{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.2, "global_step": 6000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.0000000000000001e-07, "loss": 0.9461, "step": 1 }, { "epoch": 0.0, "learning_rate": 2.0000000000000002e-07, "loss": 1.3496, "step": 2 }, { "epoch": 0.0, "learning_rate": 3.0000000000000004e-07, "loss": 1.2774, "step": 3 }, { "epoch": 0.0, "learning_rate": 4.0000000000000003e-07, "loss": 1.211, "step": 4 }, { "epoch": 0.0, "learning_rate": 5.000000000000001e-07, "loss": 1.2436, "step": 5 }, { "epoch": 0.0, "learning_rate": 6.000000000000001e-07, "loss": 1.0027, "step": 6 }, { "epoch": 0.0, "learning_rate": 7.000000000000001e-07, "loss": 1.3211, "step": 7 }, { "epoch": 0.0, "learning_rate": 8.000000000000001e-07, "loss": 0.4273, "step": 8 }, { "epoch": 0.0, "learning_rate": 9e-07, "loss": 1.3863, "step": 9 }, { "epoch": 0.0, "learning_rate": 1.0000000000000002e-06, "loss": 1.2353, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.1e-06, "loss": 1.3583, "step": 11 }, { "epoch": 0.0, "learning_rate": 1.2000000000000002e-06, "loss": 1.1541, "step": 12 }, { "epoch": 0.0, "learning_rate": 1.3e-06, "loss": 1.1323, "step": 13 }, { "epoch": 0.0, "learning_rate": 1.4000000000000001e-06, "loss": 1.2773, "step": 14 }, { "epoch": 0.0, "learning_rate": 1.5e-06, "loss": 1.1155, "step": 15 }, { "epoch": 0.0, "learning_rate": 1.6000000000000001e-06, "loss": 1.141, "step": 16 }, { "epoch": 0.0, "learning_rate": 1.7000000000000002e-06, "loss": 0.9867, "step": 17 }, { "epoch": 0.0, "learning_rate": 1.8e-06, "loss": 1.6636, "step": 18 }, { "epoch": 0.0, "learning_rate": 1.9e-06, "loss": 1.1943, "step": 19 }, { "epoch": 0.0, "learning_rate": 2.0000000000000003e-06, "loss": 0.995, "step": 20 }, { "epoch": 0.0, "learning_rate": 2.1000000000000002e-06, "loss": 1.1643, "step": 21 }, { "epoch": 0.0, "learning_rate": 2.2e-06, "loss": 1.1868, "step": 22 }, { "epoch": 0.0, "learning_rate": 2.3e-06, "loss": 1.3742, "step": 23 }, { "epoch": 0.0, "learning_rate": 2.4000000000000003e-06, "loss": 0.7619, "step": 24 }, { "epoch": 0.0, "learning_rate": 2.5e-06, "loss": 1.2239, "step": 25 }, { "epoch": 0.0, "learning_rate": 2.6e-06, "loss": 1.1577, "step": 26 }, { "epoch": 0.0, "learning_rate": 2.7e-06, "loss": 1.687, "step": 27 }, { "epoch": 0.0, "learning_rate": 2.8000000000000003e-06, "loss": 1.4415, "step": 28 }, { "epoch": 0.0, "learning_rate": 2.9e-06, "loss": 1.2303, "step": 29 }, { "epoch": 0.0, "learning_rate": 3e-06, "loss": 1.3985, "step": 30 }, { "epoch": 0.0, "learning_rate": 3.1e-06, "loss": 1.4026, "step": 31 }, { "epoch": 0.0, "learning_rate": 3.2000000000000003e-06, "loss": 3.4687, "step": 32 }, { "epoch": 0.0, "learning_rate": 3.3e-06, "loss": 1.0276, "step": 33 }, { "epoch": 0.0, "learning_rate": 3.4000000000000005e-06, "loss": 0.8491, "step": 34 }, { "epoch": 0.0, "learning_rate": 3.5000000000000004e-06, "loss": 1.4081, "step": 35 }, { "epoch": 0.0, "learning_rate": 3.6e-06, "loss": 1.7093, "step": 36 }, { "epoch": 0.0, "learning_rate": 3.7e-06, "loss": 1.8678, "step": 37 }, { "epoch": 0.0, "learning_rate": 3.8e-06, "loss": 1.4196, "step": 38 }, { "epoch": 0.0, "learning_rate": 3.9e-06, "loss": 0.9082, "step": 39 }, { "epoch": 0.0, "learning_rate": 4.000000000000001e-06, "loss": 0.4231, "step": 40 }, { "epoch": 0.0, "learning_rate": 4.1000000000000006e-06, "loss": 0.4003, "step": 41 }, { "epoch": 0.0, "learning_rate": 4.2000000000000004e-06, "loss": 0.41, "step": 42 }, { "epoch": 0.0, "learning_rate": 4.2999999999999995e-06, "loss": 0.5899, "step": 43 }, { "epoch": 0.0, "learning_rate": 4.4e-06, "loss": 1.2554, "step": 44 }, { "epoch": 0.0, "learning_rate": 4.5e-06, "loss": 0.9803, "step": 45 }, { "epoch": 0.0, "learning_rate": 4.6e-06, "loss": 1.2746, "step": 46 }, { "epoch": 0.0, "learning_rate": 4.7e-06, "loss": 0.9028, "step": 47 }, { "epoch": 0.0, "learning_rate": 4.800000000000001e-06, "loss": 0.9819, "step": 48 }, { "epoch": 0.0, "learning_rate": 4.9000000000000005e-06, "loss": 1.1507, "step": 49 }, { "epoch": 0.0, "learning_rate": 5e-06, "loss": 1.1253, "step": 50 }, { "epoch": 0.0, "learning_rate": 5.1e-06, "loss": 0.8535, "step": 51 }, { "epoch": 0.0, "learning_rate": 5.2e-06, "loss": 1.1735, "step": 52 }, { "epoch": 0.0, "learning_rate": 5.3e-06, "loss": 1.2691, "step": 53 }, { "epoch": 0.0, "learning_rate": 5.4e-06, "loss": 1.4009, "step": 54 }, { "epoch": 0.0, "learning_rate": 5.500000000000001e-06, "loss": 1.1556, "step": 55 }, { "epoch": 0.0, "learning_rate": 5.600000000000001e-06, "loss": 1.648, "step": 56 }, { "epoch": 0.0, "learning_rate": 5.7000000000000005e-06, "loss": 1.5436, "step": 57 }, { "epoch": 0.0, "learning_rate": 5.8e-06, "loss": 1.4983, "step": 58 }, { "epoch": 0.0, "learning_rate": 5.9e-06, "loss": 1.3034, "step": 59 }, { "epoch": 0.0, "learning_rate": 6e-06, "loss": 1.4251, "step": 60 }, { "epoch": 0.0, "learning_rate": 6.1e-06, "loss": 1.2505, "step": 61 }, { "epoch": 0.0, "learning_rate": 6.2e-06, "loss": 1.1089, "step": 62 }, { "epoch": 0.0, "learning_rate": 6.300000000000001e-06, "loss": 0.9032, "step": 63 }, { "epoch": 0.0, "learning_rate": 6.4000000000000006e-06, "loss": 0.7938, "step": 64 }, { "epoch": 0.0, "learning_rate": 6.5000000000000004e-06, "loss": 0.9209, "step": 65 }, { "epoch": 0.0, "learning_rate": 6.6e-06, "loss": 1.0905, "step": 66 }, { "epoch": 0.0, "learning_rate": 6.700000000000001e-06, "loss": 0.8349, "step": 67 }, { "epoch": 0.0, "learning_rate": 6.800000000000001e-06, "loss": 1.1338, "step": 68 }, { "epoch": 0.0, "learning_rate": 6.900000000000001e-06, "loss": 1.1966, "step": 69 }, { "epoch": 0.0, "learning_rate": 7.000000000000001e-06, "loss": 0.8467, "step": 70 }, { "epoch": 0.0, "learning_rate": 7.1e-06, "loss": 0.5181, "step": 71 }, { "epoch": 0.0, "learning_rate": 7.2e-06, "loss": 1.1727, "step": 72 }, { "epoch": 0.0, "learning_rate": 7.2999999999999996e-06, "loss": 0.6945, "step": 73 }, { "epoch": 0.0, "learning_rate": 7.4e-06, "loss": 1.0493, "step": 74 }, { "epoch": 0.0, "learning_rate": 7.5e-06, "loss": 0.1662, "step": 75 }, { "epoch": 0.0, "learning_rate": 7.6e-06, "loss": 0.5925, "step": 76 }, { "epoch": 0.0, "learning_rate": 7.7e-06, "loss": 0.978, "step": 77 }, { "epoch": 0.0, "learning_rate": 7.8e-06, "loss": 1.0219, "step": 78 }, { "epoch": 0.0, "learning_rate": 7.9e-06, "loss": 1.1421, "step": 79 }, { "epoch": 0.0, "learning_rate": 8.000000000000001e-06, "loss": 1.3981, "step": 80 }, { "epoch": 0.0, "learning_rate": 8.1e-06, "loss": 1.1028, "step": 81 }, { "epoch": 0.0, "learning_rate": 8.200000000000001e-06, "loss": 1.1964, "step": 82 }, { "epoch": 0.0, "learning_rate": 8.3e-06, "loss": 1.3686, "step": 83 }, { "epoch": 0.0, "learning_rate": 8.400000000000001e-06, "loss": 0.824, "step": 84 }, { "epoch": 0.0, "learning_rate": 8.500000000000002e-06, "loss": 0.8732, "step": 85 }, { "epoch": 0.0, "learning_rate": 8.599999999999999e-06, "loss": 1.2033, "step": 86 }, { "epoch": 0.0, "learning_rate": 8.7e-06, "loss": 1.2988, "step": 87 }, { "epoch": 0.0, "learning_rate": 8.8e-06, "loss": 1.0759, "step": 88 }, { "epoch": 0.0, "learning_rate": 8.9e-06, "loss": 1.4702, "step": 89 }, { "epoch": 0.0, "learning_rate": 9e-06, "loss": 1.2807, "step": 90 }, { "epoch": 0.0, "learning_rate": 9.100000000000001e-06, "loss": 1.0049, "step": 91 }, { "epoch": 0.0, "learning_rate": 9.2e-06, "loss": 0.889, "step": 92 }, { "epoch": 0.0, "learning_rate": 9.3e-06, "loss": 0.9553, "step": 93 }, { "epoch": 0.0, "learning_rate": 9.4e-06, "loss": 0.7751, "step": 94 }, { "epoch": 0.0, "learning_rate": 9.5e-06, "loss": 1.2485, "step": 95 }, { "epoch": 0.0, "learning_rate": 9.600000000000001e-06, "loss": 1.1222, "step": 96 }, { "epoch": 0.0, "learning_rate": 9.7e-06, "loss": 1.1102, "step": 97 }, { "epoch": 0.0, "learning_rate": 9.800000000000001e-06, "loss": 2.5687, "step": 98 }, { "epoch": 0.0, "learning_rate": 9.900000000000002e-06, "loss": 0.95, "step": 99 }, { "epoch": 0.0, "learning_rate": 1e-05, "loss": 0.8134, "step": 100 }, { "epoch": 0.0, "learning_rate": 1.0100000000000002e-05, "loss": 1.0486, "step": 101 }, { "epoch": 0.0, "learning_rate": 1.02e-05, "loss": 1.2589, "step": 102 }, { "epoch": 0.0, "learning_rate": 1.03e-05, "loss": 1.0614, "step": 103 }, { "epoch": 0.0, "learning_rate": 1.04e-05, "loss": 1.0108, "step": 104 }, { "epoch": 0.0, "learning_rate": 1.05e-05, "loss": 1.0774, "step": 105 }, { "epoch": 0.0, "learning_rate": 1.06e-05, "loss": 1.0711, "step": 106 }, { "epoch": 0.0, "learning_rate": 1.0700000000000001e-05, "loss": 0.9267, "step": 107 }, { "epoch": 0.0, "learning_rate": 1.08e-05, "loss": 1.3796, "step": 108 }, { "epoch": 0.0, "learning_rate": 1.09e-05, "loss": 1.5741, "step": 109 }, { "epoch": 0.0, "learning_rate": 1.1000000000000001e-05, "loss": 1.6481, "step": 110 }, { "epoch": 0.0, "learning_rate": 1.11e-05, "loss": 1.8009, "step": 111 }, { "epoch": 0.0, "learning_rate": 1.1200000000000001e-05, "loss": 1.7468, "step": 112 }, { "epoch": 0.0, "learning_rate": 1.13e-05, "loss": 1.5331, "step": 113 }, { "epoch": 0.0, "learning_rate": 1.1400000000000001e-05, "loss": 0.946, "step": 114 }, { "epoch": 0.0, "learning_rate": 1.1500000000000002e-05, "loss": 1.1601, "step": 115 }, { "epoch": 0.0, "learning_rate": 1.16e-05, "loss": 1.0723, "step": 116 }, { "epoch": 0.0, "learning_rate": 1.1700000000000001e-05, "loss": 1.2539, "step": 117 }, { "epoch": 0.0, "learning_rate": 1.18e-05, "loss": 1.1103, "step": 118 }, { "epoch": 0.0, "learning_rate": 1.19e-05, "loss": 1.0899, "step": 119 }, { "epoch": 0.0, "learning_rate": 1.2e-05, "loss": 1.0965, "step": 120 }, { "epoch": 0.0, "learning_rate": 1.2100000000000001e-05, "loss": 1.1505, "step": 121 }, { "epoch": 0.0, "learning_rate": 1.22e-05, "loss": 1.2067, "step": 122 }, { "epoch": 0.0, "learning_rate": 1.23e-05, "loss": 1.09, "step": 123 }, { "epoch": 0.0, "learning_rate": 1.24e-05, "loss": 1.0543, "step": 124 }, { "epoch": 0.0, "learning_rate": 1.25e-05, "loss": 1.011, "step": 125 }, { "epoch": 0.0, "learning_rate": 1.2600000000000001e-05, "loss": 5.1819, "step": 126 }, { "epoch": 0.0, "learning_rate": 1.27e-05, "loss": 5.789, "step": 127 }, { "epoch": 0.0, "learning_rate": 1.2800000000000001e-05, "loss": 3.6985, "step": 128 }, { "epoch": 0.0, "learning_rate": 1.29e-05, "loss": 1.0669, "step": 129 }, { "epoch": 0.0, "learning_rate": 1.3000000000000001e-05, "loss": 1.2551, "step": 130 }, { "epoch": 0.0, "learning_rate": 1.3100000000000002e-05, "loss": 1.293, "step": 131 }, { "epoch": 0.0, "learning_rate": 1.32e-05, "loss": 1.2676, "step": 132 }, { "epoch": 0.0, "learning_rate": 1.3300000000000001e-05, "loss": 0.9782, "step": 133 }, { "epoch": 0.0, "learning_rate": 1.3400000000000002e-05, "loss": 0.8504, "step": 134 }, { "epoch": 0.0, "learning_rate": 1.3500000000000001e-05, "loss": 1.1576, "step": 135 }, { "epoch": 0.0, "learning_rate": 1.3600000000000002e-05, "loss": 0.6864, "step": 136 }, { "epoch": 0.0, "learning_rate": 1.3700000000000001e-05, "loss": 0.4582, "step": 137 }, { "epoch": 0.0, "learning_rate": 1.3800000000000002e-05, "loss": 0.9545, "step": 138 }, { "epoch": 0.0, "learning_rate": 1.3900000000000002e-05, "loss": 0.917, "step": 139 }, { "epoch": 0.0, "learning_rate": 1.4000000000000001e-05, "loss": 0.9662, "step": 140 }, { "epoch": 0.0, "learning_rate": 1.4099999999999999e-05, "loss": 1.1957, "step": 141 }, { "epoch": 0.0, "learning_rate": 1.42e-05, "loss": 1.1352, "step": 142 }, { "epoch": 0.0, "learning_rate": 1.43e-05, "loss": 0.831, "step": 143 }, { "epoch": 0.0, "learning_rate": 1.44e-05, "loss": 0.9509, "step": 144 }, { "epoch": 0.0, "learning_rate": 1.45e-05, "loss": 0.9987, "step": 145 }, { "epoch": 0.0, "learning_rate": 1.4599999999999999e-05, "loss": 0.3846, "step": 146 }, { "epoch": 0.0, "learning_rate": 1.47e-05, "loss": 1.2029, "step": 147 }, { "epoch": 0.0, "learning_rate": 1.48e-05, "loss": 1.3474, "step": 148 }, { "epoch": 0.0, "learning_rate": 1.49e-05, "loss": 1.1899, "step": 149 }, { "epoch": 0.01, "learning_rate": 1.5e-05, "loss": 0.9888, "step": 150 }, { "epoch": 0.01, "learning_rate": 1.51e-05, "loss": 1.0562, "step": 151 }, { "epoch": 0.01, "learning_rate": 1.52e-05, "loss": 1.1173, "step": 152 }, { "epoch": 0.01, "learning_rate": 1.53e-05, "loss": 1.1626, "step": 153 }, { "epoch": 0.01, "learning_rate": 1.54e-05, "loss": 1.1179, "step": 154 }, { "epoch": 0.01, "learning_rate": 1.55e-05, "loss": 1.0846, "step": 155 }, { "epoch": 0.01, "learning_rate": 1.56e-05, "loss": 0.901, "step": 156 }, { "epoch": 0.01, "learning_rate": 1.5700000000000002e-05, "loss": 1.1554, "step": 157 }, { "epoch": 0.01, "learning_rate": 1.58e-05, "loss": 1.0619, "step": 158 }, { "epoch": 0.01, "learning_rate": 1.59e-05, "loss": 0.9347, "step": 159 }, { "epoch": 0.01, "learning_rate": 1.6000000000000003e-05, "loss": 1.2387, "step": 160 }, { "epoch": 0.01, "learning_rate": 1.6100000000000002e-05, "loss": 1.0537, "step": 161 }, { "epoch": 0.01, "learning_rate": 1.62e-05, "loss": 2.1056, "step": 162 }, { "epoch": 0.01, "learning_rate": 1.63e-05, "loss": 5.7752, "step": 163 }, { "epoch": 0.01, "learning_rate": 1.6400000000000002e-05, "loss": 1.2195, "step": 164 }, { "epoch": 0.01, "learning_rate": 1.65e-05, "loss": 0.7435, "step": 165 }, { "epoch": 0.01, "learning_rate": 1.66e-05, "loss": 0.9002, "step": 166 }, { "epoch": 0.01, "learning_rate": 1.6700000000000003e-05, "loss": 1.0278, "step": 167 }, { "epoch": 0.01, "learning_rate": 1.6800000000000002e-05, "loss": 0.9595, "step": 168 }, { "epoch": 0.01, "learning_rate": 1.69e-05, "loss": 0.8287, "step": 169 }, { "epoch": 0.01, "learning_rate": 1.7000000000000003e-05, "loss": 0.8583, "step": 170 }, { "epoch": 0.01, "learning_rate": 1.7100000000000002e-05, "loss": 0.8835, "step": 171 }, { "epoch": 0.01, "learning_rate": 1.7199999999999998e-05, "loss": 1.2537, "step": 172 }, { "epoch": 0.01, "learning_rate": 1.73e-05, "loss": 1.0748, "step": 173 }, { "epoch": 0.01, "learning_rate": 1.74e-05, "loss": 0.662, "step": 174 }, { "epoch": 0.01, "learning_rate": 1.75e-05, "loss": 0.9929, "step": 175 }, { "epoch": 0.01, "learning_rate": 1.76e-05, "loss": 0.9605, "step": 176 }, { "epoch": 0.01, "learning_rate": 1.77e-05, "loss": 1.0392, "step": 177 }, { "epoch": 0.01, "learning_rate": 1.78e-05, "loss": 0.9988, "step": 178 }, { "epoch": 0.01, "learning_rate": 1.79e-05, "loss": 0.6803, "step": 179 }, { "epoch": 0.01, "learning_rate": 1.8e-05, "loss": 1.1256, "step": 180 }, { "epoch": 0.01, "learning_rate": 1.81e-05, "loss": 0.952, "step": 181 }, { "epoch": 0.01, "learning_rate": 1.8200000000000002e-05, "loss": 1.104, "step": 182 }, { "epoch": 0.01, "learning_rate": 1.83e-05, "loss": 1.0527, "step": 183 }, { "epoch": 0.01, "learning_rate": 1.84e-05, "loss": 1.1046, "step": 184 }, { "epoch": 0.01, "learning_rate": 1.85e-05, "loss": 1.2225, "step": 185 }, { "epoch": 0.01, "learning_rate": 1.86e-05, "loss": 1.1119, "step": 186 }, { "epoch": 0.01, "learning_rate": 1.87e-05, "loss": 1.0216, "step": 187 }, { "epoch": 0.01, "learning_rate": 1.88e-05, "loss": 0.7849, "step": 188 }, { "epoch": 0.01, "learning_rate": 1.8900000000000002e-05, "loss": 0.9815, "step": 189 }, { "epoch": 0.01, "learning_rate": 1.9e-05, "loss": 1.1248, "step": 190 }, { "epoch": 0.01, "learning_rate": 1.91e-05, "loss": 1.0008, "step": 191 }, { "epoch": 0.01, "learning_rate": 1.9200000000000003e-05, "loss": 0.6773, "step": 192 }, { "epoch": 0.01, "learning_rate": 1.93e-05, "loss": 0.9639, "step": 193 }, { "epoch": 0.01, "learning_rate": 1.94e-05, "loss": 1.1191, "step": 194 }, { "epoch": 0.01, "learning_rate": 1.9500000000000003e-05, "loss": 1.1947, "step": 195 }, { "epoch": 0.01, "learning_rate": 1.9600000000000002e-05, "loss": 1.0382, "step": 196 }, { "epoch": 0.01, "learning_rate": 1.97e-05, "loss": 1.1452, "step": 197 }, { "epoch": 0.01, "learning_rate": 1.9800000000000004e-05, "loss": 0.9233, "step": 198 }, { "epoch": 0.01, "learning_rate": 1.9900000000000003e-05, "loss": 2.6423, "step": 199 }, { "epoch": 0.01, "learning_rate": 2e-05, "loss": 5.1126, "step": 200 }, { "epoch": 0.01, "learning_rate": 2.01e-05, "loss": 1.1374, "step": 201 }, { "epoch": 0.01, "learning_rate": 2.0200000000000003e-05, "loss": 1.0686, "step": 202 }, { "epoch": 0.01, "learning_rate": 2.0300000000000002e-05, "loss": 1.1784, "step": 203 }, { "epoch": 0.01, "learning_rate": 2.04e-05, "loss": 1.0791, "step": 204 }, { "epoch": 0.01, "learning_rate": 2.05e-05, "loss": 1.1932, "step": 205 }, { "epoch": 0.01, "learning_rate": 2.06e-05, "loss": 0.9416, "step": 206 }, { "epoch": 0.01, "learning_rate": 2.07e-05, "loss": 0.8702, "step": 207 }, { "epoch": 0.01, "learning_rate": 2.08e-05, "loss": 1.0199, "step": 208 }, { "epoch": 0.01, "learning_rate": 2.09e-05, "loss": 1.117, "step": 209 }, { "epoch": 0.01, "learning_rate": 2.1e-05, "loss": 1.144, "step": 210 }, { "epoch": 0.01, "learning_rate": 2.11e-05, "loss": 0.9199, "step": 211 }, { "epoch": 0.01, "learning_rate": 2.12e-05, "loss": 1.1141, "step": 212 }, { "epoch": 0.01, "learning_rate": 2.13e-05, "loss": 0.9948, "step": 213 }, { "epoch": 0.01, "learning_rate": 2.1400000000000002e-05, "loss": 0.1665, "step": 214 }, { "epoch": 0.01, "learning_rate": 2.15e-05, "loss": 0.1837, "step": 215 }, { "epoch": 0.01, "learning_rate": 2.16e-05, "loss": 1.1209, "step": 216 }, { "epoch": 0.01, "learning_rate": 2.1700000000000002e-05, "loss": 1.0737, "step": 217 }, { "epoch": 0.01, "learning_rate": 2.18e-05, "loss": 0.9554, "step": 218 }, { "epoch": 0.01, "learning_rate": 2.19e-05, "loss": 1.002, "step": 219 }, { "epoch": 0.01, "learning_rate": 2.2000000000000003e-05, "loss": 1.1532, "step": 220 }, { "epoch": 0.01, "learning_rate": 2.2100000000000002e-05, "loss": 1.1194, "step": 221 }, { "epoch": 0.01, "learning_rate": 2.22e-05, "loss": 1.1584, "step": 222 }, { "epoch": 0.01, "learning_rate": 2.23e-05, "loss": 0.9822, "step": 223 }, { "epoch": 0.01, "learning_rate": 2.2400000000000002e-05, "loss": 1.1193, "step": 224 }, { "epoch": 0.01, "learning_rate": 2.25e-05, "loss": 1.1607, "step": 225 }, { "epoch": 0.01, "learning_rate": 2.26e-05, "loss": 1.1488, "step": 226 }, { "epoch": 0.01, "learning_rate": 2.2700000000000003e-05, "loss": 1.0595, "step": 227 }, { "epoch": 0.01, "learning_rate": 2.2800000000000002e-05, "loss": 0.9694, "step": 228 }, { "epoch": 0.01, "learning_rate": 2.29e-05, "loss": 1.6577, "step": 229 }, { "epoch": 0.01, "learning_rate": 2.3000000000000003e-05, "loss": 3.5556, "step": 230 }, { "epoch": 0.01, "learning_rate": 2.3100000000000002e-05, "loss": 0.7221, "step": 231 }, { "epoch": 0.01, "learning_rate": 2.32e-05, "loss": 0.6124, "step": 232 }, { "epoch": 0.01, "learning_rate": 2.3300000000000004e-05, "loss": 0.9817, "step": 233 }, { "epoch": 0.01, "learning_rate": 2.3400000000000003e-05, "loss": 0.7419, "step": 234 }, { "epoch": 0.01, "learning_rate": 2.35e-05, "loss": 0.67, "step": 235 }, { "epoch": 0.01, "learning_rate": 2.36e-05, "loss": 0.7989, "step": 236 }, { "epoch": 0.01, "learning_rate": 2.37e-05, "loss": 0.7062, "step": 237 }, { "epoch": 0.01, "learning_rate": 2.38e-05, "loss": 0.95, "step": 238 }, { "epoch": 0.01, "learning_rate": 2.39e-05, "loss": 0.7981, "step": 239 }, { "epoch": 0.01, "learning_rate": 2.4e-05, "loss": 0.8443, "step": 240 }, { "epoch": 0.01, "learning_rate": 2.41e-05, "loss": 0.0885, "step": 241 }, { "epoch": 0.01, "learning_rate": 2.4200000000000002e-05, "loss": 0.4862, "step": 242 }, { "epoch": 0.01, "learning_rate": 2.43e-05, "loss": 0.6152, "step": 243 }, { "epoch": 0.01, "learning_rate": 2.44e-05, "loss": 0.5029, "step": 244 }, { "epoch": 0.01, "learning_rate": 2.45e-05, "loss": 1.0336, "step": 245 }, { "epoch": 0.01, "learning_rate": 2.46e-05, "loss": 1.0236, "step": 246 }, { "epoch": 0.01, "learning_rate": 2.47e-05, "loss": 4.8409, "step": 247 }, { "epoch": 0.01, "learning_rate": 2.48e-05, "loss": 0.9615, "step": 248 }, { "epoch": 0.01, "learning_rate": 2.4900000000000002e-05, "loss": 0.8824, "step": 249 }, { "epoch": 0.01, "learning_rate": 2.5e-05, "loss": 0.8217, "step": 250 }, { "epoch": 0.01, "learning_rate": 2.51e-05, "loss": 1.0676, "step": 251 }, { "epoch": 0.01, "learning_rate": 2.5200000000000003e-05, "loss": 0.8521, "step": 252 }, { "epoch": 0.01, "learning_rate": 2.5300000000000002e-05, "loss": 1.1088, "step": 253 }, { "epoch": 0.01, "learning_rate": 2.54e-05, "loss": 1.2951, "step": 254 }, { "epoch": 0.01, "learning_rate": 2.5500000000000003e-05, "loss": 0.9594, "step": 255 }, { "epoch": 0.01, "learning_rate": 2.5600000000000002e-05, "loss": 0.3187, "step": 256 }, { "epoch": 0.01, "learning_rate": 2.57e-05, "loss": 1.0238, "step": 257 }, { "epoch": 0.01, "learning_rate": 2.58e-05, "loss": 1.1193, "step": 258 }, { "epoch": 0.01, "learning_rate": 2.5900000000000003e-05, "loss": 1.0977, "step": 259 }, { "epoch": 0.01, "learning_rate": 2.6000000000000002e-05, "loss": 1.1718, "step": 260 }, { "epoch": 0.01, "learning_rate": 2.61e-05, "loss": 1.011, "step": 261 }, { "epoch": 0.01, "learning_rate": 2.6200000000000003e-05, "loss": 0.9029, "step": 262 }, { "epoch": 0.01, "learning_rate": 2.6300000000000002e-05, "loss": 0.9414, "step": 263 }, { "epoch": 0.01, "learning_rate": 2.64e-05, "loss": 1.0517, "step": 264 }, { "epoch": 0.01, "learning_rate": 2.6500000000000004e-05, "loss": 1.2453, "step": 265 }, { "epoch": 0.01, "learning_rate": 2.6600000000000003e-05, "loss": 1.0625, "step": 266 }, { "epoch": 0.01, "learning_rate": 2.6700000000000002e-05, "loss": 1.1513, "step": 267 }, { "epoch": 0.01, "learning_rate": 2.6800000000000004e-05, "loss": 0.9663, "step": 268 }, { "epoch": 0.01, "learning_rate": 2.6900000000000003e-05, "loss": 0.9564, "step": 269 }, { "epoch": 0.01, "learning_rate": 2.7000000000000002e-05, "loss": 1.2888, "step": 270 }, { "epoch": 0.01, "learning_rate": 2.7100000000000005e-05, "loss": 0.9917, "step": 271 }, { "epoch": 0.01, "learning_rate": 2.7200000000000004e-05, "loss": 1.0223, "step": 272 }, { "epoch": 0.01, "learning_rate": 2.7300000000000003e-05, "loss": 0.9949, "step": 273 }, { "epoch": 0.01, "learning_rate": 2.7400000000000002e-05, "loss": 0.638, "step": 274 }, { "epoch": 0.01, "learning_rate": 2.7500000000000004e-05, "loss": 1.119, "step": 275 }, { "epoch": 0.01, "learning_rate": 2.7600000000000003e-05, "loss": 1.1363, "step": 276 }, { "epoch": 0.01, "learning_rate": 2.7700000000000002e-05, "loss": 0.9719, "step": 277 }, { "epoch": 0.01, "learning_rate": 2.7800000000000005e-05, "loss": 1.1989, "step": 278 }, { "epoch": 0.01, "learning_rate": 2.7900000000000004e-05, "loss": 1.2684, "step": 279 }, { "epoch": 0.01, "learning_rate": 2.8000000000000003e-05, "loss": 1.1805, "step": 280 }, { "epoch": 0.01, "learning_rate": 2.8100000000000005e-05, "loss": 1.0837, "step": 281 }, { "epoch": 0.01, "learning_rate": 2.8199999999999998e-05, "loss": 0.7239, "step": 282 }, { "epoch": 0.01, "learning_rate": 2.83e-05, "loss": 0.8885, "step": 283 }, { "epoch": 0.01, "learning_rate": 2.84e-05, "loss": 0.8504, "step": 284 }, { "epoch": 0.01, "learning_rate": 2.8499999999999998e-05, "loss": 1.0383, "step": 285 }, { "epoch": 0.01, "learning_rate": 2.86e-05, "loss": 1.1825, "step": 286 }, { "epoch": 0.01, "learning_rate": 2.87e-05, "loss": 0.9655, "step": 287 }, { "epoch": 0.01, "learning_rate": 2.88e-05, "loss": 0.9802, "step": 288 }, { "epoch": 0.01, "learning_rate": 2.8899999999999998e-05, "loss": 1.3984, "step": 289 }, { "epoch": 0.01, "learning_rate": 2.9e-05, "loss": 0.745, "step": 290 }, { "epoch": 0.01, "learning_rate": 2.91e-05, "loss": 0.7342, "step": 291 }, { "epoch": 0.01, "learning_rate": 2.9199999999999998e-05, "loss": 1.1584, "step": 292 }, { "epoch": 0.01, "learning_rate": 2.93e-05, "loss": 1.0013, "step": 293 }, { "epoch": 0.01, "learning_rate": 2.94e-05, "loss": 1.0114, "step": 294 }, { "epoch": 0.01, "learning_rate": 2.95e-05, "loss": 1.0517, "step": 295 }, { "epoch": 0.01, "learning_rate": 2.96e-05, "loss": 1.0444, "step": 296 }, { "epoch": 0.01, "learning_rate": 2.97e-05, "loss": 1.0286, "step": 297 }, { "epoch": 0.01, "learning_rate": 2.98e-05, "loss": 0.8234, "step": 298 }, { "epoch": 0.01, "learning_rate": 2.9900000000000002e-05, "loss": 0.9624, "step": 299 }, { "epoch": 0.01, "learning_rate": 3e-05, "loss": 1.0837, "step": 300 }, { "epoch": 0.01, "learning_rate": 3.01e-05, "loss": 1.0141, "step": 301 }, { "epoch": 0.01, "learning_rate": 3.02e-05, "loss": 1.0117, "step": 302 }, { "epoch": 0.01, "learning_rate": 3.03e-05, "loss": 0.7706, "step": 303 }, { "epoch": 0.01, "learning_rate": 3.04e-05, "loss": 0.4922, "step": 304 }, { "epoch": 0.01, "learning_rate": 3.05e-05, "loss": 0.4712, "step": 305 }, { "epoch": 0.01, "learning_rate": 3.06e-05, "loss": 0.4997, "step": 306 }, { "epoch": 0.01, "learning_rate": 3.07e-05, "loss": 1.0335, "step": 307 }, { "epoch": 0.01, "learning_rate": 3.08e-05, "loss": 1.1343, "step": 308 }, { "epoch": 0.01, "learning_rate": 3.09e-05, "loss": 1.9085, "step": 309 }, { "epoch": 0.01, "learning_rate": 3.1e-05, "loss": 1.1252, "step": 310 }, { "epoch": 0.01, "learning_rate": 3.1100000000000004e-05, "loss": 1.1442, "step": 311 }, { "epoch": 0.01, "learning_rate": 3.12e-05, "loss": 1.0668, "step": 312 }, { "epoch": 0.01, "learning_rate": 3.13e-05, "loss": 1.2954, "step": 313 }, { "epoch": 0.01, "learning_rate": 3.1400000000000004e-05, "loss": 1.069, "step": 314 }, { "epoch": 0.01, "learning_rate": 3.15e-05, "loss": 1.1253, "step": 315 }, { "epoch": 0.01, "learning_rate": 3.16e-05, "loss": 0.9456, "step": 316 }, { "epoch": 0.01, "learning_rate": 3.1700000000000005e-05, "loss": 0.9377, "step": 317 }, { "epoch": 0.01, "learning_rate": 3.18e-05, "loss": 0.9714, "step": 318 }, { "epoch": 0.01, "learning_rate": 3.19e-05, "loss": 0.6232, "step": 319 }, { "epoch": 0.01, "learning_rate": 3.2000000000000005e-05, "loss": 1.0865, "step": 320 }, { "epoch": 0.01, "learning_rate": 3.21e-05, "loss": 1.2696, "step": 321 }, { "epoch": 0.01, "learning_rate": 3.2200000000000003e-05, "loss": 0.8133, "step": 322 }, { "epoch": 0.01, "learning_rate": 3.2300000000000006e-05, "loss": 0.941, "step": 323 }, { "epoch": 0.01, "learning_rate": 3.24e-05, "loss": 1.0122, "step": 324 }, { "epoch": 0.01, "learning_rate": 3.2500000000000004e-05, "loss": 1.1509, "step": 325 }, { "epoch": 0.01, "learning_rate": 3.26e-05, "loss": 1.1134, "step": 326 }, { "epoch": 0.01, "learning_rate": 3.27e-05, "loss": 1.1887, "step": 327 }, { "epoch": 0.01, "learning_rate": 3.2800000000000004e-05, "loss": 1.0315, "step": 328 }, { "epoch": 0.01, "learning_rate": 3.29e-05, "loss": 0.8791, "step": 329 }, { "epoch": 0.01, "learning_rate": 3.3e-05, "loss": 0.9155, "step": 330 }, { "epoch": 0.01, "learning_rate": 3.3100000000000005e-05, "loss": 0.7578, "step": 331 }, { "epoch": 0.01, "learning_rate": 3.32e-05, "loss": 1.0643, "step": 332 }, { "epoch": 0.01, "learning_rate": 3.33e-05, "loss": 0.9133, "step": 333 }, { "epoch": 0.01, "learning_rate": 3.3400000000000005e-05, "loss": 2.0044, "step": 334 }, { "epoch": 0.01, "learning_rate": 3.35e-05, "loss": 1.0864, "step": 335 }, { "epoch": 0.01, "learning_rate": 3.3600000000000004e-05, "loss": 1.2835, "step": 336 }, { "epoch": 0.01, "learning_rate": 3.3700000000000006e-05, "loss": 0.9378, "step": 337 }, { "epoch": 0.01, "learning_rate": 3.38e-05, "loss": 1.109, "step": 338 }, { "epoch": 0.01, "learning_rate": 3.3900000000000004e-05, "loss": 1.2929, "step": 339 }, { "epoch": 0.01, "learning_rate": 3.4000000000000007e-05, "loss": 1.3857, "step": 340 }, { "epoch": 0.01, "learning_rate": 3.41e-05, "loss": 0.981, "step": 341 }, { "epoch": 0.01, "learning_rate": 3.4200000000000005e-05, "loss": 2.257, "step": 342 }, { "epoch": 0.01, "learning_rate": 3.430000000000001e-05, "loss": 0.5908, "step": 343 }, { "epoch": 0.01, "learning_rate": 3.4399999999999996e-05, "loss": 0.7566, "step": 344 }, { "epoch": 0.01, "learning_rate": 3.45e-05, "loss": 1.1229, "step": 345 }, { "epoch": 0.01, "learning_rate": 3.46e-05, "loss": 0.7835, "step": 346 }, { "epoch": 0.01, "learning_rate": 3.4699999999999996e-05, "loss": 0.9341, "step": 347 }, { "epoch": 0.01, "learning_rate": 3.48e-05, "loss": 1.1787, "step": 348 }, { "epoch": 0.01, "learning_rate": 3.49e-05, "loss": 1.1112, "step": 349 }, { "epoch": 0.01, "learning_rate": 3.5e-05, "loss": 3.2219, "step": 350 }, { "epoch": 0.01, "learning_rate": 3.51e-05, "loss": 1.091, "step": 351 }, { "epoch": 0.01, "learning_rate": 3.52e-05, "loss": 1.2185, "step": 352 }, { "epoch": 0.01, "learning_rate": 3.53e-05, "loss": 1.1731, "step": 353 }, { "epoch": 0.01, "learning_rate": 3.54e-05, "loss": 1.0625, "step": 354 }, { "epoch": 0.01, "learning_rate": 3.55e-05, "loss": 0.9566, "step": 355 }, { "epoch": 0.01, "learning_rate": 3.56e-05, "loss": 1.1454, "step": 356 }, { "epoch": 0.01, "learning_rate": 3.57e-05, "loss": 1.1279, "step": 357 }, { "epoch": 0.01, "learning_rate": 3.58e-05, "loss": 0.928, "step": 358 }, { "epoch": 0.01, "learning_rate": 3.59e-05, "loss": 1.1628, "step": 359 }, { "epoch": 0.01, "learning_rate": 3.6e-05, "loss": 1.0322, "step": 360 }, { "epoch": 0.01, "learning_rate": 3.61e-05, "loss": 1.0848, "step": 361 }, { "epoch": 0.01, "learning_rate": 3.62e-05, "loss": 0.9059, "step": 362 }, { "epoch": 0.01, "learning_rate": 3.63e-05, "loss": 0.6308, "step": 363 }, { "epoch": 0.01, "learning_rate": 3.6400000000000004e-05, "loss": 1.3706, "step": 364 }, { "epoch": 0.01, "learning_rate": 3.65e-05, "loss": 1.0869, "step": 365 }, { "epoch": 0.01, "learning_rate": 3.66e-05, "loss": 1.1262, "step": 366 }, { "epoch": 0.01, "learning_rate": 3.6700000000000004e-05, "loss": 0.9867, "step": 367 }, { "epoch": 0.01, "learning_rate": 3.68e-05, "loss": 0.9007, "step": 368 }, { "epoch": 0.01, "learning_rate": 3.69e-05, "loss": 1.9711, "step": 369 }, { "epoch": 0.01, "learning_rate": 3.7e-05, "loss": 1.3015, "step": 370 }, { "epoch": 0.01, "learning_rate": 3.71e-05, "loss": 1.1176, "step": 371 }, { "epoch": 0.01, "learning_rate": 3.72e-05, "loss": 1.0123, "step": 372 }, { "epoch": 0.01, "learning_rate": 3.73e-05, "loss": 0.9507, "step": 373 }, { "epoch": 0.01, "learning_rate": 3.74e-05, "loss": 1.2132, "step": 374 }, { "epoch": 0.01, "learning_rate": 3.7500000000000003e-05, "loss": 1.1541, "step": 375 }, { "epoch": 0.01, "learning_rate": 3.76e-05, "loss": 1.0509, "step": 376 }, { "epoch": 0.01, "learning_rate": 3.77e-05, "loss": 2.0036, "step": 377 }, { "epoch": 0.01, "learning_rate": 3.7800000000000004e-05, "loss": 5.1656, "step": 378 }, { "epoch": 0.01, "learning_rate": 3.79e-05, "loss": 1.163, "step": 379 }, { "epoch": 0.01, "learning_rate": 3.8e-05, "loss": 0.9722, "step": 380 }, { "epoch": 0.01, "learning_rate": 3.8100000000000005e-05, "loss": 0.8693, "step": 381 }, { "epoch": 0.01, "learning_rate": 3.82e-05, "loss": 1.1957, "step": 382 }, { "epoch": 0.01, "learning_rate": 3.83e-05, "loss": 0.5759, "step": 383 }, { "epoch": 0.01, "learning_rate": 3.8400000000000005e-05, "loss": 0.7519, "step": 384 }, { "epoch": 0.01, "learning_rate": 3.85e-05, "loss": 1.1128, "step": 385 }, { "epoch": 0.01, "learning_rate": 3.86e-05, "loss": 1.0514, "step": 386 }, { "epoch": 0.01, "learning_rate": 3.8700000000000006e-05, "loss": 0.9645, "step": 387 }, { "epoch": 0.01, "learning_rate": 3.88e-05, "loss": 1.037, "step": 388 }, { "epoch": 0.01, "learning_rate": 3.8900000000000004e-05, "loss": 0.914, "step": 389 }, { "epoch": 0.01, "learning_rate": 3.9000000000000006e-05, "loss": 1.0831, "step": 390 }, { "epoch": 0.01, "learning_rate": 3.91e-05, "loss": 1.0921, "step": 391 }, { "epoch": 0.01, "learning_rate": 3.9200000000000004e-05, "loss": 0.9368, "step": 392 }, { "epoch": 0.01, "learning_rate": 3.9300000000000007e-05, "loss": 0.8841, "step": 393 }, { "epoch": 0.01, "learning_rate": 3.94e-05, "loss": 0.9968, "step": 394 }, { "epoch": 0.01, "learning_rate": 3.9500000000000005e-05, "loss": 0.8618, "step": 395 }, { "epoch": 0.01, "learning_rate": 3.960000000000001e-05, "loss": 1.1933, "step": 396 }, { "epoch": 0.01, "learning_rate": 3.97e-05, "loss": 0.9643, "step": 397 }, { "epoch": 0.01, "learning_rate": 3.9800000000000005e-05, "loss": 1.079, "step": 398 }, { "epoch": 0.01, "learning_rate": 3.99e-05, "loss": 1.0366, "step": 399 }, { "epoch": 0.01, "learning_rate": 4e-05, "loss": 1.0924, "step": 400 }, { "epoch": 0.01, "learning_rate": 4.0100000000000006e-05, "loss": 0.6042, "step": 401 }, { "epoch": 0.01, "learning_rate": 4.02e-05, "loss": 1.1285, "step": 402 }, { "epoch": 0.01, "learning_rate": 4.0300000000000004e-05, "loss": 0.5503, "step": 403 }, { "epoch": 0.01, "learning_rate": 4.0400000000000006e-05, "loss": 0.3643, "step": 404 }, { "epoch": 0.01, "learning_rate": 4.05e-05, "loss": 0.3732, "step": 405 }, { "epoch": 0.01, "learning_rate": 4.0600000000000004e-05, "loss": 0.4976, "step": 406 }, { "epoch": 0.01, "learning_rate": 4.07e-05, "loss": 0.8022, "step": 407 }, { "epoch": 0.01, "learning_rate": 4.08e-05, "loss": 1.0256, "step": 408 }, { "epoch": 0.01, "learning_rate": 4.09e-05, "loss": 1.0692, "step": 409 }, { "epoch": 0.01, "learning_rate": 4.1e-05, "loss": 1.0508, "step": 410 }, { "epoch": 0.01, "learning_rate": 4.11e-05, "loss": 1.1401, "step": 411 }, { "epoch": 0.01, "learning_rate": 4.12e-05, "loss": 0.9171, "step": 412 }, { "epoch": 0.01, "learning_rate": 4.13e-05, "loss": 0.7977, "step": 413 }, { "epoch": 0.01, "learning_rate": 4.14e-05, "loss": 1.2948, "step": 414 }, { "epoch": 0.01, "learning_rate": 4.15e-05, "loss": 1.0664, "step": 415 }, { "epoch": 0.01, "learning_rate": 4.16e-05, "loss": 1.0121, "step": 416 }, { "epoch": 0.01, "learning_rate": 4.17e-05, "loss": 1.0377, "step": 417 }, { "epoch": 0.01, "learning_rate": 4.18e-05, "loss": 0.9814, "step": 418 }, { "epoch": 0.01, "learning_rate": 4.19e-05, "loss": 0.9039, "step": 419 }, { "epoch": 0.01, "learning_rate": 4.2e-05, "loss": 1.1005, "step": 420 }, { "epoch": 0.01, "learning_rate": 4.21e-05, "loss": 0.8384, "step": 421 }, { "epoch": 0.01, "learning_rate": 4.22e-05, "loss": 0.9081, "step": 422 }, { "epoch": 0.01, "learning_rate": 4.23e-05, "loss": 0.3466, "step": 423 }, { "epoch": 0.01, "learning_rate": 4.24e-05, "loss": 0.2945, "step": 424 }, { "epoch": 0.01, "learning_rate": 4.25e-05, "loss": 0.2931, "step": 425 }, { "epoch": 0.01, "learning_rate": 4.26e-05, "loss": 0.6122, "step": 426 }, { "epoch": 0.01, "learning_rate": 4.27e-05, "loss": 0.922, "step": 427 }, { "epoch": 0.01, "learning_rate": 4.2800000000000004e-05, "loss": 1.0912, "step": 428 }, { "epoch": 0.01, "learning_rate": 4.29e-05, "loss": 0.763, "step": 429 }, { "epoch": 0.01, "learning_rate": 4.3e-05, "loss": 0.7723, "step": 430 }, { "epoch": 0.01, "learning_rate": 4.3100000000000004e-05, "loss": 1.3065, "step": 431 }, { "epoch": 0.01, "learning_rate": 4.32e-05, "loss": 1.4166, "step": 432 }, { "epoch": 0.01, "learning_rate": 4.33e-05, "loss": 1.5187, "step": 433 }, { "epoch": 0.01, "learning_rate": 4.3400000000000005e-05, "loss": 0.9958, "step": 434 }, { "epoch": 0.01, "learning_rate": 4.35e-05, "loss": 1.1421, "step": 435 }, { "epoch": 0.01, "learning_rate": 4.36e-05, "loss": 1.2012, "step": 436 }, { "epoch": 0.01, "learning_rate": 4.3700000000000005e-05, "loss": 1.0703, "step": 437 }, { "epoch": 0.01, "learning_rate": 4.38e-05, "loss": 1.3227, "step": 438 }, { "epoch": 0.01, "learning_rate": 4.39e-05, "loss": 1.1361, "step": 439 }, { "epoch": 0.01, "learning_rate": 4.4000000000000006e-05, "loss": 1.0903, "step": 440 }, { "epoch": 0.01, "learning_rate": 4.41e-05, "loss": 1.0382, "step": 441 }, { "epoch": 0.01, "learning_rate": 4.4200000000000004e-05, "loss": 1.1548, "step": 442 }, { "epoch": 0.01, "learning_rate": 4.43e-05, "loss": 1.0122, "step": 443 }, { "epoch": 0.01, "learning_rate": 4.44e-05, "loss": 1.0829, "step": 444 }, { "epoch": 0.01, "learning_rate": 4.4500000000000004e-05, "loss": 0.8663, "step": 445 }, { "epoch": 0.01, "learning_rate": 4.46e-05, "loss": 0.8373, "step": 446 }, { "epoch": 0.01, "learning_rate": 4.47e-05, "loss": 0.8238, "step": 447 }, { "epoch": 0.01, "learning_rate": 4.4800000000000005e-05, "loss": 1.2691, "step": 448 }, { "epoch": 0.01, "learning_rate": 4.49e-05, "loss": 1.1696, "step": 449 }, { "epoch": 0.01, "learning_rate": 4.5e-05, "loss": 1.0971, "step": 450 }, { "epoch": 0.02, "learning_rate": 4.5100000000000005e-05, "loss": 0.9564, "step": 451 }, { "epoch": 0.02, "learning_rate": 4.52e-05, "loss": 1.2204, "step": 452 }, { "epoch": 0.02, "learning_rate": 4.53e-05, "loss": 1.7533, "step": 453 }, { "epoch": 0.02, "learning_rate": 4.5400000000000006e-05, "loss": 1.9244, "step": 454 }, { "epoch": 0.02, "learning_rate": 4.55e-05, "loss": 1.2036, "step": 455 }, { "epoch": 0.02, "learning_rate": 4.5600000000000004e-05, "loss": 1.0282, "step": 456 }, { "epoch": 0.02, "learning_rate": 4.5700000000000006e-05, "loss": 1.1927, "step": 457 }, { "epoch": 0.02, "learning_rate": 4.58e-05, "loss": 1.1088, "step": 458 }, { "epoch": 0.02, "learning_rate": 4.5900000000000004e-05, "loss": 1.1382, "step": 459 }, { "epoch": 0.02, "learning_rate": 4.600000000000001e-05, "loss": 1.0705, "step": 460 }, { "epoch": 0.02, "learning_rate": 4.61e-05, "loss": 1.2408, "step": 461 }, { "epoch": 0.02, "learning_rate": 4.6200000000000005e-05, "loss": 1.6399, "step": 462 }, { "epoch": 0.02, "learning_rate": 4.630000000000001e-05, "loss": 1.68, "step": 463 }, { "epoch": 0.02, "learning_rate": 4.64e-05, "loss": 1.7153, "step": 464 }, { "epoch": 0.02, "learning_rate": 4.6500000000000005e-05, "loss": 1.7273, "step": 465 }, { "epoch": 0.02, "learning_rate": 4.660000000000001e-05, "loss": 1.6837, "step": 466 }, { "epoch": 0.02, "learning_rate": 4.6700000000000003e-05, "loss": 1.1287, "step": 467 }, { "epoch": 0.02, "learning_rate": 4.6800000000000006e-05, "loss": 0.4856, "step": 468 }, { "epoch": 0.02, "learning_rate": 4.69e-05, "loss": 0.9443, "step": 469 }, { "epoch": 0.02, "learning_rate": 4.7e-05, "loss": 1.0555, "step": 470 }, { "epoch": 0.02, "learning_rate": 4.71e-05, "loss": 0.645, "step": 471 }, { "epoch": 0.02, "learning_rate": 4.72e-05, "loss": 1.1149, "step": 472 }, { "epoch": 0.02, "learning_rate": 4.73e-05, "loss": 0.8703, "step": 473 }, { "epoch": 0.02, "learning_rate": 4.74e-05, "loss": 0.7289, "step": 474 }, { "epoch": 0.02, "learning_rate": 4.75e-05, "loss": 0.9002, "step": 475 }, { "epoch": 0.02, "learning_rate": 4.76e-05, "loss": 0.6147, "step": 476 }, { "epoch": 0.02, "learning_rate": 4.77e-05, "loss": 0.536, "step": 477 }, { "epoch": 0.02, "learning_rate": 4.78e-05, "loss": 0.5452, "step": 478 }, { "epoch": 0.02, "learning_rate": 4.79e-05, "loss": 0.5007, "step": 479 }, { "epoch": 0.02, "learning_rate": 4.8e-05, "loss": 1.2486, "step": 480 }, { "epoch": 0.02, "learning_rate": 4.8100000000000004e-05, "loss": 1.1059, "step": 481 }, { "epoch": 0.02, "learning_rate": 4.82e-05, "loss": 1.0986, "step": 482 }, { "epoch": 0.02, "learning_rate": 4.83e-05, "loss": 0.7648, "step": 483 }, { "epoch": 0.02, "learning_rate": 4.8400000000000004e-05, "loss": 0.944, "step": 484 }, { "epoch": 0.02, "learning_rate": 4.85e-05, "loss": 1.0316, "step": 485 }, { "epoch": 0.02, "learning_rate": 4.86e-05, "loss": 1.0771, "step": 486 }, { "epoch": 0.02, "learning_rate": 4.87e-05, "loss": 1.0432, "step": 487 }, { "epoch": 0.02, "learning_rate": 4.88e-05, "loss": 1.0193, "step": 488 }, { "epoch": 0.02, "learning_rate": 4.89e-05, "loss": 1.3888, "step": 489 }, { "epoch": 0.02, "learning_rate": 4.9e-05, "loss": 1.0071, "step": 490 }, { "epoch": 0.02, "learning_rate": 4.91e-05, "loss": 1.0176, "step": 491 }, { "epoch": 0.02, "learning_rate": 4.92e-05, "loss": 0.7121, "step": 492 }, { "epoch": 0.02, "learning_rate": 4.93e-05, "loss": 0.9969, "step": 493 }, { "epoch": 0.02, "learning_rate": 4.94e-05, "loss": 0.9738, "step": 494 }, { "epoch": 0.02, "learning_rate": 4.9500000000000004e-05, "loss": 1.1143, "step": 495 }, { "epoch": 0.02, "learning_rate": 4.96e-05, "loss": 1.0727, "step": 496 }, { "epoch": 0.02, "learning_rate": 4.97e-05, "loss": 0.8705, "step": 497 }, { "epoch": 0.02, "learning_rate": 4.9800000000000004e-05, "loss": 1.1296, "step": 498 }, { "epoch": 0.02, "learning_rate": 4.99e-05, "loss": 1.2015, "step": 499 }, { "epoch": 0.02, "learning_rate": 5e-05, "loss": 1.4, "step": 500 }, { "epoch": 0.02, "learning_rate": 4.999999985823608e-05, "loss": 1.1035, "step": 501 }, { "epoch": 0.02, "learning_rate": 4.999999943294431e-05, "loss": 1.0822, "step": 502 }, { "epoch": 0.02, "learning_rate": 4.9999998724124696e-05, "loss": 2.0827, "step": 503 }, { "epoch": 0.02, "learning_rate": 4.999999773177725e-05, "loss": 0.8651, "step": 504 }, { "epoch": 0.02, "learning_rate": 4.9999996455901984e-05, "loss": 0.3208, "step": 505 }, { "epoch": 0.02, "learning_rate": 4.9999994896498905e-05, "loss": 0.7302, "step": 506 }, { "epoch": 0.02, "learning_rate": 4.999999305356804e-05, "loss": 1.0614, "step": 507 }, { "epoch": 0.02, "learning_rate": 4.999999092710941e-05, "loss": 1.1216, "step": 508 }, { "epoch": 0.02, "learning_rate": 4.999998851712302e-05, "loss": 1.1721, "step": 509 }, { "epoch": 0.02, "learning_rate": 4.999998582360892e-05, "loss": 1.8053, "step": 510 }, { "epoch": 0.02, "learning_rate": 4.999998284656714e-05, "loss": 1.5432, "step": 511 }, { "epoch": 0.02, "learning_rate": 4.99999795859977e-05, "loss": 1.2041, "step": 512 }, { "epoch": 0.02, "learning_rate": 4.999997604190064e-05, "loss": 1.2164, "step": 513 }, { "epoch": 0.02, "learning_rate": 4.999997221427601e-05, "loss": 1.159, "step": 514 }, { "epoch": 0.02, "learning_rate": 4.9999968103123844e-05, "loss": 1.5075, "step": 515 }, { "epoch": 0.02, "learning_rate": 4.9999963708444196e-05, "loss": 1.5984, "step": 516 }, { "epoch": 0.02, "learning_rate": 4.999995903023711e-05, "loss": 1.7598, "step": 517 }, { "epoch": 0.02, "learning_rate": 4.999995406850263e-05, "loss": 1.7227, "step": 518 }, { "epoch": 0.02, "learning_rate": 4.9999948823240835e-05, "loss": 3.535, "step": 519 }, { "epoch": 0.02, "learning_rate": 4.9999943294451766e-05, "loss": 2.7366, "step": 520 }, { "epoch": 0.02, "learning_rate": 4.9999937482135494e-05, "loss": 1.3016, "step": 521 }, { "epoch": 0.02, "learning_rate": 4.9999931386292086e-05, "loss": 1.0784, "step": 522 }, { "epoch": 0.02, "learning_rate": 4.999992500692161e-05, "loss": 1.1245, "step": 523 }, { "epoch": 0.02, "learning_rate": 4.999991834402413e-05, "loss": 1.1599, "step": 524 }, { "epoch": 0.02, "learning_rate": 4.999991139759973e-05, "loss": 1.0401, "step": 525 }, { "epoch": 0.02, "learning_rate": 4.999990416764848e-05, "loss": 1.2025, "step": 526 }, { "epoch": 0.02, "learning_rate": 4.999989665417048e-05, "loss": 0.951, "step": 527 }, { "epoch": 0.02, "learning_rate": 4.99998888571658e-05, "loss": 1.0453, "step": 528 }, { "epoch": 0.02, "learning_rate": 4.9999880776634535e-05, "loss": 0.5765, "step": 529 }, { "epoch": 0.02, "learning_rate": 4.9999872412576764e-05, "loss": 1.0687, "step": 530 }, { "epoch": 0.02, "learning_rate": 4.99998637649926e-05, "loss": 1.0548, "step": 531 }, { "epoch": 0.02, "learning_rate": 4.9999854833882134e-05, "loss": 0.9824, "step": 532 }, { "epoch": 0.02, "learning_rate": 4.9999845619245464e-05, "loss": 1.058, "step": 533 }, { "epoch": 0.02, "learning_rate": 4.9999836121082696e-05, "loss": 0.9737, "step": 534 }, { "epoch": 0.02, "learning_rate": 4.999982633939394e-05, "loss": 0.9848, "step": 535 }, { "epoch": 0.02, "learning_rate": 4.9999816274179306e-05, "loss": 0.9737, "step": 536 }, { "epoch": 0.02, "learning_rate": 4.99998059254389e-05, "loss": 0.9113, "step": 537 }, { "epoch": 0.02, "learning_rate": 4.9999795293172857e-05, "loss": 1.6386, "step": 538 }, { "epoch": 0.02, "learning_rate": 4.9999784377381284e-05, "loss": 5.1346, "step": 539 }, { "epoch": 0.02, "learning_rate": 4.999977317806431e-05, "loss": 2.55, "step": 540 }, { "epoch": 0.02, "learning_rate": 4.999976169522206e-05, "loss": 1.0977, "step": 541 }, { "epoch": 0.02, "learning_rate": 4.999974992885466e-05, "loss": 1.2365, "step": 542 }, { "epoch": 0.02, "learning_rate": 4.999973787896225e-05, "loss": 0.983, "step": 543 }, { "epoch": 0.02, "learning_rate": 4.999972554554497e-05, "loss": 0.3399, "step": 544 }, { "epoch": 0.02, "learning_rate": 4.9999712928602946e-05, "loss": 1.1137, "step": 545 }, { "epoch": 0.02, "learning_rate": 4.9999700028136336e-05, "loss": 0.8944, "step": 546 }, { "epoch": 0.02, "learning_rate": 4.999968684414528e-05, "loss": 0.9436, "step": 547 }, { "epoch": 0.02, "learning_rate": 4.999967337662992e-05, "loss": 1.1348, "step": 548 }, { "epoch": 0.02, "learning_rate": 4.999965962559042e-05, "loss": 1.8587, "step": 549 }, { "epoch": 0.02, "learning_rate": 4.999964559102694e-05, "loss": 1.2981, "step": 550 }, { "epoch": 0.02, "learning_rate": 4.999963127293962e-05, "loss": 1.5095, "step": 551 }, { "epoch": 0.02, "learning_rate": 4.999961667132864e-05, "loss": 0.9359, "step": 552 }, { "epoch": 0.02, "learning_rate": 4.999960178619416e-05, "loss": 1.3416, "step": 553 }, { "epoch": 0.02, "learning_rate": 4.999958661753634e-05, "loss": 0.7874, "step": 554 }, { "epoch": 0.02, "learning_rate": 4.9999571165355375e-05, "loss": 1.1931, "step": 555 }, { "epoch": 0.02, "learning_rate": 4.999955542965141e-05, "loss": 1.014, "step": 556 }, { "epoch": 0.02, "learning_rate": 4.9999539410424646e-05, "loss": 0.9826, "step": 557 }, { "epoch": 0.02, "learning_rate": 4.999952310767526e-05, "loss": 1.1048, "step": 558 }, { "epoch": 0.02, "learning_rate": 4.999950652140343e-05, "loss": 0.9856, "step": 559 }, { "epoch": 0.02, "learning_rate": 4.999948965160935e-05, "loss": 0.5809, "step": 560 }, { "epoch": 0.02, "learning_rate": 4.999947249829321e-05, "loss": 1.1025, "step": 561 }, { "epoch": 0.02, "learning_rate": 4.99994550614552e-05, "loss": 1.2854, "step": 562 }, { "epoch": 0.02, "learning_rate": 4.9999437341095535e-05, "loss": 1.0803, "step": 563 }, { "epoch": 0.02, "learning_rate": 4.9999419337214394e-05, "loss": 5.298, "step": 564 }, { "epoch": 0.02, "learning_rate": 4.999940104981199e-05, "loss": 5.5241, "step": 565 }, { "epoch": 0.02, "learning_rate": 4.999938247888852e-05, "loss": 1.1165, "step": 566 }, { "epoch": 0.02, "learning_rate": 4.999936362444422e-05, "loss": 0.1644, "step": 567 }, { "epoch": 0.02, "learning_rate": 4.999934448647929e-05, "loss": 0.0298, "step": 568 }, { "epoch": 0.02, "learning_rate": 4.999932506499394e-05, "loss": 0.0189, "step": 569 }, { "epoch": 0.02, "learning_rate": 4.99993053599884e-05, "loss": 0.7365, "step": 570 }, { "epoch": 0.02, "learning_rate": 4.999928537146289e-05, "loss": 3.3343, "step": 571 }, { "epoch": 0.02, "learning_rate": 4.999926509941763e-05, "loss": 7.1695, "step": 572 }, { "epoch": 0.02, "learning_rate": 4.9999244543852867e-05, "loss": 1.5502, "step": 573 }, { "epoch": 0.02, "learning_rate": 4.9999223704768814e-05, "loss": 0.9496, "step": 574 }, { "epoch": 0.02, "learning_rate": 4.999920258216572e-05, "loss": 1.1089, "step": 575 }, { "epoch": 0.02, "learning_rate": 4.999918117604382e-05, "loss": 1.3243, "step": 576 }, { "epoch": 0.02, "learning_rate": 4.9999159486403366e-05, "loss": 2.0218, "step": 577 }, { "epoch": 0.02, "learning_rate": 4.9999137513244586e-05, "loss": 0.8848, "step": 578 }, { "epoch": 0.02, "learning_rate": 4.9999115256567745e-05, "loss": 0.9969, "step": 579 }, { "epoch": 0.02, "learning_rate": 4.999909271637309e-05, "loss": 0.9872, "step": 580 }, { "epoch": 0.02, "learning_rate": 4.999906989266088e-05, "loss": 1.0756, "step": 581 }, { "epoch": 0.02, "learning_rate": 4.999904678543136e-05, "loss": 1.0085, "step": 582 }, { "epoch": 0.02, "learning_rate": 4.999902339468481e-05, "loss": 0.929, "step": 583 }, { "epoch": 0.02, "learning_rate": 4.999899972042149e-05, "loss": 0.8327, "step": 584 }, { "epoch": 0.02, "learning_rate": 4.999897576264166e-05, "loss": 1.0558, "step": 585 }, { "epoch": 0.02, "learning_rate": 4.99989515213456e-05, "loss": 0.739, "step": 586 }, { "epoch": 0.02, "learning_rate": 4.999892699653358e-05, "loss": 1.0294, "step": 587 }, { "epoch": 0.02, "learning_rate": 4.999890218820589e-05, "loss": 0.8908, "step": 588 }, { "epoch": 0.02, "learning_rate": 4.9998877096362795e-05, "loss": 1.3354, "step": 589 }, { "epoch": 0.02, "learning_rate": 4.999885172100459e-05, "loss": 0.7489, "step": 590 }, { "epoch": 0.02, "learning_rate": 4.999882606213155e-05, "loss": 1.0825, "step": 591 }, { "epoch": 0.02, "learning_rate": 4.999880011974399e-05, "loss": 1.0044, "step": 592 }, { "epoch": 0.02, "learning_rate": 4.999877389384218e-05, "loss": 0.9195, "step": 593 }, { "epoch": 0.02, "learning_rate": 4.999874738442643e-05, "loss": 0.9995, "step": 594 }, { "epoch": 0.02, "learning_rate": 4.999872059149704e-05, "loss": 1.134, "step": 595 }, { "epoch": 0.02, "learning_rate": 4.9998693515054315e-05, "loss": 1.0598, "step": 596 }, { "epoch": 0.02, "learning_rate": 4.999866615509855e-05, "loss": 3.6495, "step": 597 }, { "epoch": 0.02, "learning_rate": 4.999863851163007e-05, "loss": 6.0466, "step": 598 }, { "epoch": 0.02, "learning_rate": 4.9998610584649184e-05, "loss": 5.9407, "step": 599 }, { "epoch": 0.02, "learning_rate": 4.999858237415621e-05, "loss": 3.7008, "step": 600 }, { "epoch": 0.02, "learning_rate": 4.999855388015146e-05, "loss": 1.0635, "step": 601 }, { "epoch": 0.02, "learning_rate": 4.9998525102635254e-05, "loss": 1.1143, "step": 602 }, { "epoch": 0.02, "learning_rate": 4.999849604160794e-05, "loss": 1.1094, "step": 603 }, { "epoch": 0.02, "learning_rate": 4.9998466697069836e-05, "loss": 1.0941, "step": 604 }, { "epoch": 0.02, "learning_rate": 4.9998437069021266e-05, "loss": 0.8967, "step": 605 }, { "epoch": 0.02, "learning_rate": 4.999840715746257e-05, "loss": 0.7643, "step": 606 }, { "epoch": 0.02, "learning_rate": 4.99983769623941e-05, "loss": 1.1887, "step": 607 }, { "epoch": 0.02, "learning_rate": 4.9998346483816185e-05, "loss": 1.1365, "step": 608 }, { "epoch": 0.02, "learning_rate": 4.9998315721729174e-05, "loss": 1.146, "step": 609 }, { "epoch": 0.02, "learning_rate": 4.999828467613341e-05, "loss": 1.1914, "step": 610 }, { "epoch": 0.02, "learning_rate": 4.999825334702926e-05, "loss": 0.4841, "step": 611 }, { "epoch": 0.02, "learning_rate": 4.9998221734417074e-05, "loss": 1.2752, "step": 612 }, { "epoch": 0.02, "learning_rate": 4.99981898382972e-05, "loss": 1.3246, "step": 613 }, { "epoch": 0.02, "learning_rate": 4.9998157658670005e-05, "loss": 1.0093, "step": 614 }, { "epoch": 0.02, "learning_rate": 4.999812519553586e-05, "loss": 1.0018, "step": 615 }, { "epoch": 0.02, "learning_rate": 4.9998092448895136e-05, "loss": 0.5941, "step": 616 }, { "epoch": 0.02, "learning_rate": 4.9998059418748186e-05, "loss": 1.1514, "step": 617 }, { "epoch": 0.02, "learning_rate": 4.9998026105095405e-05, "loss": 1.0478, "step": 618 }, { "epoch": 0.02, "learning_rate": 4.999799250793716e-05, "loss": 1.3763, "step": 619 }, { "epoch": 0.02, "learning_rate": 4.999795862727383e-05, "loss": 1.0823, "step": 620 }, { "epoch": 0.02, "learning_rate": 4.999792446310581e-05, "loss": 1.5328, "step": 621 }, { "epoch": 0.02, "learning_rate": 4.999789001543348e-05, "loss": 1.1532, "step": 622 }, { "epoch": 0.02, "learning_rate": 4.9997855284257225e-05, "loss": 0.9416, "step": 623 }, { "epoch": 0.02, "learning_rate": 4.999782026957745e-05, "loss": 1.1897, "step": 624 }, { "epoch": 0.02, "learning_rate": 4.999778497139455e-05, "loss": 0.9818, "step": 625 }, { "epoch": 0.02, "learning_rate": 4.9997749389708913e-05, "loss": 1.0783, "step": 626 }, { "epoch": 0.02, "learning_rate": 4.9997713524520964e-05, "loss": 0.9242, "step": 627 }, { "epoch": 0.02, "learning_rate": 4.9997677375831095e-05, "loss": 1.1113, "step": 628 }, { "epoch": 0.02, "learning_rate": 4.999764094363971e-05, "loss": 0.843, "step": 629 }, { "epoch": 0.02, "learning_rate": 4.999760422794724e-05, "loss": 0.7661, "step": 630 }, { "epoch": 0.02, "learning_rate": 4.9997567228754095e-05, "loss": 1.1915, "step": 631 }, { "epoch": 0.02, "learning_rate": 4.9997529946060683e-05, "loss": 0.9509, "step": 632 }, { "epoch": 0.02, "learning_rate": 4.9997492379867445e-05, "loss": 1.0885, "step": 633 }, { "epoch": 0.02, "learning_rate": 4.999745453017479e-05, "loss": 1.1266, "step": 634 }, { "epoch": 0.02, "learning_rate": 4.999741639698317e-05, "loss": 1.3212, "step": 635 }, { "epoch": 0.02, "learning_rate": 4.999737798029299e-05, "loss": 1.1455, "step": 636 }, { "epoch": 0.02, "learning_rate": 4.999733928010469e-05, "loss": 1.0126, "step": 637 }, { "epoch": 0.02, "learning_rate": 4.9997300296418736e-05, "loss": 1.1852, "step": 638 }, { "epoch": 0.02, "learning_rate": 4.999726102923554e-05, "loss": 3.8548, "step": 639 }, { "epoch": 0.02, "learning_rate": 4.9997221478555564e-05, "loss": 4.9868, "step": 640 }, { "epoch": 0.02, "learning_rate": 4.9997181644379255e-05, "loss": 0.9864, "step": 641 }, { "epoch": 0.02, "learning_rate": 4.999714152670706e-05, "loss": 1.36, "step": 642 }, { "epoch": 0.02, "learning_rate": 4.999710112553942e-05, "loss": 1.1334, "step": 643 }, { "epoch": 0.02, "learning_rate": 4.999706044087683e-05, "loss": 0.9677, "step": 644 }, { "epoch": 0.02, "learning_rate": 4.999701947271972e-05, "loss": 1.091, "step": 645 }, { "epoch": 0.02, "learning_rate": 4.999697822106856e-05, "loss": 0.8693, "step": 646 }, { "epoch": 0.02, "learning_rate": 4.9996936685923826e-05, "loss": 0.8486, "step": 647 }, { "epoch": 0.02, "learning_rate": 4.999689486728599e-05, "loss": 1.0339, "step": 648 }, { "epoch": 0.02, "learning_rate": 4.9996852765155515e-05, "loss": 0.7757, "step": 649 }, { "epoch": 0.02, "learning_rate": 4.999681037953289e-05, "loss": 0.9295, "step": 650 }, { "epoch": 0.02, "learning_rate": 4.999676771041859e-05, "loss": 0.8866, "step": 651 }, { "epoch": 0.02, "learning_rate": 4.99967247578131e-05, "loss": 0.9513, "step": 652 }, { "epoch": 0.02, "learning_rate": 4.99966815217169e-05, "loss": 0.4406, "step": 653 }, { "epoch": 0.02, "learning_rate": 4.999663800213049e-05, "loss": 0.8982, "step": 654 }, { "epoch": 0.02, "learning_rate": 4.999659419905437e-05, "loss": 1.0514, "step": 655 }, { "epoch": 0.02, "learning_rate": 4.9996550112489016e-05, "loss": 1.2258, "step": 656 }, { "epoch": 0.02, "learning_rate": 4.999650574243494e-05, "loss": 1.2742, "step": 657 }, { "epoch": 0.02, "learning_rate": 4.9996461088892653e-05, "loss": 1.1957, "step": 658 }, { "epoch": 0.02, "learning_rate": 4.9996416151862645e-05, "loss": 0.8459, "step": 659 }, { "epoch": 0.02, "learning_rate": 4.999637093134544e-05, "loss": 1.0066, "step": 660 }, { "epoch": 0.02, "learning_rate": 4.999632542734154e-05, "loss": 2.5795, "step": 661 }, { "epoch": 0.02, "learning_rate": 4.999627963985147e-05, "loss": 5.9296, "step": 662 }, { "epoch": 0.02, "learning_rate": 4.999623356887574e-05, "loss": 5.9331, "step": 663 }, { "epoch": 0.02, "learning_rate": 4.999618721441488e-05, "loss": 5.0072, "step": 664 }, { "epoch": 0.02, "learning_rate": 4.999614057646942e-05, "loss": 2.0132, "step": 665 }, { "epoch": 0.02, "learning_rate": 4.999609365503987e-05, "loss": 1.9424, "step": 666 }, { "epoch": 0.02, "learning_rate": 4.999604645012678e-05, "loss": 1.0953, "step": 667 }, { "epoch": 0.02, "learning_rate": 4.999599896173068e-05, "loss": 0.9892, "step": 668 }, { "epoch": 0.02, "learning_rate": 4.999595118985211e-05, "loss": 1.2581, "step": 669 }, { "epoch": 0.02, "learning_rate": 4.999590313449161e-05, "loss": 0.9823, "step": 670 }, { "epoch": 0.02, "learning_rate": 4.9995854795649714e-05, "loss": 1.1271, "step": 671 }, { "epoch": 0.02, "learning_rate": 4.999580617332699e-05, "loss": 0.9738, "step": 672 }, { "epoch": 0.02, "learning_rate": 4.999575726752398e-05, "loss": 1.0559, "step": 673 }, { "epoch": 0.02, "learning_rate": 4.999570807824124e-05, "loss": 0.3781, "step": 674 }, { "epoch": 0.02, "learning_rate": 4.999565860547932e-05, "loss": 0.9622, "step": 675 }, { "epoch": 0.02, "learning_rate": 4.99956088492388e-05, "loss": 1.4014, "step": 676 }, { "epoch": 0.02, "learning_rate": 4.999555880952023e-05, "loss": 1.6963, "step": 677 }, { "epoch": 0.02, "learning_rate": 4.999550848632417e-05, "loss": 1.6848, "step": 678 }, { "epoch": 0.02, "learning_rate": 4.9995457879651205e-05, "loss": 1.6055, "step": 679 }, { "epoch": 0.02, "learning_rate": 4.999540698950191e-05, "loss": 1.59, "step": 680 }, { "epoch": 0.02, "learning_rate": 4.9995355815876856e-05, "loss": 1.3511, "step": 681 }, { "epoch": 0.02, "learning_rate": 4.9995304358776624e-05, "loss": 1.0763, "step": 682 }, { "epoch": 0.02, "learning_rate": 4.999525261820179e-05, "loss": 1.0664, "step": 683 }, { "epoch": 0.02, "learning_rate": 4.999520059415296e-05, "loss": 1.1856, "step": 684 }, { "epoch": 0.02, "learning_rate": 4.999514828663071e-05, "loss": 1.0733, "step": 685 }, { "epoch": 0.02, "learning_rate": 4.999509569563563e-05, "loss": 1.0524, "step": 686 }, { "epoch": 0.02, "learning_rate": 4.999504282116833e-05, "loss": 1.6952, "step": 687 }, { "epoch": 0.02, "learning_rate": 4.999498966322939e-05, "loss": 1.039, "step": 688 }, { "epoch": 0.02, "learning_rate": 4.999493622181943e-05, "loss": 0.8589, "step": 689 }, { "epoch": 0.02, "learning_rate": 4.9994882496939053e-05, "loss": 1.0662, "step": 690 }, { "epoch": 0.02, "learning_rate": 4.9994828488588866e-05, "loss": 1.086, "step": 691 }, { "epoch": 0.02, "learning_rate": 4.999477419676948e-05, "loss": 0.7508, "step": 692 }, { "epoch": 0.02, "learning_rate": 4.9994719621481514e-05, "loss": 0.8504, "step": 693 }, { "epoch": 0.02, "learning_rate": 4.999466476272558e-05, "loss": 0.8165, "step": 694 }, { "epoch": 0.02, "learning_rate": 4.9994609620502316e-05, "loss": 1.2648, "step": 695 }, { "epoch": 0.02, "learning_rate": 4.9994554194812324e-05, "loss": 0.7299, "step": 696 }, { "epoch": 0.02, "learning_rate": 4.999449848565625e-05, "loss": 1.1341, "step": 697 }, { "epoch": 0.02, "learning_rate": 4.999444249303473e-05, "loss": 0.9238, "step": 698 }, { "epoch": 0.02, "learning_rate": 4.9994386216948386e-05, "loss": 1.1114, "step": 699 }, { "epoch": 0.02, "learning_rate": 4.999432965739785e-05, "loss": 0.9728, "step": 700 }, { "epoch": 0.02, "learning_rate": 4.9994272814383785e-05, "loss": 1.0707, "step": 701 }, { "epoch": 0.02, "learning_rate": 4.999421568790682e-05, "loss": 1.258, "step": 702 }, { "epoch": 0.02, "learning_rate": 4.999415827796761e-05, "loss": 1.5377, "step": 703 }, { "epoch": 0.02, "learning_rate": 4.999410058456681e-05, "loss": 1.0112, "step": 704 }, { "epoch": 0.02, "learning_rate": 4.999404260770505e-05, "loss": 0.8529, "step": 705 }, { "epoch": 0.02, "learning_rate": 4.9993984347383024e-05, "loss": 0.9574, "step": 706 }, { "epoch": 0.02, "learning_rate": 4.999392580360136e-05, "loss": 1.0201, "step": 707 }, { "epoch": 0.02, "learning_rate": 4.9993866976360746e-05, "loss": 0.8943, "step": 708 }, { "epoch": 0.02, "learning_rate": 4.999380786566183e-05, "loss": 1.057, "step": 709 }, { "epoch": 0.02, "learning_rate": 4.9993748471505295e-05, "loss": 0.6514, "step": 710 }, { "epoch": 0.02, "learning_rate": 4.999368879389182e-05, "loss": 1.2567, "step": 711 }, { "epoch": 0.02, "learning_rate": 4.9993628832822064e-05, "loss": 1.6651, "step": 712 }, { "epoch": 0.02, "learning_rate": 4.999356858829672e-05, "loss": 1.7762, "step": 713 }, { "epoch": 0.02, "learning_rate": 4.999350806031646e-05, "loss": 1.3754, "step": 714 }, { "epoch": 0.02, "learning_rate": 4.999344724888199e-05, "loss": 0.8887, "step": 715 }, { "epoch": 0.02, "learning_rate": 4.999338615399398e-05, "loss": 1.0226, "step": 716 }, { "epoch": 0.02, "learning_rate": 4.999332477565313e-05, "loss": 0.5723, "step": 717 }, { "epoch": 0.02, "learning_rate": 4.9993263113860134e-05, "loss": 0.8042, "step": 718 }, { "epoch": 0.02, "learning_rate": 4.99932011686157e-05, "loss": 0.4557, "step": 719 }, { "epoch": 0.02, "learning_rate": 4.9993138939920516e-05, "loss": 0.7832, "step": 720 }, { "epoch": 0.02, "learning_rate": 4.99930764277753e-05, "loss": 1.7556, "step": 721 }, { "epoch": 0.02, "learning_rate": 4.9993013632180766e-05, "loss": 1.0122, "step": 722 }, { "epoch": 0.02, "learning_rate": 4.9992950553137604e-05, "loss": 1.0788, "step": 723 }, { "epoch": 0.02, "learning_rate": 4.9992887190646555e-05, "loss": 0.8928, "step": 724 }, { "epoch": 0.02, "learning_rate": 4.999282354470831e-05, "loss": 1.1244, "step": 725 }, { "epoch": 0.02, "learning_rate": 4.9992759615323616e-05, "loss": 1.1244, "step": 726 }, { "epoch": 0.02, "learning_rate": 4.999269540249318e-05, "loss": 1.0085, "step": 727 }, { "epoch": 0.02, "learning_rate": 4.999263090621775e-05, "loss": 1.0098, "step": 728 }, { "epoch": 0.02, "learning_rate": 4.999256612649804e-05, "loss": 1.0846, "step": 729 }, { "epoch": 0.02, "learning_rate": 4.999250106333479e-05, "loss": 1.0779, "step": 730 }, { "epoch": 0.02, "learning_rate": 4.9992435716728734e-05, "loss": 0.9044, "step": 731 }, { "epoch": 0.02, "learning_rate": 4.9992370086680626e-05, "loss": 0.7695, "step": 732 }, { "epoch": 0.02, "learning_rate": 4.99923041731912e-05, "loss": 0.5065, "step": 733 }, { "epoch": 0.02, "learning_rate": 4.99922379762612e-05, "loss": 1.0083, "step": 734 }, { "epoch": 0.02, "learning_rate": 4.999217149589138e-05, "loss": 1.0828, "step": 735 }, { "epoch": 0.02, "learning_rate": 4.99921047320825e-05, "loss": 1.2457, "step": 736 }, { "epoch": 0.02, "learning_rate": 4.999203768483531e-05, "loss": 1.0432, "step": 737 }, { "epoch": 0.02, "learning_rate": 4.999197035415058e-05, "loss": 1.0739, "step": 738 }, { "epoch": 0.02, "learning_rate": 4.999190274002906e-05, "loss": 1.305, "step": 739 }, { "epoch": 0.02, "learning_rate": 4.999183484247153e-05, "loss": 1.0634, "step": 740 }, { "epoch": 0.02, "learning_rate": 4.999176666147876e-05, "loss": 0.603, "step": 741 }, { "epoch": 0.02, "learning_rate": 4.99916981970515e-05, "loss": 0.2903, "step": 742 }, { "epoch": 0.02, "learning_rate": 4.999162944919056e-05, "loss": 0.2822, "step": 743 }, { "epoch": 0.02, "learning_rate": 4.999156041789669e-05, "loss": 1.0935, "step": 744 }, { "epoch": 0.02, "learning_rate": 4.999149110317069e-05, "loss": 1.0742, "step": 745 }, { "epoch": 0.02, "learning_rate": 4.999142150501335e-05, "loss": 1.0379, "step": 746 }, { "epoch": 0.02, "learning_rate": 4.999135162342544e-05, "loss": 0.7829, "step": 747 }, { "epoch": 0.02, "learning_rate": 4.999128145840777e-05, "loss": 0.8828, "step": 748 }, { "epoch": 0.02, "learning_rate": 4.999121100996113e-05, "loss": 1.1007, "step": 749 }, { "epoch": 0.03, "learning_rate": 4.9991140278086316e-05, "loss": 1.2393, "step": 750 }, { "epoch": 0.03, "learning_rate": 4.999106926278414e-05, "loss": 1.0302, "step": 751 }, { "epoch": 0.03, "learning_rate": 4.999099796405539e-05, "loss": 1.0873, "step": 752 }, { "epoch": 0.03, "learning_rate": 4.999092638190089e-05, "loss": 1.0347, "step": 753 }, { "epoch": 0.03, "learning_rate": 4.999085451632145e-05, "loss": 0.9964, "step": 754 }, { "epoch": 0.03, "learning_rate": 4.999078236731788e-05, "loss": 1.0707, "step": 755 }, { "epoch": 0.03, "learning_rate": 4.9990709934891e-05, "loss": 0.5079, "step": 756 }, { "epoch": 0.03, "learning_rate": 4.999063721904163e-05, "loss": 3.7136, "step": 757 }, { "epoch": 0.03, "learning_rate": 4.9990564219770596e-05, "loss": 2.0047, "step": 758 }, { "epoch": 0.03, "learning_rate": 4.9990490937078724e-05, "loss": 0.9224, "step": 759 }, { "epoch": 0.03, "learning_rate": 4.9990417370966856e-05, "loss": 0.8483, "step": 760 }, { "epoch": 0.03, "learning_rate": 4.999034352143581e-05, "loss": 1.0651, "step": 761 }, { "epoch": 0.03, "learning_rate": 4.999026938848644e-05, "loss": 1.0003, "step": 762 }, { "epoch": 0.03, "learning_rate": 4.999019497211957e-05, "loss": 1.3056, "step": 763 }, { "epoch": 0.03, "learning_rate": 4.999012027233605e-05, "loss": 0.4875, "step": 764 }, { "epoch": 0.03, "learning_rate": 4.999004528913673e-05, "loss": 0.6589, "step": 765 }, { "epoch": 0.03, "learning_rate": 4.998997002252247e-05, "loss": 4.3198, "step": 766 }, { "epoch": 0.03, "learning_rate": 4.9989894472494105e-05, "loss": 5.6637, "step": 767 }, { "epoch": 0.03, "learning_rate": 4.99898186390525e-05, "loss": 1.6974, "step": 768 }, { "epoch": 0.03, "learning_rate": 4.998974252219851e-05, "loss": 0.918, "step": 769 }, { "epoch": 0.03, "learning_rate": 4.998966612193301e-05, "loss": 0.9248, "step": 770 }, { "epoch": 0.03, "learning_rate": 4.9989589438256856e-05, "loss": 0.9166, "step": 771 }, { "epoch": 0.03, "learning_rate": 4.9989512471170926e-05, "loss": 0.866, "step": 772 }, { "epoch": 0.03, "learning_rate": 4.9989435220676086e-05, "loss": 1.0056, "step": 773 }, { "epoch": 0.03, "learning_rate": 4.998935768677321e-05, "loss": 1.0067, "step": 774 }, { "epoch": 0.03, "learning_rate": 4.998927986946319e-05, "loss": 1.0082, "step": 775 }, { "epoch": 0.03, "learning_rate": 4.9989201768746884e-05, "loss": 1.0761, "step": 776 }, { "epoch": 0.03, "learning_rate": 4.998912338462521e-05, "loss": 0.7052, "step": 777 }, { "epoch": 0.03, "learning_rate": 4.9989044717099034e-05, "loss": 0.7946, "step": 778 }, { "epoch": 0.03, "learning_rate": 4.998896576616926e-05, "loss": 0.0242, "step": 779 }, { "epoch": 0.03, "learning_rate": 4.9988886531836775e-05, "loss": 0.0242, "step": 780 }, { "epoch": 0.03, "learning_rate": 4.998880701410248e-05, "loss": 1.8697, "step": 781 }, { "epoch": 0.03, "learning_rate": 4.9988727212967277e-05, "loss": 0.9932, "step": 782 }, { "epoch": 0.03, "learning_rate": 4.9988647128432075e-05, "loss": 1.4304, "step": 783 }, { "epoch": 0.03, "learning_rate": 4.9988566760497776e-05, "loss": 1.0616, "step": 784 }, { "epoch": 0.03, "learning_rate": 4.99884861091653e-05, "loss": 0.86, "step": 785 }, { "epoch": 0.03, "learning_rate": 4.998840517443555e-05, "loss": 0.8413, "step": 786 }, { "epoch": 0.03, "learning_rate": 4.998832395630946e-05, "loss": 1.6091, "step": 787 }, { "epoch": 0.03, "learning_rate": 4.9988242454787926e-05, "loss": 1.1217, "step": 788 }, { "epoch": 0.03, "learning_rate": 4.99881606698719e-05, "loss": 1.0512, "step": 789 }, { "epoch": 0.03, "learning_rate": 4.9988078601562294e-05, "loss": 1.099, "step": 790 }, { "epoch": 0.03, "learning_rate": 4.9987996249860034e-05, "loss": 0.9169, "step": 791 }, { "epoch": 0.03, "learning_rate": 4.998791361476607e-05, "loss": 1.1508, "step": 792 }, { "epoch": 0.03, "learning_rate": 4.9987830696281335e-05, "loss": 1.0625, "step": 793 }, { "epoch": 0.03, "learning_rate": 4.998774749440675e-05, "loss": 1.0979, "step": 794 }, { "epoch": 0.03, "learning_rate": 4.998766400914329e-05, "loss": 1.1011, "step": 795 }, { "epoch": 0.03, "learning_rate": 4.9987580240491886e-05, "loss": 0.9401, "step": 796 }, { "epoch": 0.03, "learning_rate": 4.998749618845348e-05, "loss": 1.0996, "step": 797 }, { "epoch": 0.03, "learning_rate": 4.998741185302903e-05, "loss": 0.7629, "step": 798 }, { "epoch": 0.03, "learning_rate": 4.9987327234219506e-05, "loss": 0.9754, "step": 799 }, { "epoch": 0.03, "learning_rate": 4.9987242332025846e-05, "loss": 1.0664, "step": 800 }, { "epoch": 0.03, "learning_rate": 4.9987157146449026e-05, "loss": 1.0725, "step": 801 }, { "epoch": 0.03, "learning_rate": 4.9987071677490015e-05, "loss": 1.1946, "step": 802 }, { "epoch": 0.03, "learning_rate": 4.9986985925149775e-05, "loss": 0.8215, "step": 803 }, { "epoch": 0.03, "learning_rate": 4.998689988942928e-05, "loss": 1.0311, "step": 804 }, { "epoch": 0.03, "learning_rate": 4.998681357032952e-05, "loss": 0.9088, "step": 805 }, { "epoch": 0.03, "learning_rate": 4.998672696785144e-05, "loss": 0.8205, "step": 806 }, { "epoch": 0.03, "learning_rate": 4.998664008199606e-05, "loss": 1.0748, "step": 807 }, { "epoch": 0.03, "learning_rate": 4.998655291276434e-05, "loss": 1.0655, "step": 808 }, { "epoch": 0.03, "learning_rate": 4.998646546015727e-05, "loss": 0.9902, "step": 809 }, { "epoch": 0.03, "learning_rate": 4.9986377724175857e-05, "loss": 0.9229, "step": 810 }, { "epoch": 0.03, "learning_rate": 4.998628970482109e-05, "loss": 0.8658, "step": 811 }, { "epoch": 0.03, "learning_rate": 4.998620140209396e-05, "loss": 0.6486, "step": 812 }, { "epoch": 0.03, "learning_rate": 4.998611281599548e-05, "loss": 0.7194, "step": 813 }, { "epoch": 0.03, "learning_rate": 4.998602394652664e-05, "loss": 1.0978, "step": 814 }, { "epoch": 0.03, "learning_rate": 4.998593479368846e-05, "loss": 0.9391, "step": 815 }, { "epoch": 0.03, "learning_rate": 4.998584535748195e-05, "loss": 1.0839, "step": 816 }, { "epoch": 0.03, "learning_rate": 4.998575563790812e-05, "loss": 1.1805, "step": 817 }, { "epoch": 0.03, "learning_rate": 4.998566563496798e-05, "loss": 1.0149, "step": 818 }, { "epoch": 0.03, "learning_rate": 4.998557534866257e-05, "loss": 1.1287, "step": 819 }, { "epoch": 0.03, "learning_rate": 4.99854847789929e-05, "loss": 0.4336, "step": 820 }, { "epoch": 0.03, "learning_rate": 4.998539392596e-05, "loss": 0.293, "step": 821 }, { "epoch": 0.03, "learning_rate": 4.99853027895649e-05, "loss": 1.1152, "step": 822 }, { "epoch": 0.03, "learning_rate": 4.9985211369808636e-05, "loss": 0.8716, "step": 823 }, { "epoch": 0.03, "learning_rate": 4.998511966669224e-05, "loss": 0.8151, "step": 824 }, { "epoch": 0.03, "learning_rate": 4.9985027680216765e-05, "loss": 0.7003, "step": 825 }, { "epoch": 0.03, "learning_rate": 4.998493541038324e-05, "loss": 0.9177, "step": 826 }, { "epoch": 0.03, "learning_rate": 4.998484285719271e-05, "loss": 1.0679, "step": 827 }, { "epoch": 0.03, "learning_rate": 4.998475002064624e-05, "loss": 1.064, "step": 828 }, { "epoch": 0.03, "learning_rate": 4.998465690074487e-05, "loss": 1.0429, "step": 829 }, { "epoch": 0.03, "learning_rate": 4.998456349748966e-05, "loss": 0.9084, "step": 830 }, { "epoch": 0.03, "learning_rate": 4.998446981088166e-05, "loss": 1.3639, "step": 831 }, { "epoch": 0.03, "learning_rate": 4.998437584092196e-05, "loss": 1.7693, "step": 832 }, { "epoch": 0.03, "learning_rate": 4.9984281587611594e-05, "loss": 1.1641, "step": 833 }, { "epoch": 0.03, "learning_rate": 4.998418705095165e-05, "loss": 1.0059, "step": 834 }, { "epoch": 0.03, "learning_rate": 4.998409223094319e-05, "loss": 1.0959, "step": 835 }, { "epoch": 0.03, "learning_rate": 4.998399712758729e-05, "loss": 1.0846, "step": 836 }, { "epoch": 0.03, "learning_rate": 4.998390174088504e-05, "loss": 1.6591, "step": 837 }, { "epoch": 0.03, "learning_rate": 4.998380607083751e-05, "loss": 1.9079, "step": 838 }, { "epoch": 0.03, "learning_rate": 4.9983710117445795e-05, "loss": 2.6471, "step": 839 }, { "epoch": 0.03, "learning_rate": 4.9983613880710966e-05, "loss": 1.1417, "step": 840 }, { "epoch": 0.03, "learning_rate": 4.998351736063414e-05, "loss": 1.1277, "step": 841 }, { "epoch": 0.03, "learning_rate": 4.998342055721639e-05, "loss": 0.9206, "step": 842 }, { "epoch": 0.03, "learning_rate": 4.998332347045882e-05, "loss": 0.7786, "step": 843 }, { "epoch": 0.03, "learning_rate": 4.998322610036253e-05, "loss": 0.7362, "step": 844 }, { "epoch": 0.03, "learning_rate": 4.998312844692863e-05, "loss": 0.8869, "step": 845 }, { "epoch": 0.03, "learning_rate": 4.998303051015822e-05, "loss": 0.9924, "step": 846 }, { "epoch": 0.03, "learning_rate": 4.9982932290052426e-05, "loss": 1.2144, "step": 847 }, { "epoch": 0.03, "learning_rate": 4.998283378661235e-05, "loss": 1.0188, "step": 848 }, { "epoch": 0.03, "learning_rate": 4.99827349998391e-05, "loss": 0.5631, "step": 849 }, { "epoch": 0.03, "learning_rate": 4.998263592973381e-05, "loss": 0.9233, "step": 850 }, { "epoch": 0.03, "learning_rate": 4.99825365762976e-05, "loss": 1.7673, "step": 851 }, { "epoch": 0.03, "learning_rate": 4.99824369395316e-05, "loss": 2.3103, "step": 852 }, { "epoch": 0.03, "learning_rate": 4.998233701943693e-05, "loss": 1.3813, "step": 853 }, { "epoch": 0.03, "learning_rate": 4.998223681601473e-05, "loss": 0.9617, "step": 854 }, { "epoch": 0.03, "learning_rate": 4.9982136329266144e-05, "loss": 0.9735, "step": 855 }, { "epoch": 0.03, "learning_rate": 4.99820355591923e-05, "loss": 3.1264, "step": 856 }, { "epoch": 0.03, "learning_rate": 4.998193450579434e-05, "loss": 1.7254, "step": 857 }, { "epoch": 0.03, "learning_rate": 4.998183316907342e-05, "loss": 1.1576, "step": 858 }, { "epoch": 0.03, "learning_rate": 4.9981731549030684e-05, "loss": 0.6205, "step": 859 }, { "epoch": 0.03, "learning_rate": 4.9981629645667285e-05, "loss": 0.433, "step": 860 }, { "epoch": 0.03, "learning_rate": 4.9981527458984375e-05, "loss": 0.6958, "step": 861 }, { "epoch": 0.03, "learning_rate": 4.9981424988983114e-05, "loss": 1.342, "step": 862 }, { "epoch": 0.03, "learning_rate": 4.998132223566467e-05, "loss": 1.013, "step": 863 }, { "epoch": 0.03, "learning_rate": 4.9981219199030205e-05, "loss": 1.0718, "step": 864 }, { "epoch": 0.03, "learning_rate": 4.998111587908089e-05, "loss": 1.2388, "step": 865 }, { "epoch": 0.03, "learning_rate": 4.998101227581789e-05, "loss": 1.0322, "step": 866 }, { "epoch": 0.03, "learning_rate": 4.9980908389242386e-05, "loss": 1.0696, "step": 867 }, { "epoch": 0.03, "learning_rate": 4.9980804219355545e-05, "loss": 0.7743, "step": 868 }, { "epoch": 0.03, "learning_rate": 4.9980699766158565e-05, "loss": 0.6875, "step": 869 }, { "epoch": 0.03, "learning_rate": 4.9980595029652625e-05, "loss": 0.9857, "step": 870 }, { "epoch": 0.03, "learning_rate": 4.998049000983891e-05, "loss": 1.067, "step": 871 }, { "epoch": 0.03, "learning_rate": 4.998038470671861e-05, "loss": 1.0825, "step": 872 }, { "epoch": 0.03, "learning_rate": 4.9980279120292925e-05, "loss": 0.8276, "step": 873 }, { "epoch": 0.03, "learning_rate": 4.9980173250563045e-05, "loss": 0.9464, "step": 874 }, { "epoch": 0.03, "learning_rate": 4.998006709753017e-05, "loss": 0.7202, "step": 875 }, { "epoch": 0.03, "learning_rate": 4.9979960661195514e-05, "loss": 0.8827, "step": 876 }, { "epoch": 0.03, "learning_rate": 4.997985394156028e-05, "loss": 0.769, "step": 877 }, { "epoch": 0.03, "learning_rate": 4.9979746938625674e-05, "loss": 0.9303, "step": 878 }, { "epoch": 0.03, "learning_rate": 4.997963965239291e-05, "loss": 1.0076, "step": 879 }, { "epoch": 0.03, "learning_rate": 4.997953208286321e-05, "loss": 0.9663, "step": 880 }, { "epoch": 0.03, "learning_rate": 4.9979424230037794e-05, "loss": 1.0237, "step": 881 }, { "epoch": 0.03, "learning_rate": 4.997931609391787e-05, "loss": 0.8674, "step": 882 }, { "epoch": 0.03, "learning_rate": 4.997920767450469e-05, "loss": 0.9624, "step": 883 }, { "epoch": 0.03, "learning_rate": 4.997909897179946e-05, "loss": 1.239, "step": 884 }, { "epoch": 0.03, "learning_rate": 4.997898998580343e-05, "loss": 0.9411, "step": 885 }, { "epoch": 0.03, "learning_rate": 4.997888071651783e-05, "loss": 0.4863, "step": 886 }, { "epoch": 0.03, "learning_rate": 4.9978771163943895e-05, "loss": 0.655, "step": 887 }, { "epoch": 0.03, "learning_rate": 4.997866132808287e-05, "loss": 1.0665, "step": 888 }, { "epoch": 0.03, "learning_rate": 4.9978551208936e-05, "loss": 0.859, "step": 889 }, { "epoch": 0.03, "learning_rate": 4.9978440806504533e-05, "loss": 0.5385, "step": 890 }, { "epoch": 0.03, "learning_rate": 4.997833012078973e-05, "loss": 0.9614, "step": 891 }, { "epoch": 0.03, "learning_rate": 4.997821915179284e-05, "loss": 1.0716, "step": 892 }, { "epoch": 0.03, "learning_rate": 4.9978107899515116e-05, "loss": 1.4625, "step": 893 }, { "epoch": 0.03, "learning_rate": 4.9977996363957824e-05, "loss": 1.2037, "step": 894 }, { "epoch": 0.03, "learning_rate": 4.9977884545122233e-05, "loss": 1.3755, "step": 895 }, { "epoch": 0.03, "learning_rate": 4.9977772443009607e-05, "loss": 0.9815, "step": 896 }, { "epoch": 0.03, "learning_rate": 4.997766005762121e-05, "loss": 1.0318, "step": 897 }, { "epoch": 0.03, "learning_rate": 4.997754738895833e-05, "loss": 1.0689, "step": 898 }, { "epoch": 0.03, "learning_rate": 4.9977434437022245e-05, "loss": 1.0287, "step": 899 }, { "epoch": 0.03, "learning_rate": 4.997732120181423e-05, "loss": 0.9911, "step": 900 }, { "epoch": 0.03, "learning_rate": 4.997720768333557e-05, "loss": 0.7412, "step": 901 }, { "epoch": 0.03, "learning_rate": 4.997709388158754e-05, "loss": 0.8223, "step": 902 }, { "epoch": 0.03, "learning_rate": 4.9976979796571456e-05, "loss": 0.4464, "step": 903 }, { "epoch": 0.03, "learning_rate": 4.99768654282886e-05, "loss": 0.6912, "step": 904 }, { "epoch": 0.03, "learning_rate": 4.9976750776740253e-05, "loss": 0.9775, "step": 905 }, { "epoch": 0.03, "learning_rate": 4.9976635841927745e-05, "loss": 1.1354, "step": 906 }, { "epoch": 0.03, "learning_rate": 4.997652062385236e-05, "loss": 1.3721, "step": 907 }, { "epoch": 0.03, "learning_rate": 4.9976405122515406e-05, "loss": 0.9926, "step": 908 }, { "epoch": 0.03, "learning_rate": 4.9976289337918205e-05, "loss": 0.9205, "step": 909 }, { "epoch": 0.03, "learning_rate": 4.9976173270062046e-05, "loss": 0.5192, "step": 910 }, { "epoch": 0.03, "learning_rate": 4.997605691894827e-05, "loss": 0.9495, "step": 911 }, { "epoch": 0.03, "learning_rate": 4.997594028457819e-05, "loss": 0.8935, "step": 912 }, { "epoch": 0.03, "learning_rate": 4.9975823366953124e-05, "loss": 0.9985, "step": 913 }, { "epoch": 0.03, "learning_rate": 4.9975706166074395e-05, "loss": 0.9489, "step": 914 }, { "epoch": 0.03, "learning_rate": 4.997558868194334e-05, "loss": 1.1144, "step": 915 }, { "epoch": 0.03, "learning_rate": 4.997547091456128e-05, "loss": 0.85, "step": 916 }, { "epoch": 0.03, "learning_rate": 4.997535286392958e-05, "loss": 1.0494, "step": 917 }, { "epoch": 0.03, "learning_rate": 4.9975234530049544e-05, "loss": 0.927, "step": 918 }, { "epoch": 0.03, "learning_rate": 4.9975115912922525e-05, "loss": 1.3081, "step": 919 }, { "epoch": 0.03, "learning_rate": 4.997499701254987e-05, "loss": 1.3798, "step": 920 }, { "epoch": 0.03, "learning_rate": 4.997487782893294e-05, "loss": 2.9096, "step": 921 }, { "epoch": 0.03, "learning_rate": 4.997475836207307e-05, "loss": 2.0541, "step": 922 }, { "epoch": 0.03, "learning_rate": 4.997463861197162e-05, "loss": 2.2006, "step": 923 }, { "epoch": 0.03, "learning_rate": 4.997451857862995e-05, "loss": 2.2202, "step": 924 }, { "epoch": 0.03, "learning_rate": 4.997439826204942e-05, "loss": 1.9072, "step": 925 }, { "epoch": 0.03, "learning_rate": 4.997427766223138e-05, "loss": 0.9502, "step": 926 }, { "epoch": 0.03, "learning_rate": 4.9974156779177226e-05, "loss": 0.9424, "step": 927 }, { "epoch": 0.03, "learning_rate": 4.9974035612888316e-05, "loss": 0.9353, "step": 928 }, { "epoch": 0.03, "learning_rate": 4.997391416336602e-05, "loss": 1.0159, "step": 929 }, { "epoch": 0.03, "learning_rate": 4.997379243061172e-05, "loss": 1.121, "step": 930 }, { "epoch": 0.03, "learning_rate": 4.997367041462679e-05, "loss": 0.7087, "step": 931 }, { "epoch": 0.03, "learning_rate": 4.997354811541262e-05, "loss": 0.667, "step": 932 }, { "epoch": 0.03, "learning_rate": 4.99734255329706e-05, "loss": 0.9358, "step": 933 }, { "epoch": 0.03, "learning_rate": 4.9973302667302114e-05, "loss": 1.0585, "step": 934 }, { "epoch": 0.03, "learning_rate": 4.997317951840855e-05, "loss": 1.1548, "step": 935 }, { "epoch": 0.03, "learning_rate": 4.997305608629132e-05, "loss": 0.957, "step": 936 }, { "epoch": 0.03, "learning_rate": 4.9972932370951816e-05, "loss": 0.9012, "step": 937 }, { "epoch": 0.03, "learning_rate": 4.997280837239144e-05, "loss": 1.0312, "step": 938 }, { "epoch": 0.03, "learning_rate": 4.99726840906116e-05, "loss": 1.0875, "step": 939 }, { "epoch": 0.03, "learning_rate": 4.99725595256137e-05, "loss": 1.2106, "step": 940 }, { "epoch": 0.03, "learning_rate": 4.9972434677399166e-05, "loss": 0.8837, "step": 941 }, { "epoch": 0.03, "learning_rate": 4.99723095459694e-05, "loss": 0.0995, "step": 942 }, { "epoch": 0.03, "learning_rate": 4.997218413132583e-05, "loss": 0.917, "step": 943 }, { "epoch": 0.03, "learning_rate": 4.997205843346987e-05, "loss": 0.8453, "step": 944 }, { "epoch": 0.03, "learning_rate": 4.9971932452402956e-05, "loss": 1.147, "step": 945 }, { "epoch": 0.03, "learning_rate": 4.99718061881265e-05, "loss": 1.0417, "step": 946 }, { "epoch": 0.03, "learning_rate": 4.9971679640641956e-05, "loss": 0.3027, "step": 947 }, { "epoch": 0.03, "learning_rate": 4.997155280995074e-05, "loss": 0.1117, "step": 948 }, { "epoch": 0.03, "learning_rate": 4.997142569605431e-05, "loss": 0.0565, "step": 949 }, { "epoch": 0.03, "learning_rate": 4.997129829895409e-05, "loss": 0.0794, "step": 950 }, { "epoch": 0.03, "learning_rate": 4.997117061865153e-05, "loss": 0.0464, "step": 951 }, { "epoch": 0.03, "learning_rate": 4.9971042655148083e-05, "loss": 0.0327, "step": 952 }, { "epoch": 0.03, "learning_rate": 4.9970914408445194e-05, "loss": 0.0483, "step": 953 }, { "epoch": 0.03, "learning_rate": 4.997078587854431e-05, "loss": 1.0637, "step": 954 }, { "epoch": 0.03, "learning_rate": 4.997065706544691e-05, "loss": 0.9498, "step": 955 }, { "epoch": 0.03, "learning_rate": 4.9970527969154443e-05, "loss": 0.7049, "step": 956 }, { "epoch": 0.03, "learning_rate": 4.9970398589668376e-05, "loss": 0.9363, "step": 957 }, { "epoch": 0.03, "learning_rate": 4.997026892699017e-05, "loss": 0.9893, "step": 958 }, { "epoch": 0.03, "learning_rate": 4.99701389811213e-05, "loss": 1.0413, "step": 959 }, { "epoch": 0.03, "learning_rate": 4.997000875206323e-05, "loss": 1.1246, "step": 960 }, { "epoch": 0.03, "learning_rate": 4.996987823981746e-05, "loss": 1.0624, "step": 961 }, { "epoch": 0.03, "learning_rate": 4.9969747444385453e-05, "loss": 0.8815, "step": 962 }, { "epoch": 0.03, "learning_rate": 4.9969616365768695e-05, "loss": 0.726, "step": 963 }, { "epoch": 0.03, "learning_rate": 4.996948500396868e-05, "loss": 0.8177, "step": 964 }, { "epoch": 0.03, "learning_rate": 4.9969353358986876e-05, "loss": 0.9334, "step": 965 }, { "epoch": 0.03, "learning_rate": 4.9969221430824796e-05, "loss": 1.0906, "step": 966 }, { "epoch": 0.03, "learning_rate": 4.9969089219483935e-05, "loss": 0.9816, "step": 967 }, { "epoch": 0.03, "learning_rate": 4.9968956724965784e-05, "loss": 0.855, "step": 968 }, { "epoch": 0.03, "learning_rate": 4.996882394727186e-05, "loss": 0.6776, "step": 969 }, { "epoch": 0.03, "learning_rate": 4.996869088640365e-05, "loss": 1.7443, "step": 970 }, { "epoch": 0.03, "learning_rate": 4.996855754236267e-05, "loss": 1.8524, "step": 971 }, { "epoch": 0.03, "learning_rate": 4.996842391515044e-05, "loss": 1.8876, "step": 972 }, { "epoch": 0.03, "learning_rate": 4.996829000476847e-05, "loss": 1.1181, "step": 973 }, { "epoch": 0.03, "learning_rate": 4.996815581121828e-05, "loss": 1.5167, "step": 974 }, { "epoch": 0.03, "learning_rate": 4.996802133450138e-05, "loss": 0.8539, "step": 975 }, { "epoch": 0.03, "learning_rate": 4.996788657461931e-05, "loss": 1.0526, "step": 976 }, { "epoch": 0.03, "learning_rate": 4.9967751531573604e-05, "loss": 1.102, "step": 977 }, { "epoch": 0.03, "learning_rate": 4.996761620536578e-05, "loss": 1.0377, "step": 978 }, { "epoch": 0.03, "learning_rate": 4.996748059599737e-05, "loss": 1.0049, "step": 979 }, { "epoch": 0.03, "learning_rate": 4.996734470346992e-05, "loss": 0.9249, "step": 980 }, { "epoch": 0.03, "learning_rate": 4.996720852778497e-05, "loss": 1.0987, "step": 981 }, { "epoch": 0.03, "learning_rate": 4.9967072068944065e-05, "loss": 0.9846, "step": 982 }, { "epoch": 0.03, "learning_rate": 4.996693532694875e-05, "loss": 1.1603, "step": 983 }, { "epoch": 0.03, "learning_rate": 4.9966798301800575e-05, "loss": 0.9765, "step": 984 }, { "epoch": 0.03, "learning_rate": 4.99666609935011e-05, "loss": 0.8321, "step": 985 }, { "epoch": 0.03, "learning_rate": 4.996652340205188e-05, "loss": 0.8429, "step": 986 }, { "epoch": 0.03, "learning_rate": 4.9966385527454466e-05, "loss": 0.8801, "step": 987 }, { "epoch": 0.03, "learning_rate": 4.996624736971044e-05, "loss": 0.8495, "step": 988 }, { "epoch": 0.03, "learning_rate": 4.996610892882135e-05, "loss": 1.0207, "step": 989 }, { "epoch": 0.03, "learning_rate": 4.9965970204788775e-05, "loss": 0.8426, "step": 990 }, { "epoch": 0.03, "learning_rate": 4.996583119761429e-05, "loss": 0.4789, "step": 991 }, { "epoch": 0.03, "learning_rate": 4.9965691907299474e-05, "loss": 0.5263, "step": 992 }, { "epoch": 0.03, "learning_rate": 4.996555233384589e-05, "loss": 1.0147, "step": 993 }, { "epoch": 0.03, "learning_rate": 4.9965412477255145e-05, "loss": 0.2394, "step": 994 }, { "epoch": 0.03, "learning_rate": 4.9965272337528803e-05, "loss": 0.6407, "step": 995 }, { "epoch": 0.03, "learning_rate": 4.996513191466847e-05, "loss": 0.7453, "step": 996 }, { "epoch": 0.03, "learning_rate": 4.9964991208675724e-05, "loss": 0.7422, "step": 997 }, { "epoch": 0.03, "learning_rate": 4.9964850219552176e-05, "loss": 0.9518, "step": 998 }, { "epoch": 0.03, "learning_rate": 4.9964708947299424e-05, "loss": 0.8321, "step": 999 }, { "epoch": 0.03, "learning_rate": 4.996456739191905e-05, "loss": 1.2666, "step": 1000 }, { "epoch": 0.03, "learning_rate": 4.996442555341268e-05, "loss": 0.828, "step": 1001 }, { "epoch": 0.03, "learning_rate": 4.996428343178191e-05, "loss": 0.5721, "step": 1002 }, { "epoch": 0.03, "learning_rate": 4.9964141027028364e-05, "loss": 0.8101, "step": 1003 }, { "epoch": 0.03, "learning_rate": 4.996399833915365e-05, "loss": 0.9418, "step": 1004 }, { "epoch": 0.03, "learning_rate": 4.996385536815938e-05, "loss": 1.1385, "step": 1005 }, { "epoch": 0.03, "learning_rate": 4.9963712114047196e-05, "loss": 0.9054, "step": 1006 }, { "epoch": 0.03, "learning_rate": 4.99635685768187e-05, "loss": 1.1033, "step": 1007 }, { "epoch": 0.03, "learning_rate": 4.996342475647553e-05, "loss": 0.7889, "step": 1008 }, { "epoch": 0.03, "learning_rate": 4.996328065301932e-05, "loss": 0.9089, "step": 1009 }, { "epoch": 0.03, "learning_rate": 4.996313626645169e-05, "loss": 1.0402, "step": 1010 }, { "epoch": 0.03, "learning_rate": 4.9962991596774296e-05, "loss": 1.1941, "step": 1011 }, { "epoch": 0.03, "learning_rate": 4.996284664398877e-05, "loss": 1.083, "step": 1012 }, { "epoch": 0.03, "learning_rate": 4.996270140809675e-05, "loss": 0.9324, "step": 1013 }, { "epoch": 0.03, "learning_rate": 4.9962555889099895e-05, "loss": 1.1161, "step": 1014 }, { "epoch": 0.03, "learning_rate": 4.996241008699985e-05, "loss": 1.0825, "step": 1015 }, { "epoch": 0.03, "learning_rate": 4.996226400179827e-05, "loss": 1.1013, "step": 1016 }, { "epoch": 0.03, "learning_rate": 4.996211763349681e-05, "loss": 0.9478, "step": 1017 }, { "epoch": 0.03, "learning_rate": 4.996197098209712e-05, "loss": 1.3722, "step": 1018 }, { "epoch": 0.03, "learning_rate": 4.996182404760088e-05, "loss": 1.5267, "step": 1019 }, { "epoch": 0.03, "learning_rate": 4.9961676830009754e-05, "loss": 1.4267, "step": 1020 }, { "epoch": 0.03, "learning_rate": 4.9961529329325406e-05, "loss": 1.3323, "step": 1021 }, { "epoch": 0.03, "learning_rate": 4.996138154554951e-05, "loss": 1.1731, "step": 1022 }, { "epoch": 0.03, "learning_rate": 4.996123347868373e-05, "loss": 1.0806, "step": 1023 }, { "epoch": 0.03, "learning_rate": 4.9961085128729774e-05, "loss": 1.92, "step": 1024 }, { "epoch": 0.03, "learning_rate": 4.99609364956893e-05, "loss": 2.0663, "step": 1025 }, { "epoch": 0.03, "learning_rate": 4.996078757956401e-05, "loss": 1.0603, "step": 1026 }, { "epoch": 0.03, "learning_rate": 4.996063838035557e-05, "loss": 0.7219, "step": 1027 }, { "epoch": 0.03, "learning_rate": 4.996048889806569e-05, "loss": 1.0208, "step": 1028 }, { "epoch": 0.03, "learning_rate": 4.9960339132696077e-05, "loss": 0.8598, "step": 1029 }, { "epoch": 0.03, "learning_rate": 4.99601890842484e-05, "loss": 1.1062, "step": 1030 }, { "epoch": 0.03, "learning_rate": 4.996003875272438e-05, "loss": 0.4962, "step": 1031 }, { "epoch": 0.03, "learning_rate": 4.995988813812571e-05, "loss": 0.8626, "step": 1032 }, { "epoch": 0.03, "learning_rate": 4.995973724045412e-05, "loss": 1.0223, "step": 1033 }, { "epoch": 0.03, "learning_rate": 4.995958605971129e-05, "loss": 0.9248, "step": 1034 }, { "epoch": 0.03, "learning_rate": 4.9959434595898955e-05, "loss": 0.8516, "step": 1035 }, { "epoch": 0.03, "learning_rate": 4.995928284901884e-05, "loss": 1.2557, "step": 1036 }, { "epoch": 0.03, "learning_rate": 4.995913081907264e-05, "loss": 1.0032, "step": 1037 }, { "epoch": 0.03, "learning_rate": 4.9958978506062105e-05, "loss": 1.05, "step": 1038 }, { "epoch": 0.03, "learning_rate": 4.995882590998895e-05, "loss": 1.0103, "step": 1039 }, { "epoch": 0.03, "learning_rate": 4.9958673030854896e-05, "loss": 0.9441, "step": 1040 }, { "epoch": 0.03, "learning_rate": 4.99585198686617e-05, "loss": 0.5888, "step": 1041 }, { "epoch": 0.03, "learning_rate": 4.995836642341108e-05, "loss": 0.2198, "step": 1042 }, { "epoch": 0.03, "learning_rate": 4.9958212695104786e-05, "loss": 0.9303, "step": 1043 }, { "epoch": 0.03, "learning_rate": 4.9958058683744555e-05, "loss": 0.9888, "step": 1044 }, { "epoch": 0.03, "learning_rate": 4.9957904389332144e-05, "loss": 1.1609, "step": 1045 }, { "epoch": 0.03, "learning_rate": 4.9957749811869294e-05, "loss": 0.9949, "step": 1046 }, { "epoch": 0.03, "learning_rate": 4.9957594951357755e-05, "loss": 1.0034, "step": 1047 }, { "epoch": 0.03, "learning_rate": 4.995743980779929e-05, "loss": 1.0339, "step": 1048 }, { "epoch": 0.03, "learning_rate": 4.9957284381195665e-05, "loss": 0.4992, "step": 1049 }, { "epoch": 0.04, "learning_rate": 4.9957128671548625e-05, "loss": 1.015, "step": 1050 }, { "epoch": 0.04, "learning_rate": 4.995697267885995e-05, "loss": 1.1863, "step": 1051 }, { "epoch": 0.04, "learning_rate": 4.99568164031314e-05, "loss": 0.6663, "step": 1052 }, { "epoch": 0.04, "learning_rate": 4.995665984436476e-05, "loss": 0.7808, "step": 1053 }, { "epoch": 0.04, "learning_rate": 4.99565030025618e-05, "loss": 1.0138, "step": 1054 }, { "epoch": 0.04, "learning_rate": 4.995634587772428e-05, "loss": 0.9296, "step": 1055 }, { "epoch": 0.04, "learning_rate": 4.995618846985401e-05, "loss": 1.0527, "step": 1056 }, { "epoch": 0.04, "learning_rate": 4.995603077895276e-05, "loss": 0.3789, "step": 1057 }, { "epoch": 0.04, "learning_rate": 4.995587280502232e-05, "loss": 0.8297, "step": 1058 }, { "epoch": 0.04, "learning_rate": 4.9955714548064495e-05, "loss": 1.0038, "step": 1059 }, { "epoch": 0.04, "learning_rate": 4.995555600808105e-05, "loss": 0.9593, "step": 1060 }, { "epoch": 0.04, "learning_rate": 4.995539718507382e-05, "loss": 0.2574, "step": 1061 }, { "epoch": 0.04, "learning_rate": 4.9955238079044576e-05, "loss": 0.5138, "step": 1062 }, { "epoch": 0.04, "learning_rate": 4.995507868999513e-05, "loss": 0.6825, "step": 1063 }, { "epoch": 0.04, "learning_rate": 4.995491901792731e-05, "loss": 0.9169, "step": 1064 }, { "epoch": 0.04, "learning_rate": 4.995475906284289e-05, "loss": 0.8881, "step": 1065 }, { "epoch": 0.04, "learning_rate": 4.995459882474372e-05, "loss": 0.9913, "step": 1066 }, { "epoch": 0.04, "learning_rate": 4.995443830363159e-05, "loss": 1.0093, "step": 1067 }, { "epoch": 0.04, "learning_rate": 4.995427749950835e-05, "loss": 0.735, "step": 1068 }, { "epoch": 0.04, "learning_rate": 4.995411641237579e-05, "loss": 0.9745, "step": 1069 }, { "epoch": 0.04, "learning_rate": 4.995395504223575e-05, "loss": 1.1063, "step": 1070 }, { "epoch": 0.04, "learning_rate": 4.9953793389090083e-05, "loss": 0.7945, "step": 1071 }, { "epoch": 0.04, "learning_rate": 4.995363145294058e-05, "loss": 0.9971, "step": 1072 }, { "epoch": 0.04, "learning_rate": 4.995346923378912e-05, "loss": 0.9985, "step": 1073 }, { "epoch": 0.04, "learning_rate": 4.995330673163751e-05, "loss": 0.9424, "step": 1074 }, { "epoch": 0.04, "learning_rate": 4.995314394648761e-05, "loss": 0.9348, "step": 1075 }, { "epoch": 0.04, "learning_rate": 4.995298087834126e-05, "loss": 0.9152, "step": 1076 }, { "epoch": 0.04, "learning_rate": 4.9952817527200313e-05, "loss": 0.905, "step": 1077 }, { "epoch": 0.04, "learning_rate": 4.995265389306661e-05, "loss": 1.0326, "step": 1078 }, { "epoch": 0.04, "learning_rate": 4.995248997594203e-05, "loss": 1.1274, "step": 1079 }, { "epoch": 0.04, "learning_rate": 4.995232577582841e-05, "loss": 1.1464, "step": 1080 }, { "epoch": 0.04, "learning_rate": 4.9952161292727627e-05, "loss": 1.3016, "step": 1081 }, { "epoch": 0.04, "learning_rate": 4.9951996526641544e-05, "loss": 0.7613, "step": 1082 }, { "epoch": 0.04, "learning_rate": 4.995183147757201e-05, "loss": 0.788, "step": 1083 }, { "epoch": 0.04, "learning_rate": 4.995166614552092e-05, "loss": 1.0831, "step": 1084 }, { "epoch": 0.04, "learning_rate": 4.9951500530490146e-05, "loss": 0.9928, "step": 1085 }, { "epoch": 0.04, "learning_rate": 4.995133463248156e-05, "loss": 0.4036, "step": 1086 }, { "epoch": 0.04, "learning_rate": 4.9951168451497044e-05, "loss": 0.8213, "step": 1087 }, { "epoch": 0.04, "learning_rate": 4.995100198753849e-05, "loss": 1.0138, "step": 1088 }, { "epoch": 0.04, "learning_rate": 4.995083524060777e-05, "loss": 1.2604, "step": 1089 }, { "epoch": 0.04, "learning_rate": 4.995066821070679e-05, "loss": 0.9863, "step": 1090 }, { "epoch": 0.04, "learning_rate": 4.995050089783744e-05, "loss": 1.3446, "step": 1091 }, { "epoch": 0.04, "learning_rate": 4.995033330200162e-05, "loss": 1.2619, "step": 1092 }, { "epoch": 0.04, "learning_rate": 4.995016542320122e-05, "loss": 1.1565, "step": 1093 }, { "epoch": 0.04, "learning_rate": 4.9949997261438155e-05, "loss": 1.2085, "step": 1094 }, { "epoch": 0.04, "learning_rate": 4.994982881671433e-05, "loss": 1.0044, "step": 1095 }, { "epoch": 0.04, "learning_rate": 4.994966008903165e-05, "loss": 1.0966, "step": 1096 }, { "epoch": 0.04, "learning_rate": 4.9949491078392034e-05, "loss": 0.9092, "step": 1097 }, { "epoch": 0.04, "learning_rate": 4.9949321784797395e-05, "loss": 1.6385, "step": 1098 }, { "epoch": 0.04, "learning_rate": 4.994915220824966e-05, "loss": 3.0598, "step": 1099 }, { "epoch": 0.04, "learning_rate": 4.994898234875075e-05, "loss": 1.8819, "step": 1100 }, { "epoch": 0.04, "learning_rate": 4.9948812206302584e-05, "loss": 0.691, "step": 1101 }, { "epoch": 0.04, "learning_rate": 4.99486417809071e-05, "loss": 0.8952, "step": 1102 }, { "epoch": 0.04, "learning_rate": 4.9948471072566226e-05, "loss": 1.2006, "step": 1103 }, { "epoch": 0.04, "learning_rate": 4.9948300081281896e-05, "loss": 0.6909, "step": 1104 }, { "epoch": 0.04, "learning_rate": 4.994812880705606e-05, "loss": 1.02, "step": 1105 }, { "epoch": 0.04, "learning_rate": 4.994795724989064e-05, "loss": 1.7347, "step": 1106 }, { "epoch": 0.04, "learning_rate": 4.99477854097876e-05, "loss": 1.2827, "step": 1107 }, { "epoch": 0.04, "learning_rate": 4.9947613286748894e-05, "loss": 1.0047, "step": 1108 }, { "epoch": 0.04, "learning_rate": 4.994744088077646e-05, "loss": 0.9319, "step": 1109 }, { "epoch": 0.04, "learning_rate": 4.9947268191872256e-05, "loss": 0.8484, "step": 1110 }, { "epoch": 0.04, "learning_rate": 4.994709522003823e-05, "loss": 0.9491, "step": 1111 }, { "epoch": 0.04, "learning_rate": 4.994692196527637e-05, "loss": 0.9874, "step": 1112 }, { "epoch": 0.04, "learning_rate": 4.994674842758862e-05, "loss": 1.2731, "step": 1113 }, { "epoch": 0.04, "learning_rate": 4.994657460697695e-05, "loss": 1.2117, "step": 1114 }, { "epoch": 0.04, "learning_rate": 4.9946400503443336e-05, "loss": 1.2264, "step": 1115 }, { "epoch": 0.04, "learning_rate": 4.9946226116989755e-05, "loss": 0.6435, "step": 1116 }, { "epoch": 0.04, "learning_rate": 4.994605144761818e-05, "loss": 0.4793, "step": 1117 }, { "epoch": 0.04, "learning_rate": 4.99458764953306e-05, "loss": 2.1027, "step": 1118 }, { "epoch": 0.04, "learning_rate": 4.994570126012898e-05, "loss": 1.3123, "step": 1119 }, { "epoch": 0.04, "learning_rate": 4.994552574201533e-05, "loss": 1.1505, "step": 1120 }, { "epoch": 0.04, "learning_rate": 4.9945349940991634e-05, "loss": 0.9336, "step": 1121 }, { "epoch": 0.04, "learning_rate": 4.9945173857059876e-05, "loss": 0.6542, "step": 1122 }, { "epoch": 0.04, "learning_rate": 4.9944997490222055e-05, "loss": 0.2621, "step": 1123 }, { "epoch": 0.04, "learning_rate": 4.994482084048018e-05, "loss": 1.848, "step": 1124 }, { "epoch": 0.04, "learning_rate": 4.994464390783625e-05, "loss": 2.3333, "step": 1125 }, { "epoch": 0.04, "learning_rate": 4.994446669229228e-05, "loss": 1.4109, "step": 1126 }, { "epoch": 0.04, "learning_rate": 4.994428919385026e-05, "loss": 0.9536, "step": 1127 }, { "epoch": 0.04, "learning_rate": 4.994411141251222e-05, "loss": 1.2559, "step": 1128 }, { "epoch": 0.04, "learning_rate": 4.994393334828017e-05, "loss": 1.2865, "step": 1129 }, { "epoch": 0.04, "learning_rate": 4.994375500115613e-05, "loss": 1.0191, "step": 1130 }, { "epoch": 0.04, "learning_rate": 4.994357637114212e-05, "loss": 1.1061, "step": 1131 }, { "epoch": 0.04, "learning_rate": 4.994339745824017e-05, "loss": 1.1862, "step": 1132 }, { "epoch": 0.04, "learning_rate": 4.9943218262452314e-05, "loss": 1.0008, "step": 1133 }, { "epoch": 0.04, "learning_rate": 4.994303878378057e-05, "loss": 1.277, "step": 1134 }, { "epoch": 0.04, "learning_rate": 4.9942859022226985e-05, "loss": 4.4138, "step": 1135 }, { "epoch": 0.04, "learning_rate": 4.9942678977793596e-05, "loss": 0.8809, "step": 1136 }, { "epoch": 0.04, "learning_rate": 4.994249865048244e-05, "loss": 1.0296, "step": 1137 }, { "epoch": 0.04, "learning_rate": 4.994231804029557e-05, "loss": 1.2013, "step": 1138 }, { "epoch": 0.04, "learning_rate": 4.9942137147235026e-05, "loss": 0.9212, "step": 1139 }, { "epoch": 0.04, "learning_rate": 4.994195597130286e-05, "loss": 1.0998, "step": 1140 }, { "epoch": 0.04, "learning_rate": 4.994177451250113e-05, "loss": 0.299, "step": 1141 }, { "epoch": 0.04, "learning_rate": 4.99415927708319e-05, "loss": 0.6218, "step": 1142 }, { "epoch": 0.04, "learning_rate": 4.9941410746297226e-05, "loss": 0.8122, "step": 1143 }, { "epoch": 0.04, "learning_rate": 4.994122843889916e-05, "loss": 1.1579, "step": 1144 }, { "epoch": 0.04, "learning_rate": 4.994104584863979e-05, "loss": 1.1219, "step": 1145 }, { "epoch": 0.04, "learning_rate": 4.9940862975521175e-05, "loss": 1.1209, "step": 1146 }, { "epoch": 0.04, "learning_rate": 4.994067981954539e-05, "loss": 1.2457, "step": 1147 }, { "epoch": 0.04, "learning_rate": 4.994049638071452e-05, "loss": 0.9496, "step": 1148 }, { "epoch": 0.04, "learning_rate": 4.994031265903063e-05, "loss": 1.1184, "step": 1149 }, { "epoch": 0.04, "learning_rate": 4.994012865449582e-05, "loss": 1.1248, "step": 1150 }, { "epoch": 0.04, "learning_rate": 4.993994436711217e-05, "loss": 1.0949, "step": 1151 }, { "epoch": 0.04, "learning_rate": 4.993975979688177e-05, "loss": 1.0358, "step": 1152 }, { "epoch": 0.04, "learning_rate": 4.9939574943806706e-05, "loss": 0.9905, "step": 1153 }, { "epoch": 0.04, "learning_rate": 4.993938980788909e-05, "loss": 0.879, "step": 1154 }, { "epoch": 0.04, "learning_rate": 4.9939204389131004e-05, "loss": 0.9315, "step": 1155 }, { "epoch": 0.04, "learning_rate": 4.993901868753457e-05, "loss": 0.8385, "step": 1156 }, { "epoch": 0.04, "learning_rate": 4.9938832703101874e-05, "loss": 1.0189, "step": 1157 }, { "epoch": 0.04, "learning_rate": 4.993864643583504e-05, "loss": 1.0706, "step": 1158 }, { "epoch": 0.04, "learning_rate": 4.993845988573618e-05, "loss": 0.9937, "step": 1159 }, { "epoch": 0.04, "learning_rate": 4.99382730528074e-05, "loss": 1.0458, "step": 1160 }, { "epoch": 0.04, "learning_rate": 4.993808593705082e-05, "loss": 0.8601, "step": 1161 }, { "epoch": 0.04, "learning_rate": 4.993789853846858e-05, "loss": 0.8442, "step": 1162 }, { "epoch": 0.04, "learning_rate": 4.9937710857062775e-05, "loss": 0.893, "step": 1163 }, { "epoch": 0.04, "learning_rate": 4.993752289283556e-05, "loss": 0.7185, "step": 1164 }, { "epoch": 0.04, "learning_rate": 4.993733464578906e-05, "loss": 0.6028, "step": 1165 }, { "epoch": 0.04, "learning_rate": 4.9937146115925396e-05, "loss": 0.0194, "step": 1166 }, { "epoch": 0.04, "learning_rate": 4.993695730324672e-05, "loss": 0.0242, "step": 1167 }, { "epoch": 0.04, "learning_rate": 4.993676820775518e-05, "loss": 0.0152, "step": 1168 }, { "epoch": 0.04, "learning_rate": 4.99365788294529e-05, "loss": 0.3786, "step": 1169 }, { "epoch": 0.04, "learning_rate": 4.993638916834205e-05, "loss": 1.2387, "step": 1170 }, { "epoch": 0.04, "learning_rate": 4.9936199224424756e-05, "loss": 0.8125, "step": 1171 }, { "epoch": 0.04, "learning_rate": 4.9936008997703195e-05, "loss": 0.5072, "step": 1172 }, { "epoch": 0.04, "learning_rate": 4.993581848817951e-05, "loss": 0.4166, "step": 1173 }, { "epoch": 0.04, "learning_rate": 4.9935627695855866e-05, "loss": 0.3911, "step": 1174 }, { "epoch": 0.04, "learning_rate": 4.9935436620734436e-05, "loss": 2.5928, "step": 1175 }, { "epoch": 0.04, "learning_rate": 4.993524526281737e-05, "loss": 2.0296, "step": 1176 }, { "epoch": 0.04, "learning_rate": 4.993505362210685e-05, "loss": 0.9435, "step": 1177 }, { "epoch": 0.04, "learning_rate": 4.993486169860505e-05, "loss": 0.8242, "step": 1178 }, { "epoch": 0.04, "learning_rate": 4.993466949231413e-05, "loss": 0.9484, "step": 1179 }, { "epoch": 0.04, "learning_rate": 4.9934477003236294e-05, "loss": 0.2676, "step": 1180 }, { "epoch": 0.04, "learning_rate": 4.993428423137372e-05, "loss": 0.8271, "step": 1181 }, { "epoch": 0.04, "learning_rate": 4.993409117672857e-05, "loss": 1.1009, "step": 1182 }, { "epoch": 0.04, "learning_rate": 4.9933897839303065e-05, "loss": 0.8749, "step": 1183 }, { "epoch": 0.04, "learning_rate": 4.993370421909938e-05, "loss": 2.9929, "step": 1184 }, { "epoch": 0.04, "learning_rate": 4.9933510316119716e-05, "loss": 2.2705, "step": 1185 }, { "epoch": 0.04, "learning_rate": 4.993331613036627e-05, "loss": 1.0333, "step": 1186 }, { "epoch": 0.04, "learning_rate": 4.9933121661841244e-05, "loss": 1.0639, "step": 1187 }, { "epoch": 0.04, "learning_rate": 4.993292691054685e-05, "loss": 0.9613, "step": 1188 }, { "epoch": 0.04, "learning_rate": 4.993273187648528e-05, "loss": 1.0493, "step": 1189 }, { "epoch": 0.04, "learning_rate": 4.9932536559658767e-05, "loss": 0.9476, "step": 1190 }, { "epoch": 0.04, "learning_rate": 4.9932340960069513e-05, "loss": 0.9123, "step": 1191 }, { "epoch": 0.04, "learning_rate": 4.993214507771974e-05, "loss": 1.1115, "step": 1192 }, { "epoch": 0.04, "learning_rate": 4.993194891261167e-05, "loss": 0.9555, "step": 1193 }, { "epoch": 0.04, "learning_rate": 4.993175246474753e-05, "loss": 0.8114, "step": 1194 }, { "epoch": 0.04, "learning_rate": 4.9931555734129546e-05, "loss": 1.0436, "step": 1195 }, { "epoch": 0.04, "learning_rate": 4.993135872075994e-05, "loss": 1.0547, "step": 1196 }, { "epoch": 0.04, "learning_rate": 4.993116142464096e-05, "loss": 0.6865, "step": 1197 }, { "epoch": 0.04, "learning_rate": 4.993096384577484e-05, "loss": 0.1845, "step": 1198 }, { "epoch": 0.04, "learning_rate": 4.9930765984163804e-05, "loss": 1.0427, "step": 1199 }, { "epoch": 0.04, "learning_rate": 4.993056783981013e-05, "loss": 0.9595, "step": 1200 }, { "epoch": 0.04, "learning_rate": 4.993036941271604e-05, "loss": 0.3246, "step": 1201 }, { "epoch": 0.04, "learning_rate": 4.9930170702883786e-05, "loss": 0.1931, "step": 1202 }, { "epoch": 0.04, "learning_rate": 4.992997171031563e-05, "loss": 0.2012, "step": 1203 }, { "epoch": 0.04, "learning_rate": 4.992977243501382e-05, "loss": 0.6391, "step": 1204 }, { "epoch": 0.04, "learning_rate": 4.9929572876980624e-05, "loss": 1.0887, "step": 1205 }, { "epoch": 0.04, "learning_rate": 4.9929373036218305e-05, "loss": 0.9474, "step": 1206 }, { "epoch": 0.04, "learning_rate": 4.992917291272912e-05, "loss": 0.8244, "step": 1207 }, { "epoch": 0.04, "learning_rate": 4.992897250651535e-05, "loss": 1.3461, "step": 1208 }, { "epoch": 0.04, "learning_rate": 4.992877181757926e-05, "loss": 1.518, "step": 1209 }, { "epoch": 0.04, "learning_rate": 4.992857084592313e-05, "loss": 1.6051, "step": 1210 }, { "epoch": 0.04, "learning_rate": 4.992836959154924e-05, "loss": 1.756, "step": 1211 }, { "epoch": 0.04, "learning_rate": 4.992816805445987e-05, "loss": 1.6814, "step": 1212 }, { "epoch": 0.04, "learning_rate": 4.9927966234657306e-05, "loss": 1.3864, "step": 1213 }, { "epoch": 0.04, "learning_rate": 4.9927764132143836e-05, "loss": 0.8439, "step": 1214 }, { "epoch": 0.04, "learning_rate": 4.992756174692175e-05, "loss": 0.9397, "step": 1215 }, { "epoch": 0.04, "learning_rate": 4.992735907899335e-05, "loss": 1.0659, "step": 1216 }, { "epoch": 0.04, "learning_rate": 4.992715612836093e-05, "loss": 0.9282, "step": 1217 }, { "epoch": 0.04, "learning_rate": 4.992695289502679e-05, "loss": 0.9695, "step": 1218 }, { "epoch": 0.04, "learning_rate": 4.992674937899324e-05, "loss": 0.3807, "step": 1219 }, { "epoch": 0.04, "learning_rate": 4.992654558026259e-05, "loss": 1.3879, "step": 1220 }, { "epoch": 0.04, "learning_rate": 4.992634149883714e-05, "loss": 0.756, "step": 1221 }, { "epoch": 0.04, "learning_rate": 4.992613713471922e-05, "loss": 1.0646, "step": 1222 }, { "epoch": 0.04, "learning_rate": 4.9925932487911134e-05, "loss": 0.9037, "step": 1223 }, { "epoch": 0.04, "learning_rate": 4.99257275584152e-05, "loss": 0.8726, "step": 1224 }, { "epoch": 0.04, "learning_rate": 4.992552234623376e-05, "loss": 1.0681, "step": 1225 }, { "epoch": 0.04, "learning_rate": 4.992531685136913e-05, "loss": 0.9777, "step": 1226 }, { "epoch": 0.04, "learning_rate": 4.9925111073823635e-05, "loss": 1.1274, "step": 1227 }, { "epoch": 0.04, "learning_rate": 4.992490501359963e-05, "loss": 0.7641, "step": 1228 }, { "epoch": 0.04, "learning_rate": 4.9924698670699425e-05, "loss": 0.0801, "step": 1229 }, { "epoch": 0.04, "learning_rate": 4.992449204512538e-05, "loss": 0.0483, "step": 1230 }, { "epoch": 0.04, "learning_rate": 4.992428513687983e-05, "loss": 0.0435, "step": 1231 }, { "epoch": 0.04, "learning_rate": 4.992407794596512e-05, "loss": 1.3583, "step": 1232 }, { "epoch": 0.04, "learning_rate": 4.99238704723836e-05, "loss": 0.8749, "step": 1233 }, { "epoch": 0.04, "learning_rate": 4.992366271613763e-05, "loss": 1.0693, "step": 1234 }, { "epoch": 0.04, "learning_rate": 4.9923454677229565e-05, "loss": 1.1039, "step": 1235 }, { "epoch": 0.04, "learning_rate": 4.992324635566176e-05, "loss": 0.9452, "step": 1236 }, { "epoch": 0.04, "learning_rate": 4.9923037751436576e-05, "loss": 0.9702, "step": 1237 }, { "epoch": 0.04, "learning_rate": 4.992282886455638e-05, "loss": 1.1392, "step": 1238 }, { "epoch": 0.04, "learning_rate": 4.992261969502354e-05, "loss": 1.1463, "step": 1239 }, { "epoch": 0.04, "learning_rate": 4.992241024284043e-05, "loss": 1.0391, "step": 1240 }, { "epoch": 0.04, "learning_rate": 4.9922200508009434e-05, "loss": 0.9342, "step": 1241 }, { "epoch": 0.04, "learning_rate": 4.992199049053292e-05, "loss": 1.0399, "step": 1242 }, { "epoch": 0.04, "learning_rate": 4.9921780190413277e-05, "loss": 0.9033, "step": 1243 }, { "epoch": 0.04, "learning_rate": 4.992156960765288e-05, "loss": 0.9092, "step": 1244 }, { "epoch": 0.04, "learning_rate": 4.992135874225412e-05, "loss": 0.6527, "step": 1245 }, { "epoch": 0.04, "learning_rate": 4.99211475942194e-05, "loss": 1.1502, "step": 1246 }, { "epoch": 0.04, "learning_rate": 4.99209361635511e-05, "loss": 1.3126, "step": 1247 }, { "epoch": 0.04, "learning_rate": 4.992072445025163e-05, "loss": 0.9164, "step": 1248 }, { "epoch": 0.04, "learning_rate": 4.992051245432338e-05, "loss": 1.1828, "step": 1249 }, { "epoch": 0.04, "learning_rate": 4.992030017576875e-05, "loss": 0.878, "step": 1250 }, { "epoch": 0.04, "learning_rate": 4.9920087614590174e-05, "loss": 1.0192, "step": 1251 }, { "epoch": 0.04, "learning_rate": 4.991987477079004e-05, "loss": 2.4228, "step": 1252 }, { "epoch": 0.04, "learning_rate": 4.991966164437076e-05, "loss": 1.5922, "step": 1253 }, { "epoch": 0.04, "learning_rate": 4.991944823533476e-05, "loss": 0.5178, "step": 1254 }, { "epoch": 0.04, "learning_rate": 4.991923454368447e-05, "loss": 0.8466, "step": 1255 }, { "epoch": 0.04, "learning_rate": 4.991902056942229e-05, "loss": 0.8312, "step": 1256 }, { "epoch": 0.04, "learning_rate": 4.991880631255066e-05, "loss": 0.8743, "step": 1257 }, { "epoch": 0.04, "learning_rate": 4.991859177307202e-05, "loss": 0.6596, "step": 1258 }, { "epoch": 0.04, "learning_rate": 4.991837695098878e-05, "loss": 1.065, "step": 1259 }, { "epoch": 0.04, "learning_rate": 4.9918161846303394e-05, "loss": 0.6398, "step": 1260 }, { "epoch": 0.04, "learning_rate": 4.991794645901829e-05, "loss": 0.6722, "step": 1261 }, { "epoch": 0.04, "learning_rate": 4.991773078913592e-05, "loss": 0.6733, "step": 1262 }, { "epoch": 0.04, "learning_rate": 4.991751483665873e-05, "loss": 0.2395, "step": 1263 }, { "epoch": 0.04, "learning_rate": 4.991729860158916e-05, "loss": 0.3247, "step": 1264 }, { "epoch": 0.04, "learning_rate": 4.991708208392967e-05, "loss": 1.0852, "step": 1265 }, { "epoch": 0.04, "learning_rate": 4.991686528368271e-05, "loss": 0.9783, "step": 1266 }, { "epoch": 0.04, "learning_rate": 4.991664820085074e-05, "loss": 0.8992, "step": 1267 }, { "epoch": 0.04, "learning_rate": 4.991643083543624e-05, "loss": 1.1258, "step": 1268 }, { "epoch": 0.04, "learning_rate": 4.991621318744165e-05, "loss": 1.0352, "step": 1269 }, { "epoch": 0.04, "learning_rate": 4.9915995256869444e-05, "loss": 1.0848, "step": 1270 }, { "epoch": 0.04, "learning_rate": 4.9915777043722104e-05, "loss": 0.5921, "step": 1271 }, { "epoch": 0.04, "learning_rate": 4.991555854800209e-05, "loss": 1.0344, "step": 1272 }, { "epoch": 0.04, "learning_rate": 4.99153397697119e-05, "loss": 0.9875, "step": 1273 }, { "epoch": 0.04, "learning_rate": 4.991512070885399e-05, "loss": 1.1137, "step": 1274 }, { "epoch": 0.04, "learning_rate": 4.9914901365430866e-05, "loss": 0.9504, "step": 1275 }, { "epoch": 0.04, "learning_rate": 4.9914681739445004e-05, "loss": 1.5212, "step": 1276 }, { "epoch": 0.04, "learning_rate": 4.99144618308989e-05, "loss": 3.5277, "step": 1277 }, { "epoch": 0.04, "learning_rate": 4.991424163979504e-05, "loss": 4.272, "step": 1278 }, { "epoch": 0.04, "learning_rate": 4.991402116613593e-05, "loss": 0.8987, "step": 1279 }, { "epoch": 0.04, "learning_rate": 4.991380040992407e-05, "loss": 0.7538, "step": 1280 }, { "epoch": 0.04, "learning_rate": 4.991357937116196e-05, "loss": 0.7749, "step": 1281 }, { "epoch": 0.04, "learning_rate": 4.991335804985211e-05, "loss": 1.0445, "step": 1282 }, { "epoch": 0.04, "learning_rate": 4.9913136445997026e-05, "loss": 0.9559, "step": 1283 }, { "epoch": 0.04, "learning_rate": 4.991291455959922e-05, "loss": 0.6606, "step": 1284 }, { "epoch": 0.04, "learning_rate": 4.991269239066122e-05, "loss": 0.1803, "step": 1285 }, { "epoch": 0.04, "learning_rate": 4.991246993918553e-05, "loss": 0.1935, "step": 1286 }, { "epoch": 0.04, "learning_rate": 4.991224720517469e-05, "loss": 0.7538, "step": 1287 }, { "epoch": 0.04, "learning_rate": 4.991202418863121e-05, "loss": 0.4911, "step": 1288 }, { "epoch": 0.04, "learning_rate": 4.9911800889557625e-05, "loss": 0.4034, "step": 1289 }, { "epoch": 0.04, "learning_rate": 4.991157730795647e-05, "loss": 0.6197, "step": 1290 }, { "epoch": 0.04, "learning_rate": 4.991135344383028e-05, "loss": 0.6044, "step": 1291 }, { "epoch": 0.04, "learning_rate": 4.991112929718159e-05, "loss": 1.0932, "step": 1292 }, { "epoch": 0.04, "learning_rate": 4.9910904868012945e-05, "loss": 1.1647, "step": 1293 }, { "epoch": 0.04, "learning_rate": 4.9910680156326885e-05, "loss": 1.1889, "step": 1294 }, { "epoch": 0.04, "learning_rate": 4.991045516212597e-05, "loss": 0.3902, "step": 1295 }, { "epoch": 0.04, "learning_rate": 4.991022988541274e-05, "loss": 0.9406, "step": 1296 }, { "epoch": 0.04, "learning_rate": 4.9910004326189765e-05, "loss": 0.6146, "step": 1297 }, { "epoch": 0.04, "learning_rate": 4.990977848445959e-05, "loss": 0.3704, "step": 1298 }, { "epoch": 0.04, "learning_rate": 4.990955236022477e-05, "loss": 0.6857, "step": 1299 }, { "epoch": 0.04, "learning_rate": 4.990932595348789e-05, "loss": 0.6499, "step": 1300 }, { "epoch": 0.04, "learning_rate": 4.9909099264251495e-05, "loss": 0.7977, "step": 1301 }, { "epoch": 0.04, "learning_rate": 4.990887229251817e-05, "loss": 0.747, "step": 1302 }, { "epoch": 0.04, "learning_rate": 4.99086450382905e-05, "loss": 1.0061, "step": 1303 }, { "epoch": 0.04, "learning_rate": 4.990841750157104e-05, "loss": 0.9476, "step": 1304 }, { "epoch": 0.04, "learning_rate": 4.990818968236237e-05, "loss": 0.7076, "step": 1305 }, { "epoch": 0.04, "learning_rate": 4.9907961580667097e-05, "loss": 1.0782, "step": 1306 }, { "epoch": 0.04, "learning_rate": 4.990773319648779e-05, "loss": 1.1651, "step": 1307 }, { "epoch": 0.04, "learning_rate": 4.9907504529827044e-05, "loss": 1.0496, "step": 1308 }, { "epoch": 0.04, "learning_rate": 4.990727558068745e-05, "loss": 0.7212, "step": 1309 }, { "epoch": 0.04, "learning_rate": 4.990704634907161e-05, "loss": 1.1282, "step": 1310 }, { "epoch": 0.04, "learning_rate": 4.990681683498211e-05, "loss": 1.0484, "step": 1311 }, { "epoch": 0.04, "learning_rate": 4.990658703842157e-05, "loss": 1.0725, "step": 1312 }, { "epoch": 0.04, "learning_rate": 4.9906356959392586e-05, "loss": 0.9914, "step": 1313 }, { "epoch": 0.04, "learning_rate": 4.990612659789777e-05, "loss": 1.0545, "step": 1314 }, { "epoch": 0.04, "learning_rate": 4.990589595393974e-05, "loss": 0.8583, "step": 1315 }, { "epoch": 0.04, "learning_rate": 4.9905665027521107e-05, "loss": 0.9565, "step": 1316 }, { "epoch": 0.04, "learning_rate": 4.9905433818644484e-05, "loss": 0.5485, "step": 1317 }, { "epoch": 0.04, "learning_rate": 4.99052023273125e-05, "loss": 1.1803, "step": 1318 }, { "epoch": 0.04, "learning_rate": 4.990497055352778e-05, "loss": 0.9748, "step": 1319 }, { "epoch": 0.04, "learning_rate": 4.990473849729295e-05, "loss": 0.8765, "step": 1320 }, { "epoch": 0.04, "learning_rate": 4.9904506158610646e-05, "loss": 0.1, "step": 1321 }, { "epoch": 0.04, "learning_rate": 4.9904273537483505e-05, "loss": 0.3819, "step": 1322 }, { "epoch": 0.04, "learning_rate": 4.990404063391415e-05, "loss": 1.0074, "step": 1323 }, { "epoch": 0.04, "learning_rate": 4.9903807447905236e-05, "loss": 1.0199, "step": 1324 }, { "epoch": 0.04, "learning_rate": 4.9903573979459414e-05, "loss": 1.0891, "step": 1325 }, { "epoch": 0.04, "learning_rate": 4.990334022857932e-05, "loss": 1.2718, "step": 1326 }, { "epoch": 0.04, "learning_rate": 4.990310619526759e-05, "loss": 0.9434, "step": 1327 }, { "epoch": 0.04, "learning_rate": 4.9902871879526913e-05, "loss": 1.0431, "step": 1328 }, { "epoch": 0.04, "learning_rate": 4.9902637281359924e-05, "loss": 0.9503, "step": 1329 }, { "epoch": 0.04, "learning_rate": 4.9902402400769286e-05, "loss": 1.1888, "step": 1330 }, { "epoch": 0.04, "learning_rate": 4.990216723775768e-05, "loss": 0.9416, "step": 1331 }, { "epoch": 0.04, "learning_rate": 4.990193179232774e-05, "loss": 1.3064, "step": 1332 }, { "epoch": 0.04, "learning_rate": 4.9901696064482165e-05, "loss": 0.8979, "step": 1333 }, { "epoch": 0.04, "learning_rate": 4.9901460054223614e-05, "loss": 0.8765, "step": 1334 }, { "epoch": 0.04, "learning_rate": 4.9901223761554765e-05, "loss": 1.1112, "step": 1335 }, { "epoch": 0.04, "learning_rate": 4.990098718647831e-05, "loss": 1.2958, "step": 1336 }, { "epoch": 0.04, "learning_rate": 4.990075032899691e-05, "loss": 1.6193, "step": 1337 }, { "epoch": 0.04, "learning_rate": 4.9900513189113275e-05, "loss": 0.5965, "step": 1338 }, { "epoch": 0.04, "learning_rate": 4.990027576683007e-05, "loss": 0.9657, "step": 1339 }, { "epoch": 0.04, "learning_rate": 4.990003806215001e-05, "loss": 0.853, "step": 1340 }, { "epoch": 0.04, "learning_rate": 4.989980007507579e-05, "loss": 2.1462, "step": 1341 }, { "epoch": 0.04, "learning_rate": 4.989956180561009e-05, "loss": 1.2795, "step": 1342 }, { "epoch": 0.04, "learning_rate": 4.989932325375562e-05, "loss": 0.9189, "step": 1343 }, { "epoch": 0.04, "learning_rate": 4.98990844195151e-05, "loss": 0.9098, "step": 1344 }, { "epoch": 0.04, "learning_rate": 4.989884530289121e-05, "loss": 0.5024, "step": 1345 }, { "epoch": 0.04, "learning_rate": 4.989860590388669e-05, "loss": 0.8822, "step": 1346 }, { "epoch": 0.04, "learning_rate": 4.989836622250424e-05, "loss": 1.0274, "step": 1347 }, { "epoch": 0.04, "learning_rate": 4.989812625874659e-05, "loss": 0.9965, "step": 1348 }, { "epoch": 0.04, "learning_rate": 4.9897886012616444e-05, "loss": 0.9971, "step": 1349 }, { "epoch": 0.04, "learning_rate": 4.989764548411654e-05, "loss": 1.18, "step": 1350 }, { "epoch": 0.05, "learning_rate": 4.98974046732496e-05, "loss": 0.9868, "step": 1351 }, { "epoch": 0.05, "learning_rate": 4.9897163580018366e-05, "loss": 0.981, "step": 1352 }, { "epoch": 0.05, "learning_rate": 4.9896922204425554e-05, "loss": 0.9864, "step": 1353 }, { "epoch": 0.05, "learning_rate": 4.9896680546473916e-05, "loss": 0.7005, "step": 1354 }, { "epoch": 0.05, "learning_rate": 4.989643860616618e-05, "loss": 0.7474, "step": 1355 }, { "epoch": 0.05, "learning_rate": 4.989619638350511e-05, "loss": 0.2252, "step": 1356 }, { "epoch": 0.05, "learning_rate": 4.9895953878493434e-05, "loss": 0.9851, "step": 1357 }, { "epoch": 0.05, "learning_rate": 4.9895711091133915e-05, "loss": 1.3878, "step": 1358 }, { "epoch": 0.05, "learning_rate": 4.9895468021429286e-05, "loss": 0.9857, "step": 1359 }, { "epoch": 0.05, "learning_rate": 4.9895224669382324e-05, "loss": 0.68, "step": 1360 }, { "epoch": 0.05, "learning_rate": 4.9894981034995794e-05, "loss": 0.3798, "step": 1361 }, { "epoch": 0.05, "learning_rate": 4.989473711827243e-05, "loss": 0.9092, "step": 1362 }, { "epoch": 0.05, "learning_rate": 4.989449291921503e-05, "loss": 1.2222, "step": 1363 }, { "epoch": 0.05, "learning_rate": 4.989424843782634e-05, "loss": 1.6918, "step": 1364 }, { "epoch": 0.05, "learning_rate": 4.989400367410915e-05, "loss": 0.8202, "step": 1365 }, { "epoch": 0.05, "learning_rate": 4.989375862806622e-05, "loss": 0.3283, "step": 1366 }, { "epoch": 0.05, "learning_rate": 4.989351329970034e-05, "loss": 0.5602, "step": 1367 }, { "epoch": 0.05, "learning_rate": 4.98932676890143e-05, "loss": 1.3094, "step": 1368 }, { "epoch": 0.05, "learning_rate": 4.989302179601087e-05, "loss": 0.7332, "step": 1369 }, { "epoch": 0.05, "learning_rate": 4.9892775620692844e-05, "loss": 0.7788, "step": 1370 }, { "epoch": 0.05, "learning_rate": 4.989252916306301e-05, "loss": 0.9668, "step": 1371 }, { "epoch": 0.05, "learning_rate": 4.989228242312417e-05, "loss": 0.9729, "step": 1372 }, { "epoch": 0.05, "learning_rate": 4.989203540087911e-05, "loss": 0.7261, "step": 1373 }, { "epoch": 0.05, "learning_rate": 4.9891788096330654e-05, "loss": 1.0176, "step": 1374 }, { "epoch": 0.05, "learning_rate": 4.989154050948159e-05, "loss": 0.9757, "step": 1375 }, { "epoch": 0.05, "learning_rate": 4.989129264033472e-05, "loss": 0.9224, "step": 1376 }, { "epoch": 0.05, "learning_rate": 4.989104448889288e-05, "loss": 0.7957, "step": 1377 }, { "epoch": 0.05, "learning_rate": 4.989079605515886e-05, "loss": 0.7235, "step": 1378 }, { "epoch": 0.05, "learning_rate": 4.9890547339135486e-05, "loss": 0.141, "step": 1379 }, { "epoch": 0.05, "learning_rate": 4.989029834082558e-05, "loss": 0.2266, "step": 1380 }, { "epoch": 0.05, "learning_rate": 4.989004906023196e-05, "loss": 0.2316, "step": 1381 }, { "epoch": 0.05, "learning_rate": 4.988979949735747e-05, "loss": 0.6501, "step": 1382 }, { "epoch": 0.05, "learning_rate": 4.988954965220491e-05, "loss": 0.8336, "step": 1383 }, { "epoch": 0.05, "learning_rate": 4.988929952477715e-05, "loss": 0.9508, "step": 1384 }, { "epoch": 0.05, "learning_rate": 4.9889049115077005e-05, "loss": 0.8577, "step": 1385 }, { "epoch": 0.05, "learning_rate": 4.988879842310731e-05, "loss": 1.1633, "step": 1386 }, { "epoch": 0.05, "learning_rate": 4.9888547448870925e-05, "loss": 1.6373, "step": 1387 }, { "epoch": 0.05, "learning_rate": 4.9888296192370686e-05, "loss": 1.0132, "step": 1388 }, { "epoch": 0.05, "learning_rate": 4.9888044653609444e-05, "loss": 1.15, "step": 1389 }, { "epoch": 0.05, "learning_rate": 4.988779283259005e-05, "loss": 0.8613, "step": 1390 }, { "epoch": 0.05, "learning_rate": 4.988754072931536e-05, "loss": 1.1475, "step": 1391 }, { "epoch": 0.05, "learning_rate": 4.9887288343788244e-05, "loss": 0.9769, "step": 1392 }, { "epoch": 0.05, "learning_rate": 4.988703567601155e-05, "loss": 1.5783, "step": 1393 }, { "epoch": 0.05, "learning_rate": 4.9886782725988153e-05, "loss": 2.3945, "step": 1394 }, { "epoch": 0.05, "learning_rate": 4.988652949372091e-05, "loss": 3.0384, "step": 1395 }, { "epoch": 0.05, "learning_rate": 4.988627597921271e-05, "loss": 2.4414, "step": 1396 }, { "epoch": 0.05, "learning_rate": 4.988602218246641e-05, "loss": 1.4148, "step": 1397 }, { "epoch": 0.05, "learning_rate": 4.988576810348491e-05, "loss": 1.2044, "step": 1398 }, { "epoch": 0.05, "learning_rate": 4.988551374227107e-05, "loss": 1.0139, "step": 1399 }, { "epoch": 0.05, "learning_rate": 4.988525909882779e-05, "loss": 1.0169, "step": 1400 }, { "epoch": 0.05, "learning_rate": 4.9885004173157934e-05, "loss": 1.0097, "step": 1401 }, { "epoch": 0.05, "learning_rate": 4.988474896526443e-05, "loss": 0.9927, "step": 1402 }, { "epoch": 0.05, "learning_rate": 4.988449347515014e-05, "loss": 1.1575, "step": 1403 }, { "epoch": 0.05, "learning_rate": 4.988423770281798e-05, "loss": 0.6286, "step": 1404 }, { "epoch": 0.05, "learning_rate": 4.988398164827085e-05, "loss": 0.4965, "step": 1405 }, { "epoch": 0.05, "learning_rate": 4.988372531151164e-05, "loss": 0.5469, "step": 1406 }, { "epoch": 0.05, "learning_rate": 4.9883468692543276e-05, "loss": 0.7861, "step": 1407 }, { "epoch": 0.05, "learning_rate": 4.988321179136865e-05, "loss": 1.1328, "step": 1408 }, { "epoch": 0.05, "learning_rate": 4.988295460799069e-05, "loss": 0.9624, "step": 1409 }, { "epoch": 0.05, "learning_rate": 4.988269714241231e-05, "loss": 0.9233, "step": 1410 }, { "epoch": 0.05, "learning_rate": 4.988243939463642e-05, "loss": 0.8709, "step": 1411 }, { "epoch": 0.05, "learning_rate": 4.9882181364665956e-05, "loss": 0.8607, "step": 1412 }, { "epoch": 0.05, "learning_rate": 4.988192305250383e-05, "loss": 1.0129, "step": 1413 }, { "epoch": 0.05, "learning_rate": 4.988166445815299e-05, "loss": 1.0198, "step": 1414 }, { "epoch": 0.05, "learning_rate": 4.988140558161635e-05, "loss": 1.0518, "step": 1415 }, { "epoch": 0.05, "learning_rate": 4.9881146422896864e-05, "loss": 1.1047, "step": 1416 }, { "epoch": 0.05, "learning_rate": 4.9880886981997456e-05, "loss": 0.8122, "step": 1417 }, { "epoch": 0.05, "learning_rate": 4.988062725892108e-05, "loss": 0.3187, "step": 1418 }, { "epoch": 0.05, "learning_rate": 4.988036725367066e-05, "loss": 0.2918, "step": 1419 }, { "epoch": 0.05, "learning_rate": 4.9880106966249176e-05, "loss": 1.0519, "step": 1420 }, { "epoch": 0.05, "learning_rate": 4.9879846396659555e-05, "loss": 1.6364, "step": 1421 }, { "epoch": 0.05, "learning_rate": 4.9879585544904774e-05, "loss": 1.8557, "step": 1422 }, { "epoch": 0.05, "learning_rate": 4.9879324410987766e-05, "loss": 0.9235, "step": 1423 }, { "epoch": 0.05, "learning_rate": 4.987906299491151e-05, "loss": 0.2305, "step": 1424 }, { "epoch": 0.05, "learning_rate": 4.9878801296678965e-05, "loss": 0.3089, "step": 1425 }, { "epoch": 0.05, "learning_rate": 4.987853931629311e-05, "loss": 0.8169, "step": 1426 }, { "epoch": 0.05, "learning_rate": 4.9878277053756895e-05, "loss": 1.0241, "step": 1427 }, { "epoch": 0.05, "learning_rate": 4.987801450907331e-05, "loss": 0.8808, "step": 1428 }, { "epoch": 0.05, "learning_rate": 4.987775168224533e-05, "loss": 1.0808, "step": 1429 }, { "epoch": 0.05, "learning_rate": 4.987748857327593e-05, "loss": 1.2501, "step": 1430 }, { "epoch": 0.05, "learning_rate": 4.9877225182168095e-05, "loss": 0.974, "step": 1431 }, { "epoch": 0.05, "learning_rate": 4.987696150892482e-05, "loss": 0.9767, "step": 1432 }, { "epoch": 0.05, "learning_rate": 4.987669755354909e-05, "loss": 0.9247, "step": 1433 }, { "epoch": 0.05, "learning_rate": 4.98764333160439e-05, "loss": 0.7467, "step": 1434 }, { "epoch": 0.05, "learning_rate": 4.987616879641225e-05, "loss": 0.9277, "step": 1435 }, { "epoch": 0.05, "learning_rate": 4.9875903994657125e-05, "loss": 0.8298, "step": 1436 }, { "epoch": 0.05, "learning_rate": 4.9875638910781546e-05, "loss": 1.2742, "step": 1437 }, { "epoch": 0.05, "learning_rate": 4.987537354478851e-05, "loss": 1.1217, "step": 1438 }, { "epoch": 0.05, "learning_rate": 4.987510789668103e-05, "loss": 1.069, "step": 1439 }, { "epoch": 0.05, "learning_rate": 4.987484196646212e-05, "loss": 1.156, "step": 1440 }, { "epoch": 0.05, "learning_rate": 4.987457575413479e-05, "loss": 1.013, "step": 1441 }, { "epoch": 0.05, "learning_rate": 4.987430925970206e-05, "loss": 0.7779, "step": 1442 }, { "epoch": 0.05, "learning_rate": 4.987404248316696e-05, "loss": 1.2606, "step": 1443 }, { "epoch": 0.05, "learning_rate": 4.987377542453251e-05, "loss": 1.819, "step": 1444 }, { "epoch": 0.05, "learning_rate": 4.987350808380173e-05, "loss": 1.3703, "step": 1445 }, { "epoch": 0.05, "learning_rate": 4.987324046097767e-05, "loss": 1.3459, "step": 1446 }, { "epoch": 0.05, "learning_rate": 4.987297255606336e-05, "loss": 0.9806, "step": 1447 }, { "epoch": 0.05, "learning_rate": 4.987270436906183e-05, "loss": 1.0654, "step": 1448 }, { "epoch": 0.05, "learning_rate": 4.987243589997612e-05, "loss": 0.8681, "step": 1449 }, { "epoch": 0.05, "learning_rate": 4.987216714880929e-05, "loss": 0.4193, "step": 1450 }, { "epoch": 0.05, "learning_rate": 4.987189811556438e-05, "loss": 0.9448, "step": 1451 }, { "epoch": 0.05, "learning_rate": 4.987162880024443e-05, "loss": 0.7649, "step": 1452 }, { "epoch": 0.05, "learning_rate": 4.9871359202852505e-05, "loss": 0.6175, "step": 1453 }, { "epoch": 0.05, "learning_rate": 4.987108932339167e-05, "loss": 0.4866, "step": 1454 }, { "epoch": 0.05, "learning_rate": 4.9870819161864976e-05, "loss": 0.8662, "step": 1455 }, { "epoch": 0.05, "learning_rate": 4.987054871827549e-05, "loss": 0.8377, "step": 1456 }, { "epoch": 0.05, "learning_rate": 4.987027799262627e-05, "loss": 0.8554, "step": 1457 }, { "epoch": 0.05, "learning_rate": 4.98700069849204e-05, "loss": 0.7403, "step": 1458 }, { "epoch": 0.05, "learning_rate": 4.9869735695160946e-05, "loss": 0.9973, "step": 1459 }, { "epoch": 0.05, "learning_rate": 4.9869464123350985e-05, "loss": 1.3116, "step": 1460 }, { "epoch": 0.05, "learning_rate": 4.98691922694936e-05, "loss": 0.8716, "step": 1461 }, { "epoch": 0.05, "learning_rate": 4.986892013359187e-05, "loss": 0.6008, "step": 1462 }, { "epoch": 0.05, "learning_rate": 4.986864771564889e-05, "loss": 1.0457, "step": 1463 }, { "epoch": 0.05, "learning_rate": 4.9868375015667735e-05, "loss": 0.8559, "step": 1464 }, { "epoch": 0.05, "learning_rate": 4.986810203365151e-05, "loss": 1.0727, "step": 1465 }, { "epoch": 0.05, "learning_rate": 4.986782876960331e-05, "loss": 1.211, "step": 1466 }, { "epoch": 0.05, "learning_rate": 4.986755522352622e-05, "loss": 3.3965, "step": 1467 }, { "epoch": 0.05, "learning_rate": 4.986728139542336e-05, "loss": 1.1333, "step": 1468 }, { "epoch": 0.05, "learning_rate": 4.986700728529783e-05, "loss": 1.1514, "step": 1469 }, { "epoch": 0.05, "learning_rate": 4.9866732893152735e-05, "loss": 0.9949, "step": 1470 }, { "epoch": 0.05, "learning_rate": 4.98664582189912e-05, "loss": 1.3149, "step": 1471 }, { "epoch": 0.05, "learning_rate": 4.986618326281631e-05, "loss": 1.3411, "step": 1472 }, { "epoch": 0.05, "learning_rate": 4.986590802463121e-05, "loss": 1.3611, "step": 1473 }, { "epoch": 0.05, "learning_rate": 4.986563250443902e-05, "loss": 1.3195, "step": 1474 }, { "epoch": 0.05, "learning_rate": 4.986535670224286e-05, "loss": 1.3131, "step": 1475 }, { "epoch": 0.05, "learning_rate": 4.9865080618045854e-05, "loss": 1.1951, "step": 1476 }, { "epoch": 0.05, "learning_rate": 4.9864804251851134e-05, "loss": 1.1641, "step": 1477 }, { "epoch": 0.05, "learning_rate": 4.986452760366184e-05, "loss": 4.2416, "step": 1478 }, { "epoch": 0.05, "learning_rate": 4.98642506734811e-05, "loss": 5.8442, "step": 1479 }, { "epoch": 0.05, "learning_rate": 4.986397346131206e-05, "loss": 5.7912, "step": 1480 }, { "epoch": 0.05, "learning_rate": 4.9863695967157874e-05, "loss": 5.7445, "step": 1481 }, { "epoch": 0.05, "learning_rate": 4.9863418191021675e-05, "loss": 5.2343, "step": 1482 }, { "epoch": 0.05, "learning_rate": 4.9863140132906615e-05, "loss": 1.0116, "step": 1483 }, { "epoch": 0.05, "learning_rate": 4.986286179281585e-05, "loss": 1.444, "step": 1484 }, { "epoch": 0.05, "learning_rate": 4.986258317075254e-05, "loss": 1.0526, "step": 1485 }, { "epoch": 0.05, "learning_rate": 4.9862304266719836e-05, "loss": 1.1327, "step": 1486 }, { "epoch": 0.05, "learning_rate": 4.986202508072091e-05, "loss": 0.9781, "step": 1487 }, { "epoch": 0.05, "learning_rate": 4.986174561275893e-05, "loss": 0.8287, "step": 1488 }, { "epoch": 0.05, "learning_rate": 4.986146586283705e-05, "loss": 1.0006, "step": 1489 }, { "epoch": 0.05, "learning_rate": 4.9861185830958466e-05, "loss": 1.0323, "step": 1490 }, { "epoch": 0.05, "learning_rate": 4.9860905517126336e-05, "loss": 0.5915, "step": 1491 }, { "epoch": 0.05, "learning_rate": 4.9860624921343846e-05, "loss": 0.593, "step": 1492 }, { "epoch": 0.05, "learning_rate": 4.986034404361417e-05, "loss": 0.7546, "step": 1493 }, { "epoch": 0.05, "learning_rate": 4.986006288394051e-05, "loss": 0.9171, "step": 1494 }, { "epoch": 0.05, "learning_rate": 4.9859781442326034e-05, "loss": 1.0336, "step": 1495 }, { "epoch": 0.05, "learning_rate": 4.985949971877395e-05, "loss": 1.0336, "step": 1496 }, { "epoch": 0.05, "learning_rate": 4.985921771328745e-05, "loss": 1.0008, "step": 1497 }, { "epoch": 0.05, "learning_rate": 4.985893542586972e-05, "loss": 0.8483, "step": 1498 }, { "epoch": 0.05, "learning_rate": 4.9858652856523985e-05, "loss": 0.6683, "step": 1499 }, { "epoch": 0.05, "learning_rate": 4.985837000525343e-05, "loss": 0.5985, "step": 1500 }, { "epoch": 0.05, "learning_rate": 4.985808687206127e-05, "loss": 0.9305, "step": 1501 }, { "epoch": 0.05, "learning_rate": 4.985780345695071e-05, "loss": 0.9598, "step": 1502 }, { "epoch": 0.05, "learning_rate": 4.9857519759924974e-05, "loss": 0.9984, "step": 1503 }, { "epoch": 0.05, "learning_rate": 4.985723578098728e-05, "loss": 0.9376, "step": 1504 }, { "epoch": 0.05, "learning_rate": 4.985695152014083e-05, "loss": 0.964, "step": 1505 }, { "epoch": 0.05, "learning_rate": 4.9856666977388865e-05, "loss": 1.0308, "step": 1506 }, { "epoch": 0.05, "learning_rate": 4.985638215273461e-05, "loss": 1.1223, "step": 1507 }, { "epoch": 0.05, "learning_rate": 4.985609704618129e-05, "loss": 0.7733, "step": 1508 }, { "epoch": 0.05, "learning_rate": 4.985581165773216e-05, "loss": 0.3021, "step": 1509 }, { "epoch": 0.05, "learning_rate": 4.985552598739042e-05, "loss": 0.0499, "step": 1510 }, { "epoch": 0.05, "learning_rate": 4.985524003515932e-05, "loss": 0.0203, "step": 1511 }, { "epoch": 0.05, "learning_rate": 4.985495380104213e-05, "loss": 0.3669, "step": 1512 }, { "epoch": 0.05, "learning_rate": 4.985466728504207e-05, "loss": 1.2336, "step": 1513 }, { "epoch": 0.05, "learning_rate": 4.98543804871624e-05, "loss": 0.8988, "step": 1514 }, { "epoch": 0.05, "learning_rate": 4.985409340740636e-05, "loss": 0.9802, "step": 1515 }, { "epoch": 0.05, "learning_rate": 4.985380604577722e-05, "loss": 0.4828, "step": 1516 }, { "epoch": 0.05, "learning_rate": 4.985351840227824e-05, "loss": 0.0573, "step": 1517 }, { "epoch": 0.05, "learning_rate": 4.985323047691267e-05, "loss": 0.7636, "step": 1518 }, { "epoch": 0.05, "learning_rate": 4.985294226968379e-05, "loss": 0.973, "step": 1519 }, { "epoch": 0.05, "learning_rate": 4.985265378059485e-05, "loss": 0.8547, "step": 1520 }, { "epoch": 0.05, "learning_rate": 4.985236500964914e-05, "loss": 0.6839, "step": 1521 }, { "epoch": 0.05, "learning_rate": 4.985207595684992e-05, "loss": 0.4831, "step": 1522 }, { "epoch": 0.05, "learning_rate": 4.9851786622200484e-05, "loss": 0.9678, "step": 1523 }, { "epoch": 0.05, "learning_rate": 4.9851497005704097e-05, "loss": 1.8123, "step": 1524 }, { "epoch": 0.05, "learning_rate": 4.985120710736405e-05, "loss": 1.2979, "step": 1525 }, { "epoch": 0.05, "learning_rate": 4.985091692718364e-05, "loss": 1.372, "step": 1526 }, { "epoch": 0.05, "learning_rate": 4.9850626465166145e-05, "loss": 1.5735, "step": 1527 }, { "epoch": 0.05, "learning_rate": 4.985033572131487e-05, "loss": 1.616, "step": 1528 }, { "epoch": 0.05, "learning_rate": 4.9850044695633105e-05, "loss": 1.7601, "step": 1529 }, { "epoch": 0.05, "learning_rate": 4.984975338812414e-05, "loss": 1.697, "step": 1530 }, { "epoch": 0.05, "learning_rate": 4.98494617987913e-05, "loss": 1.0492, "step": 1531 }, { "epoch": 0.05, "learning_rate": 4.984916992763788e-05, "loss": 1.0786, "step": 1532 }, { "epoch": 0.05, "learning_rate": 4.98488777746672e-05, "loss": 0.807, "step": 1533 }, { "epoch": 0.05, "learning_rate": 4.984858533988256e-05, "loss": 0.601, "step": 1534 }, { "epoch": 0.05, "learning_rate": 4.984829262328728e-05, "loss": 0.7548, "step": 1535 }, { "epoch": 0.05, "learning_rate": 4.98479996248847e-05, "loss": 1.0541, "step": 1536 }, { "epoch": 0.05, "learning_rate": 4.9847706344678106e-05, "loss": 1.0339, "step": 1537 }, { "epoch": 0.05, "learning_rate": 4.984741278267086e-05, "loss": 0.8743, "step": 1538 }, { "epoch": 0.05, "learning_rate": 4.9847118938866255e-05, "loss": 1.7654, "step": 1539 }, { "epoch": 0.05, "learning_rate": 4.984682481326766e-05, "loss": 1.1901, "step": 1540 }, { "epoch": 0.05, "learning_rate": 4.984653040587839e-05, "loss": 1.0885, "step": 1541 }, { "epoch": 0.05, "learning_rate": 4.984623571670178e-05, "loss": 0.9764, "step": 1542 }, { "epoch": 0.05, "learning_rate": 4.984594074574119e-05, "loss": 1.1157, "step": 1543 }, { "epoch": 0.05, "learning_rate": 4.984564549299995e-05, "loss": 0.9492, "step": 1544 }, { "epoch": 0.05, "learning_rate": 4.984534995848141e-05, "loss": 0.7835, "step": 1545 }, { "epoch": 0.05, "learning_rate": 4.9845054142188936e-05, "loss": 0.2739, "step": 1546 }, { "epoch": 0.05, "learning_rate": 4.984475804412587e-05, "loss": 0.3971, "step": 1547 }, { "epoch": 0.05, "learning_rate": 4.984446166429557e-05, "loss": 0.9957, "step": 1548 }, { "epoch": 0.05, "learning_rate": 4.9844165002701395e-05, "loss": 1.0274, "step": 1549 }, { "epoch": 0.05, "learning_rate": 4.984386805934672e-05, "loss": 0.3857, "step": 1550 }, { "epoch": 0.05, "learning_rate": 4.984357083423491e-05, "loss": 0.4797, "step": 1551 }, { "epoch": 0.05, "learning_rate": 4.9843273327369334e-05, "loss": 1.0098, "step": 1552 }, { "epoch": 0.05, "learning_rate": 4.9842975538753357e-05, "loss": 0.8714, "step": 1553 }, { "epoch": 0.05, "learning_rate": 4.984267746839036e-05, "loss": 0.7299, "step": 1554 }, { "epoch": 0.05, "learning_rate": 4.9842379116283744e-05, "loss": 1.0099, "step": 1555 }, { "epoch": 0.05, "learning_rate": 4.984208048243687e-05, "loss": 0.8071, "step": 1556 }, { "epoch": 0.05, "learning_rate": 4.984178156685313e-05, "loss": 0.5762, "step": 1557 }, { "epoch": 0.05, "learning_rate": 4.984148236953591e-05, "loss": 1.1744, "step": 1558 }, { "epoch": 0.05, "learning_rate": 4.9841182890488615e-05, "loss": 0.2281, "step": 1559 }, { "epoch": 0.05, "learning_rate": 4.984088312971463e-05, "loss": 0.0903, "step": 1560 }, { "epoch": 0.05, "learning_rate": 4.984058308721736e-05, "loss": 0.5313, "step": 1561 }, { "epoch": 0.05, "learning_rate": 4.984028276300021e-05, "loss": 2.2169, "step": 1562 }, { "epoch": 0.05, "learning_rate": 4.983998215706659e-05, "loss": 0.7238, "step": 1563 }, { "epoch": 0.05, "learning_rate": 4.9839681269419894e-05, "loss": 1.0911, "step": 1564 }, { "epoch": 0.05, "learning_rate": 4.983938010006354e-05, "loss": 0.8726, "step": 1565 }, { "epoch": 0.05, "learning_rate": 4.983907864900095e-05, "loss": 0.7078, "step": 1566 }, { "epoch": 0.05, "learning_rate": 4.983877691623554e-05, "loss": 1.1525, "step": 1567 }, { "epoch": 0.05, "learning_rate": 4.983847490177073e-05, "loss": 0.9838, "step": 1568 }, { "epoch": 0.05, "learning_rate": 4.9838172605609946e-05, "loss": 1.1127, "step": 1569 }, { "epoch": 0.05, "learning_rate": 4.983787002775662e-05, "loss": 0.9721, "step": 1570 }, { "epoch": 0.05, "learning_rate": 4.9837567168214186e-05, "loss": 0.8515, "step": 1571 }, { "epoch": 0.05, "learning_rate": 4.983726402698606e-05, "loss": 0.7673, "step": 1572 }, { "epoch": 0.05, "learning_rate": 4.9836960604075697e-05, "loss": 0.8076, "step": 1573 }, { "epoch": 0.05, "learning_rate": 4.9836656899486536e-05, "loss": 1.0863, "step": 1574 }, { "epoch": 0.05, "learning_rate": 4.9836352913222024e-05, "loss": 0.8612, "step": 1575 }, { "epoch": 0.05, "learning_rate": 4.9836048645285596e-05, "loss": 0.6, "step": 1576 }, { "epoch": 0.05, "learning_rate": 4.9835744095680715e-05, "loss": 0.945, "step": 1577 }, { "epoch": 0.05, "learning_rate": 4.9835439264410835e-05, "loss": 1.68, "step": 1578 }, { "epoch": 0.05, "learning_rate": 4.98351341514794e-05, "loss": 1.2805, "step": 1579 }, { "epoch": 0.05, "learning_rate": 4.983482875688988e-05, "loss": 1.175, "step": 1580 }, { "epoch": 0.05, "learning_rate": 4.983452308064575e-05, "loss": 2.9601, "step": 1581 }, { "epoch": 0.05, "learning_rate": 4.983421712275046e-05, "loss": 0.8887, "step": 1582 }, { "epoch": 0.05, "learning_rate": 4.983391088320748e-05, "loss": 0.6523, "step": 1583 }, { "epoch": 0.05, "learning_rate": 4.9833604362020283e-05, "loss": 1.5759, "step": 1584 }, { "epoch": 0.05, "learning_rate": 4.983329755919236e-05, "loss": 2.6364, "step": 1585 }, { "epoch": 0.05, "learning_rate": 4.983299047472717e-05, "loss": 1.0767, "step": 1586 }, { "epoch": 0.05, "learning_rate": 4.9832683108628216e-05, "loss": 0.8773, "step": 1587 }, { "epoch": 0.05, "learning_rate": 4.9832375460898964e-05, "loss": 0.9996, "step": 1588 }, { "epoch": 0.05, "learning_rate": 4.9832067531542925e-05, "loss": 0.8333, "step": 1589 }, { "epoch": 0.05, "learning_rate": 4.983175932056357e-05, "loss": 0.9023, "step": 1590 }, { "epoch": 0.05, "learning_rate": 4.983145082796441e-05, "loss": 1.0165, "step": 1591 }, { "epoch": 0.05, "learning_rate": 4.983114205374893e-05, "loss": 0.9198, "step": 1592 }, { "epoch": 0.05, "learning_rate": 4.983083299792064e-05, "loss": 0.3629, "step": 1593 }, { "epoch": 0.05, "learning_rate": 4.983052366048305e-05, "loss": 0.5331, "step": 1594 }, { "epoch": 0.05, "learning_rate": 4.983021404143966e-05, "loss": 1.1124, "step": 1595 }, { "epoch": 0.05, "learning_rate": 4.982990414079398e-05, "loss": 1.1943, "step": 1596 }, { "epoch": 0.05, "learning_rate": 4.982959395854953e-05, "loss": 1.028, "step": 1597 }, { "epoch": 0.05, "learning_rate": 4.9829283494709824e-05, "loss": 0.7365, "step": 1598 }, { "epoch": 0.05, "learning_rate": 4.982897274927839e-05, "loss": 0.6439, "step": 1599 }, { "epoch": 0.05, "learning_rate": 4.982866172225875e-05, "loss": 0.976, "step": 1600 }, { "epoch": 0.05, "learning_rate": 4.9828350413654424e-05, "loss": 0.8051, "step": 1601 }, { "epoch": 0.05, "learning_rate": 4.982803882346895e-05, "loss": 1.1982, "step": 1602 }, { "epoch": 0.05, "learning_rate": 4.982772695170586e-05, "loss": 0.6319, "step": 1603 }, { "epoch": 0.05, "learning_rate": 4.982741479836869e-05, "loss": 0.7718, "step": 1604 }, { "epoch": 0.05, "learning_rate": 4.982710236346098e-05, "loss": 0.7584, "step": 1605 }, { "epoch": 0.05, "learning_rate": 4.982678964698626e-05, "loss": 0.9665, "step": 1606 }, { "epoch": 0.05, "learning_rate": 4.982647664894811e-05, "loss": 0.7209, "step": 1607 }, { "epoch": 0.05, "learning_rate": 4.982616336935005e-05, "loss": 0.5638, "step": 1608 }, { "epoch": 0.05, "learning_rate": 4.982584980819565e-05, "loss": 1.0946, "step": 1609 }, { "epoch": 0.05, "learning_rate": 4.982553596548845e-05, "loss": 0.9233, "step": 1610 }, { "epoch": 0.05, "learning_rate": 4.9825221841232014e-05, "loss": 0.2228, "step": 1611 }, { "epoch": 0.05, "learning_rate": 4.982490743542992e-05, "loss": 0.2612, "step": 1612 }, { "epoch": 0.05, "learning_rate": 4.9824592748085725e-05, "loss": 0.2561, "step": 1613 }, { "epoch": 0.05, "learning_rate": 4.982427777920299e-05, "loss": 1.0074, "step": 1614 }, { "epoch": 0.05, "learning_rate": 4.982396252878529e-05, "loss": 0.9489, "step": 1615 }, { "epoch": 0.05, "learning_rate": 4.982364699683621e-05, "loss": 0.2895, "step": 1616 }, { "epoch": 0.05, "learning_rate": 4.9823331183359315e-05, "loss": 0.2334, "step": 1617 }, { "epoch": 0.05, "learning_rate": 4.9823015088358196e-05, "loss": 0.3132, "step": 1618 }, { "epoch": 0.05, "learning_rate": 4.9822698711836434e-05, "loss": 0.8158, "step": 1619 }, { "epoch": 0.05, "learning_rate": 4.982238205379762e-05, "loss": 1.0653, "step": 1620 }, { "epoch": 0.05, "learning_rate": 4.982206511424534e-05, "loss": 1.0622, "step": 1621 }, { "epoch": 0.05, "learning_rate": 4.98217478931832e-05, "loss": 0.7539, "step": 1622 }, { "epoch": 0.05, "learning_rate": 4.982143039061478e-05, "loss": 0.6701, "step": 1623 }, { "epoch": 0.05, "learning_rate": 4.9821112606543694e-05, "loss": 1.4192, "step": 1624 }, { "epoch": 0.05, "learning_rate": 4.982079454097354e-05, "loss": 1.3348, "step": 1625 }, { "epoch": 0.05, "learning_rate": 4.982047619390793e-05, "loss": 1.1027, "step": 1626 }, { "epoch": 0.05, "learning_rate": 4.982015756535047e-05, "loss": 1.0238, "step": 1627 }, { "epoch": 0.05, "learning_rate": 4.981983865530478e-05, "loss": 0.9426, "step": 1628 }, { "epoch": 0.05, "learning_rate": 4.9819519463774464e-05, "loss": 1.2231, "step": 1629 }, { "epoch": 0.05, "learning_rate": 4.981919999076315e-05, "loss": 0.8629, "step": 1630 }, { "epoch": 0.05, "learning_rate": 4.981888023627447e-05, "loss": 1.0496, "step": 1631 }, { "epoch": 0.05, "learning_rate": 4.9818560200312045e-05, "loss": 1.1133, "step": 1632 }, { "epoch": 0.05, "learning_rate": 4.9818239882879494e-05, "loss": 1.0447, "step": 1633 }, { "epoch": 0.05, "learning_rate": 4.9817919283980455e-05, "loss": 0.9578, "step": 1634 }, { "epoch": 0.05, "learning_rate": 4.981759840361857e-05, "loss": 0.8971, "step": 1635 }, { "epoch": 0.05, "learning_rate": 4.981727724179747e-05, "loss": 0.9165, "step": 1636 }, { "epoch": 0.05, "learning_rate": 4.981695579852081e-05, "loss": 1.1604, "step": 1637 }, { "epoch": 0.05, "learning_rate": 4.9816634073792216e-05, "loss": 0.9122, "step": 1638 }, { "epoch": 0.05, "learning_rate": 4.981631206761536e-05, "loss": 1.101, "step": 1639 }, { "epoch": 0.05, "learning_rate": 4.981598977999386e-05, "loss": 1.5501, "step": 1640 }, { "epoch": 0.05, "learning_rate": 4.981566721093142e-05, "loss": 1.648, "step": 1641 }, { "epoch": 0.05, "learning_rate": 4.981534436043166e-05, "loss": 0.9591, "step": 1642 }, { "epoch": 0.05, "learning_rate": 4.981502122849825e-05, "loss": 1.5281, "step": 1643 }, { "epoch": 0.05, "learning_rate": 4.981469781513486e-05, "loss": 1.1153, "step": 1644 }, { "epoch": 0.05, "learning_rate": 4.9814374120345144e-05, "loss": 1.1248, "step": 1645 }, { "epoch": 0.05, "learning_rate": 4.98140501441328e-05, "loss": 2.2173, "step": 1646 }, { "epoch": 0.05, "learning_rate": 4.981372588650148e-05, "loss": 0.9515, "step": 1647 }, { "epoch": 0.05, "learning_rate": 4.981340134745487e-05, "loss": 0.9375, "step": 1648 }, { "epoch": 0.05, "learning_rate": 4.9813076526996646e-05, "loss": 1.0476, "step": 1649 }, { "epoch": 0.06, "learning_rate": 4.981275142513049e-05, "loss": 0.2951, "step": 1650 }, { "epoch": 0.06, "learning_rate": 4.98124260418601e-05, "loss": 1.0148, "step": 1651 }, { "epoch": 0.06, "learning_rate": 4.9812100377189156e-05, "loss": 1.0974, "step": 1652 }, { "epoch": 0.06, "learning_rate": 4.981177443112136e-05, "loss": 1.0279, "step": 1653 }, { "epoch": 0.06, "learning_rate": 4.98114482036604e-05, "loss": 0.8535, "step": 1654 }, { "epoch": 0.06, "learning_rate": 4.981112169480998e-05, "loss": 1.0297, "step": 1655 }, { "epoch": 0.06, "learning_rate": 4.98107949045738e-05, "loss": 1.0864, "step": 1656 }, { "epoch": 0.06, "learning_rate": 4.981046783295557e-05, "loss": 1.0162, "step": 1657 }, { "epoch": 0.06, "learning_rate": 4.9810140479959e-05, "loss": 0.9089, "step": 1658 }, { "epoch": 0.06, "learning_rate": 4.98098128455878e-05, "loss": 0.8858, "step": 1659 }, { "epoch": 0.06, "learning_rate": 4.9809484929845686e-05, "loss": 1.04, "step": 1660 }, { "epoch": 0.06, "learning_rate": 4.980915673273637e-05, "loss": 0.7955, "step": 1661 }, { "epoch": 0.06, "learning_rate": 4.980882825426358e-05, "loss": 1.0113, "step": 1662 }, { "epoch": 0.06, "learning_rate": 4.980849949443106e-05, "loss": 0.9116, "step": 1663 }, { "epoch": 0.06, "learning_rate": 4.980817045324251e-05, "loss": 1.0117, "step": 1664 }, { "epoch": 0.06, "learning_rate": 4.980784113070167e-05, "loss": 1.1002, "step": 1665 }, { "epoch": 0.06, "learning_rate": 4.980751152681228e-05, "loss": 0.9736, "step": 1666 }, { "epoch": 0.06, "learning_rate": 4.980718164157808e-05, "loss": 0.7428, "step": 1667 }, { "epoch": 0.06, "learning_rate": 4.98068514750028e-05, "loss": 0.2466, "step": 1668 }, { "epoch": 0.06, "learning_rate": 4.980652102709019e-05, "loss": 0.3907, "step": 1669 }, { "epoch": 0.06, "learning_rate": 4.980619029784401e-05, "loss": 0.5169, "step": 1670 }, { "epoch": 0.06, "learning_rate": 4.980585928726799e-05, "loss": 0.8333, "step": 1671 }, { "epoch": 0.06, "learning_rate": 4.9805527995365896e-05, "loss": 1.0715, "step": 1672 }, { "epoch": 0.06, "learning_rate": 4.9805196422141486e-05, "loss": 1.0371, "step": 1673 }, { "epoch": 0.06, "learning_rate": 4.9804864567598516e-05, "loss": 0.8337, "step": 1674 }, { "epoch": 0.06, "learning_rate": 4.9804532431740745e-05, "loss": 0.4631, "step": 1675 }, { "epoch": 0.06, "learning_rate": 4.9804200014571955e-05, "loss": 0.8012, "step": 1676 }, { "epoch": 0.06, "learning_rate": 4.980386731609591e-05, "loss": 0.8881, "step": 1677 }, { "epoch": 0.06, "learning_rate": 4.980353433631637e-05, "loss": 1.1128, "step": 1678 }, { "epoch": 0.06, "learning_rate": 4.980320107523712e-05, "loss": 0.8717, "step": 1679 }, { "epoch": 0.06, "learning_rate": 4.980286753286195e-05, "loss": 1.1727, "step": 1680 }, { "epoch": 0.06, "learning_rate": 4.9802533709194625e-05, "loss": 1.0288, "step": 1681 }, { "epoch": 0.06, "learning_rate": 4.980219960423895e-05, "loss": 0.9388, "step": 1682 }, { "epoch": 0.06, "learning_rate": 4.9801865217998694e-05, "loss": 1.8551, "step": 1683 }, { "epoch": 0.06, "learning_rate": 4.980153055047766e-05, "loss": 1.0372, "step": 1684 }, { "epoch": 0.06, "learning_rate": 4.980119560167965e-05, "loss": 0.9304, "step": 1685 }, { "epoch": 0.06, "learning_rate": 4.980086037160845e-05, "loss": 0.7368, "step": 1686 }, { "epoch": 0.06, "learning_rate": 4.980052486026787e-05, "loss": 0.8437, "step": 1687 }, { "epoch": 0.06, "learning_rate": 4.9800189067661705e-05, "loss": 0.7202, "step": 1688 }, { "epoch": 0.06, "learning_rate": 4.979985299379378e-05, "loss": 1.0339, "step": 1689 }, { "epoch": 0.06, "learning_rate": 4.9799516638667884e-05, "loss": 0.9861, "step": 1690 }, { "epoch": 0.06, "learning_rate": 4.979918000228786e-05, "loss": 0.9242, "step": 1691 }, { "epoch": 0.06, "learning_rate": 4.97988430846575e-05, "loss": 0.9723, "step": 1692 }, { "epoch": 0.06, "learning_rate": 4.979850588578065e-05, "loss": 1.1281, "step": 1693 }, { "epoch": 0.06, "learning_rate": 4.979816840566109e-05, "loss": 1.0443, "step": 1694 }, { "epoch": 0.06, "learning_rate": 4.97978306443027e-05, "loss": 0.4622, "step": 1695 }, { "epoch": 0.06, "learning_rate": 4.9797492601709285e-05, "loss": 0.7653, "step": 1696 }, { "epoch": 0.06, "learning_rate": 4.979715427788468e-05, "loss": 1.1502, "step": 1697 }, { "epoch": 0.06, "learning_rate": 4.979681567283272e-05, "loss": 1.0394, "step": 1698 }, { "epoch": 0.06, "learning_rate": 4.979647678655726e-05, "loss": 0.4947, "step": 1699 }, { "epoch": 0.06, "learning_rate": 4.979613761906212e-05, "loss": 1.1224, "step": 1700 }, { "epoch": 0.06, "learning_rate": 4.979579817035116e-05, "loss": 1.3988, "step": 1701 }, { "epoch": 0.06, "learning_rate": 4.9795458440428225e-05, "loss": 1.206, "step": 1702 }, { "epoch": 0.06, "learning_rate": 4.9795118429297185e-05, "loss": 1.1989, "step": 1703 }, { "epoch": 0.06, "learning_rate": 4.9794778136961866e-05, "loss": 1.034, "step": 1704 }, { "epoch": 0.06, "learning_rate": 4.979443756342616e-05, "loss": 1.0815, "step": 1705 }, { "epoch": 0.06, "learning_rate": 4.9794096708693894e-05, "loss": 0.8098, "step": 1706 }, { "epoch": 0.06, "learning_rate": 4.9793755572768966e-05, "loss": 0.8337, "step": 1707 }, { "epoch": 0.06, "learning_rate": 4.979341415565523e-05, "loss": 1.0901, "step": 1708 }, { "epoch": 0.06, "learning_rate": 4.979307245735655e-05, "loss": 1.0683, "step": 1709 }, { "epoch": 0.06, "learning_rate": 4.9792730477876816e-05, "loss": 1.2007, "step": 1710 }, { "epoch": 0.06, "learning_rate": 4.979238821721991e-05, "loss": 1.1644, "step": 1711 }, { "epoch": 0.06, "learning_rate": 4.979204567538969e-05, "loss": 0.8962, "step": 1712 }, { "epoch": 0.06, "learning_rate": 4.979170285239006e-05, "loss": 1.3423, "step": 1713 }, { "epoch": 0.06, "learning_rate": 4.979135974822491e-05, "loss": 1.5147, "step": 1714 }, { "epoch": 0.06, "learning_rate": 4.979101636289812e-05, "loss": 1.041, "step": 1715 }, { "epoch": 0.06, "learning_rate": 4.979067269641359e-05, "loss": 0.6835, "step": 1716 }, { "epoch": 0.06, "learning_rate": 4.9790328748775214e-05, "loss": 1.0306, "step": 1717 }, { "epoch": 0.06, "learning_rate": 4.9789984519986905e-05, "loss": 0.9681, "step": 1718 }, { "epoch": 0.06, "learning_rate": 4.978964001005255e-05, "loss": 0.8081, "step": 1719 }, { "epoch": 0.06, "learning_rate": 4.978929521897607e-05, "loss": 0.9513, "step": 1720 }, { "epoch": 0.06, "learning_rate": 4.978895014676136e-05, "loss": 0.9314, "step": 1721 }, { "epoch": 0.06, "learning_rate": 4.9788604793412344e-05, "loss": 0.9104, "step": 1722 }, { "epoch": 0.06, "learning_rate": 4.978825915893294e-05, "loss": 1.3535, "step": 1723 }, { "epoch": 0.06, "learning_rate": 4.978791324332707e-05, "loss": 0.8559, "step": 1724 }, { "epoch": 0.06, "learning_rate": 4.978756704659864e-05, "loss": 0.9422, "step": 1725 }, { "epoch": 0.06, "learning_rate": 4.9787220568751594e-05, "loss": 1.2105, "step": 1726 }, { "epoch": 0.06, "learning_rate": 4.978687380978986e-05, "loss": 1.1622, "step": 1727 }, { "epoch": 0.06, "learning_rate": 4.978652676971736e-05, "loss": 0.9497, "step": 1728 }, { "epoch": 0.06, "learning_rate": 4.978617944853804e-05, "loss": 0.861, "step": 1729 }, { "epoch": 0.06, "learning_rate": 4.9785831846255823e-05, "loss": 0.7502, "step": 1730 }, { "epoch": 0.06, "learning_rate": 4.978548396287468e-05, "loss": 0.9545, "step": 1731 }, { "epoch": 0.06, "learning_rate": 4.978513579839853e-05, "loss": 1.1598, "step": 1732 }, { "epoch": 0.06, "learning_rate": 4.978478735283134e-05, "loss": 1.1171, "step": 1733 }, { "epoch": 0.06, "learning_rate": 4.978443862617704e-05, "loss": 0.9829, "step": 1734 }, { "epoch": 0.06, "learning_rate": 4.978408961843961e-05, "loss": 1.0621, "step": 1735 }, { "epoch": 0.06, "learning_rate": 4.9783740329623e-05, "loss": 0.8093, "step": 1736 }, { "epoch": 0.06, "learning_rate": 4.978339075973115e-05, "loss": 1.083, "step": 1737 }, { "epoch": 0.06, "learning_rate": 4.978304090876805e-05, "loss": 1.7968, "step": 1738 }, { "epoch": 0.06, "learning_rate": 4.978269077673767e-05, "loss": 0.5659, "step": 1739 }, { "epoch": 0.06, "learning_rate": 4.978234036364396e-05, "loss": 0.9427, "step": 1740 }, { "epoch": 0.06, "learning_rate": 4.97819896694909e-05, "loss": 0.8732, "step": 1741 }, { "epoch": 0.06, "learning_rate": 4.978163869428248e-05, "loss": 1.0518, "step": 1742 }, { "epoch": 0.06, "learning_rate": 4.9781287438022673e-05, "loss": 0.8445, "step": 1743 }, { "epoch": 0.06, "learning_rate": 4.978093590071546e-05, "loss": 0.5226, "step": 1744 }, { "epoch": 0.06, "learning_rate": 4.9780584082364825e-05, "loss": 0.5069, "step": 1745 }, { "epoch": 0.06, "learning_rate": 4.9780231982974776e-05, "loss": 0.4447, "step": 1746 }, { "epoch": 0.06, "learning_rate": 4.9779879602549284e-05, "loss": 0.6823, "step": 1747 }, { "epoch": 0.06, "learning_rate": 4.977952694109235e-05, "loss": 1.1026, "step": 1748 }, { "epoch": 0.06, "learning_rate": 4.977917399860799e-05, "loss": 0.7453, "step": 1749 }, { "epoch": 0.06, "learning_rate": 4.9778820775100185e-05, "loss": 0.9964, "step": 1750 }, { "epoch": 0.06, "learning_rate": 4.9778467270572946e-05, "loss": 0.5863, "step": 1751 }, { "epoch": 0.06, "learning_rate": 4.9778113485030296e-05, "loss": 0.9612, "step": 1752 }, { "epoch": 0.06, "learning_rate": 4.977775941847624e-05, "loss": 0.9794, "step": 1753 }, { "epoch": 0.06, "learning_rate": 4.977740507091479e-05, "loss": 0.831, "step": 1754 }, { "epoch": 0.06, "learning_rate": 4.977705044234996e-05, "loss": 0.9185, "step": 1755 }, { "epoch": 0.06, "learning_rate": 4.9776695532785784e-05, "loss": 1.1892, "step": 1756 }, { "epoch": 0.06, "learning_rate": 4.977634034222628e-05, "loss": 1.0582, "step": 1757 }, { "epoch": 0.06, "learning_rate": 4.9775984870675475e-05, "loss": 1.0688, "step": 1758 }, { "epoch": 0.06, "learning_rate": 4.9775629118137404e-05, "loss": 0.8016, "step": 1759 }, { "epoch": 0.06, "learning_rate": 4.97752730846161e-05, "loss": 0.6913, "step": 1760 }, { "epoch": 0.06, "learning_rate": 4.977491677011561e-05, "loss": 0.8736, "step": 1761 }, { "epoch": 0.06, "learning_rate": 4.977456017463996e-05, "loss": 0.8933, "step": 1762 }, { "epoch": 0.06, "learning_rate": 4.97742032981932e-05, "loss": 0.9786, "step": 1763 }, { "epoch": 0.06, "learning_rate": 4.9773846140779376e-05, "loss": 1.0699, "step": 1764 }, { "epoch": 0.06, "learning_rate": 4.977348870240254e-05, "loss": 0.5272, "step": 1765 }, { "epoch": 0.06, "learning_rate": 4.977313098306675e-05, "loss": 0.9288, "step": 1766 }, { "epoch": 0.06, "learning_rate": 4.9772772982776064e-05, "loss": 0.7425, "step": 1767 }, { "epoch": 0.06, "learning_rate": 4.977241470153453e-05, "loss": 1.0234, "step": 1768 }, { "epoch": 0.06, "learning_rate": 4.977205613934622e-05, "loss": 1.203, "step": 1769 }, { "epoch": 0.06, "learning_rate": 4.97716972962152e-05, "loss": 0.843, "step": 1770 }, { "epoch": 0.06, "learning_rate": 4.977133817214554e-05, "loss": 0.5561, "step": 1771 }, { "epoch": 0.06, "learning_rate": 4.9770978767141316e-05, "loss": 1.0513, "step": 1772 }, { "epoch": 0.06, "learning_rate": 4.977061908120659e-05, "loss": 0.9698, "step": 1773 }, { "epoch": 0.06, "learning_rate": 4.977025911434546e-05, "loss": 0.4744, "step": 1774 }, { "epoch": 0.06, "learning_rate": 4.976989886656199e-05, "loss": 1.0234, "step": 1775 }, { "epoch": 0.06, "learning_rate": 4.976953833786028e-05, "loss": 0.9268, "step": 1776 }, { "epoch": 0.06, "learning_rate": 4.9769177528244415e-05, "loss": 1.0681, "step": 1777 }, { "epoch": 0.06, "learning_rate": 4.976881643771849e-05, "loss": 1.3164, "step": 1778 }, { "epoch": 0.06, "learning_rate": 4.976845506628659e-05, "loss": 3.6034, "step": 1779 }, { "epoch": 0.06, "learning_rate": 4.9768093413952824e-05, "loss": 1.1208, "step": 1780 }, { "epoch": 0.06, "learning_rate": 4.9767731480721284e-05, "loss": 0.6424, "step": 1781 }, { "epoch": 0.06, "learning_rate": 4.9767369266596074e-05, "loss": 0.5293, "step": 1782 }, { "epoch": 0.06, "learning_rate": 4.9767006771581315e-05, "loss": 0.8546, "step": 1783 }, { "epoch": 0.06, "learning_rate": 4.9766643995681114e-05, "loss": 1.0664, "step": 1784 }, { "epoch": 0.06, "learning_rate": 4.9766280938899576e-05, "loss": 1.0726, "step": 1785 }, { "epoch": 0.06, "learning_rate": 4.976591760124083e-05, "loss": 0.9699, "step": 1786 }, { "epoch": 0.06, "learning_rate": 4.9765553982708985e-05, "loss": 1.0041, "step": 1787 }, { "epoch": 0.06, "learning_rate": 4.976519008330818e-05, "loss": 0.9182, "step": 1788 }, { "epoch": 0.06, "learning_rate": 4.976482590304252e-05, "loss": 0.8621, "step": 1789 }, { "epoch": 0.06, "learning_rate": 4.976446144191616e-05, "loss": 1.2986, "step": 1790 }, { "epoch": 0.06, "learning_rate": 4.976409669993322e-05, "loss": 0.6683, "step": 1791 }, { "epoch": 0.06, "learning_rate": 4.976373167709783e-05, "loss": 0.9147, "step": 1792 }, { "epoch": 0.06, "learning_rate": 4.976336637341414e-05, "loss": 0.9376, "step": 1793 }, { "epoch": 0.06, "learning_rate": 4.9763000788886295e-05, "loss": 1.091, "step": 1794 }, { "epoch": 0.06, "learning_rate": 4.976263492351843e-05, "loss": 1.1336, "step": 1795 }, { "epoch": 0.06, "learning_rate": 4.976226877731471e-05, "loss": 1.1005, "step": 1796 }, { "epoch": 0.06, "learning_rate": 4.9761902350279275e-05, "loss": 1.0492, "step": 1797 }, { "epoch": 0.06, "learning_rate": 4.976153564241628e-05, "loss": 0.9621, "step": 1798 }, { "epoch": 0.06, "learning_rate": 4.97611686537299e-05, "loss": 0.8767, "step": 1799 }, { "epoch": 0.06, "learning_rate": 4.976080138422428e-05, "loss": 0.9561, "step": 1800 }, { "epoch": 0.06, "learning_rate": 4.976043383390359e-05, "loss": 0.8869, "step": 1801 }, { "epoch": 0.06, "learning_rate": 4.9760066002771985e-05, "loss": 0.9506, "step": 1802 }, { "epoch": 0.06, "learning_rate": 4.975969789083367e-05, "loss": 0.5252, "step": 1803 }, { "epoch": 0.06, "learning_rate": 4.9759329498092794e-05, "loss": 1.2461, "step": 1804 }, { "epoch": 0.06, "learning_rate": 4.9758960824553546e-05, "loss": 1.0692, "step": 1805 }, { "epoch": 0.06, "learning_rate": 4.975859187022009e-05, "loss": 0.8431, "step": 1806 }, { "epoch": 0.06, "learning_rate": 4.975822263509664e-05, "loss": 0.8342, "step": 1807 }, { "epoch": 0.06, "learning_rate": 4.975785311918736e-05, "loss": 0.7853, "step": 1808 }, { "epoch": 0.06, "learning_rate": 4.975748332249644e-05, "loss": 0.6348, "step": 1809 }, { "epoch": 0.06, "learning_rate": 4.9757113245028086e-05, "loss": 0.2761, "step": 1810 }, { "epoch": 0.06, "learning_rate": 4.9756742886786486e-05, "loss": 0.9661, "step": 1811 }, { "epoch": 0.06, "learning_rate": 4.975637224777585e-05, "loss": 2.2045, "step": 1812 }, { "epoch": 0.06, "learning_rate": 4.975600132800038e-05, "loss": 1.5047, "step": 1813 }, { "epoch": 0.06, "learning_rate": 4.975563012746427e-05, "loss": 1.2459, "step": 1814 }, { "epoch": 0.06, "learning_rate": 4.975525864617175e-05, "loss": 0.9204, "step": 1815 }, { "epoch": 0.06, "learning_rate": 4.975488688412701e-05, "loss": 1.0266, "step": 1816 }, { "epoch": 0.06, "learning_rate": 4.975451484133429e-05, "loss": 1.0032, "step": 1817 }, { "epoch": 0.06, "learning_rate": 4.975414251779779e-05, "loss": 1.0603, "step": 1818 }, { "epoch": 0.06, "learning_rate": 4.975376991352173e-05, "loss": 0.9285, "step": 1819 }, { "epoch": 0.06, "learning_rate": 4.975339702851035e-05, "loss": 1.1511, "step": 1820 }, { "epoch": 0.06, "learning_rate": 4.9753023862767884e-05, "loss": 0.98, "step": 1821 }, { "epoch": 0.06, "learning_rate": 4.975265041629855e-05, "loss": 1.1861, "step": 1822 }, { "epoch": 0.06, "learning_rate": 4.975227668910658e-05, "loss": 1.1481, "step": 1823 }, { "epoch": 0.06, "learning_rate": 4.975190268119623e-05, "loss": 0.9501, "step": 1824 }, { "epoch": 0.06, "learning_rate": 4.975152839257173e-05, "loss": 2.8888, "step": 1825 }, { "epoch": 0.06, "learning_rate": 4.975115382323732e-05, "loss": 1.5749, "step": 1826 }, { "epoch": 0.06, "learning_rate": 4.975077897319726e-05, "loss": 0.8758, "step": 1827 }, { "epoch": 0.06, "learning_rate": 4.9750403842455804e-05, "loss": 1.2639, "step": 1828 }, { "epoch": 0.06, "learning_rate": 4.975002843101719e-05, "loss": 1.1697, "step": 1829 }, { "epoch": 0.06, "learning_rate": 4.9749652738885676e-05, "loss": 1.1092, "step": 1830 }, { "epoch": 0.06, "learning_rate": 4.9749276766065545e-05, "loss": 0.9141, "step": 1831 }, { "epoch": 0.06, "learning_rate": 4.9748900512561037e-05, "loss": 1.021, "step": 1832 }, { "epoch": 0.06, "learning_rate": 4.974852397837644e-05, "loss": 0.9212, "step": 1833 }, { "epoch": 0.06, "learning_rate": 4.9748147163516e-05, "loss": 0.5412, "step": 1834 }, { "epoch": 0.06, "learning_rate": 4.974777006798401e-05, "loss": 0.5518, "step": 1835 }, { "epoch": 0.06, "learning_rate": 4.974739269178474e-05, "loss": 0.6737, "step": 1836 }, { "epoch": 0.06, "learning_rate": 4.974701503492247e-05, "loss": 0.8955, "step": 1837 }, { "epoch": 0.06, "learning_rate": 4.974663709740148e-05, "loss": 0.9512, "step": 1838 }, { "epoch": 0.06, "learning_rate": 4.974625887922606e-05, "loss": 0.975, "step": 1839 }, { "epoch": 0.06, "learning_rate": 4.9745880380400504e-05, "loss": 1.0648, "step": 1840 }, { "epoch": 0.06, "learning_rate": 4.97455016009291e-05, "loss": 0.8131, "step": 1841 }, { "epoch": 0.06, "learning_rate": 4.9745122540816137e-05, "loss": 1.0592, "step": 1842 }, { "epoch": 0.06, "learning_rate": 4.9744743200065925e-05, "loss": 1.5236, "step": 1843 }, { "epoch": 0.06, "learning_rate": 4.9744363578682755e-05, "loss": 1.1624, "step": 1844 }, { "epoch": 0.06, "learning_rate": 4.974398367667095e-05, "loss": 1.096, "step": 1845 }, { "epoch": 0.06, "learning_rate": 4.9743603494034796e-05, "loss": 1.0619, "step": 1846 }, { "epoch": 0.06, "learning_rate": 4.974322303077862e-05, "loss": 1.062, "step": 1847 }, { "epoch": 0.06, "learning_rate": 4.974284228690673e-05, "loss": 0.9457, "step": 1848 }, { "epoch": 0.06, "learning_rate": 4.974246126242346e-05, "loss": 0.9696, "step": 1849 }, { "epoch": 0.06, "learning_rate": 4.97420799573331e-05, "loss": 1.007, "step": 1850 }, { "epoch": 0.06, "learning_rate": 4.974169837164e-05, "loss": 0.902, "step": 1851 }, { "epoch": 0.06, "learning_rate": 4.9741316505348484e-05, "loss": 0.5207, "step": 1852 }, { "epoch": 0.06, "learning_rate": 4.9740934358462876e-05, "loss": 0.9401, "step": 1853 }, { "epoch": 0.06, "learning_rate": 4.9740551930987514e-05, "loss": 0.7948, "step": 1854 }, { "epoch": 0.06, "learning_rate": 4.974016922292673e-05, "loss": 0.2402, "step": 1855 }, { "epoch": 0.06, "learning_rate": 4.973978623428487e-05, "loss": 0.9477, "step": 1856 }, { "epoch": 0.06, "learning_rate": 4.9739402965066276e-05, "loss": 0.8847, "step": 1857 }, { "epoch": 0.06, "learning_rate": 4.9739019415275295e-05, "loss": 1.269, "step": 1858 }, { "epoch": 0.06, "learning_rate": 4.973863558491628e-05, "loss": 0.8335, "step": 1859 }, { "epoch": 0.06, "learning_rate": 4.9738251473993576e-05, "loss": 0.897, "step": 1860 }, { "epoch": 0.06, "learning_rate": 4.9737867082511546e-05, "loss": 1.6219, "step": 1861 }, { "epoch": 0.06, "learning_rate": 4.9737482410474545e-05, "loss": 0.929, "step": 1862 }, { "epoch": 0.06, "learning_rate": 4.9737097457886936e-05, "loss": 0.9719, "step": 1863 }, { "epoch": 0.06, "learning_rate": 4.973671222475309e-05, "loss": 0.9394, "step": 1864 }, { "epoch": 0.06, "learning_rate": 4.9736326711077374e-05, "loss": 0.8133, "step": 1865 }, { "epoch": 0.06, "learning_rate": 4.9735940916864155e-05, "loss": 0.7774, "step": 1866 }, { "epoch": 0.06, "learning_rate": 4.973555484211782e-05, "loss": 0.8124, "step": 1867 }, { "epoch": 0.06, "learning_rate": 4.9735168486842734e-05, "loss": 1.018, "step": 1868 }, { "epoch": 0.06, "learning_rate": 4.973478185104328e-05, "loss": 1.286, "step": 1869 }, { "epoch": 0.06, "learning_rate": 4.973439493472385e-05, "loss": 0.9795, "step": 1870 }, { "epoch": 0.06, "learning_rate": 4.973400773788883e-05, "loss": 0.7584, "step": 1871 }, { "epoch": 0.06, "learning_rate": 4.9733620260542614e-05, "loss": 0.8842, "step": 1872 }, { "epoch": 0.06, "learning_rate": 4.9733232502689586e-05, "loss": 0.6666, "step": 1873 }, { "epoch": 0.06, "learning_rate": 4.973284446433416e-05, "loss": 1.2097, "step": 1874 }, { "epoch": 0.06, "learning_rate": 4.973245614548072e-05, "loss": 1.4788, "step": 1875 }, { "epoch": 0.06, "learning_rate": 4.9732067546133675e-05, "loss": 0.9592, "step": 1876 }, { "epoch": 0.06, "learning_rate": 4.973167866629744e-05, "loss": 0.6447, "step": 1877 }, { "epoch": 0.06, "learning_rate": 4.973128950597641e-05, "loss": 0.8289, "step": 1878 }, { "epoch": 0.06, "learning_rate": 4.9730900065175027e-05, "loss": 0.1061, "step": 1879 }, { "epoch": 0.06, "learning_rate": 4.9730510343897675e-05, "loss": 0.323, "step": 1880 }, { "epoch": 0.06, "learning_rate": 4.973012034214879e-05, "loss": 0.248, "step": 1881 }, { "epoch": 0.06, "learning_rate": 4.9729730059932787e-05, "loss": 0.9319, "step": 1882 }, { "epoch": 0.06, "learning_rate": 4.972933949725411e-05, "loss": 1.0814, "step": 1883 }, { "epoch": 0.06, "learning_rate": 4.972894865411717e-05, "loss": 1.0173, "step": 1884 }, { "epoch": 0.06, "learning_rate": 4.972855753052641e-05, "loss": 0.8254, "step": 1885 }, { "epoch": 0.06, "learning_rate": 4.9728166126486254e-05, "loss": 0.9041, "step": 1886 }, { "epoch": 0.06, "learning_rate": 4.9727774442001156e-05, "loss": 0.821, "step": 1887 }, { "epoch": 0.06, "learning_rate": 4.9727382477075544e-05, "loss": 1.2968, "step": 1888 }, { "epoch": 0.06, "learning_rate": 4.972699023171388e-05, "loss": 1.3189, "step": 1889 }, { "epoch": 0.06, "learning_rate": 4.9726597705920597e-05, "loss": 1.1282, "step": 1890 }, { "epoch": 0.06, "learning_rate": 4.972620489970016e-05, "loss": 0.8202, "step": 1891 }, { "epoch": 0.06, "learning_rate": 4.9725811813057007e-05, "loss": 0.1099, "step": 1892 }, { "epoch": 0.06, "learning_rate": 4.972541844599561e-05, "loss": 0.1628, "step": 1893 }, { "epoch": 0.06, "learning_rate": 4.972502479852042e-05, "loss": 0.7715, "step": 1894 }, { "epoch": 0.06, "learning_rate": 4.972463087063591e-05, "loss": 1.0, "step": 1895 }, { "epoch": 0.06, "learning_rate": 4.972423666234655e-05, "loss": 1.0539, "step": 1896 }, { "epoch": 0.06, "learning_rate": 4.97238421736568e-05, "loss": 1.1094, "step": 1897 }, { "epoch": 0.06, "learning_rate": 4.972344740457114e-05, "loss": 1.0836, "step": 1898 }, { "epoch": 0.06, "learning_rate": 4.972305235509405e-05, "loss": 1.2499, "step": 1899 }, { "epoch": 0.06, "learning_rate": 4.972265702523e-05, "loss": 0.7804, "step": 1900 }, { "epoch": 0.06, "learning_rate": 4.972226141498349e-05, "loss": 1.3408, "step": 1901 }, { "epoch": 0.06, "learning_rate": 4.972186552435898e-05, "loss": 1.6654, "step": 1902 }, { "epoch": 0.06, "learning_rate": 4.972146935336099e-05, "loss": 1.3437, "step": 1903 }, { "epoch": 0.06, "learning_rate": 4.972107290199399e-05, "loss": 0.9382, "step": 1904 }, { "epoch": 0.06, "learning_rate": 4.97206761702625e-05, "loss": 0.9764, "step": 1905 }, { "epoch": 0.06, "learning_rate": 4.9720279158171e-05, "loss": 0.7199, "step": 1906 }, { "epoch": 0.06, "learning_rate": 4.9719881865723996e-05, "loss": 0.8798, "step": 1907 }, { "epoch": 0.06, "learning_rate": 4.971948429292599e-05, "loss": 0.8338, "step": 1908 }, { "epoch": 0.06, "learning_rate": 4.97190864397815e-05, "loss": 0.9866, "step": 1909 }, { "epoch": 0.06, "learning_rate": 4.971868830629504e-05, "loss": 0.6995, "step": 1910 }, { "epoch": 0.06, "learning_rate": 4.971828989247111e-05, "loss": 0.9371, "step": 1911 }, { "epoch": 0.06, "learning_rate": 4.971789119831425e-05, "loss": 1.2297, "step": 1912 }, { "epoch": 0.06, "learning_rate": 4.971749222382896e-05, "loss": 1.0217, "step": 1913 }, { "epoch": 0.06, "learning_rate": 4.971709296901978e-05, "loss": 1.0218, "step": 1914 }, { "epoch": 0.06, "learning_rate": 4.971669343389124e-05, "loss": 0.755, "step": 1915 }, { "epoch": 0.06, "learning_rate": 4.971629361844785e-05, "loss": 0.4661, "step": 1916 }, { "epoch": 0.06, "learning_rate": 4.971589352269417e-05, "loss": 0.9652, "step": 1917 }, { "epoch": 0.06, "learning_rate": 4.9715493146634715e-05, "loss": 0.7638, "step": 1918 }, { "epoch": 0.06, "learning_rate": 4.9715092490274046e-05, "loss": 0.0722, "step": 1919 }, { "epoch": 0.06, "learning_rate": 4.971469155361669e-05, "loss": 0.9505, "step": 1920 }, { "epoch": 0.06, "learning_rate": 4.971429033666721e-05, "loss": 1.0644, "step": 1921 }, { "epoch": 0.06, "learning_rate": 4.971388883943014e-05, "loss": 1.2604, "step": 1922 }, { "epoch": 0.06, "learning_rate": 4.971348706191005e-05, "loss": 1.1162, "step": 1923 }, { "epoch": 0.06, "learning_rate": 4.971308500411147e-05, "loss": 0.8687, "step": 1924 }, { "epoch": 0.06, "learning_rate": 4.971268266603899e-05, "loss": 0.6559, "step": 1925 }, { "epoch": 0.06, "learning_rate": 4.9712280047697164e-05, "loss": 0.9843, "step": 1926 }, { "epoch": 0.06, "learning_rate": 4.971187714909055e-05, "loss": 0.9175, "step": 1927 }, { "epoch": 0.06, "learning_rate": 4.971147397022372e-05, "loss": 1.0574, "step": 1928 }, { "epoch": 0.06, "learning_rate": 4.9711070511101256e-05, "loss": 2.4941, "step": 1929 }, { "epoch": 0.06, "learning_rate": 4.971066677172772e-05, "loss": 4.082, "step": 1930 }, { "epoch": 0.06, "learning_rate": 4.971026275210769e-05, "loss": 3.1191, "step": 1931 }, { "epoch": 0.06, "learning_rate": 4.970985845224577e-05, "loss": 0.8652, "step": 1932 }, { "epoch": 0.06, "learning_rate": 4.970945387214652e-05, "loss": 0.9294, "step": 1933 }, { "epoch": 0.06, "learning_rate": 4.9709049011814546e-05, "loss": 1.0153, "step": 1934 }, { "epoch": 0.06, "learning_rate": 4.970864387125442e-05, "loss": 1.4666, "step": 1935 }, { "epoch": 0.06, "learning_rate": 4.9708238450470766e-05, "loss": 1.5473, "step": 1936 }, { "epoch": 0.06, "learning_rate": 4.9707832749468155e-05, "loss": 1.7093, "step": 1937 }, { "epoch": 0.06, "learning_rate": 4.9707426768251194e-05, "loss": 1.6961, "step": 1938 }, { "epoch": 0.06, "learning_rate": 4.97070205068245e-05, "loss": 1.0313, "step": 1939 }, { "epoch": 0.06, "learning_rate": 4.970661396519267e-05, "loss": 0.9253, "step": 1940 }, { "epoch": 0.06, "learning_rate": 4.9706207143360314e-05, "loss": 0.6958, "step": 1941 }, { "epoch": 0.06, "learning_rate": 4.970580004133205e-05, "loss": 0.9239, "step": 1942 }, { "epoch": 0.06, "learning_rate": 4.970539265911249e-05, "loss": 0.9273, "step": 1943 }, { "epoch": 0.06, "learning_rate": 4.970498499670626e-05, "loss": 1.0097, "step": 1944 }, { "epoch": 0.06, "learning_rate": 4.970457705411798e-05, "loss": 0.639, "step": 1945 }, { "epoch": 0.06, "learning_rate": 4.970416883135227e-05, "loss": 1.0732, "step": 1946 }, { "epoch": 0.06, "learning_rate": 4.9703760328413777e-05, "loss": 1.0331, "step": 1947 }, { "epoch": 0.06, "learning_rate": 4.970335154530712e-05, "loss": 0.8256, "step": 1948 }, { "epoch": 0.06, "learning_rate": 4.970294248203694e-05, "loss": 1.0106, "step": 1949 }, { "epoch": 0.07, "learning_rate": 4.970253313860788e-05, "loss": 1.061, "step": 1950 }, { "epoch": 0.07, "learning_rate": 4.970212351502457e-05, "loss": 1.1649, "step": 1951 }, { "epoch": 0.07, "learning_rate": 4.9701713611291665e-05, "loss": 3.1729, "step": 1952 }, { "epoch": 0.07, "learning_rate": 4.9701303427413814e-05, "loss": 3.1686, "step": 1953 }, { "epoch": 0.07, "learning_rate": 4.970089296339566e-05, "loss": 1.6163, "step": 1954 }, { "epoch": 0.07, "learning_rate": 4.9700482219241875e-05, "loss": 0.9907, "step": 1955 }, { "epoch": 0.07, "learning_rate": 4.97000711949571e-05, "loss": 0.9848, "step": 1956 }, { "epoch": 0.07, "learning_rate": 4.9699659890546e-05, "loss": 1.2307, "step": 1957 }, { "epoch": 0.07, "learning_rate": 4.969924830601325e-05, "loss": 0.9923, "step": 1958 }, { "epoch": 0.07, "learning_rate": 4.969883644136351e-05, "loss": 0.6886, "step": 1959 }, { "epoch": 0.07, "learning_rate": 4.969842429660145e-05, "loss": 0.7155, "step": 1960 }, { "epoch": 0.07, "learning_rate": 4.969801187173175e-05, "loss": 0.9058, "step": 1961 }, { "epoch": 0.07, "learning_rate": 4.9697599166759084e-05, "loss": 0.9737, "step": 1962 }, { "epoch": 0.07, "learning_rate": 4.9697186181688124e-05, "loss": 0.8902, "step": 1963 }, { "epoch": 0.07, "learning_rate": 4.969677291652357e-05, "loss": 0.1406, "step": 1964 }, { "epoch": 0.07, "learning_rate": 4.969635937127011e-05, "loss": 0.6348, "step": 1965 }, { "epoch": 0.07, "learning_rate": 4.969594554593241e-05, "loss": 1.1187, "step": 1966 }, { "epoch": 0.07, "learning_rate": 4.969553144051518e-05, "loss": 0.8796, "step": 1967 }, { "epoch": 0.07, "learning_rate": 4.9695117055023116e-05, "loss": 0.7633, "step": 1968 }, { "epoch": 0.07, "learning_rate": 4.969470238946092e-05, "loss": 1.8496, "step": 1969 }, { "epoch": 0.07, "learning_rate": 4.9694287443833285e-05, "loss": 1.5347, "step": 1970 }, { "epoch": 0.07, "learning_rate": 4.969387221814492e-05, "loss": 1.2617, "step": 1971 }, { "epoch": 0.07, "learning_rate": 4.969345671240055e-05, "loss": 0.9759, "step": 1972 }, { "epoch": 0.07, "learning_rate": 4.969304092660486e-05, "loss": 0.9136, "step": 1973 }, { "epoch": 0.07, "learning_rate": 4.969262486076259e-05, "loss": 0.8937, "step": 1974 }, { "epoch": 0.07, "learning_rate": 4.9692208514878444e-05, "loss": 1.5323, "step": 1975 }, { "epoch": 0.07, "learning_rate": 4.9691791888957154e-05, "loss": 0.8231, "step": 1976 }, { "epoch": 0.07, "learning_rate": 4.969137498300343e-05, "loss": 0.6307, "step": 1977 }, { "epoch": 0.07, "learning_rate": 4.969095779702201e-05, "loss": 1.3552, "step": 1978 }, { "epoch": 0.07, "learning_rate": 4.9690540331017634e-05, "loss": 0.9519, "step": 1979 }, { "epoch": 0.07, "learning_rate": 4.969012258499502e-05, "loss": 0.8778, "step": 1980 }, { "epoch": 0.07, "learning_rate": 4.9689704558958916e-05, "loss": 0.9796, "step": 1981 }, { "epoch": 0.07, "learning_rate": 4.968928625291406e-05, "loss": 0.9916, "step": 1982 }, { "epoch": 0.07, "learning_rate": 4.968886766686519e-05, "loss": 1.2722, "step": 1983 }, { "epoch": 0.07, "learning_rate": 4.968844880081707e-05, "loss": 0.8063, "step": 1984 }, { "epoch": 0.07, "learning_rate": 4.968802965477443e-05, "loss": 1.3403, "step": 1985 }, { "epoch": 0.07, "learning_rate": 4.968761022874203e-05, "loss": 0.909, "step": 1986 }, { "epoch": 0.07, "learning_rate": 4.968719052272464e-05, "loss": 3.1816, "step": 1987 }, { "epoch": 0.07, "learning_rate": 4.9686770536727006e-05, "loss": 1.9074, "step": 1988 }, { "epoch": 0.07, "learning_rate": 4.9686350270753894e-05, "loss": 1.1522, "step": 1989 }, { "epoch": 0.07, "learning_rate": 4.9685929724810075e-05, "loss": 1.1621, "step": 1990 }, { "epoch": 0.07, "learning_rate": 4.968550889890031e-05, "loss": 0.8671, "step": 1991 }, { "epoch": 0.07, "learning_rate": 4.968508779302939e-05, "loss": 1.1294, "step": 1992 }, { "epoch": 0.07, "learning_rate": 4.968466640720205e-05, "loss": 0.949, "step": 1993 }, { "epoch": 0.07, "learning_rate": 4.968424474142311e-05, "loss": 1.523, "step": 1994 }, { "epoch": 0.07, "learning_rate": 4.9683822795697344e-05, "loss": 1.1109, "step": 1995 }, { "epoch": 0.07, "learning_rate": 4.9683400570029524e-05, "loss": 1.3075, "step": 1996 }, { "epoch": 0.07, "learning_rate": 4.968297806442446e-05, "loss": 1.7254, "step": 1997 }, { "epoch": 0.07, "learning_rate": 4.9682555278886914e-05, "loss": 1.0074, "step": 1998 }, { "epoch": 0.07, "learning_rate": 4.9682132213421705e-05, "loss": 1.1178, "step": 1999 }, { "epoch": 0.07, "learning_rate": 4.9681708868033616e-05, "loss": 0.7343, "step": 2000 }, { "epoch": 0.07, "learning_rate": 4.968128524272746e-05, "loss": 0.8748, "step": 2001 }, { "epoch": 0.07, "learning_rate": 4.968086133750803e-05, "loss": 0.6784, "step": 2002 }, { "epoch": 0.07, "learning_rate": 4.968043715238015e-05, "loss": 1.2095, "step": 2003 }, { "epoch": 0.07, "learning_rate": 4.9680012687348614e-05, "loss": 1.135, "step": 2004 }, { "epoch": 0.07, "learning_rate": 4.9679587942418245e-05, "loss": 0.7196, "step": 2005 }, { "epoch": 0.07, "learning_rate": 4.967916291759386e-05, "loss": 0.6213, "step": 2006 }, { "epoch": 0.07, "learning_rate": 4.967873761288027e-05, "loss": 1.0185, "step": 2007 }, { "epoch": 0.07, "learning_rate": 4.96783120282823e-05, "loss": 0.9038, "step": 2008 }, { "epoch": 0.07, "learning_rate": 4.96778861638048e-05, "loss": 0.4479, "step": 2009 }, { "epoch": 0.07, "learning_rate": 4.967746001945257e-05, "loss": 0.0339, "step": 2010 }, { "epoch": 0.07, "learning_rate": 4.967703359523045e-05, "loss": 0.5268, "step": 2011 }, { "epoch": 0.07, "learning_rate": 4.9676606891143294e-05, "loss": 1.0796, "step": 2012 }, { "epoch": 0.07, "learning_rate": 4.967617990719592e-05, "loss": 1.1595, "step": 2013 }, { "epoch": 0.07, "learning_rate": 4.967575264339318e-05, "loss": 0.9784, "step": 2014 }, { "epoch": 0.07, "learning_rate": 4.9675325099739915e-05, "loss": 1.1021, "step": 2015 }, { "epoch": 0.07, "learning_rate": 4.9674897276240984e-05, "loss": 1.2265, "step": 2016 }, { "epoch": 0.07, "learning_rate": 4.9674469172901226e-05, "loss": 1.0019, "step": 2017 }, { "epoch": 0.07, "learning_rate": 4.9674040789725505e-05, "loss": 0.9923, "step": 2018 }, { "epoch": 0.07, "learning_rate": 4.9673612126718674e-05, "loss": 1.1175, "step": 2019 }, { "epoch": 0.07, "learning_rate": 4.96731831838856e-05, "loss": 0.4512, "step": 2020 }, { "epoch": 0.07, "learning_rate": 4.967275396123114e-05, "loss": 0.2009, "step": 2021 }, { "epoch": 0.07, "learning_rate": 4.967232445876017e-05, "loss": 0.9005, "step": 2022 }, { "epoch": 0.07, "learning_rate": 4.967189467647756e-05, "loss": 0.8911, "step": 2023 }, { "epoch": 0.07, "learning_rate": 4.9671464614388174e-05, "loss": 0.5538, "step": 2024 }, { "epoch": 0.07, "learning_rate": 4.96710342724969e-05, "loss": 1.0677, "step": 2025 }, { "epoch": 0.07, "learning_rate": 4.9670603650808624e-05, "loss": 0.9176, "step": 2026 }, { "epoch": 0.07, "learning_rate": 4.967017274932821e-05, "loss": 0.8786, "step": 2027 }, { "epoch": 0.07, "learning_rate": 4.966974156806056e-05, "loss": 0.8195, "step": 2028 }, { "epoch": 0.07, "learning_rate": 4.966931010701056e-05, "loss": 1.0367, "step": 2029 }, { "epoch": 0.07, "learning_rate": 4.966887836618311e-05, "loss": 0.8956, "step": 2030 }, { "epoch": 0.07, "learning_rate": 4.96684463455831e-05, "loss": 1.0338, "step": 2031 }, { "epoch": 0.07, "learning_rate": 4.9668014045215416e-05, "loss": 0.8831, "step": 2032 }, { "epoch": 0.07, "learning_rate": 4.966758146508499e-05, "loss": 0.8924, "step": 2033 }, { "epoch": 0.07, "learning_rate": 4.96671486051967e-05, "loss": 0.9514, "step": 2034 }, { "epoch": 0.07, "learning_rate": 4.966671546555548e-05, "loss": 0.5085, "step": 2035 }, { "epoch": 0.07, "learning_rate": 4.966628204616622e-05, "loss": 0.4552, "step": 2036 }, { "epoch": 0.07, "learning_rate": 4.966584834703385e-05, "loss": 1.1246, "step": 2037 }, { "epoch": 0.07, "learning_rate": 4.966541436816328e-05, "loss": 0.8173, "step": 2038 }, { "epoch": 0.07, "learning_rate": 4.966498010955944e-05, "loss": 0.9175, "step": 2039 }, { "epoch": 0.07, "learning_rate": 4.966454557122725e-05, "loss": 1.5095, "step": 2040 }, { "epoch": 0.07, "learning_rate": 4.966411075317163e-05, "loss": 1.698, "step": 2041 }, { "epoch": 0.07, "learning_rate": 4.966367565539753e-05, "loss": 1.4871, "step": 2042 }, { "epoch": 0.07, "learning_rate": 4.9663240277909875e-05, "loss": 1.4628, "step": 2043 }, { "epoch": 0.07, "learning_rate": 4.96628046207136e-05, "loss": 2.785, "step": 2044 }, { "epoch": 0.07, "learning_rate": 4.966236868381365e-05, "loss": 5.4314, "step": 2045 }, { "epoch": 0.07, "learning_rate": 4.9661932467214965e-05, "loss": 1.1213, "step": 2046 }, { "epoch": 0.07, "learning_rate": 4.966149597092249e-05, "loss": 0.9036, "step": 2047 }, { "epoch": 0.07, "learning_rate": 4.966105919494118e-05, "loss": 1.0946, "step": 2048 }, { "epoch": 0.07, "learning_rate": 4.9660622139276005e-05, "loss": 0.4428, "step": 2049 }, { "epoch": 0.07, "learning_rate": 4.9660184803931885e-05, "loss": 0.2519, "step": 2050 }, { "epoch": 0.07, "learning_rate": 4.965974718891381e-05, "loss": 0.2532, "step": 2051 }, { "epoch": 0.07, "learning_rate": 4.965930929422673e-05, "loss": 0.6488, "step": 2052 }, { "epoch": 0.07, "learning_rate": 4.965887111987561e-05, "loss": 0.8332, "step": 2053 }, { "epoch": 0.07, "learning_rate": 4.965843266586543e-05, "loss": 0.8781, "step": 2054 }, { "epoch": 0.07, "learning_rate": 4.9657993932201153e-05, "loss": 1.0795, "step": 2055 }, { "epoch": 0.07, "learning_rate": 4.965755491888775e-05, "loss": 1.5386, "step": 2056 }, { "epoch": 0.07, "learning_rate": 4.965711562593022e-05, "loss": 1.7001, "step": 2057 }, { "epoch": 0.07, "learning_rate": 4.965667605333353e-05, "loss": 1.753, "step": 2058 }, { "epoch": 0.07, "learning_rate": 4.965623620110267e-05, "loss": 1.4256, "step": 2059 }, { "epoch": 0.07, "learning_rate": 4.965579606924262e-05, "loss": 0.8681, "step": 2060 }, { "epoch": 0.07, "learning_rate": 4.9655355657758376e-05, "loss": 0.2955, "step": 2061 }, { "epoch": 0.07, "learning_rate": 4.965491496665494e-05, "loss": 1.117, "step": 2062 }, { "epoch": 0.07, "learning_rate": 4.9654473995937304e-05, "loss": 1.9555, "step": 2063 }, { "epoch": 0.07, "learning_rate": 4.9654032745610466e-05, "loss": 0.9118, "step": 2064 }, { "epoch": 0.07, "learning_rate": 4.965359121567944e-05, "loss": 0.9596, "step": 2065 }, { "epoch": 0.07, "learning_rate": 4.965314940614922e-05, "loss": 0.8791, "step": 2066 }, { "epoch": 0.07, "learning_rate": 4.9652707317024834e-05, "loss": 1.0214, "step": 2067 }, { "epoch": 0.07, "learning_rate": 4.9652264948311275e-05, "loss": 0.8958, "step": 2068 }, { "epoch": 0.07, "learning_rate": 4.9651822300013586e-05, "loss": 0.8326, "step": 2069 }, { "epoch": 0.07, "learning_rate": 4.9651379372136756e-05, "loss": 0.815, "step": 2070 }, { "epoch": 0.07, "learning_rate": 4.965093616468583e-05, "loss": 0.8448, "step": 2071 }, { "epoch": 0.07, "learning_rate": 4.965049267766583e-05, "loss": 1.4906, "step": 2072 }, { "epoch": 0.07, "learning_rate": 4.965004891108178e-05, "loss": 1.8659, "step": 2073 }, { "epoch": 0.07, "learning_rate": 4.9649604864938726e-05, "loss": 1.933, "step": 2074 }, { "epoch": 0.07, "learning_rate": 4.964916053924168e-05, "loss": 0.9044, "step": 2075 }, { "epoch": 0.07, "learning_rate": 4.964871593399571e-05, "loss": 1.513, "step": 2076 }, { "epoch": 0.07, "learning_rate": 4.964827104920584e-05, "loss": 1.6313, "step": 2077 }, { "epoch": 0.07, "learning_rate": 4.9647825884877117e-05, "loss": 1.4377, "step": 2078 }, { "epoch": 0.07, "learning_rate": 4.96473804410146e-05, "loss": 1.2618, "step": 2079 }, { "epoch": 0.07, "learning_rate": 4.964693471762333e-05, "loss": 1.1856, "step": 2080 }, { "epoch": 0.07, "learning_rate": 4.964648871470836e-05, "loss": 1.0188, "step": 2081 }, { "epoch": 0.07, "learning_rate": 4.964604243227477e-05, "loss": 0.8947, "step": 2082 }, { "epoch": 0.07, "learning_rate": 4.9645595870327584e-05, "loss": 0.351, "step": 2083 }, { "epoch": 0.07, "learning_rate": 4.9645149028871895e-05, "loss": 0.4171, "step": 2084 }, { "epoch": 0.07, "learning_rate": 4.964470190791277e-05, "loss": 1.0616, "step": 2085 }, { "epoch": 0.07, "learning_rate": 4.9644254507455266e-05, "loss": 1.1026, "step": 2086 }, { "epoch": 0.07, "learning_rate": 4.9643806827504464e-05, "loss": 0.9564, "step": 2087 }, { "epoch": 0.07, "learning_rate": 4.9643358868065444e-05, "loss": 0.8416, "step": 2088 }, { "epoch": 0.07, "learning_rate": 4.964291062914328e-05, "loss": 0.9641, "step": 2089 }, { "epoch": 0.07, "learning_rate": 4.964246211074307e-05, "loss": 1.1381, "step": 2090 }, { "epoch": 0.07, "learning_rate": 4.964201331286988e-05, "loss": 3.9055, "step": 2091 }, { "epoch": 0.07, "learning_rate": 4.964156423552881e-05, "loss": 5.301, "step": 2092 }, { "epoch": 0.07, "learning_rate": 4.9641114878724956e-05, "loss": 4.9497, "step": 2093 }, { "epoch": 0.07, "learning_rate": 4.964066524246341e-05, "loss": 3.1069, "step": 2094 }, { "epoch": 0.07, "learning_rate": 4.964021532674927e-05, "loss": 0.8642, "step": 2095 }, { "epoch": 0.07, "learning_rate": 4.963976513158764e-05, "loss": 0.6282, "step": 2096 }, { "epoch": 0.07, "learning_rate": 4.9639314656983625e-05, "loss": 0.8217, "step": 2097 }, { "epoch": 0.07, "learning_rate": 4.963886390294235e-05, "loss": 0.6443, "step": 2098 }, { "epoch": 0.07, "learning_rate": 4.96384128694689e-05, "loss": 0.8923, "step": 2099 }, { "epoch": 0.07, "learning_rate": 4.9637961556568404e-05, "loss": 0.8598, "step": 2100 }, { "epoch": 0.07, "learning_rate": 4.9637509964245975e-05, "loss": 0.8204, "step": 2101 }, { "epoch": 0.07, "learning_rate": 4.963705809250675e-05, "loss": 0.8897, "step": 2102 }, { "epoch": 0.07, "learning_rate": 4.963660594135584e-05, "loss": 0.834, "step": 2103 }, { "epoch": 0.07, "learning_rate": 4.963615351079837e-05, "loss": 1.0061, "step": 2104 }, { "epoch": 0.07, "learning_rate": 4.963570080083947e-05, "loss": 1.0121, "step": 2105 }, { "epoch": 0.07, "learning_rate": 4.963524781148429e-05, "loss": 0.2558, "step": 2106 }, { "epoch": 0.07, "learning_rate": 4.963479454273796e-05, "loss": 0.3408, "step": 2107 }, { "epoch": 0.07, "learning_rate": 4.963434099460561e-05, "loss": 0.4055, "step": 2108 }, { "epoch": 0.07, "learning_rate": 4.9633887167092404e-05, "loss": 1.0708, "step": 2109 }, { "epoch": 0.07, "learning_rate": 4.963343306020347e-05, "loss": 1.1339, "step": 2110 }, { "epoch": 0.07, "learning_rate": 4.9632978673943964e-05, "loss": 0.9495, "step": 2111 }, { "epoch": 0.07, "learning_rate": 4.9632524008319045e-05, "loss": 0.9357, "step": 2112 }, { "epoch": 0.07, "learning_rate": 4.9632069063333864e-05, "loss": 1.1094, "step": 2113 }, { "epoch": 0.07, "learning_rate": 4.963161383899358e-05, "loss": 0.9951, "step": 2114 }, { "epoch": 0.07, "learning_rate": 4.963115833530336e-05, "loss": 0.8298, "step": 2115 }, { "epoch": 0.07, "learning_rate": 4.963070255226836e-05, "loss": 0.2721, "step": 2116 }, { "epoch": 0.07, "learning_rate": 4.963024648989376e-05, "loss": 0.3857, "step": 2117 }, { "epoch": 0.07, "learning_rate": 4.962979014818473e-05, "loss": 1.0268, "step": 2118 }, { "epoch": 0.07, "learning_rate": 4.962933352714645e-05, "loss": 0.8444, "step": 2119 }, { "epoch": 0.07, "learning_rate": 4.962887662678408e-05, "loss": 0.8812, "step": 2120 }, { "epoch": 0.07, "learning_rate": 4.9628419447102824e-05, "loss": 0.1471, "step": 2121 }, { "epoch": 0.07, "learning_rate": 4.962796198810785e-05, "loss": 1.0435, "step": 2122 }, { "epoch": 0.07, "learning_rate": 4.9627504249804365e-05, "loss": 0.7184, "step": 2123 }, { "epoch": 0.07, "learning_rate": 4.962704623219754e-05, "loss": 1.023, "step": 2124 }, { "epoch": 0.07, "learning_rate": 4.962658793529258e-05, "loss": 1.054, "step": 2125 }, { "epoch": 0.07, "learning_rate": 4.962612935909468e-05, "loss": 0.7296, "step": 2126 }, { "epoch": 0.07, "learning_rate": 4.962567050360904e-05, "loss": 1.2766, "step": 2127 }, { "epoch": 0.07, "learning_rate": 4.962521136884087e-05, "loss": 1.0125, "step": 2128 }, { "epoch": 0.07, "learning_rate": 4.9624751954795364e-05, "loss": 0.864, "step": 2129 }, { "epoch": 0.07, "learning_rate": 4.962429226147775e-05, "loss": 0.5849, "step": 2130 }, { "epoch": 0.07, "learning_rate": 4.962383228889323e-05, "loss": 0.8335, "step": 2131 }, { "epoch": 0.07, "learning_rate": 4.9623372037047026e-05, "loss": 0.8422, "step": 2132 }, { "epoch": 0.07, "learning_rate": 4.962291150594434e-05, "loss": 0.9573, "step": 2133 }, { "epoch": 0.07, "learning_rate": 4.962245069559042e-05, "loss": 1.1188, "step": 2134 }, { "epoch": 0.07, "learning_rate": 4.962198960599048e-05, "loss": 1.0163, "step": 2135 }, { "epoch": 0.07, "learning_rate": 4.962152823714975e-05, "loss": 0.9676, "step": 2136 }, { "epoch": 0.07, "learning_rate": 4.962106658907347e-05, "loss": 0.8574, "step": 2137 }, { "epoch": 0.07, "learning_rate": 4.962060466176687e-05, "loss": 1.5937, "step": 2138 }, { "epoch": 0.07, "learning_rate": 4.962014245523517e-05, "loss": 1.762, "step": 2139 }, { "epoch": 0.07, "learning_rate": 4.9619679969483646e-05, "loss": 1.7882, "step": 2140 }, { "epoch": 0.07, "learning_rate": 4.961921720451752e-05, "loss": 1.2215, "step": 2141 }, { "epoch": 0.07, "learning_rate": 4.961875416034205e-05, "loss": 0.9992, "step": 2142 }, { "epoch": 0.07, "learning_rate": 4.961829083696248e-05, "loss": 1.0223, "step": 2143 }, { "epoch": 0.07, "learning_rate": 4.961782723438407e-05, "loss": 0.8906, "step": 2144 }, { "epoch": 0.07, "learning_rate": 4.961736335261209e-05, "loss": 0.6203, "step": 2145 }, { "epoch": 0.07, "learning_rate": 4.9616899191651775e-05, "loss": 1.3863, "step": 2146 }, { "epoch": 0.07, "learning_rate": 4.96164347515084e-05, "loss": 0.9227, "step": 2147 }, { "epoch": 0.07, "learning_rate": 4.9615970032187244e-05, "loss": 0.9012, "step": 2148 }, { "epoch": 0.07, "learning_rate": 4.9615505033693554e-05, "loss": 0.973, "step": 2149 }, { "epoch": 0.07, "learning_rate": 4.9615039756032625e-05, "loss": 0.9991, "step": 2150 }, { "epoch": 0.07, "learning_rate": 4.961457419920973e-05, "loss": 0.9429, "step": 2151 }, { "epoch": 0.07, "learning_rate": 4.9614108363230135e-05, "loss": 0.8452, "step": 2152 }, { "epoch": 0.07, "learning_rate": 4.961364224809914e-05, "loss": 0.7396, "step": 2153 }, { "epoch": 0.07, "learning_rate": 4.9613175853822024e-05, "loss": 0.9901, "step": 2154 }, { "epoch": 0.07, "learning_rate": 4.961270918040407e-05, "loss": 0.5027, "step": 2155 }, { "epoch": 0.07, "learning_rate": 4.961224222785058e-05, "loss": 1.1846, "step": 2156 }, { "epoch": 0.07, "learning_rate": 4.9611774996166854e-05, "loss": 0.8844, "step": 2157 }, { "epoch": 0.07, "learning_rate": 4.961130748535818e-05, "loss": 1.1019, "step": 2158 }, { "epoch": 0.07, "learning_rate": 4.961083969542986e-05, "loss": 1.1182, "step": 2159 }, { "epoch": 0.07, "learning_rate": 4.961037162638721e-05, "loss": 1.8784, "step": 2160 }, { "epoch": 0.07, "learning_rate": 4.9609903278235526e-05, "loss": 4.3328, "step": 2161 }, { "epoch": 0.07, "learning_rate": 4.9609434650980124e-05, "loss": 0.9051, "step": 2162 }, { "epoch": 0.07, "learning_rate": 4.960896574462632e-05, "loss": 0.9241, "step": 2163 }, { "epoch": 0.07, "learning_rate": 4.960849655917944e-05, "loss": 0.9009, "step": 2164 }, { "epoch": 0.07, "learning_rate": 4.9608027094644796e-05, "loss": 0.886, "step": 2165 }, { "epoch": 0.07, "learning_rate": 4.960755735102771e-05, "loss": 1.0919, "step": 2166 }, { "epoch": 0.07, "learning_rate": 4.960708732833351e-05, "loss": 1.4732, "step": 2167 }, { "epoch": 0.07, "learning_rate": 4.9606617026567534e-05, "loss": 1.396, "step": 2168 }, { "epoch": 0.07, "learning_rate": 4.960614644573511e-05, "loss": 1.4205, "step": 2169 }, { "epoch": 0.07, "learning_rate": 4.960567558584157e-05, "loss": 1.3897, "step": 2170 }, { "epoch": 0.07, "learning_rate": 4.960520444689226e-05, "loss": 1.3593, "step": 2171 }, { "epoch": 0.07, "learning_rate": 4.960473302889254e-05, "loss": 1.2127, "step": 2172 }, { "epoch": 0.07, "learning_rate": 4.9604261331847724e-05, "loss": 0.934, "step": 2173 }, { "epoch": 0.07, "learning_rate": 4.960378935576317e-05, "loss": 0.9591, "step": 2174 }, { "epoch": 0.07, "learning_rate": 4.960331710064425e-05, "loss": 0.9399, "step": 2175 }, { "epoch": 0.07, "learning_rate": 4.960284456649631e-05, "loss": 0.6119, "step": 2176 }, { "epoch": 0.07, "learning_rate": 4.960237175332471e-05, "loss": 0.9262, "step": 2177 }, { "epoch": 0.07, "learning_rate": 4.96018986611348e-05, "loss": 1.2139, "step": 2178 }, { "epoch": 0.07, "learning_rate": 4.9601425289931955e-05, "loss": 0.7921, "step": 2179 }, { "epoch": 0.07, "learning_rate": 4.9600951639721546e-05, "loss": 0.6226, "step": 2180 }, { "epoch": 0.07, "learning_rate": 4.960047771050894e-05, "loss": 1.036, "step": 2181 }, { "epoch": 0.07, "learning_rate": 4.960000350229952e-05, "loss": 0.7531, "step": 2182 }, { "epoch": 0.07, "learning_rate": 4.959952901509865e-05, "loss": 0.9118, "step": 2183 }, { "epoch": 0.07, "learning_rate": 4.959905424891172e-05, "loss": 2.0265, "step": 2184 }, { "epoch": 0.07, "learning_rate": 4.959857920374412e-05, "loss": 1.1565, "step": 2185 }, { "epoch": 0.07, "learning_rate": 4.959810387960122e-05, "loss": 0.8424, "step": 2186 }, { "epoch": 0.07, "learning_rate": 4.959762827648843e-05, "loss": 1.0198, "step": 2187 }, { "epoch": 0.07, "learning_rate": 4.9597152394411136e-05, "loss": 1.5395, "step": 2188 }, { "epoch": 0.07, "learning_rate": 4.959667623337473e-05, "loss": 1.1709, "step": 2189 }, { "epoch": 0.07, "learning_rate": 4.959619979338461e-05, "loss": 0.8787, "step": 2190 }, { "epoch": 0.07, "learning_rate": 4.95957230744462e-05, "loss": 0.4031, "step": 2191 }, { "epoch": 0.07, "learning_rate": 4.959524607656489e-05, "loss": 0.8967, "step": 2192 }, { "epoch": 0.07, "learning_rate": 4.959476879974608e-05, "loss": 1.1441, "step": 2193 }, { "epoch": 0.07, "learning_rate": 4.959429124399521e-05, "loss": 0.9792, "step": 2194 }, { "epoch": 0.07, "learning_rate": 4.959381340931768e-05, "loss": 0.7157, "step": 2195 }, { "epoch": 0.07, "learning_rate": 4.95933352957189e-05, "loss": 0.9456, "step": 2196 }, { "epoch": 0.07, "learning_rate": 4.959285690320431e-05, "loss": 0.5399, "step": 2197 }, { "epoch": 0.07, "learning_rate": 4.959237823177933e-05, "loss": 1.2102, "step": 2198 }, { "epoch": 0.07, "learning_rate": 4.959189928144938e-05, "loss": 0.622, "step": 2199 }, { "epoch": 0.07, "learning_rate": 4.959142005221991e-05, "loss": 0.6693, "step": 2200 }, { "epoch": 0.07, "learning_rate": 4.9590940544096326e-05, "loss": 0.9032, "step": 2201 }, { "epoch": 0.07, "learning_rate": 4.9590460757084096e-05, "loss": 0.8814, "step": 2202 }, { "epoch": 0.07, "learning_rate": 4.958998069118864e-05, "loss": 0.7885, "step": 2203 }, { "epoch": 0.07, "learning_rate": 4.9589500346415425e-05, "loss": 0.8543, "step": 2204 }, { "epoch": 0.07, "learning_rate": 4.958901972276988e-05, "loss": 0.4653, "step": 2205 }, { "epoch": 0.07, "learning_rate": 4.9588538820257455e-05, "loss": 0.4139, "step": 2206 }, { "epoch": 0.07, "learning_rate": 4.9588057638883616e-05, "loss": 1.1948, "step": 2207 }, { "epoch": 0.07, "learning_rate": 4.958757617865382e-05, "loss": 1.1112, "step": 2208 }, { "epoch": 0.07, "learning_rate": 4.95870944395735e-05, "loss": 0.9271, "step": 2209 }, { "epoch": 0.07, "learning_rate": 4.958661242164816e-05, "loss": 1.0909, "step": 2210 }, { "epoch": 0.07, "learning_rate": 4.958613012488324e-05, "loss": 0.8324, "step": 2211 }, { "epoch": 0.07, "learning_rate": 4.958564754928423e-05, "loss": 0.8743, "step": 2212 }, { "epoch": 0.07, "learning_rate": 4.9585164694856576e-05, "loss": 1.0307, "step": 2213 }, { "epoch": 0.07, "learning_rate": 4.958468156160577e-05, "loss": 0.887, "step": 2214 }, { "epoch": 0.07, "learning_rate": 4.958419814953729e-05, "loss": 1.0556, "step": 2215 }, { "epoch": 0.07, "learning_rate": 4.958371445865663e-05, "loss": 0.9152, "step": 2216 }, { "epoch": 0.07, "learning_rate": 4.958323048896924e-05, "loss": 0.7663, "step": 2217 }, { "epoch": 0.07, "learning_rate": 4.958274624048065e-05, "loss": 0.0893, "step": 2218 }, { "epoch": 0.07, "learning_rate": 4.958226171319633e-05, "loss": 0.7447, "step": 2219 }, { "epoch": 0.07, "learning_rate": 4.9581776907121784e-05, "loss": 1.7754, "step": 2220 }, { "epoch": 0.07, "learning_rate": 4.95812918222625e-05, "loss": 0.974, "step": 2221 }, { "epoch": 0.07, "learning_rate": 4.958080645862399e-05, "loss": 1.0459, "step": 2222 }, { "epoch": 0.07, "learning_rate": 4.958032081621175e-05, "loss": 0.8309, "step": 2223 }, { "epoch": 0.07, "learning_rate": 4.957983489503129e-05, "loss": 0.9008, "step": 2224 }, { "epoch": 0.07, "learning_rate": 4.9579348695088124e-05, "loss": 1.0311, "step": 2225 }, { "epoch": 0.07, "learning_rate": 4.957886221638777e-05, "loss": 0.8339, "step": 2226 }, { "epoch": 0.07, "learning_rate": 4.957837545893573e-05, "loss": 0.7398, "step": 2227 }, { "epoch": 0.07, "learning_rate": 4.957788842273754e-05, "loss": 1.0405, "step": 2228 }, { "epoch": 0.07, "learning_rate": 4.957740110779872e-05, "loss": 1.0143, "step": 2229 }, { "epoch": 0.07, "learning_rate": 4.957691351412478e-05, "loss": 1.0468, "step": 2230 }, { "epoch": 0.07, "learning_rate": 4.957642564172127e-05, "loss": 0.9698, "step": 2231 }, { "epoch": 0.07, "learning_rate": 4.9575937490593726e-05, "loss": 0.8154, "step": 2232 }, { "epoch": 0.07, "learning_rate": 4.957544906074767e-05, "loss": 0.7287, "step": 2233 }, { "epoch": 0.07, "learning_rate": 4.9574960352188645e-05, "loss": 1.5939, "step": 2234 }, { "epoch": 0.07, "learning_rate": 4.9574471364922195e-05, "loss": 0.9146, "step": 2235 }, { "epoch": 0.07, "learning_rate": 4.9573982098953866e-05, "loss": 0.8242, "step": 2236 }, { "epoch": 0.07, "learning_rate": 4.9573492554289206e-05, "loss": 0.89, "step": 2237 }, { "epoch": 0.07, "learning_rate": 4.957300273093377e-05, "loss": 0.9018, "step": 2238 }, { "epoch": 0.07, "learning_rate": 4.957251262889311e-05, "loss": 0.9982, "step": 2239 }, { "epoch": 0.07, "learning_rate": 4.9572022248172786e-05, "loss": 0.6377, "step": 2240 }, { "epoch": 0.07, "learning_rate": 4.957153158877835e-05, "loss": 0.6488, "step": 2241 }, { "epoch": 0.07, "learning_rate": 4.957104065071538e-05, "loss": 0.5501, "step": 2242 }, { "epoch": 0.07, "learning_rate": 4.957054943398944e-05, "loss": 0.9687, "step": 2243 }, { "epoch": 0.07, "learning_rate": 4.957005793860611e-05, "loss": 0.7523, "step": 2244 }, { "epoch": 0.07, "learning_rate": 4.9569566164570933e-05, "loss": 0.6036, "step": 2245 }, { "epoch": 0.07, "learning_rate": 4.9569074111889524e-05, "loss": 0.8774, "step": 2246 }, { "epoch": 0.07, "learning_rate": 4.956858178056744e-05, "loss": 0.8445, "step": 2247 }, { "epoch": 0.07, "learning_rate": 4.9568089170610264e-05, "loss": 1.0625, "step": 2248 }, { "epoch": 0.07, "learning_rate": 4.9567596282023596e-05, "loss": 2.0525, "step": 2249 }, { "epoch": 0.07, "learning_rate": 4.956710311481303e-05, "loss": 5.3786, "step": 2250 }, { "epoch": 0.08, "learning_rate": 4.9566609668984134e-05, "loss": 5.4003, "step": 2251 }, { "epoch": 0.08, "learning_rate": 4.956611594454253e-05, "loss": 5.2536, "step": 2252 }, { "epoch": 0.08, "learning_rate": 4.95656219414938e-05, "loss": 2.2886, "step": 2253 }, { "epoch": 0.08, "learning_rate": 4.956512765984355e-05, "loss": 0.6737, "step": 2254 }, { "epoch": 0.08, "learning_rate": 4.956463309959739e-05, "loss": 1.1335, "step": 2255 }, { "epoch": 0.08, "learning_rate": 4.9564138260760935e-05, "loss": 1.2091, "step": 2256 }, { "epoch": 0.08, "learning_rate": 4.9563643143339786e-05, "loss": 1.1221, "step": 2257 }, { "epoch": 0.08, "learning_rate": 4.9563147747339556e-05, "loss": 1.1912, "step": 2258 }, { "epoch": 0.08, "learning_rate": 4.956265207276588e-05, "loss": 1.0041, "step": 2259 }, { "epoch": 0.08, "learning_rate": 4.956215611962436e-05, "loss": 0.987, "step": 2260 }, { "epoch": 0.08, "learning_rate": 4.956165988792063e-05, "loss": 0.7852, "step": 2261 }, { "epoch": 0.08, "learning_rate": 4.956116337766031e-05, "loss": 0.8924, "step": 2262 }, { "epoch": 0.08, "learning_rate": 4.956066658884905e-05, "loss": 0.7657, "step": 2263 }, { "epoch": 0.08, "learning_rate": 4.956016952149247e-05, "loss": 0.8155, "step": 2264 }, { "epoch": 0.08, "learning_rate": 4.9559672175596214e-05, "loss": 0.7761, "step": 2265 }, { "epoch": 0.08, "learning_rate": 4.955917455116591e-05, "loss": 0.9653, "step": 2266 }, { "epoch": 0.08, "learning_rate": 4.955867664820721e-05, "loss": 0.9433, "step": 2267 }, { "epoch": 0.08, "learning_rate": 4.9558178466725754e-05, "loss": 0.9152, "step": 2268 }, { "epoch": 0.08, "learning_rate": 4.955768000672722e-05, "loss": 1.016, "step": 2269 }, { "epoch": 0.08, "learning_rate": 4.9557181268217227e-05, "loss": 0.6376, "step": 2270 }, { "epoch": 0.08, "learning_rate": 4.9556682251201434e-05, "loss": 0.7779, "step": 2271 }, { "epoch": 0.08, "learning_rate": 4.955618295568552e-05, "loss": 0.9329, "step": 2272 }, { "epoch": 0.08, "learning_rate": 4.955568338167513e-05, "loss": 1.0999, "step": 2273 }, { "epoch": 0.08, "learning_rate": 4.955518352917595e-05, "loss": 1.1448, "step": 2274 }, { "epoch": 0.08, "learning_rate": 4.9554683398193626e-05, "loss": 0.8572, "step": 2275 }, { "epoch": 0.08, "learning_rate": 4.9554182988733846e-05, "loss": 0.3256, "step": 2276 }, { "epoch": 0.08, "learning_rate": 4.9553682300802274e-05, "loss": 0.9316, "step": 2277 }, { "epoch": 0.08, "learning_rate": 4.9553181334404596e-05, "loss": 0.9046, "step": 2278 }, { "epoch": 0.08, "learning_rate": 4.9552680089546496e-05, "loss": 0.8338, "step": 2279 }, { "epoch": 0.08, "learning_rate": 4.955217856623365e-05, "loss": 1.0285, "step": 2280 }, { "epoch": 0.08, "learning_rate": 4.955167676447174e-05, "loss": 0.7089, "step": 2281 }, { "epoch": 0.08, "learning_rate": 4.955117468426649e-05, "loss": 0.8696, "step": 2282 }, { "epoch": 0.08, "learning_rate": 4.955067232562355e-05, "loss": 1.0711, "step": 2283 }, { "epoch": 0.08, "learning_rate": 4.9550169688548645e-05, "loss": 0.7661, "step": 2284 }, { "epoch": 0.08, "learning_rate": 4.954966677304747e-05, "loss": 1.0913, "step": 2285 }, { "epoch": 0.08, "learning_rate": 4.954916357912573e-05, "loss": 1.1768, "step": 2286 }, { "epoch": 0.08, "learning_rate": 4.954866010678912e-05, "loss": 0.9411, "step": 2287 }, { "epoch": 0.08, "learning_rate": 4.9548156356043375e-05, "loss": 0.8748, "step": 2288 }, { "epoch": 0.08, "learning_rate": 4.954765232689418e-05, "loss": 0.9327, "step": 2289 }, { "epoch": 0.08, "learning_rate": 4.954714801934726e-05, "loss": 0.9998, "step": 2290 }, { "epoch": 0.08, "learning_rate": 4.954664343340834e-05, "loss": 0.9379, "step": 2291 }, { "epoch": 0.08, "learning_rate": 4.954613856908315e-05, "loss": 0.8453, "step": 2292 }, { "epoch": 0.08, "learning_rate": 4.9545633426377404e-05, "loss": 0.5321, "step": 2293 }, { "epoch": 0.08, "learning_rate": 4.954512800529683e-05, "loss": 0.4201, "step": 2294 }, { "epoch": 0.08, "learning_rate": 4.954462230584716e-05, "loss": 0.4703, "step": 2295 }, { "epoch": 0.08, "learning_rate": 4.954411632803413e-05, "loss": 1.0868, "step": 2296 }, { "epoch": 0.08, "learning_rate": 4.954361007186349e-05, "loss": 0.9219, "step": 2297 }, { "epoch": 0.08, "learning_rate": 4.954310353734096e-05, "loss": 0.7483, "step": 2298 }, { "epoch": 0.08, "learning_rate": 4.9542596724472306e-05, "loss": 0.8627, "step": 2299 }, { "epoch": 0.08, "learning_rate": 4.954208963326327e-05, "loss": 0.8595, "step": 2300 }, { "epoch": 0.08, "learning_rate": 4.9541582263719586e-05, "loss": 0.6844, "step": 2301 }, { "epoch": 0.08, "learning_rate": 4.954107461584703e-05, "loss": 0.7554, "step": 2302 }, { "epoch": 0.08, "learning_rate": 4.954056668965135e-05, "loss": 1.0599, "step": 2303 }, { "epoch": 0.08, "learning_rate": 4.9540058485138306e-05, "loss": 1.1298, "step": 2304 }, { "epoch": 0.08, "learning_rate": 4.9539550002313666e-05, "loss": 0.0848, "step": 2305 }, { "epoch": 0.08, "learning_rate": 4.953904124118319e-05, "loss": 0.2406, "step": 2306 }, { "epoch": 0.08, "learning_rate": 4.9538532201752665e-05, "loss": 1.0493, "step": 2307 }, { "epoch": 0.08, "learning_rate": 4.9538022884027835e-05, "loss": 1.0107, "step": 2308 }, { "epoch": 0.08, "learning_rate": 4.95375132880145e-05, "loss": 0.9121, "step": 2309 }, { "epoch": 0.08, "learning_rate": 4.953700341371842e-05, "loss": 0.5396, "step": 2310 }, { "epoch": 0.08, "learning_rate": 4.95364932611454e-05, "loss": 0.4276, "step": 2311 }, { "epoch": 0.08, "learning_rate": 4.953598283030121e-05, "loss": 1.0514, "step": 2312 }, { "epoch": 0.08, "learning_rate": 4.953547212119165e-05, "loss": 0.9164, "step": 2313 }, { "epoch": 0.08, "learning_rate": 4.95349611338225e-05, "loss": 0.8919, "step": 2314 }, { "epoch": 0.08, "learning_rate": 4.953444986819956e-05, "loss": 0.7691, "step": 2315 }, { "epoch": 0.08, "learning_rate": 4.953393832432863e-05, "loss": 0.9234, "step": 2316 }, { "epoch": 0.08, "learning_rate": 4.953342650221552e-05, "loss": 0.9591, "step": 2317 }, { "epoch": 0.08, "learning_rate": 4.953291440186601e-05, "loss": 0.8501, "step": 2318 }, { "epoch": 0.08, "learning_rate": 4.953240202328593e-05, "loss": 0.9662, "step": 2319 }, { "epoch": 0.08, "learning_rate": 4.9531889366481086e-05, "loss": 0.8571, "step": 2320 }, { "epoch": 0.08, "learning_rate": 4.953137643145729e-05, "loss": 1.007, "step": 2321 }, { "epoch": 0.08, "learning_rate": 4.9530863218220355e-05, "loss": 0.626, "step": 2322 }, { "epoch": 0.08, "learning_rate": 4.9530349726776105e-05, "loss": 0.7934, "step": 2323 }, { "epoch": 0.08, "learning_rate": 4.952983595713037e-05, "loss": 0.8092, "step": 2324 }, { "epoch": 0.08, "learning_rate": 4.952932190928896e-05, "loss": 0.8935, "step": 2325 }, { "epoch": 0.08, "learning_rate": 4.9528807583257724e-05, "loss": 1.7077, "step": 2326 }, { "epoch": 0.08, "learning_rate": 4.952829297904249e-05, "loss": 0.7928, "step": 2327 }, { "epoch": 0.08, "learning_rate": 4.952777809664908e-05, "loss": 0.8949, "step": 2328 }, { "epoch": 0.08, "learning_rate": 4.952726293608335e-05, "loss": 1.1347, "step": 2329 }, { "epoch": 0.08, "learning_rate": 4.9526747497351146e-05, "loss": 1.0083, "step": 2330 }, { "epoch": 0.08, "learning_rate": 4.952623178045829e-05, "loss": 3.0696, "step": 2331 }, { "epoch": 0.08, "learning_rate": 4.952571578541066e-05, "loss": 2.988, "step": 2332 }, { "epoch": 0.08, "learning_rate": 4.952519951221408e-05, "loss": 1.1619, "step": 2333 }, { "epoch": 0.08, "learning_rate": 4.952468296087443e-05, "loss": 0.9229, "step": 2334 }, { "epoch": 0.08, "learning_rate": 4.952416613139754e-05, "loss": 0.6426, "step": 2335 }, { "epoch": 0.08, "learning_rate": 4.9523649023789295e-05, "loss": 1.0379, "step": 2336 }, { "epoch": 0.08, "learning_rate": 4.952313163805556e-05, "loss": 0.9368, "step": 2337 }, { "epoch": 0.08, "learning_rate": 4.952261397420219e-05, "loss": 0.64, "step": 2338 }, { "epoch": 0.08, "learning_rate": 4.952209603223506e-05, "loss": 1.1005, "step": 2339 }, { "epoch": 0.08, "learning_rate": 4.9521577812160046e-05, "loss": 0.8774, "step": 2340 }, { "epoch": 0.08, "learning_rate": 4.952105931398303e-05, "loss": 0.8741, "step": 2341 }, { "epoch": 0.08, "learning_rate": 4.952054053770988e-05, "loss": 0.9794, "step": 2342 }, { "epoch": 0.08, "learning_rate": 4.952002148334649e-05, "loss": 0.8704, "step": 2343 }, { "epoch": 0.08, "learning_rate": 4.951950215089874e-05, "loss": 0.7921, "step": 2344 }, { "epoch": 0.08, "learning_rate": 4.9518982540372525e-05, "loss": 0.9508, "step": 2345 }, { "epoch": 0.08, "learning_rate": 4.951846265177373e-05, "loss": 0.6698, "step": 2346 }, { "epoch": 0.08, "learning_rate": 4.951794248510827e-05, "loss": 1.0788, "step": 2347 }, { "epoch": 0.08, "learning_rate": 4.951742204038202e-05, "loss": 0.663, "step": 2348 }, { "epoch": 0.08, "learning_rate": 4.9516901317600895e-05, "loss": 0.8145, "step": 2349 }, { "epoch": 0.08, "learning_rate": 4.9516380316770804e-05, "loss": 2.5243, "step": 2350 }, { "epoch": 0.08, "learning_rate": 4.951585903789764e-05, "loss": 2.8179, "step": 2351 }, { "epoch": 0.08, "learning_rate": 4.951533748098734e-05, "loss": 2.3861, "step": 2352 }, { "epoch": 0.08, "learning_rate": 4.9514815646045795e-05, "loss": 1.0639, "step": 2353 }, { "epoch": 0.08, "learning_rate": 4.9514293533078934e-05, "loss": 1.1249, "step": 2354 }, { "epoch": 0.08, "learning_rate": 4.951377114209268e-05, "loss": 1.6037, "step": 2355 }, { "epoch": 0.08, "learning_rate": 4.9513248473092956e-05, "loss": 1.0743, "step": 2356 }, { "epoch": 0.08, "learning_rate": 4.951272552608569e-05, "loss": 0.7678, "step": 2357 }, { "epoch": 0.08, "learning_rate": 4.9512202301076806e-05, "loss": 0.9279, "step": 2358 }, { "epoch": 0.08, "learning_rate": 4.951167879807224e-05, "loss": 0.7408, "step": 2359 }, { "epoch": 0.08, "learning_rate": 4.951115501707794e-05, "loss": 0.9533, "step": 2360 }, { "epoch": 0.08, "learning_rate": 4.951063095809983e-05, "loss": 0.7675, "step": 2361 }, { "epoch": 0.08, "learning_rate": 4.951010662114387e-05, "loss": 0.6476, "step": 2362 }, { "epoch": 0.08, "learning_rate": 4.9509582006216e-05, "loss": 0.9298, "step": 2363 }, { "epoch": 0.08, "learning_rate": 4.9509057113322165e-05, "loss": 0.3509, "step": 2364 }, { "epoch": 0.08, "learning_rate": 4.950853194246832e-05, "loss": 0.4155, "step": 2365 }, { "epoch": 0.08, "learning_rate": 4.950800649366042e-05, "loss": 0.7083, "step": 2366 }, { "epoch": 0.08, "learning_rate": 4.9507480766904435e-05, "loss": 0.2517, "step": 2367 }, { "epoch": 0.08, "learning_rate": 4.950695476220631e-05, "loss": 0.2173, "step": 2368 }, { "epoch": 0.08, "learning_rate": 4.950642847957202e-05, "loss": 0.3221, "step": 2369 }, { "epoch": 0.08, "learning_rate": 4.950590191900754e-05, "loss": 0.2676, "step": 2370 }, { "epoch": 0.08, "learning_rate": 4.9505375080518824e-05, "loss": 0.2508, "step": 2371 }, { "epoch": 0.08, "learning_rate": 4.950484796411187e-05, "loss": 0.716, "step": 2372 }, { "epoch": 0.08, "learning_rate": 4.950432056979263e-05, "loss": 0.8179, "step": 2373 }, { "epoch": 0.08, "learning_rate": 4.9503792897567106e-05, "loss": 0.9227, "step": 2374 }, { "epoch": 0.08, "learning_rate": 4.9503264947441275e-05, "loss": 1.2101, "step": 2375 }, { "epoch": 0.08, "learning_rate": 4.950273671942112e-05, "loss": 0.9721, "step": 2376 }, { "epoch": 0.08, "learning_rate": 4.950220821351265e-05, "loss": 0.9804, "step": 2377 }, { "epoch": 0.08, "learning_rate": 4.950167942972183e-05, "loss": 0.7685, "step": 2378 }, { "epoch": 0.08, "learning_rate": 4.950115036805468e-05, "loss": 0.9828, "step": 2379 }, { "epoch": 0.08, "learning_rate": 4.950062102851718e-05, "loss": 1.4533, "step": 2380 }, { "epoch": 0.08, "learning_rate": 4.950009141111536e-05, "loss": 0.827, "step": 2381 }, { "epoch": 0.08, "learning_rate": 4.949956151585521e-05, "loss": 1.2817, "step": 2382 }, { "epoch": 0.08, "learning_rate": 4.949903134274274e-05, "loss": 3.4427, "step": 2383 }, { "epoch": 0.08, "learning_rate": 4.949850089178397e-05, "loss": 1.1142, "step": 2384 }, { "epoch": 0.08, "learning_rate": 4.94979701629849e-05, "loss": 1.0201, "step": 2385 }, { "epoch": 0.08, "learning_rate": 4.949743915635157e-05, "loss": 0.8144, "step": 2386 }, { "epoch": 0.08, "learning_rate": 4.949690787188998e-05, "loss": 1.7174, "step": 2387 }, { "epoch": 0.08, "learning_rate": 4.949637630960617e-05, "loss": 1.679, "step": 2388 }, { "epoch": 0.08, "learning_rate": 4.949584446950617e-05, "loss": 1.4041, "step": 2389 }, { "epoch": 0.08, "learning_rate": 4.949531235159601e-05, "loss": 1.5324, "step": 2390 }, { "epoch": 0.08, "learning_rate": 4.949477995588172e-05, "loss": 1.7057, "step": 2391 }, { "epoch": 0.08, "learning_rate": 4.9494247282369336e-05, "loss": 1.7011, "step": 2392 }, { "epoch": 0.08, "learning_rate": 4.94937143310649e-05, "loss": 1.5231, "step": 2393 }, { "epoch": 0.08, "learning_rate": 4.949318110197447e-05, "loss": 1.0359, "step": 2394 }, { "epoch": 0.08, "learning_rate": 4.9492647595104074e-05, "loss": 1.0794, "step": 2395 }, { "epoch": 0.08, "learning_rate": 4.9492113810459775e-05, "loss": 1.3096, "step": 2396 }, { "epoch": 0.08, "learning_rate": 4.949157974804762e-05, "loss": 1.0493, "step": 2397 }, { "epoch": 0.08, "learning_rate": 4.949104540787367e-05, "loss": 1.1613, "step": 2398 }, { "epoch": 0.08, "learning_rate": 4.949051078994399e-05, "loss": 0.987, "step": 2399 }, { "epoch": 0.08, "learning_rate": 4.9489975894264636e-05, "loss": 3.2561, "step": 2400 }, { "epoch": 0.08, "learning_rate": 4.948944072084167e-05, "loss": 5.8947, "step": 2401 }, { "epoch": 0.08, "learning_rate": 4.948890526968117e-05, "loss": 5.7773, "step": 2402 }, { "epoch": 0.08, "learning_rate": 4.94883695407892e-05, "loss": 5.709, "step": 2403 }, { "epoch": 0.08, "learning_rate": 4.948783353417184e-05, "loss": 5.6518, "step": 2404 }, { "epoch": 0.08, "learning_rate": 4.948729724983519e-05, "loss": 5.6251, "step": 2405 }, { "epoch": 0.08, "learning_rate": 4.94867606877853e-05, "loss": 1.9653, "step": 2406 }, { "epoch": 0.08, "learning_rate": 4.9486223848028255e-05, "loss": 0.9904, "step": 2407 }, { "epoch": 0.08, "learning_rate": 4.948568673057017e-05, "loss": 0.655, "step": 2408 }, { "epoch": 0.08, "learning_rate": 4.948514933541711e-05, "loss": 0.9831, "step": 2409 }, { "epoch": 0.08, "learning_rate": 4.94846116625752e-05, "loss": 0.9544, "step": 2410 }, { "epoch": 0.08, "learning_rate": 4.9484073712050504e-05, "loss": 0.8454, "step": 2411 }, { "epoch": 0.08, "learning_rate": 4.948353548384915e-05, "loss": 0.4862, "step": 2412 }, { "epoch": 0.08, "learning_rate": 4.9482996977977224e-05, "loss": 0.7477, "step": 2413 }, { "epoch": 0.08, "learning_rate": 4.948245819444084e-05, "loss": 1.0784, "step": 2414 }, { "epoch": 0.08, "learning_rate": 4.948191913324611e-05, "loss": 1.0762, "step": 2415 }, { "epoch": 0.08, "learning_rate": 4.948137979439915e-05, "loss": 0.9141, "step": 2416 }, { "epoch": 0.08, "learning_rate": 4.948084017790607e-05, "loss": 0.8849, "step": 2417 }, { "epoch": 0.08, "learning_rate": 4.948030028377299e-05, "loss": 1.0311, "step": 2418 }, { "epoch": 0.08, "learning_rate": 4.947976011200604e-05, "loss": 1.1914, "step": 2419 }, { "epoch": 0.08, "learning_rate": 4.9479219662611334e-05, "loss": 0.6636, "step": 2420 }, { "epoch": 0.08, "learning_rate": 4.947867893559502e-05, "loss": 1.1805, "step": 2421 }, { "epoch": 0.08, "learning_rate": 4.947813793096321e-05, "loss": 0.8821, "step": 2422 }, { "epoch": 0.08, "learning_rate": 4.947759664872206e-05, "loss": 0.8904, "step": 2423 }, { "epoch": 0.08, "learning_rate": 4.947705508887769e-05, "loss": 0.8082, "step": 2424 }, { "epoch": 0.08, "learning_rate": 4.9476513251436244e-05, "loss": 0.8376, "step": 2425 }, { "epoch": 0.08, "learning_rate": 4.9475971136403874e-05, "loss": 0.2098, "step": 2426 }, { "epoch": 0.08, "learning_rate": 4.947542874378674e-05, "loss": 0.2797, "step": 2427 }, { "epoch": 0.08, "learning_rate": 4.947488607359097e-05, "loss": 1.209, "step": 2428 }, { "epoch": 0.08, "learning_rate": 4.947434312582273e-05, "loss": 0.9583, "step": 2429 }, { "epoch": 0.08, "learning_rate": 4.9473799900488185e-05, "loss": 1.5715, "step": 2430 }, { "epoch": 0.08, "learning_rate": 4.947325639759347e-05, "loss": 1.2618, "step": 2431 }, { "epoch": 0.08, "learning_rate": 4.947271261714478e-05, "loss": 0.9234, "step": 2432 }, { "epoch": 0.08, "learning_rate": 4.9472168559148264e-05, "loss": 0.9786, "step": 2433 }, { "epoch": 0.08, "learning_rate": 4.94716242236101e-05, "loss": 0.7317, "step": 2434 }, { "epoch": 0.08, "learning_rate": 4.947107961053645e-05, "loss": 0.6432, "step": 2435 }, { "epoch": 0.08, "learning_rate": 4.9470534719933505e-05, "loss": 0.853, "step": 2436 }, { "epoch": 0.08, "learning_rate": 4.9469989551807436e-05, "loss": 1.7939, "step": 2437 }, { "epoch": 0.08, "learning_rate": 4.9469444106164424e-05, "loss": 0.6877, "step": 2438 }, { "epoch": 0.08, "learning_rate": 4.9468898383010665e-05, "loss": 0.5666, "step": 2439 }, { "epoch": 0.08, "learning_rate": 4.946835238235234e-05, "loss": 0.5627, "step": 2440 }, { "epoch": 0.08, "learning_rate": 4.9467806104195634e-05, "loss": 0.7565, "step": 2441 }, { "epoch": 0.08, "learning_rate": 4.9467259548546764e-05, "loss": 1.2749, "step": 2442 }, { "epoch": 0.08, "learning_rate": 4.946671271541191e-05, "loss": 1.1304, "step": 2443 }, { "epoch": 0.08, "learning_rate": 4.946616560479729e-05, "loss": 0.8139, "step": 2444 }, { "epoch": 0.08, "learning_rate": 4.946561821670909e-05, "loss": 0.812, "step": 2445 }, { "epoch": 0.08, "learning_rate": 4.946507055115352e-05, "loss": 0.6871, "step": 2446 }, { "epoch": 0.08, "learning_rate": 4.9464522608136805e-05, "loss": 0.8843, "step": 2447 }, { "epoch": 0.08, "learning_rate": 4.946397438766515e-05, "loss": 1.061, "step": 2448 }, { "epoch": 0.08, "learning_rate": 4.946342588974477e-05, "loss": 0.9746, "step": 2449 }, { "epoch": 0.08, "learning_rate": 4.94628771143819e-05, "loss": 0.9009, "step": 2450 }, { "epoch": 0.08, "learning_rate": 4.946232806158275e-05, "loss": 0.8891, "step": 2451 }, { "epoch": 0.08, "learning_rate": 4.946177873135355e-05, "loss": 0.8578, "step": 2452 }, { "epoch": 0.08, "learning_rate": 4.9461229123700535e-05, "loss": 0.7999, "step": 2453 }, { "epoch": 0.08, "learning_rate": 4.9460679238629925e-05, "loss": 1.2052, "step": 2454 }, { "epoch": 0.08, "learning_rate": 4.9460129076147974e-05, "loss": 1.2379, "step": 2455 }, { "epoch": 0.08, "learning_rate": 4.9459578636260904e-05, "loss": 0.8584, "step": 2456 }, { "epoch": 0.08, "learning_rate": 4.9459027918974975e-05, "loss": 0.8787, "step": 2457 }, { "epoch": 0.08, "learning_rate": 4.9458476924296415e-05, "loss": 0.9607, "step": 2458 }, { "epoch": 0.08, "learning_rate": 4.945792565223149e-05, "loss": 0.9285, "step": 2459 }, { "epoch": 0.08, "learning_rate": 4.945737410278644e-05, "loss": 0.4283, "step": 2460 }, { "epoch": 0.08, "learning_rate": 4.9456822275967526e-05, "loss": 0.8703, "step": 2461 }, { "epoch": 0.08, "learning_rate": 4.9456270171781007e-05, "loss": 0.9261, "step": 2462 }, { "epoch": 0.08, "learning_rate": 4.945571779023313e-05, "loss": 0.9488, "step": 2463 }, { "epoch": 0.08, "learning_rate": 4.9455165131330186e-05, "loss": 1.0519, "step": 2464 }, { "epoch": 0.08, "learning_rate": 4.9454612195078415e-05, "loss": 1.3102, "step": 2465 }, { "epoch": 0.08, "learning_rate": 4.945405898148411e-05, "loss": 0.9715, "step": 2466 }, { "epoch": 0.08, "learning_rate": 4.945350549055353e-05, "loss": 1.2824, "step": 2467 }, { "epoch": 0.08, "learning_rate": 4.945295172229296e-05, "loss": 1.072, "step": 2468 }, { "epoch": 0.08, "learning_rate": 4.945239767670868e-05, "loss": 0.9667, "step": 2469 }, { "epoch": 0.08, "learning_rate": 4.9451843353806974e-05, "loss": 0.5891, "step": 2470 }, { "epoch": 0.08, "learning_rate": 4.9451288753594126e-05, "loss": 0.7415, "step": 2471 }, { "epoch": 0.08, "learning_rate": 4.945073387607643e-05, "loss": 0.8753, "step": 2472 }, { "epoch": 0.08, "learning_rate": 4.945017872126017e-05, "loss": 1.0001, "step": 2473 }, { "epoch": 0.08, "learning_rate": 4.944962328915165e-05, "loss": 0.9673, "step": 2474 }, { "epoch": 0.08, "learning_rate": 4.9449067579757166e-05, "loss": 0.9971, "step": 2475 }, { "epoch": 0.08, "learning_rate": 4.9448511593083026e-05, "loss": 0.9798, "step": 2476 }, { "epoch": 0.08, "learning_rate": 4.944795532913552e-05, "loss": 1.0882, "step": 2477 }, { "epoch": 0.08, "learning_rate": 4.9447398787920976e-05, "loss": 1.2251, "step": 2478 }, { "epoch": 0.08, "learning_rate": 4.9446841969445696e-05, "loss": 0.968, "step": 2479 }, { "epoch": 0.08, "learning_rate": 4.9446284873716e-05, "loss": 1.0017, "step": 2480 }, { "epoch": 0.08, "learning_rate": 4.944572750073819e-05, "loss": 0.5091, "step": 2481 }, { "epoch": 0.08, "learning_rate": 4.944516985051861e-05, "loss": 0.1212, "step": 2482 }, { "epoch": 0.08, "learning_rate": 4.9444611923063574e-05, "loss": 0.2503, "step": 2483 }, { "epoch": 0.08, "learning_rate": 4.9444053718379405e-05, "loss": 0.7162, "step": 2484 }, { "epoch": 0.08, "learning_rate": 4.944349523647244e-05, "loss": 1.0017, "step": 2485 }, { "epoch": 0.08, "learning_rate": 4.944293647734901e-05, "loss": 0.9482, "step": 2486 }, { "epoch": 0.08, "learning_rate": 4.944237744101545e-05, "loss": 1.0616, "step": 2487 }, { "epoch": 0.08, "learning_rate": 4.944181812747811e-05, "loss": 1.1537, "step": 2488 }, { "epoch": 0.08, "learning_rate": 4.944125853674332e-05, "loss": 1.2082, "step": 2489 }, { "epoch": 0.08, "learning_rate": 4.944069866881743e-05, "loss": 0.8584, "step": 2490 }, { "epoch": 0.08, "learning_rate": 4.94401385237068e-05, "loss": 0.8501, "step": 2491 }, { "epoch": 0.08, "learning_rate": 4.943957810141777e-05, "loss": 0.64, "step": 2492 }, { "epoch": 0.08, "learning_rate": 4.94390174019567e-05, "loss": 0.8143, "step": 2493 }, { "epoch": 0.08, "learning_rate": 4.9438456425329955e-05, "loss": 0.5683, "step": 2494 }, { "epoch": 0.08, "learning_rate": 4.9437895171543894e-05, "loss": 0.9611, "step": 2495 }, { "epoch": 0.08, "learning_rate": 4.9437333640604864e-05, "loss": 0.8048, "step": 2496 }, { "epoch": 0.08, "learning_rate": 4.943677183251926e-05, "loss": 2.9976, "step": 2497 }, { "epoch": 0.08, "learning_rate": 4.9436209747293446e-05, "loss": 1.1072, "step": 2498 }, { "epoch": 0.08, "learning_rate": 4.9435647384933784e-05, "loss": 0.8418, "step": 2499 }, { "epoch": 0.08, "learning_rate": 4.9435084745446666e-05, "loss": 1.2177, "step": 2500 }, { "epoch": 0.08, "learning_rate": 4.943452182883847e-05, "loss": 0.7314, "step": 2501 }, { "epoch": 0.08, "learning_rate": 4.9433958635115576e-05, "loss": 0.7264, "step": 2502 }, { "epoch": 0.08, "learning_rate": 4.943339516428438e-05, "loss": 1.0306, "step": 2503 }, { "epoch": 0.08, "learning_rate": 4.9432831416351256e-05, "loss": 1.044, "step": 2504 }, { "epoch": 0.08, "learning_rate": 4.943226739132261e-05, "loss": 1.0747, "step": 2505 }, { "epoch": 0.08, "learning_rate": 4.943170308920484e-05, "loss": 0.9535, "step": 2506 }, { "epoch": 0.08, "learning_rate": 4.943113851000434e-05, "loss": 1.0532, "step": 2507 }, { "epoch": 0.08, "learning_rate": 4.943057365372751e-05, "loss": 1.0522, "step": 2508 }, { "epoch": 0.08, "learning_rate": 4.9430008520380764e-05, "loss": 2.9953, "step": 2509 }, { "epoch": 0.08, "learning_rate": 4.9429443109970514e-05, "loss": 0.7206, "step": 2510 }, { "epoch": 0.08, "learning_rate": 4.942887742250316e-05, "loss": 0.8306, "step": 2511 }, { "epoch": 0.08, "learning_rate": 4.942831145798513e-05, "loss": 0.8365, "step": 2512 }, { "epoch": 0.08, "learning_rate": 4.942774521642283e-05, "loss": 0.9764, "step": 2513 }, { "epoch": 0.08, "learning_rate": 4.942717869782269e-05, "loss": 0.8734, "step": 2514 }, { "epoch": 0.08, "learning_rate": 4.942661190219114e-05, "loss": 0.8928, "step": 2515 }, { "epoch": 0.08, "learning_rate": 4.942604482953459e-05, "loss": 0.4512, "step": 2516 }, { "epoch": 0.08, "learning_rate": 4.942547747985949e-05, "loss": 0.6705, "step": 2517 }, { "epoch": 0.08, "learning_rate": 4.9424909853172276e-05, "loss": 0.8126, "step": 2518 }, { "epoch": 0.08, "learning_rate": 4.9424341949479366e-05, "loss": 3.2372, "step": 2519 }, { "epoch": 0.08, "learning_rate": 4.9423773768787214e-05, "loss": 5.875, "step": 2520 }, { "epoch": 0.08, "learning_rate": 4.9423205311102264e-05, "loss": 5.8365, "step": 2521 }, { "epoch": 0.08, "learning_rate": 4.942263657643095e-05, "loss": 5.8006, "step": 2522 }, { "epoch": 0.08, "learning_rate": 4.942206756477974e-05, "loss": 5.0128, "step": 2523 }, { "epoch": 0.08, "learning_rate": 4.942149827615508e-05, "loss": 0.765, "step": 2524 }, { "epoch": 0.08, "learning_rate": 4.942092871056343e-05, "loss": 2.8745, "step": 2525 }, { "epoch": 0.08, "learning_rate": 4.9420358868011226e-05, "loss": 0.941, "step": 2526 }, { "epoch": 0.08, "learning_rate": 4.941978874850497e-05, "loss": 0.9466, "step": 2527 }, { "epoch": 0.08, "learning_rate": 4.94192183520511e-05, "loss": 0.8305, "step": 2528 }, { "epoch": 0.08, "learning_rate": 4.9418647678656096e-05, "loss": 0.8124, "step": 2529 }, { "epoch": 0.08, "learning_rate": 4.9418076728326424e-05, "loss": 0.603, "step": 2530 }, { "epoch": 0.08, "learning_rate": 4.941750550106857e-05, "loss": 1.468, "step": 2531 }, { "epoch": 0.08, "learning_rate": 4.9416933996888994e-05, "loss": 1.0015, "step": 2532 }, { "epoch": 0.08, "learning_rate": 4.941636221579419e-05, "loss": 0.9277, "step": 2533 }, { "epoch": 0.08, "learning_rate": 4.9415790157790645e-05, "loss": 1.115, "step": 2534 }, { "epoch": 0.08, "learning_rate": 4.941521782288484e-05, "loss": 0.9729, "step": 2535 }, { "epoch": 0.08, "learning_rate": 4.941464521108326e-05, "loss": 0.8496, "step": 2536 }, { "epoch": 0.08, "learning_rate": 4.9414072322392424e-05, "loss": 4.0358, "step": 2537 }, { "epoch": 0.08, "learning_rate": 4.94134991568188e-05, "loss": 5.3047, "step": 2538 }, { "epoch": 0.08, "learning_rate": 4.941292571436891e-05, "loss": 1.1371, "step": 2539 }, { "epoch": 0.08, "learning_rate": 4.941235199504923e-05, "loss": 0.9076, "step": 2540 }, { "epoch": 0.08, "learning_rate": 4.9411777998866304e-05, "loss": 0.966, "step": 2541 }, { "epoch": 0.08, "learning_rate": 4.941120372582662e-05, "loss": 1.4019, "step": 2542 }, { "epoch": 0.08, "learning_rate": 4.941062917593668e-05, "loss": 1.6117, "step": 2543 }, { "epoch": 0.08, "learning_rate": 4.9410054349203025e-05, "loss": 1.1154, "step": 2544 }, { "epoch": 0.08, "learning_rate": 4.940947924563216e-05, "loss": 1.0107, "step": 2545 }, { "epoch": 0.08, "learning_rate": 4.940890386523062e-05, "loss": 0.9089, "step": 2546 }, { "epoch": 0.08, "learning_rate": 4.940832820800491e-05, "loss": 0.8067, "step": 2547 }, { "epoch": 0.08, "learning_rate": 4.9407752273961565e-05, "loss": 1.2449, "step": 2548 }, { "epoch": 0.08, "learning_rate": 4.940717606310713e-05, "loss": 1.1181, "step": 2549 }, { "epoch": 0.09, "learning_rate": 4.940659957544813e-05, "loss": 1.0549, "step": 2550 }, { "epoch": 0.09, "learning_rate": 4.9406022810991093e-05, "loss": 0.9429, "step": 2551 }, { "epoch": 0.09, "learning_rate": 4.940544576974258e-05, "loss": 0.9914, "step": 2552 }, { "epoch": 0.09, "learning_rate": 4.940486845170913e-05, "loss": 0.9632, "step": 2553 }, { "epoch": 0.09, "learning_rate": 4.940429085689728e-05, "loss": 0.6179, "step": 2554 }, { "epoch": 0.09, "learning_rate": 4.9403712985313585e-05, "loss": 1.0443, "step": 2555 }, { "epoch": 0.09, "learning_rate": 4.9403134836964606e-05, "loss": 0.6018, "step": 2556 }, { "epoch": 0.09, "learning_rate": 4.9402556411856896e-05, "loss": 0.7831, "step": 2557 }, { "epoch": 0.09, "learning_rate": 4.9401977709997005e-05, "loss": 0.9571, "step": 2558 }, { "epoch": 0.09, "learning_rate": 4.9401398731391516e-05, "loss": 1.0163, "step": 2559 }, { "epoch": 0.09, "learning_rate": 4.940081947604698e-05, "loss": 0.9185, "step": 2560 }, { "epoch": 0.09, "learning_rate": 4.940023994396997e-05, "loss": 1.0066, "step": 2561 }, { "epoch": 0.09, "learning_rate": 4.9399660135167056e-05, "loss": 0.9805, "step": 2562 }, { "epoch": 0.09, "learning_rate": 4.9399080049644816e-05, "loss": 1.5677, "step": 2563 }, { "epoch": 0.09, "learning_rate": 4.939849968740984e-05, "loss": 0.9511, "step": 2564 }, { "epoch": 0.09, "learning_rate": 4.939791904846869e-05, "loss": 0.9954, "step": 2565 }, { "epoch": 0.09, "learning_rate": 4.9397338132827966e-05, "loss": 1.4579, "step": 2566 }, { "epoch": 0.09, "learning_rate": 4.9396756940494245e-05, "loss": 1.5849, "step": 2567 }, { "epoch": 0.09, "learning_rate": 4.939617547147413e-05, "loss": 1.6628, "step": 2568 }, { "epoch": 0.09, "learning_rate": 4.9395593725774204e-05, "loss": 1.6614, "step": 2569 }, { "epoch": 0.09, "learning_rate": 4.939501170340107e-05, "loss": 1.7534, "step": 2570 }, { "epoch": 0.09, "learning_rate": 4.9394429404361334e-05, "loss": 1.422, "step": 2571 }, { "epoch": 0.09, "learning_rate": 4.939384682866159e-05, "loss": 1.0189, "step": 2572 }, { "epoch": 0.09, "learning_rate": 4.939326397630845e-05, "loss": 0.9464, "step": 2573 }, { "epoch": 0.09, "learning_rate": 4.939268084730853e-05, "loss": 0.9955, "step": 2574 }, { "epoch": 0.09, "learning_rate": 4.939209744166843e-05, "loss": 0.8205, "step": 2575 }, { "epoch": 0.09, "learning_rate": 4.9391513759394784e-05, "loss": 0.8476, "step": 2576 }, { "epoch": 0.09, "learning_rate": 4.9390929800494193e-05, "loss": 1.0763, "step": 2577 }, { "epoch": 0.09, "learning_rate": 4.93903455649733e-05, "loss": 1.0186, "step": 2578 }, { "epoch": 0.09, "learning_rate": 4.938976105283871e-05, "loss": 2.1895, "step": 2579 }, { "epoch": 0.09, "learning_rate": 4.938917626409706e-05, "loss": 2.0744, "step": 2580 }, { "epoch": 0.09, "learning_rate": 4.9388591198754986e-05, "loss": 1.5795, "step": 2581 }, { "epoch": 0.09, "learning_rate": 4.938800585681912e-05, "loss": 0.3727, "step": 2582 }, { "epoch": 0.09, "learning_rate": 4.93874202382961e-05, "loss": 0.8105, "step": 2583 }, { "epoch": 0.09, "learning_rate": 4.9386834343192576e-05, "loss": 1.0442, "step": 2584 }, { "epoch": 0.09, "learning_rate": 4.938624817151518e-05, "loss": 0.9214, "step": 2585 }, { "epoch": 0.09, "learning_rate": 4.938566172327057e-05, "loss": 0.8459, "step": 2586 }, { "epoch": 0.09, "learning_rate": 4.938507499846539e-05, "loss": 1.1163, "step": 2587 }, { "epoch": 0.09, "learning_rate": 4.9384487997106287e-05, "loss": 0.9379, "step": 2588 }, { "epoch": 0.09, "learning_rate": 4.938390071919994e-05, "loss": 0.9983, "step": 2589 }, { "epoch": 0.09, "learning_rate": 4.9383313164752985e-05, "loss": 0.9143, "step": 2590 }, { "epoch": 0.09, "learning_rate": 4.9382725333772105e-05, "loss": 0.8254, "step": 2591 }, { "epoch": 0.09, "learning_rate": 4.938213722626396e-05, "loss": 1.1141, "step": 2592 }, { "epoch": 0.09, "learning_rate": 4.938154884223522e-05, "loss": 1.1485, "step": 2593 }, { "epoch": 0.09, "learning_rate": 4.9380960181692556e-05, "loss": 0.607, "step": 2594 }, { "epoch": 0.09, "learning_rate": 4.9380371244642644e-05, "loss": 1.1212, "step": 2595 }, { "epoch": 0.09, "learning_rate": 4.937978203109216e-05, "loss": 1.6216, "step": 2596 }, { "epoch": 0.09, "learning_rate": 4.93791925410478e-05, "loss": 1.7492, "step": 2597 }, { "epoch": 0.09, "learning_rate": 4.9378602774516235e-05, "loss": 1.7687, "step": 2598 }, { "epoch": 0.09, "learning_rate": 4.937801273150415e-05, "loss": 1.7367, "step": 2599 }, { "epoch": 0.09, "learning_rate": 4.937742241201826e-05, "loss": 0.4467, "step": 2600 }, { "epoch": 0.09, "learning_rate": 4.9376831816065236e-05, "loss": 1.084, "step": 2601 }, { "epoch": 0.09, "learning_rate": 4.937624094365179e-05, "loss": 0.8288, "step": 2602 }, { "epoch": 0.09, "learning_rate": 4.937564979478462e-05, "loss": 0.8307, "step": 2603 }, { "epoch": 0.09, "learning_rate": 4.937505836947043e-05, "loss": 0.7973, "step": 2604 }, { "epoch": 0.09, "learning_rate": 4.937446666771591e-05, "loss": 0.6778, "step": 2605 }, { "epoch": 0.09, "learning_rate": 4.93738746895278e-05, "loss": 0.7086, "step": 2606 }, { "epoch": 0.09, "learning_rate": 4.93732824349128e-05, "loss": 1.3947, "step": 2607 }, { "epoch": 0.09, "learning_rate": 4.937268990387763e-05, "loss": 2.2156, "step": 2608 }, { "epoch": 0.09, "learning_rate": 4.9372097096429e-05, "loss": 2.5021, "step": 2609 }, { "epoch": 0.09, "learning_rate": 4.9371504012573646e-05, "loss": 0.752, "step": 2610 }, { "epoch": 0.09, "learning_rate": 4.9370910652318287e-05, "loss": 0.5713, "step": 2611 }, { "epoch": 0.09, "learning_rate": 4.937031701566965e-05, "loss": 0.7653, "step": 2612 }, { "epoch": 0.09, "learning_rate": 4.936972310263448e-05, "loss": 0.403, "step": 2613 }, { "epoch": 0.09, "learning_rate": 4.9369128913219497e-05, "loss": 0.2673, "step": 2614 }, { "epoch": 0.09, "learning_rate": 4.9368534447431446e-05, "loss": 0.3628, "step": 2615 }, { "epoch": 0.09, "learning_rate": 4.936793970527707e-05, "loss": 0.2689, "step": 2616 }, { "epoch": 0.09, "learning_rate": 4.936734468676312e-05, "loss": 1.0002, "step": 2617 }, { "epoch": 0.09, "learning_rate": 4.936674939189633e-05, "loss": 1.0221, "step": 2618 }, { "epoch": 0.09, "learning_rate": 4.9366153820683467e-05, "loss": 0.9809, "step": 2619 }, { "epoch": 0.09, "learning_rate": 4.9365557973131274e-05, "loss": 0.9198, "step": 2620 }, { "epoch": 0.09, "learning_rate": 4.936496184924651e-05, "loss": 1.1874, "step": 2621 }, { "epoch": 0.09, "learning_rate": 4.936436544903594e-05, "loss": 1.021, "step": 2622 }, { "epoch": 0.09, "learning_rate": 4.936376877250632e-05, "loss": 0.8992, "step": 2623 }, { "epoch": 0.09, "learning_rate": 4.9363171819664434e-05, "loss": 0.5899, "step": 2624 }, { "epoch": 0.09, "learning_rate": 4.936257459051703e-05, "loss": 0.7087, "step": 2625 }, { "epoch": 0.09, "learning_rate": 4.93619770850709e-05, "loss": 0.9577, "step": 2626 }, { "epoch": 0.09, "learning_rate": 4.936137930333281e-05, "loss": 0.9324, "step": 2627 }, { "epoch": 0.09, "learning_rate": 4.936078124530954e-05, "loss": 0.8132, "step": 2628 }, { "epoch": 0.09, "learning_rate": 4.936018291100788e-05, "loss": 1.0605, "step": 2629 }, { "epoch": 0.09, "learning_rate": 4.93595843004346e-05, "loss": 0.4339, "step": 2630 }, { "epoch": 0.09, "learning_rate": 4.935898541359651e-05, "loss": 0.7836, "step": 2631 }, { "epoch": 0.09, "learning_rate": 4.9358386250500385e-05, "loss": 0.8984, "step": 2632 }, { "epoch": 0.09, "learning_rate": 4.935778681115303e-05, "loss": 0.9739, "step": 2633 }, { "epoch": 0.09, "learning_rate": 4.935718709556124e-05, "loss": 0.8931, "step": 2634 }, { "epoch": 0.09, "learning_rate": 4.935658710373181e-05, "loss": 0.9133, "step": 2635 }, { "epoch": 0.09, "learning_rate": 4.935598683567156e-05, "loss": 0.6448, "step": 2636 }, { "epoch": 0.09, "learning_rate": 4.935538629138728e-05, "loss": 0.7879, "step": 2637 }, { "epoch": 0.09, "learning_rate": 4.93547854708858e-05, "loss": 0.841, "step": 2638 }, { "epoch": 0.09, "learning_rate": 4.935418437417391e-05, "loss": 1.0864, "step": 2639 }, { "epoch": 0.09, "learning_rate": 4.9353583001258455e-05, "loss": 0.9608, "step": 2640 }, { "epoch": 0.09, "learning_rate": 4.935298135214623e-05, "loss": 1.1416, "step": 2641 }, { "epoch": 0.09, "learning_rate": 4.9352379426844074e-05, "loss": 0.9194, "step": 2642 }, { "epoch": 0.09, "learning_rate": 4.935177722535881e-05, "loss": 1.1134, "step": 2643 }, { "epoch": 0.09, "learning_rate": 4.935117474769726e-05, "loss": 0.9413, "step": 2644 }, { "epoch": 0.09, "learning_rate": 4.935057199386627e-05, "loss": 0.8957, "step": 2645 }, { "epoch": 0.09, "learning_rate": 4.9349968963872664e-05, "loss": 0.9631, "step": 2646 }, { "epoch": 0.09, "learning_rate": 4.934936565772328e-05, "loss": 0.8814, "step": 2647 }, { "epoch": 0.09, "learning_rate": 4.934876207542498e-05, "loss": 1.0734, "step": 2648 }, { "epoch": 0.09, "learning_rate": 4.934815821698459e-05, "loss": 0.8548, "step": 2649 }, { "epoch": 0.09, "learning_rate": 4.9347554082408956e-05, "loss": 0.9378, "step": 2650 }, { "epoch": 0.09, "learning_rate": 4.934694967170495e-05, "loss": 0.9778, "step": 2651 }, { "epoch": 0.09, "learning_rate": 4.93463449848794e-05, "loss": 1.0049, "step": 2652 }, { "epoch": 0.09, "learning_rate": 4.934574002193919e-05, "loss": 0.7698, "step": 2653 }, { "epoch": 0.09, "learning_rate": 4.934513478289117e-05, "loss": 0.9925, "step": 2654 }, { "epoch": 0.09, "learning_rate": 4.934452926774219e-05, "loss": 0.9429, "step": 2655 }, { "epoch": 0.09, "learning_rate": 4.934392347649914e-05, "loss": 0.8157, "step": 2656 }, { "epoch": 0.09, "learning_rate": 4.934331740916888e-05, "loss": 0.7718, "step": 2657 }, { "epoch": 0.09, "learning_rate": 4.9342711065758286e-05, "loss": 3.981, "step": 2658 }, { "epoch": 0.09, "learning_rate": 4.9342104446274224e-05, "loss": 0.9364, "step": 2659 }, { "epoch": 0.09, "learning_rate": 4.934149755072359e-05, "loss": 0.8958, "step": 2660 }, { "epoch": 0.09, "learning_rate": 4.934089037911326e-05, "loss": 1.0266, "step": 2661 }, { "epoch": 0.09, "learning_rate": 4.934028293145011e-05, "loss": 0.8757, "step": 2662 }, { "epoch": 0.09, "learning_rate": 4.933967520774104e-05, "loss": 0.8909, "step": 2663 }, { "epoch": 0.09, "learning_rate": 4.933906720799295e-05, "loss": 0.8537, "step": 2664 }, { "epoch": 0.09, "learning_rate": 4.9338458932212715e-05, "loss": 0.1429, "step": 2665 }, { "epoch": 0.09, "learning_rate": 4.933785038040726e-05, "loss": 0.137, "step": 2666 }, { "epoch": 0.09, "learning_rate": 4.9337241552583456e-05, "loss": 0.8423, "step": 2667 }, { "epoch": 0.09, "learning_rate": 4.9336632448748235e-05, "loss": 0.4207, "step": 2668 }, { "epoch": 0.09, "learning_rate": 4.933602306890848e-05, "loss": 0.7686, "step": 2669 }, { "epoch": 0.09, "learning_rate": 4.933541341307112e-05, "loss": 0.9063, "step": 2670 }, { "epoch": 0.09, "learning_rate": 4.933480348124307e-05, "loss": 1.1161, "step": 2671 }, { "epoch": 0.09, "learning_rate": 4.9334193273431235e-05, "loss": 1.077, "step": 2672 }, { "epoch": 0.09, "learning_rate": 4.9333582789642544e-05, "loss": 2.9253, "step": 2673 }, { "epoch": 0.09, "learning_rate": 4.933297202988392e-05, "loss": 0.9583, "step": 2674 }, { "epoch": 0.09, "learning_rate": 4.9332360994162285e-05, "loss": 1.2572, "step": 2675 }, { "epoch": 0.09, "learning_rate": 4.9331749682484574e-05, "loss": 1.0272, "step": 2676 }, { "epoch": 0.09, "learning_rate": 4.933113809485771e-05, "loss": 0.7707, "step": 2677 }, { "epoch": 0.09, "learning_rate": 4.933052623128865e-05, "loss": 1.234, "step": 2678 }, { "epoch": 0.09, "learning_rate": 4.9329914091784316e-05, "loss": 1.7824, "step": 2679 }, { "epoch": 0.09, "learning_rate": 4.9329301676351646e-05, "loss": 1.6173, "step": 2680 }, { "epoch": 0.09, "learning_rate": 4.932868898499761e-05, "loss": 1.4374, "step": 2681 }, { "epoch": 0.09, "learning_rate": 4.932807601772913e-05, "loss": 0.9614, "step": 2682 }, { "epoch": 0.09, "learning_rate": 4.9327462774553166e-05, "loss": 0.9807, "step": 2683 }, { "epoch": 0.09, "learning_rate": 4.932684925547668e-05, "loss": 0.9216, "step": 2684 }, { "epoch": 0.09, "learning_rate": 4.932623546050662e-05, "loss": 0.9987, "step": 2685 }, { "epoch": 0.09, "learning_rate": 4.932562138964995e-05, "loss": 3.761, "step": 2686 }, { "epoch": 0.09, "learning_rate": 4.932500704291365e-05, "loss": 0.6567, "step": 2687 }, { "epoch": 0.09, "learning_rate": 4.932439242030467e-05, "loss": 0.7983, "step": 2688 }, { "epoch": 0.09, "learning_rate": 4.932377752182998e-05, "loss": 1.0064, "step": 2689 }, { "epoch": 0.09, "learning_rate": 4.9323162347496566e-05, "loss": 0.9857, "step": 2690 }, { "epoch": 0.09, "learning_rate": 4.9322546897311386e-05, "loss": 0.9207, "step": 2691 }, { "epoch": 0.09, "learning_rate": 4.932193117128143e-05, "loss": 0.9494, "step": 2692 }, { "epoch": 0.09, "learning_rate": 4.93213151694137e-05, "loss": 1.0021, "step": 2693 }, { "epoch": 0.09, "learning_rate": 4.9320698891715146e-05, "loss": 0.9253, "step": 2694 }, { "epoch": 0.09, "learning_rate": 4.932008233819278e-05, "loss": 0.8239, "step": 2695 }, { "epoch": 0.09, "learning_rate": 4.9319465508853585e-05, "loss": 0.9772, "step": 2696 }, { "epoch": 0.09, "learning_rate": 4.931884840370457e-05, "loss": 1.0543, "step": 2697 }, { "epoch": 0.09, "learning_rate": 4.931823102275272e-05, "loss": 0.9357, "step": 2698 }, { "epoch": 0.09, "learning_rate": 4.9317613366005036e-05, "loss": 0.9221, "step": 2699 }, { "epoch": 0.09, "learning_rate": 4.931699543346854e-05, "loss": 1.0161, "step": 2700 }, { "epoch": 0.09, "learning_rate": 4.9316377225150226e-05, "loss": 0.9522, "step": 2701 }, { "epoch": 0.09, "learning_rate": 4.93157587410571e-05, "loss": 1.4277, "step": 2702 }, { "epoch": 0.09, "learning_rate": 4.931513998119619e-05, "loss": 0.985, "step": 2703 }, { "epoch": 0.09, "learning_rate": 4.93145209455745e-05, "loss": 0.3431, "step": 2704 }, { "epoch": 0.09, "learning_rate": 4.9313901634199065e-05, "loss": 0.9333, "step": 2705 }, { "epoch": 0.09, "learning_rate": 4.9313282047076904e-05, "loss": 0.9946, "step": 2706 }, { "epoch": 0.09, "learning_rate": 4.931266218421504e-05, "loss": 0.9522, "step": 2707 }, { "epoch": 0.09, "learning_rate": 4.9312042045620496e-05, "loss": 1.0258, "step": 2708 }, { "epoch": 0.09, "learning_rate": 4.9311421631300324e-05, "loss": 0.9268, "step": 2709 }, { "epoch": 0.09, "learning_rate": 4.931080094126155e-05, "loss": 1.0346, "step": 2710 }, { "epoch": 0.09, "learning_rate": 4.9310179975511206e-05, "loss": 0.8268, "step": 2711 }, { "epoch": 0.09, "learning_rate": 4.930955873405634e-05, "loss": 1.1073, "step": 2712 }, { "epoch": 0.09, "learning_rate": 4.9308937216904e-05, "loss": 1.0383, "step": 2713 }, { "epoch": 0.09, "learning_rate": 4.930831542406124e-05, "loss": 1.0159, "step": 2714 }, { "epoch": 0.09, "learning_rate": 4.93076933555351e-05, "loss": 0.9683, "step": 2715 }, { "epoch": 0.09, "learning_rate": 4.9307071011332645e-05, "loss": 0.9283, "step": 2716 }, { "epoch": 0.09, "learning_rate": 4.930644839146092e-05, "loss": 0.9075, "step": 2717 }, { "epoch": 0.09, "learning_rate": 4.9305825495927e-05, "loss": 0.8581, "step": 2718 }, { "epoch": 0.09, "learning_rate": 4.9305202324737953e-05, "loss": 1.0572, "step": 2719 }, { "epoch": 0.09, "learning_rate": 4.930457887790083e-05, "loss": 0.9071, "step": 2720 }, { "epoch": 0.09, "learning_rate": 4.930395515542271e-05, "loss": 0.944, "step": 2721 }, { "epoch": 0.09, "learning_rate": 4.930333115731066e-05, "loss": 0.963, "step": 2722 }, { "epoch": 0.09, "learning_rate": 4.930270688357177e-05, "loss": 1.2442, "step": 2723 }, { "epoch": 0.09, "learning_rate": 4.9302082334213107e-05, "loss": 1.0454, "step": 2724 }, { "epoch": 0.09, "learning_rate": 4.930145750924177e-05, "loss": 0.9947, "step": 2725 }, { "epoch": 0.09, "learning_rate": 4.930083240866482e-05, "loss": 0.9753, "step": 2726 }, { "epoch": 0.09, "learning_rate": 4.9300207032489363e-05, "loss": 0.7404, "step": 2727 }, { "epoch": 0.09, "learning_rate": 4.9299581380722505e-05, "loss": 1.2239, "step": 2728 }, { "epoch": 0.09, "learning_rate": 4.929895545337131e-05, "loss": 3.8417, "step": 2729 }, { "epoch": 0.09, "learning_rate": 4.92983292504429e-05, "loss": 0.9457, "step": 2730 }, { "epoch": 0.09, "learning_rate": 4.929770277194437e-05, "loss": 0.7944, "step": 2731 }, { "epoch": 0.09, "learning_rate": 4.9297076017882824e-05, "loss": 0.9517, "step": 2732 }, { "epoch": 0.09, "learning_rate": 4.9296448988265373e-05, "loss": 0.387, "step": 2733 }, { "epoch": 0.09, "learning_rate": 4.929582168309912e-05, "loss": 2.316, "step": 2734 }, { "epoch": 0.09, "learning_rate": 4.929519410239119e-05, "loss": 1.7384, "step": 2735 }, { "epoch": 0.09, "learning_rate": 4.929456624614869e-05, "loss": 1.015, "step": 2736 }, { "epoch": 0.09, "learning_rate": 4.929393811437876e-05, "loss": 0.9172, "step": 2737 }, { "epoch": 0.09, "learning_rate": 4.92933097070885e-05, "loss": 1.0776, "step": 2738 }, { "epoch": 0.09, "learning_rate": 4.929268102428505e-05, "loss": 1.4835, "step": 2739 }, { "epoch": 0.09, "learning_rate": 4.9292052065975536e-05, "loss": 1.2658, "step": 2740 }, { "epoch": 0.09, "learning_rate": 4.9291422832167086e-05, "loss": 0.8769, "step": 2741 }, { "epoch": 0.09, "learning_rate": 4.929079332286685e-05, "loss": 1.1273, "step": 2742 }, { "epoch": 0.09, "learning_rate": 4.929016353808196e-05, "loss": 0.9085, "step": 2743 }, { "epoch": 0.09, "learning_rate": 4.928953347781955e-05, "loss": 0.6934, "step": 2744 }, { "epoch": 0.09, "learning_rate": 4.9288903142086785e-05, "loss": 0.8932, "step": 2745 }, { "epoch": 0.09, "learning_rate": 4.92882725308908e-05, "loss": 1.0571, "step": 2746 }, { "epoch": 0.09, "learning_rate": 4.928764164423874e-05, "loss": 0.9084, "step": 2747 }, { "epoch": 0.09, "learning_rate": 4.928701048213777e-05, "loss": 1.0115, "step": 2748 }, { "epoch": 0.09, "learning_rate": 4.928637904459506e-05, "loss": 0.8411, "step": 2749 }, { "epoch": 0.09, "learning_rate": 4.9285747331617746e-05, "loss": 0.7865, "step": 2750 }, { "epoch": 0.09, "learning_rate": 4.9285115343213015e-05, "loss": 2.6638, "step": 2751 }, { "epoch": 0.09, "learning_rate": 4.928448307938802e-05, "loss": 1.5303, "step": 2752 }, { "epoch": 0.09, "learning_rate": 4.9283850540149936e-05, "loss": 0.9874, "step": 2753 }, { "epoch": 0.09, "learning_rate": 4.9283217725505935e-05, "loss": 0.6683, "step": 2754 }, { "epoch": 0.09, "learning_rate": 4.928258463546319e-05, "loss": 1.0297, "step": 2755 }, { "epoch": 0.09, "learning_rate": 4.92819512700289e-05, "loss": 0.8824, "step": 2756 }, { "epoch": 0.09, "learning_rate": 4.9281317629210224e-05, "loss": 1.4003, "step": 2757 }, { "epoch": 0.09, "learning_rate": 4.928068371301436e-05, "loss": 1.4942, "step": 2758 }, { "epoch": 0.09, "learning_rate": 4.92800495214485e-05, "loss": 1.3702, "step": 2759 }, { "epoch": 0.09, "learning_rate": 4.927941505451983e-05, "loss": 1.0437, "step": 2760 }, { "epoch": 0.09, "learning_rate": 4.927878031223555e-05, "loss": 1.0859, "step": 2761 }, { "epoch": 0.09, "learning_rate": 4.927814529460285e-05, "loss": 0.8115, "step": 2762 }, { "epoch": 0.09, "learning_rate": 4.9277510001628944e-05, "loss": 0.9104, "step": 2763 }, { "epoch": 0.09, "learning_rate": 4.927687443332103e-05, "loss": 1.1831, "step": 2764 }, { "epoch": 0.09, "learning_rate": 4.927623858968631e-05, "loss": 0.9452, "step": 2765 }, { "epoch": 0.09, "learning_rate": 4.927560247073201e-05, "loss": 0.8426, "step": 2766 }, { "epoch": 0.09, "learning_rate": 4.9274966076465336e-05, "loss": 0.7588, "step": 2767 }, { "epoch": 0.09, "learning_rate": 4.92743294068935e-05, "loss": 1.0291, "step": 2768 }, { "epoch": 0.09, "learning_rate": 4.9273692462023744e-05, "loss": 1.0236, "step": 2769 }, { "epoch": 0.09, "learning_rate": 4.927305524186326e-05, "loss": 1.1457, "step": 2770 }, { "epoch": 0.09, "learning_rate": 4.927241774641931e-05, "loss": 0.8993, "step": 2771 }, { "epoch": 0.09, "learning_rate": 4.927177997569909e-05, "loss": 0.6802, "step": 2772 }, { "epoch": 0.09, "learning_rate": 4.927114192970985e-05, "loss": 0.5258, "step": 2773 }, { "epoch": 0.09, "learning_rate": 4.927050360845883e-05, "loss": 0.9401, "step": 2774 }, { "epoch": 0.09, "learning_rate": 4.926986501195326e-05, "loss": 0.7244, "step": 2775 }, { "epoch": 0.09, "learning_rate": 4.926922614020039e-05, "loss": 0.6992, "step": 2776 }, { "epoch": 0.09, "learning_rate": 4.9268586993207455e-05, "loss": 1.0735, "step": 2777 }, { "epoch": 0.09, "learning_rate": 4.926794757098171e-05, "loss": 0.9043, "step": 2778 }, { "epoch": 0.09, "learning_rate": 4.926730787353041e-05, "loss": 0.8299, "step": 2779 }, { "epoch": 0.09, "learning_rate": 4.926666790086081e-05, "loss": 0.9879, "step": 2780 }, { "epoch": 0.09, "learning_rate": 4.926602765298016e-05, "loss": 1.1569, "step": 2781 }, { "epoch": 0.09, "learning_rate": 4.926538712989572e-05, "loss": 0.8694, "step": 2782 }, { "epoch": 0.09, "learning_rate": 4.9264746331614776e-05, "loss": 1.2527, "step": 2783 }, { "epoch": 0.09, "learning_rate": 4.926410525814457e-05, "loss": 1.0149, "step": 2784 }, { "epoch": 0.09, "learning_rate": 4.926346390949238e-05, "loss": 0.947, "step": 2785 }, { "epoch": 0.09, "learning_rate": 4.9262822285665486e-05, "loss": 0.8974, "step": 2786 }, { "epoch": 0.09, "learning_rate": 4.926218038667115e-05, "loss": 1.0949, "step": 2787 }, { "epoch": 0.09, "learning_rate": 4.926153821251668e-05, "loss": 1.1872, "step": 2788 }, { "epoch": 0.09, "learning_rate": 4.926089576320933e-05, "loss": 1.094, "step": 2789 }, { "epoch": 0.09, "learning_rate": 4.92602530387564e-05, "loss": 1.0931, "step": 2790 }, { "epoch": 0.09, "learning_rate": 4.925961003916517e-05, "loss": 0.952, "step": 2791 }, { "epoch": 0.09, "learning_rate": 4.925896676444295e-05, "loss": 0.8579, "step": 2792 }, { "epoch": 0.09, "learning_rate": 4.9258323214597015e-05, "loss": 1.0744, "step": 2793 }, { "epoch": 0.09, "learning_rate": 4.925767938963468e-05, "loss": 0.9895, "step": 2794 }, { "epoch": 0.09, "learning_rate": 4.9257035289563236e-05, "loss": 0.8815, "step": 2795 }, { "epoch": 0.09, "learning_rate": 4.9256390914389994e-05, "loss": 0.7036, "step": 2796 }, { "epoch": 0.09, "learning_rate": 4.925574626412226e-05, "loss": 1.32, "step": 2797 }, { "epoch": 0.09, "learning_rate": 4.925510133876734e-05, "loss": 0.8565, "step": 2798 }, { "epoch": 0.09, "learning_rate": 4.925445613833255e-05, "loss": 0.8055, "step": 2799 }, { "epoch": 0.09, "learning_rate": 4.9253810662825216e-05, "loss": 0.4658, "step": 2800 }, { "epoch": 0.09, "learning_rate": 4.925316491225265e-05, "loss": 0.8552, "step": 2801 }, { "epoch": 0.09, "learning_rate": 4.925251888662218e-05, "loss": 0.8971, "step": 2802 }, { "epoch": 0.09, "learning_rate": 4.9251872585941125e-05, "loss": 0.6418, "step": 2803 }, { "epoch": 0.09, "learning_rate": 4.925122601021682e-05, "loss": 1.3775, "step": 2804 }, { "epoch": 0.09, "learning_rate": 4.9250579159456603e-05, "loss": 1.3229, "step": 2805 }, { "epoch": 0.09, "learning_rate": 4.92499320336678e-05, "loss": 1.2406, "step": 2806 }, { "epoch": 0.09, "learning_rate": 4.924928463285776e-05, "loss": 1.1109, "step": 2807 }, { "epoch": 0.09, "learning_rate": 4.924863695703382e-05, "loss": 0.9184, "step": 2808 }, { "epoch": 0.09, "learning_rate": 4.924798900620332e-05, "loss": 1.0575, "step": 2809 }, { "epoch": 0.09, "learning_rate": 4.9247340780373624e-05, "loss": 0.8842, "step": 2810 }, { "epoch": 0.09, "learning_rate": 4.9246692279552066e-05, "loss": 0.6088, "step": 2811 }, { "epoch": 0.09, "learning_rate": 4.924604350374601e-05, "loss": 0.1506, "step": 2812 }, { "epoch": 0.09, "learning_rate": 4.924539445296282e-05, "loss": 0.6162, "step": 2813 }, { "epoch": 0.09, "learning_rate": 4.9244745127209843e-05, "loss": 1.024, "step": 2814 }, { "epoch": 0.09, "learning_rate": 4.924409552649445e-05, "loss": 0.9777, "step": 2815 }, { "epoch": 0.09, "learning_rate": 4.9243445650824006e-05, "loss": 0.9069, "step": 2816 }, { "epoch": 0.09, "learning_rate": 4.924279550020588e-05, "loss": 0.9309, "step": 2817 }, { "epoch": 0.09, "learning_rate": 4.924214507464745e-05, "loss": 1.0514, "step": 2818 }, { "epoch": 0.09, "learning_rate": 4.92414943741561e-05, "loss": 1.0149, "step": 2819 }, { "epoch": 0.09, "learning_rate": 4.924084339873919e-05, "loss": 0.6206, "step": 2820 }, { "epoch": 0.09, "learning_rate": 4.924019214840412e-05, "loss": 1.0927, "step": 2821 }, { "epoch": 0.09, "learning_rate": 4.923954062315827e-05, "loss": 0.9969, "step": 2822 }, { "epoch": 0.09, "learning_rate": 4.923888882300902e-05, "loss": 0.7359, "step": 2823 }, { "epoch": 0.09, "learning_rate": 4.923823674796378e-05, "loss": 0.9941, "step": 2824 }, { "epoch": 0.09, "learning_rate": 4.923758439802993e-05, "loss": 0.7313, "step": 2825 }, { "epoch": 0.09, "learning_rate": 4.923693177321488e-05, "loss": 0.7045, "step": 2826 }, { "epoch": 0.09, "learning_rate": 4.923627887352602e-05, "loss": 0.9152, "step": 2827 }, { "epoch": 0.09, "learning_rate": 4.9235625698970756e-05, "loss": 1.0098, "step": 2828 }, { "epoch": 0.09, "learning_rate": 4.923497224955651e-05, "loss": 0.4909, "step": 2829 }, { "epoch": 0.09, "learning_rate": 4.923431852529068e-05, "loss": 0.4423, "step": 2830 }, { "epoch": 0.09, "learning_rate": 4.923366452618068e-05, "loss": 0.3981, "step": 2831 }, { "epoch": 0.09, "learning_rate": 4.9233010252233925e-05, "loss": 0.6304, "step": 2832 }, { "epoch": 0.09, "learning_rate": 4.923235570345785e-05, "loss": 0.538, "step": 2833 }, { "epoch": 0.09, "learning_rate": 4.923170087985986e-05, "loss": 1.02, "step": 2834 }, { "epoch": 0.09, "learning_rate": 4.9231045781447396e-05, "loss": 0.955, "step": 2835 }, { "epoch": 0.09, "learning_rate": 4.9230390408227876e-05, "loss": 0.8046, "step": 2836 }, { "epoch": 0.09, "learning_rate": 4.922973476020874e-05, "loss": 2.421, "step": 2837 }, { "epoch": 0.09, "learning_rate": 4.922907883739743e-05, "loss": 0.9007, "step": 2838 }, { "epoch": 0.09, "learning_rate": 4.922842263980137e-05, "loss": 0.9158, "step": 2839 }, { "epoch": 0.09, "learning_rate": 4.9227766167428005e-05, "loss": 0.6816, "step": 2840 }, { "epoch": 0.09, "learning_rate": 4.922710942028479e-05, "loss": 1.0316, "step": 2841 }, { "epoch": 0.09, "learning_rate": 4.922645239837916e-05, "loss": 0.8904, "step": 2842 }, { "epoch": 0.09, "learning_rate": 4.922579510171858e-05, "loss": 0.6514, "step": 2843 }, { "epoch": 0.09, "learning_rate": 4.9225137530310495e-05, "loss": 0.7509, "step": 2844 }, { "epoch": 0.09, "learning_rate": 4.922447968416237e-05, "loss": 0.594, "step": 2845 }, { "epoch": 0.09, "learning_rate": 4.922382156328165e-05, "loss": 0.9327, "step": 2846 }, { "epoch": 0.09, "learning_rate": 4.922316316767582e-05, "loss": 0.9428, "step": 2847 }, { "epoch": 0.09, "learning_rate": 4.922250449735233e-05, "loss": 0.9938, "step": 2848 }, { "epoch": 0.09, "learning_rate": 4.922184555231867e-05, "loss": 0.9336, "step": 2849 }, { "epoch": 0.1, "learning_rate": 4.922118633258229e-05, "loss": 0.9147, "step": 2850 }, { "epoch": 0.1, "learning_rate": 4.9220526838150674e-05, "loss": 0.926, "step": 2851 }, { "epoch": 0.1, "learning_rate": 4.9219867069031304e-05, "loss": 0.6321, "step": 2852 }, { "epoch": 0.1, "learning_rate": 4.9219207025231665e-05, "loss": 0.8453, "step": 2853 }, { "epoch": 0.1, "learning_rate": 4.9218546706759245e-05, "loss": 1.0118, "step": 2854 }, { "epoch": 0.1, "learning_rate": 4.9217886113621525e-05, "loss": 0.9823, "step": 2855 }, { "epoch": 0.1, "learning_rate": 4.921722524582599e-05, "loss": 1.1283, "step": 2856 }, { "epoch": 0.1, "learning_rate": 4.921656410338016e-05, "loss": 0.8909, "step": 2857 }, { "epoch": 0.1, "learning_rate": 4.921590268629151e-05, "loss": 0.8999, "step": 2858 }, { "epoch": 0.1, "learning_rate": 4.9215240994567546e-05, "loss": 0.7668, "step": 2859 }, { "epoch": 0.1, "learning_rate": 4.9214579028215776e-05, "loss": 0.6374, "step": 2860 }, { "epoch": 0.1, "learning_rate": 4.9213916787243714e-05, "loss": 0.8829, "step": 2861 }, { "epoch": 0.1, "learning_rate": 4.921325427165886e-05, "loss": 1.1832, "step": 2862 }, { "epoch": 0.1, "learning_rate": 4.9212591481468725e-05, "loss": 1.2229, "step": 2863 }, { "epoch": 0.1, "learning_rate": 4.921192841668084e-05, "loss": 1.372, "step": 2864 }, { "epoch": 0.1, "learning_rate": 4.921126507730271e-05, "loss": 1.0113, "step": 2865 }, { "epoch": 0.1, "learning_rate": 4.9210601463341864e-05, "loss": 0.7927, "step": 2866 }, { "epoch": 0.1, "learning_rate": 4.9209937574805834e-05, "loss": 0.7784, "step": 2867 }, { "epoch": 0.1, "learning_rate": 4.920927341170214e-05, "loss": 1.1544, "step": 2868 }, { "epoch": 0.1, "learning_rate": 4.920860897403832e-05, "loss": 0.9399, "step": 2869 }, { "epoch": 0.1, "learning_rate": 4.920794426182191e-05, "loss": 0.9297, "step": 2870 }, { "epoch": 0.1, "learning_rate": 4.920727927506044e-05, "loss": 1.0331, "step": 2871 }, { "epoch": 0.1, "learning_rate": 4.920661401376146e-05, "loss": 1.0363, "step": 2872 }, { "epoch": 0.1, "learning_rate": 4.920594847793252e-05, "loss": 0.8228, "step": 2873 }, { "epoch": 0.1, "learning_rate": 4.9205282667581156e-05, "loss": 0.7683, "step": 2874 }, { "epoch": 0.1, "learning_rate": 4.920461658271492e-05, "loss": 0.6805, "step": 2875 }, { "epoch": 0.1, "learning_rate": 4.9203950223341373e-05, "loss": 1.0565, "step": 2876 }, { "epoch": 0.1, "learning_rate": 4.920328358946806e-05, "loss": 0.8444, "step": 2877 }, { "epoch": 0.1, "learning_rate": 4.920261668110257e-05, "loss": 1.1722, "step": 2878 }, { "epoch": 0.1, "learning_rate": 4.920194949825243e-05, "loss": 4.8996, "step": 2879 }, { "epoch": 0.1, "learning_rate": 4.9201282040925234e-05, "loss": 5.5252, "step": 2880 }, { "epoch": 0.1, "learning_rate": 4.920061430912853e-05, "loss": 0.938, "step": 2881 }, { "epoch": 0.1, "learning_rate": 4.9199946302869914e-05, "loss": 0.6108, "step": 2882 }, { "epoch": 0.1, "learning_rate": 4.919927802215695e-05, "loss": 0.1166, "step": 2883 }, { "epoch": 0.1, "learning_rate": 4.91986094669972e-05, "loss": 0.1218, "step": 2884 }, { "epoch": 0.1, "learning_rate": 4.9197940637398284e-05, "loss": 0.8457, "step": 2885 }, { "epoch": 0.1, "learning_rate": 4.9197271533367754e-05, "loss": 0.7483, "step": 2886 }, { "epoch": 0.1, "learning_rate": 4.919660215491322e-05, "loss": 0.8558, "step": 2887 }, { "epoch": 0.1, "learning_rate": 4.919593250204225e-05, "loss": 2.978, "step": 2888 }, { "epoch": 0.1, "learning_rate": 4.919526257476247e-05, "loss": 1.3606, "step": 2889 }, { "epoch": 0.1, "learning_rate": 4.919459237308145e-05, "loss": 0.8484, "step": 2890 }, { "epoch": 0.1, "learning_rate": 4.919392189700681e-05, "loss": 0.8162, "step": 2891 }, { "epoch": 0.1, "learning_rate": 4.919325114654614e-05, "loss": 0.9619, "step": 2892 }, { "epoch": 0.1, "learning_rate": 4.9192580121707046e-05, "loss": 0.3779, "step": 2893 }, { "epoch": 0.1, "learning_rate": 4.9191908822497155e-05, "loss": 0.5408, "step": 2894 }, { "epoch": 0.1, "learning_rate": 4.919123724892406e-05, "loss": 0.8357, "step": 2895 }, { "epoch": 0.1, "learning_rate": 4.9190565400995406e-05, "loss": 0.3116, "step": 2896 }, { "epoch": 0.1, "learning_rate": 4.918989327871878e-05, "loss": 0.9584, "step": 2897 }, { "epoch": 0.1, "learning_rate": 4.9189220882101813e-05, "loss": 0.9753, "step": 2898 }, { "epoch": 0.1, "learning_rate": 4.9188548211152144e-05, "loss": 1.0388, "step": 2899 }, { "epoch": 0.1, "learning_rate": 4.918787526587739e-05, "loss": 0.876, "step": 2900 }, { "epoch": 0.1, "learning_rate": 4.9187202046285195e-05, "loss": 1.1392, "step": 2901 }, { "epoch": 0.1, "learning_rate": 4.918652855238318e-05, "loss": 1.0369, "step": 2902 }, { "epoch": 0.1, "learning_rate": 4.9185854784178996e-05, "loss": 1.1057, "step": 2903 }, { "epoch": 0.1, "learning_rate": 4.918518074168027e-05, "loss": 1.4788, "step": 2904 }, { "epoch": 0.1, "learning_rate": 4.9184506424894664e-05, "loss": 0.9652, "step": 2905 }, { "epoch": 0.1, "learning_rate": 4.9183831833829804e-05, "loss": 0.7187, "step": 2906 }, { "epoch": 0.1, "learning_rate": 4.918315696849335e-05, "loss": 0.7212, "step": 2907 }, { "epoch": 0.1, "learning_rate": 4.918248182889297e-05, "loss": 1.0003, "step": 2908 }, { "epoch": 0.1, "learning_rate": 4.9181806415036305e-05, "loss": 1.0107, "step": 2909 }, { "epoch": 0.1, "learning_rate": 4.918113072693101e-05, "loss": 0.8184, "step": 2910 }, { "epoch": 0.1, "learning_rate": 4.918045476458477e-05, "loss": 0.7845, "step": 2911 }, { "epoch": 0.1, "learning_rate": 4.9179778528005224e-05, "loss": 0.6977, "step": 2912 }, { "epoch": 0.1, "learning_rate": 4.9179102017200065e-05, "loss": 0.867, "step": 2913 }, { "epoch": 0.1, "learning_rate": 4.917842523217695e-05, "loss": 0.93, "step": 2914 }, { "epoch": 0.1, "learning_rate": 4.917774817294356e-05, "loss": 0.434, "step": 2915 }, { "epoch": 0.1, "learning_rate": 4.917707083950758e-05, "loss": 0.8969, "step": 2916 }, { "epoch": 0.1, "learning_rate": 4.9176393231876685e-05, "loss": 2.1638, "step": 2917 }, { "epoch": 0.1, "learning_rate": 4.917571535005855e-05, "loss": 0.8652, "step": 2918 }, { "epoch": 0.1, "learning_rate": 4.917503719406088e-05, "loss": 0.9903, "step": 2919 }, { "epoch": 0.1, "learning_rate": 4.917435876389136e-05, "loss": 0.5729, "step": 2920 }, { "epoch": 0.1, "learning_rate": 4.9173680059557684e-05, "loss": 0.943, "step": 2921 }, { "epoch": 0.1, "learning_rate": 4.917300108106754e-05, "loss": 0.9406, "step": 2922 }, { "epoch": 0.1, "learning_rate": 4.917232182842865e-05, "loss": 0.6409, "step": 2923 }, { "epoch": 0.1, "learning_rate": 4.917164230164869e-05, "loss": 0.8875, "step": 2924 }, { "epoch": 0.1, "learning_rate": 4.917096250073538e-05, "loss": 1.0291, "step": 2925 }, { "epoch": 0.1, "learning_rate": 4.917028242569645e-05, "loss": 3.0572, "step": 2926 }, { "epoch": 0.1, "learning_rate": 4.916960207653958e-05, "loss": 2.8362, "step": 2927 }, { "epoch": 0.1, "learning_rate": 4.91689214532725e-05, "loss": 4.6547, "step": 2928 }, { "epoch": 0.1, "learning_rate": 4.9168240555902937e-05, "loss": 2.9581, "step": 2929 }, { "epoch": 0.1, "learning_rate": 4.916755938443859e-05, "loss": 0.9473, "step": 2930 }, { "epoch": 0.1, "learning_rate": 4.916687793888721e-05, "loss": 1.048, "step": 2931 }, { "epoch": 0.1, "learning_rate": 4.916619621925651e-05, "loss": 0.9375, "step": 2932 }, { "epoch": 0.1, "learning_rate": 4.916551422555422e-05, "loss": 0.9951, "step": 2933 }, { "epoch": 0.1, "learning_rate": 4.916483195778809e-05, "loss": 1.0724, "step": 2934 }, { "epoch": 0.1, "learning_rate": 4.9164149415965846e-05, "loss": 0.6018, "step": 2935 }, { "epoch": 0.1, "learning_rate": 4.916346660009523e-05, "loss": 0.9831, "step": 2936 }, { "epoch": 0.1, "learning_rate": 4.916278351018398e-05, "loss": 1.0309, "step": 2937 }, { "epoch": 0.1, "learning_rate": 4.9162100146239854e-05, "loss": 0.8304, "step": 2938 }, { "epoch": 0.1, "learning_rate": 4.9161416508270594e-05, "loss": 0.7391, "step": 2939 }, { "epoch": 0.1, "learning_rate": 4.916073259628397e-05, "loss": 1.0772, "step": 2940 }, { "epoch": 0.1, "learning_rate": 4.916004841028771e-05, "loss": 0.9868, "step": 2941 }, { "epoch": 0.1, "learning_rate": 4.915936395028959e-05, "loss": 1.1374, "step": 2942 }, { "epoch": 0.1, "learning_rate": 4.915867921629738e-05, "loss": 0.9696, "step": 2943 }, { "epoch": 0.1, "learning_rate": 4.915799420831884e-05, "loss": 1.1736, "step": 2944 }, { "epoch": 0.1, "learning_rate": 4.915730892636172e-05, "loss": 0.4926, "step": 2945 }, { "epoch": 0.1, "learning_rate": 4.915662337043382e-05, "loss": 0.4304, "step": 2946 }, { "epoch": 0.1, "learning_rate": 4.9155937540542894e-05, "loss": 0.3792, "step": 2947 }, { "epoch": 0.1, "learning_rate": 4.915525143669674e-05, "loss": 0.8053, "step": 2948 }, { "epoch": 0.1, "learning_rate": 4.9154565058903115e-05, "loss": 0.939, "step": 2949 }, { "epoch": 0.1, "learning_rate": 4.915387840716982e-05, "loss": 0.9854, "step": 2950 }, { "epoch": 0.1, "learning_rate": 4.915319148150464e-05, "loss": 0.9995, "step": 2951 }, { "epoch": 0.1, "learning_rate": 4.9152504281915365e-05, "loss": 0.9472, "step": 2952 }, { "epoch": 0.1, "learning_rate": 4.915181680840978e-05, "loss": 0.8589, "step": 2953 }, { "epoch": 0.1, "learning_rate": 4.915112906099569e-05, "loss": 0.7328, "step": 2954 }, { "epoch": 0.1, "learning_rate": 4.915044103968091e-05, "loss": 0.5568, "step": 2955 }, { "epoch": 0.1, "learning_rate": 4.914975274447321e-05, "loss": 0.9883, "step": 2956 }, { "epoch": 0.1, "learning_rate": 4.914906417538042e-05, "loss": 0.6185, "step": 2957 }, { "epoch": 0.1, "learning_rate": 4.914837533241034e-05, "loss": 1.0483, "step": 2958 }, { "epoch": 0.1, "learning_rate": 4.9147686215570785e-05, "loss": 1.4448, "step": 2959 }, { "epoch": 0.1, "learning_rate": 4.914699682486957e-05, "loss": 1.3869, "step": 2960 }, { "epoch": 0.1, "learning_rate": 4.914630716031451e-05, "loss": 1.4336, "step": 2961 }, { "epoch": 0.1, "learning_rate": 4.914561722191343e-05, "loss": 1.4368, "step": 2962 }, { "epoch": 0.1, "learning_rate": 4.914492700967416e-05, "loss": 1.4251, "step": 2963 }, { "epoch": 0.1, "learning_rate": 4.914423652360452e-05, "loss": 1.3724, "step": 2964 }, { "epoch": 0.1, "learning_rate": 4.914354576371234e-05, "loss": 0.6619, "step": 2965 }, { "epoch": 0.1, "learning_rate": 4.914285473000545e-05, "loss": 1.0889, "step": 2966 }, { "epoch": 0.1, "learning_rate": 4.91421634224917e-05, "loss": 0.9954, "step": 2967 }, { "epoch": 0.1, "learning_rate": 4.914147184117893e-05, "loss": 0.6238, "step": 2968 }, { "epoch": 0.1, "learning_rate": 4.914077998607498e-05, "loss": 0.8328, "step": 2969 }, { "epoch": 0.1, "learning_rate": 4.914008785718769e-05, "loss": 1.0113, "step": 2970 }, { "epoch": 0.1, "learning_rate": 4.9139395454524905e-05, "loss": 1.051, "step": 2971 }, { "epoch": 0.1, "learning_rate": 4.91387027780945e-05, "loss": 1.1452, "step": 2972 }, { "epoch": 0.1, "learning_rate": 4.913800982790431e-05, "loss": 0.8898, "step": 2973 }, { "epoch": 0.1, "learning_rate": 4.91373166039622e-05, "loss": 0.8241, "step": 2974 }, { "epoch": 0.1, "learning_rate": 4.9136623106276034e-05, "loss": 1.0008, "step": 2975 }, { "epoch": 0.1, "learning_rate": 4.913592933485367e-05, "loss": 0.9618, "step": 2976 }, { "epoch": 0.1, "learning_rate": 4.913523528970299e-05, "loss": 0.6841, "step": 2977 }, { "epoch": 0.1, "learning_rate": 4.913454097083185e-05, "loss": 0.8788, "step": 2978 }, { "epoch": 0.1, "learning_rate": 4.913384637824813e-05, "loss": 0.9323, "step": 2979 }, { "epoch": 0.1, "learning_rate": 4.9133151511959715e-05, "loss": 1.2273, "step": 2980 }, { "epoch": 0.1, "learning_rate": 4.9132456371974475e-05, "loss": 0.7171, "step": 2981 }, { "epoch": 0.1, "learning_rate": 4.913176095830031e-05, "loss": 0.8205, "step": 2982 }, { "epoch": 0.1, "learning_rate": 4.913106527094508e-05, "loss": 0.8919, "step": 2983 }, { "epoch": 0.1, "learning_rate": 4.91303693099167e-05, "loss": 0.9342, "step": 2984 }, { "epoch": 0.1, "learning_rate": 4.912967307522305e-05, "loss": 1.0375, "step": 2985 }, { "epoch": 0.1, "learning_rate": 4.912897656687202e-05, "loss": 1.1325, "step": 2986 }, { "epoch": 0.1, "learning_rate": 4.912827978487152e-05, "loss": 0.9109, "step": 2987 }, { "epoch": 0.1, "learning_rate": 4.912758272922946e-05, "loss": 0.9195, "step": 2988 }, { "epoch": 0.1, "learning_rate": 4.912688539995373e-05, "loss": 0.7258, "step": 2989 }, { "epoch": 0.1, "learning_rate": 4.9126187797052245e-05, "loss": 1.2197, "step": 2990 }, { "epoch": 0.1, "learning_rate": 4.9125489920532916e-05, "loss": 0.8844, "step": 2991 }, { "epoch": 0.1, "learning_rate": 4.9124791770403655e-05, "loss": 0.7179, "step": 2992 }, { "epoch": 0.1, "learning_rate": 4.9124093346672385e-05, "loss": 0.4801, "step": 2993 }, { "epoch": 0.1, "learning_rate": 4.912339464934702e-05, "loss": 1.0047, "step": 2994 }, { "epoch": 0.1, "learning_rate": 4.9122695678435494e-05, "loss": 1.0545, "step": 2995 }, { "epoch": 0.1, "learning_rate": 4.912199643394573e-05, "loss": 0.887, "step": 2996 }, { "epoch": 0.1, "learning_rate": 4.912129691588564e-05, "loss": 0.8997, "step": 2997 }, { "epoch": 0.1, "learning_rate": 4.912059712426319e-05, "loss": 0.8285, "step": 2998 }, { "epoch": 0.1, "learning_rate": 4.9119897059086297e-05, "loss": 0.8743, "step": 2999 }, { "epoch": 0.1, "learning_rate": 4.91191967203629e-05, "loss": 0.9389, "step": 3000 }, { "epoch": 0.1, "learning_rate": 4.911849610810095e-05, "loss": 0.9675, "step": 3001 }, { "epoch": 0.1, "learning_rate": 4.9117795222308383e-05, "loss": 0.7254, "step": 3002 }, { "epoch": 0.1, "learning_rate": 4.911709406299315e-05, "loss": 1.0201, "step": 3003 }, { "epoch": 0.1, "learning_rate": 4.911639263016321e-05, "loss": 1.0423, "step": 3004 }, { "epoch": 0.1, "learning_rate": 4.911569092382652e-05, "loss": 0.9483, "step": 3005 }, { "epoch": 0.1, "learning_rate": 4.911498894399103e-05, "loss": 0.9704, "step": 3006 }, { "epoch": 0.1, "learning_rate": 4.9114286690664695e-05, "loss": 0.8727, "step": 3007 }, { "epoch": 0.1, "learning_rate": 4.9113584163855494e-05, "loss": 0.9124, "step": 3008 }, { "epoch": 0.1, "learning_rate": 4.9112881363571386e-05, "loss": 0.7466, "step": 3009 }, { "epoch": 0.1, "learning_rate": 4.9112178289820346e-05, "loss": 0.9963, "step": 3010 }, { "epoch": 0.1, "learning_rate": 4.911147494261034e-05, "loss": 1.1421, "step": 3011 }, { "epoch": 0.1, "learning_rate": 4.911077132194935e-05, "loss": 0.937, "step": 3012 }, { "epoch": 0.1, "learning_rate": 4.911006742784536e-05, "loss": 1.1851, "step": 3013 }, { "epoch": 0.1, "learning_rate": 4.9109363260306343e-05, "loss": 1.4971, "step": 3014 }, { "epoch": 0.1, "learning_rate": 4.9108658819340294e-05, "loss": 1.569, "step": 3015 }, { "epoch": 0.1, "learning_rate": 4.91079541049552e-05, "loss": 1.0889, "step": 3016 }, { "epoch": 0.1, "learning_rate": 4.9107249117159046e-05, "loss": 1.0091, "step": 3017 }, { "epoch": 0.1, "learning_rate": 4.910654385595984e-05, "loss": 1.0186, "step": 3018 }, { "epoch": 0.1, "learning_rate": 4.910583832136557e-05, "loss": 0.4901, "step": 3019 }, { "epoch": 0.1, "learning_rate": 4.910513251338424e-05, "loss": 0.6791, "step": 3020 }, { "epoch": 0.1, "learning_rate": 4.9104426432023864e-05, "loss": 0.8656, "step": 3021 }, { "epoch": 0.1, "learning_rate": 4.910372007729244e-05, "loss": 0.9544, "step": 3022 }, { "epoch": 0.1, "learning_rate": 4.9103013449197973e-05, "loss": 0.9493, "step": 3023 }, { "epoch": 0.1, "learning_rate": 4.9102306547748494e-05, "loss": 0.4715, "step": 3024 }, { "epoch": 0.1, "learning_rate": 4.9101599372952e-05, "loss": 3.2232, "step": 3025 }, { "epoch": 0.1, "learning_rate": 4.9100891924816537e-05, "loss": 5.8427, "step": 3026 }, { "epoch": 0.1, "learning_rate": 4.91001842033501e-05, "loss": 5.8202, "step": 3027 }, { "epoch": 0.1, "learning_rate": 4.9099476208560736e-05, "loss": 3.7644, "step": 3028 }, { "epoch": 0.1, "learning_rate": 4.909876794045646e-05, "loss": 0.9529, "step": 3029 }, { "epoch": 0.1, "learning_rate": 4.909805939904532e-05, "loss": 0.9736, "step": 3030 }, { "epoch": 0.1, "learning_rate": 4.909735058433533e-05, "loss": 0.9766, "step": 3031 }, { "epoch": 0.1, "learning_rate": 4.909664149633455e-05, "loss": 0.8643, "step": 3032 }, { "epoch": 0.1, "learning_rate": 4.9095932135051014e-05, "loss": 1.4726, "step": 3033 }, { "epoch": 0.1, "learning_rate": 4.9095222500492765e-05, "loss": 1.6492, "step": 3034 }, { "epoch": 0.1, "learning_rate": 4.909451259266785e-05, "loss": 1.5725, "step": 3035 }, { "epoch": 0.1, "learning_rate": 4.909380241158433e-05, "loss": 0.6699, "step": 3036 }, { "epoch": 0.1, "learning_rate": 4.909309195725025e-05, "loss": 1.4404, "step": 3037 }, { "epoch": 0.1, "learning_rate": 4.909238122967366e-05, "loss": 1.1583, "step": 3038 }, { "epoch": 0.1, "learning_rate": 4.909167022886264e-05, "loss": 0.9407, "step": 3039 }, { "epoch": 0.1, "learning_rate": 4.909095895482524e-05, "loss": 0.4411, "step": 3040 }, { "epoch": 0.1, "learning_rate": 4.909024740756953e-05, "loss": 0.2169, "step": 3041 }, { "epoch": 0.1, "learning_rate": 4.908953558710358e-05, "loss": 0.3919, "step": 3042 }, { "epoch": 0.1, "learning_rate": 4.9088823493435465e-05, "loss": 0.8697, "step": 3043 }, { "epoch": 0.1, "learning_rate": 4.908811112657326e-05, "loss": 1.5277, "step": 3044 }, { "epoch": 0.1, "learning_rate": 4.908739848652503e-05, "loss": 0.796, "step": 3045 }, { "epoch": 0.1, "learning_rate": 4.908668557329888e-05, "loss": 0.7943, "step": 3046 }, { "epoch": 0.1, "learning_rate": 4.908597238690288e-05, "loss": 0.9939, "step": 3047 }, { "epoch": 0.1, "learning_rate": 4.908525892734513e-05, "loss": 0.9243, "step": 3048 }, { "epoch": 0.1, "learning_rate": 4.9084545194633715e-05, "loss": 1.3031, "step": 3049 }, { "epoch": 0.1, "learning_rate": 4.908383118877672e-05, "loss": 1.194, "step": 3050 }, { "epoch": 0.1, "learning_rate": 4.9083116909782254e-05, "loss": 0.9784, "step": 3051 }, { "epoch": 0.1, "learning_rate": 4.908240235765842e-05, "loss": 0.9512, "step": 3052 }, { "epoch": 0.1, "learning_rate": 4.908168753241331e-05, "loss": 1.0836, "step": 3053 }, { "epoch": 0.1, "learning_rate": 4.908097243405505e-05, "loss": 0.7402, "step": 3054 }, { "epoch": 0.1, "learning_rate": 4.908025706259173e-05, "loss": 0.6285, "step": 3055 }, { "epoch": 0.1, "learning_rate": 4.907954141803147e-05, "loss": 0.654, "step": 3056 }, { "epoch": 0.1, "learning_rate": 4.90788255003824e-05, "loss": 1.0529, "step": 3057 }, { "epoch": 0.1, "learning_rate": 4.907810930965261e-05, "loss": 1.0722, "step": 3058 }, { "epoch": 0.1, "learning_rate": 4.907739284585025e-05, "loss": 0.8973, "step": 3059 }, { "epoch": 0.1, "learning_rate": 4.907667610898343e-05, "loss": 1.1458, "step": 3060 }, { "epoch": 0.1, "learning_rate": 4.907595909906029e-05, "loss": 1.1446, "step": 3061 }, { "epoch": 0.1, "learning_rate": 4.907524181608895e-05, "loss": 1.8787, "step": 3062 }, { "epoch": 0.1, "learning_rate": 4.907452426007755e-05, "loss": 0.6397, "step": 3063 }, { "epoch": 0.1, "learning_rate": 4.9073806431034224e-05, "loss": 0.9195, "step": 3064 }, { "epoch": 0.1, "learning_rate": 4.907308832896712e-05, "loss": 0.9018, "step": 3065 }, { "epoch": 0.1, "learning_rate": 4.907236995388438e-05, "loss": 0.7634, "step": 3066 }, { "epoch": 0.1, "learning_rate": 4.907165130579415e-05, "loss": 0.6919, "step": 3067 }, { "epoch": 0.1, "learning_rate": 4.907093238470458e-05, "loss": 0.744, "step": 3068 }, { "epoch": 0.1, "learning_rate": 4.907021319062383e-05, "loss": 0.8948, "step": 3069 }, { "epoch": 0.1, "learning_rate": 4.9069493723560036e-05, "loss": 1.0167, "step": 3070 }, { "epoch": 0.1, "learning_rate": 4.9068773983521376e-05, "loss": 0.9496, "step": 3071 }, { "epoch": 0.1, "learning_rate": 4.9068053970516016e-05, "loss": 0.9259, "step": 3072 }, { "epoch": 0.1, "learning_rate": 4.9067333684552106e-05, "loss": 1.0425, "step": 3073 }, { "epoch": 0.1, "learning_rate": 4.9066613125637827e-05, "loss": 0.7556, "step": 3074 }, { "epoch": 0.1, "learning_rate": 4.906589229378135e-05, "loss": 1.0281, "step": 3075 }, { "epoch": 0.1, "learning_rate": 4.906517118899083e-05, "loss": 1.2577, "step": 3076 }, { "epoch": 0.1, "learning_rate": 4.906444981127448e-05, "loss": 1.5615, "step": 3077 }, { "epoch": 0.1, "learning_rate": 4.9063728160640455e-05, "loss": 1.7236, "step": 3078 }, { "epoch": 0.1, "learning_rate": 4.9063006237096954e-05, "loss": 1.7394, "step": 3079 }, { "epoch": 0.1, "learning_rate": 4.906228404065215e-05, "loss": 0.6572, "step": 3080 }, { "epoch": 0.1, "learning_rate": 4.9061561571314254e-05, "loss": 0.9779, "step": 3081 }, { "epoch": 0.1, "learning_rate": 4.9060838829091434e-05, "loss": 1.0343, "step": 3082 }, { "epoch": 0.1, "learning_rate": 4.90601158139919e-05, "loss": 0.9418, "step": 3083 }, { "epoch": 0.1, "learning_rate": 4.9059392526023864e-05, "loss": 0.8865, "step": 3084 }, { "epoch": 0.1, "learning_rate": 4.9058668965195505e-05, "loss": 1.0439, "step": 3085 }, { "epoch": 0.1, "learning_rate": 4.905794513151504e-05, "loss": 1.2827, "step": 3086 }, { "epoch": 0.1, "learning_rate": 4.905722102499068e-05, "loss": 0.9717, "step": 3087 }, { "epoch": 0.1, "learning_rate": 4.905649664563064e-05, "loss": 0.435, "step": 3088 }, { "epoch": 0.1, "learning_rate": 4.9055771993443134e-05, "loss": 0.7312, "step": 3089 }, { "epoch": 0.1, "learning_rate": 4.9055047068436365e-05, "loss": 0.8812, "step": 3090 }, { "epoch": 0.1, "learning_rate": 4.905432187061858e-05, "loss": 0.6801, "step": 3091 }, { "epoch": 0.1, "learning_rate": 4.9053596399997984e-05, "loss": 0.6533, "step": 3092 }, { "epoch": 0.1, "learning_rate": 4.9052870656582804e-05, "loss": 0.9127, "step": 3093 }, { "epoch": 0.1, "learning_rate": 4.9052144640381294e-05, "loss": 0.9505, "step": 3094 }, { "epoch": 0.1, "learning_rate": 4.905141835140166e-05, "loss": 0.4576, "step": 3095 }, { "epoch": 0.1, "learning_rate": 4.905069178965215e-05, "loss": 0.3328, "step": 3096 }, { "epoch": 0.1, "learning_rate": 4.9049964955141004e-05, "loss": 0.2618, "step": 3097 }, { "epoch": 0.1, "learning_rate": 4.904923784787647e-05, "loss": 0.5302, "step": 3098 }, { "epoch": 0.1, "learning_rate": 4.904851046786679e-05, "loss": 0.8032, "step": 3099 }, { "epoch": 0.1, "learning_rate": 4.9047782815120214e-05, "loss": 0.9541, "step": 3100 }, { "epoch": 0.1, "learning_rate": 4.904705488964499e-05, "loss": 0.7552, "step": 3101 }, { "epoch": 0.1, "learning_rate": 4.9046326691449376e-05, "loss": 0.8295, "step": 3102 }, { "epoch": 0.1, "learning_rate": 4.9045598220541634e-05, "loss": 1.484, "step": 3103 }, { "epoch": 0.1, "learning_rate": 4.904486947693002e-05, "loss": 0.9762, "step": 3104 }, { "epoch": 0.1, "learning_rate": 4.9044140460622815e-05, "loss": 1.9138, "step": 3105 }, { "epoch": 0.1, "learning_rate": 4.9043411171628264e-05, "loss": 1.4186, "step": 3106 }, { "epoch": 0.1, "learning_rate": 4.9042681609954644e-05, "loss": 0.829, "step": 3107 }, { "epoch": 0.1, "learning_rate": 4.9041951775610236e-05, "loss": 0.9931, "step": 3108 }, { "epoch": 0.1, "learning_rate": 4.904122166860331e-05, "loss": 0.8522, "step": 3109 }, { "epoch": 0.1, "learning_rate": 4.904049128894216e-05, "loss": 0.7948, "step": 3110 }, { "epoch": 0.1, "learning_rate": 4.9039760636635054e-05, "loss": 0.9879, "step": 3111 }, { "epoch": 0.1, "learning_rate": 4.9039029711690284e-05, "loss": 1.1056, "step": 3112 }, { "epoch": 0.1, "learning_rate": 4.9038298514116136e-05, "loss": 0.9172, "step": 3113 }, { "epoch": 0.1, "learning_rate": 4.9037567043920904e-05, "loss": 3.2036, "step": 3114 }, { "epoch": 0.1, "learning_rate": 4.90368353011129e-05, "loss": 3.674, "step": 3115 }, { "epoch": 0.1, "learning_rate": 4.9036103285700396e-05, "loss": 1.5073, "step": 3116 }, { "epoch": 0.1, "learning_rate": 4.903537099769171e-05, "loss": 1.1077, "step": 3117 }, { "epoch": 0.1, "learning_rate": 4.903463843709515e-05, "loss": 1.0851, "step": 3118 }, { "epoch": 0.1, "learning_rate": 4.9033905603919006e-05, "loss": 1.0632, "step": 3119 }, { "epoch": 0.1, "learning_rate": 4.9033172498171606e-05, "loss": 0.8252, "step": 3120 }, { "epoch": 0.1, "learning_rate": 4.9032439119861256e-05, "loss": 0.813, "step": 3121 }, { "epoch": 0.1, "learning_rate": 4.903170546899628e-05, "loss": 0.9663, "step": 3122 }, { "epoch": 0.1, "learning_rate": 4.903097154558499e-05, "loss": 0.9661, "step": 3123 }, { "epoch": 0.1, "learning_rate": 4.9030237349635724e-05, "loss": 0.7889, "step": 3124 }, { "epoch": 0.1, "learning_rate": 4.902950288115679e-05, "loss": 0.9484, "step": 3125 }, { "epoch": 0.1, "learning_rate": 4.902876814015652e-05, "loss": 0.9578, "step": 3126 }, { "epoch": 0.1, "learning_rate": 4.9028033126643266e-05, "loss": 1.1105, "step": 3127 }, { "epoch": 0.1, "learning_rate": 4.902729784062534e-05, "loss": 0.8453, "step": 3128 }, { "epoch": 0.1, "learning_rate": 4.90265622821111e-05, "loss": 1.0961, "step": 3129 }, { "epoch": 0.1, "learning_rate": 4.902582645110888e-05, "loss": 1.3779, "step": 3130 }, { "epoch": 0.1, "learning_rate": 4.902509034762702e-05, "loss": 0.6189, "step": 3131 }, { "epoch": 0.1, "learning_rate": 4.9024353971673875e-05, "loss": 0.7565, "step": 3132 }, { "epoch": 0.1, "learning_rate": 4.90236173232578e-05, "loss": 1.4753, "step": 3133 }, { "epoch": 0.1, "learning_rate": 4.902288040238714e-05, "loss": 1.1348, "step": 3134 }, { "epoch": 0.1, "learning_rate": 4.902214320907025e-05, "loss": 1.1631, "step": 3135 }, { "epoch": 0.1, "learning_rate": 4.902140574331551e-05, "loss": 0.9187, "step": 3136 }, { "epoch": 0.1, "learning_rate": 4.902066800513127e-05, "loss": 0.7045, "step": 3137 }, { "epoch": 0.1, "learning_rate": 4.901992999452588e-05, "loss": 1.0456, "step": 3138 }, { "epoch": 0.1, "learning_rate": 4.9019191711507745e-05, "loss": 3.6306, "step": 3139 }, { "epoch": 0.1, "learning_rate": 4.901845315608522e-05, "loss": 5.6194, "step": 3140 }, { "epoch": 0.1, "learning_rate": 4.901771432826667e-05, "loss": 5.6596, "step": 3141 }, { "epoch": 0.1, "learning_rate": 4.901697522806049e-05, "loss": 5.6974, "step": 3142 }, { "epoch": 0.1, "learning_rate": 4.901623585547506e-05, "loss": 2.9655, "step": 3143 }, { "epoch": 0.1, "learning_rate": 4.9015496210518764e-05, "loss": 0.9461, "step": 3144 }, { "epoch": 0.1, "learning_rate": 4.901475629319999e-05, "loss": 0.7112, "step": 3145 }, { "epoch": 0.1, "learning_rate": 4.901401610352713e-05, "loss": 0.8072, "step": 3146 }, { "epoch": 0.1, "learning_rate": 4.9013275641508574e-05, "loss": 0.7742, "step": 3147 }, { "epoch": 0.1, "learning_rate": 4.901253490715272e-05, "loss": 1.0452, "step": 3148 }, { "epoch": 0.1, "learning_rate": 4.9011793900467985e-05, "loss": 1.2812, "step": 3149 }, { "epoch": 0.1, "learning_rate": 4.901105262146275e-05, "loss": 0.8219, "step": 3150 }, { "epoch": 0.11, "learning_rate": 4.901031107014543e-05, "loss": 0.871, "step": 3151 }, { "epoch": 0.11, "learning_rate": 4.900956924652445e-05, "loss": 0.9846, "step": 3152 }, { "epoch": 0.11, "learning_rate": 4.9008827150608195e-05, "loss": 1.0204, "step": 3153 }, { "epoch": 0.11, "learning_rate": 4.90080847824051e-05, "loss": 0.9388, "step": 3154 }, { "epoch": 0.11, "learning_rate": 4.900734214192358e-05, "loss": 0.824, "step": 3155 }, { "epoch": 0.11, "learning_rate": 4.900659922917206e-05, "loss": 0.8926, "step": 3156 }, { "epoch": 0.11, "learning_rate": 4.900585604415896e-05, "loss": 0.9076, "step": 3157 }, { "epoch": 0.11, "learning_rate": 4.900511258689271e-05, "loss": 0.9625, "step": 3158 }, { "epoch": 0.11, "learning_rate": 4.900436885738175e-05, "loss": 0.8787, "step": 3159 }, { "epoch": 0.11, "learning_rate": 4.900362485563451e-05, "loss": 1.7495, "step": 3160 }, { "epoch": 0.11, "learning_rate": 4.900288058165942e-05, "loss": 1.2659, "step": 3161 }, { "epoch": 0.11, "learning_rate": 4.900213603546493e-05, "loss": 1.1625, "step": 3162 }, { "epoch": 0.11, "learning_rate": 4.900139121705948e-05, "loss": 0.8639, "step": 3163 }, { "epoch": 0.11, "learning_rate": 4.9000646126451515e-05, "loss": 0.931, "step": 3164 }, { "epoch": 0.11, "learning_rate": 4.899990076364949e-05, "loss": 0.4821, "step": 3165 }, { "epoch": 0.11, "learning_rate": 4.899915512866186e-05, "loss": 0.7097, "step": 3166 }, { "epoch": 0.11, "learning_rate": 4.8998409221497076e-05, "loss": 1.0915, "step": 3167 }, { "epoch": 0.11, "learning_rate": 4.899766304216359e-05, "loss": 0.8249, "step": 3168 }, { "epoch": 0.11, "learning_rate": 4.899691659066988e-05, "loss": 1.0894, "step": 3169 }, { "epoch": 0.11, "learning_rate": 4.899616986702441e-05, "loss": 1.6386, "step": 3170 }, { "epoch": 0.11, "learning_rate": 4.899542287123564e-05, "loss": 1.7271, "step": 3171 }, { "epoch": 0.11, "learning_rate": 4.899467560331204e-05, "loss": 1.7489, "step": 3172 }, { "epoch": 0.11, "learning_rate": 4.899392806326211e-05, "loss": 1.2105, "step": 3173 }, { "epoch": 0.11, "learning_rate": 4.899318025109429e-05, "loss": 0.9461, "step": 3174 }, { "epoch": 0.11, "learning_rate": 4.8992432166817084e-05, "loss": 0.5829, "step": 3175 }, { "epoch": 0.11, "learning_rate": 4.899168381043897e-05, "loss": 0.3004, "step": 3176 }, { "epoch": 0.11, "learning_rate": 4.8990935181968435e-05, "loss": 0.8916, "step": 3177 }, { "epoch": 0.11, "learning_rate": 4.899018628141398e-05, "loss": 0.8906, "step": 3178 }, { "epoch": 0.11, "learning_rate": 4.898943710878407e-05, "loss": 0.8808, "step": 3179 }, { "epoch": 0.11, "learning_rate": 4.898868766408724e-05, "loss": 0.7613, "step": 3180 }, { "epoch": 0.11, "learning_rate": 4.898793794733196e-05, "loss": 1.0238, "step": 3181 }, { "epoch": 0.11, "learning_rate": 4.898718795852675e-05, "loss": 0.7726, "step": 3182 }, { "epoch": 0.11, "learning_rate": 4.898643769768011e-05, "loss": 0.9722, "step": 3183 }, { "epoch": 0.11, "learning_rate": 4.8985687164800545e-05, "loss": 1.2964, "step": 3184 }, { "epoch": 0.11, "learning_rate": 4.8984936359896565e-05, "loss": 0.8917, "step": 3185 }, { "epoch": 0.11, "learning_rate": 4.898418528297669e-05, "loss": 0.7199, "step": 3186 }, { "epoch": 0.11, "learning_rate": 4.898343393404944e-05, "loss": 0.683, "step": 3187 }, { "epoch": 0.11, "learning_rate": 4.898268231312333e-05, "loss": 0.8777, "step": 3188 }, { "epoch": 0.11, "learning_rate": 4.8981930420206894e-05, "loss": 0.9102, "step": 3189 }, { "epoch": 0.11, "learning_rate": 4.898117825530865e-05, "loss": 0.9366, "step": 3190 }, { "epoch": 0.11, "learning_rate": 4.898042581843713e-05, "loss": 0.7614, "step": 3191 }, { "epoch": 0.11, "learning_rate": 4.8979673109600874e-05, "loss": 1.0673, "step": 3192 }, { "epoch": 0.11, "learning_rate": 4.89789201288084e-05, "loss": 1.1875, "step": 3193 }, { "epoch": 0.11, "learning_rate": 4.897816687606828e-05, "loss": 0.8773, "step": 3194 }, { "epoch": 0.11, "learning_rate": 4.897741335138902e-05, "loss": 0.1393, "step": 3195 }, { "epoch": 0.11, "learning_rate": 4.89766595547792e-05, "loss": 0.7604, "step": 3196 }, { "epoch": 0.11, "learning_rate": 4.897590548624734e-05, "loss": 0.5386, "step": 3197 }, { "epoch": 0.11, "learning_rate": 4.8975151145802015e-05, "loss": 0.6556, "step": 3198 }, { "epoch": 0.11, "learning_rate": 4.897439653345176e-05, "loss": 0.9685, "step": 3199 }, { "epoch": 0.11, "learning_rate": 4.897364164920514e-05, "loss": 0.8823, "step": 3200 }, { "epoch": 0.11, "learning_rate": 4.897288649307072e-05, "loss": 0.9692, "step": 3201 }, { "epoch": 0.11, "learning_rate": 4.897213106505707e-05, "loss": 1.682, "step": 3202 }, { "epoch": 0.11, "learning_rate": 4.897137536517276e-05, "loss": 2.7436, "step": 3203 }, { "epoch": 0.11, "learning_rate": 4.897061939342633e-05, "loss": 0.9526, "step": 3204 }, { "epoch": 0.11, "learning_rate": 4.896986314982639e-05, "loss": 0.8988, "step": 3205 }, { "epoch": 0.11, "learning_rate": 4.896910663438149e-05, "loss": 0.6876, "step": 3206 }, { "epoch": 0.11, "learning_rate": 4.896834984710023e-05, "loss": 1.3531, "step": 3207 }, { "epoch": 0.11, "learning_rate": 4.896759278799118e-05, "loss": 1.1211, "step": 3208 }, { "epoch": 0.11, "learning_rate": 4.896683545706293e-05, "loss": 0.7829, "step": 3209 }, { "epoch": 0.11, "learning_rate": 4.896607785432407e-05, "loss": 0.6558, "step": 3210 }, { "epoch": 0.11, "learning_rate": 4.8965319979783185e-05, "loss": 0.4686, "step": 3211 }, { "epoch": 0.11, "learning_rate": 4.8964561833448885e-05, "loss": 0.5671, "step": 3212 }, { "epoch": 0.11, "learning_rate": 4.896380341532976e-05, "loss": 1.0266, "step": 3213 }, { "epoch": 0.11, "learning_rate": 4.89630447254344e-05, "loss": 0.9765, "step": 3214 }, { "epoch": 0.11, "learning_rate": 4.896228576377142e-05, "loss": 0.6265, "step": 3215 }, { "epoch": 0.11, "learning_rate": 4.896152653034943e-05, "loss": 0.4311, "step": 3216 }, { "epoch": 0.11, "learning_rate": 4.896076702517705e-05, "loss": 0.9135, "step": 3217 }, { "epoch": 0.11, "learning_rate": 4.896000724826287e-05, "loss": 0.8111, "step": 3218 }, { "epoch": 0.11, "learning_rate": 4.8959247199615524e-05, "loss": 0.8485, "step": 3219 }, { "epoch": 0.11, "learning_rate": 4.8958486879243615e-05, "loss": 0.6469, "step": 3220 }, { "epoch": 0.11, "learning_rate": 4.895772628715578e-05, "loss": 0.8429, "step": 3221 }, { "epoch": 0.11, "learning_rate": 4.895696542336064e-05, "loss": 0.6276, "step": 3222 }, { "epoch": 0.11, "learning_rate": 4.895620428786684e-05, "loss": 0.8571, "step": 3223 }, { "epoch": 0.11, "learning_rate": 4.895544288068299e-05, "loss": 0.7716, "step": 3224 }, { "epoch": 0.11, "learning_rate": 4.895468120181773e-05, "loss": 0.9464, "step": 3225 }, { "epoch": 0.11, "learning_rate": 4.895391925127971e-05, "loss": 1.2516, "step": 3226 }, { "epoch": 0.11, "learning_rate": 4.895315702907756e-05, "loss": 0.1879, "step": 3227 }, { "epoch": 0.11, "learning_rate": 4.895239453521991e-05, "loss": 0.203, "step": 3228 }, { "epoch": 0.11, "learning_rate": 4.8951631769715447e-05, "loss": 0.8298, "step": 3229 }, { "epoch": 0.11, "learning_rate": 4.8950868732572794e-05, "loss": 1.033, "step": 3230 }, { "epoch": 0.11, "learning_rate": 4.895010542380061e-05, "loss": 0.7431, "step": 3231 }, { "epoch": 0.11, "learning_rate": 4.894934184340754e-05, "loss": 0.6961, "step": 3232 }, { "epoch": 0.11, "learning_rate": 4.8948577991402264e-05, "loss": 0.155, "step": 3233 }, { "epoch": 0.11, "learning_rate": 4.894781386779345e-05, "loss": 0.1255, "step": 3234 }, { "epoch": 0.11, "learning_rate": 4.894704947258973e-05, "loss": 0.1161, "step": 3235 }, { "epoch": 0.11, "learning_rate": 4.89462848057998e-05, "loss": 0.0952, "step": 3236 }, { "epoch": 0.11, "learning_rate": 4.894551986743233e-05, "loss": 1.8567, "step": 3237 }, { "epoch": 0.11, "learning_rate": 4.8944754657495997e-05, "loss": 0.9991, "step": 3238 }, { "epoch": 0.11, "learning_rate": 4.894398917599946e-05, "loss": 0.977, "step": 3239 }, { "epoch": 0.11, "learning_rate": 4.894322342295142e-05, "loss": 0.9501, "step": 3240 }, { "epoch": 0.11, "learning_rate": 4.8942457398360554e-05, "loss": 0.8971, "step": 3241 }, { "epoch": 0.11, "learning_rate": 4.894169110223556e-05, "loss": 0.8896, "step": 3242 }, { "epoch": 0.11, "learning_rate": 4.894092453458511e-05, "loss": 0.8756, "step": 3243 }, { "epoch": 0.11, "learning_rate": 4.894015769541791e-05, "loss": 1.2378, "step": 3244 }, { "epoch": 0.11, "learning_rate": 4.893939058474265e-05, "loss": 0.632, "step": 3245 }, { "epoch": 0.11, "learning_rate": 4.893862320256804e-05, "loss": 0.8153, "step": 3246 }, { "epoch": 0.11, "learning_rate": 4.8937855548902776e-05, "loss": 0.7, "step": 3247 }, { "epoch": 0.11, "learning_rate": 4.8937087623755564e-05, "loss": 0.1159, "step": 3248 }, { "epoch": 0.11, "learning_rate": 4.893631942713511e-05, "loss": 0.2336, "step": 3249 }, { "epoch": 0.11, "learning_rate": 4.893555095905014e-05, "loss": 1.3234, "step": 3250 }, { "epoch": 0.11, "learning_rate": 4.8934782219509356e-05, "loss": 1.0233, "step": 3251 }, { "epoch": 0.11, "learning_rate": 4.893401320852148e-05, "loss": 1.1258, "step": 3252 }, { "epoch": 0.11, "learning_rate": 4.893324392609523e-05, "loss": 1.2418, "step": 3253 }, { "epoch": 0.11, "learning_rate": 4.893247437223934e-05, "loss": 0.9436, "step": 3254 }, { "epoch": 0.11, "learning_rate": 4.893170454696253e-05, "loss": 0.9955, "step": 3255 }, { "epoch": 0.11, "learning_rate": 4.893093445027352e-05, "loss": 0.843, "step": 3256 }, { "epoch": 0.11, "learning_rate": 4.893016408218107e-05, "loss": 0.9247, "step": 3257 }, { "epoch": 0.11, "learning_rate": 4.89293934426939e-05, "loss": 3.3012, "step": 3258 }, { "epoch": 0.11, "learning_rate": 4.8928622531820754e-05, "loss": 4.2434, "step": 3259 }, { "epoch": 0.11, "learning_rate": 4.892785134957038e-05, "loss": 1.0008, "step": 3260 }, { "epoch": 0.11, "learning_rate": 4.89270798959515e-05, "loss": 1.2378, "step": 3261 }, { "epoch": 0.11, "learning_rate": 4.89263081709729e-05, "loss": 1.3278, "step": 3262 }, { "epoch": 0.11, "learning_rate": 4.89255361746433e-05, "loss": 1.1096, "step": 3263 }, { "epoch": 0.11, "learning_rate": 4.8924763906971474e-05, "loss": 0.8463, "step": 3264 }, { "epoch": 0.11, "learning_rate": 4.892399136796617e-05, "loss": 1.1792, "step": 3265 }, { "epoch": 0.11, "learning_rate": 4.892321855763616e-05, "loss": 1.0535, "step": 3266 }, { "epoch": 0.11, "learning_rate": 4.8922445475990206e-05, "loss": 0.7033, "step": 3267 }, { "epoch": 0.11, "learning_rate": 4.8921672123037065e-05, "loss": 0.7147, "step": 3268 }, { "epoch": 0.11, "learning_rate": 4.892089849878552e-05, "loss": 1.0111, "step": 3269 }, { "epoch": 0.11, "learning_rate": 4.8920124603244334e-05, "loss": 1.0022, "step": 3270 }, { "epoch": 0.11, "learning_rate": 4.8919350436422296e-05, "loss": 0.7785, "step": 3271 }, { "epoch": 0.11, "learning_rate": 4.891857599832818e-05, "loss": 0.5168, "step": 3272 }, { "epoch": 0.11, "learning_rate": 4.891780128897077e-05, "loss": 1.6403, "step": 3273 }, { "epoch": 0.11, "learning_rate": 4.8917026308358845e-05, "loss": 0.9321, "step": 3274 }, { "epoch": 0.11, "learning_rate": 4.89162510565012e-05, "loss": 1.005, "step": 3275 }, { "epoch": 0.11, "learning_rate": 4.8915475533406626e-05, "loss": 0.6105, "step": 3276 }, { "epoch": 0.11, "learning_rate": 4.891469973908392e-05, "loss": 1.0372, "step": 3277 }, { "epoch": 0.11, "learning_rate": 4.8913923673541886e-05, "loss": 0.8745, "step": 3278 }, { "epoch": 0.11, "learning_rate": 4.891314733678932e-05, "loss": 0.7878, "step": 3279 }, { "epoch": 0.11, "learning_rate": 4.891237072883502e-05, "loss": 0.8994, "step": 3280 }, { "epoch": 0.11, "learning_rate": 4.89115938496878e-05, "loss": 0.817, "step": 3281 }, { "epoch": 0.11, "learning_rate": 4.891081669935647e-05, "loss": 0.8464, "step": 3282 }, { "epoch": 0.11, "learning_rate": 4.891003927784985e-05, "loss": 0.9125, "step": 3283 }, { "epoch": 0.11, "learning_rate": 4.890926158517674e-05, "loss": 0.9958, "step": 3284 }, { "epoch": 0.11, "learning_rate": 4.890848362134598e-05, "loss": 2.9345, "step": 3285 }, { "epoch": 0.11, "learning_rate": 4.8907705386366374e-05, "loss": 0.5961, "step": 3286 }, { "epoch": 0.11, "learning_rate": 4.8906926880246765e-05, "loss": 0.5034, "step": 3287 }, { "epoch": 0.11, "learning_rate": 4.8906148102995975e-05, "loss": 0.7401, "step": 3288 }, { "epoch": 0.11, "learning_rate": 4.890536905462283e-05, "loss": 0.9091, "step": 3289 }, { "epoch": 0.11, "learning_rate": 4.890458973513618e-05, "loss": 0.7726, "step": 3290 }, { "epoch": 0.11, "learning_rate": 4.890381014454485e-05, "loss": 0.9815, "step": 3291 }, { "epoch": 0.11, "learning_rate": 4.8903030282857686e-05, "loss": 1.3185, "step": 3292 }, { "epoch": 0.11, "learning_rate": 4.890225015008353e-05, "loss": 1.4655, "step": 3293 }, { "epoch": 0.11, "learning_rate": 4.890146974623123e-05, "loss": 1.1088, "step": 3294 }, { "epoch": 0.11, "learning_rate": 4.8900689071309645e-05, "loss": 1.0806, "step": 3295 }, { "epoch": 0.11, "learning_rate": 4.889990812532762e-05, "loss": 0.3523, "step": 3296 }, { "epoch": 0.11, "learning_rate": 4.8899126908294014e-05, "loss": 0.1297, "step": 3297 }, { "epoch": 0.11, "learning_rate": 4.889834542021768e-05, "loss": 1.1678, "step": 3298 }, { "epoch": 0.11, "learning_rate": 4.8897563661107504e-05, "loss": 0.6567, "step": 3299 }, { "epoch": 0.11, "learning_rate": 4.889678163097233e-05, "loss": 0.7636, "step": 3300 }, { "epoch": 0.11, "learning_rate": 4.889599932982103e-05, "loss": 0.898, "step": 3301 }, { "epoch": 0.11, "learning_rate": 4.889521675766248e-05, "loss": 0.869, "step": 3302 }, { "epoch": 0.11, "learning_rate": 4.8894433914505546e-05, "loss": 0.545, "step": 3303 }, { "epoch": 0.11, "learning_rate": 4.889365080035913e-05, "loss": 0.9193, "step": 3304 }, { "epoch": 0.11, "learning_rate": 4.889286741523209e-05, "loss": 1.6086, "step": 3305 }, { "epoch": 0.11, "learning_rate": 4.889208375913332e-05, "loss": 1.3979, "step": 3306 }, { "epoch": 0.11, "learning_rate": 4.8891299832071704e-05, "loss": 0.9559, "step": 3307 }, { "epoch": 0.11, "learning_rate": 4.889051563405614e-05, "loss": 0.6155, "step": 3308 }, { "epoch": 0.11, "learning_rate": 4.888973116509551e-05, "loss": 0.5867, "step": 3309 }, { "epoch": 0.11, "learning_rate": 4.888894642519872e-05, "loss": 1.1748, "step": 3310 }, { "epoch": 0.11, "learning_rate": 4.888816141437467e-05, "loss": 1.0551, "step": 3311 }, { "epoch": 0.11, "learning_rate": 4.888737613263226e-05, "loss": 0.949, "step": 3312 }, { "epoch": 0.11, "learning_rate": 4.888659057998039e-05, "loss": 0.9925, "step": 3313 }, { "epoch": 0.11, "learning_rate": 4.888580475642798e-05, "loss": 0.9863, "step": 3314 }, { "epoch": 0.11, "learning_rate": 4.8885018661983936e-05, "loss": 0.8223, "step": 3315 }, { "epoch": 0.11, "learning_rate": 4.888423229665717e-05, "loss": 1.0095, "step": 3316 }, { "epoch": 0.11, "learning_rate": 4.888344566045662e-05, "loss": 0.8694, "step": 3317 }, { "epoch": 0.11, "learning_rate": 4.8882658753391175e-05, "loss": 0.907, "step": 3318 }, { "epoch": 0.11, "learning_rate": 4.8881871575469784e-05, "loss": 1.5624, "step": 3319 }, { "epoch": 0.11, "learning_rate": 4.888108412670136e-05, "loss": 1.2294, "step": 3320 }, { "epoch": 0.11, "learning_rate": 4.888029640709485e-05, "loss": 0.6926, "step": 3321 }, { "epoch": 0.11, "learning_rate": 4.887950841665917e-05, "loss": 0.8464, "step": 3322 }, { "epoch": 0.11, "learning_rate": 4.887872015540327e-05, "loss": 1.1371, "step": 3323 }, { "epoch": 0.11, "learning_rate": 4.887793162333608e-05, "loss": 0.8666, "step": 3324 }, { "epoch": 0.11, "learning_rate": 4.8877142820466546e-05, "loss": 0.9795, "step": 3325 }, { "epoch": 0.11, "learning_rate": 4.887635374680362e-05, "loss": 0.961, "step": 3326 }, { "epoch": 0.11, "learning_rate": 4.887556440235625e-05, "loss": 1.0295, "step": 3327 }, { "epoch": 0.11, "learning_rate": 4.887477478713337e-05, "loss": 0.9288, "step": 3328 }, { "epoch": 0.11, "learning_rate": 4.887398490114396e-05, "loss": 0.8729, "step": 3329 }, { "epoch": 0.11, "learning_rate": 4.8873194744396964e-05, "loss": 0.5927, "step": 3330 }, { "epoch": 0.11, "learning_rate": 4.887240431690135e-05, "loss": 0.7822, "step": 3331 }, { "epoch": 0.11, "learning_rate": 4.887161361866608e-05, "loss": 0.7973, "step": 3332 }, { "epoch": 0.11, "learning_rate": 4.887082264970011e-05, "loss": 0.5207, "step": 3333 }, { "epoch": 0.11, "learning_rate": 4.887003141001243e-05, "loss": 0.9494, "step": 3334 }, { "epoch": 0.11, "learning_rate": 4.8869239899612005e-05, "loss": 1.0438, "step": 3335 }, { "epoch": 0.11, "learning_rate": 4.886844811850781e-05, "loss": 1.0083, "step": 3336 }, { "epoch": 0.11, "learning_rate": 4.8867656066708824e-05, "loss": 0.794, "step": 3337 }, { "epoch": 0.11, "learning_rate": 4.8866863744224035e-05, "loss": 0.4157, "step": 3338 }, { "epoch": 0.11, "learning_rate": 4.886607115106242e-05, "loss": 0.0083, "step": 3339 }, { "epoch": 0.11, "learning_rate": 4.886527828723297e-05, "loss": 0.0323, "step": 3340 }, { "epoch": 0.11, "learning_rate": 4.886448515274469e-05, "loss": 0.0108, "step": 3341 }, { "epoch": 0.11, "learning_rate": 4.886369174760655e-05, "loss": 0.4113, "step": 3342 }, { "epoch": 0.11, "learning_rate": 4.886289807182758e-05, "loss": 0.9686, "step": 3343 }, { "epoch": 0.11, "learning_rate": 4.8862104125416765e-05, "loss": 0.9281, "step": 3344 }, { "epoch": 0.11, "learning_rate": 4.8861309908383094e-05, "loss": 0.9423, "step": 3345 }, { "epoch": 0.11, "learning_rate": 4.8860515420735595e-05, "loss": 1.3212, "step": 3346 }, { "epoch": 0.11, "learning_rate": 4.8859720662483276e-05, "loss": 0.8141, "step": 3347 }, { "epoch": 0.11, "learning_rate": 4.885892563363514e-05, "loss": 0.8063, "step": 3348 }, { "epoch": 0.11, "learning_rate": 4.8858130334200225e-05, "loss": 1.1077, "step": 3349 }, { "epoch": 0.11, "learning_rate": 4.8857334764187525e-05, "loss": 1.002, "step": 3350 }, { "epoch": 0.11, "learning_rate": 4.8856538923606074e-05, "loss": 0.9519, "step": 3351 }, { "epoch": 0.11, "learning_rate": 4.88557428124649e-05, "loss": 0.7621, "step": 3352 }, { "epoch": 0.11, "learning_rate": 4.885494643077302e-05, "loss": 0.9801, "step": 3353 }, { "epoch": 0.11, "learning_rate": 4.885414977853949e-05, "loss": 0.8198, "step": 3354 }, { "epoch": 0.11, "learning_rate": 4.8853352855773315e-05, "loss": 0.842, "step": 3355 }, { "epoch": 0.11, "learning_rate": 4.8852555662483554e-05, "loss": 0.7901, "step": 3356 }, { "epoch": 0.11, "learning_rate": 4.8851758198679246e-05, "loss": 0.5944, "step": 3357 }, { "epoch": 0.11, "learning_rate": 4.885096046436943e-05, "loss": 0.4471, "step": 3358 }, { "epoch": 0.11, "learning_rate": 4.885016245956315e-05, "loss": 0.9762, "step": 3359 }, { "epoch": 0.11, "learning_rate": 4.8849364184269465e-05, "loss": 0.9743, "step": 3360 }, { "epoch": 0.11, "learning_rate": 4.8848565638497416e-05, "loss": 0.8785, "step": 3361 }, { "epoch": 0.11, "learning_rate": 4.884776682225607e-05, "loss": 4.8193, "step": 3362 }, { "epoch": 0.11, "learning_rate": 4.884696773555449e-05, "loss": 0.953, "step": 3363 }, { "epoch": 0.11, "learning_rate": 4.884616837840172e-05, "loss": 0.8485, "step": 3364 }, { "epoch": 0.11, "learning_rate": 4.884536875080685e-05, "loss": 1.2547, "step": 3365 }, { "epoch": 0.11, "learning_rate": 4.8844568852778937e-05, "loss": 0.8717, "step": 3366 }, { "epoch": 0.11, "learning_rate": 4.884376868432704e-05, "loss": 0.8893, "step": 3367 }, { "epoch": 0.11, "learning_rate": 4.884296824546026e-05, "loss": 0.6592, "step": 3368 }, { "epoch": 0.11, "learning_rate": 4.884216753618765e-05, "loss": 0.8402, "step": 3369 }, { "epoch": 0.11, "learning_rate": 4.884136655651831e-05, "loss": 0.6448, "step": 3370 }, { "epoch": 0.11, "learning_rate": 4.88405653064613e-05, "loss": 0.8616, "step": 3371 }, { "epoch": 0.11, "learning_rate": 4.883976378602574e-05, "loss": 0.881, "step": 3372 }, { "epoch": 0.11, "learning_rate": 4.88389619952207e-05, "loss": 0.9066, "step": 3373 }, { "epoch": 0.11, "learning_rate": 4.883815993405527e-05, "loss": 0.8199, "step": 3374 }, { "epoch": 0.11, "learning_rate": 4.883735760253856e-05, "loss": 1.5413, "step": 3375 }, { "epoch": 0.11, "learning_rate": 4.883655500067966e-05, "loss": 0.5303, "step": 3376 }, { "epoch": 0.11, "learning_rate": 4.8835752128487674e-05, "loss": 0.805, "step": 3377 }, { "epoch": 0.11, "learning_rate": 4.883494898597171e-05, "loss": 0.7983, "step": 3378 }, { "epoch": 0.11, "learning_rate": 4.883414557314086e-05, "loss": 1.0368, "step": 3379 }, { "epoch": 0.11, "learning_rate": 4.883334189000426e-05, "loss": 1.584, "step": 3380 }, { "epoch": 0.11, "learning_rate": 4.883253793657102e-05, "loss": 0.9869, "step": 3381 }, { "epoch": 0.11, "learning_rate": 4.883173371285025e-05, "loss": 0.8753, "step": 3382 }, { "epoch": 0.11, "learning_rate": 4.883092921885106e-05, "loss": 0.8207, "step": 3383 }, { "epoch": 0.11, "learning_rate": 4.88301244545826e-05, "loss": 0.991, "step": 3384 }, { "epoch": 0.11, "learning_rate": 4.8829319420053975e-05, "loss": 0.9747, "step": 3385 }, { "epoch": 0.11, "learning_rate": 4.882851411527434e-05, "loss": 1.4289, "step": 3386 }, { "epoch": 0.11, "learning_rate": 4.8827708540252795e-05, "loss": 0.8667, "step": 3387 }, { "epoch": 0.11, "learning_rate": 4.88269026949985e-05, "loss": 1.0113, "step": 3388 }, { "epoch": 0.11, "learning_rate": 4.882609657952059e-05, "loss": 0.8944, "step": 3389 }, { "epoch": 0.11, "learning_rate": 4.88252901938282e-05, "loss": 0.82, "step": 3390 }, { "epoch": 0.11, "learning_rate": 4.882448353793048e-05, "loss": 1.1896, "step": 3391 }, { "epoch": 0.11, "learning_rate": 4.882367661183657e-05, "loss": 1.448, "step": 3392 }, { "epoch": 0.11, "learning_rate": 4.8822869415555636e-05, "loss": 1.0646, "step": 3393 }, { "epoch": 0.11, "learning_rate": 4.8822061949096834e-05, "loss": 0.8119, "step": 3394 }, { "epoch": 0.11, "learning_rate": 4.8821254212469304e-05, "loss": 1.0199, "step": 3395 }, { "epoch": 0.11, "learning_rate": 4.8820446205682216e-05, "loss": 0.8874, "step": 3396 }, { "epoch": 0.11, "learning_rate": 4.8819637928744736e-05, "loss": 0.9877, "step": 3397 }, { "epoch": 0.11, "learning_rate": 4.881882938166603e-05, "loss": 0.8131, "step": 3398 }, { "epoch": 0.11, "learning_rate": 4.881802056445527e-05, "loss": 0.3046, "step": 3399 }, { "epoch": 0.11, "learning_rate": 4.881721147712162e-05, "loss": 0.3933, "step": 3400 }, { "epoch": 0.11, "learning_rate": 4.881640211967426e-05, "loss": 0.7424, "step": 3401 }, { "epoch": 0.11, "learning_rate": 4.8815592492122373e-05, "loss": 0.9355, "step": 3402 }, { "epoch": 0.11, "learning_rate": 4.881478259447514e-05, "loss": 1.0364, "step": 3403 }, { "epoch": 0.11, "learning_rate": 4.881397242674174e-05, "loss": 0.9261, "step": 3404 }, { "epoch": 0.11, "learning_rate": 4.8813161988931365e-05, "loss": 0.6688, "step": 3405 }, { "epoch": 0.11, "learning_rate": 4.8812351281053216e-05, "loss": 0.8641, "step": 3406 }, { "epoch": 0.11, "learning_rate": 4.8811540303116474e-05, "loss": 0.7999, "step": 3407 }, { "epoch": 0.11, "learning_rate": 4.8810729055130335e-05, "loss": 0.8846, "step": 3408 }, { "epoch": 0.11, "learning_rate": 4.8809917537104014e-05, "loss": 1.1535, "step": 3409 }, { "epoch": 0.11, "learning_rate": 4.88091057490467e-05, "loss": 1.042, "step": 3410 }, { "epoch": 0.11, "learning_rate": 4.880829369096761e-05, "loss": 0.818, "step": 3411 }, { "epoch": 0.11, "learning_rate": 4.880748136287594e-05, "loss": 0.7729, "step": 3412 }, { "epoch": 0.11, "learning_rate": 4.880666876478092e-05, "loss": 0.883, "step": 3413 }, { "epoch": 0.11, "learning_rate": 4.8805855896691755e-05, "loss": 0.8294, "step": 3414 }, { "epoch": 0.11, "learning_rate": 4.880504275861766e-05, "loss": 1.1448, "step": 3415 }, { "epoch": 0.11, "learning_rate": 4.880422935056788e-05, "loss": 0.8699, "step": 3416 }, { "epoch": 0.11, "learning_rate": 4.88034156725516e-05, "loss": 0.9296, "step": 3417 }, { "epoch": 0.11, "learning_rate": 4.880260172457809e-05, "loss": 0.6416, "step": 3418 }, { "epoch": 0.11, "learning_rate": 4.8801787506656554e-05, "loss": 0.927, "step": 3419 }, { "epoch": 0.11, "learning_rate": 4.880097301879624e-05, "loss": 0.9226, "step": 3420 }, { "epoch": 0.11, "learning_rate": 4.8800158261006365e-05, "loss": 0.9872, "step": 3421 }, { "epoch": 0.11, "learning_rate": 4.87993432332962e-05, "loss": 1.4817, "step": 3422 }, { "epoch": 0.11, "learning_rate": 4.879852793567496e-05, "loss": 5.8175, "step": 3423 }, { "epoch": 0.11, "learning_rate": 4.8797712368151915e-05, "loss": 2.8297, "step": 3424 }, { "epoch": 0.11, "learning_rate": 4.87968965307363e-05, "loss": 1.0686, "step": 3425 }, { "epoch": 0.11, "learning_rate": 4.8796080423437365e-05, "loss": 0.6869, "step": 3426 }, { "epoch": 0.11, "learning_rate": 4.8795264046264375e-05, "loss": 0.8479, "step": 3427 }, { "epoch": 0.11, "learning_rate": 4.879444739922658e-05, "loss": 0.6928, "step": 3428 }, { "epoch": 0.11, "learning_rate": 4.879363048233325e-05, "loss": 0.9204, "step": 3429 }, { "epoch": 0.11, "learning_rate": 4.879281329559365e-05, "loss": 0.5546, "step": 3430 }, { "epoch": 0.11, "learning_rate": 4.879199583901704e-05, "loss": 0.5309, "step": 3431 }, { "epoch": 0.11, "learning_rate": 4.879117811261269e-05, "loss": 0.7489, "step": 3432 }, { "epoch": 0.11, "learning_rate": 4.8790360116389886e-05, "loss": 1.7949, "step": 3433 }, { "epoch": 0.11, "learning_rate": 4.878954185035789e-05, "loss": 0.8138, "step": 3434 }, { "epoch": 0.11, "learning_rate": 4.8788723314526e-05, "loss": 0.9887, "step": 3435 }, { "epoch": 0.11, "learning_rate": 4.878790450890348e-05, "loss": 0.7812, "step": 3436 }, { "epoch": 0.11, "learning_rate": 4.878708543349963e-05, "loss": 1.0275, "step": 3437 }, { "epoch": 0.11, "learning_rate": 4.878626608832373e-05, "loss": 0.972, "step": 3438 }, { "epoch": 0.11, "learning_rate": 4.878544647338508e-05, "loss": 0.8362, "step": 3439 }, { "epoch": 0.11, "learning_rate": 4.878462658869297e-05, "loss": 1.0399, "step": 3440 }, { "epoch": 0.11, "learning_rate": 4.8783806434256704e-05, "loss": 0.5856, "step": 3441 }, { "epoch": 0.11, "learning_rate": 4.878298601008557e-05, "loss": 0.7694, "step": 3442 }, { "epoch": 0.11, "learning_rate": 4.8782165316188895e-05, "loss": 0.8923, "step": 3443 }, { "epoch": 0.11, "learning_rate": 4.878134435257597e-05, "loss": 1.0586, "step": 3444 }, { "epoch": 0.11, "learning_rate": 4.878052311925611e-05, "loss": 0.9553, "step": 3445 }, { "epoch": 0.11, "learning_rate": 4.877970161623861e-05, "loss": 0.6625, "step": 3446 }, { "epoch": 0.11, "learning_rate": 4.8778879843532835e-05, "loss": 0.8026, "step": 3447 }, { "epoch": 0.11, "learning_rate": 4.877805780114805e-05, "loss": 1.1574, "step": 3448 }, { "epoch": 0.11, "learning_rate": 4.8777235489093616e-05, "loss": 0.5033, "step": 3449 }, { "epoch": 0.12, "learning_rate": 4.877641290737884e-05, "loss": 0.788, "step": 3450 }, { "epoch": 0.12, "learning_rate": 4.877559005601306e-05, "loss": 0.8596, "step": 3451 }, { "epoch": 0.12, "learning_rate": 4.87747669350056e-05, "loss": 0.9611, "step": 3452 }, { "epoch": 0.12, "learning_rate": 4.877394354436581e-05, "loss": 0.9375, "step": 3453 }, { "epoch": 0.12, "learning_rate": 4.8773119884103006e-05, "loss": 0.8549, "step": 3454 }, { "epoch": 0.12, "learning_rate": 4.8772295954226546e-05, "loss": 0.8627, "step": 3455 }, { "epoch": 0.12, "learning_rate": 4.8771471754745766e-05, "loss": 0.9809, "step": 3456 }, { "epoch": 0.12, "learning_rate": 4.8770647285670026e-05, "loss": 0.9699, "step": 3457 }, { "epoch": 0.12, "learning_rate": 4.876982254700866e-05, "loss": 0.6806, "step": 3458 }, { "epoch": 0.12, "learning_rate": 4.8768997538771026e-05, "loss": 0.885, "step": 3459 }, { "epoch": 0.12, "learning_rate": 4.876817226096649e-05, "loss": 2.0687, "step": 3460 }, { "epoch": 0.12, "learning_rate": 4.87673467136044e-05, "loss": 0.9587, "step": 3461 }, { "epoch": 0.12, "learning_rate": 4.8766520896694126e-05, "loss": 0.9811, "step": 3462 }, { "epoch": 0.12, "learning_rate": 4.876569481024503e-05, "loss": 0.808, "step": 3463 }, { "epoch": 0.12, "learning_rate": 4.876486845426648e-05, "loss": 1.1773, "step": 3464 }, { "epoch": 0.12, "learning_rate": 4.876404182876785e-05, "loss": 0.7668, "step": 3465 }, { "epoch": 0.12, "learning_rate": 4.8763214933758515e-05, "loss": 0.8485, "step": 3466 }, { "epoch": 0.12, "learning_rate": 4.8762387769247855e-05, "loss": 0.7735, "step": 3467 }, { "epoch": 0.12, "learning_rate": 4.876156033524524e-05, "loss": 1.0194, "step": 3468 }, { "epoch": 0.12, "learning_rate": 4.876073263176007e-05, "loss": 0.7611, "step": 3469 }, { "epoch": 0.12, "learning_rate": 4.8759904658801726e-05, "loss": 1.125, "step": 3470 }, { "epoch": 0.12, "learning_rate": 4.875907641637959e-05, "loss": 1.1295, "step": 3471 }, { "epoch": 0.12, "learning_rate": 4.875824790450306e-05, "loss": 0.8283, "step": 3472 }, { "epoch": 0.12, "learning_rate": 4.8757419123181535e-05, "loss": 0.8344, "step": 3473 }, { "epoch": 0.12, "learning_rate": 4.875659007242443e-05, "loss": 0.7779, "step": 3474 }, { "epoch": 0.12, "learning_rate": 4.8755760752241106e-05, "loss": 0.8473, "step": 3475 }, { "epoch": 0.12, "learning_rate": 4.8754931162641006e-05, "loss": 0.8739, "step": 3476 }, { "epoch": 0.12, "learning_rate": 4.8754101303633526e-05, "loss": 0.9661, "step": 3477 }, { "epoch": 0.12, "learning_rate": 4.875327117522807e-05, "loss": 2.7039, "step": 3478 }, { "epoch": 0.12, "learning_rate": 4.875244077743406e-05, "loss": 1.1856, "step": 3479 }, { "epoch": 0.12, "learning_rate": 4.875161011026092e-05, "loss": 0.9262, "step": 3480 }, { "epoch": 0.12, "learning_rate": 4.8750779173718065e-05, "loss": 0.9219, "step": 3481 }, { "epoch": 0.12, "learning_rate": 4.874994796781491e-05, "loss": 2.0802, "step": 3482 }, { "epoch": 0.12, "learning_rate": 4.8749116492560895e-05, "loss": 2.0076, "step": 3483 }, { "epoch": 0.12, "learning_rate": 4.874828474796544e-05, "loss": 0.6156, "step": 3484 }, { "epoch": 0.12, "learning_rate": 4.874745273403798e-05, "loss": 0.8308, "step": 3485 }, { "epoch": 0.12, "learning_rate": 4.874662045078796e-05, "loss": 1.0148, "step": 3486 }, { "epoch": 0.12, "learning_rate": 4.874578789822481e-05, "loss": 0.8769, "step": 3487 }, { "epoch": 0.12, "learning_rate": 4.874495507635797e-05, "loss": 0.9672, "step": 3488 }, { "epoch": 0.12, "learning_rate": 4.87441219851969e-05, "loss": 1.0707, "step": 3489 }, { "epoch": 0.12, "learning_rate": 4.874328862475103e-05, "loss": 1.0274, "step": 3490 }, { "epoch": 0.12, "learning_rate": 4.874245499502982e-05, "loss": 1.5361, "step": 3491 }, { "epoch": 0.12, "learning_rate": 4.874162109604272e-05, "loss": 1.3555, "step": 3492 }, { "epoch": 0.12, "learning_rate": 4.874078692779919e-05, "loss": 0.9091, "step": 3493 }, { "epoch": 0.12, "learning_rate": 4.8739952490308694e-05, "loss": 0.8827, "step": 3494 }, { "epoch": 0.12, "learning_rate": 4.8739117783580695e-05, "loss": 1.0816, "step": 3495 }, { "epoch": 0.12, "learning_rate": 4.8738282807624656e-05, "loss": 0.9176, "step": 3496 }, { "epoch": 0.12, "learning_rate": 4.8737447562450044e-05, "loss": 0.8086, "step": 3497 }, { "epoch": 0.12, "learning_rate": 4.873661204806633e-05, "loss": 4.4961, "step": 3498 }, { "epoch": 0.12, "learning_rate": 4.8735776264483005e-05, "loss": 0.7988, "step": 3499 }, { "epoch": 0.12, "learning_rate": 4.873494021170953e-05, "loss": 0.8863, "step": 3500 }, { "epoch": 0.12, "learning_rate": 4.87341038897554e-05, "loss": 0.7175, "step": 3501 }, { "epoch": 0.12, "learning_rate": 4.873326729863009e-05, "loss": 0.1114, "step": 3502 }, { "epoch": 0.12, "learning_rate": 4.87324304383431e-05, "loss": 0.1226, "step": 3503 }, { "epoch": 0.12, "learning_rate": 4.87315933089039e-05, "loss": 0.0981, "step": 3504 }, { "epoch": 0.12, "learning_rate": 4.8730755910322004e-05, "loss": 0.4886, "step": 3505 }, { "epoch": 0.12, "learning_rate": 4.87299182426069e-05, "loss": 1.2213, "step": 3506 }, { "epoch": 0.12, "learning_rate": 4.872908030576809e-05, "loss": 0.9784, "step": 3507 }, { "epoch": 0.12, "learning_rate": 4.872824209981508e-05, "loss": 1.0497, "step": 3508 }, { "epoch": 0.12, "learning_rate": 4.8727403624757365e-05, "loss": 0.773, "step": 3509 }, { "epoch": 0.12, "learning_rate": 4.872656488060447e-05, "loss": 0.8598, "step": 3510 }, { "epoch": 0.12, "learning_rate": 4.8725725867365894e-05, "loss": 0.5777, "step": 3511 }, { "epoch": 0.12, "learning_rate": 4.8724886585051166e-05, "loss": 1.0659, "step": 3512 }, { "epoch": 0.12, "learning_rate": 4.872404703366979e-05, "loss": 0.9535, "step": 3513 }, { "epoch": 0.12, "learning_rate": 4.8723207213231295e-05, "loss": 0.7844, "step": 3514 }, { "epoch": 0.12, "learning_rate": 4.872236712374521e-05, "loss": 1.4692, "step": 3515 }, { "epoch": 0.12, "learning_rate": 4.8721526765221046e-05, "loss": 0.7654, "step": 3516 }, { "epoch": 0.12, "learning_rate": 4.872068613766835e-05, "loss": 0.8553, "step": 3517 }, { "epoch": 0.12, "learning_rate": 4.871984524109665e-05, "loss": 0.8379, "step": 3518 }, { "epoch": 0.12, "learning_rate": 4.8719004075515484e-05, "loss": 0.8082, "step": 3519 }, { "epoch": 0.12, "learning_rate": 4.871816264093439e-05, "loss": 0.8556, "step": 3520 }, { "epoch": 0.12, "learning_rate": 4.871732093736291e-05, "loss": 0.8569, "step": 3521 }, { "epoch": 0.12, "learning_rate": 4.8716478964810595e-05, "loss": 0.9284, "step": 3522 }, { "epoch": 0.12, "learning_rate": 4.871563672328699e-05, "loss": 0.6723, "step": 3523 }, { "epoch": 0.12, "learning_rate": 4.871479421280164e-05, "loss": 1.0093, "step": 3524 }, { "epoch": 0.12, "learning_rate": 4.871395143336411e-05, "loss": 1.2117, "step": 3525 }, { "epoch": 0.12, "learning_rate": 4.8713108384983964e-05, "loss": 0.8468, "step": 3526 }, { "epoch": 0.12, "learning_rate": 4.8712265067670745e-05, "loss": 1.0486, "step": 3527 }, { "epoch": 0.12, "learning_rate": 4.8711421481434025e-05, "loss": 0.7816, "step": 3528 }, { "epoch": 0.12, "learning_rate": 4.871057762628338e-05, "loss": 0.7499, "step": 3529 }, { "epoch": 0.12, "learning_rate": 4.8709733502228375e-05, "loss": 0.9622, "step": 3530 }, { "epoch": 0.12, "learning_rate": 4.8708889109278574e-05, "loss": 0.9548, "step": 3531 }, { "epoch": 0.12, "learning_rate": 4.870804444744356e-05, "loss": 0.917, "step": 3532 }, { "epoch": 0.12, "learning_rate": 4.8707199516732914e-05, "loss": 1.0365, "step": 3533 }, { "epoch": 0.12, "learning_rate": 4.870635431715622e-05, "loss": 0.8878, "step": 3534 }, { "epoch": 0.12, "learning_rate": 4.870550884872306e-05, "loss": 0.9214, "step": 3535 }, { "epoch": 0.12, "learning_rate": 4.870466311144303e-05, "loss": 0.2678, "step": 3536 }, { "epoch": 0.12, "learning_rate": 4.870381710532571e-05, "loss": 0.4671, "step": 3537 }, { "epoch": 0.12, "learning_rate": 4.870297083038069e-05, "loss": 1.1748, "step": 3538 }, { "epoch": 0.12, "learning_rate": 4.870212428661759e-05, "loss": 0.8615, "step": 3539 }, { "epoch": 0.12, "learning_rate": 4.870127747404599e-05, "loss": 1.525, "step": 3540 }, { "epoch": 0.12, "learning_rate": 4.870043039267551e-05, "loss": 0.9943, "step": 3541 }, { "epoch": 0.12, "learning_rate": 4.8699583042515745e-05, "loss": 0.9611, "step": 3542 }, { "epoch": 0.12, "learning_rate": 4.8698735423576306e-05, "loss": 0.8575, "step": 3543 }, { "epoch": 0.12, "learning_rate": 4.869788753586681e-05, "loss": 0.9014, "step": 3544 }, { "epoch": 0.12, "learning_rate": 4.8697039379396884e-05, "loss": 0.7017, "step": 3545 }, { "epoch": 0.12, "learning_rate": 4.869619095417612e-05, "loss": 0.8848, "step": 3546 }, { "epoch": 0.12, "learning_rate": 4.869534226021416e-05, "loss": 0.9302, "step": 3547 }, { "epoch": 0.12, "learning_rate": 4.869449329752062e-05, "loss": 0.9003, "step": 3548 }, { "epoch": 0.12, "learning_rate": 4.869364406610514e-05, "loss": 1.0494, "step": 3549 }, { "epoch": 0.12, "learning_rate": 4.869279456597733e-05, "loss": 0.8978, "step": 3550 }, { "epoch": 0.12, "learning_rate": 4.869194479714685e-05, "loss": 0.9165, "step": 3551 }, { "epoch": 0.12, "learning_rate": 4.8691094759623324e-05, "loss": 0.8156, "step": 3552 }, { "epoch": 0.12, "learning_rate": 4.869024445341639e-05, "loss": 0.9842, "step": 3553 }, { "epoch": 0.12, "learning_rate": 4.868939387853569e-05, "loss": 1.1496, "step": 3554 }, { "epoch": 0.12, "learning_rate": 4.868854303499089e-05, "loss": 0.9366, "step": 3555 }, { "epoch": 0.12, "learning_rate": 4.8687691922791615e-05, "loss": 0.876, "step": 3556 }, { "epoch": 0.12, "learning_rate": 4.868684054194752e-05, "loss": 0.7839, "step": 3557 }, { "epoch": 0.12, "learning_rate": 4.8685988892468284e-05, "loss": 0.6438, "step": 3558 }, { "epoch": 0.12, "learning_rate": 4.868513697436354e-05, "loss": 0.6103, "step": 3559 }, { "epoch": 0.12, "learning_rate": 4.8684284787642955e-05, "loss": 0.6652, "step": 3560 }, { "epoch": 0.12, "learning_rate": 4.86834323323162e-05, "loss": 0.5773, "step": 3561 }, { "epoch": 0.12, "learning_rate": 4.868257960839295e-05, "loss": 0.5176, "step": 3562 }, { "epoch": 0.12, "learning_rate": 4.868172661588286e-05, "loss": 0.4467, "step": 3563 }, { "epoch": 0.12, "learning_rate": 4.8680873354795616e-05, "loss": 0.7543, "step": 3564 }, { "epoch": 0.12, "learning_rate": 4.868001982514088e-05, "loss": 0.8412, "step": 3565 }, { "epoch": 0.12, "learning_rate": 4.867916602692835e-05, "loss": 0.7115, "step": 3566 }, { "epoch": 0.12, "learning_rate": 4.86783119601677e-05, "loss": 2.6219, "step": 3567 }, { "epoch": 0.12, "learning_rate": 4.867745762486861e-05, "loss": 0.9031, "step": 3568 }, { "epoch": 0.12, "learning_rate": 4.867660302104078e-05, "loss": 0.9652, "step": 3569 }, { "epoch": 0.12, "learning_rate": 4.867574814869389e-05, "loss": 0.1686, "step": 3570 }, { "epoch": 0.12, "learning_rate": 4.8674893007837655e-05, "loss": 0.8701, "step": 3571 }, { "epoch": 0.12, "learning_rate": 4.867403759848176e-05, "loss": 0.7649, "step": 3572 }, { "epoch": 0.12, "learning_rate": 4.8673181920635904e-05, "loss": 0.8314, "step": 3573 }, { "epoch": 0.12, "learning_rate": 4.86723259743098e-05, "loss": 0.9737, "step": 3574 }, { "epoch": 0.12, "learning_rate": 4.867146975951314e-05, "loss": 0.7238, "step": 3575 }, { "epoch": 0.12, "learning_rate": 4.867061327625565e-05, "loss": 0.9438, "step": 3576 }, { "epoch": 0.12, "learning_rate": 4.866975652454704e-05, "loss": 1.1198, "step": 3577 }, { "epoch": 0.12, "learning_rate": 4.8668899504397015e-05, "loss": 0.7638, "step": 3578 }, { "epoch": 0.12, "learning_rate": 4.866804221581531e-05, "loss": 0.9101, "step": 3579 }, { "epoch": 0.12, "learning_rate": 4.8667184658811645e-05, "loss": 0.861, "step": 3580 }, { "epoch": 0.12, "learning_rate": 4.866632683339575e-05, "loss": 0.4812, "step": 3581 }, { "epoch": 0.12, "learning_rate": 4.8665468739577334e-05, "loss": 0.949, "step": 3582 }, { "epoch": 0.12, "learning_rate": 4.866461037736614e-05, "loss": 0.9332, "step": 3583 }, { "epoch": 0.12, "learning_rate": 4.866375174677191e-05, "loss": 1.027, "step": 3584 }, { "epoch": 0.12, "learning_rate": 4.866289284780437e-05, "loss": 0.7449, "step": 3585 }, { "epoch": 0.12, "learning_rate": 4.8662033680473276e-05, "loss": 0.9655, "step": 3586 }, { "epoch": 0.12, "learning_rate": 4.866117424478836e-05, "loss": 1.0311, "step": 3587 }, { "epoch": 0.12, "learning_rate": 4.866031454075937e-05, "loss": 0.9361, "step": 3588 }, { "epoch": 0.12, "learning_rate": 4.865945456839605e-05, "loss": 0.8468, "step": 3589 }, { "epoch": 0.12, "learning_rate": 4.865859432770817e-05, "loss": 0.8623, "step": 3590 }, { "epoch": 0.12, "learning_rate": 4.865773381870547e-05, "loss": 0.9004, "step": 3591 }, { "epoch": 0.12, "learning_rate": 4.865687304139772e-05, "loss": 0.8029, "step": 3592 }, { "epoch": 0.12, "learning_rate": 4.8656011995794674e-05, "loss": 0.9292, "step": 3593 }, { "epoch": 0.12, "learning_rate": 4.86551506819061e-05, "loss": 0.9704, "step": 3594 }, { "epoch": 0.12, "learning_rate": 4.865428909974177e-05, "loss": 0.9323, "step": 3595 }, { "epoch": 0.12, "learning_rate": 4.8653427249311454e-05, "loss": 1.1049, "step": 3596 }, { "epoch": 0.12, "learning_rate": 4.865256513062493e-05, "loss": 0.8285, "step": 3597 }, { "epoch": 0.12, "learning_rate": 4.8651702743691966e-05, "loss": 0.9572, "step": 3598 }, { "epoch": 0.12, "learning_rate": 4.865084008852234e-05, "loss": 0.8337, "step": 3599 }, { "epoch": 0.12, "learning_rate": 4.864997716512584e-05, "loss": 0.8468, "step": 3600 }, { "epoch": 0.12, "learning_rate": 4.864911397351226e-05, "loss": 0.1752, "step": 3601 }, { "epoch": 0.12, "learning_rate": 4.864825051369139e-05, "loss": 0.8671, "step": 3602 }, { "epoch": 0.12, "learning_rate": 4.864738678567301e-05, "loss": 1.0556, "step": 3603 }, { "epoch": 0.12, "learning_rate": 4.864652278946693e-05, "loss": 1.1522, "step": 3604 }, { "epoch": 0.12, "learning_rate": 4.864565852508294e-05, "loss": 1.0331, "step": 3605 }, { "epoch": 0.12, "learning_rate": 4.864479399253083e-05, "loss": 1.0195, "step": 3606 }, { "epoch": 0.12, "learning_rate": 4.8643929191820426e-05, "loss": 0.9206, "step": 3607 }, { "epoch": 0.12, "learning_rate": 4.8643064122961533e-05, "loss": 1.172, "step": 3608 }, { "epoch": 0.12, "learning_rate": 4.8642198785963946e-05, "loss": 2.5369, "step": 3609 }, { "epoch": 0.12, "learning_rate": 4.86413331808375e-05, "loss": 2.2544, "step": 3610 }, { "epoch": 0.12, "learning_rate": 4.8640467307592e-05, "loss": 2.4215, "step": 3611 }, { "epoch": 0.12, "learning_rate": 4.863960116623726e-05, "loss": 1.8118, "step": 3612 }, { "epoch": 0.12, "learning_rate": 4.863873475678311e-05, "loss": 0.5406, "step": 3613 }, { "epoch": 0.12, "learning_rate": 4.863786807923938e-05, "loss": 0.7555, "step": 3614 }, { "epoch": 0.12, "learning_rate": 4.863700113361589e-05, "loss": 0.8199, "step": 3615 }, { "epoch": 0.12, "learning_rate": 4.863613391992248e-05, "loss": 0.9835, "step": 3616 }, { "epoch": 0.12, "learning_rate": 4.8635266438168984e-05, "loss": 0.9435, "step": 3617 }, { "epoch": 0.12, "learning_rate": 4.863439868836524e-05, "loss": 0.7163, "step": 3618 }, { "epoch": 0.12, "learning_rate": 4.863353067052109e-05, "loss": 1.2812, "step": 3619 }, { "epoch": 0.12, "learning_rate": 4.863266238464637e-05, "loss": 1.0453, "step": 3620 }, { "epoch": 0.12, "learning_rate": 4.863179383075094e-05, "loss": 1.485, "step": 3621 }, { "epoch": 0.12, "learning_rate": 4.8630925008844634e-05, "loss": 0.5423, "step": 3622 }, { "epoch": 0.12, "learning_rate": 4.863005591893732e-05, "loss": 0.8369, "step": 3623 }, { "epoch": 0.12, "learning_rate": 4.862918656103885e-05, "loss": 1.3674, "step": 3624 }, { "epoch": 0.12, "learning_rate": 4.862831693515909e-05, "loss": 1.481, "step": 3625 }, { "epoch": 0.12, "learning_rate": 4.8627447041307896e-05, "loss": 1.5993, "step": 3626 }, { "epoch": 0.12, "learning_rate": 4.862657687949512e-05, "loss": 1.6464, "step": 3627 }, { "epoch": 0.12, "learning_rate": 4.862570644973066e-05, "loss": 1.4144, "step": 3628 }, { "epoch": 0.12, "learning_rate": 4.8624835752024356e-05, "loss": 2.8441, "step": 3629 }, { "epoch": 0.12, "learning_rate": 4.862396478638611e-05, "loss": 0.864, "step": 3630 }, { "epoch": 0.12, "learning_rate": 4.862309355282578e-05, "loss": 0.9116, "step": 3631 }, { "epoch": 0.12, "learning_rate": 4.862222205135326e-05, "loss": 0.7983, "step": 3632 }, { "epoch": 0.12, "learning_rate": 4.8621350281978425e-05, "loss": 0.6035, "step": 3633 }, { "epoch": 0.12, "learning_rate": 4.862047824471117e-05, "loss": 0.3228, "step": 3634 }, { "epoch": 0.12, "learning_rate": 4.861960593956138e-05, "loss": 0.9727, "step": 3635 }, { "epoch": 0.12, "learning_rate": 4.861873336653895e-05, "loss": 0.7548, "step": 3636 }, { "epoch": 0.12, "learning_rate": 4.861786052565377e-05, "loss": 0.865, "step": 3637 }, { "epoch": 0.12, "learning_rate": 4.861698741691574e-05, "loss": 0.8133, "step": 3638 }, { "epoch": 0.12, "learning_rate": 4.861611404033477e-05, "loss": 0.9719, "step": 3639 }, { "epoch": 0.12, "learning_rate": 4.861524039592076e-05, "loss": 0.7815, "step": 3640 }, { "epoch": 0.12, "learning_rate": 4.8614366483683616e-05, "loss": 0.5868, "step": 3641 }, { "epoch": 0.12, "learning_rate": 4.8613492303633254e-05, "loss": 0.8139, "step": 3642 }, { "epoch": 0.12, "learning_rate": 4.861261785577959e-05, "loss": 0.8485, "step": 3643 }, { "epoch": 0.12, "learning_rate": 4.8611743140132526e-05, "loss": 0.9446, "step": 3644 }, { "epoch": 0.12, "learning_rate": 4.8610868156701995e-05, "loss": 0.8886, "step": 3645 }, { "epoch": 0.12, "learning_rate": 4.860999290549793e-05, "loss": 0.773, "step": 3646 }, { "epoch": 0.12, "learning_rate": 4.860911738653023e-05, "loss": 0.742, "step": 3647 }, { "epoch": 0.12, "learning_rate": 4.860824159980884e-05, "loss": 0.9361, "step": 3648 }, { "epoch": 0.12, "learning_rate": 4.860736554534371e-05, "loss": 0.997, "step": 3649 }, { "epoch": 0.12, "learning_rate": 4.8606489223144744e-05, "loss": 0.9511, "step": 3650 }, { "epoch": 0.12, "learning_rate": 4.8605612633221896e-05, "loss": 0.9533, "step": 3651 }, { "epoch": 0.12, "learning_rate": 4.8604735775585117e-05, "loss": 0.8606, "step": 3652 }, { "epoch": 0.12, "learning_rate": 4.8603858650244326e-05, "loss": 1.3301, "step": 3653 }, { "epoch": 0.12, "learning_rate": 4.860298125720949e-05, "loss": 1.4585, "step": 3654 }, { "epoch": 0.12, "learning_rate": 4.860210359649056e-05, "loss": 0.8882, "step": 3655 }, { "epoch": 0.12, "learning_rate": 4.860122566809748e-05, "loss": 0.9347, "step": 3656 }, { "epoch": 0.12, "learning_rate": 4.860034747204021e-05, "loss": 0.6392, "step": 3657 }, { "epoch": 0.12, "learning_rate": 4.8599469008328715e-05, "loss": 0.8535, "step": 3658 }, { "epoch": 0.12, "learning_rate": 4.859859027697294e-05, "loss": 1.0082, "step": 3659 }, { "epoch": 0.12, "learning_rate": 4.8597711277982885e-05, "loss": 0.9451, "step": 3660 }, { "epoch": 0.12, "learning_rate": 4.859683201136848e-05, "loss": 1.0458, "step": 3661 }, { "epoch": 0.12, "learning_rate": 4.8595952477139726e-05, "loss": 1.0239, "step": 3662 }, { "epoch": 0.12, "learning_rate": 4.859507267530659e-05, "loss": 0.7141, "step": 3663 }, { "epoch": 0.12, "learning_rate": 4.859419260587903e-05, "loss": 0.7117, "step": 3664 }, { "epoch": 0.12, "learning_rate": 4.8593312268867055e-05, "loss": 2.1788, "step": 3665 }, { "epoch": 0.12, "learning_rate": 4.859243166428063e-05, "loss": 3.646, "step": 3666 }, { "epoch": 0.12, "learning_rate": 4.8591550792129756e-05, "loss": 0.9128, "step": 3667 }, { "epoch": 0.12, "learning_rate": 4.859066965242443e-05, "loss": 2.5021, "step": 3668 }, { "epoch": 0.12, "learning_rate": 4.858978824517461e-05, "loss": 2.1594, "step": 3669 }, { "epoch": 0.12, "learning_rate": 4.858890657039033e-05, "loss": 0.6664, "step": 3670 }, { "epoch": 0.12, "learning_rate": 4.8588024628081566e-05, "loss": 1.0146, "step": 3671 }, { "epoch": 0.12, "learning_rate": 4.8587142418258326e-05, "loss": 0.8334, "step": 3672 }, { "epoch": 0.12, "learning_rate": 4.858625994093061e-05, "loss": 0.7169, "step": 3673 }, { "epoch": 0.12, "learning_rate": 4.8585377196108444e-05, "loss": 0.8213, "step": 3674 }, { "epoch": 0.12, "learning_rate": 4.8584494183801825e-05, "loss": 0.773, "step": 3675 }, { "epoch": 0.12, "learning_rate": 4.858361090402077e-05, "loss": 0.873, "step": 3676 }, { "epoch": 0.12, "learning_rate": 4.8582727356775304e-05, "loss": 0.9129, "step": 3677 }, { "epoch": 0.12, "learning_rate": 4.858184354207543e-05, "loss": 0.9228, "step": 3678 }, { "epoch": 0.12, "learning_rate": 4.858095945993119e-05, "loss": 0.8912, "step": 3679 }, { "epoch": 0.12, "learning_rate": 4.8580075110352595e-05, "loss": 0.6585, "step": 3680 }, { "epoch": 0.12, "learning_rate": 4.857919049334969e-05, "loss": 1.1379, "step": 3681 }, { "epoch": 0.12, "learning_rate": 4.8578305608932496e-05, "loss": 1.1072, "step": 3682 }, { "epoch": 0.12, "learning_rate": 4.857742045711104e-05, "loss": 0.8921, "step": 3683 }, { "epoch": 0.12, "learning_rate": 4.8576535037895394e-05, "loss": 0.8275, "step": 3684 }, { "epoch": 0.12, "learning_rate": 4.857564935129557e-05, "loss": 1.2631, "step": 3685 }, { "epoch": 0.12, "learning_rate": 4.8574763397321614e-05, "loss": 1.2042, "step": 3686 }, { "epoch": 0.12, "learning_rate": 4.8573877175983586e-05, "loss": 0.7937, "step": 3687 }, { "epoch": 0.12, "learning_rate": 4.857299068729153e-05, "loss": 0.7628, "step": 3688 }, { "epoch": 0.12, "learning_rate": 4.8572103931255504e-05, "loss": 0.195, "step": 3689 }, { "epoch": 0.12, "learning_rate": 4.857121690788556e-05, "loss": 0.1786, "step": 3690 }, { "epoch": 0.12, "learning_rate": 4.8570329617191754e-05, "loss": 0.1692, "step": 3691 }, { "epoch": 0.12, "learning_rate": 4.856944205918417e-05, "loss": 0.1613, "step": 3692 }, { "epoch": 0.12, "learning_rate": 4.8568554233872846e-05, "loss": 0.8475, "step": 3693 }, { "epoch": 0.12, "learning_rate": 4.856766614126786e-05, "loss": 0.7177, "step": 3694 }, { "epoch": 0.12, "learning_rate": 4.856677778137929e-05, "loss": 1.1627, "step": 3695 }, { "epoch": 0.12, "learning_rate": 4.8565889154217214e-05, "loss": 1.2511, "step": 3696 }, { "epoch": 0.12, "learning_rate": 4.856500025979171e-05, "loss": 0.9446, "step": 3697 }, { "epoch": 0.12, "learning_rate": 4.8564111098112844e-05, "loss": 0.9275, "step": 3698 }, { "epoch": 0.12, "learning_rate": 4.85632216691907e-05, "loss": 0.583, "step": 3699 }, { "epoch": 0.12, "learning_rate": 4.856233197303539e-05, "loss": 0.5783, "step": 3700 }, { "epoch": 0.12, "learning_rate": 4.856144200965699e-05, "loss": 1.9908, "step": 3701 }, { "epoch": 0.12, "learning_rate": 4.8560551779065583e-05, "loss": 0.9608, "step": 3702 }, { "epoch": 0.12, "learning_rate": 4.855966128127128e-05, "loss": 0.8884, "step": 3703 }, { "epoch": 0.12, "learning_rate": 4.855877051628417e-05, "loss": 1.154, "step": 3704 }, { "epoch": 0.12, "learning_rate": 4.855787948411436e-05, "loss": 0.7808, "step": 3705 }, { "epoch": 0.12, "learning_rate": 4.8556988184771953e-05, "loss": 0.749, "step": 3706 }, { "epoch": 0.12, "learning_rate": 4.8556096618267065e-05, "loss": 0.3039, "step": 3707 }, { "epoch": 0.12, "learning_rate": 4.85552047846098e-05, "loss": 0.7391, "step": 3708 }, { "epoch": 0.12, "learning_rate": 4.855431268381028e-05, "loss": 1.3029, "step": 3709 }, { "epoch": 0.12, "learning_rate": 4.8553420315878606e-05, "loss": 0.9534, "step": 3710 }, { "epoch": 0.12, "learning_rate": 4.855252768082492e-05, "loss": 1.1214, "step": 3711 }, { "epoch": 0.12, "learning_rate": 4.855163477865932e-05, "loss": 0.9503, "step": 3712 }, { "epoch": 0.12, "learning_rate": 4.855074160939196e-05, "loss": 0.9007, "step": 3713 }, { "epoch": 0.12, "learning_rate": 4.8549848173032946e-05, "loss": 0.5417, "step": 3714 }, { "epoch": 0.12, "learning_rate": 4.854895446959243e-05, "loss": 0.955, "step": 3715 }, { "epoch": 0.12, "learning_rate": 4.8548060499080534e-05, "loss": 0.9236, "step": 3716 }, { "epoch": 0.12, "learning_rate": 4.85471662615074e-05, "loss": 0.8725, "step": 3717 }, { "epoch": 0.12, "learning_rate": 4.854627175688317e-05, "loss": 2.4386, "step": 3718 }, { "epoch": 0.12, "learning_rate": 4.8545376985217996e-05, "loss": 2.642, "step": 3719 }, { "epoch": 0.12, "learning_rate": 4.854448194652201e-05, "loss": 0.7923, "step": 3720 }, { "epoch": 0.12, "learning_rate": 4.8543586640805384e-05, "loss": 1.1386, "step": 3721 }, { "epoch": 0.12, "learning_rate": 4.854269106807825e-05, "loss": 1.0761, "step": 3722 }, { "epoch": 0.12, "learning_rate": 4.8541795228350775e-05, "loss": 1.0853, "step": 3723 }, { "epoch": 0.12, "learning_rate": 4.8540899121633123e-05, "loss": 1.0894, "step": 3724 }, { "epoch": 0.12, "learning_rate": 4.854000274793545e-05, "loss": 0.8954, "step": 3725 }, { "epoch": 0.12, "learning_rate": 4.853910610726792e-05, "loss": 0.9487, "step": 3726 }, { "epoch": 0.12, "learning_rate": 4.853820919964072e-05, "loss": 0.808, "step": 3727 }, { "epoch": 0.12, "learning_rate": 4.853731202506399e-05, "loss": 0.7009, "step": 3728 }, { "epoch": 0.12, "learning_rate": 4.8536414583547936e-05, "loss": 0.6914, "step": 3729 }, { "epoch": 0.12, "learning_rate": 4.853551687510272e-05, "loss": 1.0455, "step": 3730 }, { "epoch": 0.12, "learning_rate": 4.853461889973851e-05, "loss": 1.3859, "step": 3731 }, { "epoch": 0.12, "learning_rate": 4.853372065746553e-05, "loss": 0.8219, "step": 3732 }, { "epoch": 0.12, "learning_rate": 4.853282214829393e-05, "loss": 0.8636, "step": 3733 }, { "epoch": 0.12, "learning_rate": 4.853192337223392e-05, "loss": 0.9023, "step": 3734 }, { "epoch": 0.12, "learning_rate": 4.8531024329295685e-05, "loss": 1.7222, "step": 3735 }, { "epoch": 0.12, "learning_rate": 4.8530125019489415e-05, "loss": 4.0363, "step": 3736 }, { "epoch": 0.12, "learning_rate": 4.852922544282532e-05, "loss": 1.9838, "step": 3737 }, { "epoch": 0.12, "learning_rate": 4.85283255993136e-05, "loss": 0.6971, "step": 3738 }, { "epoch": 0.12, "learning_rate": 4.8527425488964454e-05, "loss": 0.4155, "step": 3739 }, { "epoch": 0.12, "learning_rate": 4.85265251117881e-05, "loss": 0.2993, "step": 3740 }, { "epoch": 0.12, "learning_rate": 4.852562446779475e-05, "loss": 0.3371, "step": 3741 }, { "epoch": 0.12, "learning_rate": 4.85247235569946e-05, "loss": 0.5361, "step": 3742 }, { "epoch": 0.12, "learning_rate": 4.8523822379397886e-05, "loss": 0.5219, "step": 3743 }, { "epoch": 0.12, "learning_rate": 4.8522920935014826e-05, "loss": 0.7377, "step": 3744 }, { "epoch": 0.12, "learning_rate": 4.852201922385564e-05, "loss": 0.6705, "step": 3745 }, { "epoch": 0.12, "learning_rate": 4.8521117245930546e-05, "loss": 0.9283, "step": 3746 }, { "epoch": 0.12, "learning_rate": 4.852021500124979e-05, "loss": 1.0117, "step": 3747 }, { "epoch": 0.12, "learning_rate": 4.8519312489823596e-05, "loss": 0.4555, "step": 3748 }, { "epoch": 0.12, "learning_rate": 4.851840971166219e-05, "loss": 0.622, "step": 3749 }, { "epoch": 0.12, "learning_rate": 4.851750666677584e-05, "loss": 0.9558, "step": 3750 }, { "epoch": 0.13, "learning_rate": 4.851660335517475e-05, "loss": 1.0062, "step": 3751 }, { "epoch": 0.13, "learning_rate": 4.851569977686918e-05, "loss": 1.1186, "step": 3752 }, { "epoch": 0.13, "learning_rate": 4.8514795931869394e-05, "loss": 1.3103, "step": 3753 }, { "epoch": 0.13, "learning_rate": 4.851389182018562e-05, "loss": 1.0394, "step": 3754 }, { "epoch": 0.13, "learning_rate": 4.8512987441828126e-05, "loss": 1.1588, "step": 3755 }, { "epoch": 0.13, "learning_rate": 4.8512082796807156e-05, "loss": 0.7285, "step": 3756 }, { "epoch": 0.13, "learning_rate": 4.851117788513298e-05, "loss": 1.0033, "step": 3757 }, { "epoch": 0.13, "learning_rate": 4.851027270681585e-05, "loss": 0.9363, "step": 3758 }, { "epoch": 0.13, "learning_rate": 4.8509367261866046e-05, "loss": 0.8139, "step": 3759 }, { "epoch": 0.13, "learning_rate": 4.850846155029383e-05, "loss": 2.9861, "step": 3760 }, { "epoch": 0.13, "learning_rate": 4.850755557210948e-05, "loss": 1.5414, "step": 3761 }, { "epoch": 0.13, "learning_rate": 4.850664932732325e-05, "loss": 0.753, "step": 3762 }, { "epoch": 0.13, "learning_rate": 4.8505742815945436e-05, "loss": 0.1218, "step": 3763 }, { "epoch": 0.13, "learning_rate": 4.8504836037986326e-05, "loss": 0.6913, "step": 3764 }, { "epoch": 0.13, "learning_rate": 4.850392899345618e-05, "loss": 1.2042, "step": 3765 }, { "epoch": 0.13, "learning_rate": 4.85030216823653e-05, "loss": 0.8662, "step": 3766 }, { "epoch": 0.13, "learning_rate": 4.850211410472397e-05, "loss": 0.8822, "step": 3767 }, { "epoch": 0.13, "learning_rate": 4.8501206260542495e-05, "loss": 1.7519, "step": 3768 }, { "epoch": 0.13, "learning_rate": 4.850029814983116e-05, "loss": 1.8665, "step": 3769 }, { "epoch": 0.13, "learning_rate": 4.849938977260027e-05, "loss": 1.082, "step": 3770 }, { "epoch": 0.13, "learning_rate": 4.849848112886011e-05, "loss": 0.8454, "step": 3771 }, { "epoch": 0.13, "learning_rate": 4.849757221862101e-05, "loss": 0.9324, "step": 3772 }, { "epoch": 0.13, "learning_rate": 4.8496663041893265e-05, "loss": 2.6007, "step": 3773 }, { "epoch": 0.13, "learning_rate": 4.8495753598687185e-05, "loss": 1.0955, "step": 3774 }, { "epoch": 0.13, "learning_rate": 4.8494843889013086e-05, "loss": 1.0441, "step": 3775 }, { "epoch": 0.13, "learning_rate": 4.849393391288129e-05, "loss": 0.8748, "step": 3776 }, { "epoch": 0.13, "learning_rate": 4.849302367030211e-05, "loss": 1.704, "step": 3777 }, { "epoch": 0.13, "learning_rate": 4.849211316128587e-05, "loss": 0.9221, "step": 3778 }, { "epoch": 0.13, "learning_rate": 4.849120238584291e-05, "loss": 0.7922, "step": 3779 }, { "epoch": 0.13, "learning_rate": 4.849029134398354e-05, "loss": 0.4587, "step": 3780 }, { "epoch": 0.13, "learning_rate": 4.848938003571809e-05, "loss": 0.304, "step": 3781 }, { "epoch": 0.13, "learning_rate": 4.848846846105692e-05, "loss": 0.865, "step": 3782 }, { "epoch": 0.13, "learning_rate": 4.848755662001034e-05, "loss": 1.4756, "step": 3783 }, { "epoch": 0.13, "learning_rate": 4.848664451258872e-05, "loss": 1.6047, "step": 3784 }, { "epoch": 0.13, "learning_rate": 4.848573213880238e-05, "loss": 1.1366, "step": 3785 }, { "epoch": 0.13, "learning_rate": 4.848481949866167e-05, "loss": 0.8104, "step": 3786 }, { "epoch": 0.13, "learning_rate": 4.848390659217696e-05, "loss": 0.9296, "step": 3787 }, { "epoch": 0.13, "learning_rate": 4.848299341935858e-05, "loss": 1.3314, "step": 3788 }, { "epoch": 0.13, "learning_rate": 4.84820799802169e-05, "loss": 0.8149, "step": 3789 }, { "epoch": 0.13, "learning_rate": 4.848116627476228e-05, "loss": 0.4216, "step": 3790 }, { "epoch": 0.13, "learning_rate": 4.848025230300507e-05, "loss": 0.8036, "step": 3791 }, { "epoch": 0.13, "learning_rate": 4.847933806495565e-05, "loss": 0.9524, "step": 3792 }, { "epoch": 0.13, "learning_rate": 4.847842356062438e-05, "loss": 0.428, "step": 3793 }, { "epoch": 0.13, "learning_rate": 4.847750879002163e-05, "loss": 0.6271, "step": 3794 }, { "epoch": 0.13, "learning_rate": 4.847659375315779e-05, "loss": 0.7565, "step": 3795 }, { "epoch": 0.13, "learning_rate": 4.847567845004321e-05, "loss": 0.9534, "step": 3796 }, { "epoch": 0.13, "learning_rate": 4.84747628806883e-05, "loss": 0.9861, "step": 3797 }, { "epoch": 0.13, "learning_rate": 4.847384704510342e-05, "loss": 0.5457, "step": 3798 }, { "epoch": 0.13, "learning_rate": 4.8472930943298963e-05, "loss": 0.2949, "step": 3799 }, { "epoch": 0.13, "learning_rate": 4.847201457528533e-05, "loss": 0.2888, "step": 3800 }, { "epoch": 0.13, "learning_rate": 4.8471097941072906e-05, "loss": 0.7595, "step": 3801 }, { "epoch": 0.13, "learning_rate": 4.8470181040672086e-05, "loss": 1.0275, "step": 3802 }, { "epoch": 0.13, "learning_rate": 4.846926387409327e-05, "loss": 1.5985, "step": 3803 }, { "epoch": 0.13, "learning_rate": 4.846834644134686e-05, "loss": 1.2035, "step": 3804 }, { "epoch": 0.13, "learning_rate": 4.8467428742443256e-05, "loss": 0.9006, "step": 3805 }, { "epoch": 0.13, "learning_rate": 4.846651077739287e-05, "loss": 0.8603, "step": 3806 }, { "epoch": 0.13, "learning_rate": 4.846559254620611e-05, "loss": 0.9263, "step": 3807 }, { "epoch": 0.13, "learning_rate": 4.846467404889339e-05, "loss": 0.715, "step": 3808 }, { "epoch": 0.13, "learning_rate": 4.846375528546513e-05, "loss": 0.4928, "step": 3809 }, { "epoch": 0.13, "learning_rate": 4.846283625593175e-05, "loss": 0.6946, "step": 3810 }, { "epoch": 0.13, "learning_rate": 4.846191696030366e-05, "loss": 0.8801, "step": 3811 }, { "epoch": 0.13, "learning_rate": 4.846099739859131e-05, "loss": 0.5999, "step": 3812 }, { "epoch": 0.13, "learning_rate": 4.846007757080512e-05, "loss": 0.8658, "step": 3813 }, { "epoch": 0.13, "learning_rate": 4.8459157476955505e-05, "loss": 0.7138, "step": 3814 }, { "epoch": 0.13, "learning_rate": 4.8458237117052924e-05, "loss": 1.0024, "step": 3815 }, { "epoch": 0.13, "learning_rate": 4.84573164911078e-05, "loss": 1.2334, "step": 3816 }, { "epoch": 0.13, "learning_rate": 4.845639559913057e-05, "loss": 0.9082, "step": 3817 }, { "epoch": 0.13, "learning_rate": 4.8455474441131695e-05, "loss": 1.7044, "step": 3818 }, { "epoch": 0.13, "learning_rate": 4.8454553017121605e-05, "loss": 1.611, "step": 3819 }, { "epoch": 0.13, "learning_rate": 4.845363132711076e-05, "loss": 1.4487, "step": 3820 }, { "epoch": 0.13, "learning_rate": 4.8452709371109606e-05, "loss": 1.4681, "step": 3821 }, { "epoch": 0.13, "learning_rate": 4.8451787149128614e-05, "loss": 0.8031, "step": 3822 }, { "epoch": 0.13, "learning_rate": 4.845086466117822e-05, "loss": 2.7472, "step": 3823 }, { "epoch": 0.13, "learning_rate": 4.8449941907268906e-05, "loss": 0.9273, "step": 3824 }, { "epoch": 0.13, "learning_rate": 4.844901888741112e-05, "loss": 0.9863, "step": 3825 }, { "epoch": 0.13, "learning_rate": 4.844809560161535e-05, "loss": 0.6017, "step": 3826 }, { "epoch": 0.13, "learning_rate": 4.844717204989205e-05, "loss": 0.8474, "step": 3827 }, { "epoch": 0.13, "learning_rate": 4.844624823225171e-05, "loss": 0.9445, "step": 3828 }, { "epoch": 0.13, "learning_rate": 4.844532414870479e-05, "loss": 0.4674, "step": 3829 }, { "epoch": 0.13, "learning_rate": 4.8444399799261784e-05, "loss": 0.7958, "step": 3830 }, { "epoch": 0.13, "learning_rate": 4.844347518393316e-05, "loss": 0.7051, "step": 3831 }, { "epoch": 0.13, "learning_rate": 4.844255030272942e-05, "loss": 0.9214, "step": 3832 }, { "epoch": 0.13, "learning_rate": 4.844162515566104e-05, "loss": 0.9376, "step": 3833 }, { "epoch": 0.13, "learning_rate": 4.844069974273853e-05, "loss": 1.4291, "step": 3834 }, { "epoch": 0.13, "learning_rate": 4.843977406397236e-05, "loss": 0.5619, "step": 3835 }, { "epoch": 0.13, "learning_rate": 4.843884811937305e-05, "loss": 0.9372, "step": 3836 }, { "epoch": 0.13, "learning_rate": 4.8437921908951086e-05, "loss": 1.1829, "step": 3837 }, { "epoch": 0.13, "learning_rate": 4.843699543271698e-05, "loss": 0.9897, "step": 3838 }, { "epoch": 0.13, "learning_rate": 4.8436068690681244e-05, "loss": 0.5894, "step": 3839 }, { "epoch": 0.13, "learning_rate": 4.843514168285439e-05, "loss": 0.7758, "step": 3840 }, { "epoch": 0.13, "learning_rate": 4.843421440924691e-05, "loss": 0.8073, "step": 3841 }, { "epoch": 0.13, "learning_rate": 4.8433286869869346e-05, "loss": 0.921, "step": 3842 }, { "epoch": 0.13, "learning_rate": 4.8432359064732204e-05, "loss": 0.8884, "step": 3843 }, { "epoch": 0.13, "learning_rate": 4.8431430993846004e-05, "loss": 0.9082, "step": 3844 }, { "epoch": 0.13, "learning_rate": 4.8430502657221275e-05, "loss": 0.8399, "step": 3845 }, { "epoch": 0.13, "learning_rate": 4.8429574054868555e-05, "loss": 0.5377, "step": 3846 }, { "epoch": 0.13, "learning_rate": 4.842864518679836e-05, "loss": 0.2233, "step": 3847 }, { "epoch": 0.13, "learning_rate": 4.8427716053021234e-05, "loss": 0.9174, "step": 3848 }, { "epoch": 0.13, "learning_rate": 4.842678665354771e-05, "loss": 1.745, "step": 3849 }, { "epoch": 0.13, "learning_rate": 4.842585698838832e-05, "loss": 1.7256, "step": 3850 }, { "epoch": 0.13, "learning_rate": 4.8424927057553624e-05, "loss": 0.9249, "step": 3851 }, { "epoch": 0.13, "learning_rate": 4.8423996861054166e-05, "loss": 0.8269, "step": 3852 }, { "epoch": 0.13, "learning_rate": 4.8423066398900484e-05, "loss": 0.846, "step": 3853 }, { "epoch": 0.13, "learning_rate": 4.842213567110314e-05, "loss": 1.14, "step": 3854 }, { "epoch": 0.13, "learning_rate": 4.8421204677672695e-05, "loss": 0.831, "step": 3855 }, { "epoch": 0.13, "learning_rate": 4.842027341861969e-05, "loss": 0.8624, "step": 3856 }, { "epoch": 0.13, "learning_rate": 4.84193418939547e-05, "loss": 1.0354, "step": 3857 }, { "epoch": 0.13, "learning_rate": 4.841841010368828e-05, "loss": 0.6857, "step": 3858 }, { "epoch": 0.13, "learning_rate": 4.841747804783101e-05, "loss": 0.5882, "step": 3859 }, { "epoch": 0.13, "learning_rate": 4.8416545726393456e-05, "loss": 1.0384, "step": 3860 }, { "epoch": 0.13, "learning_rate": 4.841561313938618e-05, "loss": 0.6689, "step": 3861 }, { "epoch": 0.13, "learning_rate": 4.841468028681978e-05, "loss": 0.8645, "step": 3862 }, { "epoch": 0.13, "learning_rate": 4.841374716870481e-05, "loss": 0.4489, "step": 3863 }, { "epoch": 0.13, "learning_rate": 4.841281378505187e-05, "loss": 0.2158, "step": 3864 }, { "epoch": 0.13, "learning_rate": 4.841188013587155e-05, "loss": 0.1949, "step": 3865 }, { "epoch": 0.13, "learning_rate": 4.841094622117442e-05, "loss": 0.9398, "step": 3866 }, { "epoch": 0.13, "learning_rate": 4.8410012040971084e-05, "loss": 0.5806, "step": 3867 }, { "epoch": 0.13, "learning_rate": 4.840907759527214e-05, "loss": 0.7211, "step": 3868 }, { "epoch": 0.13, "learning_rate": 4.840814288408817e-05, "loss": 0.7043, "step": 3869 }, { "epoch": 0.13, "learning_rate": 4.84072079074298e-05, "loss": 0.8887, "step": 3870 }, { "epoch": 0.13, "learning_rate": 4.8406272665307606e-05, "loss": 0.363, "step": 3871 }, { "epoch": 0.13, "learning_rate": 4.840533715773221e-05, "loss": 0.9002, "step": 3872 }, { "epoch": 0.13, "learning_rate": 4.8404401384714216e-05, "loss": 0.9588, "step": 3873 }, { "epoch": 0.13, "learning_rate": 4.8403465346264246e-05, "loss": 0.7214, "step": 3874 }, { "epoch": 0.13, "learning_rate": 4.840252904239291e-05, "loss": 0.8959, "step": 3875 }, { "epoch": 0.13, "learning_rate": 4.840159247311082e-05, "loss": 0.8677, "step": 3876 }, { "epoch": 0.13, "learning_rate": 4.84006556384286e-05, "loss": 0.2033, "step": 3877 }, { "epoch": 0.13, "learning_rate": 4.839971853835689e-05, "loss": 0.309, "step": 3878 }, { "epoch": 0.13, "learning_rate": 4.83987811729063e-05, "loss": 0.8129, "step": 3879 }, { "epoch": 0.13, "learning_rate": 4.839784354208747e-05, "loss": 0.8077, "step": 3880 }, { "epoch": 0.13, "learning_rate": 4.839690564591103e-05, "loss": 0.607, "step": 3881 }, { "epoch": 0.13, "learning_rate": 4.8395967484387614e-05, "loss": 0.4718, "step": 3882 }, { "epoch": 0.13, "learning_rate": 4.839502905752786e-05, "loss": 0.9409, "step": 3883 }, { "epoch": 0.13, "learning_rate": 4.839409036534243e-05, "loss": 1.1055, "step": 3884 }, { "epoch": 0.13, "learning_rate": 4.8393151407841955e-05, "loss": 1.0526, "step": 3885 }, { "epoch": 0.13, "learning_rate": 4.839221218503708e-05, "loss": 1.049, "step": 3886 }, { "epoch": 0.13, "learning_rate": 4.839127269693846e-05, "loss": 0.8966, "step": 3887 }, { "epoch": 0.13, "learning_rate": 4.839033294355675e-05, "loss": 0.9392, "step": 3888 }, { "epoch": 0.13, "learning_rate": 4.838939292490261e-05, "loss": 1.1048, "step": 3889 }, { "epoch": 0.13, "learning_rate": 4.83884526409867e-05, "loss": 1.2873, "step": 3890 }, { "epoch": 0.13, "learning_rate": 4.8387512091819696e-05, "loss": 1.113, "step": 3891 }, { "epoch": 0.13, "learning_rate": 4.838657127741224e-05, "loss": 0.7005, "step": 3892 }, { "epoch": 0.13, "learning_rate": 4.8385630197775026e-05, "loss": 0.5694, "step": 3893 }, { "epoch": 0.13, "learning_rate": 4.83846888529187e-05, "loss": 0.7969, "step": 3894 }, { "epoch": 0.13, "learning_rate": 4.8383747242853966e-05, "loss": 1.0621, "step": 3895 }, { "epoch": 0.13, "learning_rate": 4.838280536759149e-05, "loss": 0.9117, "step": 3896 }, { "epoch": 0.13, "learning_rate": 4.838186322714195e-05, "loss": 0.9613, "step": 3897 }, { "epoch": 0.13, "learning_rate": 4.838092082151604e-05, "loss": 0.8338, "step": 3898 }, { "epoch": 0.13, "learning_rate": 4.837997815072444e-05, "loss": 1.0102, "step": 3899 }, { "epoch": 0.13, "learning_rate": 4.8379035214777836e-05, "loss": 0.8145, "step": 3900 }, { "epoch": 0.13, "learning_rate": 4.8378092013686936e-05, "loss": 0.5489, "step": 3901 }, { "epoch": 0.13, "learning_rate": 4.837714854746244e-05, "loss": 0.871, "step": 3902 }, { "epoch": 0.13, "learning_rate": 4.8376204816115035e-05, "loss": 0.8615, "step": 3903 }, { "epoch": 0.13, "learning_rate": 4.837526081965542e-05, "loss": 0.8266, "step": 3904 }, { "epoch": 0.13, "learning_rate": 4.837431655809432e-05, "loss": 0.7242, "step": 3905 }, { "epoch": 0.13, "learning_rate": 4.837337203144243e-05, "loss": 0.8404, "step": 3906 }, { "epoch": 0.13, "learning_rate": 4.837242723971046e-05, "loss": 0.5839, "step": 3907 }, { "epoch": 0.13, "learning_rate": 4.8371482182909135e-05, "loss": 0.9265, "step": 3908 }, { "epoch": 0.13, "learning_rate": 4.8370536861049166e-05, "loss": 0.6965, "step": 3909 }, { "epoch": 0.13, "learning_rate": 4.836959127414127e-05, "loss": 0.7088, "step": 3910 }, { "epoch": 0.13, "learning_rate": 4.8368645422196186e-05, "loss": 0.9774, "step": 3911 }, { "epoch": 0.13, "learning_rate": 4.836769930522463e-05, "loss": 0.9853, "step": 3912 }, { "epoch": 0.13, "learning_rate": 4.836675292323734e-05, "loss": 1.3107, "step": 3913 }, { "epoch": 0.13, "learning_rate": 4.836580627624503e-05, "loss": 0.8033, "step": 3914 }, { "epoch": 0.13, "learning_rate": 4.836485936425845e-05, "loss": 1.0286, "step": 3915 }, { "epoch": 0.13, "learning_rate": 4.836391218728835e-05, "loss": 1.0854, "step": 3916 }, { "epoch": 0.13, "learning_rate": 4.836296474534545e-05, "loss": 0.7925, "step": 3917 }, { "epoch": 0.13, "learning_rate": 4.836201703844051e-05, "loss": 0.8343, "step": 3918 }, { "epoch": 0.13, "learning_rate": 4.836106906658427e-05, "loss": 0.7061, "step": 3919 }, { "epoch": 0.13, "learning_rate": 4.836012082978749e-05, "loss": 1.0766, "step": 3920 }, { "epoch": 0.13, "learning_rate": 4.8359172328060915e-05, "loss": 0.8021, "step": 3921 }, { "epoch": 0.13, "learning_rate": 4.8358223561415304e-05, "loss": 1.903, "step": 3922 }, { "epoch": 0.13, "learning_rate": 4.835727452986142e-05, "loss": 3.578, "step": 3923 }, { "epoch": 0.13, "learning_rate": 4.835632523341003e-05, "loss": 0.8328, "step": 3924 }, { "epoch": 0.13, "learning_rate": 4.835537567207188e-05, "loss": 0.9087, "step": 3925 }, { "epoch": 0.13, "learning_rate": 4.835442584585777e-05, "loss": 1.028, "step": 3926 }, { "epoch": 0.13, "learning_rate": 4.8353475754778444e-05, "loss": 0.9628, "step": 3927 }, { "epoch": 0.13, "learning_rate": 4.83525253988447e-05, "loss": 1.0444, "step": 3928 }, { "epoch": 0.13, "learning_rate": 4.8351574778067296e-05, "loss": 0.7072, "step": 3929 }, { "epoch": 0.13, "learning_rate": 4.835062389245703e-05, "loss": 0.6911, "step": 3930 }, { "epoch": 0.13, "learning_rate": 4.8349672742024666e-05, "loss": 0.464, "step": 3931 }, { "epoch": 0.13, "learning_rate": 4.834872132678101e-05, "loss": 0.8692, "step": 3932 }, { "epoch": 0.13, "learning_rate": 4.834776964673685e-05, "loss": 1.1934, "step": 3933 }, { "epoch": 0.13, "learning_rate": 4.834681770190297e-05, "loss": 1.0095, "step": 3934 }, { "epoch": 0.13, "learning_rate": 4.8345865492290174e-05, "loss": 0.8974, "step": 3935 }, { "epoch": 0.13, "learning_rate": 4.834491301790925e-05, "loss": 0.9532, "step": 3936 }, { "epoch": 0.13, "learning_rate": 4.834396027877102e-05, "loss": 1.0299, "step": 3937 }, { "epoch": 0.13, "learning_rate": 4.834300727488627e-05, "loss": 2.4181, "step": 3938 }, { "epoch": 0.13, "learning_rate": 4.8342054006265814e-05, "loss": 1.5486, "step": 3939 }, { "epoch": 0.13, "learning_rate": 4.834110047292047e-05, "loss": 2.961, "step": 3940 }, { "epoch": 0.13, "learning_rate": 4.834014667486103e-05, "loss": 2.6218, "step": 3941 }, { "epoch": 0.13, "learning_rate": 4.833919261209834e-05, "loss": 1.6678, "step": 3942 }, { "epoch": 0.13, "learning_rate": 4.8338238284643214e-05, "loss": 0.8602, "step": 3943 }, { "epoch": 0.13, "learning_rate": 4.8337283692506466e-05, "loss": 0.8569, "step": 3944 }, { "epoch": 0.13, "learning_rate": 4.833632883569892e-05, "loss": 1.0265, "step": 3945 }, { "epoch": 0.13, "learning_rate": 4.83353737142314e-05, "loss": 0.9874, "step": 3946 }, { "epoch": 0.13, "learning_rate": 4.8334418328114765e-05, "loss": 0.9167, "step": 3947 }, { "epoch": 0.13, "learning_rate": 4.833346267735983e-05, "loss": 0.7643, "step": 3948 }, { "epoch": 0.13, "learning_rate": 4.833250676197744e-05, "loss": 1.0942, "step": 3949 }, { "epoch": 0.13, "learning_rate": 4.833155058197842e-05, "loss": 1.4395, "step": 3950 }, { "epoch": 0.13, "learning_rate": 4.833059413737364e-05, "loss": 0.7939, "step": 3951 }, { "epoch": 0.13, "learning_rate": 4.832963742817392e-05, "loss": 0.8463, "step": 3952 }, { "epoch": 0.13, "learning_rate": 4.8328680454390136e-05, "loss": 1.0912, "step": 3953 }, { "epoch": 0.13, "learning_rate": 4.832772321603313e-05, "loss": 0.9993, "step": 3954 }, { "epoch": 0.13, "learning_rate": 4.832676571311375e-05, "loss": 0.948, "step": 3955 }, { "epoch": 0.13, "learning_rate": 4.8325807945642866e-05, "loss": 1.2866, "step": 3956 }, { "epoch": 0.13, "learning_rate": 4.832484991363133e-05, "loss": 1.2744, "step": 3957 }, { "epoch": 0.13, "learning_rate": 4.832389161709002e-05, "loss": 0.8159, "step": 3958 }, { "epoch": 0.13, "learning_rate": 4.832293305602981e-05, "loss": 0.7054, "step": 3959 }, { "epoch": 0.13, "learning_rate": 4.832197423046154e-05, "loss": 1.0259, "step": 3960 }, { "epoch": 0.13, "learning_rate": 4.8321015140396105e-05, "loss": 1.2321, "step": 3961 }, { "epoch": 0.13, "learning_rate": 4.8320055785844394e-05, "loss": 2.0009, "step": 3962 }, { "epoch": 0.13, "learning_rate": 4.831909616681726e-05, "loss": 0.9798, "step": 3963 }, { "epoch": 0.13, "learning_rate": 4.831813628332561e-05, "loss": 0.957, "step": 3964 }, { "epoch": 0.13, "learning_rate": 4.8317176135380314e-05, "loss": 0.9444, "step": 3965 }, { "epoch": 0.13, "learning_rate": 4.831621572299227e-05, "loss": 0.7919, "step": 3966 }, { "epoch": 0.13, "learning_rate": 4.831525504617237e-05, "loss": 0.9833, "step": 3967 }, { "epoch": 0.13, "learning_rate": 4.83142941049315e-05, "loss": 0.536, "step": 3968 }, { "epoch": 0.13, "learning_rate": 4.831333289928056e-05, "loss": 0.4996, "step": 3969 }, { "epoch": 0.13, "learning_rate": 4.8312371429230466e-05, "loss": 0.8433, "step": 3970 }, { "epoch": 0.13, "learning_rate": 4.831140969479211e-05, "loss": 0.7308, "step": 3971 }, { "epoch": 0.13, "learning_rate": 4.8310447695976404e-05, "loss": 1.0697, "step": 3972 }, { "epoch": 0.13, "learning_rate": 4.8309485432794244e-05, "loss": 0.6797, "step": 3973 }, { "epoch": 0.13, "learning_rate": 4.830852290525656e-05, "loss": 0.9986, "step": 3974 }, { "epoch": 0.13, "learning_rate": 4.8307560113374264e-05, "loss": 0.9615, "step": 3975 }, { "epoch": 0.13, "learning_rate": 4.830659705715828e-05, "loss": 0.7929, "step": 3976 }, { "epoch": 0.13, "learning_rate": 4.830563373661951e-05, "loss": 0.7358, "step": 3977 }, { "epoch": 0.13, "learning_rate": 4.83046701517689e-05, "loss": 1.0033, "step": 3978 }, { "epoch": 0.13, "learning_rate": 4.830370630261737e-05, "loss": 0.948, "step": 3979 }, { "epoch": 0.13, "learning_rate": 4.8302742189175846e-05, "loss": 3.2597, "step": 3980 }, { "epoch": 0.13, "learning_rate": 4.8301777811455276e-05, "loss": 0.8038, "step": 3981 }, { "epoch": 0.13, "learning_rate": 4.8300813169466585e-05, "loss": 0.8774, "step": 3982 }, { "epoch": 0.13, "learning_rate": 4.829984826322072e-05, "loss": 1.0474, "step": 3983 }, { "epoch": 0.13, "learning_rate": 4.829888309272862e-05, "loss": 0.7578, "step": 3984 }, { "epoch": 0.13, "learning_rate": 4.829791765800123e-05, "loss": 0.8437, "step": 3985 }, { "epoch": 0.13, "learning_rate": 4.8296951959049505e-05, "loss": 0.9117, "step": 3986 }, { "epoch": 0.13, "learning_rate": 4.82959859958844e-05, "loss": 0.9864, "step": 3987 }, { "epoch": 0.13, "learning_rate": 4.829501976851686e-05, "loss": 0.8598, "step": 3988 }, { "epoch": 0.13, "learning_rate": 4.829405327695784e-05, "loss": 0.9535, "step": 3989 }, { "epoch": 0.13, "learning_rate": 4.829308652121831e-05, "loss": 0.5532, "step": 3990 }, { "epoch": 0.13, "learning_rate": 4.829211950130924e-05, "loss": 0.9525, "step": 3991 }, { "epoch": 0.13, "learning_rate": 4.829115221724158e-05, "loss": 0.8077, "step": 3992 }, { "epoch": 0.13, "learning_rate": 4.8290184669026314e-05, "loss": 0.8331, "step": 3993 }, { "epoch": 0.13, "learning_rate": 4.8289216856674414e-05, "loss": 0.9622, "step": 3994 }, { "epoch": 0.13, "learning_rate": 4.8288248780196844e-05, "loss": 0.5614, "step": 3995 }, { "epoch": 0.13, "learning_rate": 4.82872804396046e-05, "loss": 0.1359, "step": 3996 }, { "epoch": 0.13, "learning_rate": 4.828631183490865e-05, "loss": 1.3686, "step": 3997 }, { "epoch": 0.13, "learning_rate": 4.8285342966119984e-05, "loss": 1.0356, "step": 3998 }, { "epoch": 0.13, "learning_rate": 4.8284373833249594e-05, "loss": 0.6972, "step": 3999 }, { "epoch": 0.13, "learning_rate": 4.8283404436308464e-05, "loss": 0.9796, "step": 4000 }, { "epoch": 0.13, "learning_rate": 4.82824347753076e-05, "loss": 0.8064, "step": 4001 }, { "epoch": 0.13, "learning_rate": 4.828146485025797e-05, "loss": 0.8935, "step": 4002 }, { "epoch": 0.13, "learning_rate": 4.828049466117062e-05, "loss": 0.8925, "step": 4003 }, { "epoch": 0.13, "learning_rate": 4.8279524208056516e-05, "loss": 0.9964, "step": 4004 }, { "epoch": 0.13, "learning_rate": 4.8278553490926676e-05, "loss": 1.7328, "step": 4005 }, { "epoch": 0.13, "learning_rate": 4.827758250979211e-05, "loss": 2.0897, "step": 4006 }, { "epoch": 0.13, "learning_rate": 4.827661126466383e-05, "loss": 0.9991, "step": 4007 }, { "epoch": 0.13, "learning_rate": 4.827563975555285e-05, "loss": 0.9175, "step": 4008 }, { "epoch": 0.13, "learning_rate": 4.8274667982470185e-05, "loss": 1.0543, "step": 4009 }, { "epoch": 0.13, "learning_rate": 4.827369594542686e-05, "loss": 0.8224, "step": 4010 }, { "epoch": 0.13, "learning_rate": 4.82727236444339e-05, "loss": 0.8898, "step": 4011 }, { "epoch": 0.13, "learning_rate": 4.827175107950232e-05, "loss": 0.6607, "step": 4012 }, { "epoch": 0.13, "learning_rate": 4.827077825064318e-05, "loss": 1.0643, "step": 4013 }, { "epoch": 0.13, "learning_rate": 4.826980515786748e-05, "loss": 4.2295, "step": 4014 }, { "epoch": 0.13, "learning_rate": 4.8268831801186265e-05, "loss": 4.1748, "step": 4015 }, { "epoch": 0.13, "learning_rate": 4.826785818061058e-05, "loss": 0.9742, "step": 4016 }, { "epoch": 0.13, "learning_rate": 4.826688429615147e-05, "loss": 0.8931, "step": 4017 }, { "epoch": 0.13, "learning_rate": 4.826591014781997e-05, "loss": 0.828, "step": 4018 }, { "epoch": 0.13, "learning_rate": 4.826493573562714e-05, "loss": 0.6571, "step": 4019 }, { "epoch": 0.13, "learning_rate": 4.8263961059584015e-05, "loss": 0.8715, "step": 4020 }, { "epoch": 0.13, "learning_rate": 4.8262986119701656e-05, "loss": 1.0842, "step": 4021 }, { "epoch": 0.13, "learning_rate": 4.8262010915991126e-05, "loss": 0.8532, "step": 4022 }, { "epoch": 0.13, "learning_rate": 4.8261035448463484e-05, "loss": 0.9835, "step": 4023 }, { "epoch": 0.13, "learning_rate": 4.826005971712978e-05, "loss": 0.8976, "step": 4024 }, { "epoch": 0.13, "learning_rate": 4.82590837220011e-05, "loss": 1.0314, "step": 4025 }, { "epoch": 0.13, "learning_rate": 4.825810746308849e-05, "loss": 1.0686, "step": 4026 }, { "epoch": 0.13, "learning_rate": 4.825713094040304e-05, "loss": 1.8379, "step": 4027 }, { "epoch": 0.13, "learning_rate": 4.8256154153955815e-05, "loss": 1.1156, "step": 4028 }, { "epoch": 0.13, "learning_rate": 4.82551771037579e-05, "loss": 0.5774, "step": 4029 }, { "epoch": 0.13, "learning_rate": 4.8254199789820364e-05, "loss": 1.0241, "step": 4030 }, { "epoch": 0.13, "learning_rate": 4.825322221215431e-05, "loss": 1.1402, "step": 4031 }, { "epoch": 0.13, "learning_rate": 4.82522443707708e-05, "loss": 1.6083, "step": 4032 }, { "epoch": 0.13, "learning_rate": 4.8251266265680954e-05, "loss": 0.8761, "step": 4033 }, { "epoch": 0.13, "learning_rate": 4.825028789689584e-05, "loss": 0.9501, "step": 4034 }, { "epoch": 0.13, "learning_rate": 4.8249309264426566e-05, "loss": 0.8525, "step": 4035 }, { "epoch": 0.13, "learning_rate": 4.8248330368284226e-05, "loss": 0.7858, "step": 4036 }, { "epoch": 0.13, "learning_rate": 4.8247351208479915e-05, "loss": 0.9295, "step": 4037 }, { "epoch": 0.13, "learning_rate": 4.8246371785024755e-05, "loss": 0.9253, "step": 4038 }, { "epoch": 0.13, "learning_rate": 4.8245392097929845e-05, "loss": 1.2157, "step": 4039 }, { "epoch": 0.13, "learning_rate": 4.8244412147206284e-05, "loss": 1.4728, "step": 4040 }, { "epoch": 0.13, "learning_rate": 4.8243431932865206e-05, "loss": 1.6789, "step": 4041 }, { "epoch": 0.13, "learning_rate": 4.8242451454917717e-05, "loss": 1.6854, "step": 4042 }, { "epoch": 0.13, "learning_rate": 4.8241470713374936e-05, "loss": 0.9443, "step": 4043 }, { "epoch": 0.13, "learning_rate": 4.824048970824799e-05, "loss": 0.9808, "step": 4044 }, { "epoch": 0.13, "learning_rate": 4.8239508439548e-05, "loss": 0.8997, "step": 4045 }, { "epoch": 0.13, "learning_rate": 4.82385269072861e-05, "loss": 0.93, "step": 4046 }, { "epoch": 0.13, "learning_rate": 4.823754511147342e-05, "loss": 1.4145, "step": 4047 }, { "epoch": 0.13, "learning_rate": 4.823656305212109e-05, "loss": 1.593, "step": 4048 }, { "epoch": 0.13, "learning_rate": 4.823558072924025e-05, "loss": 1.6247, "step": 4049 }, { "epoch": 0.14, "learning_rate": 4.8234598142842046e-05, "loss": 0.778, "step": 4050 }, { "epoch": 0.14, "learning_rate": 4.8233615292937616e-05, "loss": 1.0013, "step": 4051 }, { "epoch": 0.14, "learning_rate": 4.823263217953811e-05, "loss": 1.8636, "step": 4052 }, { "epoch": 0.14, "learning_rate": 4.823164880265468e-05, "loss": 0.2871, "step": 4053 }, { "epoch": 0.14, "learning_rate": 4.823066516229847e-05, "loss": 0.845, "step": 4054 }, { "epoch": 0.14, "learning_rate": 4.822968125848063e-05, "loss": 0.907, "step": 4055 }, { "epoch": 0.14, "learning_rate": 4.8228697091212336e-05, "loss": 1.0153, "step": 4056 }, { "epoch": 0.14, "learning_rate": 4.822771266050475e-05, "loss": 0.9082, "step": 4057 }, { "epoch": 0.14, "learning_rate": 4.822672796636902e-05, "loss": 0.9671, "step": 4058 }, { "epoch": 0.14, "learning_rate": 4.822574300881632e-05, "loss": 0.8324, "step": 4059 }, { "epoch": 0.14, "learning_rate": 4.8224757787857825e-05, "loss": 1.023, "step": 4060 }, { "epoch": 0.14, "learning_rate": 4.822377230350471e-05, "loss": 0.8708, "step": 4061 }, { "epoch": 0.14, "learning_rate": 4.822278655576814e-05, "loss": 0.9262, "step": 4062 }, { "epoch": 0.14, "learning_rate": 4.8221800544659315e-05, "loss": 0.8341, "step": 4063 }, { "epoch": 0.14, "learning_rate": 4.82208142701894e-05, "loss": 0.8618, "step": 4064 }, { "epoch": 0.14, "learning_rate": 4.821982773236957e-05, "loss": 0.8585, "step": 4065 }, { "epoch": 0.14, "learning_rate": 4.821884093121104e-05, "loss": 0.8487, "step": 4066 }, { "epoch": 0.14, "learning_rate": 4.8217853866724996e-05, "loss": 0.9089, "step": 4067 }, { "epoch": 0.14, "learning_rate": 4.8216866538922615e-05, "loss": 0.7683, "step": 4068 }, { "epoch": 0.14, "learning_rate": 4.821587894781511e-05, "loss": 0.7469, "step": 4069 }, { "epoch": 0.14, "learning_rate": 4.821489109341368e-05, "loss": 0.8119, "step": 4070 }, { "epoch": 0.14, "learning_rate": 4.821390297572952e-05, "loss": 0.9227, "step": 4071 }, { "epoch": 0.14, "learning_rate": 4.821291459477384e-05, "loss": 1.0148, "step": 4072 }, { "epoch": 0.14, "learning_rate": 4.8211925950557856e-05, "loss": 1.0837, "step": 4073 }, { "epoch": 0.14, "learning_rate": 4.821093704309278e-05, "loss": 0.8115, "step": 4074 }, { "epoch": 0.14, "learning_rate": 4.8209947872389816e-05, "loss": 0.6116, "step": 4075 }, { "epoch": 0.14, "learning_rate": 4.820895843846019e-05, "loss": 0.7546, "step": 4076 }, { "epoch": 0.14, "learning_rate": 4.820796874131513e-05, "loss": 1.0088, "step": 4077 }, { "epoch": 0.14, "learning_rate": 4.820697878096584e-05, "loss": 1.6793, "step": 4078 }, { "epoch": 0.14, "learning_rate": 4.8205988557423566e-05, "loss": 0.1945, "step": 4079 }, { "epoch": 0.14, "learning_rate": 4.8204998070699534e-05, "loss": 0.3237, "step": 4080 }, { "epoch": 0.14, "learning_rate": 4.820400732080497e-05, "loss": 0.7689, "step": 4081 }, { "epoch": 0.14, "learning_rate": 4.820301630775112e-05, "loss": 0.5639, "step": 4082 }, { "epoch": 0.14, "learning_rate": 4.820202503154921e-05, "loss": 0.8356, "step": 4083 }, { "epoch": 0.14, "learning_rate": 4.82010334922105e-05, "loss": 0.5411, "step": 4084 }, { "epoch": 0.14, "learning_rate": 4.8200041689746225e-05, "loss": 0.8411, "step": 4085 }, { "epoch": 0.14, "learning_rate": 4.819904962416763e-05, "loss": 0.8203, "step": 4086 }, { "epoch": 0.14, "learning_rate": 4.819805729548598e-05, "loss": 0.852, "step": 4087 }, { "epoch": 0.14, "learning_rate": 4.8197064703712514e-05, "loss": 0.7565, "step": 4088 }, { "epoch": 0.14, "learning_rate": 4.819607184885849e-05, "loss": 0.7958, "step": 4089 }, { "epoch": 0.14, "learning_rate": 4.819507873093517e-05, "loss": 0.9953, "step": 4090 }, { "epoch": 0.14, "learning_rate": 4.819408534995383e-05, "loss": 1.1763, "step": 4091 }, { "epoch": 0.14, "learning_rate": 4.819309170592572e-05, "loss": 1.1263, "step": 4092 }, { "epoch": 0.14, "learning_rate": 4.8192097798862117e-05, "loss": 1.1246, "step": 4093 }, { "epoch": 0.14, "learning_rate": 4.819110362877429e-05, "loss": 0.8528, "step": 4094 }, { "epoch": 0.14, "learning_rate": 4.819010919567352e-05, "loss": 0.8284, "step": 4095 }, { "epoch": 0.14, "learning_rate": 4.818911449957107e-05, "loss": 0.9595, "step": 4096 }, { "epoch": 0.14, "learning_rate": 4.818811954047824e-05, "loss": 0.7795, "step": 4097 }, { "epoch": 0.14, "learning_rate": 4.8187124318406296e-05, "loss": 0.6278, "step": 4098 }, { "epoch": 0.14, "learning_rate": 4.818612883336654e-05, "loss": 0.7256, "step": 4099 }, { "epoch": 0.14, "learning_rate": 4.8185133085370246e-05, "loss": 0.917, "step": 4100 }, { "epoch": 0.14, "learning_rate": 4.8184137074428724e-05, "loss": 0.9096, "step": 4101 }, { "epoch": 0.14, "learning_rate": 4.818314080055326e-05, "loss": 0.8233, "step": 4102 }, { "epoch": 0.14, "learning_rate": 4.818214426375516e-05, "loss": 0.6552, "step": 4103 }, { "epoch": 0.14, "learning_rate": 4.818114746404572e-05, "loss": 0.9823, "step": 4104 }, { "epoch": 0.14, "learning_rate": 4.818015040143624e-05, "loss": 0.7658, "step": 4105 }, { "epoch": 0.14, "learning_rate": 4.8179153075938034e-05, "loss": 0.7497, "step": 4106 }, { "epoch": 0.14, "learning_rate": 4.817815548756241e-05, "loss": 0.7518, "step": 4107 }, { "epoch": 0.14, "learning_rate": 4.8177157636320694e-05, "loss": 0.8835, "step": 4108 }, { "epoch": 0.14, "learning_rate": 4.817615952222419e-05, "loss": 1.1814, "step": 4109 }, { "epoch": 0.14, "learning_rate": 4.8175161145284215e-05, "loss": 0.9313, "step": 4110 }, { "epoch": 0.14, "learning_rate": 4.81741625055121e-05, "loss": 0.7256, "step": 4111 }, { "epoch": 0.14, "learning_rate": 4.817316360291917e-05, "loss": 0.5219, "step": 4112 }, { "epoch": 0.14, "learning_rate": 4.817216443751675e-05, "loss": 0.8497, "step": 4113 }, { "epoch": 0.14, "learning_rate": 4.817116500931617e-05, "loss": 0.5902, "step": 4114 }, { "epoch": 0.14, "learning_rate": 4.817016531832878e-05, "loss": 0.8165, "step": 4115 }, { "epoch": 0.14, "learning_rate": 4.816916536456589e-05, "loss": 0.8511, "step": 4116 }, { "epoch": 0.14, "learning_rate": 4.816816514803886e-05, "loss": 0.812, "step": 4117 }, { "epoch": 0.14, "learning_rate": 4.816716466875904e-05, "loss": 0.9505, "step": 4118 }, { "epoch": 0.14, "learning_rate": 4.816616392673775e-05, "loss": 1.151, "step": 4119 }, { "epoch": 0.14, "learning_rate": 4.8165162921986365e-05, "loss": 0.8944, "step": 4120 }, { "epoch": 0.14, "learning_rate": 4.8164161654516225e-05, "loss": 0.8972, "step": 4121 }, { "epoch": 0.14, "learning_rate": 4.8163160124338694e-05, "loss": 0.7696, "step": 4122 }, { "epoch": 0.14, "learning_rate": 4.8162158331465124e-05, "loss": 1.405, "step": 4123 }, { "epoch": 0.14, "learning_rate": 4.8161156275906876e-05, "loss": 1.0791, "step": 4124 }, { "epoch": 0.14, "learning_rate": 4.8160153957675316e-05, "loss": 0.9378, "step": 4125 }, { "epoch": 0.14, "learning_rate": 4.815915137678181e-05, "loss": 1.0594, "step": 4126 }, { "epoch": 0.14, "learning_rate": 4.8158148533237734e-05, "loss": 0.9318, "step": 4127 }, { "epoch": 0.14, "learning_rate": 4.815714542705445e-05, "loss": 1.2718, "step": 4128 }, { "epoch": 0.14, "learning_rate": 4.815614205824335e-05, "loss": 0.8289, "step": 4129 }, { "epoch": 0.14, "learning_rate": 4.81551384268158e-05, "loss": 0.8802, "step": 4130 }, { "epoch": 0.14, "learning_rate": 4.815413453278319e-05, "loss": 0.8704, "step": 4131 }, { "epoch": 0.14, "learning_rate": 4.8153130376156903e-05, "loss": 0.9666, "step": 4132 }, { "epoch": 0.14, "learning_rate": 4.8152125956948324e-05, "loss": 0.9304, "step": 4133 }, { "epoch": 0.14, "learning_rate": 4.815112127516884e-05, "loss": 0.8812, "step": 4134 }, { "epoch": 0.14, "learning_rate": 4.815011633082986e-05, "loss": 0.9369, "step": 4135 }, { "epoch": 0.14, "learning_rate": 4.814911112394278e-05, "loss": 0.8389, "step": 4136 }, { "epoch": 0.14, "learning_rate": 4.814810565451898e-05, "loss": 0.8391, "step": 4137 }, { "epoch": 0.14, "learning_rate": 4.814709992256989e-05, "loss": 1.267, "step": 4138 }, { "epoch": 0.14, "learning_rate": 4.814609392810689e-05, "loss": 0.8453, "step": 4139 }, { "epoch": 0.14, "learning_rate": 4.814508767114141e-05, "loss": 1.1551, "step": 4140 }, { "epoch": 0.14, "learning_rate": 4.814408115168485e-05, "loss": 1.0678, "step": 4141 }, { "epoch": 0.14, "learning_rate": 4.814307436974864e-05, "loss": 0.9509, "step": 4142 }, { "epoch": 0.14, "learning_rate": 4.8142067325344173e-05, "loss": 0.9266, "step": 4143 }, { "epoch": 0.14, "learning_rate": 4.8141060018482885e-05, "loss": 0.8654, "step": 4144 }, { "epoch": 0.14, "learning_rate": 4.814005244917621e-05, "loss": 4.2232, "step": 4145 }, { "epoch": 0.14, "learning_rate": 4.813904461743556e-05, "loss": 1.4498, "step": 4146 }, { "epoch": 0.14, "learning_rate": 4.8138036523272354e-05, "loss": 0.8102, "step": 4147 }, { "epoch": 0.14, "learning_rate": 4.813702816669806e-05, "loss": 1.0766, "step": 4148 }, { "epoch": 0.14, "learning_rate": 4.8136019547724084e-05, "loss": 0.6724, "step": 4149 }, { "epoch": 0.14, "learning_rate": 4.8135010666361876e-05, "loss": 0.6332, "step": 4150 }, { "epoch": 0.14, "learning_rate": 4.813400152262287e-05, "loss": 0.8446, "step": 4151 }, { "epoch": 0.14, "learning_rate": 4.813299211651853e-05, "loss": 0.7495, "step": 4152 }, { "epoch": 0.14, "learning_rate": 4.813198244806029e-05, "loss": 0.9731, "step": 4153 }, { "epoch": 0.14, "learning_rate": 4.813097251725959e-05, "loss": 0.8995, "step": 4154 }, { "epoch": 0.14, "learning_rate": 4.81299623241279e-05, "loss": 0.9297, "step": 4155 }, { "epoch": 0.14, "learning_rate": 4.812895186867668e-05, "loss": 0.7514, "step": 4156 }, { "epoch": 0.14, "learning_rate": 4.8127941150917374e-05, "loss": 0.9274, "step": 4157 }, { "epoch": 0.14, "learning_rate": 4.812693017086145e-05, "loss": 0.6826, "step": 4158 }, { "epoch": 0.14, "learning_rate": 4.8125918928520385e-05, "loss": 1.0307, "step": 4159 }, { "epoch": 0.14, "learning_rate": 4.812490742390564e-05, "loss": 0.7484, "step": 4160 }, { "epoch": 0.14, "learning_rate": 4.812389565702867e-05, "loss": 0.7903, "step": 4161 }, { "epoch": 0.14, "learning_rate": 4.8122883627900984e-05, "loss": 0.8515, "step": 4162 }, { "epoch": 0.14, "learning_rate": 4.812187133653403e-05, "loss": 0.9289, "step": 4163 }, { "epoch": 0.14, "learning_rate": 4.81208587829393e-05, "loss": 1.2928, "step": 4164 }, { "epoch": 0.14, "learning_rate": 4.8119845967128276e-05, "loss": 2.713, "step": 4165 }, { "epoch": 0.14, "learning_rate": 4.8118832889112445e-05, "loss": 3.6104, "step": 4166 }, { "epoch": 0.14, "learning_rate": 4.81178195489033e-05, "loss": 1.0339, "step": 4167 }, { "epoch": 0.14, "learning_rate": 4.8116805946512325e-05, "loss": 0.8405, "step": 4168 }, { "epoch": 0.14, "learning_rate": 4.811579208195103e-05, "loss": 0.8465, "step": 4169 }, { "epoch": 0.14, "learning_rate": 4.8114777955230896e-05, "loss": 1.05, "step": 4170 }, { "epoch": 0.14, "learning_rate": 4.811376356636343e-05, "loss": 1.078, "step": 4171 }, { "epoch": 0.14, "learning_rate": 4.811274891536014e-05, "loss": 1.0413, "step": 4172 }, { "epoch": 0.14, "learning_rate": 4.811173400223254e-05, "loss": 0.6058, "step": 4173 }, { "epoch": 0.14, "learning_rate": 4.8110718826992116e-05, "loss": 0.8137, "step": 4174 }, { "epoch": 0.14, "learning_rate": 4.81097033896504e-05, "loss": 0.7608, "step": 4175 }, { "epoch": 0.14, "learning_rate": 4.810868769021891e-05, "loss": 0.8549, "step": 4176 }, { "epoch": 0.14, "learning_rate": 4.810767172870916e-05, "loss": 0.7293, "step": 4177 }, { "epoch": 0.14, "learning_rate": 4.8106655505132674e-05, "loss": 0.8811, "step": 4178 }, { "epoch": 0.14, "learning_rate": 4.810563901950098e-05, "loss": 0.723, "step": 4179 }, { "epoch": 0.14, "learning_rate": 4.810462227182559e-05, "loss": 0.9379, "step": 4180 }, { "epoch": 0.14, "learning_rate": 4.810360526211806e-05, "loss": 0.8521, "step": 4181 }, { "epoch": 0.14, "learning_rate": 4.81025879903899e-05, "loss": 0.9222, "step": 4182 }, { "epoch": 0.14, "learning_rate": 4.810157045665266e-05, "loss": 0.8602, "step": 4183 }, { "epoch": 0.14, "learning_rate": 4.8100552660917874e-05, "loss": 0.8972, "step": 4184 }, { "epoch": 0.14, "learning_rate": 4.8099534603197094e-05, "loss": 0.9204, "step": 4185 }, { "epoch": 0.14, "learning_rate": 4.809851628350186e-05, "loss": 0.8868, "step": 4186 }, { "epoch": 0.14, "learning_rate": 4.809749770184372e-05, "loss": 0.8799, "step": 4187 }, { "epoch": 0.14, "learning_rate": 4.809647885823423e-05, "loss": 0.6785, "step": 4188 }, { "epoch": 0.14, "learning_rate": 4.809545975268494e-05, "loss": 0.8702, "step": 4189 }, { "epoch": 0.14, "learning_rate": 4.8094440385207404e-05, "loss": 0.8304, "step": 4190 }, { "epoch": 0.14, "learning_rate": 4.80934207558132e-05, "loss": 0.6499, "step": 4191 }, { "epoch": 0.14, "learning_rate": 4.809240086451388e-05, "loss": 0.7228, "step": 4192 }, { "epoch": 0.14, "learning_rate": 4.8091380711321e-05, "loss": 0.2514, "step": 4193 }, { "epoch": 0.14, "learning_rate": 4.809036029624615e-05, "loss": 0.9013, "step": 4194 }, { "epoch": 0.14, "learning_rate": 4.808933961930089e-05, "loss": 0.9461, "step": 4195 }, { "epoch": 0.14, "learning_rate": 4.808831868049679e-05, "loss": 0.7579, "step": 4196 }, { "epoch": 0.14, "learning_rate": 4.8087297479845453e-05, "loss": 0.8525, "step": 4197 }, { "epoch": 0.14, "learning_rate": 4.808627601735843e-05, "loss": 0.7775, "step": 4198 }, { "epoch": 0.14, "learning_rate": 4.808525429304733e-05, "loss": 0.8058, "step": 4199 }, { "epoch": 0.14, "learning_rate": 4.808423230692374e-05, "loss": 0.7805, "step": 4200 }, { "epoch": 0.14, "learning_rate": 4.808321005899923e-05, "loss": 1.4327, "step": 4201 }, { "epoch": 0.14, "learning_rate": 4.8082187549285406e-05, "loss": 0.7961, "step": 4202 }, { "epoch": 0.14, "learning_rate": 4.808116477779386e-05, "loss": 0.7898, "step": 4203 }, { "epoch": 0.14, "learning_rate": 4.80801417445362e-05, "loss": 1.3829, "step": 4204 }, { "epoch": 0.14, "learning_rate": 4.8079118449524025e-05, "loss": 0.5216, "step": 4205 }, { "epoch": 0.14, "learning_rate": 4.807809489276893e-05, "loss": 0.4047, "step": 4206 }, { "epoch": 0.14, "learning_rate": 4.807707107428254e-05, "loss": 0.7978, "step": 4207 }, { "epoch": 0.14, "learning_rate": 4.8076046994076455e-05, "loss": 0.8087, "step": 4208 }, { "epoch": 0.14, "learning_rate": 4.8075022652162294e-05, "loss": 0.6892, "step": 4209 }, { "epoch": 0.14, "learning_rate": 4.807399804855167e-05, "loss": 1.0262, "step": 4210 }, { "epoch": 0.14, "learning_rate": 4.8072973183256206e-05, "loss": 0.8931, "step": 4211 }, { "epoch": 0.14, "learning_rate": 4.807194805628752e-05, "loss": 0.7229, "step": 4212 }, { "epoch": 0.14, "learning_rate": 4.8070922667657255e-05, "loss": 0.8999, "step": 4213 }, { "epoch": 0.14, "learning_rate": 4.806989701737702e-05, "loss": 1.1059, "step": 4214 }, { "epoch": 0.14, "learning_rate": 4.806887110545845e-05, "loss": 0.9635, "step": 4215 }, { "epoch": 0.14, "learning_rate": 4.806784493191319e-05, "loss": 0.6257, "step": 4216 }, { "epoch": 0.14, "learning_rate": 4.8066818496752875e-05, "loss": 1.1222, "step": 4217 }, { "epoch": 0.14, "learning_rate": 4.8065791799989136e-05, "loss": 1.0651, "step": 4218 }, { "epoch": 0.14, "learning_rate": 4.8064764841633637e-05, "loss": 1.3645, "step": 4219 }, { "epoch": 0.14, "learning_rate": 4.8063737621698e-05, "loss": 0.9834, "step": 4220 }, { "epoch": 0.14, "learning_rate": 4.806271014019389e-05, "loss": 1.2408, "step": 4221 }, { "epoch": 0.14, "learning_rate": 4.8061682397132954e-05, "loss": 1.0722, "step": 4222 }, { "epoch": 0.14, "learning_rate": 4.806065439252686e-05, "loss": 0.9918, "step": 4223 }, { "epoch": 0.14, "learning_rate": 4.805962612638725e-05, "loss": 0.9011, "step": 4224 }, { "epoch": 0.14, "learning_rate": 4.8058597598725794e-05, "loss": 0.8754, "step": 4225 }, { "epoch": 0.14, "learning_rate": 4.805756880955416e-05, "loss": 0.966, "step": 4226 }, { "epoch": 0.14, "learning_rate": 4.8056539758884014e-05, "loss": 0.8727, "step": 4227 }, { "epoch": 0.14, "learning_rate": 4.8055510446727014e-05, "loss": 1.027, "step": 4228 }, { "epoch": 0.14, "learning_rate": 4.805448087309485e-05, "loss": 1.1517, "step": 4229 }, { "epoch": 0.14, "learning_rate": 4.805345103799919e-05, "loss": 2.1788, "step": 4230 }, { "epoch": 0.14, "learning_rate": 4.805242094145171e-05, "loss": 0.7272, "step": 4231 }, { "epoch": 0.14, "learning_rate": 4.805139058346411e-05, "loss": 0.9049, "step": 4232 }, { "epoch": 0.14, "learning_rate": 4.805035996404805e-05, "loss": 0.8789, "step": 4233 }, { "epoch": 0.14, "learning_rate": 4.804932908321524e-05, "loss": 0.7773, "step": 4234 }, { "epoch": 0.14, "learning_rate": 4.804829794097737e-05, "loss": 1.0846, "step": 4235 }, { "epoch": 0.14, "learning_rate": 4.804726653734611e-05, "loss": 1.1673, "step": 4236 }, { "epoch": 0.14, "learning_rate": 4.8046234872333185e-05, "loss": 1.0751, "step": 4237 }, { "epoch": 0.14, "learning_rate": 4.8045202945950286e-05, "loss": 1.0438, "step": 4238 }, { "epoch": 0.14, "learning_rate": 4.804417075820911e-05, "loss": 1.1478, "step": 4239 }, { "epoch": 0.14, "learning_rate": 4.804313830912137e-05, "loss": 0.9488, "step": 4240 }, { "epoch": 0.14, "learning_rate": 4.804210559869877e-05, "loss": 1.4516, "step": 4241 }, { "epoch": 0.14, "learning_rate": 4.804107262695303e-05, "loss": 0.9643, "step": 4242 }, { "epoch": 0.14, "learning_rate": 4.804003939389586e-05, "loss": 0.5366, "step": 4243 }, { "epoch": 0.14, "learning_rate": 4.803900589953898e-05, "loss": 0.7314, "step": 4244 }, { "epoch": 0.14, "learning_rate": 4.8037972143894106e-05, "loss": 1.1103, "step": 4245 }, { "epoch": 0.14, "learning_rate": 4.803693812697297e-05, "loss": 0.8748, "step": 4246 }, { "epoch": 0.14, "learning_rate": 4.803590384878729e-05, "loss": 0.9345, "step": 4247 }, { "epoch": 0.14, "learning_rate": 4.8034869309348804e-05, "loss": 0.9074, "step": 4248 }, { "epoch": 0.14, "learning_rate": 4.803383450866923e-05, "loss": 0.9702, "step": 4249 }, { "epoch": 0.14, "learning_rate": 4.803279944676032e-05, "loss": 0.9175, "step": 4250 }, { "epoch": 0.14, "learning_rate": 4.8031764123633815e-05, "loss": 0.6787, "step": 4251 }, { "epoch": 0.14, "learning_rate": 4.803072853930145e-05, "loss": 0.458, "step": 4252 }, { "epoch": 0.14, "learning_rate": 4.802969269377496e-05, "loss": 0.4374, "step": 4253 }, { "epoch": 0.14, "learning_rate": 4.8028656587066104e-05, "loss": 0.6044, "step": 4254 }, { "epoch": 0.14, "learning_rate": 4.802762021918663e-05, "loss": 1.1239, "step": 4255 }, { "epoch": 0.14, "learning_rate": 4.80265835901483e-05, "loss": 0.956, "step": 4256 }, { "epoch": 0.14, "learning_rate": 4.802554669996285e-05, "loss": 4.1288, "step": 4257 }, { "epoch": 0.14, "learning_rate": 4.8024509548642056e-05, "loss": 0.6109, "step": 4258 }, { "epoch": 0.14, "learning_rate": 4.8023472136197675e-05, "loss": 0.961, "step": 4259 }, { "epoch": 0.14, "learning_rate": 4.802243446264148e-05, "loss": 1.2848, "step": 4260 }, { "epoch": 0.14, "learning_rate": 4.8021396527985224e-05, "loss": 0.9018, "step": 4261 }, { "epoch": 0.14, "learning_rate": 4.802035833224069e-05, "loss": 0.9146, "step": 4262 }, { "epoch": 0.14, "learning_rate": 4.801931987541965e-05, "loss": 0.5353, "step": 4263 }, { "epoch": 0.14, "learning_rate": 4.8018281157533887e-05, "loss": 0.462, "step": 4264 }, { "epoch": 0.14, "learning_rate": 4.801724217859517e-05, "loss": 0.5353, "step": 4265 }, { "epoch": 0.14, "learning_rate": 4.801620293861528e-05, "loss": 1.0096, "step": 4266 }, { "epoch": 0.14, "learning_rate": 4.8015163437606026e-05, "loss": 0.9852, "step": 4267 }, { "epoch": 0.14, "learning_rate": 4.801412367557917e-05, "loss": 0.9689, "step": 4268 }, { "epoch": 0.14, "learning_rate": 4.801308365254652e-05, "loss": 0.973, "step": 4269 }, { "epoch": 0.14, "learning_rate": 4.801204336851986e-05, "loss": 0.4597, "step": 4270 }, { "epoch": 0.14, "learning_rate": 4.801100282351101e-05, "loss": 0.7203, "step": 4271 }, { "epoch": 0.14, "learning_rate": 4.800996201753174e-05, "loss": 1.729, "step": 4272 }, { "epoch": 0.14, "learning_rate": 4.800892095059388e-05, "loss": 2.0579, "step": 4273 }, { "epoch": 0.14, "learning_rate": 4.8007879622709215e-05, "loss": 0.88, "step": 4274 }, { "epoch": 0.14, "learning_rate": 4.800683803388958e-05, "loss": 0.8338, "step": 4275 }, { "epoch": 0.14, "learning_rate": 4.800579618414676e-05, "loss": 0.9316, "step": 4276 }, { "epoch": 0.14, "learning_rate": 4.800475407349259e-05, "loss": 1.0156, "step": 4277 }, { "epoch": 0.14, "learning_rate": 4.8003711701938886e-05, "loss": 0.7925, "step": 4278 }, { "epoch": 0.14, "learning_rate": 4.800266906949746e-05, "loss": 0.1117, "step": 4279 }, { "epoch": 0.14, "learning_rate": 4.800162617618015e-05, "loss": 0.9049, "step": 4280 }, { "epoch": 0.14, "learning_rate": 4.8000583021998776e-05, "loss": 0.7091, "step": 4281 }, { "epoch": 0.14, "learning_rate": 4.799953960696517e-05, "loss": 0.6322, "step": 4282 }, { "epoch": 0.14, "learning_rate": 4.7998495931091166e-05, "loss": 0.9455, "step": 4283 }, { "epoch": 0.14, "learning_rate": 4.79974519943886e-05, "loss": 0.9485, "step": 4284 }, { "epoch": 0.14, "learning_rate": 4.7996407796869304e-05, "loss": 0.8494, "step": 4285 }, { "epoch": 0.14, "learning_rate": 4.7995363338545134e-05, "loss": 0.3396, "step": 4286 }, { "epoch": 0.14, "learning_rate": 4.799431861942792e-05, "loss": 0.4347, "step": 4287 }, { "epoch": 0.14, "learning_rate": 4.799327363952953e-05, "loss": 1.1285, "step": 4288 }, { "epoch": 0.14, "learning_rate": 4.7992228398861805e-05, "loss": 0.8879, "step": 4289 }, { "epoch": 0.14, "learning_rate": 4.799118289743659e-05, "loss": 1.1483, "step": 4290 }, { "epoch": 0.14, "learning_rate": 4.7990137135265755e-05, "loss": 1.048, "step": 4291 }, { "epoch": 0.14, "learning_rate": 4.798909111236115e-05, "loss": 0.9842, "step": 4292 }, { "epoch": 0.14, "learning_rate": 4.798804482873465e-05, "loss": 1.0858, "step": 4293 }, { "epoch": 0.14, "learning_rate": 4.7986998284398116e-05, "loss": 1.0002, "step": 4294 }, { "epoch": 0.14, "learning_rate": 4.798595147936341e-05, "loss": 0.7879, "step": 4295 }, { "epoch": 0.14, "learning_rate": 4.798490441364242e-05, "loss": 0.758, "step": 4296 }, { "epoch": 0.14, "learning_rate": 4.7983857087247e-05, "loss": 1.1087, "step": 4297 }, { "epoch": 0.14, "learning_rate": 4.7982809500189044e-05, "loss": 0.9275, "step": 4298 }, { "epoch": 0.14, "learning_rate": 4.798176165248042e-05, "loss": 1.0006, "step": 4299 }, { "epoch": 0.14, "learning_rate": 4.798071354413302e-05, "loss": 0.9067, "step": 4300 }, { "epoch": 0.14, "learning_rate": 4.797966517515874e-05, "loss": 0.785, "step": 4301 }, { "epoch": 0.14, "learning_rate": 4.797861654556944e-05, "loss": 0.9986, "step": 4302 }, { "epoch": 0.14, "learning_rate": 4.7977567655377055e-05, "loss": 1.1186, "step": 4303 }, { "epoch": 0.14, "learning_rate": 4.7976518504593444e-05, "loss": 0.6517, "step": 4304 }, { "epoch": 0.14, "learning_rate": 4.797546909323052e-05, "loss": 4.9957, "step": 4305 }, { "epoch": 0.14, "learning_rate": 4.7974419421300186e-05, "loss": 5.8123, "step": 4306 }, { "epoch": 0.14, "learning_rate": 4.7973369488814344e-05, "loss": 3.1839, "step": 4307 }, { "epoch": 0.14, "learning_rate": 4.79723192957849e-05, "loss": 0.7826, "step": 4308 }, { "epoch": 0.14, "learning_rate": 4.797126884222377e-05, "loss": 1.0789, "step": 4309 }, { "epoch": 0.14, "learning_rate": 4.797021812814286e-05, "loss": 0.7996, "step": 4310 }, { "epoch": 0.14, "learning_rate": 4.796916715355408e-05, "loss": 0.9485, "step": 4311 }, { "epoch": 0.14, "learning_rate": 4.796811591846937e-05, "loss": 0.4947, "step": 4312 }, { "epoch": 0.14, "learning_rate": 4.7967064422900645e-05, "loss": 0.9325, "step": 4313 }, { "epoch": 0.14, "learning_rate": 4.796601266685982e-05, "loss": 2.0378, "step": 4314 }, { "epoch": 0.14, "learning_rate": 4.7964960650358825e-05, "loss": 1.005, "step": 4315 }, { "epoch": 0.14, "learning_rate": 4.7963908373409594e-05, "loss": 0.9236, "step": 4316 }, { "epoch": 0.14, "learning_rate": 4.7962855836024065e-05, "loss": 0.9627, "step": 4317 }, { "epoch": 0.14, "learning_rate": 4.796180303821418e-05, "loss": 1.0042, "step": 4318 }, { "epoch": 0.14, "learning_rate": 4.796074997999186e-05, "loss": 1.0873, "step": 4319 }, { "epoch": 0.14, "learning_rate": 4.795969666136907e-05, "loss": 0.6016, "step": 4320 }, { "epoch": 0.14, "learning_rate": 4.7958643082357735e-05, "loss": 0.516, "step": 4321 }, { "epoch": 0.14, "learning_rate": 4.7957589242969815e-05, "loss": 0.8863, "step": 4322 }, { "epoch": 0.14, "learning_rate": 4.795653514321726e-05, "loss": 0.8016, "step": 4323 }, { "epoch": 0.14, "learning_rate": 4.795548078311203e-05, "loss": 0.9817, "step": 4324 }, { "epoch": 0.14, "learning_rate": 4.7954426162666065e-05, "loss": 0.9778, "step": 4325 }, { "epoch": 0.14, "learning_rate": 4.7953371281891346e-05, "loss": 0.8507, "step": 4326 }, { "epoch": 0.14, "learning_rate": 4.795231614079983e-05, "loss": 0.8128, "step": 4327 }, { "epoch": 0.14, "learning_rate": 4.795126073940348e-05, "loss": 0.8478, "step": 4328 }, { "epoch": 0.14, "learning_rate": 4.7950205077714264e-05, "loss": 0.9618, "step": 4329 }, { "epoch": 0.14, "learning_rate": 4.7949149155744164e-05, "loss": 0.8911, "step": 4330 }, { "epoch": 0.14, "learning_rate": 4.7948092973505145e-05, "loss": 0.5128, "step": 4331 }, { "epoch": 0.14, "learning_rate": 4.7947036531009195e-05, "loss": 0.7965, "step": 4332 }, { "epoch": 0.14, "learning_rate": 4.794597982826828e-05, "loss": 0.8693, "step": 4333 }, { "epoch": 0.14, "learning_rate": 4.79449228652944e-05, "loss": 0.8612, "step": 4334 }, { "epoch": 0.14, "learning_rate": 4.794386564209953e-05, "loss": 0.5343, "step": 4335 }, { "epoch": 0.14, "learning_rate": 4.7942808158695675e-05, "loss": 0.6844, "step": 4336 }, { "epoch": 0.14, "learning_rate": 4.794175041509481e-05, "loss": 1.4195, "step": 4337 }, { "epoch": 0.14, "learning_rate": 4.794069241130894e-05, "loss": 1.0245, "step": 4338 }, { "epoch": 0.14, "learning_rate": 4.793963414735007e-05, "loss": 0.8701, "step": 4339 }, { "epoch": 0.14, "learning_rate": 4.79385756232302e-05, "loss": 0.8541, "step": 4340 }, { "epoch": 0.14, "learning_rate": 4.7937516838961316e-05, "loss": 1.037, "step": 4341 }, { "epoch": 0.14, "learning_rate": 4.793645779455545e-05, "loss": 0.3547, "step": 4342 }, { "epoch": 0.14, "learning_rate": 4.79353984900246e-05, "loss": 1.1691, "step": 4343 }, { "epoch": 0.14, "learning_rate": 4.793433892538078e-05, "loss": 0.7357, "step": 4344 }, { "epoch": 0.14, "learning_rate": 4.793327910063602e-05, "loss": 0.9279, "step": 4345 }, { "epoch": 0.14, "learning_rate": 4.793221901580233e-05, "loss": 0.5917, "step": 4346 }, { "epoch": 0.14, "learning_rate": 4.793115867089172e-05, "loss": 0.5167, "step": 4347 }, { "epoch": 0.14, "learning_rate": 4.793009806591623e-05, "loss": 1.1391, "step": 4348 }, { "epoch": 0.14, "learning_rate": 4.7929037200887886e-05, "loss": 1.0568, "step": 4349 }, { "epoch": 0.14, "learning_rate": 4.792797607581872e-05, "loss": 1.0088, "step": 4350 }, { "epoch": 0.15, "learning_rate": 4.792691469072077e-05, "loss": 0.9033, "step": 4351 }, { "epoch": 0.15, "learning_rate": 4.792585304560606e-05, "loss": 1.1568, "step": 4352 }, { "epoch": 0.15, "learning_rate": 4.792479114048665e-05, "loss": 5.9868, "step": 4353 }, { "epoch": 0.15, "learning_rate": 4.792372897537456e-05, "loss": 5.8805, "step": 4354 }, { "epoch": 0.15, "learning_rate": 4.792266655028185e-05, "loss": 5.8527, "step": 4355 }, { "epoch": 0.15, "learning_rate": 4.7921603865220575e-05, "loss": 4.9089, "step": 4356 }, { "epoch": 0.15, "learning_rate": 4.792054092020277e-05, "loss": 0.9151, "step": 4357 }, { "epoch": 0.15, "learning_rate": 4.79194777152405e-05, "loss": 0.9125, "step": 4358 }, { "epoch": 0.15, "learning_rate": 4.7918414250345836e-05, "loss": 1.0488, "step": 4359 }, { "epoch": 0.15, "learning_rate": 4.791735052553081e-05, "loss": 0.9973, "step": 4360 }, { "epoch": 0.15, "learning_rate": 4.791628654080751e-05, "loss": 0.8998, "step": 4361 }, { "epoch": 0.15, "learning_rate": 4.7915222296187986e-05, "loss": 1.2094, "step": 4362 }, { "epoch": 0.15, "learning_rate": 4.7914157791684316e-05, "loss": 1.0264, "step": 4363 }, { "epoch": 0.15, "learning_rate": 4.7913093027308574e-05, "loss": 0.921, "step": 4364 }, { "epoch": 0.15, "learning_rate": 4.791202800307284e-05, "loss": 0.7918, "step": 4365 }, { "epoch": 0.15, "learning_rate": 4.791096271898918e-05, "loss": 0.7803, "step": 4366 }, { "epoch": 0.15, "learning_rate": 4.790989717506968e-05, "loss": 0.9231, "step": 4367 }, { "epoch": 0.15, "learning_rate": 4.7908831371326426e-05, "loss": 0.8123, "step": 4368 }, { "epoch": 0.15, "learning_rate": 4.7907765307771515e-05, "loss": 0.8515, "step": 4369 }, { "epoch": 0.15, "learning_rate": 4.790669898441702e-05, "loss": 0.8797, "step": 4370 }, { "epoch": 0.15, "learning_rate": 4.790563240127504e-05, "loss": 0.7787, "step": 4371 }, { "epoch": 0.15, "learning_rate": 4.790456555835768e-05, "loss": 0.7064, "step": 4372 }, { "epoch": 0.15, "learning_rate": 4.790349845567703e-05, "loss": 0.8989, "step": 4373 }, { "epoch": 0.15, "learning_rate": 4.790243109324519e-05, "loss": 0.8157, "step": 4374 }, { "epoch": 0.15, "learning_rate": 4.790136347107427e-05, "loss": 0.7071, "step": 4375 }, { "epoch": 0.15, "learning_rate": 4.7900295589176385e-05, "loss": 0.6285, "step": 4376 }, { "epoch": 0.15, "learning_rate": 4.7899227447563635e-05, "loss": 0.7536, "step": 4377 }, { "epoch": 0.15, "learning_rate": 4.789815904624814e-05, "loss": 0.8425, "step": 4378 }, { "epoch": 0.15, "learning_rate": 4.7897090385242014e-05, "loss": 1.0646, "step": 4379 }, { "epoch": 0.15, "learning_rate": 4.7896021464557376e-05, "loss": 1.039, "step": 4380 }, { "epoch": 0.15, "learning_rate": 4.789495228420635e-05, "loss": 1.3638, "step": 4381 }, { "epoch": 0.15, "learning_rate": 4.789388284420107e-05, "loss": 1.2435, "step": 4382 }, { "epoch": 0.15, "learning_rate": 4.789281314455364e-05, "loss": 0.8185, "step": 4383 }, { "epoch": 0.15, "learning_rate": 4.789174318527622e-05, "loss": 0.8155, "step": 4384 }, { "epoch": 0.15, "learning_rate": 4.789067296638093e-05, "loss": 0.7119, "step": 4385 }, { "epoch": 0.15, "learning_rate": 4.788960248787991e-05, "loss": 0.5637, "step": 4386 }, { "epoch": 0.15, "learning_rate": 4.7888531749785303e-05, "loss": 0.7848, "step": 4387 }, { "epoch": 0.15, "learning_rate": 4.788746075210925e-05, "loss": 0.8927, "step": 4388 }, { "epoch": 0.15, "learning_rate": 4.7886389494863895e-05, "loss": 1.0297, "step": 4389 }, { "epoch": 0.15, "learning_rate": 4.788531797806138e-05, "loss": 1.0111, "step": 4390 }, { "epoch": 0.15, "learning_rate": 4.788424620171388e-05, "loss": 0.7495, "step": 4391 }, { "epoch": 0.15, "learning_rate": 4.788317416583353e-05, "loss": 0.4018, "step": 4392 }, { "epoch": 0.15, "learning_rate": 4.788210187043249e-05, "loss": 0.8648, "step": 4393 }, { "epoch": 0.15, "learning_rate": 4.788102931552294e-05, "loss": 2.4222, "step": 4394 }, { "epoch": 0.15, "learning_rate": 4.7879956501117026e-05, "loss": 3.7377, "step": 4395 }, { "epoch": 0.15, "learning_rate": 4.787888342722692e-05, "loss": 2.075, "step": 4396 }, { "epoch": 0.15, "learning_rate": 4.787781009386478e-05, "loss": 0.8313, "step": 4397 }, { "epoch": 0.15, "learning_rate": 4.78767365010428e-05, "loss": 1.0574, "step": 4398 }, { "epoch": 0.15, "learning_rate": 4.787566264877313e-05, "loss": 1.1169, "step": 4399 }, { "epoch": 0.15, "learning_rate": 4.787458853706798e-05, "loss": 0.8625, "step": 4400 }, { "epoch": 0.15, "learning_rate": 4.7873514165939514e-05, "loss": 0.918, "step": 4401 }, { "epoch": 0.15, "learning_rate": 4.7872439535399916e-05, "loss": 0.8696, "step": 4402 }, { "epoch": 0.15, "learning_rate": 4.787136464546137e-05, "loss": 0.9802, "step": 4403 }, { "epoch": 0.15, "learning_rate": 4.7870289496136076e-05, "loss": 0.8003, "step": 4404 }, { "epoch": 0.15, "learning_rate": 4.786921408743622e-05, "loss": 0.7497, "step": 4405 }, { "epoch": 0.15, "learning_rate": 4.786813841937401e-05, "loss": 0.907, "step": 4406 }, { "epoch": 0.15, "learning_rate": 4.7867062491961625e-05, "loss": 0.9459, "step": 4407 }, { "epoch": 0.15, "learning_rate": 4.7865986305211294e-05, "loss": 0.7436, "step": 4408 }, { "epoch": 0.15, "learning_rate": 4.78649098591352e-05, "loss": 1.1676, "step": 4409 }, { "epoch": 0.15, "learning_rate": 4.786383315374555e-05, "loss": 0.69, "step": 4410 }, { "epoch": 0.15, "learning_rate": 4.7862756189054574e-05, "loss": 1.1161, "step": 4411 }, { "epoch": 0.15, "learning_rate": 4.786167896507447e-05, "loss": 0.8509, "step": 4412 }, { "epoch": 0.15, "learning_rate": 4.786060148181747e-05, "loss": 1.1413, "step": 4413 }, { "epoch": 0.15, "learning_rate": 4.785952373929578e-05, "loss": 0.8771, "step": 4414 }, { "epoch": 0.15, "learning_rate": 4.7858445737521615e-05, "loss": 0.7599, "step": 4415 }, { "epoch": 0.15, "learning_rate": 4.7857367476507224e-05, "loss": 0.9709, "step": 4416 }, { "epoch": 0.15, "learning_rate": 4.785628895626483e-05, "loss": 1.5571, "step": 4417 }, { "epoch": 0.15, "learning_rate": 4.7855210176806645e-05, "loss": 2.0498, "step": 4418 }, { "epoch": 0.15, "learning_rate": 4.7854131138144926e-05, "loss": 0.9209, "step": 4419 }, { "epoch": 0.15, "learning_rate": 4.7853051840291904e-05, "loss": 0.9296, "step": 4420 }, { "epoch": 0.15, "learning_rate": 4.785197228325981e-05, "loss": 0.9072, "step": 4421 }, { "epoch": 0.15, "learning_rate": 4.78508924670609e-05, "loss": 1.8549, "step": 4422 }, { "epoch": 0.15, "learning_rate": 4.7849812391707415e-05, "loss": 1.223, "step": 4423 }, { "epoch": 0.15, "learning_rate": 4.7848732057211596e-05, "loss": 0.85, "step": 4424 }, { "epoch": 0.15, "learning_rate": 4.784765146358572e-05, "loss": 1.0511, "step": 4425 }, { "epoch": 0.15, "learning_rate": 4.784657061084201e-05, "loss": 0.6633, "step": 4426 }, { "epoch": 0.15, "learning_rate": 4.784548949899275e-05, "loss": 0.9983, "step": 4427 }, { "epoch": 0.15, "learning_rate": 4.7844408128050186e-05, "loss": 0.9559, "step": 4428 }, { "epoch": 0.15, "learning_rate": 4.7843326498026595e-05, "loss": 0.9327, "step": 4429 }, { "epoch": 0.15, "learning_rate": 4.784224460893423e-05, "loss": 0.8954, "step": 4430 }, { "epoch": 0.15, "learning_rate": 4.784116246078537e-05, "loss": 0.2745, "step": 4431 }, { "epoch": 0.15, "learning_rate": 4.784008005359227e-05, "loss": 0.8212, "step": 4432 }, { "epoch": 0.15, "learning_rate": 4.7838997387367234e-05, "loss": 0.7087, "step": 4433 }, { "epoch": 0.15, "learning_rate": 4.7837914462122524e-05, "loss": 0.7295, "step": 4434 }, { "epoch": 0.15, "learning_rate": 4.783683127787043e-05, "loss": 0.7189, "step": 4435 }, { "epoch": 0.15, "learning_rate": 4.7835747834623225e-05, "loss": 0.9472, "step": 4436 }, { "epoch": 0.15, "learning_rate": 4.78346641323932e-05, "loss": 0.6789, "step": 4437 }, { "epoch": 0.15, "learning_rate": 4.783358017119265e-05, "loss": 0.8451, "step": 4438 }, { "epoch": 0.15, "learning_rate": 4.783249595103386e-05, "loss": 0.4961, "step": 4439 }, { "epoch": 0.15, "learning_rate": 4.7831411471929144e-05, "loss": 1.4585, "step": 4440 }, { "epoch": 0.15, "learning_rate": 4.7830326733890785e-05, "loss": 0.7109, "step": 4441 }, { "epoch": 0.15, "learning_rate": 4.782924173693108e-05, "loss": 0.51, "step": 4442 }, { "epoch": 0.15, "learning_rate": 4.782815648106236e-05, "loss": 1.0026, "step": 4443 }, { "epoch": 0.15, "learning_rate": 4.782707096629691e-05, "loss": 0.7924, "step": 4444 }, { "epoch": 0.15, "learning_rate": 4.7825985192647044e-05, "loss": 0.9582, "step": 4445 }, { "epoch": 0.15, "learning_rate": 4.782489916012508e-05, "loss": 1.2476, "step": 4446 }, { "epoch": 0.15, "learning_rate": 4.782381286874334e-05, "loss": 2.3179, "step": 4447 }, { "epoch": 0.15, "learning_rate": 4.782272631851413e-05, "loss": 0.7008, "step": 4448 }, { "epoch": 0.15, "learning_rate": 4.782163950944979e-05, "loss": 0.7948, "step": 4449 }, { "epoch": 0.15, "learning_rate": 4.782055244156263e-05, "loss": 0.7429, "step": 4450 }, { "epoch": 0.15, "learning_rate": 4.781946511486499e-05, "loss": 0.6637, "step": 4451 }, { "epoch": 0.15, "learning_rate": 4.781837752936919e-05, "loss": 0.8773, "step": 4452 }, { "epoch": 0.15, "learning_rate": 4.7817289685087577e-05, "loss": 0.9023, "step": 4453 }, { "epoch": 0.15, "learning_rate": 4.781620158203247e-05, "loss": 0.6541, "step": 4454 }, { "epoch": 0.15, "learning_rate": 4.781511322021623e-05, "loss": 0.8161, "step": 4455 }, { "epoch": 0.15, "learning_rate": 4.7814024599651194e-05, "loss": 1.12, "step": 4456 }, { "epoch": 0.15, "learning_rate": 4.7812935720349704e-05, "loss": 0.924, "step": 4457 }, { "epoch": 0.15, "learning_rate": 4.7811846582324114e-05, "loss": 0.8385, "step": 4458 }, { "epoch": 0.15, "learning_rate": 4.781075718558678e-05, "loss": 1.3023, "step": 4459 }, { "epoch": 0.15, "learning_rate": 4.7809667530150035e-05, "loss": 1.0204, "step": 4460 }, { "epoch": 0.15, "learning_rate": 4.780857761602626e-05, "loss": 0.8246, "step": 4461 }, { "epoch": 0.15, "learning_rate": 4.78074874432278e-05, "loss": 1.3124, "step": 4462 }, { "epoch": 0.15, "learning_rate": 4.780639701176703e-05, "loss": 1.5613, "step": 4463 }, { "epoch": 0.15, "learning_rate": 4.780530632165632e-05, "loss": 1.3839, "step": 4464 }, { "epoch": 0.15, "learning_rate": 4.7804215372908023e-05, "loss": 0.9703, "step": 4465 }, { "epoch": 0.15, "learning_rate": 4.780312416553453e-05, "loss": 0.4951, "step": 4466 }, { "epoch": 0.15, "learning_rate": 4.78020326995482e-05, "loss": 0.4247, "step": 4467 }, { "epoch": 0.15, "learning_rate": 4.7800940974961425e-05, "loss": 1.1975, "step": 4468 }, { "epoch": 0.15, "learning_rate": 4.779984899178659e-05, "loss": 0.9035, "step": 4469 }, { "epoch": 0.15, "learning_rate": 4.7798756750036054e-05, "loss": 0.6131, "step": 4470 }, { "epoch": 0.15, "learning_rate": 4.7797664249722226e-05, "loss": 0.9192, "step": 4471 }, { "epoch": 0.15, "learning_rate": 4.7796571490857493e-05, "loss": 0.9363, "step": 4472 }, { "epoch": 0.15, "learning_rate": 4.779547847345425e-05, "loss": 1.0036, "step": 4473 }, { "epoch": 0.15, "learning_rate": 4.7794385197524875e-05, "loss": 0.8573, "step": 4474 }, { "epoch": 0.15, "learning_rate": 4.779329166308179e-05, "loss": 1.0044, "step": 4475 }, { "epoch": 0.15, "learning_rate": 4.779219787013739e-05, "loss": 0.6674, "step": 4476 }, { "epoch": 0.15, "learning_rate": 4.779110381870407e-05, "loss": 0.5165, "step": 4477 }, { "epoch": 0.15, "learning_rate": 4.7790009508794246e-05, "loss": 0.7231, "step": 4478 }, { "epoch": 0.15, "learning_rate": 4.778891494042033e-05, "loss": 0.692, "step": 4479 }, { "epoch": 0.15, "learning_rate": 4.778782011359474e-05, "loss": 0.8172, "step": 4480 }, { "epoch": 0.15, "learning_rate": 4.7786725028329874e-05, "loss": 0.9233, "step": 4481 }, { "epoch": 0.15, "learning_rate": 4.778562968463817e-05, "loss": 1.2419, "step": 4482 }, { "epoch": 0.15, "learning_rate": 4.778453408253205e-05, "loss": 1.175, "step": 4483 }, { "epoch": 0.15, "learning_rate": 4.778343822202392e-05, "loss": 1.0361, "step": 4484 }, { "epoch": 0.15, "learning_rate": 4.7782342103126234e-05, "loss": 0.751, "step": 4485 }, { "epoch": 0.15, "learning_rate": 4.77812457258514e-05, "loss": 1.3682, "step": 4486 }, { "epoch": 0.15, "learning_rate": 4.778014909021187e-05, "loss": 0.9019, "step": 4487 }, { "epoch": 0.15, "learning_rate": 4.777905219622007e-05, "loss": 0.6675, "step": 4488 }, { "epoch": 0.15, "learning_rate": 4.777795504388844e-05, "loss": 0.8389, "step": 4489 }, { "epoch": 0.15, "learning_rate": 4.777685763322943e-05, "loss": 0.6601, "step": 4490 }, { "epoch": 0.15, "learning_rate": 4.7775759964255496e-05, "loss": 0.9932, "step": 4491 }, { "epoch": 0.15, "learning_rate": 4.777466203697906e-05, "loss": 0.8628, "step": 4492 }, { "epoch": 0.15, "learning_rate": 4.777356385141259e-05, "loss": 1.0592, "step": 4493 }, { "epoch": 0.15, "learning_rate": 4.777246540756853e-05, "loss": 0.6658, "step": 4494 }, { "epoch": 0.15, "learning_rate": 4.777136670545936e-05, "loss": 0.6068, "step": 4495 }, { "epoch": 0.15, "learning_rate": 4.777026774509752e-05, "loss": 0.8721, "step": 4496 }, { "epoch": 0.15, "learning_rate": 4.7769168526495476e-05, "loss": 1.0518, "step": 4497 }, { "epoch": 0.15, "learning_rate": 4.7768069049665705e-05, "loss": 0.9663, "step": 4498 }, { "epoch": 0.15, "learning_rate": 4.776696931462067e-05, "loss": 0.9679, "step": 4499 }, { "epoch": 0.15, "learning_rate": 4.7765869321372836e-05, "loss": 0.899, "step": 4500 }, { "epoch": 0.15, "learning_rate": 4.776476906993468e-05, "loss": 0.5529, "step": 4501 }, { "epoch": 0.15, "learning_rate": 4.776366856031869e-05, "loss": 0.6398, "step": 4502 }, { "epoch": 0.15, "learning_rate": 4.776256779253735e-05, "loss": 1.0881, "step": 4503 }, { "epoch": 0.15, "learning_rate": 4.7761466766603124e-05, "loss": 0.5568, "step": 4504 }, { "epoch": 0.15, "learning_rate": 4.7760365482528516e-05, "loss": 0.6635, "step": 4505 }, { "epoch": 0.15, "learning_rate": 4.7759263940326004e-05, "loss": 0.6128, "step": 4506 }, { "epoch": 0.15, "learning_rate": 4.775816214000809e-05, "loss": 0.2832, "step": 4507 }, { "epoch": 0.15, "learning_rate": 4.7757060081587265e-05, "loss": 0.8874, "step": 4508 }, { "epoch": 0.15, "learning_rate": 4.775595776507603e-05, "loss": 0.7059, "step": 4509 }, { "epoch": 0.15, "learning_rate": 4.775485519048689e-05, "loss": 0.7573, "step": 4510 }, { "epoch": 0.15, "learning_rate": 4.775375235783234e-05, "loss": 0.969, "step": 4511 }, { "epoch": 0.15, "learning_rate": 4.775264926712489e-05, "loss": 0.5724, "step": 4512 }, { "epoch": 0.15, "learning_rate": 4.7751545918377064e-05, "loss": 0.8952, "step": 4513 }, { "epoch": 0.15, "learning_rate": 4.775044231160135e-05, "loss": 0.9013, "step": 4514 }, { "epoch": 0.15, "learning_rate": 4.774933844681029e-05, "loss": 0.9884, "step": 4515 }, { "epoch": 0.15, "learning_rate": 4.7748234324016385e-05, "loss": 1.011, "step": 4516 }, { "epoch": 0.15, "learning_rate": 4.774712994323216e-05, "loss": 0.9603, "step": 4517 }, { "epoch": 0.15, "learning_rate": 4.774602530447014e-05, "loss": 1.0976, "step": 4518 }, { "epoch": 0.15, "learning_rate": 4.774492040774287e-05, "loss": 0.8583, "step": 4519 }, { "epoch": 0.15, "learning_rate": 4.774381525306286e-05, "loss": 0.9015, "step": 4520 }, { "epoch": 0.15, "learning_rate": 4.774270984044264e-05, "loss": 0.8882, "step": 4521 }, { "epoch": 0.15, "learning_rate": 4.774160416989477e-05, "loss": 0.8085, "step": 4522 }, { "epoch": 0.15, "learning_rate": 4.7740498241431775e-05, "loss": 0.9345, "step": 4523 }, { "epoch": 0.15, "learning_rate": 4.773939205506619e-05, "loss": 0.9918, "step": 4524 }, { "epoch": 0.15, "learning_rate": 4.773828561081058e-05, "loss": 0.4166, "step": 4525 }, { "epoch": 0.15, "learning_rate": 4.773717890867747e-05, "loss": 0.9831, "step": 4526 }, { "epoch": 0.15, "learning_rate": 4.773607194867944e-05, "loss": 1.0042, "step": 4527 }, { "epoch": 0.15, "learning_rate": 4.773496473082901e-05, "loss": 1.1249, "step": 4528 }, { "epoch": 0.15, "learning_rate": 4.773385725513877e-05, "loss": 0.7935, "step": 4529 }, { "epoch": 0.15, "learning_rate": 4.773274952162126e-05, "loss": 0.7129, "step": 4530 }, { "epoch": 0.15, "learning_rate": 4.773164153028904e-05, "loss": 0.9082, "step": 4531 }, { "epoch": 0.15, "learning_rate": 4.7730533281154694e-05, "loss": 0.9126, "step": 4532 }, { "epoch": 0.15, "learning_rate": 4.772942477423078e-05, "loss": 0.7181, "step": 4533 }, { "epoch": 0.15, "learning_rate": 4.772831600952986e-05, "loss": 0.9919, "step": 4534 }, { "epoch": 0.15, "learning_rate": 4.772720698706453e-05, "loss": 0.9285, "step": 4535 }, { "epoch": 0.15, "learning_rate": 4.772609770684735e-05, "loss": 1.3617, "step": 4536 }, { "epoch": 0.15, "learning_rate": 4.772498816889091e-05, "loss": 0.9202, "step": 4537 }, { "epoch": 0.15, "learning_rate": 4.7723878373207784e-05, "loss": 0.8969, "step": 4538 }, { "epoch": 0.15, "learning_rate": 4.772276831981057e-05, "loss": 1.0066, "step": 4539 }, { "epoch": 0.15, "learning_rate": 4.7721658008711855e-05, "loss": 0.8053, "step": 4540 }, { "epoch": 0.15, "learning_rate": 4.772054743992422e-05, "loss": 1.1042, "step": 4541 }, { "epoch": 0.15, "learning_rate": 4.771943661346028e-05, "loss": 0.5824, "step": 4542 }, { "epoch": 0.15, "learning_rate": 4.7718325529332615e-05, "loss": 0.9603, "step": 4543 }, { "epoch": 0.15, "learning_rate": 4.7717214187553826e-05, "loss": 0.8739, "step": 4544 }, { "epoch": 0.15, "learning_rate": 4.771610258813653e-05, "loss": 0.7829, "step": 4545 }, { "epoch": 0.15, "learning_rate": 4.7714990731093326e-05, "loss": 0.9512, "step": 4546 }, { "epoch": 0.15, "learning_rate": 4.7713878616436824e-05, "loss": 0.4939, "step": 4547 }, { "epoch": 0.15, "learning_rate": 4.771276624417963e-05, "loss": 0.6156, "step": 4548 }, { "epoch": 0.15, "learning_rate": 4.771165361433437e-05, "loss": 0.8309, "step": 4549 }, { "epoch": 0.15, "learning_rate": 4.7710540726913666e-05, "loss": 0.8892, "step": 4550 }, { "epoch": 0.15, "learning_rate": 4.770942758193013e-05, "loss": 1.1214, "step": 4551 }, { "epoch": 0.15, "learning_rate": 4.7708314179396384e-05, "loss": 1.6408, "step": 4552 }, { "epoch": 0.15, "learning_rate": 4.7707200519325064e-05, "loss": 1.7222, "step": 4553 }, { "epoch": 0.15, "learning_rate": 4.770608660172879e-05, "loss": 0.6609, "step": 4554 }, { "epoch": 0.15, "learning_rate": 4.770497242662021e-05, "loss": 0.8556, "step": 4555 }, { "epoch": 0.15, "learning_rate": 4.7703857994011946e-05, "loss": 0.737, "step": 4556 }, { "epoch": 0.15, "learning_rate": 4.770274330391664e-05, "loss": 1.0367, "step": 4557 }, { "epoch": 0.15, "learning_rate": 4.770162835634694e-05, "loss": 0.8194, "step": 4558 }, { "epoch": 0.15, "learning_rate": 4.770051315131549e-05, "loss": 2.9039, "step": 4559 }, { "epoch": 0.15, "learning_rate": 4.7699397688834926e-05, "loss": 0.9636, "step": 4560 }, { "epoch": 0.15, "learning_rate": 4.769828196891791e-05, "loss": 0.9599, "step": 4561 }, { "epoch": 0.15, "learning_rate": 4.7697165991577086e-05, "loss": 1.0101, "step": 4562 }, { "epoch": 0.15, "learning_rate": 4.7696049756825124e-05, "loss": 0.6571, "step": 4563 }, { "epoch": 0.15, "learning_rate": 4.769493326467468e-05, "loss": 0.8607, "step": 4564 }, { "epoch": 0.15, "learning_rate": 4.76938165151384e-05, "loss": 0.8409, "step": 4565 }, { "epoch": 0.15, "learning_rate": 4.769269950822897e-05, "loss": 0.9143, "step": 4566 }, { "epoch": 0.15, "learning_rate": 4.7691582243959055e-05, "loss": 0.7792, "step": 4567 }, { "epoch": 0.15, "learning_rate": 4.769046472234131e-05, "loss": 0.216, "step": 4568 }, { "epoch": 0.15, "learning_rate": 4.7689346943388424e-05, "loss": 0.3975, "step": 4569 }, { "epoch": 0.15, "learning_rate": 4.768822890711307e-05, "loss": 0.6703, "step": 4570 }, { "epoch": 0.15, "learning_rate": 4.7687110613527926e-05, "loss": 0.013, "step": 4571 }, { "epoch": 0.15, "learning_rate": 4.768599206264568e-05, "loss": 0.0085, "step": 4572 }, { "epoch": 0.15, "learning_rate": 4.768487325447901e-05, "loss": 0.0058, "step": 4573 }, { "epoch": 0.15, "learning_rate": 4.768375418904061e-05, "loss": 0.0038, "step": 4574 }, { "epoch": 0.15, "learning_rate": 4.768263486634316e-05, "loss": 0.4854, "step": 4575 }, { "epoch": 0.15, "learning_rate": 4.768151528639937e-05, "loss": 1.0314, "step": 4576 }, { "epoch": 0.15, "learning_rate": 4.7680395449221936e-05, "loss": 0.9897, "step": 4577 }, { "epoch": 0.15, "learning_rate": 4.767927535482355e-05, "loss": 0.682, "step": 4578 }, { "epoch": 0.15, "learning_rate": 4.767815500321692e-05, "loss": 0.46, "step": 4579 }, { "epoch": 0.15, "learning_rate": 4.767703439441474e-05, "loss": 0.4249, "step": 4580 }, { "epoch": 0.15, "learning_rate": 4.767591352842975e-05, "loss": 0.3849, "step": 4581 }, { "epoch": 0.15, "learning_rate": 4.767479240527463e-05, "loss": 0.5998, "step": 4582 }, { "epoch": 0.15, "learning_rate": 4.76736710249621e-05, "loss": 1.0526, "step": 4583 }, { "epoch": 0.15, "learning_rate": 4.76725493875049e-05, "loss": 0.8768, "step": 4584 }, { "epoch": 0.15, "learning_rate": 4.767142749291572e-05, "loss": 0.8826, "step": 4585 }, { "epoch": 0.15, "learning_rate": 4.76703053412073e-05, "loss": 0.8643, "step": 4586 }, { "epoch": 0.15, "learning_rate": 4.7669182932392366e-05, "loss": 0.7822, "step": 4587 }, { "epoch": 0.15, "learning_rate": 4.766806026648365e-05, "loss": 0.4975, "step": 4588 }, { "epoch": 0.15, "learning_rate": 4.766693734349387e-05, "loss": 1.2853, "step": 4589 }, { "epoch": 0.15, "learning_rate": 4.7665814163435785e-05, "loss": 0.886, "step": 4590 }, { "epoch": 0.15, "learning_rate": 4.766469072632211e-05, "loss": 1.022, "step": 4591 }, { "epoch": 0.15, "learning_rate": 4.76635670321656e-05, "loss": 0.8644, "step": 4592 }, { "epoch": 0.15, "learning_rate": 4.766244308097899e-05, "loss": 1.0294, "step": 4593 }, { "epoch": 0.15, "learning_rate": 4.766131887277503e-05, "loss": 0.8588, "step": 4594 }, { "epoch": 0.15, "learning_rate": 4.7660194407566476e-05, "loss": 1.0153, "step": 4595 }, { "epoch": 0.15, "learning_rate": 4.765906968536608e-05, "loss": 0.8591, "step": 4596 }, { "epoch": 0.15, "learning_rate": 4.7657944706186584e-05, "loss": 0.9952, "step": 4597 }, { "epoch": 0.15, "learning_rate": 4.765681947004076e-05, "loss": 0.9337, "step": 4598 }, { "epoch": 0.15, "learning_rate": 4.765569397694136e-05, "loss": 2.3508, "step": 4599 }, { "epoch": 0.15, "learning_rate": 4.765456822690116e-05, "loss": 2.0011, "step": 4600 }, { "epoch": 0.15, "learning_rate": 4.765344221993292e-05, "loss": 1.1872, "step": 4601 }, { "epoch": 0.15, "learning_rate": 4.765231595604941e-05, "loss": 0.9157, "step": 4602 }, { "epoch": 0.15, "learning_rate": 4.76511894352634e-05, "loss": 0.9147, "step": 4603 }, { "epoch": 0.15, "learning_rate": 4.765006265758768e-05, "loss": 0.9894, "step": 4604 }, { "epoch": 0.15, "learning_rate": 4.764893562303501e-05, "loss": 1.0365, "step": 4605 }, { "epoch": 0.15, "learning_rate": 4.7647808331618186e-05, "loss": 0.9347, "step": 4606 }, { "epoch": 0.15, "learning_rate": 4.7646680783349986e-05, "loss": 2.1444, "step": 4607 }, { "epoch": 0.15, "learning_rate": 4.76455529782432e-05, "loss": 1.304, "step": 4608 }, { "epoch": 0.15, "learning_rate": 4.764442491631061e-05, "loss": 0.965, "step": 4609 }, { "epoch": 0.15, "learning_rate": 4.7643296597565035e-05, "loss": 0.7358, "step": 4610 }, { "epoch": 0.15, "learning_rate": 4.764216802201924e-05, "loss": 0.9752, "step": 4611 }, { "epoch": 0.15, "learning_rate": 4.764103918968604e-05, "loss": 1.5292, "step": 4612 }, { "epoch": 0.15, "learning_rate": 4.7639910100578236e-05, "loss": 4.2213, "step": 4613 }, { "epoch": 0.15, "learning_rate": 4.7638780754708633e-05, "loss": 0.806, "step": 4614 }, { "epoch": 0.15, "learning_rate": 4.7637651152090037e-05, "loss": 0.4064, "step": 4615 }, { "epoch": 0.15, "learning_rate": 4.763652129273526e-05, "loss": 0.3973, "step": 4616 }, { "epoch": 0.15, "learning_rate": 4.763539117665712e-05, "loss": 0.5887, "step": 4617 }, { "epoch": 0.15, "learning_rate": 4.763426080386842e-05, "loss": 0.787, "step": 4618 }, { "epoch": 0.15, "learning_rate": 4.7633130174382e-05, "loss": 1.129, "step": 4619 }, { "epoch": 0.15, "learning_rate": 4.763199928821067e-05, "loss": 1.0089, "step": 4620 }, { "epoch": 0.15, "learning_rate": 4.7630868145367256e-05, "loss": 1.0206, "step": 4621 }, { "epoch": 0.15, "learning_rate": 4.762973674586459e-05, "loss": 0.7418, "step": 4622 }, { "epoch": 0.15, "learning_rate": 4.762860508971551e-05, "loss": 0.7549, "step": 4623 }, { "epoch": 0.15, "learning_rate": 4.7627473176932827e-05, "loss": 0.9026, "step": 4624 }, { "epoch": 0.15, "learning_rate": 4.76263410075294e-05, "loss": 0.4934, "step": 4625 }, { "epoch": 0.15, "learning_rate": 4.762520858151806e-05, "loss": 0.5836, "step": 4626 }, { "epoch": 0.15, "learning_rate": 4.7624075898911644e-05, "loss": 0.9978, "step": 4627 }, { "epoch": 0.15, "learning_rate": 4.762294295972301e-05, "loss": 0.7669, "step": 4628 }, { "epoch": 0.15, "learning_rate": 4.762180976396502e-05, "loss": 0.8359, "step": 4629 }, { "epoch": 0.15, "learning_rate": 4.762067631165049e-05, "loss": 0.9285, "step": 4630 }, { "epoch": 0.15, "learning_rate": 4.76195426027923e-05, "loss": 1.3855, "step": 4631 }, { "epoch": 0.15, "learning_rate": 4.76184086374033e-05, "loss": 0.9183, "step": 4632 }, { "epoch": 0.15, "learning_rate": 4.761727441549635e-05, "loss": 0.9869, "step": 4633 }, { "epoch": 0.15, "learning_rate": 4.7616139937084306e-05, "loss": 0.8468, "step": 4634 }, { "epoch": 0.15, "learning_rate": 4.7615005202180054e-05, "loss": 0.6791, "step": 4635 }, { "epoch": 0.15, "learning_rate": 4.761387021079645e-05, "loss": 0.918, "step": 4636 }, { "epoch": 0.15, "learning_rate": 4.761273496294636e-05, "loss": 0.7729, "step": 4637 }, { "epoch": 0.15, "learning_rate": 4.761159945864267e-05, "loss": 0.0941, "step": 4638 }, { "epoch": 0.15, "learning_rate": 4.761046369789826e-05, "loss": 0.0179, "step": 4639 }, { "epoch": 0.15, "learning_rate": 4.7609327680726e-05, "loss": 0.8148, "step": 4640 }, { "epoch": 0.15, "learning_rate": 4.760819140713878e-05, "loss": 1.2607, "step": 4641 }, { "epoch": 0.15, "learning_rate": 4.7607054877149496e-05, "loss": 0.9281, "step": 4642 }, { "epoch": 0.15, "learning_rate": 4.760591809077101e-05, "loss": 1.0006, "step": 4643 }, { "epoch": 0.15, "learning_rate": 4.760478104801624e-05, "loss": 0.5547, "step": 4644 }, { "epoch": 0.15, "learning_rate": 4.760364374889807e-05, "loss": 1.0461, "step": 4645 }, { "epoch": 0.15, "learning_rate": 4.7602506193429406e-05, "loss": 0.7543, "step": 4646 }, { "epoch": 0.15, "learning_rate": 4.760136838162314e-05, "loss": 0.6565, "step": 4647 }, { "epoch": 0.15, "learning_rate": 4.760023031349218e-05, "loss": 0.4773, "step": 4648 }, { "epoch": 0.15, "learning_rate": 4.759909198904944e-05, "loss": 0.6491, "step": 4649 }, { "epoch": 0.15, "learning_rate": 4.759795340830781e-05, "loss": 0.7204, "step": 4650 }, { "epoch": 0.16, "learning_rate": 4.7596814571280236e-05, "loss": 0.8876, "step": 4651 }, { "epoch": 0.16, "learning_rate": 4.7595675477979595e-05, "loss": 0.7861, "step": 4652 }, { "epoch": 0.16, "learning_rate": 4.759453612841883e-05, "loss": 0.8906, "step": 4653 }, { "epoch": 0.16, "learning_rate": 4.7593396522610865e-05, "loss": 0.7638, "step": 4654 }, { "epoch": 0.16, "learning_rate": 4.7592256660568606e-05, "loss": 0.7607, "step": 4655 }, { "epoch": 0.16, "learning_rate": 4.7591116542304994e-05, "loss": 0.8286, "step": 4656 }, { "epoch": 0.16, "learning_rate": 4.7589976167832956e-05, "loss": 0.8577, "step": 4657 }, { "epoch": 0.16, "learning_rate": 4.758883553716542e-05, "loss": 0.7968, "step": 4658 }, { "epoch": 0.16, "learning_rate": 4.7587694650315334e-05, "loss": 0.9163, "step": 4659 }, { "epoch": 0.16, "learning_rate": 4.758655350729563e-05, "loss": 1.1376, "step": 4660 }, { "epoch": 0.16, "learning_rate": 4.758541210811924e-05, "loss": 0.9196, "step": 4661 }, { "epoch": 0.16, "learning_rate": 4.758427045279913e-05, "loss": 0.9086, "step": 4662 }, { "epoch": 0.16, "learning_rate": 4.758312854134823e-05, "loss": 0.8916, "step": 4663 }, { "epoch": 0.16, "learning_rate": 4.75819863737795e-05, "loss": 1.6031, "step": 4664 }, { "epoch": 0.16, "learning_rate": 4.758084395010588e-05, "loss": 0.8202, "step": 4665 }, { "epoch": 0.16, "learning_rate": 4.757970127034034e-05, "loss": 0.6234, "step": 4666 }, { "epoch": 0.16, "learning_rate": 4.7578558334495836e-05, "loss": 0.8839, "step": 4667 }, { "epoch": 0.16, "learning_rate": 4.7577415142585336e-05, "loss": 1.0528, "step": 4668 }, { "epoch": 0.16, "learning_rate": 4.75762716946218e-05, "loss": 0.7482, "step": 4669 }, { "epoch": 0.16, "learning_rate": 4.757512799061818e-05, "loss": 0.6117, "step": 4670 }, { "epoch": 0.16, "learning_rate": 4.7573984030587476e-05, "loss": 0.3296, "step": 4671 }, { "epoch": 0.16, "learning_rate": 4.757283981454265e-05, "loss": 0.3144, "step": 4672 }, { "epoch": 0.16, "learning_rate": 4.7571695342496665e-05, "loss": 0.0915, "step": 4673 }, { "epoch": 0.16, "learning_rate": 4.757055061446252e-05, "loss": 0.6617, "step": 4674 }, { "epoch": 0.16, "learning_rate": 4.756940563045318e-05, "loss": 0.8953, "step": 4675 }, { "epoch": 0.16, "learning_rate": 4.756826039048165e-05, "loss": 0.8542, "step": 4676 }, { "epoch": 0.16, "learning_rate": 4.756711489456089e-05, "loss": 0.7644, "step": 4677 }, { "epoch": 0.16, "learning_rate": 4.756596914270392e-05, "loss": 0.519, "step": 4678 }, { "epoch": 0.16, "learning_rate": 4.756482313492373e-05, "loss": 0.7857, "step": 4679 }, { "epoch": 0.16, "learning_rate": 4.75636768712333e-05, "loss": 1.7156, "step": 4680 }, { "epoch": 0.16, "learning_rate": 4.756253035164565e-05, "loss": 2.3418, "step": 4681 }, { "epoch": 0.16, "learning_rate": 4.756138357617377e-05, "loss": 1.409, "step": 4682 }, { "epoch": 0.16, "learning_rate": 4.756023654483066e-05, "loss": 1.0155, "step": 4683 }, { "epoch": 0.16, "learning_rate": 4.755908925762934e-05, "loss": 1.0551, "step": 4684 }, { "epoch": 0.16, "learning_rate": 4.7557941714582825e-05, "loss": 0.5429, "step": 4685 }, { "epoch": 0.16, "learning_rate": 4.7556793915704123e-05, "loss": 0.7932, "step": 4686 }, { "epoch": 0.16, "learning_rate": 4.7555645861006245e-05, "loss": 0.5831, "step": 4687 }, { "epoch": 0.16, "learning_rate": 4.755449755050222e-05, "loss": 0.8742, "step": 4688 }, { "epoch": 0.16, "learning_rate": 4.755334898420507e-05, "loss": 0.9997, "step": 4689 }, { "epoch": 0.16, "learning_rate": 4.755220016212783e-05, "loss": 0.8563, "step": 4690 }, { "epoch": 0.16, "learning_rate": 4.7551051084283504e-05, "loss": 0.8813, "step": 4691 }, { "epoch": 0.16, "learning_rate": 4.7549901750685145e-05, "loss": 0.8259, "step": 4692 }, { "epoch": 0.16, "learning_rate": 4.754875216134578e-05, "loss": 0.7992, "step": 4693 }, { "epoch": 0.16, "learning_rate": 4.754760231627844e-05, "loss": 0.9445, "step": 4694 }, { "epoch": 0.16, "learning_rate": 4.7546452215496185e-05, "loss": 0.511, "step": 4695 }, { "epoch": 0.16, "learning_rate": 4.7545301859012036e-05, "loss": 0.804, "step": 4696 }, { "epoch": 0.16, "learning_rate": 4.7544151246839066e-05, "loss": 1.0346, "step": 4697 }, { "epoch": 0.16, "learning_rate": 4.75430003789903e-05, "loss": 0.7548, "step": 4698 }, { "epoch": 0.16, "learning_rate": 4.754184925547879e-05, "loss": 1.4358, "step": 4699 }, { "epoch": 0.16, "learning_rate": 4.754069787631761e-05, "loss": 2.6329, "step": 4700 }, { "epoch": 0.16, "learning_rate": 4.7539546241519804e-05, "loss": 0.8596, "step": 4701 }, { "epoch": 0.16, "learning_rate": 4.753839435109844e-05, "loss": 0.6891, "step": 4702 }, { "epoch": 0.16, "learning_rate": 4.7537242205066566e-05, "loss": 0.2238, "step": 4703 }, { "epoch": 0.16, "learning_rate": 4.7536089803437277e-05, "loss": 1.1452, "step": 4704 }, { "epoch": 0.16, "learning_rate": 4.7534937146223615e-05, "loss": 1.0656, "step": 4705 }, { "epoch": 0.16, "learning_rate": 4.7533784233438674e-05, "loss": 1.8949, "step": 4706 }, { "epoch": 0.16, "learning_rate": 4.753263106509551e-05, "loss": 0.6704, "step": 4707 }, { "epoch": 0.16, "learning_rate": 4.7531477641207215e-05, "loss": 0.87, "step": 4708 }, { "epoch": 0.16, "learning_rate": 4.753032396178686e-05, "loss": 0.7489, "step": 4709 }, { "epoch": 0.16, "learning_rate": 4.752917002684753e-05, "loss": 0.9148, "step": 4710 }, { "epoch": 0.16, "learning_rate": 4.752801583640233e-05, "loss": 0.7885, "step": 4711 }, { "epoch": 0.16, "learning_rate": 4.752686139046433e-05, "loss": 0.6552, "step": 4712 }, { "epoch": 0.16, "learning_rate": 4.7525706689046626e-05, "loss": 0.861, "step": 4713 }, { "epoch": 0.16, "learning_rate": 4.752455173216231e-05, "loss": 0.7059, "step": 4714 }, { "epoch": 0.16, "learning_rate": 4.75233965198245e-05, "loss": 0.7401, "step": 4715 }, { "epoch": 0.16, "learning_rate": 4.752224105204628e-05, "loss": 0.6862, "step": 4716 }, { "epoch": 0.16, "learning_rate": 4.7521085328840755e-05, "loss": 0.1987, "step": 4717 }, { "epoch": 0.16, "learning_rate": 4.751992935022104e-05, "loss": 0.6354, "step": 4718 }, { "epoch": 0.16, "learning_rate": 4.751877311620024e-05, "loss": 0.8331, "step": 4719 }, { "epoch": 0.16, "learning_rate": 4.751761662679146e-05, "loss": 0.9611, "step": 4720 }, { "epoch": 0.16, "learning_rate": 4.751645988200784e-05, "loss": 0.8058, "step": 4721 }, { "epoch": 0.16, "learning_rate": 4.751530288186247e-05, "loss": 0.7625, "step": 4722 }, { "epoch": 0.16, "learning_rate": 4.751414562636849e-05, "loss": 0.9028, "step": 4723 }, { "epoch": 0.16, "learning_rate": 4.751298811553902e-05, "loss": 0.7045, "step": 4724 }, { "epoch": 0.16, "learning_rate": 4.751183034938719e-05, "loss": 0.9868, "step": 4725 }, { "epoch": 0.16, "learning_rate": 4.751067232792612e-05, "loss": 0.7116, "step": 4726 }, { "epoch": 0.16, "learning_rate": 4.750951405116896e-05, "loss": 3.2342, "step": 4727 }, { "epoch": 0.16, "learning_rate": 4.7508355519128825e-05, "loss": 2.3907, "step": 4728 }, { "epoch": 0.16, "learning_rate": 4.750719673181887e-05, "loss": 0.8683, "step": 4729 }, { "epoch": 0.16, "learning_rate": 4.7506037689252234e-05, "loss": 0.5564, "step": 4730 }, { "epoch": 0.16, "learning_rate": 4.7504878391442066e-05, "loss": 0.6307, "step": 4731 }, { "epoch": 0.16, "learning_rate": 4.75037188384015e-05, "loss": 0.2708, "step": 4732 }, { "epoch": 0.16, "learning_rate": 4.75025590301437e-05, "loss": 0.914, "step": 4733 }, { "epoch": 0.16, "learning_rate": 4.750139896668182e-05, "loss": 0.8133, "step": 4734 }, { "epoch": 0.16, "learning_rate": 4.7500238648029e-05, "loss": 0.9534, "step": 4735 }, { "epoch": 0.16, "learning_rate": 4.749907807419841e-05, "loss": 1.0292, "step": 4736 }, { "epoch": 0.16, "learning_rate": 4.749791724520322e-05, "loss": 0.8836, "step": 4737 }, { "epoch": 0.16, "learning_rate": 4.749675616105659e-05, "loss": 0.5479, "step": 4738 }, { "epoch": 0.16, "learning_rate": 4.749559482177167e-05, "loss": 0.8735, "step": 4739 }, { "epoch": 0.16, "learning_rate": 4.7494433227361655e-05, "loss": 0.9249, "step": 4740 }, { "epoch": 0.16, "learning_rate": 4.749327137783972e-05, "loss": 0.299, "step": 4741 }, { "epoch": 0.16, "learning_rate": 4.749210927321902e-05, "loss": 0.3818, "step": 4742 }, { "epoch": 0.16, "learning_rate": 4.7490946913512744e-05, "loss": 0.954, "step": 4743 }, { "epoch": 0.16, "learning_rate": 4.7489784298734084e-05, "loss": 1.2273, "step": 4744 }, { "epoch": 0.16, "learning_rate": 4.748862142889622e-05, "loss": 0.9657, "step": 4745 }, { "epoch": 0.16, "learning_rate": 4.7487458304012336e-05, "loss": 0.9396, "step": 4746 }, { "epoch": 0.16, "learning_rate": 4.748629492409562e-05, "loss": 0.9286, "step": 4747 }, { "epoch": 0.16, "learning_rate": 4.7485131289159276e-05, "loss": 0.9896, "step": 4748 }, { "epoch": 0.16, "learning_rate": 4.74839673992165e-05, "loss": 0.9091, "step": 4749 }, { "epoch": 0.16, "learning_rate": 4.7482803254280484e-05, "loss": 0.6038, "step": 4750 }, { "epoch": 0.16, "learning_rate": 4.748163885436444e-05, "loss": 1.1053, "step": 4751 }, { "epoch": 0.16, "learning_rate": 4.748047419948156e-05, "loss": 1.4451, "step": 4752 }, { "epoch": 0.16, "learning_rate": 4.747930928964506e-05, "loss": 1.1723, "step": 4753 }, { "epoch": 0.16, "learning_rate": 4.747814412486816e-05, "loss": 0.8338, "step": 4754 }, { "epoch": 0.16, "learning_rate": 4.747697870516406e-05, "loss": 0.9405, "step": 4755 }, { "epoch": 0.16, "learning_rate": 4.747581303054599e-05, "loss": 0.824, "step": 4756 }, { "epoch": 0.16, "learning_rate": 4.747464710102717e-05, "loss": 0.9108, "step": 4757 }, { "epoch": 0.16, "learning_rate": 4.747348091662081e-05, "loss": 2.3908, "step": 4758 }, { "epoch": 0.16, "learning_rate": 4.747231447734013e-05, "loss": 0.3675, "step": 4759 }, { "epoch": 0.16, "learning_rate": 4.747114778319839e-05, "loss": 0.3348, "step": 4760 }, { "epoch": 0.16, "learning_rate": 4.74699808342088e-05, "loss": 0.2739, "step": 4761 }, { "epoch": 0.16, "learning_rate": 4.746881363038459e-05, "loss": 0.2731, "step": 4762 }, { "epoch": 0.16, "learning_rate": 4.7467646171739005e-05, "loss": 0.2365, "step": 4763 }, { "epoch": 0.16, "learning_rate": 4.746647845828529e-05, "loss": 0.6608, "step": 4764 }, { "epoch": 0.16, "learning_rate": 4.746531049003668e-05, "loss": 0.8778, "step": 4765 }, { "epoch": 0.16, "learning_rate": 4.746414226700643e-05, "loss": 0.8785, "step": 4766 }, { "epoch": 0.16, "learning_rate": 4.746297378920778e-05, "loss": 0.896, "step": 4767 }, { "epoch": 0.16, "learning_rate": 4.746180505665398e-05, "loss": 0.908, "step": 4768 }, { "epoch": 0.16, "learning_rate": 4.74606360693583e-05, "loss": 0.7476, "step": 4769 }, { "epoch": 0.16, "learning_rate": 4.745946682733398e-05, "loss": 0.7221, "step": 4770 }, { "epoch": 0.16, "learning_rate": 4.7458297330594294e-05, "loss": 0.9824, "step": 4771 }, { "epoch": 0.16, "learning_rate": 4.745712757915249e-05, "loss": 0.9262, "step": 4772 }, { "epoch": 0.16, "learning_rate": 4.7455957573021844e-05, "loss": 0.5777, "step": 4773 }, { "epoch": 0.16, "learning_rate": 4.7454787312215636e-05, "loss": 0.9043, "step": 4774 }, { "epoch": 0.16, "learning_rate": 4.745361679674712e-05, "loss": 0.8695, "step": 4775 }, { "epoch": 0.16, "learning_rate": 4.745244602662958e-05, "loss": 1.2907, "step": 4776 }, { "epoch": 0.16, "learning_rate": 4.7451275001876296e-05, "loss": 0.7675, "step": 4777 }, { "epoch": 0.16, "learning_rate": 4.745010372250054e-05, "loss": 0.8066, "step": 4778 }, { "epoch": 0.16, "learning_rate": 4.744893218851559e-05, "loss": 0.581, "step": 4779 }, { "epoch": 0.16, "learning_rate": 4.744776039993476e-05, "loss": 0.7647, "step": 4780 }, { "epoch": 0.16, "learning_rate": 4.7446588356771315e-05, "loss": 1.9288, "step": 4781 }, { "epoch": 0.16, "learning_rate": 4.744541605903855e-05, "loss": 0.8713, "step": 4782 }, { "epoch": 0.16, "learning_rate": 4.7444243506749766e-05, "loss": 1.1017, "step": 4783 }, { "epoch": 0.16, "learning_rate": 4.744307069991827e-05, "loss": 1.2738, "step": 4784 }, { "epoch": 0.16, "learning_rate": 4.744189763855734e-05, "loss": 0.9708, "step": 4785 }, { "epoch": 0.16, "learning_rate": 4.74407243226803e-05, "loss": 1.1457, "step": 4786 }, { "epoch": 0.16, "learning_rate": 4.743955075230044e-05, "loss": 1.008, "step": 4787 }, { "epoch": 0.16, "learning_rate": 4.743837692743108e-05, "loss": 1.0235, "step": 4788 }, { "epoch": 0.16, "learning_rate": 4.743720284808554e-05, "loss": 0.9481, "step": 4789 }, { "epoch": 0.16, "learning_rate": 4.743602851427712e-05, "loss": 0.3435, "step": 4790 }, { "epoch": 0.16, "learning_rate": 4.743485392601914e-05, "loss": 0.897, "step": 4791 }, { "epoch": 0.16, "learning_rate": 4.743367908332493e-05, "loss": 0.902, "step": 4792 }, { "epoch": 0.16, "learning_rate": 4.743250398620782e-05, "loss": 0.8729, "step": 4793 }, { "epoch": 0.16, "learning_rate": 4.7431328634681114e-05, "loss": 0.5893, "step": 4794 }, { "epoch": 0.16, "learning_rate": 4.7430153028758154e-05, "loss": 0.7669, "step": 4795 }, { "epoch": 0.16, "learning_rate": 4.742897716845228e-05, "loss": 1.361, "step": 4796 }, { "epoch": 0.16, "learning_rate": 4.7427801053776824e-05, "loss": 1.579, "step": 4797 }, { "epoch": 0.16, "learning_rate": 4.742662468474511e-05, "loss": 1.4652, "step": 4798 }, { "epoch": 0.16, "learning_rate": 4.742544806137049e-05, "loss": 0.9151, "step": 4799 }, { "epoch": 0.16, "learning_rate": 4.742427118366632e-05, "loss": 0.5357, "step": 4800 }, { "epoch": 0.16, "learning_rate": 4.742309405164593e-05, "loss": 0.8056, "step": 4801 }, { "epoch": 0.16, "learning_rate": 4.7421916665322674e-05, "loss": 1.0417, "step": 4802 }, { "epoch": 0.16, "learning_rate": 4.7420739024709914e-05, "loss": 0.9935, "step": 4803 }, { "epoch": 0.16, "learning_rate": 4.741956112982099e-05, "loss": 1.6194, "step": 4804 }, { "epoch": 0.16, "learning_rate": 4.741838298066927e-05, "loss": 1.4573, "step": 4805 }, { "epoch": 0.16, "learning_rate": 4.741720457726813e-05, "loss": 1.078, "step": 4806 }, { "epoch": 0.16, "learning_rate": 4.7416025919630904e-05, "loss": 0.3444, "step": 4807 }, { "epoch": 0.16, "learning_rate": 4.7414847007770966e-05, "loss": 0.7367, "step": 4808 }, { "epoch": 0.16, "learning_rate": 4.7413667841701715e-05, "loss": 1.0799, "step": 4809 }, { "epoch": 0.16, "learning_rate": 4.7412488421436494e-05, "loss": 0.8061, "step": 4810 }, { "epoch": 0.16, "learning_rate": 4.741130874698868e-05, "loss": 0.8852, "step": 4811 }, { "epoch": 0.16, "learning_rate": 4.7410128818371675e-05, "loss": 0.8897, "step": 4812 }, { "epoch": 0.16, "learning_rate": 4.740894863559884e-05, "loss": 0.8221, "step": 4813 }, { "epoch": 0.16, "learning_rate": 4.740776819868357e-05, "loss": 0.6282, "step": 4814 }, { "epoch": 0.16, "learning_rate": 4.740658750763923e-05, "loss": 0.8674, "step": 4815 }, { "epoch": 0.16, "learning_rate": 4.740540656247925e-05, "loss": 0.9283, "step": 4816 }, { "epoch": 0.16, "learning_rate": 4.7404225363216995e-05, "loss": 0.8785, "step": 4817 }, { "epoch": 0.16, "learning_rate": 4.740304390986586e-05, "loss": 0.7441, "step": 4818 }, { "epoch": 0.16, "learning_rate": 4.740186220243926e-05, "loss": 0.8031, "step": 4819 }, { "epoch": 0.16, "learning_rate": 4.740068024095059e-05, "loss": 0.9739, "step": 4820 }, { "epoch": 0.16, "learning_rate": 4.7399498025413245e-05, "loss": 0.9467, "step": 4821 }, { "epoch": 0.16, "learning_rate": 4.739831555584064e-05, "loss": 1.04, "step": 4822 }, { "epoch": 0.16, "learning_rate": 4.7397132832246206e-05, "loss": 1.0068, "step": 4823 }, { "epoch": 0.16, "learning_rate": 4.739594985464332e-05, "loss": 1.1238, "step": 4824 }, { "epoch": 0.16, "learning_rate": 4.7394766623045414e-05, "loss": 0.9456, "step": 4825 }, { "epoch": 0.16, "learning_rate": 4.739358313746592e-05, "loss": 0.1599, "step": 4826 }, { "epoch": 0.16, "learning_rate": 4.7392399397918245e-05, "loss": 0.018, "step": 4827 }, { "epoch": 0.16, "learning_rate": 4.7391215404415815e-05, "loss": 0.0725, "step": 4828 }, { "epoch": 0.16, "learning_rate": 4.7390031156972054e-05, "loss": 0.9022, "step": 4829 }, { "epoch": 0.16, "learning_rate": 4.7388846655600406e-05, "loss": 0.8946, "step": 4830 }, { "epoch": 0.16, "learning_rate": 4.7387661900314295e-05, "loss": 1.2826, "step": 4831 }, { "epoch": 0.16, "learning_rate": 4.7386476891127164e-05, "loss": 1.596, "step": 4832 }, { "epoch": 0.16, "learning_rate": 4.7385291628052444e-05, "loss": 1.7039, "step": 4833 }, { "epoch": 0.16, "learning_rate": 4.738410611110358e-05, "loss": 1.0097, "step": 4834 }, { "epoch": 0.16, "learning_rate": 4.738292034029402e-05, "loss": 0.8225, "step": 4835 }, { "epoch": 0.16, "learning_rate": 4.738173431563722e-05, "loss": 0.8931, "step": 4836 }, { "epoch": 0.16, "learning_rate": 4.7380548037146606e-05, "loss": 0.1515, "step": 4837 }, { "epoch": 0.16, "learning_rate": 4.737936150483565e-05, "loss": 0.1887, "step": 4838 }, { "epoch": 0.16, "learning_rate": 4.737817471871782e-05, "loss": 0.1953, "step": 4839 }, { "epoch": 0.16, "learning_rate": 4.737698767880655e-05, "loss": 0.3955, "step": 4840 }, { "epoch": 0.16, "learning_rate": 4.737580038511531e-05, "loss": 2.1422, "step": 4841 }, { "epoch": 0.16, "learning_rate": 4.737461283765757e-05, "loss": 0.8223, "step": 4842 }, { "epoch": 0.16, "learning_rate": 4.737342503644681e-05, "loss": 0.9212, "step": 4843 }, { "epoch": 0.16, "learning_rate": 4.737223698149647e-05, "loss": 0.7603, "step": 4844 }, { "epoch": 0.16, "learning_rate": 4.737104867282005e-05, "loss": 1.1653, "step": 4845 }, { "epoch": 0.16, "learning_rate": 4.736986011043102e-05, "loss": 1.3596, "step": 4846 }, { "epoch": 0.16, "learning_rate": 4.736867129434285e-05, "loss": 0.7731, "step": 4847 }, { "epoch": 0.16, "learning_rate": 4.7367482224569035e-05, "loss": 0.9361, "step": 4848 }, { "epoch": 0.16, "learning_rate": 4.736629290112306e-05, "loss": 0.8189, "step": 4849 }, { "epoch": 0.16, "learning_rate": 4.7365103324018407e-05, "loss": 0.4329, "step": 4850 }, { "epoch": 0.16, "learning_rate": 4.736391349326857e-05, "loss": 0.7334, "step": 4851 }, { "epoch": 0.16, "learning_rate": 4.736272340888704e-05, "loss": 0.8841, "step": 4852 }, { "epoch": 0.16, "learning_rate": 4.7361533070887314e-05, "loss": 0.7796, "step": 4853 }, { "epoch": 0.16, "learning_rate": 4.7360342479282896e-05, "loss": 0.9453, "step": 4854 }, { "epoch": 0.16, "learning_rate": 4.735915163408729e-05, "loss": 0.574, "step": 4855 }, { "epoch": 0.16, "learning_rate": 4.7357960535313994e-05, "loss": 0.7917, "step": 4856 }, { "epoch": 0.16, "learning_rate": 4.735676918297652e-05, "loss": 0.9268, "step": 4857 }, { "epoch": 0.16, "learning_rate": 4.735557757708839e-05, "loss": 0.8361, "step": 4858 }, { "epoch": 0.16, "learning_rate": 4.73543857176631e-05, "loss": 0.8595, "step": 4859 }, { "epoch": 0.16, "learning_rate": 4.7353193604714175e-05, "loss": 1.1044, "step": 4860 }, { "epoch": 0.16, "learning_rate": 4.735200123825513e-05, "loss": 0.9983, "step": 4861 }, { "epoch": 0.16, "learning_rate": 4.7350808618299495e-05, "loss": 0.9877, "step": 4862 }, { "epoch": 0.16, "learning_rate": 4.7349615744860797e-05, "loss": 1.8877, "step": 4863 }, { "epoch": 0.16, "learning_rate": 4.734842261795256e-05, "loss": 3.5586, "step": 4864 }, { "epoch": 0.16, "learning_rate": 4.7347229237588316e-05, "loss": 3.0671, "step": 4865 }, { "epoch": 0.16, "learning_rate": 4.73460356037816e-05, "loss": 0.8831, "step": 4866 }, { "epoch": 0.16, "learning_rate": 4.734484171654594e-05, "loss": 0.6677, "step": 4867 }, { "epoch": 0.16, "learning_rate": 4.7343647575894896e-05, "loss": 0.5956, "step": 4868 }, { "epoch": 0.16, "learning_rate": 4.7342453181841996e-05, "loss": 1.6144, "step": 4869 }, { "epoch": 0.16, "learning_rate": 4.734125853440079e-05, "loss": 1.7991, "step": 4870 }, { "epoch": 0.16, "learning_rate": 4.734006363358482e-05, "loss": 2.5077, "step": 4871 }, { "epoch": 0.16, "learning_rate": 4.733886847940765e-05, "loss": 1.0259, "step": 4872 }, { "epoch": 0.16, "learning_rate": 4.733767307188282e-05, "loss": 0.9698, "step": 4873 }, { "epoch": 0.16, "learning_rate": 4.73364774110239e-05, "loss": 0.8731, "step": 4874 }, { "epoch": 0.16, "learning_rate": 4.733528149684444e-05, "loss": 0.8163, "step": 4875 }, { "epoch": 0.16, "learning_rate": 4.733408532935801e-05, "loss": 0.9254, "step": 4876 }, { "epoch": 0.16, "learning_rate": 4.733288890857818e-05, "loss": 2.7818, "step": 4877 }, { "epoch": 0.16, "learning_rate": 4.733169223451851e-05, "loss": 0.8565, "step": 4878 }, { "epoch": 0.16, "learning_rate": 4.733049530719258e-05, "loss": 0.8616, "step": 4879 }, { "epoch": 0.16, "learning_rate": 4.732929812661395e-05, "loss": 0.7779, "step": 4880 }, { "epoch": 0.16, "learning_rate": 4.732810069279621e-05, "loss": 0.8848, "step": 4881 }, { "epoch": 0.16, "learning_rate": 4.732690300575293e-05, "loss": 0.7596, "step": 4882 }, { "epoch": 0.16, "learning_rate": 4.7325705065497704e-05, "loss": 0.5766, "step": 4883 }, { "epoch": 0.16, "learning_rate": 4.7324506872044116e-05, "loss": 1.0555, "step": 4884 }, { "epoch": 0.16, "learning_rate": 4.732330842540575e-05, "loss": 0.7692, "step": 4885 }, { "epoch": 0.16, "learning_rate": 4.7322109725596195e-05, "loss": 0.7397, "step": 4886 }, { "epoch": 0.16, "learning_rate": 4.7320910772629055e-05, "loss": 0.8774, "step": 4887 }, { "epoch": 0.16, "learning_rate": 4.7319711566517926e-05, "loss": 1.7695, "step": 4888 }, { "epoch": 0.16, "learning_rate": 4.73185121072764e-05, "loss": 2.1435, "step": 4889 }, { "epoch": 0.16, "learning_rate": 4.731731239491809e-05, "loss": 0.8624, "step": 4890 }, { "epoch": 0.16, "learning_rate": 4.731611242945659e-05, "loss": 0.6971, "step": 4891 }, { "epoch": 0.16, "learning_rate": 4.731491221090553e-05, "loss": 0.8597, "step": 4892 }, { "epoch": 0.16, "learning_rate": 4.73137117392785e-05, "loss": 1.0608, "step": 4893 }, { "epoch": 0.16, "learning_rate": 4.731251101458912e-05, "loss": 1.0278, "step": 4894 }, { "epoch": 0.16, "learning_rate": 4.731131003685102e-05, "loss": 0.8288, "step": 4895 }, { "epoch": 0.16, "learning_rate": 4.73101088060778e-05, "loss": 0.6642, "step": 4896 }, { "epoch": 0.16, "learning_rate": 4.7308907322283103e-05, "loss": 1.1321, "step": 4897 }, { "epoch": 0.16, "learning_rate": 4.7307705585480545e-05, "loss": 0.5663, "step": 4898 }, { "epoch": 0.16, "learning_rate": 4.730650359568376e-05, "loss": 0.8005, "step": 4899 }, { "epoch": 0.16, "learning_rate": 4.7305301352906376e-05, "loss": 0.8205, "step": 4900 }, { "epoch": 0.16, "learning_rate": 4.730409885716202e-05, "loss": 0.8832, "step": 4901 }, { "epoch": 0.16, "learning_rate": 4.730289610846434e-05, "loss": 1.0425, "step": 4902 }, { "epoch": 0.16, "learning_rate": 4.730169310682698e-05, "loss": 3.8075, "step": 4903 }, { "epoch": 0.16, "learning_rate": 4.730048985226357e-05, "loss": 4.8812, "step": 4904 }, { "epoch": 0.16, "learning_rate": 4.729928634478777e-05, "loss": 0.7509, "step": 4905 }, { "epoch": 0.16, "learning_rate": 4.729808258441322e-05, "loss": 0.7391, "step": 4906 }, { "epoch": 0.16, "learning_rate": 4.729687857115358e-05, "loss": 1.5647, "step": 4907 }, { "epoch": 0.16, "learning_rate": 4.72956743050225e-05, "loss": 1.4842, "step": 4908 }, { "epoch": 0.16, "learning_rate": 4.729446978603363e-05, "loss": 1.4823, "step": 4909 }, { "epoch": 0.16, "learning_rate": 4.729326501420064e-05, "loss": 0.6745, "step": 4910 }, { "epoch": 0.16, "learning_rate": 4.7292059989537196e-05, "loss": 1.1692, "step": 4911 }, { "epoch": 0.16, "learning_rate": 4.729085471205696e-05, "loss": 1.4515, "step": 4912 }, { "epoch": 0.16, "learning_rate": 4.72896491817736e-05, "loss": 1.0769, "step": 4913 }, { "epoch": 0.16, "learning_rate": 4.728844339870079e-05, "loss": 3.4, "step": 4914 }, { "epoch": 0.16, "learning_rate": 4.72872373628522e-05, "loss": 1.2509, "step": 4915 }, { "epoch": 0.16, "learning_rate": 4.7286031074241515e-05, "loss": 0.7933, "step": 4916 }, { "epoch": 0.16, "learning_rate": 4.728482453288241e-05, "loss": 1.4991, "step": 4917 }, { "epoch": 0.16, "learning_rate": 4.7283617738788576e-05, "loss": 0.8343, "step": 4918 }, { "epoch": 0.16, "learning_rate": 4.7282410691973694e-05, "loss": 0.779, "step": 4919 }, { "epoch": 0.16, "learning_rate": 4.728120339245146e-05, "loss": 0.9362, "step": 4920 }, { "epoch": 0.16, "learning_rate": 4.727999584023555e-05, "loss": 0.9233, "step": 4921 }, { "epoch": 0.16, "learning_rate": 4.7278788035339664e-05, "loss": 0.9835, "step": 4922 }, { "epoch": 0.16, "learning_rate": 4.727757997777752e-05, "loss": 0.6446, "step": 4923 }, { "epoch": 0.16, "learning_rate": 4.7276371667562794e-05, "loss": 0.6235, "step": 4924 }, { "epoch": 0.16, "learning_rate": 4.72751631047092e-05, "loss": 0.9884, "step": 4925 }, { "epoch": 0.16, "learning_rate": 4.727395428923044e-05, "loss": 1.4491, "step": 4926 }, { "epoch": 0.16, "learning_rate": 4.727274522114023e-05, "loss": 0.699, "step": 4927 }, { "epoch": 0.16, "learning_rate": 4.7271535900452285e-05, "loss": 0.9435, "step": 4928 }, { "epoch": 0.16, "learning_rate": 4.727032632718031e-05, "loss": 1.0346, "step": 4929 }, { "epoch": 0.16, "learning_rate": 4.726911650133803e-05, "loss": 0.8838, "step": 4930 }, { "epoch": 0.16, "learning_rate": 4.7267906422939155e-05, "loss": 0.5546, "step": 4931 }, { "epoch": 0.16, "learning_rate": 4.726669609199742e-05, "loss": 0.6186, "step": 4932 }, { "epoch": 0.16, "learning_rate": 4.726548550852655e-05, "loss": 0.5589, "step": 4933 }, { "epoch": 0.16, "learning_rate": 4.7264274672540274e-05, "loss": 0.7967, "step": 4934 }, { "epoch": 0.16, "learning_rate": 4.726306358405232e-05, "loss": 0.7206, "step": 4935 }, { "epoch": 0.16, "learning_rate": 4.7261852243076424e-05, "loss": 0.8548, "step": 4936 }, { "epoch": 0.16, "learning_rate": 4.7260640649626324e-05, "loss": 0.9983, "step": 4937 }, { "epoch": 0.16, "learning_rate": 4.7259428803715764e-05, "loss": 0.6921, "step": 4938 }, { "epoch": 0.16, "learning_rate": 4.725821670535849e-05, "loss": 0.6853, "step": 4939 }, { "epoch": 0.16, "learning_rate": 4.7257004354568245e-05, "loss": 0.4091, "step": 4940 }, { "epoch": 0.16, "learning_rate": 4.725579175135877e-05, "loss": 0.7428, "step": 4941 }, { "epoch": 0.16, "learning_rate": 4.725457889574383e-05, "loss": 0.9192, "step": 4942 }, { "epoch": 0.16, "learning_rate": 4.725336578773718e-05, "loss": 0.8669, "step": 4943 }, { "epoch": 0.16, "learning_rate": 4.725215242735257e-05, "loss": 0.6368, "step": 4944 }, { "epoch": 0.16, "learning_rate": 4.7250938814603765e-05, "loss": 4.4775, "step": 4945 }, { "epoch": 0.16, "learning_rate": 4.724972494950453e-05, "loss": 2.8331, "step": 4946 }, { "epoch": 0.16, "learning_rate": 4.7248510832068613e-05, "loss": 0.7382, "step": 4947 }, { "epoch": 0.16, "learning_rate": 4.7247296462309823e-05, "loss": 0.7985, "step": 4948 }, { "epoch": 0.16, "learning_rate": 4.7246081840241894e-05, "loss": 0.7721, "step": 4949 }, { "epoch": 0.17, "learning_rate": 4.724486696587862e-05, "loss": 0.7644, "step": 4950 }, { "epoch": 0.17, "learning_rate": 4.724365183923378e-05, "loss": 2.1351, "step": 4951 }, { "epoch": 0.17, "learning_rate": 4.724243646032115e-05, "loss": 5.8826, "step": 4952 }, { "epoch": 0.17, "learning_rate": 4.7241220829154507e-05, "loss": 2.2364, "step": 4953 }, { "epoch": 0.17, "learning_rate": 4.724000494574765e-05, "loss": 0.8398, "step": 4954 }, { "epoch": 0.17, "learning_rate": 4.7238788810114365e-05, "loss": 1.0713, "step": 4955 }, { "epoch": 0.17, "learning_rate": 4.723757242226844e-05, "loss": 0.8204, "step": 4956 }, { "epoch": 0.17, "learning_rate": 4.7236355782223676e-05, "loss": 0.9271, "step": 4957 }, { "epoch": 0.17, "learning_rate": 4.723513888999387e-05, "loss": 0.8488, "step": 4958 }, { "epoch": 0.17, "learning_rate": 4.7233921745592816e-05, "loss": 0.9155, "step": 4959 }, { "epoch": 0.17, "learning_rate": 4.7232704349034316e-05, "loss": 0.8194, "step": 4960 }, { "epoch": 0.17, "learning_rate": 4.723148670033219e-05, "loss": 0.7988, "step": 4961 }, { "epoch": 0.17, "learning_rate": 4.723026879950024e-05, "loss": 0.7664, "step": 4962 }, { "epoch": 0.17, "learning_rate": 4.722905064655228e-05, "loss": 0.6882, "step": 4963 }, { "epoch": 0.17, "learning_rate": 4.722783224150212e-05, "loss": 1.0501, "step": 4964 }, { "epoch": 0.17, "learning_rate": 4.722661358436359e-05, "loss": 0.8996, "step": 4965 }, { "epoch": 0.17, "learning_rate": 4.7225394675150495e-05, "loss": 0.9858, "step": 4966 }, { "epoch": 0.17, "learning_rate": 4.722417551387668e-05, "loss": 0.7038, "step": 4967 }, { "epoch": 0.17, "learning_rate": 4.722295610055594e-05, "loss": 0.6109, "step": 4968 }, { "epoch": 0.17, "learning_rate": 4.7221736435202134e-05, "loss": 0.7414, "step": 4969 }, { "epoch": 0.17, "learning_rate": 4.722051651782908e-05, "loss": 0.9266, "step": 4970 }, { "epoch": 0.17, "learning_rate": 4.7219296348450624e-05, "loss": 0.7392, "step": 4971 }, { "epoch": 0.17, "learning_rate": 4.721807592708059e-05, "loss": 1.0345, "step": 4972 }, { "epoch": 0.17, "learning_rate": 4.721685525373283e-05, "loss": 0.5246, "step": 4973 }, { "epoch": 0.17, "learning_rate": 4.7215634328421184e-05, "loss": 0.8157, "step": 4974 }, { "epoch": 0.17, "learning_rate": 4.721441315115949e-05, "loss": 0.9007, "step": 4975 }, { "epoch": 0.17, "learning_rate": 4.721319172196161e-05, "loss": 0.988, "step": 4976 }, { "epoch": 0.17, "learning_rate": 4.7211970040841394e-05, "loss": 0.8291, "step": 4977 }, { "epoch": 0.17, "learning_rate": 4.7210748107812695e-05, "loss": 0.651, "step": 4978 }, { "epoch": 0.17, "learning_rate": 4.7209525922889373e-05, "loss": 0.3371, "step": 4979 }, { "epoch": 0.17, "learning_rate": 4.720830348608529e-05, "loss": 0.7701, "step": 4980 }, { "epoch": 0.17, "learning_rate": 4.72070807974143e-05, "loss": 0.9916, "step": 4981 }, { "epoch": 0.17, "learning_rate": 4.720585785689028e-05, "loss": 0.874, "step": 4982 }, { "epoch": 0.17, "learning_rate": 4.720463466452709e-05, "loss": 0.8527, "step": 4983 }, { "epoch": 0.17, "learning_rate": 4.720341122033862e-05, "loss": 4.3267, "step": 4984 }, { "epoch": 0.17, "learning_rate": 4.720218752433873e-05, "loss": 0.9197, "step": 4985 }, { "epoch": 0.17, "learning_rate": 4.720096357654129e-05, "loss": 1.5709, "step": 4986 }, { "epoch": 0.17, "learning_rate": 4.7199739376960204e-05, "loss": 1.8612, "step": 4987 }, { "epoch": 0.17, "learning_rate": 4.719851492560935e-05, "loss": 1.0457, "step": 4988 }, { "epoch": 0.17, "learning_rate": 4.71972902225026e-05, "loss": 0.8585, "step": 4989 }, { "epoch": 0.17, "learning_rate": 4.719606526765386e-05, "loss": 0.7024, "step": 4990 }, { "epoch": 0.17, "learning_rate": 4.7194840061077005e-05, "loss": 0.9759, "step": 4991 }, { "epoch": 0.17, "learning_rate": 4.719361460278594e-05, "loss": 0.8433, "step": 4992 }, { "epoch": 0.17, "learning_rate": 4.719238889279457e-05, "loss": 1.0231, "step": 4993 }, { "epoch": 0.17, "learning_rate": 4.7191162931116796e-05, "loss": 0.9419, "step": 4994 }, { "epoch": 0.17, "learning_rate": 4.71899367177665e-05, "loss": 0.9517, "step": 4995 }, { "epoch": 0.17, "learning_rate": 4.718871025275762e-05, "loss": 0.7386, "step": 4996 }, { "epoch": 0.17, "learning_rate": 4.718748353610404e-05, "loss": 0.7382, "step": 4997 }, { "epoch": 0.17, "learning_rate": 4.7186256567819675e-05, "loss": 0.9992, "step": 4998 }, { "epoch": 0.17, "learning_rate": 4.718502934791846e-05, "loss": 0.7385, "step": 4999 }, { "epoch": 0.17, "learning_rate": 4.7183801876414294e-05, "loss": 0.7965, "step": 5000 }, { "epoch": 0.17, "learning_rate": 4.71825741533211e-05, "loss": 1.0488, "step": 5001 }, { "epoch": 0.17, "learning_rate": 4.718134617865281e-05, "loss": 0.7373, "step": 5002 }, { "epoch": 0.17, "learning_rate": 4.718011795242334e-05, "loss": 0.9321, "step": 5003 }, { "epoch": 0.17, "learning_rate": 4.7178889474646635e-05, "loss": 0.773, "step": 5004 }, { "epoch": 0.17, "learning_rate": 4.717766074533662e-05, "loss": 1.3636, "step": 5005 }, { "epoch": 0.17, "learning_rate": 4.7176431764507214e-05, "loss": 1.3007, "step": 5006 }, { "epoch": 0.17, "learning_rate": 4.717520253217238e-05, "loss": 0.1374, "step": 5007 }, { "epoch": 0.17, "learning_rate": 4.7173973048346044e-05, "loss": 0.228, "step": 5008 }, { "epoch": 0.17, "learning_rate": 4.717274331304215e-05, "loss": 0.2882, "step": 5009 }, { "epoch": 0.17, "learning_rate": 4.7171513326274655e-05, "loss": 0.2394, "step": 5010 }, { "epoch": 0.17, "learning_rate": 4.7170283088057496e-05, "loss": 2.5503, "step": 5011 }, { "epoch": 0.17, "learning_rate": 4.716905259840463e-05, "loss": 4.283, "step": 5012 }, { "epoch": 0.17, "learning_rate": 4.716782185733002e-05, "loss": 1.1339, "step": 5013 }, { "epoch": 0.17, "learning_rate": 4.716659086484761e-05, "loss": 0.7672, "step": 5014 }, { "epoch": 0.17, "learning_rate": 4.7165359620971375e-05, "loss": 1.0731, "step": 5015 }, { "epoch": 0.17, "learning_rate": 4.716412812571526e-05, "loss": 0.8234, "step": 5016 }, { "epoch": 0.17, "learning_rate": 4.716289637909325e-05, "loss": 0.8888, "step": 5017 }, { "epoch": 0.17, "learning_rate": 4.716166438111931e-05, "loss": 0.8116, "step": 5018 }, { "epoch": 0.17, "learning_rate": 4.7160432131807404e-05, "loss": 0.4977, "step": 5019 }, { "epoch": 0.17, "learning_rate": 4.715919963117151e-05, "loss": 0.9124, "step": 5020 }, { "epoch": 0.17, "learning_rate": 4.715796687922562e-05, "loss": 0.5677, "step": 5021 }, { "epoch": 0.17, "learning_rate": 4.715673387598369e-05, "loss": 0.8374, "step": 5022 }, { "epoch": 0.17, "learning_rate": 4.7155500621459724e-05, "loss": 0.6283, "step": 5023 }, { "epoch": 0.17, "learning_rate": 4.71542671156677e-05, "loss": 1.4394, "step": 5024 }, { "epoch": 0.17, "learning_rate": 4.715303335862161e-05, "loss": 1.5607, "step": 5025 }, { "epoch": 0.17, "learning_rate": 4.715179935033545e-05, "loss": 1.6771, "step": 5026 }, { "epoch": 0.17, "learning_rate": 4.715056509082321e-05, "loss": 1.6745, "step": 5027 }, { "epoch": 0.17, "learning_rate": 4.714933058009888e-05, "loss": 1.017, "step": 5028 }, { "epoch": 0.17, "learning_rate": 4.714809581817647e-05, "loss": 0.869, "step": 5029 }, { "epoch": 0.17, "learning_rate": 4.7146860805069984e-05, "loss": 0.5864, "step": 5030 }, { "epoch": 0.17, "learning_rate": 4.714562554079342e-05, "loss": 0.6453, "step": 5031 }, { "epoch": 0.17, "learning_rate": 4.7144390025360805e-05, "loss": 0.5345, "step": 5032 }, { "epoch": 0.17, "learning_rate": 4.7143154258786137e-05, "loss": 0.752, "step": 5033 }, { "epoch": 0.17, "learning_rate": 4.7141918241083425e-05, "loss": 1.436, "step": 5034 }, { "epoch": 0.17, "learning_rate": 4.714068197226671e-05, "loss": 0.8318, "step": 5035 }, { "epoch": 0.17, "learning_rate": 4.7139445452349984e-05, "loss": 0.877, "step": 5036 }, { "epoch": 0.17, "learning_rate": 4.7138208681347296e-05, "loss": 0.9181, "step": 5037 }, { "epoch": 0.17, "learning_rate": 4.713697165927266e-05, "loss": 0.6559, "step": 5038 }, { "epoch": 0.17, "learning_rate": 4.71357343861401e-05, "loss": 1.1878, "step": 5039 }, { "epoch": 0.17, "learning_rate": 4.713449686196366e-05, "loss": 0.939, "step": 5040 }, { "epoch": 0.17, "learning_rate": 4.713325908675736e-05, "loss": 0.9773, "step": 5041 }, { "epoch": 0.17, "learning_rate": 4.7132021060535255e-05, "loss": 1.1932, "step": 5042 }, { "epoch": 0.17, "learning_rate": 4.713078278331138e-05, "loss": 0.7641, "step": 5043 }, { "epoch": 0.17, "learning_rate": 4.712954425509978e-05, "loss": 1.056, "step": 5044 }, { "epoch": 0.17, "learning_rate": 4.712830547591449e-05, "loss": 0.9389, "step": 5045 }, { "epoch": 0.17, "learning_rate": 4.7127066445769566e-05, "loss": 0.8611, "step": 5046 }, { "epoch": 0.17, "learning_rate": 4.712582716467906e-05, "loss": 0.851, "step": 5047 }, { "epoch": 0.17, "learning_rate": 4.712458763265704e-05, "loss": 0.7706, "step": 5048 }, { "epoch": 0.17, "learning_rate": 4.712334784971754e-05, "loss": 0.4795, "step": 5049 }, { "epoch": 0.17, "learning_rate": 4.712210781587463e-05, "loss": 0.8588, "step": 5050 }, { "epoch": 0.17, "learning_rate": 4.712086753114238e-05, "loss": 0.8935, "step": 5051 }, { "epoch": 0.17, "learning_rate": 4.7119626995534845e-05, "loss": 0.9923, "step": 5052 }, { "epoch": 0.17, "learning_rate": 4.71183862090661e-05, "loss": 1.2373, "step": 5053 }, { "epoch": 0.17, "learning_rate": 4.711714517175023e-05, "loss": 1.1128, "step": 5054 }, { "epoch": 0.17, "learning_rate": 4.7115903883601286e-05, "loss": 0.4101, "step": 5055 }, { "epoch": 0.17, "learning_rate": 4.711466234463335e-05, "loss": 0.8374, "step": 5056 }, { "epoch": 0.17, "learning_rate": 4.711342055486053e-05, "loss": 0.906, "step": 5057 }, { "epoch": 0.17, "learning_rate": 4.711217851429687e-05, "loss": 5.1744, "step": 5058 }, { "epoch": 0.17, "learning_rate": 4.711093622295648e-05, "loss": 0.7442, "step": 5059 }, { "epoch": 0.17, "learning_rate": 4.710969368085345e-05, "loss": 0.5989, "step": 5060 }, { "epoch": 0.17, "learning_rate": 4.710845088800185e-05, "loss": 1.4034, "step": 5061 }, { "epoch": 0.17, "learning_rate": 4.71072078444158e-05, "loss": 1.5149, "step": 5062 }, { "epoch": 0.17, "learning_rate": 4.710596455010939e-05, "loss": 1.6182, "step": 5063 }, { "epoch": 0.17, "learning_rate": 4.710472100509672e-05, "loss": 1.6857, "step": 5064 }, { "epoch": 0.17, "learning_rate": 4.7103477209391875e-05, "loss": 1.5782, "step": 5065 }, { "epoch": 0.17, "learning_rate": 4.710223316300899e-05, "loss": 1.5743, "step": 5066 }, { "epoch": 0.17, "learning_rate": 4.710098886596216e-05, "loss": 1.0213, "step": 5067 }, { "epoch": 0.17, "learning_rate": 4.7099744318265494e-05, "loss": 1.002, "step": 5068 }, { "epoch": 0.17, "learning_rate": 4.7098499519933116e-05, "loss": 1.0079, "step": 5069 }, { "epoch": 0.17, "learning_rate": 4.709725447097913e-05, "loss": 0.9514, "step": 5070 }, { "epoch": 0.17, "learning_rate": 4.7096009171417665e-05, "loss": 0.8861, "step": 5071 }, { "epoch": 0.17, "learning_rate": 4.709476362126284e-05, "loss": 1.4126, "step": 5072 }, { "epoch": 0.17, "learning_rate": 4.709351782052879e-05, "loss": 0.9472, "step": 5073 }, { "epoch": 0.17, "learning_rate": 4.7092271769229634e-05, "loss": 0.8945, "step": 5074 }, { "epoch": 0.17, "learning_rate": 4.7091025467379514e-05, "loss": 1.3198, "step": 5075 }, { "epoch": 0.17, "learning_rate": 4.7089778914992546e-05, "loss": 0.9521, "step": 5076 }, { "epoch": 0.17, "learning_rate": 4.7088532112082885e-05, "loss": 0.933, "step": 5077 }, { "epoch": 0.17, "learning_rate": 4.7087285058664666e-05, "loss": 0.849, "step": 5078 }, { "epoch": 0.17, "learning_rate": 4.7086037754752025e-05, "loss": 1.0886, "step": 5079 }, { "epoch": 0.17, "learning_rate": 4.708479020035912e-05, "loss": 1.6958, "step": 5080 }, { "epoch": 0.17, "learning_rate": 4.708354239550009e-05, "loss": 1.8448, "step": 5081 }, { "epoch": 0.17, "learning_rate": 4.708229434018909e-05, "loss": 1.8203, "step": 5082 }, { "epoch": 0.17, "learning_rate": 4.7081046034440276e-05, "loss": 0.9925, "step": 5083 }, { "epoch": 0.17, "learning_rate": 4.7079797478267804e-05, "loss": 0.6766, "step": 5084 }, { "epoch": 0.17, "learning_rate": 4.707854867168583e-05, "loss": 0.9233, "step": 5085 }, { "epoch": 0.17, "learning_rate": 4.707729961470853e-05, "loss": 0.8194, "step": 5086 }, { "epoch": 0.17, "learning_rate": 4.707605030735005e-05, "loss": 0.9085, "step": 5087 }, { "epoch": 0.17, "learning_rate": 4.707480074962457e-05, "loss": 0.8361, "step": 5088 }, { "epoch": 0.17, "learning_rate": 4.707355094154626e-05, "loss": 0.9174, "step": 5089 }, { "epoch": 0.17, "learning_rate": 4.70723008831293e-05, "loss": 0.8521, "step": 5090 }, { "epoch": 0.17, "learning_rate": 4.707105057438785e-05, "loss": 0.9958, "step": 5091 }, { "epoch": 0.17, "learning_rate": 4.706980001533611e-05, "loss": 0.6438, "step": 5092 }, { "epoch": 0.17, "learning_rate": 4.706854920598824e-05, "loss": 0.9774, "step": 5093 }, { "epoch": 0.17, "learning_rate": 4.7067298146358454e-05, "loss": 0.962, "step": 5094 }, { "epoch": 0.17, "learning_rate": 4.706604683646092e-05, "loss": 0.9019, "step": 5095 }, { "epoch": 0.17, "learning_rate": 4.7064795276309835e-05, "loss": 0.9218, "step": 5096 }, { "epoch": 0.17, "learning_rate": 4.70635434659194e-05, "loss": 0.9588, "step": 5097 }, { "epoch": 0.17, "learning_rate": 4.70622914053038e-05, "loss": 0.8179, "step": 5098 }, { "epoch": 0.17, "learning_rate": 4.706103909447724e-05, "loss": 0.5901, "step": 5099 }, { "epoch": 0.17, "learning_rate": 4.705978653345392e-05, "loss": 0.861, "step": 5100 }, { "epoch": 0.17, "learning_rate": 4.705853372224805e-05, "loss": 0.8193, "step": 5101 }, { "epoch": 0.17, "learning_rate": 4.7057280660873835e-05, "loss": 1.4044, "step": 5102 }, { "epoch": 0.17, "learning_rate": 4.705602734934549e-05, "loss": 0.9086, "step": 5103 }, { "epoch": 0.17, "learning_rate": 4.705477378767722e-05, "loss": 0.7262, "step": 5104 }, { "epoch": 0.17, "learning_rate": 4.705351997588325e-05, "loss": 0.9309, "step": 5105 }, { "epoch": 0.17, "learning_rate": 4.70522659139778e-05, "loss": 0.7734, "step": 5106 }, { "epoch": 0.17, "learning_rate": 4.705101160197509e-05, "loss": 0.9833, "step": 5107 }, { "epoch": 0.17, "learning_rate": 4.704975703988935e-05, "loss": 1.1118, "step": 5108 }, { "epoch": 0.17, "learning_rate": 4.704850222773479e-05, "loss": 0.9225, "step": 5109 }, { "epoch": 0.17, "learning_rate": 4.704724716552567e-05, "loss": 0.607, "step": 5110 }, { "epoch": 0.17, "learning_rate": 4.7045991853276195e-05, "loss": 0.7509, "step": 5111 }, { "epoch": 0.17, "learning_rate": 4.704473629100062e-05, "loss": 1.5112, "step": 5112 }, { "epoch": 0.17, "learning_rate": 4.704348047871319e-05, "loss": 1.6093, "step": 5113 }, { "epoch": 0.17, "learning_rate": 4.704222441642812e-05, "loss": 1.793, "step": 5114 }, { "epoch": 0.17, "learning_rate": 4.704096810415968e-05, "loss": 1.6056, "step": 5115 }, { "epoch": 0.17, "learning_rate": 4.7039711541922104e-05, "loss": 0.9082, "step": 5116 }, { "epoch": 0.17, "learning_rate": 4.703845472972965e-05, "loss": 0.9238, "step": 5117 }, { "epoch": 0.17, "learning_rate": 4.7037197667596576e-05, "loss": 0.7432, "step": 5118 }, { "epoch": 0.17, "learning_rate": 4.703594035553713e-05, "loss": 0.8431, "step": 5119 }, { "epoch": 0.17, "learning_rate": 4.7034682793565574e-05, "loss": 0.7851, "step": 5120 }, { "epoch": 0.17, "learning_rate": 4.703342498169617e-05, "loss": 0.9458, "step": 5121 }, { "epoch": 0.17, "learning_rate": 4.7032166919943186e-05, "loss": 0.804, "step": 5122 }, { "epoch": 0.17, "learning_rate": 4.703090860832088e-05, "loss": 0.7435, "step": 5123 }, { "epoch": 0.17, "learning_rate": 4.702965004684353e-05, "loss": 0.9051, "step": 5124 }, { "epoch": 0.17, "learning_rate": 4.702839123552541e-05, "loss": 0.9139, "step": 5125 }, { "epoch": 0.17, "learning_rate": 4.7027132174380804e-05, "loss": 0.6594, "step": 5126 }, { "epoch": 0.17, "learning_rate": 4.702587286342398e-05, "loss": 0.9943, "step": 5127 }, { "epoch": 0.17, "learning_rate": 4.702461330266922e-05, "loss": 0.7934, "step": 5128 }, { "epoch": 0.17, "learning_rate": 4.7023353492130804e-05, "loss": 1.0052, "step": 5129 }, { "epoch": 0.17, "learning_rate": 4.702209343182304e-05, "loss": 0.9433, "step": 5130 }, { "epoch": 0.17, "learning_rate": 4.70208331217602e-05, "loss": 0.6747, "step": 5131 }, { "epoch": 0.17, "learning_rate": 4.7019572561956585e-05, "loss": 0.8363, "step": 5132 }, { "epoch": 0.17, "learning_rate": 4.701831175242648e-05, "loss": 0.6659, "step": 5133 }, { "epoch": 0.17, "learning_rate": 4.70170506931842e-05, "loss": 0.8049, "step": 5134 }, { "epoch": 0.17, "learning_rate": 4.701578938424404e-05, "loss": 0.9542, "step": 5135 }, { "epoch": 0.17, "learning_rate": 4.70145278256203e-05, "loss": 1.1294, "step": 5136 }, { "epoch": 0.17, "learning_rate": 4.70132660173273e-05, "loss": 0.8376, "step": 5137 }, { "epoch": 0.17, "learning_rate": 4.7012003959379336e-05, "loss": 0.807, "step": 5138 }, { "epoch": 0.17, "learning_rate": 4.701074165179073e-05, "loss": 0.8642, "step": 5139 }, { "epoch": 0.17, "learning_rate": 4.700947909457579e-05, "loss": 0.9508, "step": 5140 }, { "epoch": 0.17, "learning_rate": 4.7008216287748854e-05, "loss": 0.7383, "step": 5141 }, { "epoch": 0.17, "learning_rate": 4.7006953231324214e-05, "loss": 0.8461, "step": 5142 }, { "epoch": 0.17, "learning_rate": 4.700568992531622e-05, "loss": 0.9116, "step": 5143 }, { "epoch": 0.17, "learning_rate": 4.700442636973918e-05, "loss": 0.9945, "step": 5144 }, { "epoch": 0.17, "learning_rate": 4.700316256460744e-05, "loss": 0.9568, "step": 5145 }, { "epoch": 0.17, "learning_rate": 4.700189850993533e-05, "loss": 0.9155, "step": 5146 }, { "epoch": 0.17, "learning_rate": 4.700063420573718e-05, "loss": 0.8181, "step": 5147 }, { "epoch": 0.17, "learning_rate": 4.699936965202734e-05, "loss": 0.8679, "step": 5148 }, { "epoch": 0.17, "learning_rate": 4.699810484882012e-05, "loss": 0.8561, "step": 5149 }, { "epoch": 0.17, "learning_rate": 4.69968397961299e-05, "loss": 1.1029, "step": 5150 }, { "epoch": 0.17, "learning_rate": 4.699557449397102e-05, "loss": 1.2446, "step": 5151 }, { "epoch": 0.17, "learning_rate": 4.699430894235782e-05, "loss": 1.4797, "step": 5152 }, { "epoch": 0.17, "learning_rate": 4.699304314130465e-05, "loss": 1.6688, "step": 5153 }, { "epoch": 0.17, "learning_rate": 4.6991777090825876e-05, "loss": 1.6627, "step": 5154 }, { "epoch": 0.17, "learning_rate": 4.699051079093586e-05, "loss": 1.5885, "step": 5155 }, { "epoch": 0.17, "learning_rate": 4.6989244241648946e-05, "loss": 0.8593, "step": 5156 }, { "epoch": 0.17, "learning_rate": 4.698797744297951e-05, "loss": 0.6923, "step": 5157 }, { "epoch": 0.17, "learning_rate": 4.6986710394941915e-05, "loss": 0.8902, "step": 5158 }, { "epoch": 0.17, "learning_rate": 4.698544309755054e-05, "loss": 0.7512, "step": 5159 }, { "epoch": 0.17, "learning_rate": 4.698417555081974e-05, "loss": 1.7434, "step": 5160 }, { "epoch": 0.17, "learning_rate": 4.6982907754763906e-05, "loss": 1.072, "step": 5161 }, { "epoch": 0.17, "learning_rate": 4.6981639709397416e-05, "loss": 0.8941, "step": 5162 }, { "epoch": 0.17, "learning_rate": 4.6980371414734635e-05, "loss": 1.526, "step": 5163 }, { "epoch": 0.17, "learning_rate": 4.697910287078997e-05, "loss": 0.8377, "step": 5164 }, { "epoch": 0.17, "learning_rate": 4.6977834077577785e-05, "loss": 0.8512, "step": 5165 }, { "epoch": 0.17, "learning_rate": 4.697656503511249e-05, "loss": 1.3187, "step": 5166 }, { "epoch": 0.17, "learning_rate": 4.697529574340845e-05, "loss": 0.8023, "step": 5167 }, { "epoch": 0.17, "learning_rate": 4.6974026202480094e-05, "loss": 0.8456, "step": 5168 }, { "epoch": 0.17, "learning_rate": 4.69727564123418e-05, "loss": 0.9676, "step": 5169 }, { "epoch": 0.17, "learning_rate": 4.6971486373007964e-05, "loss": 0.9008, "step": 5170 }, { "epoch": 0.17, "learning_rate": 4.697021608449301e-05, "loss": 0.5179, "step": 5171 }, { "epoch": 0.17, "learning_rate": 4.696894554681133e-05, "loss": 0.5161, "step": 5172 }, { "epoch": 0.17, "learning_rate": 4.696767475997733e-05, "loss": 1.0533, "step": 5173 }, { "epoch": 0.17, "learning_rate": 4.696640372400543e-05, "loss": 1.0168, "step": 5174 }, { "epoch": 0.17, "learning_rate": 4.696513243891004e-05, "loss": 1.1297, "step": 5175 }, { "epoch": 0.17, "learning_rate": 4.696386090470559e-05, "loss": 0.9852, "step": 5176 }, { "epoch": 0.17, "learning_rate": 4.6962589121406486e-05, "loss": 0.9807, "step": 5177 }, { "epoch": 0.17, "learning_rate": 4.6961317089027154e-05, "loss": 0.9693, "step": 5178 }, { "epoch": 0.17, "learning_rate": 4.6960044807582027e-05, "loss": 1.1013, "step": 5179 }, { "epoch": 0.17, "learning_rate": 4.695877227708553e-05, "loss": 1.2592, "step": 5180 }, { "epoch": 0.17, "learning_rate": 4.6957499497552096e-05, "loss": 1.2363, "step": 5181 }, { "epoch": 0.17, "learning_rate": 4.6956226468996146e-05, "loss": 0.7328, "step": 5182 }, { "epoch": 0.17, "learning_rate": 4.695495319143214e-05, "loss": 0.5425, "step": 5183 }, { "epoch": 0.17, "learning_rate": 4.6953679664874514e-05, "loss": 0.6037, "step": 5184 }, { "epoch": 0.17, "learning_rate": 4.69524058893377e-05, "loss": 0.9022, "step": 5185 }, { "epoch": 0.17, "learning_rate": 4.695113186483615e-05, "loss": 0.7999, "step": 5186 }, { "epoch": 0.17, "learning_rate": 4.6949857591384314e-05, "loss": 0.9517, "step": 5187 }, { "epoch": 0.17, "learning_rate": 4.694858306899664e-05, "loss": 1.6363, "step": 5188 }, { "epoch": 0.17, "learning_rate": 4.694730829768759e-05, "loss": 5.6235, "step": 5189 }, { "epoch": 0.17, "learning_rate": 4.694603327747161e-05, "loss": 1.6471, "step": 5190 }, { "epoch": 0.17, "learning_rate": 4.6944758008363166e-05, "loss": 0.8734, "step": 5191 }, { "epoch": 0.17, "learning_rate": 4.694348249037673e-05, "loss": 0.9817, "step": 5192 }, { "epoch": 0.17, "learning_rate": 4.694220672352676e-05, "loss": 0.8671, "step": 5193 }, { "epoch": 0.17, "learning_rate": 4.6940930707827716e-05, "loss": 0.918, "step": 5194 }, { "epoch": 0.17, "learning_rate": 4.693965444329408e-05, "loss": 0.8102, "step": 5195 }, { "epoch": 0.17, "learning_rate": 4.693837792994032e-05, "loss": 0.6552, "step": 5196 }, { "epoch": 0.17, "learning_rate": 4.6937101167780914e-05, "loss": 0.8448, "step": 5197 }, { "epoch": 0.17, "learning_rate": 4.693582415683036e-05, "loss": 1.8984, "step": 5198 }, { "epoch": 0.17, "learning_rate": 4.693454689710311e-05, "loss": 1.4937, "step": 5199 }, { "epoch": 0.17, "learning_rate": 4.693326938861366e-05, "loss": 0.9143, "step": 5200 }, { "epoch": 0.17, "learning_rate": 4.693199163137652e-05, "loss": 0.8782, "step": 5201 }, { "epoch": 0.17, "learning_rate": 4.6930713625406157e-05, "loss": 0.7361, "step": 5202 }, { "epoch": 0.17, "learning_rate": 4.6929435370717075e-05, "loss": 0.1427, "step": 5203 }, { "epoch": 0.17, "learning_rate": 4.6928156867323766e-05, "loss": 0.5919, "step": 5204 }, { "epoch": 0.17, "learning_rate": 4.692687811524073e-05, "loss": 0.9497, "step": 5205 }, { "epoch": 0.17, "learning_rate": 4.692559911448248e-05, "loss": 0.7974, "step": 5206 }, { "epoch": 0.17, "learning_rate": 4.69243198650635e-05, "loss": 0.8911, "step": 5207 }, { "epoch": 0.17, "learning_rate": 4.6923040366998316e-05, "loss": 0.7352, "step": 5208 }, { "epoch": 0.17, "learning_rate": 4.692176062030144e-05, "loss": 0.9456, "step": 5209 }, { "epoch": 0.17, "learning_rate": 4.6920480624987364e-05, "loss": 0.9035, "step": 5210 }, { "epoch": 0.17, "learning_rate": 4.691920038107064e-05, "loss": 0.9854, "step": 5211 }, { "epoch": 0.17, "learning_rate": 4.6917919888565756e-05, "loss": 1.0489, "step": 5212 }, { "epoch": 0.17, "learning_rate": 4.691663914748724e-05, "loss": 0.9798, "step": 5213 }, { "epoch": 0.17, "learning_rate": 4.691535815784963e-05, "loss": 0.4002, "step": 5214 }, { "epoch": 0.17, "learning_rate": 4.691407691966745e-05, "loss": 0.787, "step": 5215 }, { "epoch": 0.17, "learning_rate": 4.6912795432955215e-05, "loss": 0.6408, "step": 5216 }, { "epoch": 0.17, "learning_rate": 4.691151369772748e-05, "loss": 0.8498, "step": 5217 }, { "epoch": 0.17, "learning_rate": 4.6910231713998774e-05, "loss": 0.8801, "step": 5218 }, { "epoch": 0.17, "learning_rate": 4.690894948178363e-05, "loss": 0.8859, "step": 5219 }, { "epoch": 0.17, "learning_rate": 4.690766700109659e-05, "loss": 1.7137, "step": 5220 }, { "epoch": 0.17, "learning_rate": 4.690638427195221e-05, "loss": 0.1811, "step": 5221 }, { "epoch": 0.17, "learning_rate": 4.6905101294365025e-05, "loss": 0.6481, "step": 5222 }, { "epoch": 0.17, "learning_rate": 4.6903818068349595e-05, "loss": 0.9274, "step": 5223 }, { "epoch": 0.17, "learning_rate": 4.690253459392047e-05, "loss": 0.93, "step": 5224 }, { "epoch": 0.17, "learning_rate": 4.6901250871092194e-05, "loss": 0.8319, "step": 5225 }, { "epoch": 0.17, "learning_rate": 4.689996689987935e-05, "loss": 0.7984, "step": 5226 }, { "epoch": 0.17, "learning_rate": 4.689868268029648e-05, "loss": 0.9823, "step": 5227 }, { "epoch": 0.17, "learning_rate": 4.689739821235816e-05, "loss": 0.8356, "step": 5228 }, { "epoch": 0.17, "learning_rate": 4.689611349607894e-05, "loss": 1.5884, "step": 5229 }, { "epoch": 0.17, "learning_rate": 4.689482853147341e-05, "loss": 5.9489, "step": 5230 }, { "epoch": 0.17, "learning_rate": 4.689354331855614e-05, "loss": 2.3798, "step": 5231 }, { "epoch": 0.17, "learning_rate": 4.68922578573417e-05, "loss": 0.9634, "step": 5232 }, { "epoch": 0.17, "learning_rate": 4.689097214784467e-05, "loss": 0.8267, "step": 5233 }, { "epoch": 0.17, "learning_rate": 4.688968619007962e-05, "loss": 0.9757, "step": 5234 }, { "epoch": 0.17, "learning_rate": 4.6888399984061154e-05, "loss": 0.7593, "step": 5235 }, { "epoch": 0.17, "learning_rate": 4.688711352980385e-05, "loss": 0.9005, "step": 5236 }, { "epoch": 0.17, "learning_rate": 4.6885826827322296e-05, "loss": 0.7883, "step": 5237 }, { "epoch": 0.17, "learning_rate": 4.688453987663109e-05, "loss": 1.3114, "step": 5238 }, { "epoch": 0.17, "learning_rate": 4.688325267774482e-05, "loss": 1.0753, "step": 5239 }, { "epoch": 0.17, "learning_rate": 4.6881965230678094e-05, "loss": 0.969, "step": 5240 }, { "epoch": 0.17, "learning_rate": 4.6880677535445505e-05, "loss": 0.8114, "step": 5241 }, { "epoch": 0.17, "learning_rate": 4.687938959206165e-05, "loss": 0.7877, "step": 5242 }, { "epoch": 0.17, "learning_rate": 4.6878101400541156e-05, "loss": 0.5419, "step": 5243 }, { "epoch": 0.17, "learning_rate": 4.687681296089862e-05, "loss": 0.8202, "step": 5244 }, { "epoch": 0.17, "learning_rate": 4.687552427314865e-05, "loss": 0.7481, "step": 5245 }, { "epoch": 0.17, "learning_rate": 4.687423533730587e-05, "loss": 0.3971, "step": 5246 }, { "epoch": 0.17, "learning_rate": 4.68729461533849e-05, "loss": 0.3449, "step": 5247 }, { "epoch": 0.17, "learning_rate": 4.6871656721400357e-05, "loss": 0.8764, "step": 5248 }, { "epoch": 0.17, "learning_rate": 4.687036704136686e-05, "loss": 0.7237, "step": 5249 }, { "epoch": 0.17, "learning_rate": 4.686907711329903e-05, "loss": 0.9274, "step": 5250 }, { "epoch": 0.18, "learning_rate": 4.686778693721152e-05, "loss": 0.4203, "step": 5251 }, { "epoch": 0.18, "learning_rate": 4.686649651311894e-05, "loss": 0.9743, "step": 5252 }, { "epoch": 0.18, "learning_rate": 4.6865205841035934e-05, "loss": 1.1295, "step": 5253 }, { "epoch": 0.18, "learning_rate": 4.686391492097713e-05, "loss": 0.8455, "step": 5254 }, { "epoch": 0.18, "learning_rate": 4.686262375295718e-05, "loss": 1.1015, "step": 5255 }, { "epoch": 0.18, "learning_rate": 4.686133233699073e-05, "loss": 0.791, "step": 5256 }, { "epoch": 0.18, "learning_rate": 4.686004067309241e-05, "loss": 0.9349, "step": 5257 }, { "epoch": 0.18, "learning_rate": 4.685874876127688e-05, "loss": 1.0194, "step": 5258 }, { "epoch": 0.18, "learning_rate": 4.68574566015588e-05, "loss": 1.1104, "step": 5259 }, { "epoch": 0.18, "learning_rate": 4.68561641939528e-05, "loss": 0.7342, "step": 5260 }, { "epoch": 0.18, "learning_rate": 4.685487153847357e-05, "loss": 0.6074, "step": 5261 }, { "epoch": 0.18, "learning_rate": 4.685357863513573e-05, "loss": 0.8739, "step": 5262 }, { "epoch": 0.18, "learning_rate": 4.6852285483953986e-05, "loss": 0.7532, "step": 5263 }, { "epoch": 0.18, "learning_rate": 4.685099208494297e-05, "loss": 0.9776, "step": 5264 }, { "epoch": 0.18, "learning_rate": 4.6849698438117364e-05, "loss": 1.1725, "step": 5265 }, { "epoch": 0.18, "learning_rate": 4.684840454349184e-05, "loss": 0.8503, "step": 5266 }, { "epoch": 0.18, "learning_rate": 4.684711040108107e-05, "loss": 0.8986, "step": 5267 }, { "epoch": 0.18, "learning_rate": 4.684581601089974e-05, "loss": 0.7106, "step": 5268 }, { "epoch": 0.18, "learning_rate": 4.684452137296251e-05, "loss": 1.0843, "step": 5269 }, { "epoch": 0.18, "learning_rate": 4.6843226487284085e-05, "loss": 0.6075, "step": 5270 }, { "epoch": 0.18, "learning_rate": 4.6841931353879145e-05, "loss": 0.9299, "step": 5271 }, { "epoch": 0.18, "learning_rate": 4.684063597276236e-05, "loss": 0.8968, "step": 5272 }, { "epoch": 0.18, "learning_rate": 4.6839340343948435e-05, "loss": 0.7366, "step": 5273 }, { "epoch": 0.18, "learning_rate": 4.683804446745207e-05, "loss": 0.9184, "step": 5274 }, { "epoch": 0.18, "learning_rate": 4.683674834328795e-05, "loss": 0.8849, "step": 5275 }, { "epoch": 0.18, "learning_rate": 4.6835451971470776e-05, "loss": 0.7772, "step": 5276 }, { "epoch": 0.18, "learning_rate": 4.683415535201526e-05, "loss": 0.7354, "step": 5277 }, { "epoch": 0.18, "learning_rate": 4.68328584849361e-05, "loss": 1.1104, "step": 5278 }, { "epoch": 0.18, "learning_rate": 4.683156137024801e-05, "loss": 0.7798, "step": 5279 }, { "epoch": 0.18, "learning_rate": 4.683026400796569e-05, "loss": 0.6707, "step": 5280 }, { "epoch": 0.18, "learning_rate": 4.6828966398103854e-05, "loss": 0.9787, "step": 5281 }, { "epoch": 0.18, "learning_rate": 4.682766854067723e-05, "loss": 0.9678, "step": 5282 }, { "epoch": 0.18, "learning_rate": 4.682637043570053e-05, "loss": 0.9346, "step": 5283 }, { "epoch": 0.18, "learning_rate": 4.682507208318847e-05, "loss": 0.932, "step": 5284 }, { "epoch": 0.18, "learning_rate": 4.682377348315579e-05, "loss": 0.8227, "step": 5285 }, { "epoch": 0.18, "learning_rate": 4.68224746356172e-05, "loss": 0.7574, "step": 5286 }, { "epoch": 0.18, "learning_rate": 4.6821175540587435e-05, "loss": 1.3446, "step": 5287 }, { "epoch": 0.18, "learning_rate": 4.6819876198081246e-05, "loss": 0.9954, "step": 5288 }, { "epoch": 0.18, "learning_rate": 4.681857660811335e-05, "loss": 0.7715, "step": 5289 }, { "epoch": 0.18, "learning_rate": 4.681727677069849e-05, "loss": 0.8826, "step": 5290 }, { "epoch": 0.18, "learning_rate": 4.681597668585141e-05, "loss": 0.9264, "step": 5291 }, { "epoch": 0.18, "learning_rate": 4.681467635358685e-05, "loss": 0.8743, "step": 5292 }, { "epoch": 0.18, "learning_rate": 4.681337577391956e-05, "loss": 0.865, "step": 5293 }, { "epoch": 0.18, "learning_rate": 4.68120749468643e-05, "loss": 0.83, "step": 5294 }, { "epoch": 0.18, "learning_rate": 4.6810773872435806e-05, "loss": 1.1241, "step": 5295 }, { "epoch": 0.18, "learning_rate": 4.680947255064884e-05, "loss": 1.6461, "step": 5296 }, { "epoch": 0.18, "learning_rate": 4.6808170981518165e-05, "loss": 0.9977, "step": 5297 }, { "epoch": 0.18, "learning_rate": 4.680686916505853e-05, "loss": 0.7072, "step": 5298 }, { "epoch": 0.18, "learning_rate": 4.680556710128471e-05, "loss": 0.9203, "step": 5299 }, { "epoch": 0.18, "learning_rate": 4.680426479021147e-05, "loss": 1.0281, "step": 5300 }, { "epoch": 0.18, "learning_rate": 4.680296223185359e-05, "loss": 1.2518, "step": 5301 }, { "epoch": 0.18, "learning_rate": 4.680165942622582e-05, "loss": 0.813, "step": 5302 }, { "epoch": 0.18, "learning_rate": 4.6800356373342944e-05, "loss": 0.5528, "step": 5303 }, { "epoch": 0.18, "learning_rate": 4.679905307321976e-05, "loss": 0.5584, "step": 5304 }, { "epoch": 0.18, "learning_rate": 4.6797749525871015e-05, "loss": 0.8467, "step": 5305 }, { "epoch": 0.18, "learning_rate": 4.679644573131151e-05, "loss": 0.8562, "step": 5306 }, { "epoch": 0.18, "learning_rate": 4.679514168955603e-05, "loss": 1.0014, "step": 5307 }, { "epoch": 0.18, "learning_rate": 4.679383740061938e-05, "loss": 1.0197, "step": 5308 }, { "epoch": 0.18, "learning_rate": 4.679253286451633e-05, "loss": 0.9237, "step": 5309 }, { "epoch": 0.18, "learning_rate": 4.6791228081261675e-05, "loss": 1.3902, "step": 5310 }, { "epoch": 0.18, "learning_rate": 4.6789923050870224e-05, "loss": 0.8247, "step": 5311 }, { "epoch": 0.18, "learning_rate": 4.678861777335677e-05, "loss": 0.7258, "step": 5312 }, { "epoch": 0.18, "learning_rate": 4.678731224873613e-05, "loss": 1.0151, "step": 5313 }, { "epoch": 0.18, "learning_rate": 4.678600647702309e-05, "loss": 0.9166, "step": 5314 }, { "epoch": 0.18, "learning_rate": 4.6784700458232464e-05, "loss": 1.1243, "step": 5315 }, { "epoch": 0.18, "learning_rate": 4.678339419237908e-05, "loss": 1.025, "step": 5316 }, { "epoch": 0.18, "learning_rate": 4.678208767947774e-05, "loss": 0.8545, "step": 5317 }, { "epoch": 0.18, "learning_rate": 4.678078091954325e-05, "loss": 0.9725, "step": 5318 }, { "epoch": 0.18, "learning_rate": 4.677947391259045e-05, "loss": 0.6853, "step": 5319 }, { "epoch": 0.18, "learning_rate": 4.6778166658634164e-05, "loss": 0.6729, "step": 5320 }, { "epoch": 0.18, "learning_rate": 4.67768591576892e-05, "loss": 0.8185, "step": 5321 }, { "epoch": 0.18, "learning_rate": 4.67755514097704e-05, "loss": 0.8097, "step": 5322 }, { "epoch": 0.18, "learning_rate": 4.677424341489258e-05, "loss": 1.2773, "step": 5323 }, { "epoch": 0.18, "learning_rate": 4.67729351730706e-05, "loss": 0.8701, "step": 5324 }, { "epoch": 0.18, "learning_rate": 4.677162668431927e-05, "loss": 0.9031, "step": 5325 }, { "epoch": 0.18, "learning_rate": 4.677031794865344e-05, "loss": 0.4635, "step": 5326 }, { "epoch": 0.18, "learning_rate": 4.676900896608797e-05, "loss": 0.7214, "step": 5327 }, { "epoch": 0.18, "learning_rate": 4.676769973663767e-05, "loss": 1.0035, "step": 5328 }, { "epoch": 0.18, "learning_rate": 4.6766390260317415e-05, "loss": 1.3377, "step": 5329 }, { "epoch": 0.18, "learning_rate": 4.6765080537142056e-05, "loss": 1.4332, "step": 5330 }, { "epoch": 0.18, "learning_rate": 4.676377056712643e-05, "loss": 1.5678, "step": 5331 }, { "epoch": 0.18, "learning_rate": 4.676246035028541e-05, "loss": 1.5821, "step": 5332 }, { "epoch": 0.18, "learning_rate": 4.6761149886633834e-05, "loss": 1.2516, "step": 5333 }, { "epoch": 0.18, "learning_rate": 4.675983917618659e-05, "loss": 0.7474, "step": 5334 }, { "epoch": 0.18, "learning_rate": 4.675852821895853e-05, "loss": 0.9139, "step": 5335 }, { "epoch": 0.18, "learning_rate": 4.6757217014964524e-05, "loss": 1.0374, "step": 5336 }, { "epoch": 0.18, "learning_rate": 4.675590556421943e-05, "loss": 0.9667, "step": 5337 }, { "epoch": 0.18, "learning_rate": 4.675459386673815e-05, "loss": 0.8622, "step": 5338 }, { "epoch": 0.18, "learning_rate": 4.675328192253553e-05, "loss": 1.4943, "step": 5339 }, { "epoch": 0.18, "learning_rate": 4.675196973162646e-05, "loss": 4.1457, "step": 5340 }, { "epoch": 0.18, "learning_rate": 4.6750657294025826e-05, "loss": 1.4156, "step": 5341 }, { "epoch": 0.18, "learning_rate": 4.6749344609748516e-05, "loss": 0.8118, "step": 5342 }, { "epoch": 0.18, "learning_rate": 4.6748031678809404e-05, "loss": 0.9734, "step": 5343 }, { "epoch": 0.18, "learning_rate": 4.674671850122339e-05, "loss": 0.7696, "step": 5344 }, { "epoch": 0.18, "learning_rate": 4.6745405077005366e-05, "loss": 0.6523, "step": 5345 }, { "epoch": 0.18, "learning_rate": 4.674409140617022e-05, "loss": 0.8678, "step": 5346 }, { "epoch": 0.18, "learning_rate": 4.674277748873286e-05, "loss": 0.6999, "step": 5347 }, { "epoch": 0.18, "learning_rate": 4.674146332470818e-05, "loss": 0.1464, "step": 5348 }, { "epoch": 0.18, "learning_rate": 4.6740148914111094e-05, "loss": 0.2064, "step": 5349 }, { "epoch": 0.18, "learning_rate": 4.67388342569565e-05, "loss": 0.8683, "step": 5350 }, { "epoch": 0.18, "learning_rate": 4.673751935325931e-05, "loss": 1.0513, "step": 5351 }, { "epoch": 0.18, "learning_rate": 4.6736204203034434e-05, "loss": 0.8713, "step": 5352 }, { "epoch": 0.18, "learning_rate": 4.673488880629679e-05, "loss": 0.9038, "step": 5353 }, { "epoch": 0.18, "learning_rate": 4.67335731630613e-05, "loss": 0.8466, "step": 5354 }, { "epoch": 0.18, "learning_rate": 4.673225727334288e-05, "loss": 1.1064, "step": 5355 }, { "epoch": 0.18, "learning_rate": 4.673094113715645e-05, "loss": 0.8613, "step": 5356 }, { "epoch": 0.18, "learning_rate": 4.6729624754516945e-05, "loss": 0.7757, "step": 5357 }, { "epoch": 0.18, "learning_rate": 4.672830812543929e-05, "loss": 1.1438, "step": 5358 }, { "epoch": 0.18, "learning_rate": 4.672699124993841e-05, "loss": 0.1155, "step": 5359 }, { "epoch": 0.18, "learning_rate": 4.6725674128029255e-05, "loss": 0.3136, "step": 5360 }, { "epoch": 0.18, "learning_rate": 4.672435675972675e-05, "loss": 0.9806, "step": 5361 }, { "epoch": 0.18, "learning_rate": 4.672303914504585e-05, "loss": 0.8476, "step": 5362 }, { "epoch": 0.18, "learning_rate": 4.6721721284001475e-05, "loss": 1.0275, "step": 5363 }, { "epoch": 0.18, "learning_rate": 4.6720403176608594e-05, "loss": 1.2374, "step": 5364 }, { "epoch": 0.18, "learning_rate": 4.671908482288214e-05, "loss": 0.925, "step": 5365 }, { "epoch": 0.18, "learning_rate": 4.671776622283707e-05, "loss": 0.4184, "step": 5366 }, { "epoch": 0.18, "learning_rate": 4.671644737648835e-05, "loss": 0.3813, "step": 5367 }, { "epoch": 0.18, "learning_rate": 4.671512828385092e-05, "loss": 1.0334, "step": 5368 }, { "epoch": 0.18, "learning_rate": 4.671380894493974e-05, "loss": 0.8974, "step": 5369 }, { "epoch": 0.18, "learning_rate": 4.6712489359769784e-05, "loss": 0.9253, "step": 5370 }, { "epoch": 0.18, "learning_rate": 4.6711169528356014e-05, "loss": 5.5494, "step": 5371 }, { "epoch": 0.18, "learning_rate": 4.670984945071339e-05, "loss": 5.3436, "step": 5372 }, { "epoch": 0.18, "learning_rate": 4.670852912685691e-05, "loss": 1.0898, "step": 5373 }, { "epoch": 0.18, "learning_rate": 4.67072085568015e-05, "loss": 0.8229, "step": 5374 }, { "epoch": 0.18, "learning_rate": 4.670588774056218e-05, "loss": 0.8958, "step": 5375 }, { "epoch": 0.18, "learning_rate": 4.670456667815392e-05, "loss": 0.8219, "step": 5376 }, { "epoch": 0.18, "learning_rate": 4.6703245369591686e-05, "loss": 1.0937, "step": 5377 }, { "epoch": 0.18, "learning_rate": 4.670192381489048e-05, "loss": 0.6622, "step": 5378 }, { "epoch": 0.18, "learning_rate": 4.670060201406528e-05, "loss": 0.0569, "step": 5379 }, { "epoch": 0.18, "learning_rate": 4.669927996713108e-05, "loss": 0.5422, "step": 5380 }, { "epoch": 0.18, "learning_rate": 4.669795767410287e-05, "loss": 1.9424, "step": 5381 }, { "epoch": 0.18, "learning_rate": 4.669663513499566e-05, "loss": 0.8541, "step": 5382 }, { "epoch": 0.18, "learning_rate": 4.669531234982443e-05, "loss": 0.8866, "step": 5383 }, { "epoch": 0.18, "learning_rate": 4.66939893186042e-05, "loss": 1.0051, "step": 5384 }, { "epoch": 0.18, "learning_rate": 4.669266604134996e-05, "loss": 0.7545, "step": 5385 }, { "epoch": 0.18, "learning_rate": 4.669134251807672e-05, "loss": 0.7379, "step": 5386 }, { "epoch": 0.18, "learning_rate": 4.6690018748799505e-05, "loss": 1.209, "step": 5387 }, { "epoch": 0.18, "learning_rate": 4.66886947335333e-05, "loss": 0.8587, "step": 5388 }, { "epoch": 0.18, "learning_rate": 4.6687370472293145e-05, "loss": 0.8128, "step": 5389 }, { "epoch": 0.18, "learning_rate": 4.6686045965094056e-05, "loss": 0.955, "step": 5390 }, { "epoch": 0.18, "learning_rate": 4.668472121195104e-05, "loss": 1.0045, "step": 5391 }, { "epoch": 0.18, "learning_rate": 4.6683396212879136e-05, "loss": 1.1422, "step": 5392 }, { "epoch": 0.18, "learning_rate": 4.6682070967893364e-05, "loss": 1.0922, "step": 5393 }, { "epoch": 0.18, "learning_rate": 4.668074547700876e-05, "loss": 0.9344, "step": 5394 }, { "epoch": 0.18, "learning_rate": 4.6679419740240345e-05, "loss": 2.492, "step": 5395 }, { "epoch": 0.18, "learning_rate": 4.667809375760317e-05, "loss": 0.85, "step": 5396 }, { "epoch": 0.18, "learning_rate": 4.667676752911225e-05, "loss": 0.6721, "step": 5397 }, { "epoch": 0.18, "learning_rate": 4.667544105478265e-05, "loss": 0.8648, "step": 5398 }, { "epoch": 0.18, "learning_rate": 4.6674114334629405e-05, "loss": 0.1798, "step": 5399 }, { "epoch": 0.18, "learning_rate": 4.6672787368667556e-05, "loss": 0.1672, "step": 5400 }, { "epoch": 0.18, "learning_rate": 4.667146015691215e-05, "loss": 0.1704, "step": 5401 }, { "epoch": 0.18, "learning_rate": 4.667013269937825e-05, "loss": 0.7557, "step": 5402 }, { "epoch": 0.18, "learning_rate": 4.666880499608092e-05, "loss": 1.0697, "step": 5403 }, { "epoch": 0.18, "learning_rate": 4.666747704703519e-05, "loss": 1.0112, "step": 5404 }, { "epoch": 0.18, "learning_rate": 4.6666148852256134e-05, "loss": 0.88, "step": 5405 }, { "epoch": 0.18, "learning_rate": 4.666482041175881e-05, "loss": 0.339, "step": 5406 }, { "epoch": 0.18, "learning_rate": 4.6663491725558306e-05, "loss": 0.3084, "step": 5407 }, { "epoch": 0.18, "learning_rate": 4.666216279366966e-05, "loss": 0.7124, "step": 5408 }, { "epoch": 0.18, "learning_rate": 4.666083361610797e-05, "loss": 0.4883, "step": 5409 }, { "epoch": 0.18, "learning_rate": 4.6659504192888295e-05, "loss": 0.7864, "step": 5410 }, { "epoch": 0.18, "learning_rate": 4.665817452402571e-05, "loss": 0.6335, "step": 5411 }, { "epoch": 0.18, "learning_rate": 4.66568446095353e-05, "loss": 0.8598, "step": 5412 }, { "epoch": 0.18, "learning_rate": 4.665551444943216e-05, "loss": 0.8066, "step": 5413 }, { "epoch": 0.18, "learning_rate": 4.665418404373135e-05, "loss": 0.9621, "step": 5414 }, { "epoch": 0.18, "learning_rate": 4.6652853392447976e-05, "loss": 1.0967, "step": 5415 }, { "epoch": 0.18, "learning_rate": 4.665152249559712e-05, "loss": 1.0048, "step": 5416 }, { "epoch": 0.18, "learning_rate": 4.665019135319389e-05, "loss": 1.0154, "step": 5417 }, { "epoch": 0.18, "learning_rate": 4.664885996525337e-05, "loss": 1.0603, "step": 5418 }, { "epoch": 0.18, "learning_rate": 4.6647528331790666e-05, "loss": 1.0445, "step": 5419 }, { "epoch": 0.18, "learning_rate": 4.664619645282087e-05, "loss": 0.8677, "step": 5420 }, { "epoch": 0.18, "learning_rate": 4.66448643283591e-05, "loss": 0.9548, "step": 5421 }, { "epoch": 0.18, "learning_rate": 4.664353195842045e-05, "loss": 0.9592, "step": 5422 }, { "epoch": 0.18, "learning_rate": 4.664219934302005e-05, "loss": 1.2695, "step": 5423 }, { "epoch": 0.18, "learning_rate": 4.664086648217299e-05, "loss": 0.824, "step": 5424 }, { "epoch": 0.18, "learning_rate": 4.663953337589441e-05, "loss": 0.7581, "step": 5425 }, { "epoch": 0.18, "learning_rate": 4.663820002419941e-05, "loss": 1.0496, "step": 5426 }, { "epoch": 0.18, "learning_rate": 4.663686642710312e-05, "loss": 0.905, "step": 5427 }, { "epoch": 0.18, "learning_rate": 4.6635532584620655e-05, "loss": 0.8996, "step": 5428 }, { "epoch": 0.18, "learning_rate": 4.6634198496767154e-05, "loss": 1.4561, "step": 5429 }, { "epoch": 0.18, "learning_rate": 4.6632864163557746e-05, "loss": 1.7006, "step": 5430 }, { "epoch": 0.18, "learning_rate": 4.663152958500756e-05, "loss": 0.9886, "step": 5431 }, { "epoch": 0.18, "learning_rate": 4.663019476113174e-05, "loss": 0.9934, "step": 5432 }, { "epoch": 0.18, "learning_rate": 4.66288596919454e-05, "loss": 1.007, "step": 5433 }, { "epoch": 0.18, "learning_rate": 4.662752437746372e-05, "loss": 0.9799, "step": 5434 }, { "epoch": 0.18, "learning_rate": 4.662618881770181e-05, "loss": 0.8109, "step": 5435 }, { "epoch": 0.18, "learning_rate": 4.662485301267483e-05, "loss": 1.0139, "step": 5436 }, { "epoch": 0.18, "learning_rate": 4.662351696239793e-05, "loss": 0.9312, "step": 5437 }, { "epoch": 0.18, "learning_rate": 4.662218066688625e-05, "loss": 0.7408, "step": 5438 }, { "epoch": 0.18, "learning_rate": 4.662084412615497e-05, "loss": 0.7182, "step": 5439 }, { "epoch": 0.18, "learning_rate": 4.661950734021923e-05, "loss": 0.7737, "step": 5440 }, { "epoch": 0.18, "learning_rate": 4.6618170309094186e-05, "loss": 0.8762, "step": 5441 }, { "epoch": 0.18, "learning_rate": 4.6616833032795016e-05, "loss": 0.7362, "step": 5442 }, { "epoch": 0.18, "learning_rate": 4.6615495511336884e-05, "loss": 0.9114, "step": 5443 }, { "epoch": 0.18, "learning_rate": 4.6614157744734956e-05, "loss": 0.9446, "step": 5444 }, { "epoch": 0.18, "learning_rate": 4.661281973300439e-05, "loss": 0.6978, "step": 5445 }, { "epoch": 0.18, "learning_rate": 4.6611481476160385e-05, "loss": 0.5968, "step": 5446 }, { "epoch": 0.18, "learning_rate": 4.661014297421809e-05, "loss": 0.8834, "step": 5447 }, { "epoch": 0.18, "learning_rate": 4.6608804227192726e-05, "loss": 1.1104, "step": 5448 }, { "epoch": 0.18, "learning_rate": 4.660746523509944e-05, "loss": 0.7414, "step": 5449 }, { "epoch": 0.18, "learning_rate": 4.6606125997953425e-05, "loss": 0.9229, "step": 5450 }, { "epoch": 0.18, "learning_rate": 4.660478651576988e-05, "loss": 0.8473, "step": 5451 }, { "epoch": 0.18, "learning_rate": 4.6603446788563985e-05, "loss": 0.9975, "step": 5452 }, { "epoch": 0.18, "learning_rate": 4.660210681635094e-05, "loss": 1.8035, "step": 5453 }, { "epoch": 0.18, "learning_rate": 4.6600766599145946e-05, "loss": 1.8282, "step": 5454 }, { "epoch": 0.18, "learning_rate": 4.659942613696419e-05, "loss": 1.9582, "step": 5455 }, { "epoch": 0.18, "learning_rate": 4.659808542982088e-05, "loss": 0.8778, "step": 5456 }, { "epoch": 0.18, "learning_rate": 4.659674447773124e-05, "loss": 0.8193, "step": 5457 }, { "epoch": 0.18, "learning_rate": 4.659540328071044e-05, "loss": 0.6461, "step": 5458 }, { "epoch": 0.18, "learning_rate": 4.659406183877372e-05, "loss": 0.761, "step": 5459 }, { "epoch": 0.18, "learning_rate": 4.6592720151936284e-05, "loss": 0.8974, "step": 5460 }, { "epoch": 0.18, "learning_rate": 4.659137822021334e-05, "loss": 0.8846, "step": 5461 }, { "epoch": 0.18, "learning_rate": 4.659003604362013e-05, "loss": 0.7803, "step": 5462 }, { "epoch": 0.18, "learning_rate": 4.6588693622171854e-05, "loss": 1.014, "step": 5463 }, { "epoch": 0.18, "learning_rate": 4.6587350955883746e-05, "loss": 0.7965, "step": 5464 }, { "epoch": 0.18, "learning_rate": 4.658600804477104e-05, "loss": 0.6863, "step": 5465 }, { "epoch": 0.18, "learning_rate": 4.658466488884894e-05, "loss": 0.4286, "step": 5466 }, { "epoch": 0.18, "learning_rate": 4.65833214881327e-05, "loss": 0.5446, "step": 5467 }, { "epoch": 0.18, "learning_rate": 4.6581977842637564e-05, "loss": 0.8308, "step": 5468 }, { "epoch": 0.18, "learning_rate": 4.658063395237875e-05, "loss": 0.7443, "step": 5469 }, { "epoch": 0.18, "learning_rate": 4.6579289817371506e-05, "loss": 0.9361, "step": 5470 }, { "epoch": 0.18, "learning_rate": 4.657794543763109e-05, "loss": 1.0181, "step": 5471 }, { "epoch": 0.18, "learning_rate": 4.657660081317272e-05, "loss": 0.8409, "step": 5472 }, { "epoch": 0.18, "learning_rate": 4.657525594401168e-05, "loss": 0.7626, "step": 5473 }, { "epoch": 0.18, "learning_rate": 4.6573910830163194e-05, "loss": 0.9762, "step": 5474 }, { "epoch": 0.18, "learning_rate": 4.657256547164253e-05, "loss": 0.266, "step": 5475 }, { "epoch": 0.18, "learning_rate": 4.657121986846493e-05, "loss": 0.2743, "step": 5476 }, { "epoch": 0.18, "learning_rate": 4.656987402064569e-05, "loss": 0.611, "step": 5477 }, { "epoch": 0.18, "learning_rate": 4.6568527928200036e-05, "loss": 0.9241, "step": 5478 }, { "epoch": 0.18, "learning_rate": 4.6567181591143255e-05, "loss": 0.8885, "step": 5479 }, { "epoch": 0.18, "learning_rate": 4.656583500949061e-05, "loss": 0.9631, "step": 5480 }, { "epoch": 0.18, "learning_rate": 4.656448818325738e-05, "loss": 0.8671, "step": 5481 }, { "epoch": 0.18, "learning_rate": 4.6563141112458816e-05, "loss": 0.6017, "step": 5482 }, { "epoch": 0.18, "learning_rate": 4.656179379711023e-05, "loss": 0.7654, "step": 5483 }, { "epoch": 0.18, "learning_rate": 4.656044623722687e-05, "loss": 1.0479, "step": 5484 }, { "epoch": 0.18, "learning_rate": 4.6559098432824045e-05, "loss": 1.2695, "step": 5485 }, { "epoch": 0.18, "learning_rate": 4.655775038391702e-05, "loss": 0.89, "step": 5486 }, { "epoch": 0.18, "learning_rate": 4.65564020905211e-05, "loss": 0.9433, "step": 5487 }, { "epoch": 0.18, "learning_rate": 4.655505355265156e-05, "loss": 0.9426, "step": 5488 }, { "epoch": 0.18, "learning_rate": 4.6553704770323704e-05, "loss": 0.9235, "step": 5489 }, { "epoch": 0.18, "learning_rate": 4.655235574355283e-05, "loss": 1.0318, "step": 5490 }, { "epoch": 0.18, "learning_rate": 4.655100647235423e-05, "loss": 1.1731, "step": 5491 }, { "epoch": 0.18, "learning_rate": 4.6549656956743215e-05, "loss": 0.9813, "step": 5492 }, { "epoch": 0.18, "learning_rate": 4.6548307196735086e-05, "loss": 0.8608, "step": 5493 }, { "epoch": 0.18, "learning_rate": 4.654695719234514e-05, "loss": 0.9945, "step": 5494 }, { "epoch": 0.18, "learning_rate": 4.654560694358871e-05, "loss": 0.9411, "step": 5495 }, { "epoch": 0.18, "learning_rate": 4.65442564504811e-05, "loss": 0.377, "step": 5496 }, { "epoch": 0.18, "learning_rate": 4.654290571303761e-05, "loss": 0.6723, "step": 5497 }, { "epoch": 0.18, "learning_rate": 4.654155473127357e-05, "loss": 0.3611, "step": 5498 }, { "epoch": 0.18, "learning_rate": 4.654020350520432e-05, "loss": 0.1541, "step": 5499 }, { "epoch": 0.18, "learning_rate": 4.653885203484515e-05, "loss": 0.6376, "step": 5500 }, { "epoch": 0.18, "learning_rate": 4.653750032021142e-05, "loss": 1.3291, "step": 5501 }, { "epoch": 0.18, "learning_rate": 4.653614836131844e-05, "loss": 0.602, "step": 5502 }, { "epoch": 0.18, "learning_rate": 4.653479615818155e-05, "loss": 0.5492, "step": 5503 }, { "epoch": 0.18, "learning_rate": 4.6533443710816085e-05, "loss": 0.7499, "step": 5504 }, { "epoch": 0.18, "learning_rate": 4.6532091019237375e-05, "loss": 2.1914, "step": 5505 }, { "epoch": 0.18, "learning_rate": 4.653073808346077e-05, "loss": 1.0174, "step": 5506 }, { "epoch": 0.18, "learning_rate": 4.652938490350161e-05, "loss": 0.959, "step": 5507 }, { "epoch": 0.18, "learning_rate": 4.6528031479375244e-05, "loss": 1.0887, "step": 5508 }, { "epoch": 0.18, "learning_rate": 4.6526677811097026e-05, "loss": 0.8832, "step": 5509 }, { "epoch": 0.18, "learning_rate": 4.6525323898682295e-05, "loss": 0.7422, "step": 5510 }, { "epoch": 0.18, "learning_rate": 4.6523969742146415e-05, "loss": 0.6212, "step": 5511 }, { "epoch": 0.18, "learning_rate": 4.6522615341504746e-05, "loss": 0.7562, "step": 5512 }, { "epoch": 0.18, "learning_rate": 4.652126069677264e-05, "loss": 1.1634, "step": 5513 }, { "epoch": 0.18, "learning_rate": 4.651990580796547e-05, "loss": 0.9319, "step": 5514 }, { "epoch": 0.18, "learning_rate": 4.65185506750986e-05, "loss": 0.8749, "step": 5515 }, { "epoch": 0.18, "learning_rate": 4.651719529818738e-05, "loss": 0.8666, "step": 5516 }, { "epoch": 0.18, "learning_rate": 4.6515839677247205e-05, "loss": 0.6103, "step": 5517 }, { "epoch": 0.18, "learning_rate": 4.6514483812293446e-05, "loss": 0.8493, "step": 5518 }, { "epoch": 0.18, "learning_rate": 4.651312770334147e-05, "loss": 1.0113, "step": 5519 }, { "epoch": 0.18, "learning_rate": 4.6511771350406666e-05, "loss": 0.4301, "step": 5520 }, { "epoch": 0.18, "learning_rate": 4.651041475350441e-05, "loss": 0.7443, "step": 5521 }, { "epoch": 0.18, "learning_rate": 4.650905791265009e-05, "loss": 0.7946, "step": 5522 }, { "epoch": 0.18, "learning_rate": 4.650770082785909e-05, "loss": 0.6833, "step": 5523 }, { "epoch": 0.18, "learning_rate": 4.65063434991468e-05, "loss": 0.7656, "step": 5524 }, { "epoch": 0.18, "learning_rate": 4.650498592652863e-05, "loss": 0.9059, "step": 5525 }, { "epoch": 0.18, "learning_rate": 4.650362811001996e-05, "loss": 0.9728, "step": 5526 }, { "epoch": 0.18, "learning_rate": 4.65022700496362e-05, "loss": 0.6706, "step": 5527 }, { "epoch": 0.18, "learning_rate": 4.650091174539274e-05, "loss": 0.6499, "step": 5528 }, { "epoch": 0.18, "learning_rate": 4.6499553197304984e-05, "loss": 0.9858, "step": 5529 }, { "epoch": 0.18, "learning_rate": 4.6498194405388354e-05, "loss": 0.9456, "step": 5530 }, { "epoch": 0.18, "learning_rate": 4.649683536965825e-05, "loss": 0.8052, "step": 5531 }, { "epoch": 0.18, "learning_rate": 4.649547609013009e-05, "loss": 0.9676, "step": 5532 }, { "epoch": 0.18, "learning_rate": 4.6494116566819286e-05, "loss": 0.8318, "step": 5533 }, { "epoch": 0.18, "learning_rate": 4.649275679974126e-05, "loss": 0.811, "step": 5534 }, { "epoch": 0.18, "learning_rate": 4.6491396788911426e-05, "loss": 0.9166, "step": 5535 }, { "epoch": 0.18, "learning_rate": 4.649003653434521e-05, "loss": 0.8032, "step": 5536 }, { "epoch": 0.18, "learning_rate": 4.6488676036058045e-05, "loss": 0.9761, "step": 5537 }, { "epoch": 0.18, "learning_rate": 4.6487315294065356e-05, "loss": 1.0145, "step": 5538 }, { "epoch": 0.18, "learning_rate": 4.6485954308382574e-05, "loss": 1.0199, "step": 5539 }, { "epoch": 0.18, "learning_rate": 4.648459307902514e-05, "loss": 0.8873, "step": 5540 }, { "epoch": 0.18, "learning_rate": 4.648323160600849e-05, "loss": 0.9159, "step": 5541 }, { "epoch": 0.18, "learning_rate": 4.648186988934806e-05, "loss": 0.4796, "step": 5542 }, { "epoch": 0.18, "learning_rate": 4.648050792905929e-05, "loss": 0.9409, "step": 5543 }, { "epoch": 0.18, "learning_rate": 4.647914572515765e-05, "loss": 0.6997, "step": 5544 }, { "epoch": 0.18, "learning_rate": 4.6477783277658556e-05, "loss": 0.3657, "step": 5545 }, { "epoch": 0.18, "learning_rate": 4.647642058657748e-05, "loss": 0.8699, "step": 5546 }, { "epoch": 0.18, "learning_rate": 4.6475057651929866e-05, "loss": 1.6999, "step": 5547 }, { "epoch": 0.18, "learning_rate": 4.647369447373118e-05, "loss": 1.4627, "step": 5548 }, { "epoch": 0.18, "learning_rate": 4.647233105199689e-05, "loss": 2.1009, "step": 5549 }, { "epoch": 0.18, "learning_rate": 4.647096738674243e-05, "loss": 0.8207, "step": 5550 }, { "epoch": 0.19, "learning_rate": 4.6469603477983296e-05, "loss": 1.7799, "step": 5551 }, { "epoch": 0.19, "learning_rate": 4.646823932573493e-05, "loss": 0.6043, "step": 5552 }, { "epoch": 0.19, "learning_rate": 4.6466874930012824e-05, "loss": 1.1431, "step": 5553 }, { "epoch": 0.19, "learning_rate": 4.646551029083245e-05, "loss": 0.863, "step": 5554 }, { "epoch": 0.19, "learning_rate": 4.646414540820927e-05, "loss": 0.4363, "step": 5555 }, { "epoch": 0.19, "learning_rate": 4.646278028215877e-05, "loss": 0.887, "step": 5556 }, { "epoch": 0.19, "learning_rate": 4.6461414912696446e-05, "loss": 1.2478, "step": 5557 }, { "epoch": 0.19, "learning_rate": 4.646004929983776e-05, "loss": 0.3663, "step": 5558 }, { "epoch": 0.19, "learning_rate": 4.6458683443598205e-05, "loss": 0.1714, "step": 5559 }, { "epoch": 0.19, "learning_rate": 4.6457317343993287e-05, "loss": 0.1535, "step": 5560 }, { "epoch": 0.19, "learning_rate": 4.6455951001038486e-05, "loss": 0.9322, "step": 5561 }, { "epoch": 0.19, "learning_rate": 4.64545844147493e-05, "loss": 1.1264, "step": 5562 }, { "epoch": 0.19, "learning_rate": 4.6453217585141227e-05, "loss": 0.9163, "step": 5563 }, { "epoch": 0.19, "learning_rate": 4.645185051222977e-05, "loss": 0.8231, "step": 5564 }, { "epoch": 0.19, "learning_rate": 4.6450483196030426e-05, "loss": 0.9913, "step": 5565 }, { "epoch": 0.19, "learning_rate": 4.6449115636558714e-05, "loss": 0.9132, "step": 5566 }, { "epoch": 0.19, "learning_rate": 4.6447747833830144e-05, "loss": 0.6501, "step": 5567 }, { "epoch": 0.19, "learning_rate": 4.6446379787860214e-05, "loss": 0.9756, "step": 5568 }, { "epoch": 0.19, "learning_rate": 4.644501149866445e-05, "loss": 0.952, "step": 5569 }, { "epoch": 0.19, "learning_rate": 4.644364296625837e-05, "loss": 0.6226, "step": 5570 }, { "epoch": 0.19, "learning_rate": 4.6442274190657484e-05, "loss": 0.7829, "step": 5571 }, { "epoch": 0.19, "learning_rate": 4.6440905171877335e-05, "loss": 0.9752, "step": 5572 }, { "epoch": 0.19, "learning_rate": 4.643953590993343e-05, "loss": 0.8475, "step": 5573 }, { "epoch": 0.19, "learning_rate": 4.643816640484131e-05, "loss": 1.6232, "step": 5574 }, { "epoch": 0.19, "learning_rate": 4.6436796656616496e-05, "loss": 1.1609, "step": 5575 }, { "epoch": 0.19, "learning_rate": 4.643542666527454e-05, "loss": 1.0161, "step": 5576 }, { "epoch": 0.19, "learning_rate": 4.6434056430830965e-05, "loss": 0.8878, "step": 5577 }, { "epoch": 0.19, "learning_rate": 4.64326859533013e-05, "loss": 0.4901, "step": 5578 }, { "epoch": 0.19, "learning_rate": 4.643131523270112e-05, "loss": 0.0769, "step": 5579 }, { "epoch": 0.19, "learning_rate": 4.642994426904594e-05, "loss": 0.2687, "step": 5580 }, { "epoch": 0.19, "learning_rate": 4.642857306235133e-05, "loss": 0.8778, "step": 5581 }, { "epoch": 0.19, "learning_rate": 4.642720161263283e-05, "loss": 0.8095, "step": 5582 }, { "epoch": 0.19, "learning_rate": 4.6425829919905995e-05, "loss": 0.9642, "step": 5583 }, { "epoch": 0.19, "learning_rate": 4.642445798418638e-05, "loss": 0.8983, "step": 5584 }, { "epoch": 0.19, "learning_rate": 4.6423085805489544e-05, "loss": 0.6443, "step": 5585 }, { "epoch": 0.19, "learning_rate": 4.6421713383831054e-05, "loss": 0.1142, "step": 5586 }, { "epoch": 0.19, "learning_rate": 4.642034071922647e-05, "loss": 0.0657, "step": 5587 }, { "epoch": 0.19, "learning_rate": 4.641896781169136e-05, "loss": 0.0335, "step": 5588 }, { "epoch": 0.19, "learning_rate": 4.64175946612413e-05, "loss": 0.6192, "step": 5589 }, { "epoch": 0.19, "learning_rate": 4.641622126789186e-05, "loss": 1.6275, "step": 5590 }, { "epoch": 0.19, "learning_rate": 4.6414847631658615e-05, "loss": 0.828, "step": 5591 }, { "epoch": 0.19, "learning_rate": 4.641347375255714e-05, "loss": 0.9364, "step": 5592 }, { "epoch": 0.19, "learning_rate": 4.6412099630603025e-05, "loss": 0.8395, "step": 5593 }, { "epoch": 0.19, "learning_rate": 4.6410725265811835e-05, "loss": 0.9909, "step": 5594 }, { "epoch": 0.19, "learning_rate": 4.640935065819919e-05, "loss": 0.9999, "step": 5595 }, { "epoch": 0.19, "learning_rate": 4.6407975807780646e-05, "loss": 0.8444, "step": 5596 }, { "epoch": 0.19, "learning_rate": 4.640660071457181e-05, "loss": 0.6519, "step": 5597 }, { "epoch": 0.19, "learning_rate": 4.640522537858829e-05, "loss": 0.8011, "step": 5598 }, { "epoch": 0.19, "learning_rate": 4.6403849799845655e-05, "loss": 0.7384, "step": 5599 }, { "epoch": 0.19, "learning_rate": 4.640247397835953e-05, "loss": 0.8017, "step": 5600 }, { "epoch": 0.19, "learning_rate": 4.6401097914145506e-05, "loss": 0.9291, "step": 5601 }, { "epoch": 0.19, "learning_rate": 4.639972160721919e-05, "loss": 0.8251, "step": 5602 }, { "epoch": 0.19, "learning_rate": 4.63983450575962e-05, "loss": 0.9401, "step": 5603 }, { "epoch": 0.19, "learning_rate": 4.639696826529214e-05, "loss": 0.7902, "step": 5604 }, { "epoch": 0.19, "learning_rate": 4.639559123032262e-05, "loss": 0.9672, "step": 5605 }, { "epoch": 0.19, "learning_rate": 4.639421395270326e-05, "loss": 1.0756, "step": 5606 }, { "epoch": 0.19, "learning_rate": 4.6392836432449696e-05, "loss": 0.8102, "step": 5607 }, { "epoch": 0.19, "learning_rate": 4.639145866957753e-05, "loss": 0.5618, "step": 5608 }, { "epoch": 0.19, "learning_rate": 4.6390080664102386e-05, "loss": 0.5908, "step": 5609 }, { "epoch": 0.19, "learning_rate": 4.638870241603991e-05, "loss": 0.9434, "step": 5610 }, { "epoch": 0.19, "learning_rate": 4.638732392540572e-05, "loss": 1.4987, "step": 5611 }, { "epoch": 0.19, "learning_rate": 4.6385945192215454e-05, "loss": 0.9474, "step": 5612 }, { "epoch": 0.19, "learning_rate": 4.6384566216484745e-05, "loss": 0.6852, "step": 5613 }, { "epoch": 0.19, "learning_rate": 4.638318699822923e-05, "loss": 0.8833, "step": 5614 }, { "epoch": 0.19, "learning_rate": 4.6381807537464566e-05, "loss": 0.7692, "step": 5615 }, { "epoch": 0.19, "learning_rate": 4.638042783420638e-05, "loss": 0.7657, "step": 5616 }, { "epoch": 0.19, "learning_rate": 4.637904788847033e-05, "loss": 0.6252, "step": 5617 }, { "epoch": 0.19, "learning_rate": 4.637766770027205e-05, "loss": 0.9795, "step": 5618 }, { "epoch": 0.19, "learning_rate": 4.637628726962722e-05, "loss": 0.9207, "step": 5619 }, { "epoch": 0.19, "learning_rate": 4.637490659655148e-05, "loss": 1.0507, "step": 5620 }, { "epoch": 0.19, "learning_rate": 4.637352568106048e-05, "loss": 0.8996, "step": 5621 }, { "epoch": 0.19, "learning_rate": 4.63721445231699e-05, "loss": 1.0413, "step": 5622 }, { "epoch": 0.19, "learning_rate": 4.6370763122895386e-05, "loss": 0.444, "step": 5623 }, { "epoch": 0.19, "learning_rate": 4.6369381480252616e-05, "loss": 0.06, "step": 5624 }, { "epoch": 0.19, "learning_rate": 4.636799959525726e-05, "loss": 0.7096, "step": 5625 }, { "epoch": 0.19, "learning_rate": 4.6366617467924985e-05, "loss": 0.7975, "step": 5626 }, { "epoch": 0.19, "learning_rate": 4.6365235098271465e-05, "loss": 0.8212, "step": 5627 }, { "epoch": 0.19, "learning_rate": 4.636385248631237e-05, "loss": 0.8399, "step": 5628 }, { "epoch": 0.19, "learning_rate": 4.636246963206341e-05, "loss": 0.8491, "step": 5629 }, { "epoch": 0.19, "learning_rate": 4.6361086535540234e-05, "loss": 0.7079, "step": 5630 }, { "epoch": 0.19, "learning_rate": 4.635970319675855e-05, "loss": 0.9005, "step": 5631 }, { "epoch": 0.19, "learning_rate": 4.635831961573404e-05, "loss": 0.8539, "step": 5632 }, { "epoch": 0.19, "learning_rate": 4.635693579248238e-05, "loss": 0.911, "step": 5633 }, { "epoch": 0.19, "learning_rate": 4.635555172701929e-05, "loss": 0.8841, "step": 5634 }, { "epoch": 0.19, "learning_rate": 4.635416741936045e-05, "loss": 0.6964, "step": 5635 }, { "epoch": 0.19, "learning_rate": 4.635278286952157e-05, "loss": 0.9926, "step": 5636 }, { "epoch": 0.19, "learning_rate": 4.635139807751834e-05, "loss": 0.7962, "step": 5637 }, { "epoch": 0.19, "learning_rate": 4.635001304336648e-05, "loss": 0.8685, "step": 5638 }, { "epoch": 0.19, "learning_rate": 4.634862776708168e-05, "loss": 0.8867, "step": 5639 }, { "epoch": 0.19, "learning_rate": 4.634724224867967e-05, "loss": 0.9325, "step": 5640 }, { "epoch": 0.19, "learning_rate": 4.634585648817615e-05, "loss": 1.0279, "step": 5641 }, { "epoch": 0.19, "learning_rate": 4.634447048558683e-05, "loss": 0.9213, "step": 5642 }, { "epoch": 0.19, "learning_rate": 4.634308424092745e-05, "loss": 0.73, "step": 5643 }, { "epoch": 0.19, "learning_rate": 4.634169775421372e-05, "loss": 0.8433, "step": 5644 }, { "epoch": 0.19, "learning_rate": 4.634031102546136e-05, "loss": 1.0976, "step": 5645 }, { "epoch": 0.19, "learning_rate": 4.6338924054686104e-05, "loss": 0.9669, "step": 5646 }, { "epoch": 0.19, "learning_rate": 4.633753684190368e-05, "loss": 0.6828, "step": 5647 }, { "epoch": 0.19, "learning_rate": 4.633614938712982e-05, "loss": 1.0305, "step": 5648 }, { "epoch": 0.19, "learning_rate": 4.633476169038026e-05, "loss": 0.9653, "step": 5649 }, { "epoch": 0.19, "learning_rate": 4.633337375167073e-05, "loss": 0.6753, "step": 5650 }, { "epoch": 0.19, "learning_rate": 4.633198557101699e-05, "loss": 1.035, "step": 5651 }, { "epoch": 0.19, "learning_rate": 4.6330597148434764e-05, "loss": 0.7865, "step": 5652 }, { "epoch": 0.19, "learning_rate": 4.63292084839398e-05, "loss": 0.6505, "step": 5653 }, { "epoch": 0.19, "learning_rate": 4.632781957754786e-05, "loss": 0.758, "step": 5654 }, { "epoch": 0.19, "learning_rate": 4.632643042927469e-05, "loss": 0.8912, "step": 5655 }, { "epoch": 0.19, "learning_rate": 4.632504103913604e-05, "loss": 0.6271, "step": 5656 }, { "epoch": 0.19, "learning_rate": 4.6323651407147675e-05, "loss": 0.8197, "step": 5657 }, { "epoch": 0.19, "learning_rate": 4.632226153332534e-05, "loss": 1.2257, "step": 5658 }, { "epoch": 0.19, "learning_rate": 4.632087141768482e-05, "loss": 0.9932, "step": 5659 }, { "epoch": 0.19, "learning_rate": 4.6319481060241855e-05, "loss": 0.9564, "step": 5660 }, { "epoch": 0.19, "learning_rate": 4.631809046101223e-05, "loss": 0.864, "step": 5661 }, { "epoch": 0.19, "learning_rate": 4.631669962001172e-05, "loss": 0.9261, "step": 5662 }, { "epoch": 0.19, "learning_rate": 4.631530853725609e-05, "loss": 0.7923, "step": 5663 }, { "epoch": 0.19, "learning_rate": 4.6313917212761114e-05, "loss": 0.7127, "step": 5664 }, { "epoch": 0.19, "learning_rate": 4.631252564654257e-05, "loss": 0.9247, "step": 5665 }, { "epoch": 0.19, "learning_rate": 4.6311133838616247e-05, "loss": 1.0216, "step": 5666 }, { "epoch": 0.19, "learning_rate": 4.630974178899794e-05, "loss": 0.9636, "step": 5667 }, { "epoch": 0.19, "learning_rate": 4.6308349497703406e-05, "loss": 0.8901, "step": 5668 }, { "epoch": 0.19, "learning_rate": 4.6306956964748465e-05, "loss": 0.7589, "step": 5669 }, { "epoch": 0.19, "learning_rate": 4.6305564190148896e-05, "loss": 2.3672, "step": 5670 }, { "epoch": 0.19, "learning_rate": 4.630417117392049e-05, "loss": 5.7756, "step": 5671 }, { "epoch": 0.19, "learning_rate": 4.630277791607905e-05, "loss": 5.5246, "step": 5672 }, { "epoch": 0.19, "learning_rate": 4.630138441664039e-05, "loss": 0.7682, "step": 5673 }, { "epoch": 0.19, "learning_rate": 4.62999906756203e-05, "loss": 0.5483, "step": 5674 }, { "epoch": 0.19, "learning_rate": 4.629859669303458e-05, "loss": 1.1898, "step": 5675 }, { "epoch": 0.19, "learning_rate": 4.629720246889906e-05, "loss": 0.9245, "step": 5676 }, { "epoch": 0.19, "learning_rate": 4.6295808003229527e-05, "loss": 0.8399, "step": 5677 }, { "epoch": 0.19, "learning_rate": 4.629441329604182e-05, "loss": 1.4514, "step": 5678 }, { "epoch": 0.19, "learning_rate": 4.629301834735175e-05, "loss": 0.7043, "step": 5679 }, { "epoch": 0.19, "learning_rate": 4.629162315717512e-05, "loss": 0.9942, "step": 5680 }, { "epoch": 0.19, "learning_rate": 4.629022772552777e-05, "loss": 1.1663, "step": 5681 }, { "epoch": 0.19, "learning_rate": 4.628883205242553e-05, "loss": 0.8475, "step": 5682 }, { "epoch": 0.19, "learning_rate": 4.628743613788421e-05, "loss": 1.0663, "step": 5683 }, { "epoch": 0.19, "learning_rate": 4.628603998191966e-05, "loss": 0.7177, "step": 5684 }, { "epoch": 0.19, "learning_rate": 4.62846435845477e-05, "loss": 0.8883, "step": 5685 }, { "epoch": 0.19, "learning_rate": 4.628324694578418e-05, "loss": 0.8143, "step": 5686 }, { "epoch": 0.19, "learning_rate": 4.628185006564492e-05, "loss": 0.7731, "step": 5687 }, { "epoch": 0.19, "learning_rate": 4.6280452944145783e-05, "loss": 0.8789, "step": 5688 }, { "epoch": 0.19, "learning_rate": 4.62790555813026e-05, "loss": 0.837, "step": 5689 }, { "epoch": 0.19, "learning_rate": 4.6277657977131226e-05, "loss": 0.924, "step": 5690 }, { "epoch": 0.19, "learning_rate": 4.627626013164751e-05, "loss": 0.9686, "step": 5691 }, { "epoch": 0.19, "learning_rate": 4.6274862044867304e-05, "loss": 0.7701, "step": 5692 }, { "epoch": 0.19, "learning_rate": 4.627346371680645e-05, "loss": 0.9108, "step": 5693 }, { "epoch": 0.19, "learning_rate": 4.6272065147480835e-05, "loss": 0.629, "step": 5694 }, { "epoch": 0.19, "learning_rate": 4.627066633690631e-05, "loss": 0.4302, "step": 5695 }, { "epoch": 0.19, "learning_rate": 4.6269267285098724e-05, "loss": 0.7284, "step": 5696 }, { "epoch": 0.19, "learning_rate": 4.626786799207396e-05, "loss": 0.7031, "step": 5697 }, { "epoch": 0.19, "learning_rate": 4.626646845784788e-05, "loss": 0.7477, "step": 5698 }, { "epoch": 0.19, "learning_rate": 4.6265068682436364e-05, "loss": 0.3237, "step": 5699 }, { "epoch": 0.19, "learning_rate": 4.626366866585528e-05, "loss": 0.9207, "step": 5700 }, { "epoch": 0.19, "learning_rate": 4.6262268408120496e-05, "loss": 0.8954, "step": 5701 }, { "epoch": 0.19, "learning_rate": 4.626086790924791e-05, "loss": 1.2078, "step": 5702 }, { "epoch": 0.19, "learning_rate": 4.62594671692534e-05, "loss": 0.8797, "step": 5703 }, { "epoch": 0.19, "learning_rate": 4.6258066188152855e-05, "loss": 0.6552, "step": 5704 }, { "epoch": 0.19, "learning_rate": 4.625666496596216e-05, "loss": 0.522, "step": 5705 }, { "epoch": 0.19, "learning_rate": 4.62552635026972e-05, "loss": 0.7501, "step": 5706 }, { "epoch": 0.19, "learning_rate": 4.6253861798373874e-05, "loss": 0.9168, "step": 5707 }, { "epoch": 0.19, "learning_rate": 4.6252459853008086e-05, "loss": 0.9371, "step": 5708 }, { "epoch": 0.19, "learning_rate": 4.6251057666615726e-05, "loss": 0.9656, "step": 5709 }, { "epoch": 0.19, "learning_rate": 4.62496552392127e-05, "loss": 0.7356, "step": 5710 }, { "epoch": 0.19, "learning_rate": 4.624825257081491e-05, "loss": 0.6895, "step": 5711 }, { "epoch": 0.19, "learning_rate": 4.624684966143828e-05, "loss": 0.8388, "step": 5712 }, { "epoch": 0.19, "learning_rate": 4.6245446511098686e-05, "loss": 0.8755, "step": 5713 }, { "epoch": 0.19, "learning_rate": 4.624404311981208e-05, "loss": 0.8988, "step": 5714 }, { "epoch": 0.19, "learning_rate": 4.624263948759436e-05, "loss": 0.989, "step": 5715 }, { "epoch": 0.19, "learning_rate": 4.624123561446144e-05, "loss": 0.8578, "step": 5716 }, { "epoch": 0.19, "learning_rate": 4.623983150042925e-05, "loss": 0.964, "step": 5717 }, { "epoch": 0.19, "learning_rate": 4.623842714551371e-05, "loss": 0.5156, "step": 5718 }, { "epoch": 0.19, "learning_rate": 4.6237022549730744e-05, "loss": 0.8554, "step": 5719 }, { "epoch": 0.19, "learning_rate": 4.623561771309629e-05, "loss": 1.0684, "step": 5720 }, { "epoch": 0.19, "learning_rate": 4.623421263562628e-05, "loss": 0.4654, "step": 5721 }, { "epoch": 0.19, "learning_rate": 4.623280731733664e-05, "loss": 0.5907, "step": 5722 }, { "epoch": 0.19, "learning_rate": 4.623140175824332e-05, "loss": 0.8717, "step": 5723 }, { "epoch": 0.19, "learning_rate": 4.6229995958362244e-05, "loss": 0.6934, "step": 5724 }, { "epoch": 0.19, "learning_rate": 4.622858991770937e-05, "loss": 1.1478, "step": 5725 }, { "epoch": 0.19, "learning_rate": 4.622718363630064e-05, "loss": 0.9248, "step": 5726 }, { "epoch": 0.19, "learning_rate": 4.6225777114152005e-05, "loss": 2.1893, "step": 5727 }, { "epoch": 0.19, "learning_rate": 4.6224370351279405e-05, "loss": 2.115, "step": 5728 }, { "epoch": 0.19, "learning_rate": 4.622296334769881e-05, "loss": 2.1117, "step": 5729 }, { "epoch": 0.19, "learning_rate": 4.622155610342617e-05, "loss": 1.4903, "step": 5730 }, { "epoch": 0.19, "learning_rate": 4.6220148618477443e-05, "loss": 1.3949, "step": 5731 }, { "epoch": 0.19, "learning_rate": 4.621874089286859e-05, "loss": 0.8422, "step": 5732 }, { "epoch": 0.19, "learning_rate": 4.621733292661558e-05, "loss": 0.5325, "step": 5733 }, { "epoch": 0.19, "learning_rate": 4.621592471973438e-05, "loss": 0.7336, "step": 5734 }, { "epoch": 0.19, "learning_rate": 4.621451627224096e-05, "loss": 0.761, "step": 5735 }, { "epoch": 0.19, "learning_rate": 4.6213107584151304e-05, "loss": 0.7327, "step": 5736 }, { "epoch": 0.19, "learning_rate": 4.621169865548137e-05, "loss": 0.9263, "step": 5737 }, { "epoch": 0.19, "learning_rate": 4.621028948624714e-05, "loss": 0.8873, "step": 5738 }, { "epoch": 0.19, "learning_rate": 4.6208880076464615e-05, "loss": 0.8818, "step": 5739 }, { "epoch": 0.19, "learning_rate": 4.620747042614975e-05, "loss": 0.7152, "step": 5740 }, { "epoch": 0.19, "learning_rate": 4.620606053531855e-05, "loss": 0.8977, "step": 5741 }, { "epoch": 0.19, "learning_rate": 4.620465040398701e-05, "loss": 1.1962, "step": 5742 }, { "epoch": 0.19, "learning_rate": 4.620324003217111e-05, "loss": 0.9659, "step": 5743 }, { "epoch": 0.19, "learning_rate": 4.620182941988685e-05, "loss": 0.8286, "step": 5744 }, { "epoch": 0.19, "learning_rate": 4.620041856715023e-05, "loss": 0.7641, "step": 5745 }, { "epoch": 0.19, "learning_rate": 4.6199007473977244e-05, "loss": 0.7644, "step": 5746 }, { "epoch": 0.19, "learning_rate": 4.6197596140383895e-05, "loss": 0.8573, "step": 5747 }, { "epoch": 0.19, "learning_rate": 4.619618456638619e-05, "loss": 0.5719, "step": 5748 }, { "epoch": 0.19, "learning_rate": 4.619477275200015e-05, "loss": 0.5805, "step": 5749 }, { "epoch": 0.19, "learning_rate": 4.619336069724177e-05, "loss": 0.5468, "step": 5750 }, { "epoch": 0.19, "learning_rate": 4.619194840212708e-05, "loss": 0.7783, "step": 5751 }, { "epoch": 0.19, "learning_rate": 4.619053586667209e-05, "loss": 0.7894, "step": 5752 }, { "epoch": 0.19, "learning_rate": 4.618912309089282e-05, "loss": 0.282, "step": 5753 }, { "epoch": 0.19, "learning_rate": 4.618771007480528e-05, "loss": 0.2331, "step": 5754 }, { "epoch": 0.19, "learning_rate": 4.618629681842551e-05, "loss": 0.2096, "step": 5755 }, { "epoch": 0.19, "learning_rate": 4.618488332176954e-05, "loss": 0.5414, "step": 5756 }, { "epoch": 0.19, "learning_rate": 4.618346958485339e-05, "loss": 1.0321, "step": 5757 }, { "epoch": 0.19, "learning_rate": 4.61820556076931e-05, "loss": 1.0425, "step": 5758 }, { "epoch": 0.19, "learning_rate": 4.61806413903047e-05, "loss": 0.9593, "step": 5759 }, { "epoch": 0.19, "learning_rate": 4.617922693270424e-05, "loss": 0.7879, "step": 5760 }, { "epoch": 0.19, "learning_rate": 4.617781223490775e-05, "loss": 0.9857, "step": 5761 }, { "epoch": 0.19, "learning_rate": 4.617639729693128e-05, "loss": 0.9841, "step": 5762 }, { "epoch": 0.19, "learning_rate": 4.617498211879088e-05, "loss": 0.961, "step": 5763 }, { "epoch": 0.19, "learning_rate": 4.617356670050259e-05, "loss": 1.0935, "step": 5764 }, { "epoch": 0.19, "learning_rate": 4.6172151042082465e-05, "loss": 0.9253, "step": 5765 }, { "epoch": 0.19, "learning_rate": 4.617073514354657e-05, "loss": 0.9041, "step": 5766 }, { "epoch": 0.19, "learning_rate": 4.616931900491095e-05, "loss": 0.7991, "step": 5767 }, { "epoch": 0.19, "learning_rate": 4.616790262619167e-05, "loss": 0.8734, "step": 5768 }, { "epoch": 0.19, "learning_rate": 4.616648600740481e-05, "loss": 0.7419, "step": 5769 }, { "epoch": 0.19, "learning_rate": 4.6165069148566406e-05, "loss": 2.9077, "step": 5770 }, { "epoch": 0.19, "learning_rate": 4.6163652049692545e-05, "loss": 0.7767, "step": 5771 }, { "epoch": 0.19, "learning_rate": 4.6162234710799295e-05, "loss": 0.7969, "step": 5772 }, { "epoch": 0.19, "learning_rate": 4.6160817131902735e-05, "loss": 0.5764, "step": 5773 }, { "epoch": 0.19, "learning_rate": 4.615939931301893e-05, "loss": 0.0865, "step": 5774 }, { "epoch": 0.19, "learning_rate": 4.6157981254163975e-05, "loss": 0.0786, "step": 5775 }, { "epoch": 0.19, "learning_rate": 4.6156562955353933e-05, "loss": 0.7533, "step": 5776 }, { "epoch": 0.19, "learning_rate": 4.615514441660491e-05, "loss": 1.0964, "step": 5777 }, { "epoch": 0.19, "learning_rate": 4.615372563793298e-05, "loss": 1.528, "step": 5778 }, { "epoch": 0.19, "learning_rate": 4.6152306619354237e-05, "loss": 1.4905, "step": 5779 }, { "epoch": 0.19, "learning_rate": 4.6150887360884774e-05, "loss": 1.6448, "step": 5780 }, { "epoch": 0.19, "learning_rate": 4.6149467862540696e-05, "loss": 1.1482, "step": 5781 }, { "epoch": 0.19, "learning_rate": 4.6148048124338085e-05, "loss": 0.9962, "step": 5782 }, { "epoch": 0.19, "learning_rate": 4.614662814629305e-05, "loss": 1.0814, "step": 5783 }, { "epoch": 0.19, "learning_rate": 4.61452079284217e-05, "loss": 1.2226, "step": 5784 }, { "epoch": 0.19, "learning_rate": 4.614378747074014e-05, "loss": 0.9822, "step": 5785 }, { "epoch": 0.19, "learning_rate": 4.6142366773264476e-05, "loss": 0.6166, "step": 5786 }, { "epoch": 0.19, "learning_rate": 4.614094583601082e-05, "loss": 1.0497, "step": 5787 }, { "epoch": 0.19, "learning_rate": 4.613952465899529e-05, "loss": 0.4663, "step": 5788 }, { "epoch": 0.19, "learning_rate": 4.613810324223401e-05, "loss": 0.4409, "step": 5789 }, { "epoch": 0.19, "learning_rate": 4.613668158574308e-05, "loss": 1.0503, "step": 5790 }, { "epoch": 0.19, "learning_rate": 4.613525968953864e-05, "loss": 0.704, "step": 5791 }, { "epoch": 0.19, "learning_rate": 4.613383755363682e-05, "loss": 0.9243, "step": 5792 }, { "epoch": 0.19, "learning_rate": 4.6132415178053725e-05, "loss": 0.8831, "step": 5793 }, { "epoch": 0.19, "learning_rate": 4.613099256280551e-05, "loss": 0.8465, "step": 5794 }, { "epoch": 0.19, "learning_rate": 4.61295697079083e-05, "loss": 1.0744, "step": 5795 }, { "epoch": 0.19, "learning_rate": 4.612814661337824e-05, "loss": 0.6557, "step": 5796 }, { "epoch": 0.19, "learning_rate": 4.612672327923146e-05, "loss": 0.7798, "step": 5797 }, { "epoch": 0.19, "learning_rate": 4.612529970548411e-05, "loss": 0.8499, "step": 5798 }, { "epoch": 0.19, "learning_rate": 4.612387589215231e-05, "loss": 1.1572, "step": 5799 }, { "epoch": 0.19, "learning_rate": 4.612245183925225e-05, "loss": 0.8009, "step": 5800 }, { "epoch": 0.19, "learning_rate": 4.612102754680004e-05, "loss": 0.9841, "step": 5801 }, { "epoch": 0.19, "learning_rate": 4.6119603014811855e-05, "loss": 0.8626, "step": 5802 }, { "epoch": 0.19, "learning_rate": 4.6118178243303846e-05, "loss": 1.0292, "step": 5803 }, { "epoch": 0.19, "learning_rate": 4.611675323229217e-05, "loss": 0.7678, "step": 5804 }, { "epoch": 0.19, "learning_rate": 4.611532798179299e-05, "loss": 1.4696, "step": 5805 }, { "epoch": 0.19, "learning_rate": 4.611390249182248e-05, "loss": 1.8504, "step": 5806 }, { "epoch": 0.19, "learning_rate": 4.611247676239678e-05, "loss": 1.2858, "step": 5807 }, { "epoch": 0.19, "learning_rate": 4.611105079353209e-05, "loss": 1.6257, "step": 5808 }, { "epoch": 0.19, "learning_rate": 4.610962458524456e-05, "loss": 1.7188, "step": 5809 }, { "epoch": 0.19, "learning_rate": 4.610819813755038e-05, "loss": 1.087, "step": 5810 }, { "epoch": 0.19, "learning_rate": 4.610677145046571e-05, "loss": 0.9387, "step": 5811 }, { "epoch": 0.19, "learning_rate": 4.6105344524006756e-05, "loss": 0.4019, "step": 5812 }, { "epoch": 0.19, "learning_rate": 4.610391735818967e-05, "loss": 0.6273, "step": 5813 }, { "epoch": 0.19, "learning_rate": 4.610248995303066e-05, "loss": 0.5698, "step": 5814 }, { "epoch": 0.19, "learning_rate": 4.610106230854591e-05, "loss": 0.8331, "step": 5815 }, { "epoch": 0.19, "learning_rate": 4.6099634424751605e-05, "loss": 0.6829, "step": 5816 }, { "epoch": 0.19, "learning_rate": 4.6098206301663935e-05, "loss": 0.5723, "step": 5817 }, { "epoch": 0.19, "learning_rate": 4.6096777939299114e-05, "loss": 0.0903, "step": 5818 }, { "epoch": 0.19, "learning_rate": 4.609534933767332e-05, "loss": 0.0893, "step": 5819 }, { "epoch": 0.19, "learning_rate": 4.609392049680278e-05, "loss": 1.6696, "step": 5820 }, { "epoch": 0.19, "learning_rate": 4.6092491416703674e-05, "loss": 0.9312, "step": 5821 }, { "epoch": 0.19, "learning_rate": 4.609106209739222e-05, "loss": 0.5997, "step": 5822 }, { "epoch": 0.19, "learning_rate": 4.608963253888464e-05, "loss": 0.8083, "step": 5823 }, { "epoch": 0.19, "learning_rate": 4.6088202741197125e-05, "loss": 0.9865, "step": 5824 }, { "epoch": 0.19, "learning_rate": 4.60867727043459e-05, "loss": 1.0307, "step": 5825 }, { "epoch": 0.19, "learning_rate": 4.608534242834719e-05, "loss": 1.2653, "step": 5826 }, { "epoch": 0.19, "learning_rate": 4.6083911913217206e-05, "loss": 0.7176, "step": 5827 }, { "epoch": 0.19, "learning_rate": 4.6082481158972166e-05, "loss": 0.2546, "step": 5828 }, { "epoch": 0.19, "learning_rate": 4.608105016562831e-05, "loss": 0.4757, "step": 5829 }, { "epoch": 0.19, "learning_rate": 4.6079618933201874e-05, "loss": 0.489, "step": 5830 }, { "epoch": 0.19, "learning_rate": 4.607818746170907e-05, "loss": 0.5327, "step": 5831 }, { "epoch": 0.19, "learning_rate": 4.6076755751166134e-05, "loss": 0.5833, "step": 5832 }, { "epoch": 0.19, "learning_rate": 4.6075323801589323e-05, "loss": 1.2317, "step": 5833 }, { "epoch": 0.19, "learning_rate": 4.607389161299486e-05, "loss": 1.5005, "step": 5834 }, { "epoch": 0.19, "learning_rate": 4.607245918539898e-05, "loss": 0.9974, "step": 5835 }, { "epoch": 0.19, "learning_rate": 4.607102651881795e-05, "loss": 0.8054, "step": 5836 }, { "epoch": 0.19, "learning_rate": 4.6069593613268006e-05, "loss": 0.7909, "step": 5837 }, { "epoch": 0.19, "learning_rate": 4.60681604687654e-05, "loss": 0.9162, "step": 5838 }, { "epoch": 0.19, "learning_rate": 4.606672708532639e-05, "loss": 0.8564, "step": 5839 }, { "epoch": 0.19, "learning_rate": 4.6065293462967225e-05, "loss": 0.7759, "step": 5840 }, { "epoch": 0.19, "learning_rate": 4.6063859601704165e-05, "loss": 0.9804, "step": 5841 }, { "epoch": 0.19, "learning_rate": 4.606242550155347e-05, "loss": 0.6164, "step": 5842 }, { "epoch": 0.19, "learning_rate": 4.606099116253142e-05, "loss": 1.0391, "step": 5843 }, { "epoch": 0.19, "learning_rate": 4.605955658465426e-05, "loss": 0.8667, "step": 5844 }, { "epoch": 0.19, "learning_rate": 4.6058121767938274e-05, "loss": 1.4868, "step": 5845 }, { "epoch": 0.19, "learning_rate": 4.6056686712399734e-05, "loss": 2.2015, "step": 5846 }, { "epoch": 0.19, "learning_rate": 4.60552514180549e-05, "loss": 1.6103, "step": 5847 }, { "epoch": 0.19, "learning_rate": 4.6053815884920064e-05, "loss": 4.4375, "step": 5848 }, { "epoch": 0.19, "learning_rate": 4.605238011301151e-05, "loss": 5.8342, "step": 5849 }, { "epoch": 0.2, "learning_rate": 4.605094410234551e-05, "loss": 5.6482, "step": 5850 }, { "epoch": 0.2, "learning_rate": 4.604950785293836e-05, "loss": 5.5962, "step": 5851 }, { "epoch": 0.2, "learning_rate": 4.6048071364806334e-05, "loss": 5.6097, "step": 5852 }, { "epoch": 0.2, "learning_rate": 4.604663463796574e-05, "loss": 2.2576, "step": 5853 }, { "epoch": 0.2, "learning_rate": 4.604519767243286e-05, "loss": 0.8813, "step": 5854 }, { "epoch": 0.2, "learning_rate": 4.6043760468224e-05, "loss": 0.4489, "step": 5855 }, { "epoch": 0.2, "learning_rate": 4.604232302535546e-05, "loss": 0.4222, "step": 5856 }, { "epoch": 0.2, "learning_rate": 4.6040885343843525e-05, "loss": 1.2924, "step": 5857 }, { "epoch": 0.2, "learning_rate": 4.6039447423704515e-05, "loss": 1.2039, "step": 5858 }, { "epoch": 0.2, "learning_rate": 4.603800926495474e-05, "loss": 0.8115, "step": 5859 }, { "epoch": 0.2, "learning_rate": 4.6036570867610505e-05, "loss": 0.8305, "step": 5860 }, { "epoch": 0.2, "learning_rate": 4.6035132231688126e-05, "loss": 1.1353, "step": 5861 }, { "epoch": 0.2, "learning_rate": 4.603369335720391e-05, "loss": 0.8322, "step": 5862 }, { "epoch": 0.2, "learning_rate": 4.603225424417418e-05, "loss": 1.0122, "step": 5863 }, { "epoch": 0.2, "learning_rate": 4.603081489261526e-05, "loss": 0.9698, "step": 5864 }, { "epoch": 0.2, "learning_rate": 4.6029375302543474e-05, "loss": 1.0143, "step": 5865 }, { "epoch": 0.2, "learning_rate": 4.602793547397515e-05, "loss": 0.9165, "step": 5866 }, { "epoch": 0.2, "learning_rate": 4.602649540692661e-05, "loss": 0.7865, "step": 5867 }, { "epoch": 0.2, "learning_rate": 4.6025055101414195e-05, "loss": 0.7963, "step": 5868 }, { "epoch": 0.2, "learning_rate": 4.602361455745423e-05, "loss": 0.857, "step": 5869 }, { "epoch": 0.2, "learning_rate": 4.602217377506306e-05, "loss": 0.6381, "step": 5870 }, { "epoch": 0.2, "learning_rate": 4.602073275425702e-05, "loss": 0.7903, "step": 5871 }, { "epoch": 0.2, "learning_rate": 4.6019291495052455e-05, "loss": 0.6311, "step": 5872 }, { "epoch": 0.2, "learning_rate": 4.601784999746571e-05, "loss": 0.8942, "step": 5873 }, { "epoch": 0.2, "learning_rate": 4.601640826151313e-05, "loss": 5.737, "step": 5874 }, { "epoch": 0.2, "learning_rate": 4.6014966287211084e-05, "loss": 4.6067, "step": 5875 }, { "epoch": 0.2, "learning_rate": 4.60135240745759e-05, "loss": 5.7495, "step": 5876 }, { "epoch": 0.2, "learning_rate": 4.601208162362395e-05, "loss": 5.737, "step": 5877 }, { "epoch": 0.2, "learning_rate": 4.6010638934371585e-05, "loss": 5.7036, "step": 5878 }, { "epoch": 0.2, "learning_rate": 4.600919600683517e-05, "loss": 5.6805, "step": 5879 }, { "epoch": 0.2, "learning_rate": 4.6007752841031074e-05, "loss": 4.3184, "step": 5880 }, { "epoch": 0.2, "learning_rate": 4.6006309436975656e-05, "loss": 0.7637, "step": 5881 }, { "epoch": 0.2, "learning_rate": 4.600486579468529e-05, "loss": 1.0299, "step": 5882 }, { "epoch": 0.2, "learning_rate": 4.6003421914176346e-05, "loss": 0.7432, "step": 5883 }, { "epoch": 0.2, "learning_rate": 4.6001977795465203e-05, "loss": 0.9516, "step": 5884 }, { "epoch": 0.2, "learning_rate": 4.600053343856824e-05, "loss": 0.6275, "step": 5885 }, { "epoch": 0.2, "learning_rate": 4.599908884350183e-05, "loss": 0.7503, "step": 5886 }, { "epoch": 0.2, "learning_rate": 4.5997644010282365e-05, "loss": 0.7802, "step": 5887 }, { "epoch": 0.2, "learning_rate": 4.599619893892623e-05, "loss": 0.96, "step": 5888 }, { "epoch": 0.2, "learning_rate": 4.599475362944981e-05, "loss": 0.8563, "step": 5889 }, { "epoch": 0.2, "learning_rate": 4.599330808186948e-05, "loss": 0.4912, "step": 5890 }, { "epoch": 0.2, "learning_rate": 4.5991862296201674e-05, "loss": 1.1408, "step": 5891 }, { "epoch": 0.2, "learning_rate": 4.5990416272462753e-05, "loss": 0.5609, "step": 5892 }, { "epoch": 0.2, "learning_rate": 4.598897001066914e-05, "loss": 1.0048, "step": 5893 }, { "epoch": 0.2, "learning_rate": 4.598752351083722e-05, "loss": 0.9431, "step": 5894 }, { "epoch": 0.2, "learning_rate": 4.5986076772983414e-05, "loss": 0.3607, "step": 5895 }, { "epoch": 0.2, "learning_rate": 4.5984629797124105e-05, "loss": 1.0269, "step": 5896 }, { "epoch": 0.2, "learning_rate": 4.598318258327573e-05, "loss": 0.7742, "step": 5897 }, { "epoch": 0.2, "learning_rate": 4.5981735131454685e-05, "loss": 0.3847, "step": 5898 }, { "epoch": 0.2, "learning_rate": 4.59802874416774e-05, "loss": 1.0079, "step": 5899 }, { "epoch": 0.2, "learning_rate": 4.597883951396027e-05, "loss": 1.2906, "step": 5900 }, { "epoch": 0.2, "learning_rate": 4.597739134831974e-05, "loss": 1.3713, "step": 5901 }, { "epoch": 0.2, "learning_rate": 4.597594294477223e-05, "loss": 0.5068, "step": 5902 }, { "epoch": 0.2, "learning_rate": 4.597449430333416e-05, "loss": 0.3967, "step": 5903 }, { "epoch": 0.2, "learning_rate": 4.5973045424021954e-05, "loss": 0.4278, "step": 5904 }, { "epoch": 0.2, "learning_rate": 4.597159630685205e-05, "loss": 0.7589, "step": 5905 }, { "epoch": 0.2, "learning_rate": 4.5970146951840896e-05, "loss": 1.0851, "step": 5906 }, { "epoch": 0.2, "learning_rate": 4.59686973590049e-05, "loss": 0.8692, "step": 5907 }, { "epoch": 0.2, "learning_rate": 4.596724752836052e-05, "loss": 0.7704, "step": 5908 }, { "epoch": 0.2, "learning_rate": 4.596579745992421e-05, "loss": 0.6985, "step": 5909 }, { "epoch": 0.2, "learning_rate": 4.5964347153712395e-05, "loss": 0.8316, "step": 5910 }, { "epoch": 0.2, "learning_rate": 4.5962896609741535e-05, "loss": 0.9464, "step": 5911 }, { "epoch": 0.2, "learning_rate": 4.596144582802807e-05, "loss": 0.9739, "step": 5912 }, { "epoch": 0.2, "learning_rate": 4.595999480858846e-05, "loss": 1.2406, "step": 5913 }, { "epoch": 0.2, "learning_rate": 4.5958543551439165e-05, "loss": 0.6508, "step": 5914 }, { "epoch": 0.2, "learning_rate": 4.595709205659664e-05, "loss": 1.0181, "step": 5915 }, { "epoch": 0.2, "learning_rate": 4.595564032407734e-05, "loss": 1.1426, "step": 5916 }, { "epoch": 0.2, "learning_rate": 4.595418835389774e-05, "loss": 0.486, "step": 5917 }, { "epoch": 0.2, "learning_rate": 4.5952736146074304e-05, "loss": 0.6637, "step": 5918 }, { "epoch": 0.2, "learning_rate": 4.595128370062351e-05, "loss": 0.5966, "step": 5919 }, { "epoch": 0.2, "learning_rate": 4.594983101756181e-05, "loss": 0.7778, "step": 5920 }, { "epoch": 0.2, "learning_rate": 4.594837809690568e-05, "loss": 0.8022, "step": 5921 }, { "epoch": 0.2, "learning_rate": 4.5946924938671624e-05, "loss": 0.7928, "step": 5922 }, { "epoch": 0.2, "learning_rate": 4.59454715428761e-05, "loss": 1.0309, "step": 5923 }, { "epoch": 0.2, "learning_rate": 4.594401790953559e-05, "loss": 0.9905, "step": 5924 }, { "epoch": 0.2, "learning_rate": 4.59425640386666e-05, "loss": 0.7842, "step": 5925 }, { "epoch": 0.2, "learning_rate": 4.59411099302856e-05, "loss": 0.6924, "step": 5926 }, { "epoch": 0.2, "learning_rate": 4.593965558440908e-05, "loss": 0.5434, "step": 5927 }, { "epoch": 0.2, "learning_rate": 4.593820100105355e-05, "loss": 1.1179, "step": 5928 }, { "epoch": 0.2, "learning_rate": 4.593674618023549e-05, "loss": 0.9822, "step": 5929 }, { "epoch": 0.2, "learning_rate": 4.5935291121971405e-05, "loss": 0.6958, "step": 5930 }, { "epoch": 0.2, "learning_rate": 4.593383582627781e-05, "loss": 0.9847, "step": 5931 }, { "epoch": 0.2, "learning_rate": 4.593238029317119e-05, "loss": 0.8126, "step": 5932 }, { "epoch": 0.2, "learning_rate": 4.593092452266806e-05, "loss": 0.5788, "step": 5933 }, { "epoch": 0.2, "learning_rate": 4.592946851478493e-05, "loss": 1.0241, "step": 5934 }, { "epoch": 0.2, "learning_rate": 4.592801226953831e-05, "loss": 1.9233, "step": 5935 }, { "epoch": 0.2, "learning_rate": 4.592655578694473e-05, "loss": 0.8217, "step": 5936 }, { "epoch": 0.2, "learning_rate": 4.592509906702069e-05, "loss": 0.8432, "step": 5937 }, { "epoch": 0.2, "learning_rate": 4.592364210978272e-05, "loss": 0.8868, "step": 5938 }, { "epoch": 0.2, "learning_rate": 4.592218491524734e-05, "loss": 0.7998, "step": 5939 }, { "epoch": 0.2, "learning_rate": 4.592072748343107e-05, "loss": 0.9295, "step": 5940 }, { "epoch": 0.2, "learning_rate": 4.591926981435045e-05, "loss": 0.9749, "step": 5941 }, { "epoch": 0.2, "learning_rate": 4.5917811908022016e-05, "loss": 0.8669, "step": 5942 }, { "epoch": 0.2, "learning_rate": 4.591635376446228e-05, "loss": 0.7188, "step": 5943 }, { "epoch": 0.2, "learning_rate": 4.591489538368781e-05, "loss": 0.4685, "step": 5944 }, { "epoch": 0.2, "learning_rate": 4.5913436765715126e-05, "loss": 0.6659, "step": 5945 }, { "epoch": 0.2, "learning_rate": 4.591197791056076e-05, "loss": 0.9683, "step": 5946 }, { "epoch": 0.2, "learning_rate": 4.591051881824128e-05, "loss": 0.4716, "step": 5947 }, { "epoch": 0.2, "learning_rate": 4.590905948877322e-05, "loss": 0.8108, "step": 5948 }, { "epoch": 0.2, "learning_rate": 4.5907599922173136e-05, "loss": 0.8298, "step": 5949 }, { "epoch": 0.2, "learning_rate": 4.590614011845758e-05, "loss": 0.6514, "step": 5950 }, { "epoch": 0.2, "learning_rate": 4.5904680077643105e-05, "loss": 0.4387, "step": 5951 }, { "epoch": 0.2, "learning_rate": 4.590321979974628e-05, "loss": 0.8858, "step": 5952 }, { "epoch": 0.2, "learning_rate": 4.590175928478365e-05, "loss": 3.5718, "step": 5953 }, { "epoch": 0.2, "learning_rate": 4.590029853277178e-05, "loss": 0.7316, "step": 5954 }, { "epoch": 0.2, "learning_rate": 4.589883754372726e-05, "loss": 0.7963, "step": 5955 }, { "epoch": 0.2, "learning_rate": 4.589737631766663e-05, "loss": 0.8264, "step": 5956 }, { "epoch": 0.2, "learning_rate": 4.589591485460648e-05, "loss": 1.0123, "step": 5957 }, { "epoch": 0.2, "learning_rate": 4.589445315456338e-05, "loss": 1.2956, "step": 5958 }, { "epoch": 0.2, "learning_rate": 4.58929912175539e-05, "loss": 1.9152, "step": 5959 }, { "epoch": 0.2, "learning_rate": 4.589152904359463e-05, "loss": 0.9436, "step": 5960 }, { "epoch": 0.2, "learning_rate": 4.589006663270215e-05, "loss": 0.8687, "step": 5961 }, { "epoch": 0.2, "learning_rate": 4.588860398489304e-05, "loss": 0.8442, "step": 5962 }, { "epoch": 0.2, "learning_rate": 4.5887141100183894e-05, "loss": 0.4154, "step": 5963 }, { "epoch": 0.2, "learning_rate": 4.588567797859131e-05, "loss": 0.7107, "step": 5964 }, { "epoch": 0.2, "learning_rate": 4.588421462013186e-05, "loss": 0.3832, "step": 5965 }, { "epoch": 0.2, "learning_rate": 4.5882751024822165e-05, "loss": 1.1618, "step": 5966 }, { "epoch": 0.2, "learning_rate": 4.58812871926788e-05, "loss": 0.8901, "step": 5967 }, { "epoch": 0.2, "learning_rate": 4.587982312371839e-05, "loss": 0.8985, "step": 5968 }, { "epoch": 0.2, "learning_rate": 4.5878358817957514e-05, "loss": 0.9144, "step": 5969 }, { "epoch": 0.2, "learning_rate": 4.58768942754128e-05, "loss": 0.8007, "step": 5970 }, { "epoch": 0.2, "learning_rate": 4.587542949610085e-05, "loss": 0.8877, "step": 5971 }, { "epoch": 0.2, "learning_rate": 4.587396448003828e-05, "loss": 0.8937, "step": 5972 }, { "epoch": 0.2, "learning_rate": 4.5872499227241696e-05, "loss": 0.6587, "step": 5973 }, { "epoch": 0.2, "learning_rate": 4.587103373772772e-05, "loss": 0.789, "step": 5974 }, { "epoch": 0.2, "learning_rate": 4.586956801151297e-05, "loss": 0.9914, "step": 5975 }, { "epoch": 0.2, "learning_rate": 4.586810204861407e-05, "loss": 1.7182, "step": 5976 }, { "epoch": 0.2, "learning_rate": 4.5866635849047654e-05, "loss": 0.7584, "step": 5977 }, { "epoch": 0.2, "learning_rate": 4.5865169412830344e-05, "loss": 0.908, "step": 5978 }, { "epoch": 0.2, "learning_rate": 4.586370273997876e-05, "loss": 0.9479, "step": 5979 }, { "epoch": 0.2, "learning_rate": 4.586223583050956e-05, "loss": 0.8471, "step": 5980 }, { "epoch": 0.2, "learning_rate": 4.586076868443936e-05, "loss": 0.7961, "step": 5981 }, { "epoch": 0.2, "learning_rate": 4.585930130178481e-05, "loss": 0.8926, "step": 5982 }, { "epoch": 0.2, "learning_rate": 4.585783368256254e-05, "loss": 0.9847, "step": 5983 }, { "epoch": 0.2, "learning_rate": 4.5856365826789207e-05, "loss": 0.7897, "step": 5984 }, { "epoch": 0.2, "learning_rate": 4.585489773448145e-05, "loss": 0.7816, "step": 5985 }, { "epoch": 0.2, "learning_rate": 4.5853429405655927e-05, "loss": 1.2359, "step": 5986 }, { "epoch": 0.2, "learning_rate": 4.585196084032928e-05, "loss": 1.1291, "step": 5987 }, { "epoch": 0.2, "learning_rate": 4.5850492038518175e-05, "loss": 0.8639, "step": 5988 }, { "epoch": 0.2, "learning_rate": 4.584902300023927e-05, "loss": 0.8832, "step": 5989 }, { "epoch": 0.2, "learning_rate": 4.58475537255092e-05, "loss": 1.5503, "step": 5990 }, { "epoch": 0.2, "learning_rate": 4.5846084214344666e-05, "loss": 0.9425, "step": 5991 }, { "epoch": 0.2, "learning_rate": 4.5844614466762306e-05, "loss": 3.0123, "step": 5992 }, { "epoch": 0.2, "learning_rate": 4.5843144482778804e-05, "loss": 0.9039, "step": 5993 }, { "epoch": 0.2, "learning_rate": 4.584167426241082e-05, "loss": 1.4195, "step": 5994 }, { "epoch": 0.2, "learning_rate": 4.584020380567504e-05, "loss": 1.5194, "step": 5995 }, { "epoch": 0.2, "learning_rate": 4.583873311258814e-05, "loss": 1.105, "step": 5996 }, { "epoch": 0.2, "learning_rate": 4.583726218316679e-05, "loss": 0.6371, "step": 5997 }, { "epoch": 0.2, "learning_rate": 4.5835791017427663e-05, "loss": 0.2613, "step": 5998 }, { "epoch": 0.2, "learning_rate": 4.583431961538747e-05, "loss": 5.6751, "step": 5999 }, { "epoch": 0.2, "learning_rate": 4.5832847977062874e-05, "loss": 2.2266, "step": 6000 } ], "max_steps": 30000, "num_train_epochs": 9223372036854775807, "total_flos": 4.811863388848128e+18, "trial_name": null, "trial_params": null }