loss": 0.2075, "step": 5094 }, { "epoch": 1.74, "learning_rate": 2.2573555848087513e-06, "loss": 0.205, "step": 5095 }, { "epoch": 1.74, "learning_rate": 2.2515631589348497e-06, "loss": 0.2177, "step": 5096 }, { "epoch": 1.74, "learning_rate": 2.2457778238973046e-06, "loss": 0.1893, "step": 5097 }, { "epoch": 1.74, "learning_rate": 2.239999581499455e-06, "loss": 0.21, "step": 5098 }, { "epoch": 1.74, "learning_rate": 2.2342284335424097e-06, "loss": 0.2175, "step": 5099 }, { "epoch": 1.74, "learning_rate": 2.228464381825077e-06, "loss": 0.2128, "step": 5100 }, { "epoch": 1.74, "learning_rate": 2.2227074281441525e-06, "loss": 0.2111, "step": 5101 }, { "epoch": 1.74, "learning_rate": 2.216957574294107e-06, "loss": 0.2093, "step": 5102 }, { "epoch": 1.74, "learning_rate": 2.21121482206722e-06, "loss": 0.1866, "step": 5103 }, { "epoch": 1.74, "learning_rate": 2.20547917325353e-06, "loss": 0.1894, "step": 5104 }, { "epoch": 1.74, "learning_rate": 2.199750629640887e-06, "loss": 0.212, "step": 5105 }, { "epoch": 1.74, "learning_rate": 2.1940291930149154e-06, "loss": 0.192, "step": 5106 }, { "epoch": 1.74, "learning_rate": 2.1883148651590134e-06, "loss": 0.1974, "step": 5107 }, { "epoch": 1.74, "learning_rate": 2.18260764785439e-06, "loss": 0.2234, "step": 5108 }, { "epoch": 1.74, "learning_rate": 2.176907542880011e-06, "loss": 0.1978, "step": 5109 }, { "epoch": 1.74, "learning_rate": 2.171214552012643e-06, "loss": 0.2232, "step": 5110 }, { "epoch": 1.74, "learning_rate": 2.1655286770268234e-06, "loss": 0.1711, "step": 5111 }, { "epoch": 1.74, "learning_rate": 2.1598499196948757e-06, "loss": 0.2009, "step": 5112 }, { "epoch": 1.74, "learning_rate": 2.1541782817869143e-06, "loss": 0.2127, "step": 5113 }, { "epoch": 1.74, "learning_rate": 2.148513765070817e-06, "loss": 0.2006, "step": 5114 }, { "epoch": 1.75, "learning_rate": 2.1428563713122545e-06, "loss": 0.2348, "step": 5115 }, { "epoch": 1.75, "learning_rate": 2.137206102274672e-06, "loss": 0.1983, "step": 5116 }, { "epoch": 1.75, "learning_rate": 2.1315629597192966e-06, "loss": 0.2138, "step": 5117 }, { "epoch": 1.75, "learning_rate": 2.1259269454051296e-06, "loss": 0.2011, "step": 5118 }, { "epoch": 1.75, "learning_rate": 2.1202980610889545e-06, "loss": 0.181, "step": 5119 }, { "epoch": 1.75, "learning_rate": 2.1146763085253373e-06, "loss": 0.1999, "step": 5120 }, { "epoch": 1.75, "learning_rate": 2.1090616894666013e-06, "loss": 0.2174, "step": 5121 }, { "epoch": 1.75, "learning_rate": 2.1034542056628804e-06, "loss": 0.1772, "step": 5122 }, { "epoch": 1.75, "learning_rate": 2.097853858862045e-06, "loss": 0.2295, "step": 5123 }, { "epoch": 1.75, "learning_rate": 2.0922606508097716e-06, "loss": 0.1793, "step": 5124 }, { "epoch": 1.75, "learning_rate": 2.086674583249493e-06, "loss": 0.2374, "step": 5125 }, { "epoch": 1.75, "learning_rate": 2.0810956579224245e-06, "loss": 0.2222, "step": 5126 }, { "epoch": 1.75, "learning_rate": 2.075523876567556e-06, "loss": 0.2057, "step": 5127 }, { "epoch": 1.75, "learning_rate": 2.0699592409216456e-06, "loss": 0.2078, "step": 5128 }, { "epoch": 1.75, "learning_rate": 2.0644017527192306e-06, "loss": 0.1579, "step": 5129 }, { "epoch": 1.75, "learning_rate": 2.0588514136926084e-06, "loss": 0.2337, "step": 5130 }, { "epoch": 1.75, "learning_rate": 2.0533082255718594e-06, "loss": 0.1836, "step": 5131 }, { "epoch": 1.75, "learning_rate": 2.04777219008484e-06, "loss": 0.1715, "step": 5132 }, { "epoch": 1.75, "learning_rate": 2.0422433089571562e-06, "loss": 0.1987, "step": 5133 }, { "epoch": 1.75, "learning_rate": 2.0367215839122045e-06, "loss": 0.1842, "step": 5134 }, { "epoch": 1.75, "learning_rate": 2.0312070166711328e-06, "loss": 0.2104, "step": 5135 }, { "epoch": 1.75, "learning_rate": 2.025699608952883e-06, "loss": 0.2084, "step": 5136 }, { "epoch": 1.75, "learning_rate": 2.020199362474137e-06, "loss": 0.1832, "step": 5137 }, { "epoch": 1.75, "learning_rate": 2.014706278949363e-06, "loss": 0.2122, "step": 5138 }, { "epoch": 1.75, "learning_rate": 2.009220360090794e-06, "loss": 0.2, "step": 5139 }, { "epoch": 1.75, "learning_rate": 2.0037416076084198e-06, "loss": 0.1981, "step": 5140 }, { "epoch": 1.75, "learning_rate": 1.998270023210011e-06, "loss": 0.1848, "step": 5141 }, { "epoch": 1.75, "learning_rate": 1.9928056086010866e-06, "loss": 0.204, "step": 5142 }, { "epoch": 1.75, "learning_rate": 1.9873483654849505e-06, "loss": 0.2219, "step": 5143 }, { "epoch": 1.76, "learning_rate": 1.981898295562651e-06, "loss": 0.1989, "step": 5144 }, { "epoch": 1.76, "learning_rate": 1.976455400533017e-06, "loss": 0.228, "step": 5145 }, { "epoch": 1.76, "learning_rate": 1.971019682092637e-06, "loss": 0.1928, "step": 5146 }, { "epoch": 1.76, "learning_rate": 1.9655911419358518e-06, "loss": 0.1958, "step": 5147 }, { "epoch": 1.76, "learning_rate": 1.9601697817547807e-06, "loss": 0.2052, "step": 5148 }, { "epoch": 1.76, "learning_rate": 1.9547556032392863e-06, "loss": 0.2032, "step": 5149 }, { "epoch": 1.76, "learning_rate": 1.949348608077017e-06, "loss": 0.1938, "step": 5150 }, { "epoch": 1.76, "learning_rate": 1.9439487979533597e-06, "loss": 0.2101, "step": 5151 }, { "epoch": 1.76, "learning_rate": 1.93855617455147e-06, "loss": 0.2122, "step": 5152 }, { "epoch": 1.76, "learning_rate": 1.933170739552273e-06, "loss": 0.1874, "step": 5153 }, { "epoch": 1.76, "learning_rate": 1.927792494634434e-06, "loss": 0.2028, "step": 5154 }, { "epoch": 1.76, "learning_rate": 1.9224214414743954e-06, "loss": 0.1972, "step": 5155 }, { "epoch": 1.76, "learning_rate": 1.91705758174634e-06, "loss": 0.1821, "step": 5156 }, { "epoch": 1.76, "learning_rate": 1.9117009171222223e-06, "loss": 0.2035, "step": 5157 }, { "epoch": 1.76, "learning_rate": 1.9063514492717538e-06, "loss": 0.1843, "step": 5158 }, { "epoch": 1.76, "learning_rate": 1.9010091798623897e-06, "loss": 0.2026, "step": 5159 }, { "epoch": 1.76, "learning_rate": 1.8956741105593617e-06, "loss": 0.1883, "step": 5160 }, { "epoch": 1.76, "learning_rate": 1.890346243025637e-06, "loss": 0.1935, "step": 5161 }, { "epoch": 1.76, "learning_rate": 1.8850255789219452e-06, "loss": 0.1744, "step": 5162 }, { "epoch": 1.76, "learning_rate": 1.879712119906779e-06, "loss": 0.2116, "step": 5163 }, { "epoch": 1.76, "learning_rate": 1.8744058676363752e-06, "loss": 0.1933, "step": 5164 }, { "epoch": 1.76, "learning_rate": 1.8691068237647296e-06, "loss": 0.1575, "step": 5165 }, { "epoch": 1.76, "learning_rate": 1.8638149899435853e-06, "loss": 0.1889, "step": 5166 }, { "epoch": 1.76, "learning_rate": 1.8585303678224447e-06, "loss": 0.2802, "step": 5167 }, { "epoch": 1.76, "learning_rate": 1.8532529590485543e-06, "loss": 0.1924, "step": 5168 }, { "epoch": 1.76, "learning_rate": 1.847982765266923e-06, "loss": 0.2141, "step": 5169 }, { "epoch": 1.76, "learning_rate": 1.8427197881202979e-06, "loss": 0.1856, "step": 5170 }, { "epoch": 1.76, "learning_rate": 1.8374640292491863e-06, "loss": 0.1965, "step": 5171 }, { "epoch": 1.76, "learning_rate": 1.8322154902918497e-06, "loss": 0.1938, "step": 5172 }, { "epoch": 1.76, "learning_rate": 1.82697417288428e-06, "loss": 0.1887, "step": 5173 }, { "epoch": 1.77, "learning_rate": 1.821740078660239e-06, "loss": 0.189, "step": 5174 }, { "epoch": 1.77, "learning_rate": 1.8165132092512194e-06, "loss": 0.1882, "step": 5175 }, { "epoch": 1.77, "learning_rate": 1.8112935662864762e-06, "loss": 0.1919, "step": 5176 }, { "epoch": 1.77, "learning_rate": 1.8060811513930077e-06, "loss": 0.2037, "step": 5177 }, { "epoch": 1.77, "learning_rate": 1.8008759661955533e-06, "loss": 0.2098, "step": 5178 }, { "epoch": 1.77, "learning_rate": 1.7956780123166123e-06, "loss": 0.1916, "step": 5179 }, { "epoch": 1.77, "learning_rate": 1.7904872913764081e-06, "loss": 0.1738, "step": 5180 }, { "epoch": 1.77, "learning_rate": 1.7853038049929328e-06, "loss": 0.1794, "step": 5181 }, { "epoch": 1.77, "learning_rate": 1.7801275547819047e-06, "loss": 0.2138, "step": 5182 }, { "epoch": 1.77, "learning_rate": 1.774958542356797e-06, "loss": 0.2059, "step": 5183 }, { "epoch": 1.77, "learning_rate": 1.7697967693288326e-06, "loss": 0.2126, "step": 5184 }, { "epoch": 1.77, "learning_rate": 1.7646422373069599e-06, "loss": 0.1901, "step": 5185 }, { "epoch": 1.77, "learning_rate": 1.7594949478978857e-06, "loss": 0.1819, "step": 5186 }, { "epoch": 1.77, "learning_rate": 1.7543549027060485e-06, "loss": 0.1993, "step": 5187 }, { "epoch": 1.77, "learning_rate": 1.7492221033336392e-06, "loss": 0.1852, "step": 5188 }, { "epoch": 1.77, "learning_rate": 1.7440965513805863e-06, "loss": 0.2095, "step": 5189 }, { "epoch": 1.77, "learning_rate": 1.738978248444545e-06, "loss": 0.2115, "step": 5190 }, { "epoch": 1.77, "learning_rate": 1.7338671961209417e-06, "loss": 0.2098, "step": 5191 }, { "epoch": 1.77, "learning_rate": 1.7287633960029136e-06, "loss": 0.1895, "step": 5192 }, { "epoch": 1.77, "learning_rate": 1.7236668496813574e-06, "loss": 0.1955, "step": 5193 }, { "epoch": 1.77, "learning_rate": 1.7185775587448855e-06, "loss": 0.2264, "step": 5194 }, { "epoch": 1.77, "learning_rate": 1.713495524779879e-06, "loss": 0.1895, "step": 5195 }, { "epoch": 1.77, "learning_rate": 1.7084207493704291e-06, "loss": 0.1992, "step": 5196 }, { "epoch": 1.77, "learning_rate": 1.7033532340983787e-06, "loss": 0.1911, "step": 5197 }, { "epoch": 1.77, "learning_rate": 1.6982929805433145e-06, "loss": 0.224, "step": 5198 }, { "epoch": 1.77, "learning_rate": 1.6932399902825386e-06, "loss": 0.1956, "step": 5199 }, { "epoch": 1.77, "learning_rate": 1.6881942648911076e-06, "loss": 0.1835, "step": 5200 }, { "epoch": 1.77, "learning_rate": 1.683155805941805e-06, "loss": 0.2053, "step": 5201 }, { "epoch": 1.77, "learning_rate": 1.6781246150051499e-06, "loss": 0.1998, "step": 5202 }, { "epoch": 1.78, "learning_rate": 1.673100693649407e-06, "loss": 0.1991, "step": 5203 }, { "epoch": 1.78, "learning_rate": 1.6680840434405459e-06, "loss": 0.2192, "step": 5204 }, { "epoch": 1.78, "learning_rate": 1.6630746659423103e-06, "loss": 0.1896, "step": 5205 }, { "epoch": 1.78, "learning_rate": 1.6580725627161452e-06, "loss": 0.1804, "step": 5206 }, { "epoch": 1.78, "learning_rate": 1.6530777353212429e-06, "loss": 0.1823, "step": 5207 }, { "epoch": 1.78, "learning_rate": 1.6480901853145158e-06, "loss": 0.1965, "step": 5208 }, { "epoch": 1.78, "learning_rate": 1.6431099142506234e-06, "loss": 0.2242, "step": 5209 }, { "epoch": 1.78, "learning_rate": 1.6381369236819516e-06, "loss": 0.1894, "step": 5210 }, { "epoch": 1.78, "learning_rate": 1.633171215158602e-06, "loss": 0.2149, "step": 5211 }, { "epoch": 1.78, "learning_rate": 1.6282127902284312e-06, "loss": 0.1992, "step": 5212 }, { "epoch": 1.78, "learning_rate": 1.6232616504370025e-06, "loss": 0.177, "step": 5213 }, { "epoch": 1.78, "learning_rate": 1.6183177973276259e-06, "loss": 0.2228, "step": 5214 }, { "epoch": 1.78, "learning_rate": 1.61338123244133e-06, "loss": 0.2058, "step": 5215 }, { "epoch": 1.78, "learning_rate": 1.6084519573168727e-06, "loss": 0.2128, "step": 5216 }, { "epoch": 1.78, "learning_rate": 1.6035299734907444e-06, "loss": 0.212, "step": 5217 }, { "epoch": 1.78, "learning_rate": 1.5986152824971512e-06, "loss": 0.2048, "step": 5218 }, { "epoch": 1.78, "learning_rate": 1.5937078858680483e-06, "loss": 0.2219, "step": 5219 }, { "epoch": 1.78, "learning_rate": 1.5888077851330923e-06, "loss": 0.1908, "step": 5220 }, { "epoch": 1.78, "learning_rate": 1.5839149818196842e-06, "loss": 0.1935, "step": 5221 }, { "epoch": 1.78, "learning_rate": 1.579029477452934e-06, "loss": 0.2171, "step": 5222 }, { "epoch": 1.78, "learning_rate": 1.5741512735556935e-06, "loss": 0.1974, "step": 5223 }, { "epoch": 1.78, "learning_rate": 1.5692803716485293e-06, "loss": 0.2074, "step": 5224 }, { "epoch": 1.78, "learning_rate": 1.5644167732497273e-06, "loss": 0.1724, "step": 5225 }, { "epoch": 1.78, "learning_rate": 1.5595604798753105e-06, "loss": 0.1855, "step": 5226 }, { "epoch": 1.78, "learning_rate": 1.5547114930390099e-06, "loss": 0.2011, "step": 5227 }, { "epoch": 1.78, "learning_rate": 1.5498698142522917e-06, "loss": 0.1673, "step": 5228 }, { "epoch": 1.78, "learning_rate": 1.5450354450243426e-06, "loss": 0.1802, "step": 5229 }, { "epoch": 1.78, "learning_rate": 1.5402083868620549e-06, "loss": 0.1898, "step": 5230 }, { "epoch": 1.78, "learning_rate": 1.5353886412700662e-06, "loss": 0.2107, "step": 5231 }, { "epoch": 1.79, "learning_rate": 1.5305762097507115e-06, "loss": 0.209, "step": 5232 }, { "epoch": 1.79, "learning_rate": 1.5257710938040682e-06, "loss": 0.1926, "step": 5233 }, { "epoch": 1.79, "learning_rate": 1.520973294927916e-06, "loss": 0.2005, "step": 5234 }, { "epoch": 1.79, "learning_rate": 1.5161828146177586e-06, "loss": 0.2018, "step": 5235 }, { "epoch": 1.79, "learning_rate": 1.5113996543668297e-06, "loss": 0.1711, "step": 5236 }, { "epoch": 1.79, "learning_rate": 1.5066238156660612e-06, "loss": 0.1967, "step": 5237 }, { "epoch": 1.79, "learning_rate": 1.501855300004118e-06, "loss": 0.1601, "step": 5238 }, { "epoch": 1.79, "learning_rate": 1.4970941088673751e-06, "loss": 0.2079, "step": 5239 }, { "epoch": 1.79, "learning_rate": 1.4923402437399287e-06, "loss": 0.2052, "step": 5240 }, { "epoch": 1.79, "learning_rate": 1.4875937061035904e-06, "loss": 0.1998, "step": 5241 }, { "epoch": 1.79, "learning_rate": 1.4828544974378828e-06, "loss": 0.2041, "step": 5242 }, { "epoch": 1.79, "learning_rate": 1.478122619220057e-06, "loss": 0.2029, "step": 5243 }, { "epoch": 1.79, "learning_rate": 1.4733980729250586e-06, "loss": 0.1881, "step": 5244 }, { "epoch": 1.79, "learning_rate": 1.4686808600255648e-06, "loss": 0.2222, "step": 5245 }, { "epoch": 1.79, "learning_rate": 1.4639709819919634e-06, "loss": 0.1763, "step": 5246 }, { "epoch": 1.79, "learning_rate": 1.4592684402923579e-06, "loss": 0.1923, "step": 5247 }, { "epoch": 1.79, "learning_rate": 1.4545732363925508e-06, "loss": 0.2089, "step": 5248 }, { "epoch": 1.79, "learning_rate": 1.449885371756074e-06, "loss": 0.1936, "step": 5249 }, { "epoch": 1.79, "learning_rate": 1.4452048478441698e-06, "loss": 0.1857, "step": 5250 }, { "epoch": 1.79, "learning_rate": 1.4405316661157798e-06, "loss": 0.2156, "step": 5251 }, { "epoch": 1.79, "learning_rate": 1.4358658280275716e-06, "loss": 0.1892, "step": 5252 }, { "epoch": 1.79, "learning_rate": 1.43120733503391e-06, "loss": 0.2369, "step": 5253 }, { "epoch": 1.79, "learning_rate": 1.4265561885868834e-06, "loss": 0.1888, "step": 5254 }, { "epoch": 1.79, "learning_rate": 1.4219123901362873e-06, "loss": 0.1821, "step": 5255 }, { "epoch": 1.79, "learning_rate": 1.4172759411296143e-06, "loss": 0.2182, "step": 5256 }, { "epoch": 1.79, "learning_rate": 1.412646843012086e-06, "loss": 0.2041, "step": 5257 }, { "epoch": 1.79, "learning_rate": 1.4080250972266173e-06, "loss": 0.1847, "step": 5258 }, { "epoch": 1.79, "learning_rate": 1.403410705213834e-06, "loss": 0.191, "step": 5259 }, { "epoch": 1.79, "learning_rate": 1.3988036684120769e-06, "loss": 0.2141, "step": 5260 }, { "epoch": 1.79, "learning_rate": 1.3942039882573894e-06, "loss": 0.202, "step": 5261 }, { "epoch": 1.8, "learning_rate": 1.389611666183524e-06, "loss": 0.205, "step": 5262 }, { "epoch": 1.8, "learning_rate": 1.3850267036219279e-06, "loss": 0.1936, "step": 5263 }, { "epoch": 1.8, "learning_rate": 1.3804491020017768e-06, "loss": 0.2153, "step": 5264 }, { "epoch": 1.8, "learning_rate": 1.375878862749927e-06, "loss": 0.1703, "step": 5265 }, { "epoch": 1.8, "learning_rate": 1.3713159872909581e-06, "loss": 0.1981, "step": 5266 }, { "epoch": 1.8, "learning_rate": 1.366760477047152e-06, "loss": 0.2051, "step": 5267 }, { "epoch": 1.8, "learning_rate": 1.3622123334384812e-06, "loss": 0.1801, "step": 5268 }, { "epoch": 1.8, "learning_rate": 1.3576715578826392e-06, "loss": 0.1919, "step": 5269 }, { "epoch": 1.8, "learning_rate": 1.353138151795011e-06, "loss": 0.1921, "step": 5270 }, { "epoch": 1.8, "learning_rate": 1.3486121165886934e-06, "loss": 0.2176, "step": 5271 }, { "epoch": 1.8, "learning_rate": 1.3440934536744803e-06, "loss": 0.2392, "step": 5272 }, { "epoch": 1.8, "learning_rate": 1.3395821644608619e-06, "loss": 0.2037, "step": 5273 }, { "epoch": 1.8, "learning_rate": 1.3350782503540437e-06, "loss": 0.2321, "step": 5274 }, { "epoch": 1.8, "learning_rate": 1.3305817127579223e-06, "loss": 0.1791, "step": 5275 }, { "epoch": 1.8, "learning_rate": 1.3260925530741015e-06, "loss": 0.1712, "step": 5276 }, { "epoch": 1.8, "learning_rate": 1.3216107727018755e-06, "loss": 0.1819, "step": 5277 }, { "epoch": 1.8, "learning_rate": 1.3171363730382524e-06, "loss": 0.1922, "step": 5278 }, { "epoch": 1.8, "learning_rate": 1.3126693554779219e-06, "loss": 0.2091, "step": 5279 }, { "epoch": 1.8, "learning_rate": 1.3082097214132894e-06, "loss": 0.1891, "step": 5280 }, { "epoch": 1.8, "learning_rate": 1.3037574722344543e-06, "loss": 0.1867, "step": 5281 }, { "epoch": 1.8, "learning_rate": 1.2993126093292007e-06, "loss": 0.1989, "step": 5282 }, { "epoch": 1.8, "learning_rate": 1.2948751340830373e-06, "loss": 0.2103, "step": 5283 }, { "epoch": 1.8, "learning_rate": 1.290445047879138e-06, "loss": 0.1871, "step": 5284 }, { "epoch": 1.8, "learning_rate": 1.2860223520983984e-06, "loss": 0.1916, "step": 5285 }, { "epoch": 1.8, "learning_rate": 1.2816070481194042e-06, "loss": 0.2103, "step": 5286 }, { "epoch": 1.8, "learning_rate": 1.2771991373184217e-06, "loss": 0.2053, "step": 5287 }, { "epoch": 1.8, "learning_rate": 1.2727986210694432e-06, "loss": 0.1546, "step": 5288 }, { "epoch": 1.8, "learning_rate": 1.2684055007441242e-06, "loss": 0.2044, "step": 5289 }, { "epoch": 1.8, "learning_rate": 1.264019777711839e-06, "loss": 0.1942, "step": 5290 }, { "epoch": 1.81, "learning_rate": 1.2596414533396383e-06, "loss": 0.2074, "step": 5291 }, { "epoch": 1.81, "learning_rate": 1.255270528992275e-06, "loss": 0.1994, "step": 5292 }, { "epoch": 1.81, "learning_rate": 1.2509070060322037e-06, "loss": 0.1848, "step": 5293 }, { "epoch": 1.81, "learning_rate": 1.246550885819553e-06, "loss": 0.2173, "step": 5294 }, { "epoch": 1.81, "learning_rate": 1.242202169712159e-06, "loss": 0.2015, "step": 5295 }, { "epoch": 1.81, "learning_rate": 1.2378608590655428e-06, "loss": 0.2062, "step": 5296 }, { "epoch": 1.81, "learning_rate": 1.2335269552329165e-06, "loss": 0.2004, "step": 5297 }, { "epoch": 1.81, "learning_rate": 1.2292004595651968e-06, "loss": 0.2192, "step": 5298 }, { "epoch": 1.81, "learning_rate": 1.2248813734109687e-06, "loss": 0.2054, "step": 5299 }, { "epoch": 1.81, "learning_rate": 1.2205696981165276e-06, "loss": 0.1891, "step": 5300 }, { "epoch": 1.81, "learning_rate": 1.2162654350258402e-06, "loss": 0.1714, "step": 5301 }, { "epoch": 1.81, "learning_rate": 1.2119685854805884e-06, "loss": 0.1988, "step": 5302 }, { "epoch": 1.81, "learning_rate": 1.207679150820118e-06, "loss": 0.1932, "step": 5303 }, { "epoch": 1.81, "learning_rate": 1.2033971323814786e-06, "loss": 0.2095, "step": 5304 }, { "epoch": 1.81, "learning_rate": 1.1991225314993998e-06, "loss": 0.2717, "step": 5305 }, { "epoch": 1.81, "learning_rate": 1.1948553495063047e-06, "loss": 0.1825, "step": 5306 }, { "epoch": 1.81, "learning_rate": 1.1905955877323016e-06, "loss": 0.1895, "step": 5307 }, { "epoch": 1.81, "learning_rate": 1.1863432475051838e-06, "loss": 0.1861, "step": 5308 }, { "epoch": 1.81, "learning_rate": 1.1820983301504413e-06, "loss": 0.2144, "step": 5309 }, { "epoch": 1.81, "learning_rate": 1.177860836991232e-06, "loss": 0.1725, "step": 5310 }, { "epoch": 1.81, "learning_rate": 1.1736307693484132e-06, "loss": 0.1809, "step": 5311 }, { "epoch": 1.81, "learning_rate": 1.1694081285405307e-06, "loss": 0.2092, "step": 5312 }, { "epoch": 1.81, "learning_rate": 1.1651929158838004e-06, "loss": 0.1937, "step": 5313 }, { "epoch": 1.81, "learning_rate": 1.1609851326921412e-06, "loss": 0.1991, "step": 5314 }, { "epoch": 1.81, "learning_rate": 1.1567847802771343e-06, "loss": 0.2362, "step": 5315 }, { "epoch": 1.81, "learning_rate": 1.1525918599480713e-06, "loss": 0.1973, "step": 5316 }, { "epoch": 1.81, "learning_rate": 1.148406373011901e-06, "loss": 0.1899, "step": 5317 }, { "epoch": 1.81, "learning_rate": 1.1442283207732746e-06, "loss": 0.1924, "step": 5318 }, { "epoch": 1.81, "learning_rate": 1.1400577045345167e-06, "loss": 0.2398, "step": 5319 }, { "epoch": 1.82, "learning_rate": 1.1358945255956322e-06, "loss": 0.189, "step": 5320 }, { "epoch": 1.82, "learning_rate": 1.131738785254316e-06, "loss": 0.1805, "step": 5321 }, { "epoch": 1.82, "learning_rate": 1.1275904848059322e-06, "loss": 0.2007, "step": 5322 }, { "epoch": 1.82, "learning_rate": 1.123449625543538e-06, "loss": 0.203, "step": 5323 }, { "epoch": 1.82, "learning_rate": 1.1193162087578702e-06, "loss": 0.22, "step": 5324 }, { "epoch": 1.82, "learning_rate": 1.1151902357373345e-06, "loss": 0.1801, "step": 5325 }, { "epoch": 1.82, "learning_rate": 1.1110717077680327e-06, "loss": 0.2012, "step": 5326 }, { "epoch": 1.82, "learning_rate": 1.106960626133724e-06, "loss": 0.202, "step": 5327 }, { "epoch": 1.82, "learning_rate": 1.1028569921158721e-06, "loss": 0.1964, "step": 5328 }, { "epoch": 1.82, "learning_rate": 1.0987608069935928e-06, "loss": 0.2044, "step": 5329 }, { "epoch": 1.82, "learning_rate": 1.0946720720437092e-06, "loss": 0.2042, "step": 5330 }, { "epoch": 1.82, "learning_rate": 1.0905907885406936e-06, "loss": 0.2342, "step": 5331 }, { "epoch": 1.82, "learning_rate": 1.0865169577567169e-06, "loss": 0.2223, "step": 5332 }, { "epoch": 1.82, "learning_rate": 1.0824505809616192e-06, "loss": 0.205, "step": 5333 }, { "epoch": 1.82, "learning_rate": 1.0783916594229115e-06, "loss": 0.1993, "step": 5334 }, { "epoch": 1.82, "learning_rate": 1.0743401944057901e-06, "loss": 0.229, "step": 5335 }, { "epoch": 1.82, "learning_rate": 1.070296187173117e-06, "loss": 0.1962, "step": 5336 }, { "epoch": 1.82, "learning_rate": 1.0662596389854423e-06, "loss": 0.1843, "step": 5337 }, { "epoch": 1.82, "learning_rate": 1.0622305511009872e-06, "loss": 0.1788, "step": 5338 }, { "epoch": 1.82, "learning_rate": 1.0582089247756332e-06, "loss": 0.2164, "step": 5339 }, { "epoch": 1.82, "learning_rate": 1.0541947612629582e-06, "loss": 0.191, "step": 5340 }, { "epoch": 1.82, "learning_rate": 1.0501880618141974e-06, "loss": 0.1773, "step": 5341 }, { "epoch": 1.82, "learning_rate": 1.0461888276782655e-06, "loss": 0.2201, "step": 5342 }, { "epoch": 1.82, "learning_rate": 1.0421970601017516e-06, "loss": 0.2187, "step": 5343 }, { "epoch": 1.82, "learning_rate": 1.038212760328916e-06, "loss": 0.2007, "step": 5344 }, { "epoch": 1.82, "learning_rate": 1.034235929601693e-06, "loss": 0.2076, "step": 5345 }, { "epoch": 1.82, "learning_rate": 1.0302665691596798e-06, "loss": 0.2136, "step": 5346 }, { "epoch": 1.82, "learning_rate": 1.0263046802401589e-06, "loss": 0.1973, "step": 5347 }, { "epoch": 1.82, "learning_rate": 1.022350264078073e-06, "loss": 0.198, "step": 5348 }, { "epoch": 1.82, "learning_rate": 1.0184033219060419e-06, "loss": 0.2157, "step": 5349 }, { "epoch": 1.83, "learning_rate": 1.0144638549543534e-06, "loss": 0.1768, "step": 5350 }, { "epoch": 1.83, "learning_rate": 1.0105318644509615e-06, "loss": 0.1769, "step": 5351 }, { "epoch": 1.83, "learning_rate": 1.006607351621497e-06, "loss": 0.1767, "step": 5352 }, { "epoch": 1.83, "learning_rate": 1.0026903176892532e-06, "loss": 0.1976, "step": 5353 }, { "epoch": 1.83, "learning_rate": 9.98780763875201e-07, "loss": 0.1996, "step": 5354 }, { "epoch": 1.83, "learning_rate": 9.948786913979658e-07, "loss": 0.1808, "step": 5355 }, { "epoch": 1.83, "learning_rate": 9.90984101473852e-07, "loss": 0.1915, "step": 5356 }, { "epoch": 1.83, "learning_rate": 9.870969953168302e-07, "loss": 0.1721, "step": 5357 }, { "epoch": 1.83, "learning_rate": 9.83217374138537e-07, "loss": 0.2012, "step": 5358 }, { "epoch": 1.83, "learning_rate": 9.793452391482771e-07, "loss": 0.1849, "step": 5359 }, { "epoch": 1.83, "learning_rate": 9.75480591553013e-07, "loss": 0.2259, "step": 5360 }, { "epoch": 1.83, "learning_rate": 9.71623432557392e-07, "loss": 0.1785, "step": 5361 }, { "epoch": 1.83, "learning_rate": 9.677737633637052e-07, "loss": 0.2109, "step": 5362 }, { "epoch": 1.83, "learning_rate": 9.639315851719232e-07, "loss": 0.1961, "step": 5363 }, { "epoch": 1.83, "learning_rate": 9.600968991796828e-07, "loss": 0.1942, "step": 5364 }, { "epoch": 1.83, "learning_rate": 9.562697065822744e-07, "loss": 0.1926, "step": 5365 }, { "epoch": 1.83, "learning_rate": 9.524500085726612e-07, "loss": 0.205, "step": 5366 }, { "epoch": 1.83, "learning_rate": 9.48637806341468e-07, "loss": 0.2107, "step": 5367 }, { "epoch": 1.83, "learning_rate": 9.448331010769806e-07, "loss": 0.1957, "step": 5368 }, { "epoch": 1.83, "learning_rate": 9.410358939651587e-07, "loss": 0.204, "step": 5369 }, { "epoch": 1.83, "learning_rate": 9.372461861896025e-07, "loss": 0.1972, "step": 5370 }, { "epoch": 1.83, "learning_rate": 9.334639789316035e-07, "loss": 0.2303, "step": 5371 }, { "epoch": 1.83, "learning_rate": 9.296892733700907e-07, "loss": 0.2322, "step": 5372 }, { "epoch": 1.83, "learning_rate": 9.259220706816701e-07, "loss": 0.1985, "step": 5373 }, { "epoch": 1.83, "learning_rate": 9.221623720405998e-07, "loss": 0.1973, "step": 5374 }, { "epoch": 1.83, "learning_rate": 9.184101786188031e-07, "loss": 0.178, "step": 5375 }, { "epoch": 1.83, "learning_rate": 9.146654915858671e-07, "loss": 0.2143, "step": 5376 }, { "epoch": 1.83, "learning_rate": 9.109283121090273e-07, "loss": 0.2043, "step": 5377 }, { "epoch": 1.83, "learning_rate": 9.071986413531963e-07, "loss": 0.1876, "step": 5378 }, { "epoch": 1.84, "learning_rate": 9.034764804809276e-07, "loss": 0.2156, "step": 5379 }, { "epoch": 1.84, "learning_rate": 8.997618306524486e-07, "loss": 0.2114, "step": 5380 }, { "epoch": 1.84, "learning_rate": 8.960546930256386e-07, "loss": 0.2102, "step": 5381 }, { "epoch": 1.84, "learning_rate": 8.923550687560372e-07, "loss": 0.168, "step": 5382 }, { "epoch": 1.84, "learning_rate": 8.886629589968415e-07, "loss": 0.2013, "step": 5383 }, { "epoch": 1.84, "learning_rate": 8.84978364898903e-07, "loss": 0.1942, "step": 5384 }, { "epoch": 1.84, "learning_rate": 8.813012876107391e-07, "loss": 0.2032, "step": 5385 }, { "epoch": 1.84, "learning_rate": 8.776317282785163e-07, "loss": 0.1832, "step": 5386 }, { "epoch": 1.84, "learning_rate": 8.739696880460613e-07, "loss": 0.207, "step": 5387 }, { "epoch": 1.84, "learning_rate": 8.7031516805485e-07, "loss": 0.2273, "step": 5388 }, { "epoch": 1.84, "learning_rate": 8.666681694440293e-07, "loss": 0.1927, "step": 5389 }, { "epoch": 1.84, "learning_rate": 8.630286933503873e-07, "loss": 0.2078, "step": 5390 }, { "epoch": 1.84, "learning_rate": 8.59396740908372e-07, "loss": 0.2143, "step": 5391 }, { "epoch": 1.84, "learning_rate": 8.557723132500916e-07, "loss": 0.2049, "step": 5392 }, { "epoch": 1.84, "learning_rate": 8.521554115052954e-07, "loss": 0.1994, "step": 5393 }, { "epoch": 1.84, "learning_rate": 8.485460368014036e-07, "loss": 0.1907, "step": 5394 }, { "epoch": 1.84, "learning_rate": 8.449441902634775e-07, "loss": 0.1867, "step": 5395 }, { "epoch": 1.84, "learning_rate": 8.413498730142383e-07, "loss": 0.2105, "step": 5396 }, { "epoch": 1.84, "learning_rate": 8.377630861740566e-07, "loss": 0.2069, "step": 5397 }, { "epoch": 1.84, "learning_rate": 8.341838308609517e-07, "loss": 0.2066, "step": 5398 }, { "epoch": 1.84, "learning_rate": 8.306121081906148e-07, "loss": 0.1869, "step": 5399 }, { "epoch": 1.84, "learning_rate": 8.270479192763636e-07, "loss": 0.1858, "step": 5400 }, { "epoch": 1.84, "learning_rate": 8.234912652291843e-07, "loss": 0.2038, "step": 5401 }, { "epoch": 1.84, "learning_rate": 8.199421471577068e-07, "loss": 0.2024, "step": 5402 }, { "epoch": 1.84, "learning_rate": 8.164005661682156e-07, "loss": 0.2129, "step": 5403 }, { "epoch": 1.84, "learning_rate": 8.12866523364647e-07, "loss": 0.2096, "step": 5404 }, { "epoch": 1.84, "learning_rate": 8.093400198485784e-07, "loss": 0.2084, "step": 5405 }, { "epoch": 1.84, "learning_rate": 8.058210567192498e-07, "loss": 0.2135, "step": 5406 }, { "epoch": 1.84, "learning_rate": 8.023096350735421e-07, "loss": 0.2, "step": 5407 }, { "epoch": 1.85, "learning_rate": 7.988057560059909e-07, "loss": 0.1987, "step": 5408 }, { "epoch": 1.85, "learning_rate": 7.95309420608778e-07, "loss": 0.2035, "step": 5409 }, { "epoch": 1.85, "learning_rate": 7.91820629971729e-07, "loss": 0.2234, "step": 5410 }, { "epoch": 1.85, "learning_rate": 7.883393851823295e-07, "loss": 0.2119, "step": 5411 }, { "epoch": 1.85, "learning_rate": 7.848656873256977e-07, "loss": 0.2209, "step": 5412 }, { "epoch": 1.85, "learning_rate": 7.8139953748462e-07, "loss": 0.1696, "step": 5413 }, { "epoch": 1.85, "learning_rate": 7.779409367395102e-07, "loss": 0.1841, "step": 5414 }, { "epoch": 1.85, "learning_rate": 7.744898861684363e-07, "loss": 0.2397, "step": 5415 }, { "epoch": 1.85, "learning_rate": 7.710463868471185e-07, "loss": 0.2153, "step": 5416 }, { "epoch": 1.85, "learning_rate": 7.676104398489093e-07, "loss": 0.1935, "step": 5417 }, { "epoch": 1.85, "learning_rate": 7.641820462448268e-07, "loss": 0.2035, "step": 5418 }, { "epoch": 1.85, "learning_rate": 7.607612071035136e-07, "loss": 0.2031, "step": 5419 }, { "epoch": 1.85, "learning_rate": 7.573479234912723e-07, "loss": 0.2083, "step": 5420 }, { "epoch": 1.85, "learning_rate": 7.539421964720461e-07, "loss": 0.18, "step": 5421 }, { "epoch": 1.85, "learning_rate": 7.505440271074221e-07, "loss": 0.2358, "step": 5422 }, { "epoch": 1.85, "learning_rate": 7.471534164566335e-07, "loss": 0.2064, "step": 5423 }, { "epoch": 1.85, "learning_rate": 7.437703655765488e-07, "loss": 0.2073, "step": 5424 }, { "epoch": 1.85, "learning_rate": 7.403948755216938e-07, "loss": 0.1994, "step": 5425 }, { "epoch": 1.85, "learning_rate": 7.370269473442299e-07, "loss": 0.1881, "step": 5426 }, { "epoch": 1.85, "learning_rate": 7.336665820939587e-07, "loss": 0.2274, "step": 5427 }, { "epoch": 1.85, "learning_rate": 7.303137808183341e-07, "loss": 0.1919, "step": 5428 }, { "epoch": 1.85, "learning_rate": 7.269685445624396e-07, "loss": 0.1777, "step": 5429 }, { "epoch": 1.85, "learning_rate": 7.236308743690162e-07, "loss": 0.2042, "step": 5430 }, { "epoch": 1.85, "learning_rate": 7.20300771278426e-07, "loss": 0.2074, "step": 5431 }, { "epoch": 1.85, "learning_rate": 7.169782363286915e-07, "loss": 0.2313, "step": 5432 }, { "epoch": 1.85, "learning_rate": 7.136632705554646e-07, "loss": 0.2173, "step": 5433 }, { "epoch": 1.85, "learning_rate": 7.103558749920442e-07, "loss": 0.1871, "step": 5434 }, { "epoch": 1.85, "learning_rate": 7.070560506693691e-07, "loss": 0.2148, "step": 5435 }, { "epoch": 1.85, "learning_rate": 7.037637986160112e-07, "loss": 0.2087, "step": 5436 }, { "epoch": 1.85, "learning_rate": 7.004791198581939e-07, "loss": 0.2221, "step": 5437 }, { "epoch": 1.86, "learning_rate": 6.972020154197623e-07, "loss": 0.194, "step": 5438 }, { "epoch": 1.86, "learning_rate": 6.939324863222213e-07, "loss": 0.2213, "step": 5439 }, { "epoch": 1.86, "learning_rate": 6.906705335846975e-07, "loss": 0.1874, "step": 5440 }, { "epoch": 1.86, "learning_rate": 6.874161582239663e-07, "loss": 0.224, "step": 5441 }, { "epoch": 1.86, "learning_rate": 6.841693612544415e-07, "loss": 0.1782, "step": 5442 }, { "epoch": 1.86, "learning_rate": 6.809301436881632e-07, "loss": 0.2054, "step": 5443 }, { "epoch": 1.86, "learning_rate": 6.77698506534824e-07, "loss": 0.2179, "step": 5444 }, { "epoch": 1.86, "learning_rate": 6.744744508017403e-07, "loss": 0.198, "step": 5445 }, { "epoch": 1.86, "learning_rate": 6.71257977493872e-07, "loss": 0.1958, "step": 5446 }, { "epoch": 1.86, "learning_rate": 6.680490876138201e-07, "loss": 0.2205, "step": 5447 }, { "epoch": 1.86, "learning_rate": 6.648477821618093e-07, "loss": 0.2241, "step": 5448 }, { "epoch": 1.86, "learning_rate": 6.616540621357165e-07, "loss": 0.2279, "step": 5449 }, { "epoch": 1.86, "learning_rate": 6.584679285310341e-07, "loss": 0.2145, "step": 5450 }, { "epoch": 1.86, "learning_rate": 6.552893823409095e-07, "loss": 0.2029, "step": 5451 }, { "epoch": 1.86, "learning_rate": 6.521184245561141e-07, "loss": 0.1898, "step": 5452 }, { "epoch": 1.86, "learning_rate": 6.489550561650514e-07, "loss": 0.1726, "step": 5453 }, { "epoch": 1.86, "learning_rate": 6.457992781537691e-07, "loss": 0.1981, "step": 5454 }, { "epoch": 1.86, "learning_rate": 6.426510915059414e-07, "loss": 0.1916, "step": 5455 }, { "epoch": 1.86, "learning_rate": 6.395104972028859e-07, "loss": 0.1768, "step": 5456 }, { "epoch": 1.86, "learning_rate": 6.363774962235336e-07, "loss": 0.1618, "step": 5457 }, { "epoch": 1.86, "learning_rate": 6.332520895444727e-07, "loss": 0.2064, "step": 5458 }, { "epoch": 1.86, "learning_rate": 6.301342781399045e-07, "loss": 0.1987, "step": 5459 }, { "epoch": 1.86, "learning_rate": 6.270240629816765e-07, "loss": 0.2214, "step": 5460 }, { "epoch": 1.86, "learning_rate": 6.239214450392633e-07, "loss": 0.1865, "step": 5461 }, { "epoch": 1.86, "learning_rate": 6.208264252797663e-07, "loss": 0.1905, "step": 5462 }, { "epoch": 1.86, "learning_rate": 6.177390046679304e-07, "loss": 0.2165, "step": 5463 }, { "epoch": 1.86, "learning_rate": 6.146591841661165e-07, "loss": 0.1704, "step": 5464 }, { "epoch": 1.86, "learning_rate": 6.115869647343287e-07, "loss": 0.2042, "step": 5465 }, { "epoch": 1.86, "learning_rate": 6.085223473302038e-07, "loss": 0.1673, "step": 5466 }, { "epoch": 1.87, "learning_rate": 6.054653329089916e-07, "loss": 0.2021, "step": 5467 }, { "epoch": 1.87, "learning_rate": 6.024159224235909e-07, "loss": 0.2164, "step": 5468 }, { "epoch": 1.87, "learning_rate": 5.99374116824522e-07, "loss": 0.1967, "step": 5469 }, { "epoch": 1.87, "learning_rate": 5.963399170599376e-07, "loss": 0.1902, "step": 5470 }, { "epoch": 1.87, "learning_rate": 5.933133240756117e-07, "loss": 0.1952, "step": 5471 }, { "epoch": 1.87, "learning_rate": 5.902943388149562e-07, "loss": 0.1988, "step": 5472 }, { "epoch": 1.87, "learning_rate": 5.872829622190101e-07, "loss": 0.199, "step": 5473 }, { "epoch": 1.87, "learning_rate": 5.842791952264365e-07, "loss": 0.1792, "step": 5474 }, { "epoch": 1.87, "learning_rate": 5.812830387735307e-07, "loss": 0.1683, "step": 5475 }, { "epoch": 1.87, "learning_rate": 5.782944937942153e-07, "loss": 0.2214, "step": 5476 }, { "epoch": 1.87, "learning_rate": 5.753135612200338e-07, "loss": 0.2014, "step": 5477 }, { "epoch": 1.87, "learning_rate": 5.72340241980171e-07, "loss": 0.1809, "step": 5478 }, { "epoch": 1.87, "learning_rate": 5.693745370014241e-07, "loss": 0.1988, "step": 5479 }, { "epoch": 1.87, "learning_rate": 5.664164472082262e-07, "loss": 0.1981, "step": 5480 }, { "epoch": 1.87, "learning_rate": 5.634659735226256e-07, "loss": 0.1898, "step": 5481 }, { "epoch": 1.87, "learning_rate": 5.605231168643144e-07, "loss": 0.2168, "step": 5482 }, { "epoch": 1.87, "learning_rate": 5.575878781505922e-07, "loss": 0.1934, "step": 5483 }, { "epoch": 1.87, "learning_rate": 5.546602582963966e-07, "loss": 0.1731, "step": 5484 }, { "epoch": 1.87, "learning_rate": 5.517402582142838e-07, "loss": 0.1925, "step": 5485 }, { "epoch": 1.87, "learning_rate": 5.488278788144396e-07, "loss": 0.205, "step": 5486 }, { "epoch": 1.87, "learning_rate": 5.459231210046711e-07, "loss": 0.1882, "step": 5487 }, { "epoch": 1.87, "learning_rate": 5.430259856904041e-07, "loss": 0.2024, "step": 5488 }, { "epoch": 1.87, "learning_rate": 5.40136473774705e-07, "loss": 0.2179, "step": 5489 }, { "epoch": 1.87, "learning_rate": 5.372545861582451e-07, "loss": 0.1755, "step": 5490 }, { "epoch": 1.87, "learning_rate": 5.343803237393281e-07, "loss": 0.1889, "step": 5491 }, { "epoch": 1.87, "learning_rate": 5.315136874138871e-07, "loss": 0.1883, "step": 5492 }, { "epoch": 1.87, "learning_rate": 5.286546780754659e-07, "loss": 0.1917, "step": 5493 }, { "epoch": 1.87, "learning_rate": 5.258032966152376e-07, "loss": 0.1901, "step": 5494 }, { "epoch": 1.87, "learning_rate": 5.229595439219914e-07, "loss": 0.1963, "step": 5495 }, { "epoch": 1.88, "learning_rate": 5.201234208821543e-07, "loss": 0.1953, "step": 5496 }, { "epoch": 1.88, "learning_rate": 5.172949283797579e-07, "loss": 0.2077, "step": 5497 }, { "epoch": 1.88, "learning_rate": 5.14474067296461e-07, "loss": 0.1828, "step": 5498 }, { "epoch": 1.88, "learning_rate": 5.11660838511549e-07, "loss": 0.1905, "step": 5499 }, { "epoch": 1.88, "learning_rate": 5.088552429019178e-07, "loss": 0.1968, "step": 5500 }, { "epoch": 1.88, "learning_rate": 5.060572813420927e-07, "loss": 0.1968, "step": 5501 }, { "epoch": 1.88, "learning_rate": 5.032669547042179e-07, "loss": 0.2014, "step": 5502 }, { "epoch": 1.88, "learning_rate": 5.004842638580532e-07, "loss": 0.2053, "step": 5503 }, { "epoch": 1.88, "learning_rate": 4.977092096709852e-07, "loss": 0.2301, "step": 5504 }, { "epoch": 1.88, "learning_rate": 4.949417930080136e-07, "loss": 0.1954, "step": 5505 }, { "epoch": 1.88, "learning_rate": 4.921820147317652e-07, "loss": 0.2175, "step": 5506 }, { "epoch": 1.88, "learning_rate": 4.894298757024712e-07, "loss": 0.1956, "step": 5507 }, { "epoch": 1.88, "learning_rate": 4.866853767780011e-07, "loss": 0.1772, "step": 5508 }, { "epoch": 1.88, "learning_rate": 4.839485188138259e-07, "loss": 0.2058, "step": 5509 }, { "epoch": 1.88, "learning_rate": 4.812193026630496e-07, "loss": 0.1643, "step": 5510 }, { "epoch": 1.88, "learning_rate": 4.784977291763775e-07, "loss": 0.1985, "step": 5511 }, { "epoch": 1.88, "learning_rate": 4.7578379920215077e-07, "loss": 0.2277, "step": 5512 }, { "epoch": 1.88, "learning_rate": 4.730775135863147e-07, "loss": 0.2059, "step": 5513 }, { "epoch": 1.88, "learning_rate": 4.703788731724362e-07, "loss": 0.2152, "step": 5514 }, { "epoch": 1.88, "learning_rate": 4.6768787880170086e-07, "loss": 0.1824, "step": 5515 }, { "epoch": 1.88, "learning_rate": 4.650045313129042e-07, "loss": 0.204, "step": 5516 }, { "epoch": 1.88, "learning_rate": 4.62328831542469e-07, "loss": 0.2056, "step": 5517 }, { "epoch": 1.88, "learning_rate": 4.5966078032442796e-07, "loss": 0.1984, "step": 5518 }, { "epoch": 1.88, "learning_rate": 4.5700037849042697e-07, "loss": 0.239, "step": 5519 }, { "epoch": 1.88, "learning_rate": 4.5434762686973333e-07, "loss": 0.21, "step": 5520 }, { "epoch": 1.88, "learning_rate": 4.517025262892244e-07, "loss": 0.2094, "step": 5521 }, { "epoch": 1.88, "learning_rate": 4.490650775733962e-07, "loss": 0.2013, "step": 5522 }, { "epoch": 1.88, "learning_rate": 4.464352815443579e-07, "loss": 0.2015, "step": 5523 }, { "epoch": 1.88, "learning_rate": 4.4381313902183694e-07, "loss": 0.1961, "step": 5524 }, { "epoch": 1.89, "learning_rate": 4.4119865082317413e-07, "loss": 0.1904, "step": 5525 }, { "epoch": 1.89, "learning_rate": 4.385918177633147e-07, "loss": 0.2015, "step": 5526 }, { "epoch": 1.89, "learning_rate": 4.359926406548337e-07, "loss": 0.1713, "step": 5527 }, { "epoch": 1.89, "learning_rate": 4.3340112030790525e-07, "loss": 0.2204, "step": 5528 }, { "epoch": 1.89, "learning_rate": 4.308172575303249e-07, "loss": 0.1657, "step": 5529 }, { "epoch": 1.89, "learning_rate": 4.2824105312750385e-07, "loss": 0.1993, "step": 5530 }, { "epoch": 1.89, "learning_rate": 4.256725079024554e-07, "loss": 0.1966, "step": 5531 }, { "epoch": 1.89, "learning_rate": 4.2311162265581684e-07, "loss": 0.1967, "step": 5532 }, { "epoch": 1.89, "learning_rate": 4.205583981858302e-07, "loss": 0.1924, "step": 5533 }, { "epoch": 1.89, "learning_rate": 4.1801283528835334e-07, "loss": 0.1718, "step": 5534 }, { "epoch": 1.89, "learning_rate": 4.154749347568543e-07, "loss": 0.1909, "step": 5535 }, { "epoch": 1.89, "learning_rate": 4.1294469738241413e-07, "loss": 0.2063, "step": 5536 }, { "epoch": 1.89, "learning_rate": 4.104221239537215e-07, "loss": 0.1932, "step": 5537 }, { "epoch": 1.89, "learning_rate": 4.0790721525708065e-07, "loss": 0.1988, "step": 5538 }, { "epoch": 1.89, "learning_rate": 4.05399972076409e-07, "loss": 0.1837, "step": 5539 }, { "epoch": 1.89, "learning_rate": 4.029003951932259e-07, "loss": 0.2055, "step": 5540 }, { "epoch": 1.89, "learning_rate": 4.004084853866663e-07, "loss": 0.1762, "step": 5541 }, { "epoch": 1.89, "learning_rate": 3.9792424343347567e-07, "loss": 0.2094, "step": 5542 }, { "epoch": 1.89, "learning_rate": 3.9544767010800954e-07, "loss": 0.1958, "step": 5543 }, { "epoch": 1.89, "learning_rate": 3.9297876618223097e-07, "loss": 0.1929, "step": 5544 }, { "epoch": 1.89, "learning_rate": 3.905175324257104e-07, "loss": 0.2143, "step": 5545 }, { "epoch": 1.89, "learning_rate": 3.880639696056371e-07, "loss": 0.2198, "step": 5546 }, { "epoch": 1.89, "learning_rate": 3.8561807848679634e-07, "loss": 0.2081, "step": 5547 }, { "epoch": 1.89, "learning_rate": 3.8317985983158946e-07, "loss": 0.19, "step": 5548 }, { "epoch": 1.89, "learning_rate": 3.807493144000307e-07, "loss": 0.2102, "step": 5549 }, { "epoch": 1.89, "learning_rate": 3.7832644294973075e-07, "loss": 0.2152, "step": 5550 }, { "epoch": 1.89, "learning_rate": 3.75911246235916e-07, "loss": 0.163, "step": 5551 }, { "epoch": 1.89, "learning_rate": 3.735037250114176e-07, "loss": 0.2107, "step": 5552 }, { "epoch": 1.89, "learning_rate": 3.711038800266825e-07, "loss": 0.185, "step": 5553 }, { "epoch": 1.89, "learning_rate": 3.687117120297484e-07, "loss": 0.1745, "step": 5554 }, { "epoch": 1.9, "learning_rate": 3.663272217662772e-07, "loss": 0.2122, "step": 5555 }, { "epoch": 1.9, "learning_rate": 3.6395040997952714e-07, "loss": 0.1965, "step": 5556 }, { "epoch": 1.9, "learning_rate": 3.615812774103666e-07, "loss": 0.1847, "step": 5557 }, { "epoch": 1.9, "learning_rate": 3.592198247972689e-07, "loss": 0.1997, "step": 5558 }, { "epoch": 1.9, "learning_rate": 3.568660528763146e-07, "loss": 0.2153, "step": 5559 }, { "epoch": 1.9, "learning_rate": 3.54519962381189e-07, "loss": 0.197, "step": 5560 }, { "epoch": 1.9, "learning_rate": 3.5218155404318496e-07, "loss": 0.186, "step": 5561 }, { "epoch": 1.9, "learning_rate": 3.498508285912e-07, "loss": 0.1985, "step": 5562 }, { "epoch": 1.9, "learning_rate": 3.4752778675173624e-07, "loss": 0.1919, "step": 5563 }, { "epoch": 1.9, "learning_rate": 3.4521242924889785e-07, "loss": 0.1971, "step": 5564 }, { "epoch": 1.9, "learning_rate": 3.429047568044019e-07, "loss": 0.1943, "step": 5565 }, { "epoch": 1.9, "learning_rate": 3.4060477013756197e-07, "loss": 0.1922, "step": 5566 }, { "epoch": 1.9, "learning_rate": 3.3831246996530175e-07, "loss": 0.1974, "step": 5567 }, { "epoch": 1.9, "learning_rate": 3.3602785700214133e-07, "loss": 0.1881, "step": 5568 }, { "epoch": 1.9, "learning_rate": 3.3375093196021113e-07, "loss": 0.2071, "step": 5569 }, { "epoch": 1.9, "learning_rate": 3.3148169554924614e-07, "loss": 0.2004, "step": 5570 }, { "epoch": 1.9, "learning_rate": 3.2922014847658053e-07, "loss": 0.2066, "step": 5571 }, { "epoch": 1.9, "learning_rate": 3.2696629144715607e-07, "loss": 0.187, "step": 5572 }, { "epoch": 1.9, "learning_rate": 3.247201251635079e-07, "loss": 0.1933, "step": 5573 }, { "epoch": 1.9, "learning_rate": 3.2248165032578715e-07, "loss": 0.1783, "step": 5574 }, { "epoch": 1.9, "learning_rate": 3.202508676317384e-07, "loss": 0.1912, "step": 5575 }, { "epoch": 1.9, "learning_rate": 3.18027777776711e-07, "loss": 0.1918, "step": 5576 }, { "epoch": 1.9, "learning_rate": 3.15812381453659e-07, "loss": 0.2002, "step": 5577 }, { "epoch": 1.9, "learning_rate": 3.1360467935313e-07, "loss": 0.1991, "step": 5578 }, { "epoch": 1.9, "learning_rate": 3.114046721632874e-07, "loss": 0.2023, "step": 5579 }, { "epoch": 1.9, "learning_rate": 3.092123605698827e-07, "loss": 0.206, "step": 5580 }, { "epoch": 1.9, "learning_rate": 3.07027745256272e-07, "loss": 0.2019, "step": 5581 }, { "epoch": 1.9, "learning_rate": 3.048508269034217e-07, "loss": 0.173, "step": 5582 }, { "epoch": 1.9, "learning_rate": 3.026816061898835e-07, "loss": 0.2288, "step": 5583 }, { "epoch": 1.91, "learning_rate": 3.0052008379182205e-07, "loss": 0.2082, "step": 5584 }, { "epoch": 1.91, "learning_rate": 2.9836626038299566e-07, "loss": 0.1869, "step": 5585 }, { "epoch": 1.91, "learning_rate": 2.9622013663476443e-07, "loss": 0.2179, "step": 5586 }, { "epoch": 1.91, "learning_rate": 2.940817132160961e-07, "loss": 0.188, "step": 5587 }, { "epoch": 1.91, "learning_rate": 2.919509907935408e-07, "loss": 0.179, "step": 5588 }, { "epoch": 1.91, "learning_rate": 2.8982797003126726e-07, "loss": 0.1887, "step": 5589 }, { "epoch": 1.91, "learning_rate": 2.877126515910294e-07, "loss": 0.2128, "step": 5590 }, { "epoch": 1.91, "learning_rate": 2.856050361321916e-07, "loss": 0.1866, "step": 5591 }, { "epoch": 1.91, "learning_rate": 2.8350512431170317e-07, "loss": 0.1643, "step": 5592 }, { "epoch": 1.91, "learning_rate": 2.8141291678412674e-07, "loss": 0.2188, "step": 5593 }, { "epoch": 1.91, "learning_rate": 2.793284142016156e-07, "loss": 0.1759, "step": 5594 }, { "epoch": 1.91, "learning_rate": 2.772516172139222e-07, "loss": 0.2056, "step": 5595 }, { "epoch": 1.91, "learning_rate": 2.7518252646839827e-07, "loss": 0.2072, "step": 5596 }, { "epoch": 1.91, "learning_rate": 2.731211426099917e-07, "loss": 0.1989, "step": 5597 }, { "epoch": 1.91, "learning_rate": 2.7106746628125234e-07, "loss": 0.1862, "step": 5598 }, { "epoch": 1.91, "learning_rate": 2.6902149812231827e-07, "loss": 0.2267, "step": 5599 }, { "epoch": 1.91, "learning_rate": 2.6698323877093754e-07, "loss": 0.2061, "step": 5600 }, { "epoch": 1.91, "learning_rate": 2.649526888624493e-07, "loss": 0.2111, "step": 5601 }, { "epoch": 1.91, "learning_rate": 2.6292984902978323e-07, "loss": 0.225, "step": 5602 }, { "epoch": 1.91, "learning_rate": 2.609147199034767e-07, "loss": 0.1933, "step": 5603 }, { "epoch": 1.91, "learning_rate": 2.58907302111655e-07, "loss": 0.2007, "step": 5604 }, { "epoch": 1.91, "learning_rate": 2.5690759628004545e-07, "loss": 0.1981, "step": 5605 }, { "epoch": 1.91, "learning_rate": 2.5491560303197157e-07, "loss": 0.2082, "step": 5606 }, { "epoch": 1.91, "learning_rate": 2.529313229883451e-07, "loss": 0.2179, "step": 5607 }, { "epoch": 1.91, "learning_rate": 2.5095475676768507e-07, "loss": 0.1923, "step": 5608 }, { "epoch": 1.91, "learning_rate": 2.48985904986096e-07, "loss": 0.2015, "step": 5609 }, { "epoch": 1.91, "learning_rate": 2.470247682572868e-07, "loss": 0.2017, "step": 5610 }, { "epoch": 1.91, "learning_rate": 2.450713471925492e-07, "loss": 0.1898, "step": 5611 }, { "epoch": 1.91, "learning_rate": 2.4312564240078216e-07, "loss": 0.2109, "step": 5612 }, { "epoch": 1.92, "learning_rate": 2.411876544884728e-07, "loss": 0.2291, "step": 5613 }, { "epoch": 1.92, "learning_rate": 2.392573840597073e-07, "loss": 0.2056, "step": 5614 }, { "epoch": 1.92, "learning_rate": 2.3733483171616e-07, "loss": 0.1877, "step": 5615 }, { "epoch": 1.92, "learning_rate": 2.35419998057107e-07, "loss": 0.2096, "step": 5616 }, { "epoch": 1.92, "learning_rate": 2.3351288367940982e-07, "loss": 0.2012, "step": 5617 }, { "epoch": 1.92, "learning_rate": 2.3161348917753178e-07, "loss": 0.1964, "step": 5618 }, { "epoch": 1.92, "learning_rate": 2.2972181514352432e-07, "loss": 0.2142, "step": 5619 }, { "epoch": 1.92, "learning_rate": 2.2783786216703517e-07, "loss": 0.2367, "step": 5620 }, { "epoch": 1.92, "learning_rate": 2.2596163083530575e-07, "loss": 0.1637, "step": 5621 }, { "epoch": 1.92, "learning_rate": 2.2409312173317098e-07, "loss": 0.1839, "step": 5622 }, { "epoch": 1.92, "learning_rate": 2.222323354430539e-07, "loss": 0.1986, "step": 5623 }, { "epoch": 1.92, "learning_rate": 2.2037927254497936e-07, "loss": 0.2213, "step": 5624 }, { "epoch": 1.92, "learning_rate": 2.1853393361655206e-07, "loss": 0.2075, "step": 5625 }, { "epoch": 1.92, "learning_rate": 2.1669631923297852e-07, "loss": 0.1861, "step": 5626 }, { "epoch": 1.92, "learning_rate": 2.1486642996705885e-07, "loss": 0.237, "step": 5627 }, { "epoch": 1.92, "learning_rate": 2.1304426638917574e-07, "loss": 0.2062, "step": 5628 }, { "epoch": 1.92, "learning_rate": 2.1122982906731647e-07, "loss": 0.2136, "step": 5629 }, { "epoch": 1.92, "learning_rate": 2.0942311856704532e-07, "loss": 0.2055, "step": 5630 }, { "epoch": 1.92, "learning_rate": 2.0762413545153126e-07, "loss": 0.2185, "step": 5631 }, { "epoch": 1.92, "learning_rate": 2.0583288028152848e-07, "loss": 0.1983, "step": 5632 }, { "epoch": 1.92, "learning_rate": 2.0404935361537924e-07, "loss": 0.2014, "step": 5633 }, { "epoch": 1.92, "learning_rate": 2.022735560090222e-07, "loss": 0.1726, "step": 5634 }, { "epoch": 1.92, "learning_rate": 2.005054880159868e-07, "loss": 0.1856, "step": 5635 }, { "epoch": 1.92, "learning_rate": 1.9874515018739327e-07, "loss": 0.2006, "step": 5636 }, { "epoch": 1.92, "learning_rate": 1.9699254307194436e-07, "loss": 0.2232, "step": 5637 }, { "epoch": 1.92, "learning_rate": 1.95247667215942e-07, "loss": 0.1882, "step": 5638 }, { "epoch": 1.92, "learning_rate": 1.935105231632789e-07, "loss": 0.2277, "step": 5639 }, { "epoch": 1.92, "learning_rate": 1.9178111145542743e-07, "loss": 0.203, "step": 5640 }, { "epoch": 1.92, "learning_rate": 1.9005943263146198e-07, "loss": 0.215, "step": 5641 }, { "epoch": 1.92, "learning_rate": 1.8834548722803935e-07, "loss": 0.1927, "step": 5642 }, { "epoch": 1.93, "learning_rate": 1.866392757794072e-07, "loss": 0.1919, "step": 5643 }, { "epoch": 1.93, "learning_rate": 1.8494079881740122e-07, "loss": 0.1679, "step": 5644 }, { "epoch": 1.93, "learning_rate": 1.832500568714507e-07, "loss": 0.1851, "step": 5645 }, { "epoch": 1.93, "learning_rate": 1.8156705046857293e-07, "loss": 0.1776, "step": 5646 }, { "epoch": 1.93, "learning_rate": 1.7989178013336494e-07, "loss": 0.198, "step": 5647 }, { "epoch": 1.93, "learning_rate": 1.7822424638802292e-07, "loss": 0.2285, "step": 5648 }, { "epoch": 1.93, "learning_rate": 1.7656444975232834e-07, "loss": 0.2015, "step": 5649 }, { "epoch": 1.93, "learning_rate": 1.749123907436534e-07, "loss": 0.2078, "step": 5650 }, { "epoch": 1.93, "learning_rate": 1.7326806987695288e-07, "loss": 0.209, "step": 5651 }, { "epoch": 1.93, "learning_rate": 1.7163148766476957e-07, "loss": 0.2012, "step": 5652 }, { "epoch": 1.93, "learning_rate": 1.7000264461724257e-07, "loss": 0.2048, "step": 5653 }, { "epoch": 1.93, "learning_rate": 1.6838154124208806e-07, "loss": 0.2011, "step": 5654 }, { "epoch": 1.93, "learning_rate": 1.667681780446212e-07, "loss": 0.2178, "step": 5655 }, { "epoch": 1.93, "learning_rate": 1.6516255552772864e-07, "loss": 0.186, "step": 5656 }, { "epoch": 1.93, "learning_rate": 1.6356467419189892e-07, "loss": 0.2055, "step": 5657 }, { "epoch": 1.93, "learning_rate": 1.6197453453520028e-07, "loss": 0.1954, "step": 5658 }, { "epoch": 1.93, "learning_rate": 1.603921370532918e-07, "loss": 0.2058, "step": 5659 }, { "epoch": 1.93, "learning_rate": 1.5881748223941507e-07, "loss": 0.2186, "step": 5660 }, { "epoch": 1.93, "learning_rate": 1.572505705844024e-07, "loss": 0.2105, "step": 5661 }, { "epoch": 1.93, "learning_rate": 1.556914025766687e-07, "loss": 0.2088, "step": 5662 }, { "epoch": 1.93, "learning_rate": 1.5413997870221409e-07, "loss": 0.2069, "step": 5663 }, { "epoch": 1.93, "learning_rate": 1.5259629944463226e-07, "loss": 0.1967, "step": 5664 }, { "epoch": 1.93, "learning_rate": 1.510603652850995e-07, "loss": 0.199, "step": 5665 }, { "epoch": 1.93, "learning_rate": 1.4953217670237163e-07, "loss": 0.2181, "step": 5666 }, { "epoch": 1.93, "learning_rate": 1.4801173417279547e-07, "loss": 0.1743, "step": 5667 }, { "epoch": 1.93, "learning_rate": 1.4649903817030574e-07, "loss": 0.19, "step": 5668 }, { "epoch": 1.93, "learning_rate": 1.44994089166417e-07, "loss": 0.2001, "step": 5669 }, { "epoch": 1.93, "learning_rate": 1.434968876302345e-07, "loss": 0.2019, "step": 5670 }, { "epoch": 1.93, "learning_rate": 1.420074340284433e-07, "loss": 0.2061, "step": 5671 }, { "epoch": 1.94, "learning_rate": 1.4052572882531644e-07, "loss": 0.2179, "step": 5672 }, { "epoch": 1.94, "learning_rate": 1.3905177248271228e-07, "loss": 0.2009, "step": 5673 }, { "epoch": 1.94, "learning_rate": 1.3758556546007162e-07, "loss": 0.1867, "step": 5674 }, { "epoch": 1.94, "learning_rate": 1.3612710821441777e-07, "loss": 0.1959, "step": 5675 }, { "epoch": 1.94, "learning_rate": 1.346764012003676e-07, "loss": 0.2036, "step": 5676 }, { "epoch": 1.94, "learning_rate": 1.3323344487010947e-07, "loss": 0.1875, "step": 5677 }, { "epoch": 1.94, "learning_rate": 1.3179823967342798e-07, "loss": 0.2191, "step": 5678 }, { "epoch": 1.94, "learning_rate": 1.303707860576847e-07, "loss": 0.1882, "step": 5679 }, { "epoch": 1.94, "learning_rate": 1.2895108446782378e-07, "loss": 0.2092, "step": 5680 }, { "epoch": 1.94, "learning_rate": 1.275391353463773e-07, "loss": 0.2102, "step": 5681 }, { "epoch": 1.94, "learning_rate": 1.2613493913345708e-07, "loss": 0.1924, "step": 5682 }, { "epoch": 1.94, "learning_rate": 1.2473849626676304e-07, "loss": 0.2358, "step": 5683 }, { "epoch": 1.94, "learning_rate": 1.2334980718157474e-07, "loss": 0.2468, "step": 5684 }, { "epoch": 1.94, "learning_rate": 1.2196887231075427e-07, "loss": 0.218, "step": 5685 }, { "epoch": 1.94, "learning_rate": 1.2059569208475173e-07, "loss": 0.229, "step": 5686 }, { "epoch": 1.94, "learning_rate": 1.1923026693158867e-07, "loss": 0.2418, "step": 5687 }, { "epoch": 1.94, "learning_rate": 1.1787259727688571e-07, "loss": 0.2114, "step": 5688 }, { "epoch": 1.94, "learning_rate": 1.1652268354383211e-07, "loss": 0.1743, "step": 5689 }, { "epoch": 1.94, "learning_rate": 1.1518052615320518e-07, "loss": 0.212, "step": 5690 }, { "epoch": 1.94, "learning_rate": 1.1384612552336749e-07, "loss": 0.2144, "step": 5691 }, { "epoch": 1.94, "learning_rate": 1.1251948207025575e-07, "loss": 0.2172, "step": 5692 }, { "epoch": 1.94, "learning_rate": 1.1120059620739753e-07, "loss": 0.177, "step": 5693 }, { "epoch": 1.94, "learning_rate": 1.0988946834589454e-07, "loss": 0.1961, "step": 5694 }, { "epoch": 1.94, "learning_rate": 1.0858609889443927e-07, "loss": 0.1868, "step": 5695 }, { "epoch": 1.94, "learning_rate": 1.0729048825929566e-07, "loss": 0.1937, "step": 5696 }, { "epoch": 1.94, "learning_rate": 1.0600263684431289e-07, "loss": 0.1941, "step": 5697 }, { "epoch": 1.94, "learning_rate": 1.0472254505092816e-07, "loss": 0.1892, "step": 5698 }, { "epoch": 1.94, "learning_rate": 1.0345021327815008e-07, "loss": 0.1827, "step": 5699 }, { "epoch": 1.94, "learning_rate": 1.021856419225753e-07, "loss": 0.2054, "step": 5700 }, { "epoch": 1.95, "learning_rate": 1.009288313783746e-07, "loss": 0.2222, "step": 5701 }, { "epoch": 1.95, "learning_rate": 9.967978203730965e-08, "loss": 0.2366, "step": 5702 }, { "epoch": 1.95, "learning_rate": 9.843849428871343e-08, "loss": 0.1987, "step": 5703 }, { "epoch": 1.95, "learning_rate": 9.7204968519507e-08, "loss": 0.2251, "step": 5704 }, { "epoch": 1.95, "learning_rate": 9.597920511418556e-08, "loss": 0.2154, "step": 5705 }, { "epoch": 1.95, "learning_rate": 9.476120445482683e-08, "loss": 0.1842, "step": 5706 }, { "epoch": 1.95, "learning_rate": 9.355096692109377e-08, "loss": 0.2084, "step": 5707 }, { "epoch": 1.95, "learning_rate": 9.234849289022073e-08, "loss": 0.1895, "step": 5708 }, { "epoch": 1.95, "learning_rate": 9.115378273702735e-08, "loss": 0.1867, "step": 5709 }, { "epoch": 1.95, "learning_rate": 8.996683683391572e-08, "loss": 0.1794, "step": 5710 }, { "epoch": 1.95, "learning_rate": 8.87876555508621e-08, "loss": 0.185, "step": 5711 }, { "epoch": 1.95, "learning_rate": 8.761623925542528e-08, "loss": 0.2042, "step": 5712 }, { "epoch": 1.95, "learning_rate": 8.645258831274372e-08, "loss": 0.2182, "step": 5713 }, { "epoch": 1.95, "learning_rate": 8.529670308553562e-08, "loss": 0.189, "step": 5714 }, { "epoch": 1.95, "learning_rate": 8.414858393409886e-08, "loss": 0.2032, "step": 5715 }, { "epoch": 1.95, "learning_rate": 8.300823121630552e-08, "loss": 0.206, "step": 5716 }, { "epoch": 1.95, "learning_rate": 8.187564528761293e-08, "loss": 0.2022, "step": 5717 }, { "epoch": 1.95, "learning_rate": 8.075082650105814e-08, "loss": 0.1852, "step": 5718 }, { "epoch": 1.95, "learning_rate": 7.963377520725235e-08, "loss": 0.1827, "step": 5719 }, { "epoch": 1.95, "learning_rate": 7.852449175438925e-08, "loss": 0.1942, "step": 5720 }, { "epoch": 1.95, "learning_rate": 7.742297648823671e-08, "loss": 0.1812, "step": 5721 }, { "epoch": 1.95, "learning_rate": 7.632922975214784e-08, "loss": 0.209, "step": 5722 }, { "epoch": 1.95, "learning_rate": 7.524325188704717e-08, "loss": 0.2212, "step": 5723 }, { "epoch": 1.95, "learning_rate": 7.41650432314417e-08, "loss": 0.1985, "step": 5724 }, { "epoch": 1.95, "learning_rate": 7.309460412141811e-08, "loss": 0.1753, "step": 5725 }, { "epoch": 1.95, "learning_rate": 7.203193489064009e-08, "loss": 0.2008, "step": 5726 }, { "epoch": 1.95, "learning_rate": 7.097703587034543e-08, "loss": 0.193, "step": 5727 }, { "epoch": 1.95, "learning_rate": 6.992990738935446e-08, "loss": 0.2031, "step": 5728 }, { "epoch": 1.95, "learning_rate": 6.889054977406439e-08, "loss": 0.2094, "step": 5729 }, { "epoch": 1.95, "learning_rate": 6.785896334844665e-08, "loss": 0.1883, "step": 5730 }, { "epoch": 1.96, "learning_rate": 6.683514843405791e-08, "loss": 0.2034, "step": 5731 }, { "epoch": 1.96, "learning_rate": 6.581910535002623e-08, "loss": 0.2067, "step": 5732 }, { "epoch": 1.96, "learning_rate": 6.481083441305935e-08, "loss": 0.1922, "step": 5733 }, { "epoch": 1.96, "learning_rate": 6.381033593744201e-08, "loss": 0.2366, "step": 5734 }, { "epoch": 1.96, "learning_rate": 6.281761023503308e-08, "loss": 0.2181, "step": 5735 }, { "epoch": 1.96, "learning_rate": 6.183265761527668e-08, "loss": 0.1866, "step": 5736 }, { "epoch": 1.96, "learning_rate": 6.085547838518557e-08, "loss": 0.1943, "step": 5737 }, { "epoch": 1.96, "learning_rate": 5.988607284935221e-08, "loss": 0.1888, "step": 5738 }, { "epoch": 1.96, "learning_rate": 5.892444130995156e-08, "loss": 0.2047, "step": 5739 }, { "epoch": 1.96, "learning_rate": 5.79705840667244e-08, "loss": 0.1817, "step": 5740 }, { "epoch": 1.96, "learning_rate": 5.702450141699678e-08, "loss": 0.2188, "step": 5741 }, { "epoch": 1.96, "learning_rate": 5.6086193655668917e-08, "loss": 0.1887, "step": 5742 }, { "epoch": 1.96, "learning_rate": 5.515566107521797e-08, "loss": 0.2093, "step": 5743 }, { "epoch": 1.96, "learning_rate": 5.423290396569525e-08, "loss": 0.1949, "step": 5744 }, { "epoch": 1.96, "learning_rate": 5.331792261473456e-08, "loss": 0.2358, "step": 5745 }, { "epoch": 1.96, "learning_rate": 5.241071730753555e-08, "loss": 0.1763, "step": 5746 }, { "epoch": 1.96, "learning_rate": 5.151128832688312e-08, "loss": 0.178, "step": 5747 }, { "epoch": 1.96, "learning_rate": 5.061963595313635e-08, "loss": 0.22, "step": 5748 }, { "epoch": 1.96, "learning_rate": 4.973576046422568e-08, "loss": 0.2019, "step": 5749 }, { "epoch": 1.96, "learning_rate": 4.885966213566684e-08, "loss": 0.2189, "step": 5750 }, { "epoch": 1.96, "learning_rate": 4.799134124053861e-08, "loss": 0.1885, "step": 5751 }, { "epoch": 1.96, "learning_rate": 4.713079804950504e-08, "loss": 0.1911, "step": 5752 }, { "epoch": 1.96, "learning_rate": 4.627803283080434e-08, "loss": 0.2208, "step": 5753 }, { "epoch": 1.96, "learning_rate": 4.5433045850246106e-08, "loss": 0.1771, "step": 5754 }, { "epoch": 1.96, "learning_rate": 4.4595837371222415e-08, "loss": 0.1864, "step": 5755 }, { "epoch": 1.96, "learning_rate": 4.376640765469398e-08, "loss": 0.1946, "step": 5756 }, { "epoch": 1.96, "learning_rate": 4.294475695919842e-08, "loss": 0.1869, "step": 5757 }, { "epoch": 1.96, "learning_rate": 4.213088554085309e-08, "loss": 0.1737, "step": 5758 }, { "epoch": 1.96, "learning_rate": 4.1324793653341187e-08, "loss": 0.199, "step": 5759 }, { "epoch": 1.97, "learning_rate": 4.052648154793393e-08, "loss": 0.1851, "step": 5760 }, { "epoch": 1.97, "learning_rate": 3.9735949473465596e-08, "loss": 0.2227, "step": 5761 }, { "epoch": 1.97, "learning_rate": 3.8953197676350195e-08, "loss": 0.1874, "step": 5762 }, { "epoch": 1.97, "learning_rate": 3.817822640058144e-08, "loss": 0.2026, "step": 5763 }, { "epoch": 1.97, "learning_rate": 3.74110358877161e-08, "loss": 0.2035, "step": 5764 }, { "epoch": 1.97, "learning_rate": 3.6651626376899006e-08, "loss": 0.1897, "step": 5765 }, { "epoch": 1.97, "learning_rate": 3.589999810483802e-08, "loss": 0.2212, "step": 5766 }, { "epoch": 1.97, "learning_rate": 3.515615130582073e-08, "loss": 0.1629, "step": 5767 }, { "epoch": 1.97, "learning_rate": 3.4420086211714443e-08, "loss": 0.2195, "step": 5768 }, { "epoch": 1.97, "learning_rate": 3.3691803051949525e-08, "loss": 0.1969, "step": 5769 }, { "epoch": 1.97, "learning_rate": 3.2971302053538823e-08, "loss": 0.1907, "step": 5770 }, { "epoch": 1.97, "learning_rate": 3.225858344106936e-08, "loss": 0.1827, "step": 5771 }, { "epoch": 1.97, "learning_rate": 3.155364743669675e-08, "loss": 0.1781, "step": 5772 }, { "epoch": 1.97, "learning_rate": 3.085649426015358e-08, "loss": 0.1861, "step": 5773 }, { "epoch": 1.97, "learning_rate": 3.016712412875211e-08, "loss": 0.167, "step": 5774 }, { "epoch": 1.97, "learning_rate": 2.948553725736769e-08, "loss": 0.2217, "step": 5775 }, { "epoch": 1.97, "learning_rate": 2.8811733858460922e-08, "loss": 0.2249, "step": 5776 }, { "epoch": 1.97, "learning_rate": 2.814571414205547e-08, "loss": 0.199, "step": 5777 }, { "epoch": 1.97, "learning_rate": 2.7487478315757488e-08, "loss": 0.2043, "step": 5778 }, { "epoch": 1.97, "learning_rate": 2.6837026584741743e-08, "loss": 0.2908, "step": 5779 }, { "epoch": 1.97, "learning_rate": 2.619435915175994e-08, "loss": 0.2092, "step": 5780 }, { "epoch": 1.97, "learning_rate": 2.5559476217132393e-08, "loss": 0.1903, "step": 5781 }, { "epoch": 1.97, "learning_rate": 2.4932377978761912e-08, "loss": 0.1738, "step": 5782 }, { "epoch": 1.97, "learning_rate": 2.431306463211436e-08, "loss": 0.201, "step": 5783 }, { "epoch": 1.97, "learning_rate": 2.3701536370232536e-08, "loss": 0.1948, "step": 5784 }, { "epoch": 1.97, "learning_rate": 2.3097793383738962e-08, "loss": 0.2032, "step": 5785 }, { "epoch": 1.97, "learning_rate": 2.2501835860821997e-08, "loss": 0.216, "step": 5786 }, { "epoch": 1.97, "learning_rate": 2.191366398724415e-08, "loss": 0.1894, "step": 5787 }, { "epoch": 1.97, "learning_rate": 2.13332779463421e-08, "loss": 0.1956, "step": 5788 }, { "epoch": 1.98, "learning_rate": 2.076067791902947e-08, "loss": 0.2059, "step": 5789 }, { "epoch": 1.98, "learning_rate": 2.019586408378571e-08, "loss": 0.202, "step": 5790 }, { "epoch": 1.98, "learning_rate": 1.9638836616669988e-08, "loss": 0.202, "step": 5791 }, { "epoch": 1.98, "learning_rate": 1.908959569131008e-08, "loss": 0.1903, "step": 5792 }, { "epoch": 1.98, "learning_rate": 1.8548141478907932e-08, "loss": 0.1901, "step": 5793 }, { "epoch": 1.98, "learning_rate": 1.801447414823687e-08, "loss": 0.2198, "step": 5794 }, { "epoch": 1.98, "learning_rate": 1.7488593865647163e-08, "loss": 0.1771, "step": 5795 }, { "epoch": 1.98, "learning_rate": 1.6970500795057685e-08, "loss": 0.2108, "step": 5796 }, { "epoch": 1.98, "learning_rate": 1.6460195097958708e-08, "loss": 0.2076, "step": 5797 }, { "epoch": 1.98, "learning_rate": 1.595767693342298e-08, "loss": 0.1855, "step": 5798 }, { "epoch": 1.98, "learning_rate": 1.5462946458083548e-08, "loss": 0.2009, "step": 5799 }, { "epoch": 1.98, "learning_rate": 1.4976003826150388e-08, "loss": 0.19, "step": 5800 }, { "epoch": 1.98, "learning_rate": 1.4496849189410411e-08, "loss": 0.2025, "step": 5801 }, { "epoch": 1.98, "learning_rate": 1.4025482697216375e-08, "loss": 0.1857, "step": 5802 }, { "epoch": 1.98, "learning_rate": 1.3561904496495193e-08, "loss": 0.1908, "step": 5803 }, { "epoch": 1.98, "learning_rate": 1.3106114731750718e-08, "loss": 0.2049, "step": 5804 }, { "epoch": 1.98, "learning_rate": 1.2658113545055416e-08, "loss": 0.2222, "step": 5805 }, { "epoch": 1.98, "learning_rate": 1.2217901076050364e-08, "loss": 0.1908, "step": 5806 }, { "epoch": 1.98, "learning_rate": 1.1785477461956352e-08, "loss": 0.1852, "step": 5807 }, { "epoch": 1.98, "learning_rate": 1.1360842837562779e-08, "loss": 0.1892, "step": 5808 }, { "epoch": 1.98, "learning_rate": 1.0943997335227663e-08, "loss": 0.1756, "step": 5809 }, { "epoch": 1.98, "learning_rate": 1.0534941084885952e-08, "loss": 0.1845, "step": 5810 }, { "epoch": 1.98, "learning_rate": 1.0133674214043987e-08, "loss": 0.214, "step": 5811 }, { "epoch": 1.98, "learning_rate": 9.740196847779492e-09, "loss": 0.2155, "step": 5812 }, { "epoch": 1.98, "learning_rate": 9.354509108741583e-09, "loss": 0.1857, "step": 5813 }, { "epoch": 1.98, "learning_rate": 8.976611117153532e-09, "loss": 0.2049, "step": 5814 }, { "epoch": 1.98, "learning_rate": 8.60650299080168e-09, "loss": 0.1897, "step": 5815 }, { "epoch": 1.98, "learning_rate": 8.2441848450604e-09, "loss": 0.2168, "step": 5816 }, { "epoch": 1.98, "learning_rate": 7.889656792858802e-09, "loss": 0.2032, "step": 5817 }, { "epoch": 1.98, "learning_rate": 7.542918944711264e-09, "loss": 0.2001, "step": 5818 }, { "epoch": 1.99, "learning_rate": 7.20397140869522e-09, "loss": 0.1982, "step": 5819 }, { "epoch": 1.99, "learning_rate": 6.872814290465046e-09, "loss": 0.1858, "step": 5820 }, { "epoch": 1.99, "learning_rate": 6.549447693240951e-09, "loss": 0.1903, "step": 5821 }, { "epoch": 1.99, "learning_rate": 6.233871717822859e-09, "loss": 0.183, "step": 5822 }, { "epoch": 1.99, "learning_rate": 5.926086462573755e-09, "loss": 0.212, "step": 5823 }, { "epoch": 1.99, "learning_rate": 5.62609202343356e-09, "loss": 0.2218, "step": 5824 }, { "epoch": 1.99, "learning_rate": 5.333888493916361e-09, "loss": 0.2188, "step": 5825 }, { "epoch": 1.99, "learning_rate": 5.049475965096528e-09, "loss": 0.1888, "step": 5826 }, { "epoch": 1.99, "learning_rate": 4.772854525633697e-09, "loss": 0.1977, "step": 5827 }, { "epoch": 1.99, "learning_rate": 4.504024261750561e-09, "loss": 0.2133, "step": 5828 }, { "epoch": 1.99, "learning_rate": 4.242985257241205e-09, "loss": 0.1875, "step": 5829 }, { "epoch": 1.99, "learning_rate": 3.98973759347665e-09, "loss": 0.2196, "step": 5830 }, { "epoch": 1.99, "learning_rate": 3.7442813493937525e-09, "loss": 0.2298, "step": 5831 }, { "epoch": 1.99, "learning_rate": 3.5066166015035317e-09, "loss": 0.1786, "step": 5832 }, { "epoch": 1.99, "learning_rate": 3.2767434238883953e-09, "loss": 0.1917, "step": 5833 }, { "epoch": 1.99, "learning_rate": 3.054661888196586e-09, "loss": 0.3939, "step": 5834 }, { "epoch": 1.99, "learning_rate": 2.840372063658836e-09, "loss": 0.2102, "step": 5835 }, { "epoch": 1.99, "learning_rate": 2.633874017068938e-09, "loss": 0.187, "step": 5836 }, { "epoch": 1.99, "learning_rate": 2.4351678127892964e-09, "loss": 0.2074, "step": 5837 }, { "epoch": 1.99, "learning_rate": 2.2442535127648046e-09, "loss": 0.1909, "step": 5838 }, { "epoch": 1.99, "learning_rate": 2.0611311765006416e-09, "loss": 0.1871, "step": 5839 }, { "epoch": 1.99, "learning_rate": 1.885800861076148e-09, "loss": 0.1926, "step": 5840 }, { "epoch": 1.99, "learning_rate": 1.7182626211448282e-09, "loss": 0.1993, "step": 5841 }, { "epoch": 1.99, "learning_rate": 1.5585165089287979e-09, "loss": 0.2263, "step": 5842 }, { "epoch": 1.99, "learning_rate": 1.406562574224335e-09, "loss": 0.1815, "step": 5843 }, { "epoch": 1.99, "learning_rate": 1.2624008643935537e-09, "loss": 0.2173, "step": 5844 }, { "epoch": 1.99, "learning_rate": 1.126031424372731e-09, "loss": 0.1833, "step": 5845 }, { "epoch": 1.99, "learning_rate": 9.97454296669531e-10, "loss": 0.2005, "step": 5846 }, { "epoch": 1.99, "learning_rate": 8.766695213657805e-10, "loss": 0.1769, "step": 5847 }, { "epoch": 2.0, "learning_rate": 7.63677136103591e-10, "loss": 0.1884, "step": 5848 }, { "epoch": 2.0, "learning_rate": 6.584771761103392e-10, "loss": 0.2114, "step": 5849 }, { "epoch": 2.0, "learning_rate": 5.610696741736865e-10, "loss": 0.1711, "step": 5850 }, { "epoch": 2.0, "learning_rate": 4.714546606582326e-10, "loss": 0.1924, "step": 5851 }, { "epoch": 2.0, "learning_rate": 3.896321634971889e-10, "loss": 0.1858, "step": 5852 }, { "epoch": 2.0, "learning_rate": 3.156022081923782e-10, "loss": 0.1967, "step": 5853 }, { "epoch": 2.0, "learning_rate": 2.493648178253372e-10, "loss": 0.1942, "step": 5854 }, { "epoch": 2.0, "learning_rate": 1.9092001303511188e-10, "loss": 0.1908, "step": 5855 }, { "epoch": 2.0, "learning_rate": 1.4026781204601324e-10, "loss": 0.2095, "step": 5856 }, { "epoch": 2.0, "learning_rate": 9.740823064263715e-11, "loss": 0.1726, "step": 5857 }, { "epoch": 2.0, "learning_rate": 6.234128218374214e-11, "loss": 0.1983, "step": 5858 }, { "epoch": 2.0, "learning_rate": 3.5066977605024975e-11, "loss": 0.1943, "step": 5859 }, { "epoch": 2.0, "learning_rate": 1.5585325405242933e-11, "loss": 0.1994, "step": 5860 }, { "epoch": 2.0, "learning_rate": 3.896331654540397e-12, "loss": 0.1922, "step": 5861 }, { "epoch": 2.0, "learning_rate": 0.0, "loss": 0.1528, "step": 5862 }, { "epoch": 2.0, "eval_loss": 0.3155210018157959, "eval_runtime": 77.0758, "eval_samples_per_second": 9.886, "eval_steps_per_second": 0.623, "step": 5862 }, { "epoch": 2.0, "step": 5862, "total_flos": 1.553226491795551e+19, "train_loss": 0.28463143815169845, "train_runtime": 161659.4686, "train_samples_per_second": 2.32, "train_steps_per_second": 0.036 } ], "logging_steps": 1.0, "max_steps": 5862, "num_train_epochs": 2, "save_steps": 500, "total_flos": 1.553226491795551e+19, "trial_name": null, "trial_params": null }