{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "global_step": 1784, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 0.0, "loss": 9.0449, "step": 1 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 9.193, "step": 2 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 9.3707, "step": 3 }, { "epoch": 0.0, "learning_rate": 0.0, "loss": 9.0862, "step": 4 }, { "epoch": 0.0, "learning_rate": 2e-06, "loss": 8.5378, "step": 5 }, { "epoch": 0.0, "learning_rate": 4e-06, "loss": 8.7644, "step": 6 }, { "epoch": 0.0, "learning_rate": 6e-06, "loss": 8.3226, "step": 7 }, { "epoch": 0.0, "learning_rate": 8e-06, "loss": 8.1993, "step": 8 }, { "epoch": 0.01, "learning_rate": 8e-06, "loss": 8.813, "step": 9 }, { "epoch": 0.01, "learning_rate": 1e-05, "loss": 8.0419, "step": 10 }, { "epoch": 0.01, "learning_rate": 1.2e-05, "loss": 7.8058, "step": 11 }, { "epoch": 0.01, "learning_rate": 1.4e-05, "loss": 8.1209, "step": 12 }, { "epoch": 0.01, "learning_rate": 1.4e-05, "loss": 7.704, "step": 13 }, { "epoch": 0.01, "learning_rate": 1.6e-05, "loss": 7.7845, "step": 14 }, { "epoch": 0.01, "learning_rate": 1.8e-05, "loss": 7.5155, "step": 15 }, { "epoch": 0.01, "learning_rate": 2e-05, "loss": 7.4531, "step": 16 }, { "epoch": 0.01, "learning_rate": 2.2e-05, "loss": 7.5079, "step": 17 }, { "epoch": 0.01, "learning_rate": 2.4e-05, "loss": 7.271, "step": 18 }, { "epoch": 0.01, "learning_rate": 2.6e-05, "loss": 7.3847, "step": 19 }, { "epoch": 0.01, "learning_rate": 2.8e-05, "loss": 7.4302, "step": 20 }, { "epoch": 0.01, "learning_rate": 3e-05, "loss": 7.3806, "step": 21 }, { "epoch": 0.01, "learning_rate": 3.2e-05, "loss": 7.5205, "step": 22 }, { "epoch": 0.01, "learning_rate": 3.4000000000000007e-05, "loss": 7.4454, "step": 23 }, { "epoch": 0.01, "learning_rate": 3.6e-05, "loss": 7.2264, "step": 24 }, { "epoch": 0.01, "learning_rate": 3.8e-05, "loss": 7.2687, "step": 25 }, { "epoch": 0.01, "learning_rate": 4e-05, "loss": 7.2326, "step": 26 }, { "epoch": 0.02, "learning_rate": 4.2000000000000004e-05, "loss": 7.3255, "step": 27 }, { "epoch": 0.02, "learning_rate": 4.4e-05, "loss": 7.0537, "step": 28 }, { "epoch": 0.02, "learning_rate": 4.6e-05, "loss": 7.3762, "step": 29 }, { "epoch": 0.02, "learning_rate": 4.8e-05, "loss": 7.3015, "step": 30 }, { "epoch": 0.02, "learning_rate": 5e-05, "loss": 7.2366, "step": 31 }, { "epoch": 0.02, "learning_rate": 5.2e-05, "loss": 7.238, "step": 32 }, { "epoch": 0.02, "learning_rate": 5.4e-05, "loss": 7.2901, "step": 33 }, { "epoch": 0.02, "learning_rate": 5.6e-05, "loss": 7.052, "step": 34 }, { "epoch": 0.02, "learning_rate": 5.800000000000001e-05, "loss": 7.0733, "step": 35 }, { "epoch": 0.02, "learning_rate": 6e-05, "loss": 7.2914, "step": 36 }, { "epoch": 0.02, "learning_rate": 6.2e-05, "loss": 7.1048, "step": 37 }, { "epoch": 0.02, "learning_rate": 6.4e-05, "loss": 7.2913, "step": 38 }, { "epoch": 0.02, "learning_rate": 6.6e-05, "loss": 7.2576, "step": 39 }, { "epoch": 0.02, "learning_rate": 6.800000000000001e-05, "loss": 7.1443, "step": 40 }, { "epoch": 0.02, "learning_rate": 7.000000000000001e-05, "loss": 7.164, "step": 41 }, { "epoch": 0.02, "learning_rate": 7.2e-05, "loss": 7.0462, "step": 42 }, { "epoch": 0.02, "learning_rate": 7.4e-05, "loss": 7.241, "step": 43 }, { "epoch": 0.02, "learning_rate": 7.6e-05, "loss": 7.0124, "step": 44 }, { "epoch": 0.03, "learning_rate": 7.8e-05, "loss": 7.3261, "step": 45 }, { "epoch": 0.03, "learning_rate": 8e-05, "loss": 7.0794, "step": 46 }, { "epoch": 0.03, "learning_rate": 8.2e-05, "loss": 6.8892, "step": 47 }, { "epoch": 0.03, "learning_rate": 8.400000000000001e-05, "loss": 7.2552, "step": 48 }, { "epoch": 0.03, "learning_rate": 8.599999999999999e-05, "loss": 6.7829, "step": 49 }, { "epoch": 0.03, "learning_rate": 8.8e-05, "loss": 6.68, "step": 50 }, { "epoch": 0.03, "learning_rate": 8.999999999999999e-05, "loss": 7.3143, "step": 51 }, { "epoch": 0.03, "learning_rate": 9.2e-05, "loss": 7.365, "step": 52 }, { "epoch": 0.03, "learning_rate": 9.400000000000001e-05, "loss": 7.11, "step": 53 }, { "epoch": 0.03, "learning_rate": 9.6e-05, "loss": 7.1959, "step": 54 }, { "epoch": 0.03, "learning_rate": 9.800000000000001e-05, "loss": 7.3818, "step": 55 }, { "epoch": 0.03, "learning_rate": 0.0001, "loss": 7.1187, "step": 56 }, { "epoch": 0.03, "learning_rate": 0.000102, "loss": 7.0791, "step": 57 }, { "epoch": 0.03, "learning_rate": 0.000104, "loss": 6.9848, "step": 58 }, { "epoch": 0.03, "learning_rate": 0.000106, "loss": 7.0194, "step": 59 }, { "epoch": 0.03, "learning_rate": 0.000108, "loss": 7.0304, "step": 60 }, { "epoch": 0.03, "learning_rate": 0.00011, "loss": 7.1684, "step": 61 }, { "epoch": 0.03, "learning_rate": 0.000112, "loss": 7.1688, "step": 62 }, { "epoch": 0.04, "learning_rate": 0.000114, "loss": 7.2352, "step": 63 }, { "epoch": 0.04, "learning_rate": 0.00011600000000000001, "loss": 7.0932, "step": 64 }, { "epoch": 0.04, "learning_rate": 0.000118, "loss": 7.0533, "step": 65 }, { "epoch": 0.04, "learning_rate": 0.00012, "loss": 7.0164, "step": 66 }, { "epoch": 0.04, "learning_rate": 0.000122, "loss": 7.0437, "step": 67 }, { "epoch": 0.04, "learning_rate": 0.000124, "loss": 7.0614, "step": 68 }, { "epoch": 0.04, "learning_rate": 0.000126, "loss": 7.1547, "step": 69 }, { "epoch": 0.04, "learning_rate": 0.000128, "loss": 6.9383, "step": 70 }, { "epoch": 0.04, "learning_rate": 0.00013000000000000002, "loss": 7.0698, "step": 71 }, { "epoch": 0.04, "learning_rate": 0.000132, "loss": 7.0169, "step": 72 }, { "epoch": 0.04, "learning_rate": 0.000134, "loss": 7.1251, "step": 73 }, { "epoch": 0.04, "learning_rate": 0.00013600000000000003, "loss": 7.0169, "step": 74 }, { "epoch": 0.04, "learning_rate": 0.00013800000000000002, "loss": 7.2416, "step": 75 }, { "epoch": 0.04, "learning_rate": 0.00014000000000000001, "loss": 7.0462, "step": 76 }, { "epoch": 0.04, "learning_rate": 0.00014199999999999998, "loss": 7.1708, "step": 77 }, { "epoch": 0.04, "learning_rate": 0.000144, "loss": 7.013, "step": 78 }, { "epoch": 0.04, "learning_rate": 0.000146, "loss": 6.9791, "step": 79 }, { "epoch": 0.04, "learning_rate": 0.000148, "loss": 6.9562, "step": 80 }, { "epoch": 0.05, "learning_rate": 0.00015, "loss": 7.1318, "step": 81 }, { "epoch": 0.05, "learning_rate": 0.000152, "loss": 7.245, "step": 82 }, { "epoch": 0.05, "learning_rate": 0.000154, "loss": 6.9136, "step": 83 }, { "epoch": 0.05, "learning_rate": 0.000156, "loss": 7.0522, "step": 84 }, { "epoch": 0.05, "learning_rate": 0.000158, "loss": 7.1007, "step": 85 }, { "epoch": 0.05, "learning_rate": 0.00016, "loss": 7.0951, "step": 86 }, { "epoch": 0.05, "learning_rate": 0.000162, "loss": 7.2181, "step": 87 }, { "epoch": 0.05, "learning_rate": 0.000164, "loss": 7.0154, "step": 88 }, { "epoch": 0.05, "learning_rate": 0.00016600000000000002, "loss": 7.1123, "step": 89 }, { "epoch": 0.05, "learning_rate": 0.00016800000000000002, "loss": 6.9211, "step": 90 }, { "epoch": 0.05, "learning_rate": 0.00017, "loss": 7.1791, "step": 91 }, { "epoch": 0.05, "learning_rate": 0.00017199999999999998, "loss": 6.811, "step": 92 }, { "epoch": 0.05, "learning_rate": 0.000174, "loss": 6.9279, "step": 93 }, { "epoch": 0.05, "learning_rate": 0.000176, "loss": 7.1589, "step": 94 }, { "epoch": 0.05, "learning_rate": 0.000178, "loss": 7.1641, "step": 95 }, { "epoch": 0.05, "learning_rate": 0.00017999999999999998, "loss": 6.9397, "step": 96 }, { "epoch": 0.05, "learning_rate": 0.000182, "loss": 6.8892, "step": 97 }, { "epoch": 0.05, "learning_rate": 0.000184, "loss": 6.5806, "step": 98 }, { "epoch": 0.06, "learning_rate": 0.000186, "loss": 6.8593, "step": 99 }, { "epoch": 0.06, "learning_rate": 0.00018800000000000002, "loss": 6.7064, "step": 100 }, { "epoch": 0.06, "learning_rate": 0.00019, "loss": 7.2902, "step": 101 }, { "epoch": 0.06, "learning_rate": 0.000192, "loss": 7.2133, "step": 102 }, { "epoch": 0.06, "learning_rate": 0.000194, "loss": 7.1987, "step": 103 }, { "epoch": 0.06, "learning_rate": 0.00019600000000000002, "loss": 7.1214, "step": 104 }, { "epoch": 0.06, "learning_rate": 0.00019800000000000002, "loss": 7.0401, "step": 105 }, { "epoch": 0.06, "learning_rate": 0.0002, "loss": 7.0087, "step": 106 }, { "epoch": 0.06, "learning_rate": 0.000202, "loss": 7.4653, "step": 107 }, { "epoch": 0.06, "learning_rate": 0.000204, "loss": 7.0893, "step": 108 }, { "epoch": 0.06, "learning_rate": 0.000206, "loss": 6.9914, "step": 109 }, { "epoch": 0.06, "learning_rate": 0.000208, "loss": 7.1489, "step": 110 }, { "epoch": 0.06, "learning_rate": 0.00021, "loss": 6.9508, "step": 111 }, { "epoch": 0.06, "learning_rate": 0.000212, "loss": 6.8661, "step": 112 }, { "epoch": 0.06, "learning_rate": 0.000214, "loss": 7.1181, "step": 113 }, { "epoch": 0.06, "learning_rate": 0.000216, "loss": 7.1726, "step": 114 }, { "epoch": 0.06, "learning_rate": 0.000218, "loss": 7.0043, "step": 115 }, { "epoch": 0.07, "learning_rate": 0.00022, "loss": 7.0911, "step": 116 }, { "epoch": 0.07, "learning_rate": 0.000222, "loss": 7.1937, "step": 117 }, { "epoch": 0.07, "learning_rate": 0.000224, "loss": 6.9145, "step": 118 }, { "epoch": 0.07, "learning_rate": 0.00022600000000000002, "loss": 7.0464, "step": 119 }, { "epoch": 0.07, "learning_rate": 0.000228, "loss": 7.0933, "step": 120 }, { "epoch": 0.07, "learning_rate": 0.00023, "loss": 7.279, "step": 121 }, { "epoch": 0.07, "learning_rate": 0.00023200000000000003, "loss": 7.1724, "step": 122 }, { "epoch": 0.07, "learning_rate": 0.00023400000000000002, "loss": 7.129, "step": 123 }, { "epoch": 0.07, "learning_rate": 0.000236, "loss": 7.1728, "step": 124 }, { "epoch": 0.07, "learning_rate": 0.00023799999999999998, "loss": 6.8549, "step": 125 }, { "epoch": 0.07, "learning_rate": 0.00024, "loss": 7.2698, "step": 126 }, { "epoch": 0.07, "learning_rate": 0.000242, "loss": 7.0476, "step": 127 }, { "epoch": 0.07, "learning_rate": 0.000244, "loss": 7.0853, "step": 128 }, { "epoch": 0.07, "learning_rate": 0.000246, "loss": 6.9742, "step": 129 }, { "epoch": 0.07, "learning_rate": 0.000248, "loss": 6.9995, "step": 130 }, { "epoch": 0.07, "learning_rate": 0.00025, "loss": 7.2572, "step": 131 }, { "epoch": 0.07, "learning_rate": 0.000252, "loss": 7.0498, "step": 132 }, { "epoch": 0.07, "learning_rate": 0.000254, "loss": 7.2188, "step": 133 }, { "epoch": 0.08, "learning_rate": 0.000256, "loss": 7.0341, "step": 134 }, { "epoch": 0.08, "learning_rate": 0.00025800000000000004, "loss": 6.9916, "step": 135 }, { "epoch": 0.08, "learning_rate": 0.00026000000000000003, "loss": 7.252, "step": 136 }, { "epoch": 0.08, "learning_rate": 0.000262, "loss": 7.0907, "step": 137 }, { "epoch": 0.08, "learning_rate": 0.000264, "loss": 7.021, "step": 138 }, { "epoch": 0.08, "learning_rate": 0.000266, "loss": 7.2201, "step": 139 }, { "epoch": 0.08, "learning_rate": 0.000268, "loss": 7.1205, "step": 140 }, { "epoch": 0.08, "learning_rate": 0.00027, "loss": 7.3046, "step": 141 }, { "epoch": 0.08, "learning_rate": 0.00027200000000000005, "loss": 7.2678, "step": 142 }, { "epoch": 0.08, "learning_rate": 0.00027400000000000005, "loss": 6.9769, "step": 143 }, { "epoch": 0.08, "learning_rate": 0.00027600000000000004, "loss": 7.1021, "step": 144 }, { "epoch": 0.08, "learning_rate": 0.00027800000000000004, "loss": 7.1826, "step": 145 }, { "epoch": 0.08, "learning_rate": 0.00028000000000000003, "loss": 7.1176, "step": 146 }, { "epoch": 0.08, "learning_rate": 0.00028199999999999997, "loss": 7.091, "step": 147 }, { "epoch": 0.08, "learning_rate": 0.00028399999999999996, "loss": 6.9994, "step": 148 }, { "epoch": 0.08, "learning_rate": 0.00028599999999999996, "loss": 7.0775, "step": 149 }, { "epoch": 0.08, "learning_rate": 0.000288, "loss": 6.2802, "step": 150 }, { "epoch": 0.08, "learning_rate": 0.00029, "loss": 7.038, "step": 151 }, { "epoch": 0.09, "learning_rate": 0.000292, "loss": 7.1197, "step": 152 }, { "epoch": 0.09, "learning_rate": 0.000294, "loss": 6.9578, "step": 153 }, { "epoch": 0.09, "learning_rate": 0.000296, "loss": 7.2523, "step": 154 }, { "epoch": 0.09, "learning_rate": 0.000298, "loss": 6.9292, "step": 155 }, { "epoch": 0.09, "learning_rate": 0.0003, "loss": 7.2391, "step": 156 }, { "epoch": 0.09, "learning_rate": 0.000302, "loss": 7.0842, "step": 157 }, { "epoch": 0.09, "learning_rate": 0.000304, "loss": 7.0349, "step": 158 }, { "epoch": 0.09, "learning_rate": 0.000306, "loss": 7.0477, "step": 159 }, { "epoch": 0.09, "learning_rate": 0.000308, "loss": 7.2366, "step": 160 }, { "epoch": 0.09, "learning_rate": 0.00031, "loss": 7.1977, "step": 161 }, { "epoch": 0.09, "learning_rate": 0.000312, "loss": 7.322, "step": 162 }, { "epoch": 0.09, "learning_rate": 0.000314, "loss": 7.1274, "step": 163 }, { "epoch": 0.09, "learning_rate": 0.000316, "loss": 7.1086, "step": 164 }, { "epoch": 0.09, "learning_rate": 0.00031800000000000003, "loss": 7.1218, "step": 165 }, { "epoch": 0.09, "learning_rate": 0.00032, "loss": 7.1711, "step": 166 }, { "epoch": 0.09, "learning_rate": 0.000322, "loss": 7.1521, "step": 167 }, { "epoch": 0.09, "learning_rate": 0.000324, "loss": 6.9295, "step": 168 }, { "epoch": 0.09, "learning_rate": 0.000326, "loss": 6.8792, "step": 169 }, { "epoch": 0.1, "learning_rate": 0.000328, "loss": 7.2185, "step": 170 }, { "epoch": 0.1, "learning_rate": 0.00033, "loss": 7.1299, "step": 171 }, { "epoch": 0.1, "learning_rate": 0.00033200000000000005, "loss": 7.07, "step": 172 }, { "epoch": 0.1, "learning_rate": 0.00033400000000000004, "loss": 7.0571, "step": 173 }, { "epoch": 0.1, "learning_rate": 0.00033600000000000004, "loss": 6.9991, "step": 174 }, { "epoch": 0.1, "learning_rate": 0.00033800000000000003, "loss": 7.176, "step": 175 }, { "epoch": 0.1, "learning_rate": 0.00034, "loss": 6.928, "step": 176 }, { "epoch": 0.1, "learning_rate": 0.000342, "loss": 6.957, "step": 177 }, { "epoch": 0.1, "learning_rate": 0.00034399999999999996, "loss": 7.0252, "step": 178 }, { "epoch": 0.1, "learning_rate": 0.000346, "loss": 6.9202, "step": 179 }, { "epoch": 0.1, "learning_rate": 0.000348, "loss": 7.0868, "step": 180 }, { "epoch": 0.1, "learning_rate": 0.00035, "loss": 6.9686, "step": 181 }, { "epoch": 0.1, "learning_rate": 0.000352, "loss": 6.9302, "step": 182 }, { "epoch": 0.1, "learning_rate": 0.000354, "loss": 7.1864, "step": 183 }, { "epoch": 0.1, "learning_rate": 0.000356, "loss": 6.8742, "step": 184 }, { "epoch": 0.1, "learning_rate": 0.000358, "loss": 7.0741, "step": 185 }, { "epoch": 0.1, "learning_rate": 0.00035999999999999997, "loss": 6.9962, "step": 186 }, { "epoch": 0.1, "learning_rate": 0.000362, "loss": 6.9308, "step": 187 }, { "epoch": 0.11, "learning_rate": 0.000364, "loss": 7.0139, "step": 188 }, { "epoch": 0.11, "learning_rate": 0.000366, "loss": 6.9891, "step": 189 }, { "epoch": 0.11, "learning_rate": 0.000368, "loss": 6.9435, "step": 190 }, { "epoch": 0.11, "learning_rate": 0.00037, "loss": 7.1724, "step": 191 }, { "epoch": 0.11, "learning_rate": 0.000372, "loss": 6.8671, "step": 192 }, { "epoch": 0.11, "learning_rate": 0.000374, "loss": 6.9104, "step": 193 }, { "epoch": 0.11, "learning_rate": 0.00037600000000000003, "loss": 7.0094, "step": 194 }, { "epoch": 0.11, "learning_rate": 0.000378, "loss": 7.0253, "step": 195 }, { "epoch": 0.11, "learning_rate": 0.00038, "loss": 6.9614, "step": 196 }, { "epoch": 0.11, "learning_rate": 0.000382, "loss": 6.9934, "step": 197 }, { "epoch": 0.11, "learning_rate": 0.000384, "loss": 6.7156, "step": 198 }, { "epoch": 0.11, "learning_rate": 0.000386, "loss": 6.6918, "step": 199 }, { "epoch": 0.11, "learning_rate": 0.000388, "loss": 6.5002, "step": 200 }, { "epoch": 0.11, "learning_rate": 0.00039000000000000005, "loss": 7.3564, "step": 201 }, { "epoch": 0.11, "learning_rate": 0.00039200000000000004, "loss": 7.1292, "step": 202 }, { "epoch": 0.11, "learning_rate": 0.00039400000000000004, "loss": 7.3952, "step": 203 }, { "epoch": 0.11, "learning_rate": 0.00039600000000000003, "loss": 6.98, "step": 204 }, { "epoch": 0.11, "learning_rate": 0.000398, "loss": 6.9766, "step": 205 }, { "epoch": 0.12, "learning_rate": 0.0004, "loss": 7.1108, "step": 206 }, { "epoch": 0.12, "learning_rate": 0.000402, "loss": 7.1532, "step": 207 }, { "epoch": 0.12, "learning_rate": 0.000404, "loss": 7.0252, "step": 208 }, { "epoch": 0.12, "learning_rate": 0.00040600000000000006, "loss": 7.1127, "step": 209 }, { "epoch": 0.12, "learning_rate": 0.000408, "loss": 7.236, "step": 210 }, { "epoch": 0.12, "learning_rate": 0.00041, "loss": 7.0481, "step": 211 }, { "epoch": 0.12, "learning_rate": 0.000412, "loss": 7.1797, "step": 212 }, { "epoch": 0.12, "learning_rate": 0.000414, "loss": 7.0596, "step": 213 }, { "epoch": 0.12, "learning_rate": 0.000416, "loss": 6.8977, "step": 214 }, { "epoch": 0.12, "learning_rate": 0.00041799999999999997, "loss": 7.1588, "step": 215 }, { "epoch": 0.12, "learning_rate": 0.00042, "loss": 6.8753, "step": 216 }, { "epoch": 0.12, "learning_rate": 0.000422, "loss": 7.0657, "step": 217 }, { "epoch": 0.12, "learning_rate": 0.000424, "loss": 6.9538, "step": 218 }, { "epoch": 0.12, "learning_rate": 0.000426, "loss": 7.0376, "step": 219 }, { "epoch": 0.12, "learning_rate": 0.000428, "loss": 7.1085, "step": 220 }, { "epoch": 0.12, "learning_rate": 0.00043, "loss": 7.1454, "step": 221 }, { "epoch": 0.12, "learning_rate": 0.000432, "loss": 7.2554, "step": 222 }, { "epoch": 0.12, "learning_rate": 0.00043400000000000003, "loss": 6.9149, "step": 223 }, { "epoch": 0.13, "learning_rate": 0.000436, "loss": 7.0355, "step": 224 }, { "epoch": 0.13, "learning_rate": 0.000438, "loss": 7.1304, "step": 225 }, { "epoch": 0.13, "learning_rate": 0.00044, "loss": 7.3685, "step": 226 }, { "epoch": 0.13, "learning_rate": 0.000442, "loss": 7.2402, "step": 227 }, { "epoch": 0.13, "learning_rate": 0.000444, "loss": 7.1775, "step": 228 }, { "epoch": 0.13, "learning_rate": 0.000446, "loss": 7.0138, "step": 229 }, { "epoch": 0.13, "learning_rate": 0.000448, "loss": 7.0455, "step": 230 }, { "epoch": 0.13, "learning_rate": 0.00045000000000000004, "loss": 7.0441, "step": 231 }, { "epoch": 0.13, "learning_rate": 0.00045200000000000004, "loss": 7.1548, "step": 232 }, { "epoch": 0.13, "learning_rate": 0.00045400000000000003, "loss": 7.2901, "step": 233 }, { "epoch": 0.13, "learning_rate": 0.000456, "loss": 7.0741, "step": 234 }, { "epoch": 0.13, "learning_rate": 0.000458, "loss": 6.9167, "step": 235 }, { "epoch": 0.13, "learning_rate": 0.00046, "loss": 7.3052, "step": 236 }, { "epoch": 0.13, "learning_rate": 0.000462, "loss": 7.1367, "step": 237 }, { "epoch": 0.13, "learning_rate": 0.00046400000000000006, "loss": 7.1115, "step": 238 }, { "epoch": 0.13, "learning_rate": 0.00046600000000000005, "loss": 7.0402, "step": 239 }, { "epoch": 0.13, "learning_rate": 0.00046800000000000005, "loss": 7.031, "step": 240 }, { "epoch": 0.14, "learning_rate": 0.00047, "loss": 7.0444, "step": 241 }, { "epoch": 0.14, "learning_rate": 0.000472, "loss": 7.0321, "step": 242 }, { "epoch": 0.14, "learning_rate": 0.000474, "loss": 6.9353, "step": 243 }, { "epoch": 0.14, "learning_rate": 0.00047599999999999997, "loss": 7.0772, "step": 244 }, { "epoch": 0.14, "learning_rate": 0.00047799999999999996, "loss": 6.9884, "step": 245 }, { "epoch": 0.14, "learning_rate": 0.00048, "loss": 7.0901, "step": 246 }, { "epoch": 0.14, "learning_rate": 0.000482, "loss": 6.5488, "step": 247 }, { "epoch": 0.14, "learning_rate": 0.000484, "loss": 6.9884, "step": 248 }, { "epoch": 0.14, "learning_rate": 0.000486, "loss": 6.7535, "step": 249 }, { "epoch": 0.14, "learning_rate": 0.000488, "loss": 6.5439, "step": 250 }, { "epoch": 0.14, "learning_rate": 0.00049, "loss": 7.2145, "step": 251 }, { "epoch": 0.14, "learning_rate": 0.000492, "loss": 7.4771, "step": 252 }, { "epoch": 0.14, "learning_rate": 0.000494, "loss": 7.1361, "step": 253 }, { "epoch": 0.14, "learning_rate": 0.000496, "loss": 7.266, "step": 254 }, { "epoch": 0.14, "learning_rate": 0.000498, "loss": 7.1586, "step": 255 }, { "epoch": 0.14, "learning_rate": 0.0005, "loss": 7.1224, "step": 256 }, { "epoch": 0.14, "learning_rate": 0.0005020000000000001, "loss": 7.1853, "step": 257 }, { "epoch": 0.14, "learning_rate": 0.000504, "loss": 7.1228, "step": 258 }, { "epoch": 0.15, "learning_rate": 0.000506, "loss": 7.341, "step": 259 }, { "epoch": 0.15, "learning_rate": 0.000508, "loss": 6.8782, "step": 260 }, { "epoch": 0.15, "learning_rate": 0.00051, "loss": 6.9954, "step": 261 }, { "epoch": 0.15, "learning_rate": 0.000512, "loss": 7.0171, "step": 262 }, { "epoch": 0.15, "learning_rate": 0.000514, "loss": 7.2959, "step": 263 }, { "epoch": 0.15, "learning_rate": 0.0005160000000000001, "loss": 7.2657, "step": 264 }, { "epoch": 0.15, "learning_rate": 0.000518, "loss": 7.0276, "step": 265 }, { "epoch": 0.15, "learning_rate": 0.0005200000000000001, "loss": 6.9184, "step": 266 }, { "epoch": 0.15, "learning_rate": 0.000522, "loss": 7.1513, "step": 267 }, { "epoch": 0.15, "learning_rate": 0.000524, "loss": 7.0916, "step": 268 }, { "epoch": 0.15, "learning_rate": 0.000526, "loss": 7.1834, "step": 269 }, { "epoch": 0.15, "learning_rate": 0.000528, "loss": 6.9473, "step": 270 }, { "epoch": 0.15, "learning_rate": 0.0005300000000000001, "loss": 7.0075, "step": 271 }, { "epoch": 0.15, "learning_rate": 0.000532, "loss": 7.0776, "step": 272 }, { "epoch": 0.15, "learning_rate": 0.0005340000000000001, "loss": 6.9425, "step": 273 }, { "epoch": 0.15, "learning_rate": 0.000536, "loss": 7.0327, "step": 274 }, { "epoch": 0.15, "learning_rate": 0.0005380000000000001, "loss": 7.1699, "step": 275 }, { "epoch": 0.15, "learning_rate": 0.00054, "loss": 7.244, "step": 276 }, { "epoch": 0.16, "learning_rate": 0.0005420000000000001, "loss": 7.0916, "step": 277 }, { "epoch": 0.16, "learning_rate": 0.0005440000000000001, "loss": 7.2133, "step": 278 }, { "epoch": 0.16, "learning_rate": 0.000546, "loss": 7.0134, "step": 279 }, { "epoch": 0.16, "learning_rate": 0.0005480000000000001, "loss": 7.0721, "step": 280 }, { "epoch": 0.16, "learning_rate": 0.00055, "loss": 7.1009, "step": 281 }, { "epoch": 0.16, "learning_rate": 0.0005520000000000001, "loss": 7.2677, "step": 282 }, { "epoch": 0.16, "learning_rate": 0.000554, "loss": 7.1846, "step": 283 }, { "epoch": 0.16, "learning_rate": 0.0005560000000000001, "loss": 6.8191, "step": 284 }, { "epoch": 0.16, "learning_rate": 0.000558, "loss": 7.1693, "step": 285 }, { "epoch": 0.16, "learning_rate": 0.0005600000000000001, "loss": 6.9194, "step": 286 }, { "epoch": 0.16, "learning_rate": 0.0005620000000000001, "loss": 6.9699, "step": 287 }, { "epoch": 0.16, "learning_rate": 0.0005639999999999999, "loss": 7.2756, "step": 288 }, { "epoch": 0.16, "learning_rate": 0.000566, "loss": 6.9259, "step": 289 }, { "epoch": 0.16, "learning_rate": 0.0005679999999999999, "loss": 7.1642, "step": 290 }, { "epoch": 0.16, "learning_rate": 0.00057, "loss": 7.1586, "step": 291 }, { "epoch": 0.16, "learning_rate": 0.0005719999999999999, "loss": 7.286, "step": 292 }, { "epoch": 0.16, "learning_rate": 0.000574, "loss": 7.0571, "step": 293 }, { "epoch": 0.16, "learning_rate": 0.000576, "loss": 7.1162, "step": 294 }, { "epoch": 0.17, "learning_rate": 0.000578, "loss": 6.9448, "step": 295 }, { "epoch": 0.17, "learning_rate": 0.00058, "loss": 6.841, "step": 296 }, { "epoch": 0.17, "learning_rate": 0.0005819999999999999, "loss": 6.8322, "step": 297 }, { "epoch": 0.17, "learning_rate": 0.000584, "loss": 7.172, "step": 298 }, { "epoch": 0.17, "learning_rate": 0.0005859999999999999, "loss": 6.8272, "step": 299 }, { "epoch": 0.17, "learning_rate": 0.000588, "loss": 6.4481, "step": 300 }, { "epoch": 0.17, "learning_rate": 0.00059, "loss": 7.1124, "step": 301 }, { "epoch": 0.17, "learning_rate": 0.000592, "loss": 7.0253, "step": 302 }, { "epoch": 0.17, "learning_rate": 0.000594, "loss": 7.0942, "step": 303 }, { "epoch": 0.17, "learning_rate": 0.000596, "loss": 6.9754, "step": 304 }, { "epoch": 0.17, "learning_rate": 0.000598, "loss": 7.1096, "step": 305 }, { "epoch": 0.17, "learning_rate": 0.0006, "loss": 7.0429, "step": 306 }, { "epoch": 0.17, "learning_rate": 0.000602, "loss": 7.1405, "step": 307 }, { "epoch": 0.17, "learning_rate": 0.000604, "loss": 7.2294, "step": 308 }, { "epoch": 0.17, "learning_rate": 0.000606, "loss": 7.0469, "step": 309 }, { "epoch": 0.17, "learning_rate": 0.000608, "loss": 7.2496, "step": 310 }, { "epoch": 0.17, "learning_rate": 0.00061, "loss": 7.0246, "step": 311 }, { "epoch": 0.17, "learning_rate": 0.000612, "loss": 7.1387, "step": 312 }, { "epoch": 0.18, "learning_rate": 0.000614, "loss": 7.0737, "step": 313 }, { "epoch": 0.18, "learning_rate": 0.000616, "loss": 7.0853, "step": 314 }, { "epoch": 0.18, "learning_rate": 0.0006180000000000001, "loss": 7.1228, "step": 315 }, { "epoch": 0.18, "learning_rate": 0.00062, "loss": 6.9528, "step": 316 }, { "epoch": 0.18, "learning_rate": 0.000622, "loss": 6.9567, "step": 317 }, { "epoch": 0.18, "learning_rate": 0.000624, "loss": 6.9849, "step": 318 }, { "epoch": 0.18, "learning_rate": 0.000626, "loss": 7.1837, "step": 319 }, { "epoch": 0.18, "learning_rate": 0.000628, "loss": 7.1624, "step": 320 }, { "epoch": 0.18, "learning_rate": 0.00063, "loss": 7.3353, "step": 321 }, { "epoch": 0.18, "learning_rate": 0.000632, "loss": 7.2681, "step": 322 }, { "epoch": 0.18, "learning_rate": 0.000634, "loss": 7.2671, "step": 323 }, { "epoch": 0.18, "learning_rate": 0.0006360000000000001, "loss": 7.0366, "step": 324 }, { "epoch": 0.18, "learning_rate": 0.000638, "loss": 6.9454, "step": 325 }, { "epoch": 0.18, "learning_rate": 0.00064, "loss": 7.2206, "step": 326 }, { "epoch": 0.18, "learning_rate": 0.000642, "loss": 7.0422, "step": 327 }, { "epoch": 0.18, "learning_rate": 0.000644, "loss": 7.0474, "step": 328 }, { "epoch": 0.18, "learning_rate": 0.000646, "loss": 7.0165, "step": 329 }, { "epoch": 0.18, "learning_rate": 0.000648, "loss": 7.1498, "step": 330 }, { "epoch": 0.19, "learning_rate": 0.0006500000000000001, "loss": 7.1095, "step": 331 }, { "epoch": 0.19, "learning_rate": 0.000652, "loss": 7.0933, "step": 332 }, { "epoch": 0.19, "learning_rate": 0.0006540000000000001, "loss": 7.2449, "step": 333 }, { "epoch": 0.19, "learning_rate": 0.000656, "loss": 7.2348, "step": 334 }, { "epoch": 0.19, "learning_rate": 0.0006580000000000001, "loss": 7.0659, "step": 335 }, { "epoch": 0.19, "learning_rate": 0.00066, "loss": 7.0991, "step": 336 }, { "epoch": 0.19, "learning_rate": 0.000662, "loss": 7.0924, "step": 337 }, { "epoch": 0.19, "learning_rate": 0.0006640000000000001, "loss": 6.8876, "step": 338 }, { "epoch": 0.19, "learning_rate": 0.000666, "loss": 7.0145, "step": 339 }, { "epoch": 0.19, "learning_rate": 0.0006680000000000001, "loss": 7.088, "step": 340 }, { "epoch": 0.19, "learning_rate": 0.00067, "loss": 6.8834, "step": 341 }, { "epoch": 0.19, "learning_rate": 0.0006720000000000001, "loss": 7.2134, "step": 342 }, { "epoch": 0.19, "learning_rate": 0.000674, "loss": 7.1934, "step": 343 }, { "epoch": 0.19, "learning_rate": 0.0006760000000000001, "loss": 7.1019, "step": 344 }, { "epoch": 0.19, "learning_rate": 0.0006780000000000001, "loss": 6.8289, "step": 345 }, { "epoch": 0.19, "learning_rate": 0.00068, "loss": 6.9121, "step": 346 }, { "epoch": 0.19, "learning_rate": 0.0006820000000000001, "loss": 7.5627, "step": 347 }, { "epoch": 0.2, "learning_rate": 0.000684, "loss": 6.9207, "step": 348 }, { "epoch": 0.2, "learning_rate": 0.0006860000000000001, "loss": 6.465, "step": 349 }, { "epoch": 0.2, "learning_rate": 0.0006879999999999999, "loss": 6.2971, "step": 350 }, { "epoch": 0.2, "learning_rate": 0.00069, "loss": 7.2553, "step": 351 }, { "epoch": 0.2, "learning_rate": 0.000692, "loss": 7.3464, "step": 352 }, { "epoch": 0.2, "learning_rate": 0.000694, "loss": 7.1487, "step": 353 }, { "epoch": 0.2, "learning_rate": 0.000696, "loss": 7.0629, "step": 354 }, { "epoch": 0.2, "learning_rate": 0.0006979999999999999, "loss": 7.0028, "step": 355 }, { "epoch": 0.2, "learning_rate": 0.0007, "loss": 7.1901, "step": 356 }, { "epoch": 0.2, "learning_rate": 0.0007019999999999999, "loss": 7.3033, "step": 357 }, { "epoch": 0.2, "learning_rate": 0.000704, "loss": 7.1428, "step": 358 }, { "epoch": 0.2, "learning_rate": 0.0007059999999999999, "loss": 7.2864, "step": 359 }, { "epoch": 0.2, "learning_rate": 0.000708, "loss": 7.1572, "step": 360 }, { "epoch": 0.2, "learning_rate": 0.00071, "loss": 6.9379, "step": 361 }, { "epoch": 0.2, "learning_rate": 0.000712, "loss": 6.8767, "step": 362 }, { "epoch": 0.2, "learning_rate": 0.000714, "loss": 7.087, "step": 363 }, { "epoch": 0.2, "learning_rate": 0.000716, "loss": 7.0092, "step": 364 }, { "epoch": 0.2, "learning_rate": 0.000718, "loss": 7.0475, "step": 365 }, { "epoch": 0.21, "learning_rate": 0.0007199999999999999, "loss": 7.2061, "step": 366 }, { "epoch": 0.21, "learning_rate": 0.000722, "loss": 7.0312, "step": 367 }, { "epoch": 0.21, "learning_rate": 0.000724, "loss": 7.0803, "step": 368 }, { "epoch": 0.21, "learning_rate": 0.000726, "loss": 7.1741, "step": 369 }, { "epoch": 0.21, "learning_rate": 0.000728, "loss": 7.0005, "step": 370 }, { "epoch": 0.21, "learning_rate": 0.00073, "loss": 6.9798, "step": 371 }, { "epoch": 0.21, "learning_rate": 0.000732, "loss": 7.2175, "step": 372 }, { "epoch": 0.21, "learning_rate": 0.000734, "loss": 7.109, "step": 373 }, { "epoch": 0.21, "learning_rate": 0.000736, "loss": 6.9764, "step": 374 }, { "epoch": 0.21, "learning_rate": 0.000738, "loss": 6.8523, "step": 375 }, { "epoch": 0.21, "learning_rate": 0.00074, "loss": 7.0227, "step": 376 }, { "epoch": 0.21, "learning_rate": 0.000742, "loss": 7.0662, "step": 377 }, { "epoch": 0.21, "learning_rate": 0.000744, "loss": 7.0108, "step": 378 }, { "epoch": 0.21, "learning_rate": 0.000746, "loss": 7.0721, "step": 379 }, { "epoch": 0.21, "learning_rate": 0.000748, "loss": 7.2405, "step": 380 }, { "epoch": 0.21, "learning_rate": 0.00075, "loss": 6.9626, "step": 381 }, { "epoch": 0.21, "learning_rate": 0.0007520000000000001, "loss": 6.91, "step": 382 }, { "epoch": 0.21, "learning_rate": 0.000754, "loss": 7.0326, "step": 383 }, { "epoch": 0.22, "learning_rate": 0.000756, "loss": 7.1043, "step": 384 }, { "epoch": 0.22, "learning_rate": 0.000758, "loss": 7.2248, "step": 385 }, { "epoch": 0.22, "learning_rate": 0.00076, "loss": 7.3742, "step": 386 }, { "epoch": 0.22, "learning_rate": 0.000762, "loss": 7.0994, "step": 387 }, { "epoch": 0.22, "learning_rate": 0.000764, "loss": 6.9431, "step": 388 }, { "epoch": 0.22, "learning_rate": 0.0007660000000000001, "loss": 6.9923, "step": 389 }, { "epoch": 0.22, "learning_rate": 0.000768, "loss": 7.1723, "step": 390 }, { "epoch": 0.22, "learning_rate": 0.0007700000000000001, "loss": 6.9941, "step": 391 }, { "epoch": 0.22, "learning_rate": 0.000772, "loss": 7.6231, "step": 392 }, { "epoch": 0.22, "learning_rate": 0.0007740000000000001, "loss": 7.148, "step": 393 }, { "epoch": 0.22, "learning_rate": 0.000776, "loss": 6.9496, "step": 394 }, { "epoch": 0.22, "learning_rate": 0.000778, "loss": 6.9835, "step": 395 }, { "epoch": 0.22, "learning_rate": 0.0007800000000000001, "loss": 6.8888, "step": 396 }, { "epoch": 0.22, "learning_rate": 0.000782, "loss": 6.8057, "step": 397 }, { "epoch": 0.22, "learning_rate": 0.0007840000000000001, "loss": 7.0089, "step": 398 }, { "epoch": 0.22, "learning_rate": 0.000786, "loss": 7.1334, "step": 399 }, { "epoch": 0.22, "learning_rate": 0.0007880000000000001, "loss": 6.6337, "step": 400 }, { "epoch": 0.22, "learning_rate": 0.00079, "loss": 7.2806, "step": 401 }, { "epoch": 0.23, "learning_rate": 0.0007920000000000001, "loss": 7.046, "step": 402 }, { "epoch": 0.23, "learning_rate": 0.0007940000000000001, "loss": 6.9943, "step": 403 }, { "epoch": 0.23, "learning_rate": 0.000796, "loss": 7.2447, "step": 404 }, { "epoch": 0.23, "learning_rate": 0.0007980000000000001, "loss": 7.0006, "step": 405 }, { "epoch": 0.23, "learning_rate": 0.0008, "loss": 7.117, "step": 406 }, { "epoch": 0.23, "learning_rate": 0.0008020000000000001, "loss": 7.02, "step": 407 }, { "epoch": 0.23, "learning_rate": 0.000804, "loss": 7.0829, "step": 408 }, { "epoch": 0.23, "learning_rate": 0.0008060000000000001, "loss": 7.0694, "step": 409 }, { "epoch": 0.23, "learning_rate": 0.000808, "loss": 6.9274, "step": 410 }, { "epoch": 0.23, "learning_rate": 0.0008100000000000001, "loss": 7.0126, "step": 411 }, { "epoch": 0.23, "learning_rate": 0.0008120000000000001, "loss": 6.9576, "step": 412 }, { "epoch": 0.23, "learning_rate": 0.0008139999999999999, "loss": 7.1528, "step": 413 }, { "epoch": 0.23, "learning_rate": 0.000816, "loss": 6.9551, "step": 414 }, { "epoch": 0.23, "learning_rate": 0.0008179999999999999, "loss": 7.0775, "step": 415 }, { "epoch": 0.23, "learning_rate": 0.00082, "loss": 7.0943, "step": 416 }, { "epoch": 0.23, "learning_rate": 0.0008219999999999999, "loss": 7.0036, "step": 417 }, { "epoch": 0.23, "learning_rate": 0.000824, "loss": 7.0297, "step": 418 }, { "epoch": 0.23, "learning_rate": 0.000826, "loss": 7.0089, "step": 419 }, { "epoch": 0.24, "learning_rate": 0.000828, "loss": 7.2112, "step": 420 }, { "epoch": 0.24, "learning_rate": 0.00083, "loss": 7.1332, "step": 421 }, { "epoch": 0.24, "learning_rate": 0.000832, "loss": 7.1332, "step": 422 }, { "epoch": 0.24, "learning_rate": 0.000834, "loss": 6.999, "step": 423 }, { "epoch": 0.24, "learning_rate": 0.0008359999999999999, "loss": 7.2014, "step": 424 }, { "epoch": 0.24, "learning_rate": 0.000838, "loss": 7.2448, "step": 425 }, { "epoch": 0.24, "learning_rate": 0.00084, "loss": 7.1467, "step": 426 }, { "epoch": 0.24, "learning_rate": 0.000842, "loss": 6.9952, "step": 427 }, { "epoch": 0.24, "learning_rate": 0.000844, "loss": 7.0384, "step": 428 }, { "epoch": 0.24, "learning_rate": 0.000846, "loss": 7.0374, "step": 429 }, { "epoch": 0.24, "learning_rate": 0.000848, "loss": 7.032, "step": 430 }, { "epoch": 0.24, "learning_rate": 0.00085, "loss": 7.0036, "step": 431 }, { "epoch": 0.24, "learning_rate": 0.000852, "loss": 7.0825, "step": 432 }, { "epoch": 0.24, "learning_rate": 0.000854, "loss": 7.0835, "step": 433 }, { "epoch": 0.24, "learning_rate": 0.000856, "loss": 7.2344, "step": 434 }, { "epoch": 0.24, "learning_rate": 0.000858, "loss": 7.0893, "step": 435 }, { "epoch": 0.24, "learning_rate": 0.00086, "loss": 6.9788, "step": 436 }, { "epoch": 0.24, "learning_rate": 0.000862, "loss": 7.0012, "step": 437 }, { "epoch": 0.25, "learning_rate": 0.000864, "loss": 7.0822, "step": 438 }, { "epoch": 0.25, "learning_rate": 0.000866, "loss": 6.9395, "step": 439 }, { "epoch": 0.25, "learning_rate": 0.0008680000000000001, "loss": 7.0919, "step": 440 }, { "epoch": 0.25, "learning_rate": 0.00087, "loss": 6.9875, "step": 441 }, { "epoch": 0.25, "learning_rate": 0.000872, "loss": 7.0313, "step": 442 }, { "epoch": 0.25, "learning_rate": 0.000874, "loss": 6.8639, "step": 443 }, { "epoch": 0.25, "learning_rate": 0.000876, "loss": 6.9182, "step": 444 }, { "epoch": 0.25, "learning_rate": 0.000878, "loss": 6.9461, "step": 445 }, { "epoch": 0.25, "learning_rate": 0.00088, "loss": 7.2551, "step": 446 }, { "epoch": 0.25, "learning_rate": 0.000882, "loss": 6.994, "step": 447 }, { "epoch": 0.25, "learning_rate": 0.000884, "loss": 6.944, "step": 448 }, { "epoch": 0.25, "learning_rate": 0.0008860000000000001, "loss": 7.1847, "step": 449 }, { "epoch": 0.25, "learning_rate": 0.000888, "loss": 6.4875, "step": 450 }, { "epoch": 0.25, "learning_rate": 0.0008900000000000001, "loss": 7.1992, "step": 451 }, { "epoch": 0.25, "learning_rate": 0.000892, "loss": 7.2476, "step": 452 }, { "epoch": 0.25, "learning_rate": 0.000894, "loss": 7.2239, "step": 453 }, { "epoch": 0.25, "learning_rate": 0.000896, "loss": 7.1001, "step": 454 }, { "epoch": 0.26, "learning_rate": 0.000898, "loss": 7.0063, "step": 455 }, { "epoch": 0.26, "learning_rate": 0.0009000000000000001, "loss": 7.1645, "step": 456 }, { "epoch": 0.26, "learning_rate": 0.000902, "loss": 7.003, "step": 457 }, { "epoch": 0.26, "learning_rate": 0.0009040000000000001, "loss": 7.02, "step": 458 }, { "epoch": 0.26, "learning_rate": 0.000906, "loss": 6.901, "step": 459 }, { "epoch": 0.26, "learning_rate": 0.0009080000000000001, "loss": 7.119, "step": 460 }, { "epoch": 0.26, "learning_rate": 0.00091, "loss": 6.8651, "step": 461 }, { "epoch": 0.26, "learning_rate": 0.000912, "loss": 6.8381, "step": 462 }, { "epoch": 0.26, "learning_rate": 0.0009140000000000001, "loss": 7.0494, "step": 463 }, { "epoch": 0.26, "learning_rate": 0.000916, "loss": 7.1397, "step": 464 }, { "epoch": 0.26, "learning_rate": 0.0009180000000000001, "loss": 6.9244, "step": 465 }, { "epoch": 0.26, "learning_rate": 0.00092, "loss": 7.1848, "step": 466 }, { "epoch": 0.26, "learning_rate": 0.0009220000000000001, "loss": 7.0295, "step": 467 }, { "epoch": 0.26, "learning_rate": 0.000924, "loss": 7.1669, "step": 468 }, { "epoch": 0.26, "learning_rate": 0.0009260000000000001, "loss": 6.86, "step": 469 }, { "epoch": 0.26, "learning_rate": 0.0009280000000000001, "loss": 7.1341, "step": 470 }, { "epoch": 0.26, "learning_rate": 0.00093, "loss": 7.0796, "step": 471 }, { "epoch": 0.26, "learning_rate": 0.0009320000000000001, "loss": 6.8982, "step": 472 }, { "epoch": 0.27, "learning_rate": 0.000934, "loss": 6.8949, "step": 473 }, { "epoch": 0.27, "learning_rate": 0.0009360000000000001, "loss": 6.9946, "step": 474 }, { "epoch": 0.27, "learning_rate": 0.0009379999999999999, "loss": 6.9762, "step": 475 }, { "epoch": 0.27, "learning_rate": 0.00094, "loss": 7.0324, "step": 476 }, { "epoch": 0.27, "learning_rate": 0.000942, "loss": 7.0781, "step": 477 }, { "epoch": 0.27, "learning_rate": 0.000944, "loss": 6.9753, "step": 478 }, { "epoch": 0.27, "learning_rate": 0.000946, "loss": 7.1469, "step": 479 }, { "epoch": 0.27, "learning_rate": 0.000948, "loss": 7.208, "step": 480 }, { "epoch": 0.27, "learning_rate": 0.00095, "loss": 7.2174, "step": 481 }, { "epoch": 0.27, "learning_rate": 0.0009519999999999999, "loss": 7.0258, "step": 482 }, { "epoch": 0.27, "learning_rate": 0.000954, "loss": 7.4387, "step": 483 }, { "epoch": 0.27, "learning_rate": 0.0009559999999999999, "loss": 6.9972, "step": 484 }, { "epoch": 0.27, "learning_rate": 0.000958, "loss": 7.135, "step": 485 }, { "epoch": 0.27, "learning_rate": 0.00096, "loss": 7.1005, "step": 486 }, { "epoch": 0.27, "learning_rate": 0.000962, "loss": 6.9256, "step": 487 }, { "epoch": 0.27, "learning_rate": 0.000964, "loss": 7.2561, "step": 488 }, { "epoch": 0.27, "learning_rate": 0.000966, "loss": 7.0283, "step": 489 }, { "epoch": 0.27, "learning_rate": 0.000968, "loss": 6.9452, "step": 490 }, { "epoch": 0.28, "learning_rate": 0.0009699999999999999, "loss": 6.9076, "step": 491 }, { "epoch": 0.28, "learning_rate": 0.000972, "loss": 7.1704, "step": 492 }, { "epoch": 0.28, "learning_rate": 0.000974, "loss": 7.0958, "step": 493 }, { "epoch": 0.28, "learning_rate": 0.000976, "loss": 7.1794, "step": 494 }, { "epoch": 0.28, "learning_rate": 0.000978, "loss": 6.9961, "step": 495 }, { "epoch": 0.28, "learning_rate": 0.00098, "loss": 7.0463, "step": 496 }, { "epoch": 0.28, "learning_rate": 0.000982, "loss": 6.799, "step": 497 }, { "epoch": 0.28, "learning_rate": 0.000984, "loss": 7.3124, "step": 498 }, { "epoch": 0.28, "learning_rate": 0.0009860000000000001, "loss": 6.7608, "step": 499 }, { "epoch": 0.28, "learning_rate": 0.000988, "loss": 6.9627, "step": 500 }, { "epoch": 0.28, "eval_loss": 7.060001373291016, "eval_runtime": 338.6009, "eval_samples_per_second": 7.803, "eval_steps_per_second": 0.978, "eval_wer": 1.0, "step": 500 }, { "epoch": 0.28, "learning_rate": 0.00099, "loss": 7.1393, "step": 501 }, { "epoch": 0.28, "learning_rate": 0.000992, "loss": 7.1568, "step": 502 }, { "epoch": 0.28, "learning_rate": 0.000994, "loss": 7.046, "step": 503 }, { "epoch": 0.28, "learning_rate": 0.000996, "loss": 7.1271, "step": 504 }, { "epoch": 0.28, "learning_rate": 0.000998, "loss": 6.9393, "step": 505 }, { "epoch": 0.28, "learning_rate": 0.001, "loss": 7.1512, "step": 506 }, { "epoch": 0.28, "learning_rate": 0.0009992211838006232, "loss": 6.9213, "step": 507 }, { "epoch": 0.28, "learning_rate": 0.000998442367601246, "loss": 7.2015, "step": 508 }, { "epoch": 0.29, "learning_rate": 0.0009976635514018692, "loss": 7.2331, "step": 509 }, { "epoch": 0.29, "learning_rate": 0.0009968847352024922, "loss": 7.0947, "step": 510 }, { "epoch": 0.29, "learning_rate": 0.0009961059190031153, "loss": 7.0398, "step": 511 }, { "epoch": 0.29, "learning_rate": 0.0009953271028037385, "loss": 6.9764, "step": 512 }, { "epoch": 0.29, "learning_rate": 0.0009945482866043614, "loss": 6.9555, "step": 513 }, { "epoch": 0.29, "learning_rate": 0.0009937694704049845, "loss": 6.902, "step": 514 }, { "epoch": 0.29, "learning_rate": 0.0009929906542056075, "loss": 7.0866, "step": 515 }, { "epoch": 0.29, "learning_rate": 0.0009922118380062306, "loss": 7.1105, "step": 516 }, { "epoch": 0.29, "learning_rate": 0.0009914330218068535, "loss": 7.0919, "step": 517 }, { "epoch": 0.29, "learning_rate": 0.0009906542056074767, "loss": 7.0919, "step": 518 }, { "epoch": 0.29, "learning_rate": 0.0009898753894080998, "loss": 7.0219, "step": 519 }, { "epoch": 0.29, "learning_rate": 0.0009890965732087227, "loss": 7.0181, "step": 520 }, { "epoch": 0.29, "learning_rate": 0.000988317757009346, "loss": 7.0111, "step": 521 }, { "epoch": 0.29, "learning_rate": 0.0009875389408099688, "loss": 7.1138, "step": 522 }, { "epoch": 0.29, "learning_rate": 0.000986760124610592, "loss": 6.8774, "step": 523 }, { "epoch": 0.29, "learning_rate": 0.0009859813084112151, "loss": 7.1074, "step": 524 }, { "epoch": 0.29, "learning_rate": 0.000985202492211838, "loss": 7.1315, "step": 525 }, { "epoch": 0.29, "learning_rate": 0.0009844236760124612, "loss": 6.9912, "step": 526 }, { "epoch": 0.3, "learning_rate": 0.0009836448598130841, "loss": 7.287, "step": 527 }, { "epoch": 0.3, "learning_rate": 0.0009828660436137073, "loss": 6.8933, "step": 528 }, { "epoch": 0.3, "learning_rate": 0.0009820872274143302, "loss": 7.0932, "step": 529 }, { "epoch": 0.3, "learning_rate": 0.0009813084112149533, "loss": 7.2283, "step": 530 }, { "epoch": 0.3, "learning_rate": 0.0009805295950155765, "loss": 6.848, "step": 531 }, { "epoch": 0.3, "learning_rate": 0.0009797507788161994, "loss": 6.8904, "step": 532 }, { "epoch": 0.3, "learning_rate": 0.0009789719626168223, "loss": 7.0551, "step": 533 }, { "epoch": 0.3, "learning_rate": 0.0009781931464174455, "loss": 6.9034, "step": 534 }, { "epoch": 0.3, "learning_rate": 0.0009774143302180686, "loss": 7.2267, "step": 535 }, { "epoch": 0.3, "learning_rate": 0.0009766355140186918, "loss": 7.0234, "step": 536 }, { "epoch": 0.3, "learning_rate": 0.0009758566978193146, "loss": 6.9111, "step": 537 }, { "epoch": 0.3, "learning_rate": 0.0009750778816199377, "loss": 6.918, "step": 538 }, { "epoch": 0.3, "learning_rate": 0.0009742990654205608, "loss": 7.1326, "step": 539 }, { "epoch": 0.3, "learning_rate": 0.0009735202492211839, "loss": 7.3132, "step": 540 }, { "epoch": 0.3, "learning_rate": 0.0009727414330218068, "loss": 6.92, "step": 541 }, { "epoch": 0.3, "learning_rate": 0.0009719626168224299, "loss": 7.2023, "step": 542 }, { "epoch": 0.3, "learning_rate": 0.000971183800623053, "loss": 6.9374, "step": 543 }, { "epoch": 0.3, "learning_rate": 0.0009704049844236761, "loss": 6.9632, "step": 544 }, { "epoch": 0.31, "learning_rate": 0.0009696261682242991, "loss": 6.84, "step": 545 }, { "epoch": 0.31, "learning_rate": 0.0009688473520249221, "loss": 6.7155, "step": 546 }, { "epoch": 0.31, "learning_rate": 0.0009680685358255452, "loss": 7.1808, "step": 547 }, { "epoch": 0.31, "learning_rate": 0.0009672897196261683, "loss": 6.7672, "step": 548 }, { "epoch": 0.31, "learning_rate": 0.0009665109034267912, "loss": 6.6923, "step": 549 }, { "epoch": 0.31, "learning_rate": 0.0009657320872274144, "loss": 6.4349, "step": 550 }, { "epoch": 0.31, "learning_rate": 0.0009649532710280374, "loss": 7.4589, "step": 551 }, { "epoch": 0.31, "learning_rate": 0.0009641744548286605, "loss": 7.5232, "step": 552 }, { "epoch": 0.31, "learning_rate": 0.0009633956386292835, "loss": 7.0669, "step": 553 }, { "epoch": 0.31, "learning_rate": 0.0009626168224299065, "loss": 7.0735, "step": 554 }, { "epoch": 0.31, "learning_rate": 0.0009618380062305297, "loss": 7.0951, "step": 555 }, { "epoch": 0.31, "learning_rate": 0.0009610591900311527, "loss": 7.1777, "step": 556 }, { "epoch": 0.31, "learning_rate": 0.0009602803738317756, "loss": 7.2054, "step": 557 }, { "epoch": 0.31, "learning_rate": 0.0009595015576323988, "loss": 7.1833, "step": 558 }, { "epoch": 0.31, "learning_rate": 0.0009587227414330218, "loss": 6.8455, "step": 559 }, { "epoch": 0.31, "learning_rate": 0.000957943925233645, "loss": 7.1012, "step": 560 }, { "epoch": 0.31, "learning_rate": 0.0009571651090342679, "loss": 7.1011, "step": 561 }, { "epoch": 0.32, "learning_rate": 0.000956386292834891, "loss": 6.9802, "step": 562 }, { "epoch": 0.32, "learning_rate": 0.0009556074766355141, "loss": 7.0443, "step": 563 }, { "epoch": 0.32, "learning_rate": 0.0009548286604361371, "loss": 7.3513, "step": 564 }, { "epoch": 0.32, "learning_rate": 0.0009540498442367601, "loss": 7.1066, "step": 565 }, { "epoch": 0.32, "learning_rate": 0.0009532710280373832, "loss": 6.9427, "step": 566 }, { "epoch": 0.32, "learning_rate": 0.0009524922118380063, "loss": 7.2158, "step": 567 }, { "epoch": 0.32, "learning_rate": 0.0009517133956386294, "loss": 7.1167, "step": 568 }, { "epoch": 0.32, "learning_rate": 0.0009509345794392523, "loss": 7.1497, "step": 569 }, { "epoch": 0.32, "learning_rate": 0.0009501557632398754, "loss": 7.0561, "step": 570 }, { "epoch": 0.32, "learning_rate": 0.0009493769470404985, "loss": 6.9877, "step": 571 }, { "epoch": 0.32, "learning_rate": 0.0009485981308411216, "loss": 6.9606, "step": 572 }, { "epoch": 0.32, "learning_rate": 0.0009478193146417445, "loss": 6.9436, "step": 573 }, { "epoch": 0.32, "learning_rate": 0.0009470404984423676, "loss": 6.8624, "step": 574 }, { "epoch": 0.32, "learning_rate": 0.0009462616822429907, "loss": 6.9492, "step": 575 }, { "epoch": 0.32, "learning_rate": 0.0009462616822429907, "loss": 3.5224, "step": 576 }, { "epoch": 0.32, "learning_rate": 0.0009462616822429907, "loss": 3.5195, "step": 577 }, { "epoch": 0.32, "learning_rate": 0.0009454828660436138, "loss": 6.832, "step": 578 }, { "epoch": 0.32, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 579 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 580 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 581 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 582 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 583 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 584 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 585 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 586 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 587 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 588 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 589 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 590 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 591 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 592 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 593 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 594 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 595 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 596 }, { "epoch": 0.33, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 597 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 598 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 599 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 600 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 601 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 602 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 603 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 604 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 605 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 606 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 607 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 608 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 609 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 610 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 611 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 612 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 613 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 614 }, { "epoch": 0.34, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 615 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 616 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 617 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 618 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 619 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 620 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 621 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 622 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 623 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 624 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 625 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 626 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 627 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 628 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 629 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 630 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 631 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 632 }, { "epoch": 0.35, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 633 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 634 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 635 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 636 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 637 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 638 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 639 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 640 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 641 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 642 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 643 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 644 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 645 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 646 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 647 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 648 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 649 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 650 }, { "epoch": 0.36, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 651 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 652 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 653 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 654 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 655 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 656 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 657 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 658 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 659 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 660 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 661 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 662 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 663 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 664 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 665 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 666 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 667 }, { "epoch": 0.37, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 668 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 669 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 670 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 671 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 672 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 673 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 674 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 675 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 676 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 677 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 678 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 679 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 680 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 681 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 682 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 683 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 684 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 685 }, { "epoch": 0.38, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 686 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 687 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 688 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 689 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 690 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 691 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 692 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 693 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 694 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 695 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 696 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 697 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 698 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 699 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 700 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 701 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 702 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 703 }, { "epoch": 0.39, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 704 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 705 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 706 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 707 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 708 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 709 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 710 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 711 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 712 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 713 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 714 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 715 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 716 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 717 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 718 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 719 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 720 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 721 }, { "epoch": 0.4, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 722 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 723 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 724 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 725 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 726 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 727 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 728 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 729 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 730 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 731 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 732 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 733 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 734 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 735 }, { "epoch": 0.41, "learning_rate": 0.0009454828660436138, "loss": 0.0, "step": 736 }, { "epoch": 0.41, "learning_rate": 0.0009447040498442368, "loss": 0.0, "step": 737 }, { "epoch": 0.41, "learning_rate": 0.0009439252336448598, "loss": 0.0, "step": 738 }, { "epoch": 0.41, "learning_rate": 0.0009431464174454829, "loss": 0.0, "step": 739 }, { "epoch": 0.41, "learning_rate": 0.000942367601246106, "loss": 0.0, "step": 740 }, { "epoch": 0.42, "learning_rate": 0.0009415887850467289, "loss": 0.0, "step": 741 }, { "epoch": 0.42, "learning_rate": 0.0009408099688473521, "loss": 0.0, "step": 742 }, { "epoch": 0.42, "learning_rate": 0.0009400311526479751, "loss": 0.0, "step": 743 }, { "epoch": 0.42, "learning_rate": 0.0009392523364485982, "loss": 0.0, "step": 744 }, { "epoch": 0.42, "learning_rate": 0.0009384735202492212, "loss": 0.0, "step": 745 }, { "epoch": 0.42, "learning_rate": 0.0009376947040498442, "loss": 0.0, "step": 746 }, { "epoch": 0.42, "learning_rate": 0.0009369158878504674, "loss": 0.0, "step": 747 }, { "epoch": 0.42, "learning_rate": 0.0009361370716510904, "loss": 0.0, "step": 748 }, { "epoch": 0.42, "learning_rate": 0.0009353582554517133, "loss": 0.0, "step": 749 }, { "epoch": 0.42, "learning_rate": 0.0009345794392523365, "loss": 0.0, "step": 750 }, { "epoch": 0.42, "learning_rate": 0.0009338006230529595, "loss": 0.0, "step": 751 }, { "epoch": 0.42, "learning_rate": 0.0009330218068535827, "loss": 0.0, "step": 752 }, { "epoch": 0.42, "learning_rate": 0.0009322429906542056, "loss": 0.0, "step": 753 }, { "epoch": 0.42, "learning_rate": 0.0009314641744548286, "loss": 0.0, "step": 754 }, { "epoch": 0.42, "learning_rate": 0.0009306853582554518, "loss": 0.0, "step": 755 }, { "epoch": 0.42, "learning_rate": 0.0009299065420560748, "loss": 0.0, "step": 756 }, { "epoch": 0.42, "learning_rate": 0.0009291277258566979, "loss": 0.0, "step": 757 }, { "epoch": 0.42, "learning_rate": 0.0009283489096573209, "loss": 0.0, "step": 758 }, { "epoch": 0.43, "learning_rate": 0.0009275700934579439, "loss": 0.0, "step": 759 }, { "epoch": 0.43, "learning_rate": 0.0009267912772585671, "loss": 0.0, "step": 760 }, { "epoch": 0.43, "learning_rate": 0.00092601246105919, "loss": 0.0, "step": 761 }, { "epoch": 0.43, "learning_rate": 0.0009252336448598131, "loss": 0.0, "step": 762 }, { "epoch": 0.43, "learning_rate": 0.0009244548286604362, "loss": 0.0, "step": 763 }, { "epoch": 0.43, "learning_rate": 0.0009236760124610592, "loss": 0.0, "step": 764 }, { "epoch": 0.43, "learning_rate": 0.0009228971962616823, "loss": 0.0, "step": 765 }, { "epoch": 0.43, "learning_rate": 0.0009221183800623053, "loss": 0.0, "step": 766 }, { "epoch": 0.43, "learning_rate": 0.0009213395638629284, "loss": 0.0, "step": 767 }, { "epoch": 0.43, "learning_rate": 0.0009205607476635515, "loss": 0.0, "step": 768 }, { "epoch": 0.43, "learning_rate": 0.0009197819314641744, "loss": 0.0, "step": 769 }, { "epoch": 0.43, "learning_rate": 0.0009190031152647975, "loss": 0.0, "step": 770 }, { "epoch": 0.43, "learning_rate": 0.0009182242990654206, "loss": 0.0, "step": 771 }, { "epoch": 0.43, "learning_rate": 0.0009174454828660437, "loss": 0.0, "step": 772 }, { "epoch": 0.43, "learning_rate": 0.0009166666666666666, "loss": 0.0, "step": 773 }, { "epoch": 0.43, "learning_rate": 0.0009158878504672897, "loss": 0.0, "step": 774 }, { "epoch": 0.43, "learning_rate": 0.0009151090342679128, "loss": 0.0, "step": 775 }, { "epoch": 0.43, "learning_rate": 0.0009143302180685359, "loss": 0.0, "step": 776 }, { "epoch": 0.44, "learning_rate": 0.0009135514018691589, "loss": 0.0, "step": 777 }, { "epoch": 0.44, "learning_rate": 0.0009127725856697819, "loss": 0.0, "step": 778 }, { "epoch": 0.44, "learning_rate": 0.000911993769470405, "loss": 0.0, "step": 779 }, { "epoch": 0.44, "learning_rate": 0.0009112149532710281, "loss": 0.0, "step": 780 }, { "epoch": 0.44, "learning_rate": 0.000910436137071651, "loss": 0.0, "step": 781 }, { "epoch": 0.44, "learning_rate": 0.0009096573208722742, "loss": 0.0, "step": 782 }, { "epoch": 0.44, "learning_rate": 0.0009088785046728972, "loss": 0.0, "step": 783 }, { "epoch": 0.44, "learning_rate": 0.0009080996884735203, "loss": 0.0, "step": 784 }, { "epoch": 0.44, "learning_rate": 0.0009073208722741433, "loss": 0.0, "step": 785 }, { "epoch": 0.44, "learning_rate": 0.0009065420560747663, "loss": 0.0, "step": 786 }, { "epoch": 0.44, "learning_rate": 0.0009057632398753895, "loss": 0.0, "step": 787 }, { "epoch": 0.44, "learning_rate": 0.0009049844236760125, "loss": 0.0, "step": 788 }, { "epoch": 0.44, "learning_rate": 0.0009042056074766354, "loss": 0.0, "step": 789 }, { "epoch": 0.44, "learning_rate": 0.0009034267912772586, "loss": 0.0, "step": 790 }, { "epoch": 0.44, "learning_rate": 0.0009026479750778816, "loss": 0.0, "step": 791 }, { "epoch": 0.44, "learning_rate": 0.0009018691588785048, "loss": 0.0, "step": 792 }, { "epoch": 0.44, "learning_rate": 0.0009010903426791277, "loss": 0.0, "step": 793 }, { "epoch": 0.45, "learning_rate": 0.0009003115264797507, "loss": 0.0, "step": 794 }, { "epoch": 0.45, "learning_rate": 0.0008995327102803739, "loss": 0.0, "step": 795 }, { "epoch": 0.45, "learning_rate": 0.0008987538940809969, "loss": 0.0, "step": 796 }, { "epoch": 0.45, "learning_rate": 0.00089797507788162, "loss": 0.0, "step": 797 }, { "epoch": 0.45, "learning_rate": 0.000897196261682243, "loss": 0.0, "step": 798 }, { "epoch": 0.45, "learning_rate": 0.000896417445482866, "loss": 0.0, "step": 799 }, { "epoch": 0.45, "learning_rate": 0.0008956386292834892, "loss": 0.0, "step": 800 }, { "epoch": 0.45, "learning_rate": 0.0008948598130841121, "loss": 0.0, "step": 801 }, { "epoch": 0.45, "learning_rate": 0.0008940809968847352, "loss": 0.0, "step": 802 }, { "epoch": 0.45, "learning_rate": 0.0008933021806853583, "loss": 0.0, "step": 803 }, { "epoch": 0.45, "learning_rate": 0.0008925233644859813, "loss": 0.0, "step": 804 }, { "epoch": 0.45, "learning_rate": 0.0008917445482866044, "loss": 0.0, "step": 805 }, { "epoch": 0.45, "learning_rate": 0.0008909657320872274, "loss": 0.0, "step": 806 }, { "epoch": 0.45, "learning_rate": 0.0008901869158878505, "loss": 0.0, "step": 807 }, { "epoch": 0.45, "learning_rate": 0.0008894080996884736, "loss": 0.0, "step": 808 }, { "epoch": 0.45, "learning_rate": 0.0008886292834890965, "loss": 0.0, "step": 809 }, { "epoch": 0.45, "learning_rate": 0.0008878504672897196, "loss": 0.0, "step": 810 }, { "epoch": 0.45, "learning_rate": 0.0008870716510903427, "loss": 0.0, "step": 811 }, { "epoch": 0.46, "learning_rate": 0.0008862928348909658, "loss": 0.0, "step": 812 }, { "epoch": 0.46, "learning_rate": 0.0008855140186915888, "loss": 0.0, "step": 813 }, { "epoch": 0.46, "learning_rate": 0.0008847352024922118, "loss": 0.0, "step": 814 }, { "epoch": 0.46, "learning_rate": 0.0008839563862928349, "loss": 0.0, "step": 815 }, { "epoch": 0.46, "learning_rate": 0.000883177570093458, "loss": 0.0, "step": 816 }, { "epoch": 0.46, "learning_rate": 0.000882398753894081, "loss": 0.0, "step": 817 }, { "epoch": 0.46, "learning_rate": 0.000881619937694704, "loss": 0.0, "step": 818 }, { "epoch": 0.46, "learning_rate": 0.0008808411214953271, "loss": 0.0, "step": 819 }, { "epoch": 0.46, "learning_rate": 0.0008800623052959502, "loss": 0.0, "step": 820 }, { "epoch": 0.46, "learning_rate": 0.0008792834890965732, "loss": 0.0, "step": 821 }, { "epoch": 0.46, "learning_rate": 0.0008785046728971963, "loss": 0.0, "step": 822 }, { "epoch": 0.46, "learning_rate": 0.0008777258566978193, "loss": 0.0, "step": 823 }, { "epoch": 0.46, "learning_rate": 0.0008769470404984425, "loss": 0.0, "step": 824 }, { "epoch": 0.46, "learning_rate": 0.0008761682242990654, "loss": 0.0, "step": 825 }, { "epoch": 0.46, "learning_rate": 0.0008753894080996884, "loss": 0.0, "step": 826 }, { "epoch": 0.46, "learning_rate": 0.0008746105919003116, "loss": 0.0, "step": 827 }, { "epoch": 0.46, "learning_rate": 0.0008738317757009346, "loss": 0.0, "step": 828 }, { "epoch": 0.46, "learning_rate": 0.0008730529595015576, "loss": 0.0, "step": 829 }, { "epoch": 0.47, "learning_rate": 0.0008722741433021807, "loss": 0.0, "step": 830 }, { "epoch": 0.47, "learning_rate": 0.0008714953271028037, "loss": 0.0, "step": 831 }, { "epoch": 0.47, "learning_rate": 0.0008707165109034269, "loss": 0.0, "step": 832 }, { "epoch": 0.47, "learning_rate": 0.0008699376947040498, "loss": 0.0, "step": 833 }, { "epoch": 0.47, "learning_rate": 0.000869158878504673, "loss": 0.0, "step": 834 }, { "epoch": 0.47, "learning_rate": 0.000868380062305296, "loss": 0.0, "step": 835 }, { "epoch": 0.47, "learning_rate": 0.000867601246105919, "loss": 0.0, "step": 836 }, { "epoch": 0.47, "learning_rate": 0.0008668224299065421, "loss": 0.0, "step": 837 }, { "epoch": 0.47, "learning_rate": 0.0008660436137071651, "loss": 0.0, "step": 838 }, { "epoch": 0.47, "learning_rate": 0.0008652647975077882, "loss": 0.0, "step": 839 }, { "epoch": 0.47, "learning_rate": 0.0008644859813084113, "loss": 0.0, "step": 840 }, { "epoch": 0.47, "learning_rate": 0.0008637071651090342, "loss": 0.0, "step": 841 }, { "epoch": 0.47, "learning_rate": 0.0008629283489096574, "loss": 0.0, "step": 842 }, { "epoch": 0.47, "learning_rate": 0.0008621495327102804, "loss": 0.0, "step": 843 }, { "epoch": 0.47, "learning_rate": 0.0008613707165109035, "loss": 0.0, "step": 844 }, { "epoch": 0.47, "learning_rate": 0.0008605919003115265, "loss": 0.0, "step": 845 }, { "epoch": 0.47, "learning_rate": 0.0008598130841121495, "loss": 0.0, "step": 846 }, { "epoch": 0.47, "learning_rate": 0.0008590342679127726, "loss": 0.0, "step": 847 }, { "epoch": 0.48, "learning_rate": 0.0008582554517133957, "loss": 0.0, "step": 848 }, { "epoch": 0.48, "learning_rate": 0.0008574766355140187, "loss": 0.0, "step": 849 }, { "epoch": 0.48, "learning_rate": 0.0008566978193146418, "loss": 0.0, "step": 850 }, { "epoch": 0.48, "learning_rate": 0.0008559190031152648, "loss": 0.0, "step": 851 }, { "epoch": 0.48, "learning_rate": 0.0008551401869158879, "loss": 0.0, "step": 852 }, { "epoch": 0.48, "learning_rate": 0.0008543613707165109, "loss": 0.0, "step": 853 }, { "epoch": 0.48, "learning_rate": 0.000853582554517134, "loss": 0.0, "step": 854 }, { "epoch": 0.48, "learning_rate": 0.000852803738317757, "loss": 0.0, "step": 855 }, { "epoch": 0.48, "learning_rate": 0.0008520249221183801, "loss": 0.0, "step": 856 }, { "epoch": 0.48, "learning_rate": 0.0008512461059190031, "loss": 0.0, "step": 857 }, { "epoch": 0.48, "learning_rate": 0.0008504672897196262, "loss": 0.0, "step": 858 }, { "epoch": 0.48, "learning_rate": 0.0008496884735202493, "loss": 0.0, "step": 859 }, { "epoch": 0.48, "learning_rate": 0.0008489096573208723, "loss": 0.0, "step": 860 }, { "epoch": 0.48, "learning_rate": 0.0008481308411214953, "loss": 0.0, "step": 861 }, { "epoch": 0.48, "learning_rate": 0.0008473520249221184, "loss": 0.0, "step": 862 }, { "epoch": 0.48, "learning_rate": 0.0008465732087227414, "loss": 0.0, "step": 863 }, { "epoch": 0.48, "learning_rate": 0.0008457943925233646, "loss": 0.0, "step": 864 }, { "epoch": 0.48, "learning_rate": 0.0008450155763239875, "loss": 0.0, "step": 865 }, { "epoch": 0.49, "learning_rate": 0.0008442367601246106, "loss": 0.0, "step": 866 }, { "epoch": 0.49, "learning_rate": 0.0008434579439252337, "loss": 0.0, "step": 867 }, { "epoch": 0.49, "learning_rate": 0.0008426791277258567, "loss": 0.0, "step": 868 }, { "epoch": 0.49, "learning_rate": 0.0008419003115264798, "loss": 0.0, "step": 869 }, { "epoch": 0.49, "learning_rate": 0.0008411214953271028, "loss": 0.0, "step": 870 }, { "epoch": 0.49, "learning_rate": 0.0008403426791277258, "loss": 0.0, "step": 871 }, { "epoch": 0.49, "learning_rate": 0.000839563862928349, "loss": 0.0, "step": 872 }, { "epoch": 0.49, "learning_rate": 0.0008387850467289719, "loss": 0.0, "step": 873 }, { "epoch": 0.49, "learning_rate": 0.0008380062305295951, "loss": 0.0, "step": 874 }, { "epoch": 0.49, "learning_rate": 0.0008372274143302181, "loss": 0.0, "step": 875 }, { "epoch": 0.49, "learning_rate": 0.0008364485981308411, "loss": 0.0, "step": 876 }, { "epoch": 0.49, "learning_rate": 0.0008356697819314642, "loss": 0.0, "step": 877 }, { "epoch": 0.49, "learning_rate": 0.0008348909657320872, "loss": 0.0, "step": 878 }, { "epoch": 0.49, "learning_rate": 0.0008341121495327104, "loss": 0.0, "step": 879 }, { "epoch": 0.49, "learning_rate": 0.0008333333333333334, "loss": 0.0, "step": 880 }, { "epoch": 0.49, "learning_rate": 0.0008325545171339563, "loss": 0.0, "step": 881 }, { "epoch": 0.49, "learning_rate": 0.0008317757009345795, "loss": 0.0, "step": 882 }, { "epoch": 0.49, "learning_rate": 0.0008309968847352025, "loss": 0.0, "step": 883 }, { "epoch": 0.5, "learning_rate": 0.0008302180685358256, "loss": 0.0, "step": 884 }, { "epoch": 0.5, "learning_rate": 0.0008294392523364486, "loss": 0.0, "step": 885 }, { "epoch": 0.5, "learning_rate": 0.0008286604361370716, "loss": 0.0, "step": 886 }, { "epoch": 0.5, "learning_rate": 0.0008278816199376947, "loss": 0.0, "step": 887 }, { "epoch": 0.5, "learning_rate": 0.0008271028037383178, "loss": 0.0, "step": 888 }, { "epoch": 0.5, "learning_rate": 0.0008263239875389408, "loss": 0.0, "step": 889 }, { "epoch": 0.5, "learning_rate": 0.0008255451713395639, "loss": 0.0, "step": 890 }, { "epoch": 0.5, "learning_rate": 0.0008247663551401869, "loss": 0.0, "step": 891 }, { "epoch": 0.5, "learning_rate": 0.00082398753894081, "loss": 0.0, "step": 892 }, { "epoch": 0.5, "learning_rate": 0.000823208722741433, "loss": 0.0, "step": 893 }, { "epoch": 0.5, "learning_rate": 0.0008224299065420561, "loss": 0.0, "step": 894 }, { "epoch": 0.5, "learning_rate": 0.0008216510903426791, "loss": 0.0, "step": 895 }, { "epoch": 0.5, "learning_rate": 0.0008208722741433022, "loss": 0.0, "step": 896 }, { "epoch": 0.5, "learning_rate": 0.0008200934579439252, "loss": 0.0, "step": 897 }, { "epoch": 0.5, "learning_rate": 0.0008193146417445483, "loss": 0.0, "step": 898 }, { "epoch": 0.5, "learning_rate": 0.0008185358255451714, "loss": 0.0, "step": 899 }, { "epoch": 0.5, "learning_rate": 0.0008177570093457944, "loss": 0.0, "step": 900 }, { "epoch": 0.51, "learning_rate": 0.0008169781931464174, "loss": 0.0, "step": 901 }, { "epoch": 0.51, "learning_rate": 0.0008161993769470405, "loss": 0.0, "step": 902 }, { "epoch": 0.51, "learning_rate": 0.0008154205607476635, "loss": 0.0, "step": 903 }, { "epoch": 0.51, "learning_rate": 0.0008146417445482867, "loss": 0.0, "step": 904 }, { "epoch": 0.51, "learning_rate": 0.0008138629283489096, "loss": 0.0, "step": 905 }, { "epoch": 0.51, "learning_rate": 0.0008130841121495327, "loss": 0.0, "step": 906 }, { "epoch": 0.51, "learning_rate": 0.0008123052959501558, "loss": 0.0, "step": 907 }, { "epoch": 0.51, "learning_rate": 0.0008115264797507788, "loss": 0.0, "step": 908 }, { "epoch": 0.51, "learning_rate": 0.0008107476635514019, "loss": 0.0, "step": 909 }, { "epoch": 0.51, "learning_rate": 0.0008099688473520249, "loss": 0.0, "step": 910 }, { "epoch": 0.51, "learning_rate": 0.000809190031152648, "loss": 0.0, "step": 911 }, { "epoch": 0.51, "learning_rate": 0.0008084112149532711, "loss": 0.0, "step": 912 }, { "epoch": 0.51, "learning_rate": 0.000807632398753894, "loss": 0.0, "step": 913 }, { "epoch": 0.51, "learning_rate": 0.0008068535825545172, "loss": 0.0, "step": 914 }, { "epoch": 0.51, "learning_rate": 0.0008060747663551402, "loss": 0.0, "step": 915 }, { "epoch": 0.51, "learning_rate": 0.0008052959501557632, "loss": 0.0, "step": 916 }, { "epoch": 0.51, "learning_rate": 0.0008045171339563863, "loss": 0.0, "step": 917 }, { "epoch": 0.51, "learning_rate": 0.0008037383177570093, "loss": 0.0, "step": 918 }, { "epoch": 0.52, "learning_rate": 0.0008029595015576325, "loss": 0.0, "step": 919 }, { "epoch": 0.52, "learning_rate": 0.0008021806853582555, "loss": 0.0, "step": 920 }, { "epoch": 0.52, "learning_rate": 0.0008014018691588784, "loss": 0.0, "step": 921 }, { "epoch": 0.52, "learning_rate": 0.0008006230529595016, "loss": 0.0, "step": 922 }, { "epoch": 0.52, "learning_rate": 0.0007998442367601246, "loss": 0.0, "step": 923 }, { "epoch": 0.52, "learning_rate": 0.0007990654205607477, "loss": 0.0, "step": 924 }, { "epoch": 0.52, "learning_rate": 0.0007982866043613707, "loss": 0.0, "step": 925 }, { "epoch": 0.52, "learning_rate": 0.0007975077881619937, "loss": 0.0, "step": 926 }, { "epoch": 0.52, "learning_rate": 0.0007967289719626169, "loss": 0.0, "step": 927 }, { "epoch": 0.52, "learning_rate": 0.0007959501557632399, "loss": 0.0, "step": 928 }, { "epoch": 0.52, "learning_rate": 0.0007951713395638629, "loss": 0.0, "step": 929 }, { "epoch": 0.52, "learning_rate": 0.000794392523364486, "loss": 0.0, "step": 930 }, { "epoch": 0.52, "learning_rate": 0.000793613707165109, "loss": 0.0, "step": 931 }, { "epoch": 0.52, "learning_rate": 0.0007928348909657321, "loss": 0.0, "step": 932 }, { "epoch": 0.52, "learning_rate": 0.0007920560747663551, "loss": 0.0, "step": 933 }, { "epoch": 0.52, "learning_rate": 0.0007912772585669782, "loss": 0.0, "step": 934 }, { "epoch": 0.52, "learning_rate": 0.0007904984423676013, "loss": 0.0, "step": 935 }, { "epoch": 0.52, "learning_rate": 0.0007897196261682244, "loss": 0.0, "step": 936 }, { "epoch": 0.53, "learning_rate": 0.0007889408099688473, "loss": 0.0, "step": 937 }, { "epoch": 0.53, "learning_rate": 0.0007881619937694704, "loss": 0.0, "step": 938 }, { "epoch": 0.53, "learning_rate": 0.0007873831775700935, "loss": 0.0, "step": 939 }, { "epoch": 0.53, "learning_rate": 0.0007866043613707165, "loss": 0.0, "step": 940 }, { "epoch": 0.53, "learning_rate": 0.0007858255451713396, "loss": 0.0, "step": 941 }, { "epoch": 0.53, "learning_rate": 0.0007850467289719626, "loss": 0.0, "step": 942 }, { "epoch": 0.53, "learning_rate": 0.0007842679127725857, "loss": 0.0, "step": 943 }, { "epoch": 0.53, "learning_rate": 0.0007834890965732088, "loss": 0.0, "step": 944 }, { "epoch": 0.53, "learning_rate": 0.0007827102803738317, "loss": 0.0, "step": 945 }, { "epoch": 0.53, "learning_rate": 0.0007819314641744549, "loss": 0.0, "step": 946 }, { "epoch": 0.53, "learning_rate": 0.0007811526479750779, "loss": 0.0, "step": 947 }, { "epoch": 0.53, "learning_rate": 0.0007803738317757009, "loss": 0.0, "step": 948 }, { "epoch": 0.53, "learning_rate": 0.000779595015576324, "loss": 0.0, "step": 949 }, { "epoch": 0.53, "learning_rate": 0.000778816199376947, "loss": 0.0, "step": 950 }, { "epoch": 0.53, "learning_rate": 0.0007780373831775702, "loss": 0.0, "step": 951 }, { "epoch": 0.53, "learning_rate": 0.0007772585669781932, "loss": 0.0, "step": 952 }, { "epoch": 0.53, "learning_rate": 0.0007764797507788161, "loss": 0.0, "step": 953 }, { "epoch": 0.53, "learning_rate": 0.0007757009345794393, "loss": 0.0, "step": 954 }, { "epoch": 0.54, "learning_rate": 0.0007749221183800623, "loss": 0.0, "step": 955 }, { "epoch": 0.54, "learning_rate": 0.0007741433021806855, "loss": 0.0, "step": 956 }, { "epoch": 0.54, "learning_rate": 0.0007733644859813084, "loss": 0.0, "step": 957 }, { "epoch": 0.54, "learning_rate": 0.0007725856697819314, "loss": 0.0, "step": 958 }, { "epoch": 0.54, "learning_rate": 0.0007718068535825546, "loss": 0.0, "step": 959 }, { "epoch": 0.54, "learning_rate": 0.0007710280373831776, "loss": 0.0, "step": 960 }, { "epoch": 0.54, "learning_rate": 0.0007702492211838006, "loss": 0.0, "step": 961 }, { "epoch": 0.54, "learning_rate": 0.0007694704049844237, "loss": 0.0, "step": 962 }, { "epoch": 0.54, "learning_rate": 0.0007686915887850467, "loss": 0.0, "step": 963 }, { "epoch": 0.54, "learning_rate": 0.0007679127725856699, "loss": 0.0, "step": 964 }, { "epoch": 0.54, "learning_rate": 0.0007671339563862928, "loss": 0.0, "step": 965 }, { "epoch": 0.54, "learning_rate": 0.0007663551401869159, "loss": 0.0, "step": 966 }, { "epoch": 0.54, "learning_rate": 0.000765576323987539, "loss": 0.0, "step": 967 }, { "epoch": 0.54, "learning_rate": 0.000764797507788162, "loss": 0.0, "step": 968 }, { "epoch": 0.54, "learning_rate": 0.000764018691588785, "loss": 0.0, "step": 969 }, { "epoch": 0.54, "learning_rate": 0.0007632398753894081, "loss": 0.0, "step": 970 }, { "epoch": 0.54, "learning_rate": 0.0007624610591900312, "loss": 0.0, "step": 971 }, { "epoch": 0.54, "learning_rate": 0.0007616822429906543, "loss": 0.0, "step": 972 }, { "epoch": 0.55, "learning_rate": 0.0007609034267912772, "loss": 0.0, "step": 973 }, { "epoch": 0.55, "learning_rate": 0.0007601246105919003, "loss": 0.0, "step": 974 }, { "epoch": 0.55, "learning_rate": 0.0007593457943925234, "loss": 0.0, "step": 975 }, { "epoch": 0.55, "learning_rate": 0.0007585669781931465, "loss": 0.0, "step": 976 }, { "epoch": 0.55, "learning_rate": 0.0007577881619937694, "loss": 0.0, "step": 977 }, { "epoch": 0.55, "learning_rate": 0.0007570093457943925, "loss": 0.0, "step": 978 }, { "epoch": 0.55, "learning_rate": 0.0007562305295950156, "loss": 0.0, "step": 979 }, { "epoch": 0.55, "learning_rate": 0.0007554517133956387, "loss": 0.0, "step": 980 }, { "epoch": 0.55, "learning_rate": 0.0007546728971962617, "loss": 0.0, "step": 981 }, { "epoch": 0.55, "learning_rate": 0.0007538940809968847, "loss": 0.0, "step": 982 }, { "epoch": 0.55, "learning_rate": 0.0007531152647975078, "loss": 0.0, "step": 983 }, { "epoch": 0.55, "learning_rate": 0.0007523364485981309, "loss": 0.0, "step": 984 }, { "epoch": 0.55, "learning_rate": 0.0007515576323987538, "loss": 0.0, "step": 985 }, { "epoch": 0.55, "learning_rate": 0.000750778816199377, "loss": 0.0, "step": 986 }, { "epoch": 0.55, "learning_rate": 0.00075, "loss": 0.0, "step": 987 }, { "epoch": 0.55, "learning_rate": 0.000749221183800623, "loss": 0.0, "step": 988 }, { "epoch": 0.55, "learning_rate": 0.0007484423676012462, "loss": 0.0, "step": 989 }, { "epoch": 0.55, "learning_rate": 0.0007476635514018691, "loss": 0.0, "step": 990 }, { "epoch": 0.56, "learning_rate": 0.0007468847352024923, "loss": 0.0, "step": 991 }, { "epoch": 0.56, "learning_rate": 0.0007461059190031153, "loss": 0.0, "step": 992 }, { "epoch": 0.56, "learning_rate": 0.0007453271028037383, "loss": 0.0, "step": 993 }, { "epoch": 0.56, "learning_rate": 0.0007445482866043614, "loss": 0.0, "step": 994 }, { "epoch": 0.56, "learning_rate": 0.0007437694704049844, "loss": 0.0, "step": 995 }, { "epoch": 0.56, "learning_rate": 0.0007429906542056076, "loss": 0.0, "step": 996 }, { "epoch": 0.56, "learning_rate": 0.0007422118380062306, "loss": 0.0, "step": 997 }, { "epoch": 0.56, "learning_rate": 0.0007414330218068535, "loss": 0.0, "step": 998 }, { "epoch": 0.56, "learning_rate": 0.0007406542056074767, "loss": 0.0, "step": 999 }, { "epoch": 0.56, "learning_rate": 0.0007398753894080997, "loss": 0.0, "step": 1000 }, { "epoch": 0.56, "eval_loss": NaN, "eval_runtime": 908.6669, "eval_samples_per_second": 2.908, "eval_steps_per_second": 0.364, "eval_wer": 2.0097029585094353, "step": 1000 }, { "epoch": 0.56, "learning_rate": 0.0007390965732087229, "loss": 0.0, "step": 1001 }, { "epoch": 0.56, "learning_rate": 0.0007383177570093458, "loss": 0.0, "step": 1002 }, { "epoch": 0.56, "learning_rate": 0.0007375389408099688, "loss": 0.0, "step": 1003 }, { "epoch": 0.56, "learning_rate": 0.000736760124610592, "loss": 0.0, "step": 1004 }, { "epoch": 0.56, "learning_rate": 0.000735981308411215, "loss": 0.0, "step": 1005 }, { "epoch": 0.56, "learning_rate": 0.000735202492211838, "loss": 0.0, "step": 1006 }, { "epoch": 0.56, "learning_rate": 0.0007344236760124611, "loss": 0.0, "step": 1007 }, { "epoch": 0.57, "learning_rate": 0.0007336448598130841, "loss": 0.0, "step": 1008 }, { "epoch": 0.57, "learning_rate": 0.0007328660436137072, "loss": 0.0, "step": 1009 }, { "epoch": 0.57, "learning_rate": 0.0007320872274143302, "loss": 0.0, "step": 1010 }, { "epoch": 0.57, "learning_rate": 0.0007313084112149533, "loss": 0.0, "step": 1011 }, { "epoch": 0.57, "learning_rate": 0.0007305295950155764, "loss": 0.0, "step": 1012 }, { "epoch": 0.57, "learning_rate": 0.0007297507788161994, "loss": 0.0, "step": 1013 }, { "epoch": 0.57, "learning_rate": 0.0007289719626168224, "loss": 0.0, "step": 1014 }, { "epoch": 0.57, "learning_rate": 0.0007281931464174455, "loss": 0.0, "step": 1015 }, { "epoch": 0.57, "learning_rate": 0.0007274143302180686, "loss": 0.0, "step": 1016 }, { "epoch": 0.57, "learning_rate": 0.0007266355140186916, "loss": 0.0, "step": 1017 }, { "epoch": 0.57, "learning_rate": 0.0007258566978193146, "loss": 0.0, "step": 1018 }, { "epoch": 0.57, "learning_rate": 0.0007250778816199377, "loss": 0.0, "step": 1019 }, { "epoch": 0.57, "learning_rate": 0.0007242990654205608, "loss": 0.0, "step": 1020 }, { "epoch": 0.57, "learning_rate": 0.0007235202492211839, "loss": 0.0, "step": 1021 }, { "epoch": 0.57, "learning_rate": 0.0007227414330218068, "loss": 0.0, "step": 1022 }, { "epoch": 0.57, "learning_rate": 0.0007219626168224299, "loss": 0.0, "step": 1023 }, { "epoch": 0.57, "learning_rate": 0.000721183800623053, "loss": 0.0, "step": 1024 }, { "epoch": 0.57, "learning_rate": 0.000720404984423676, "loss": 0.0, "step": 1025 }, { "epoch": 0.58, "learning_rate": 0.0007196261682242991, "loss": 0.0, "step": 1026 }, { "epoch": 0.58, "learning_rate": 0.0007188473520249221, "loss": 0.0, "step": 1027 }, { "epoch": 0.58, "learning_rate": 0.0007180685358255452, "loss": 0.0, "step": 1028 }, { "epoch": 0.58, "learning_rate": 0.0007172897196261683, "loss": 0.0, "step": 1029 }, { "epoch": 0.58, "learning_rate": 0.0007165109034267912, "loss": 0.0, "step": 1030 }, { "epoch": 0.58, "learning_rate": 0.0007157320872274144, "loss": 0.0, "step": 1031 }, { "epoch": 0.58, "learning_rate": 0.0007149532710280374, "loss": 0.0, "step": 1032 }, { "epoch": 0.58, "learning_rate": 0.0007141744548286604, "loss": 0.0, "step": 1033 }, { "epoch": 0.58, "learning_rate": 0.0007133956386292835, "loss": 0.0, "step": 1034 }, { "epoch": 0.58, "learning_rate": 0.0007126168224299065, "loss": 0.0, "step": 1035 }, { "epoch": 0.58, "learning_rate": 0.0007118380062305297, "loss": 0.0, "step": 1036 }, { "epoch": 0.58, "learning_rate": 0.0007110591900311527, "loss": 0.0, "step": 1037 }, { "epoch": 0.58, "learning_rate": 0.0007102803738317756, "loss": 0.0, "step": 1038 }, { "epoch": 0.58, "learning_rate": 0.0007095015576323988, "loss": 0.0, "step": 1039 }, { "epoch": 0.58, "learning_rate": 0.0007087227414330218, "loss": 0.0, "step": 1040 }, { "epoch": 0.58, "learning_rate": 0.000707943925233645, "loss": 0.0, "step": 1041 }, { "epoch": 0.58, "learning_rate": 0.0007071651090342679, "loss": 0.0, "step": 1042 }, { "epoch": 0.58, "learning_rate": 0.000706386292834891, "loss": 0.0, "step": 1043 }, { "epoch": 0.59, "learning_rate": 0.0007056074766355141, "loss": 0.0, "step": 1044 }, { "epoch": 0.59, "learning_rate": 0.0007048286604361371, "loss": 0.0, "step": 1045 }, { "epoch": 0.59, "learning_rate": 0.0007040498442367601, "loss": 0.0, "step": 1046 }, { "epoch": 0.59, "learning_rate": 0.0007032710280373832, "loss": 0.0, "step": 1047 }, { "epoch": 0.59, "learning_rate": 0.0007024922118380063, "loss": 0.0, "step": 1048 }, { "epoch": 0.59, "learning_rate": 0.0007017133956386294, "loss": 0.0, "step": 1049 }, { "epoch": 0.59, "learning_rate": 0.0007009345794392523, "loss": 0.0, "step": 1050 }, { "epoch": 0.59, "learning_rate": 0.0007001557632398754, "loss": 0.0, "step": 1051 }, { "epoch": 0.59, "learning_rate": 0.0006993769470404985, "loss": 0.0, "step": 1052 }, { "epoch": 0.59, "learning_rate": 0.0006985981308411216, "loss": 0.0, "step": 1053 }, { "epoch": 0.59, "learning_rate": 0.0006978193146417445, "loss": 0.0, "step": 1054 }, { "epoch": 0.59, "learning_rate": 0.0006970404984423676, "loss": 0.0, "step": 1055 }, { "epoch": 0.59, "learning_rate": 0.0006962616822429907, "loss": 0.0, "step": 1056 }, { "epoch": 0.59, "learning_rate": 0.0006954828660436138, "loss": 0.0, "step": 1057 }, { "epoch": 0.59, "learning_rate": 0.0006947040498442368, "loss": 0.0, "step": 1058 }, { "epoch": 0.59, "learning_rate": 0.0006939252336448598, "loss": 0.0, "step": 1059 }, { "epoch": 0.59, "learning_rate": 0.0006931464174454829, "loss": 0.0, "step": 1060 }, { "epoch": 0.59, "learning_rate": 0.000692367601246106, "loss": 0.0, "step": 1061 }, { "epoch": 0.6, "learning_rate": 0.0006915887850467289, "loss": 0.0, "step": 1062 }, { "epoch": 0.6, "learning_rate": 0.0006908099688473521, "loss": 0.0, "step": 1063 }, { "epoch": 0.6, "learning_rate": 0.0006900311526479751, "loss": 0.0, "step": 1064 }, { "epoch": 0.6, "learning_rate": 0.0006892523364485982, "loss": 0.0, "step": 1065 }, { "epoch": 0.6, "learning_rate": 0.0006884735202492212, "loss": 0.0, "step": 1066 }, { "epoch": 0.6, "learning_rate": 0.0006876947040498442, "loss": 0.0, "step": 1067 }, { "epoch": 0.6, "learning_rate": 0.0006869158878504674, "loss": 0.0, "step": 1068 }, { "epoch": 0.6, "learning_rate": 0.0006861370716510904, "loss": 0.0, "step": 1069 }, { "epoch": 0.6, "learning_rate": 0.0006853582554517133, "loss": 0.0, "step": 1070 }, { "epoch": 0.6, "learning_rate": 0.0006845794392523365, "loss": 0.0, "step": 1071 }, { "epoch": 0.6, "learning_rate": 0.0006838006230529595, "loss": 0.0, "step": 1072 }, { "epoch": 0.6, "learning_rate": 0.0006830218068535827, "loss": 0.0, "step": 1073 }, { "epoch": 0.6, "learning_rate": 0.0006822429906542056, "loss": 0.0, "step": 1074 }, { "epoch": 0.6, "learning_rate": 0.0006814641744548286, "loss": 0.0, "step": 1075 }, { "epoch": 0.6, "learning_rate": 0.0006806853582554518, "loss": 0.0, "step": 1076 }, { "epoch": 0.6, "learning_rate": 0.0006799065420560748, "loss": 0.0, "step": 1077 }, { "epoch": 0.6, "learning_rate": 0.0006791277258566978, "loss": 0.0, "step": 1078 }, { "epoch": 0.6, "learning_rate": 0.0006783489096573209, "loss": 0.0, "step": 1079 }, { "epoch": 0.61, "learning_rate": 0.0006775700934579439, "loss": 0.0, "step": 1080 }, { "epoch": 0.61, "learning_rate": 0.0006767912772585671, "loss": 0.0, "step": 1081 }, { "epoch": 0.61, "learning_rate": 0.00067601246105919, "loss": 0.0, "step": 1082 }, { "epoch": 0.61, "learning_rate": 0.0006752336448598131, "loss": 0.0, "step": 1083 }, { "epoch": 0.61, "learning_rate": 0.0006744548286604362, "loss": 0.0, "step": 1084 }, { "epoch": 0.61, "learning_rate": 0.0006736760124610592, "loss": 0.0, "step": 1085 }, { "epoch": 0.61, "learning_rate": 0.0006728971962616822, "loss": 0.0, "step": 1086 }, { "epoch": 0.61, "learning_rate": 0.0006721183800623053, "loss": 0.0, "step": 1087 }, { "epoch": 0.61, "learning_rate": 0.0006713395638629284, "loss": 0.0, "step": 1088 }, { "epoch": 0.61, "learning_rate": 0.0006705607476635515, "loss": 0.0, "step": 1089 }, { "epoch": 0.61, "learning_rate": 0.0006697819314641744, "loss": 0.0, "step": 1090 }, { "epoch": 0.61, "learning_rate": 0.0006690031152647975, "loss": 0.0, "step": 1091 }, { "epoch": 0.61, "learning_rate": 0.0006682242990654206, "loss": 0.0, "step": 1092 }, { "epoch": 0.61, "learning_rate": 0.0006674454828660437, "loss": 0.0, "step": 1093 }, { "epoch": 0.61, "learning_rate": 0.0006666666666666666, "loss": 0.0, "step": 1094 }, { "epoch": 0.61, "learning_rate": 0.0006658878504672897, "loss": 0.0, "step": 1095 }, { "epoch": 0.61, "learning_rate": 0.0006651090342679128, "loss": 0.0, "step": 1096 }, { "epoch": 0.61, "learning_rate": 0.0006643302180685359, "loss": 0.0, "step": 1097 }, { "epoch": 0.62, "learning_rate": 0.0006635514018691589, "loss": 0.0, "step": 1098 }, { "epoch": 0.62, "learning_rate": 0.0006627725856697819, "loss": 0.0, "step": 1099 }, { "epoch": 0.62, "learning_rate": 0.000661993769470405, "loss": 0.0, "step": 1100 }, { "epoch": 0.62, "learning_rate": 0.0006612149532710281, "loss": 0.0, "step": 1101 }, { "epoch": 0.62, "learning_rate": 0.000660436137071651, "loss": 0.0, "step": 1102 }, { "epoch": 0.62, "learning_rate": 0.0006596573208722742, "loss": 0.0, "step": 1103 }, { "epoch": 0.62, "learning_rate": 0.0006588785046728972, "loss": 0.0, "step": 1104 }, { "epoch": 0.62, "learning_rate": 0.0006580996884735203, "loss": 0.0, "step": 1105 }, { "epoch": 0.62, "learning_rate": 0.0006573208722741433, "loss": 0.0, "step": 1106 }, { "epoch": 0.62, "learning_rate": 0.0006565420560747663, "loss": 0.0, "step": 1107 }, { "epoch": 0.62, "learning_rate": 0.0006557632398753895, "loss": 0.0, "step": 1108 }, { "epoch": 0.62, "learning_rate": 0.0006549844236760125, "loss": 0.0, "step": 1109 }, { "epoch": 0.62, "learning_rate": 0.0006542056074766354, "loss": 0.0, "step": 1110 }, { "epoch": 0.62, "learning_rate": 0.0006534267912772586, "loss": 0.0, "step": 1111 }, { "epoch": 0.62, "learning_rate": 0.0006526479750778816, "loss": 0.0, "step": 1112 }, { "epoch": 0.62, "learning_rate": 0.0006518691588785048, "loss": 0.0, "step": 1113 }, { "epoch": 0.62, "learning_rate": 0.0006510903426791277, "loss": 0.0, "step": 1114 }, { "epoch": 0.62, "learning_rate": 0.0006503115264797507, "loss": 0.0, "step": 1115 }, { "epoch": 0.63, "learning_rate": 0.0006495327102803739, "loss": 0.0, "step": 1116 }, { "epoch": 0.63, "learning_rate": 0.0006487538940809969, "loss": 0.0, "step": 1117 }, { "epoch": 0.63, "learning_rate": 0.00064797507788162, "loss": 0.0, "step": 1118 }, { "epoch": 0.63, "learning_rate": 0.000647196261682243, "loss": 0.0, "step": 1119 }, { "epoch": 0.63, "learning_rate": 0.000646417445482866, "loss": 0.0, "step": 1120 }, { "epoch": 0.63, "learning_rate": 0.0006456386292834892, "loss": 0.0, "step": 1121 }, { "epoch": 0.63, "learning_rate": 0.0006448598130841121, "loss": 0.0, "step": 1122 }, { "epoch": 0.63, "learning_rate": 0.0006440809968847352, "loss": 0.0, "step": 1123 }, { "epoch": 0.63, "learning_rate": 0.0006433021806853583, "loss": 0.0, "step": 1124 }, { "epoch": 0.63, "learning_rate": 0.0006425233644859813, "loss": 0.0, "step": 1125 }, { "epoch": 0.63, "learning_rate": 0.0006417445482866044, "loss": 0.0, "step": 1126 }, { "epoch": 0.63, "learning_rate": 0.0006409657320872274, "loss": 0.0, "step": 1127 }, { "epoch": 0.63, "learning_rate": 0.0006401869158878505, "loss": 0.0, "step": 1128 }, { "epoch": 0.63, "learning_rate": 0.0006394080996884736, "loss": 0.0, "step": 1129 }, { "epoch": 0.63, "learning_rate": 0.0006386292834890965, "loss": 0.0, "step": 1130 }, { "epoch": 0.63, "learning_rate": 0.0006378504672897196, "loss": 0.0, "step": 1131 }, { "epoch": 0.63, "learning_rate": 0.0006370716510903427, "loss": 0.0, "step": 1132 }, { "epoch": 0.64, "learning_rate": 0.0006362928348909658, "loss": 0.0, "step": 1133 }, { "epoch": 0.64, "learning_rate": 0.0006355140186915888, "loss": 0.0, "step": 1134 }, { "epoch": 0.64, "learning_rate": 0.0006347352024922118, "loss": 0.0, "step": 1135 }, { "epoch": 0.64, "learning_rate": 0.0006339563862928349, "loss": 0.0, "step": 1136 }, { "epoch": 0.64, "learning_rate": 0.000633177570093458, "loss": 0.0, "step": 1137 }, { "epoch": 0.64, "learning_rate": 0.000632398753894081, "loss": 0.0, "step": 1138 }, { "epoch": 0.64, "learning_rate": 0.000631619937694704, "loss": 0.0, "step": 1139 }, { "epoch": 0.64, "learning_rate": 0.0006308411214953271, "loss": 0.0, "step": 1140 }, { "epoch": 0.64, "learning_rate": 0.0006300623052959502, "loss": 0.0, "step": 1141 }, { "epoch": 0.64, "learning_rate": 0.0006292834890965732, "loss": 0.0, "step": 1142 }, { "epoch": 0.64, "learning_rate": 0.0006285046728971963, "loss": 0.0, "step": 1143 }, { "epoch": 0.64, "learning_rate": 0.0006277258566978193, "loss": 0.0, "step": 1144 }, { "epoch": 0.64, "learning_rate": 0.0006269470404984425, "loss": 0.0, "step": 1145 }, { "epoch": 0.64, "learning_rate": 0.0006261682242990654, "loss": 0.0, "step": 1146 }, { "epoch": 0.64, "learning_rate": 0.0006253894080996884, "loss": 0.0, "step": 1147 }, { "epoch": 0.64, "learning_rate": 0.0006246105919003116, "loss": 0.0, "step": 1148 }, { "epoch": 0.64, "learning_rate": 0.0006238317757009346, "loss": 0.0, "step": 1149 }, { "epoch": 0.64, "learning_rate": 0.0006230529595015575, "loss": 0.0, "step": 1150 }, { "epoch": 0.65, "learning_rate": 0.0006222741433021807, "loss": 0.0, "step": 1151 }, { "epoch": 0.65, "learning_rate": 0.0006214953271028037, "loss": 0.0, "step": 1152 }, { "epoch": 0.65, "learning_rate": 0.0006207165109034269, "loss": 0.0, "step": 1153 }, { "epoch": 0.65, "learning_rate": 0.0006199376947040498, "loss": 0.0, "step": 1154 }, { "epoch": 0.65, "learning_rate": 0.000619158878504673, "loss": 0.0, "step": 1155 }, { "epoch": 0.65, "learning_rate": 0.000618380062305296, "loss": 0.0, "step": 1156 }, { "epoch": 0.65, "learning_rate": 0.000617601246105919, "loss": 0.0, "step": 1157 }, { "epoch": 0.65, "learning_rate": 0.0006168224299065421, "loss": 0.0, "step": 1158 }, { "epoch": 0.65, "learning_rate": 0.0006160436137071651, "loss": 0.0, "step": 1159 }, { "epoch": 0.65, "learning_rate": 0.0006152647975077882, "loss": 0.0, "step": 1160 }, { "epoch": 0.65, "learning_rate": 0.0006144859813084113, "loss": 0.0, "step": 1161 }, { "epoch": 0.65, "learning_rate": 0.0006137071651090342, "loss": 0.0, "step": 1162 }, { "epoch": 0.65, "learning_rate": 0.0006129283489096573, "loss": 0.0, "step": 1163 }, { "epoch": 0.65, "learning_rate": 0.0006121495327102804, "loss": 0.0, "step": 1164 }, { "epoch": 0.65, "learning_rate": 0.0006113707165109035, "loss": 0.0, "step": 1165 }, { "epoch": 0.65, "learning_rate": 0.0006105919003115265, "loss": 0.0, "step": 1166 }, { "epoch": 0.65, "learning_rate": 0.0006098130841121495, "loss": 0.0, "step": 1167 }, { "epoch": 0.65, "learning_rate": 0.0006090342679127726, "loss": 0.0, "step": 1168 }, { "epoch": 0.66, "learning_rate": 0.0006082554517133957, "loss": 0.0, "step": 1169 }, { "epoch": 0.66, "learning_rate": 0.0006074766355140187, "loss": 0.0, "step": 1170 }, { "epoch": 0.66, "learning_rate": 0.0006066978193146417, "loss": 0.0, "step": 1171 }, { "epoch": 0.66, "learning_rate": 0.0006059190031152648, "loss": 0.0, "step": 1172 }, { "epoch": 0.66, "learning_rate": 0.0006051401869158879, "loss": 0.0, "step": 1173 }, { "epoch": 0.66, "learning_rate": 0.0006043613707165109, "loss": 0.0, "step": 1174 }, { "epoch": 0.66, "learning_rate": 0.000603582554517134, "loss": 0.0, "step": 1175 }, { "epoch": 0.66, "learning_rate": 0.000602803738317757, "loss": 0.0, "step": 1176 }, { "epoch": 0.66, "learning_rate": 0.0006020249221183801, "loss": 0.0, "step": 1177 }, { "epoch": 0.66, "learning_rate": 0.0006012461059190031, "loss": 0.0, "step": 1178 }, { "epoch": 0.66, "learning_rate": 0.0006004672897196261, "loss": 0.0, "step": 1179 }, { "epoch": 0.66, "learning_rate": 0.0005996884735202493, "loss": 0.0, "step": 1180 }, { "epoch": 0.66, "learning_rate": 0.0005989096573208723, "loss": 0.0, "step": 1181 }, { "epoch": 0.66, "learning_rate": 0.0005981308411214953, "loss": 0.0, "step": 1182 }, { "epoch": 0.66, "learning_rate": 0.0005973520249221184, "loss": 0.0, "step": 1183 }, { "epoch": 0.66, "learning_rate": 0.0005965732087227414, "loss": 0.0, "step": 1184 }, { "epoch": 0.66, "learning_rate": 0.0005957943925233646, "loss": 0.0, "step": 1185 }, { "epoch": 0.66, "learning_rate": 0.0005950155763239875, "loss": 0.0, "step": 1186 }, { "epoch": 0.67, "learning_rate": 0.0005942367601246105, "loss": 0.0, "step": 1187 }, { "epoch": 0.67, "learning_rate": 0.0005934579439252337, "loss": 0.0, "step": 1188 }, { "epoch": 0.67, "learning_rate": 0.0005926791277258567, "loss": 0.0, "step": 1189 }, { "epoch": 0.67, "learning_rate": 0.0005919003115264798, "loss": 0.0, "step": 1190 }, { "epoch": 0.67, "learning_rate": 0.0005911214953271028, "loss": 0.0, "step": 1191 }, { "epoch": 0.67, "learning_rate": 0.0005903426791277258, "loss": 0.0, "step": 1192 }, { "epoch": 0.67, "learning_rate": 0.000589563862928349, "loss": 0.0, "step": 1193 }, { "epoch": 0.67, "learning_rate": 0.0005887850467289719, "loss": 0.0, "step": 1194 }, { "epoch": 0.67, "learning_rate": 0.0005880062305295951, "loss": 0.0, "step": 1195 }, { "epoch": 0.67, "learning_rate": 0.0005872274143302181, "loss": 0.0, "step": 1196 }, { "epoch": 0.67, "learning_rate": 0.0005864485981308411, "loss": 0.0, "step": 1197 }, { "epoch": 0.67, "learning_rate": 0.0005856697819314642, "loss": 0.0, "step": 1198 }, { "epoch": 0.67, "learning_rate": 0.0005848909657320872, "loss": 0.0, "step": 1199 }, { "epoch": 0.67, "learning_rate": 0.0005841121495327103, "loss": 0.0, "step": 1200 }, { "epoch": 0.67, "learning_rate": 0.0005833333333333334, "loss": 0.0, "step": 1201 }, { "epoch": 0.67, "learning_rate": 0.0005825545171339563, "loss": 0.0, "step": 1202 }, { "epoch": 0.67, "learning_rate": 0.0005817757009345795, "loss": 0.0, "step": 1203 }, { "epoch": 0.67, "learning_rate": 0.0005809968847352025, "loss": 0.0, "step": 1204 }, { "epoch": 0.68, "learning_rate": 0.0005802180685358256, "loss": 0.0, "step": 1205 }, { "epoch": 0.68, "learning_rate": 0.0005794392523364486, "loss": 0.0, "step": 1206 }, { "epoch": 0.68, "learning_rate": 0.0005786604361370716, "loss": 0.0, "step": 1207 }, { "epoch": 0.68, "learning_rate": 0.0005778816199376947, "loss": 0.0, "step": 1208 }, { "epoch": 0.68, "learning_rate": 0.0005771028037383178, "loss": 0.0, "step": 1209 }, { "epoch": 0.68, "learning_rate": 0.0005763239875389408, "loss": 0.0, "step": 1210 }, { "epoch": 0.68, "learning_rate": 0.0005755451713395639, "loss": 0.0, "step": 1211 }, { "epoch": 0.68, "learning_rate": 0.0005747663551401869, "loss": 0.0, "step": 1212 }, { "epoch": 0.68, "learning_rate": 0.00057398753894081, "loss": 0.0, "step": 1213 }, { "epoch": 0.68, "learning_rate": 0.000573208722741433, "loss": 0.0, "step": 1214 }, { "epoch": 0.68, "learning_rate": 0.0005724299065420561, "loss": 0.0, "step": 1215 }, { "epoch": 0.68, "learning_rate": 0.0005716510903426791, "loss": 0.0, "step": 1216 }, { "epoch": 0.68, "learning_rate": 0.0005708722741433022, "loss": 0.0, "step": 1217 }, { "epoch": 0.68, "learning_rate": 0.0005700934579439252, "loss": 0.0, "step": 1218 }, { "epoch": 0.68, "learning_rate": 0.0005693146417445483, "loss": 0.0, "step": 1219 }, { "epoch": 0.68, "learning_rate": 0.0005685358255451714, "loss": 0.0, "step": 1220 }, { "epoch": 0.68, "learning_rate": 0.0005677570093457944, "loss": 0.0, "step": 1221 }, { "epoch": 0.68, "learning_rate": 0.0005669781931464174, "loss": 0.0, "step": 1222 }, { "epoch": 0.69, "learning_rate": 0.0005661993769470405, "loss": 0.0, "step": 1223 }, { "epoch": 0.69, "learning_rate": 0.0005654205607476635, "loss": 0.0, "step": 1224 }, { "epoch": 0.69, "learning_rate": 0.0005646417445482867, "loss": 0.0, "step": 1225 }, { "epoch": 0.69, "learning_rate": 0.0005638629283489096, "loss": 0.0, "step": 1226 }, { "epoch": 0.69, "learning_rate": 0.0005630841121495327, "loss": 0.0, "step": 1227 }, { "epoch": 0.69, "learning_rate": 0.0005623052959501558, "loss": 0.0, "step": 1228 }, { "epoch": 0.69, "learning_rate": 0.0005615264797507788, "loss": 0.0, "step": 1229 }, { "epoch": 0.69, "learning_rate": 0.0005607476635514019, "loss": 0.0, "step": 1230 }, { "epoch": 0.69, "learning_rate": 0.0005599688473520249, "loss": 0.0, "step": 1231 }, { "epoch": 0.69, "learning_rate": 0.0005591900311526479, "loss": 0.0, "step": 1232 }, { "epoch": 0.69, "learning_rate": 0.0005584112149532711, "loss": 0.0, "step": 1233 }, { "epoch": 0.69, "learning_rate": 0.000557632398753894, "loss": 0.0, "step": 1234 }, { "epoch": 0.69, "learning_rate": 0.0005568535825545172, "loss": 0.0, "step": 1235 }, { "epoch": 0.69, "learning_rate": 0.0005560747663551402, "loss": 0.0, "step": 1236 }, { "epoch": 0.69, "learning_rate": 0.0005552959501557632, "loss": 0.0, "step": 1237 }, { "epoch": 0.69, "learning_rate": 0.0005545171339563863, "loss": 0.0, "step": 1238 }, { "epoch": 0.69, "learning_rate": 0.0005537383177570093, "loss": 0.0, "step": 1239 }, { "epoch": 0.7, "learning_rate": 0.0005529595015576325, "loss": 0.0, "step": 1240 }, { "epoch": 0.7, "learning_rate": 0.0005521806853582555, "loss": 0.0, "step": 1241 }, { "epoch": 0.7, "learning_rate": 0.0005514018691588784, "loss": 0.0, "step": 1242 }, { "epoch": 0.7, "learning_rate": 0.0005506230529595016, "loss": 0.0, "step": 1243 }, { "epoch": 0.7, "learning_rate": 0.0005498442367601246, "loss": 0.0, "step": 1244 }, { "epoch": 0.7, "learning_rate": 0.0005490654205607477, "loss": 0.0, "step": 1245 }, { "epoch": 0.7, "learning_rate": 0.0005482866043613707, "loss": 0.0, "step": 1246 }, { "epoch": 0.7, "learning_rate": 0.0005475077881619937, "loss": 0.0, "step": 1247 }, { "epoch": 0.7, "learning_rate": 0.0005467289719626169, "loss": 0.0, "step": 1248 }, { "epoch": 0.7, "learning_rate": 0.0005459501557632399, "loss": 0.0, "step": 1249 }, { "epoch": 0.7, "learning_rate": 0.0005451713395638629, "loss": 0.0, "step": 1250 }, { "epoch": 0.7, "learning_rate": 0.000544392523364486, "loss": 0.0, "step": 1251 }, { "epoch": 0.7, "learning_rate": 0.000543613707165109, "loss": 0.0, "step": 1252 }, { "epoch": 0.7, "learning_rate": 0.0005428348909657321, "loss": 0.0, "step": 1253 }, { "epoch": 0.7, "learning_rate": 0.0005420560747663551, "loss": 0.0, "step": 1254 }, { "epoch": 0.7, "learning_rate": 0.0005412772585669782, "loss": 0.0, "step": 1255 }, { "epoch": 0.7, "learning_rate": 0.0005404984423676013, "loss": 0.0, "step": 1256 }, { "epoch": 0.7, "learning_rate": 0.0005397196261682244, "loss": 0.0, "step": 1257 }, { "epoch": 0.71, "learning_rate": 0.0005389408099688473, "loss": 0.0, "step": 1258 }, { "epoch": 0.71, "learning_rate": 0.0005381619937694704, "loss": 0.0, "step": 1259 }, { "epoch": 0.71, "learning_rate": 0.0005373831775700935, "loss": 0.0, "step": 1260 }, { "epoch": 0.71, "learning_rate": 0.0005366043613707165, "loss": 0.0, "step": 1261 }, { "epoch": 0.71, "learning_rate": 0.0005358255451713396, "loss": 0.0, "step": 1262 }, { "epoch": 0.71, "learning_rate": 0.0005350467289719626, "loss": 0.0, "step": 1263 }, { "epoch": 0.71, "learning_rate": 0.0005342679127725857, "loss": 0.0, "step": 1264 }, { "epoch": 0.71, "learning_rate": 0.0005334890965732088, "loss": 0.0, "step": 1265 }, { "epoch": 0.71, "learning_rate": 0.0005327102803738317, "loss": 0.0, "step": 1266 }, { "epoch": 0.71, "learning_rate": 0.0005319314641744549, "loss": 0.0, "step": 1267 }, { "epoch": 0.71, "learning_rate": 0.0005311526479750779, "loss": 0.0, "step": 1268 }, { "epoch": 0.71, "learning_rate": 0.0005303738317757009, "loss": 0.0, "step": 1269 }, { "epoch": 0.71, "learning_rate": 0.000529595015576324, "loss": 0.0, "step": 1270 }, { "epoch": 0.71, "learning_rate": 0.000528816199376947, "loss": 0.0, "step": 1271 }, { "epoch": 0.71, "learning_rate": 0.0005280373831775702, "loss": 0.0, "step": 1272 }, { "epoch": 0.71, "learning_rate": 0.0005272585669781932, "loss": 0.0, "step": 1273 }, { "epoch": 0.71, "learning_rate": 0.0005264797507788161, "loss": 0.0, "step": 1274 }, { "epoch": 0.71, "learning_rate": 0.0005257009345794393, "loss": 0.0, "step": 1275 }, { "epoch": 0.72, "learning_rate": 0.0005249221183800623, "loss": 0.0, "step": 1276 }, { "epoch": 0.72, "learning_rate": 0.0005241433021806854, "loss": 0.0, "step": 1277 }, { "epoch": 0.72, "learning_rate": 0.0005233644859813084, "loss": 0.0, "step": 1278 }, { "epoch": 0.72, "learning_rate": 0.0005225856697819314, "loss": 0.0, "step": 1279 }, { "epoch": 0.72, "learning_rate": 0.0005218068535825546, "loss": 0.0, "step": 1280 }, { "epoch": 0.72, "learning_rate": 0.0005210280373831776, "loss": 0.0, "step": 1281 }, { "epoch": 0.72, "learning_rate": 0.0005202492211838006, "loss": 0.0, "step": 1282 }, { "epoch": 0.72, "learning_rate": 0.0005194704049844237, "loss": 0.0, "step": 1283 }, { "epoch": 0.72, "learning_rate": 0.0005186915887850467, "loss": 0.0, "step": 1284 }, { "epoch": 0.72, "learning_rate": 0.0005179127725856698, "loss": 0.0, "step": 1285 }, { "epoch": 0.72, "learning_rate": 0.0005171339563862928, "loss": 0.0, "step": 1286 }, { "epoch": 0.72, "learning_rate": 0.0005163551401869159, "loss": 0.0, "step": 1287 }, { "epoch": 0.72, "learning_rate": 0.000515576323987539, "loss": 0.0, "step": 1288 }, { "epoch": 0.72, "learning_rate": 0.000514797507788162, "loss": 0.0, "step": 1289 }, { "epoch": 0.72, "learning_rate": 0.000514018691588785, "loss": 0.0, "step": 1290 }, { "epoch": 0.72, "learning_rate": 0.0005132398753894081, "loss": 0.0, "step": 1291 }, { "epoch": 0.72, "learning_rate": 0.0005124610591900312, "loss": 0.0, "step": 1292 }, { "epoch": 0.72, "learning_rate": 0.0005116822429906542, "loss": 0.0, "step": 1293 }, { "epoch": 0.73, "learning_rate": 0.0005109034267912772, "loss": 0.0, "step": 1294 }, { "epoch": 0.73, "learning_rate": 0.0005101246105919003, "loss": 0.0, "step": 1295 }, { "epoch": 0.73, "learning_rate": 0.0005093457943925234, "loss": 0.0, "step": 1296 }, { "epoch": 0.73, "learning_rate": 0.0005085669781931465, "loss": 0.0, "step": 1297 }, { "epoch": 0.73, "learning_rate": 0.0005077881619937694, "loss": 0.0, "step": 1298 }, { "epoch": 0.73, "learning_rate": 0.0005070093457943925, "loss": 0.0, "step": 1299 }, { "epoch": 0.73, "learning_rate": 0.0005062305295950156, "loss": 0.0, "step": 1300 }, { "epoch": 0.73, "learning_rate": 0.0005054517133956386, "loss": 0.0, "step": 1301 }, { "epoch": 0.73, "learning_rate": 0.0005046728971962617, "loss": 0.0, "step": 1302 }, { "epoch": 0.73, "learning_rate": 0.0005038940809968847, "loss": 0.0, "step": 1303 }, { "epoch": 0.73, "learning_rate": 0.0005031152647975078, "loss": 0.0, "step": 1304 }, { "epoch": 0.73, "learning_rate": 0.0005023364485981309, "loss": 0.0, "step": 1305 }, { "epoch": 0.73, "learning_rate": 0.0005015576323987538, "loss": 0.0, "step": 1306 }, { "epoch": 0.73, "learning_rate": 0.000500778816199377, "loss": 0.0, "step": 1307 }, { "epoch": 0.73, "learning_rate": 0.0005, "loss": 0.0, "step": 1308 }, { "epoch": 0.73, "learning_rate": 0.000499221183800623, "loss": 0.0, "step": 1309 }, { "epoch": 0.73, "learning_rate": 0.0004984423676012461, "loss": 0.0, "step": 1310 }, { "epoch": 0.73, "learning_rate": 0.0004976635514018692, "loss": 0.0, "step": 1311 }, { "epoch": 0.74, "learning_rate": 0.0004968847352024923, "loss": 0.0, "step": 1312 }, { "epoch": 0.74, "learning_rate": 0.0004961059190031153, "loss": 0.0, "step": 1313 }, { "epoch": 0.74, "learning_rate": 0.0004953271028037383, "loss": 0.0, "step": 1314 }, { "epoch": 0.74, "learning_rate": 0.0004945482866043614, "loss": 0.0, "step": 1315 }, { "epoch": 0.74, "learning_rate": 0.0004937694704049844, "loss": 0.0, "step": 1316 }, { "epoch": 0.74, "learning_rate": 0.0004929906542056076, "loss": 0.0, "step": 1317 }, { "epoch": 0.74, "learning_rate": 0.0004922118380062306, "loss": 0.0, "step": 1318 }, { "epoch": 0.74, "learning_rate": 0.0004914330218068536, "loss": 0.0, "step": 1319 }, { "epoch": 0.74, "learning_rate": 0.0004906542056074767, "loss": 0.0, "step": 1320 }, { "epoch": 0.74, "learning_rate": 0.0004898753894080997, "loss": 0.0, "step": 1321 }, { "epoch": 0.74, "learning_rate": 0.0004890965732087227, "loss": 0.0, "step": 1322 }, { "epoch": 0.74, "learning_rate": 0.0004883177570093459, "loss": 0.0, "step": 1323 }, { "epoch": 0.74, "learning_rate": 0.00048753894080996886, "loss": 0.0, "step": 1324 }, { "epoch": 0.74, "learning_rate": 0.00048676012461059195, "loss": 0.0, "step": 1325 }, { "epoch": 0.74, "learning_rate": 0.00048598130841121494, "loss": 0.0, "step": 1326 }, { "epoch": 0.74, "learning_rate": 0.00048520249221183803, "loss": 0.0, "step": 1327 }, { "epoch": 0.74, "learning_rate": 0.00048442367601246106, "loss": 0.0, "step": 1328 }, { "epoch": 0.74, "learning_rate": 0.00048364485981308415, "loss": 0.0, "step": 1329 }, { "epoch": 0.75, "learning_rate": 0.0004828660436137072, "loss": 0.0, "step": 1330 }, { "epoch": 0.75, "learning_rate": 0.0004820872274143302, "loss": 0.0, "step": 1331 }, { "epoch": 0.75, "learning_rate": 0.00048130841121495326, "loss": 0.0, "step": 1332 }, { "epoch": 0.75, "learning_rate": 0.00048052959501557635, "loss": 0.0, "step": 1333 }, { "epoch": 0.75, "learning_rate": 0.0004797507788161994, "loss": 0.0, "step": 1334 }, { "epoch": 0.75, "learning_rate": 0.0004789719626168225, "loss": 0.0, "step": 1335 }, { "epoch": 0.75, "learning_rate": 0.0004781931464174455, "loss": 0.0, "step": 1336 }, { "epoch": 0.75, "learning_rate": 0.00047741433021806855, "loss": 0.0, "step": 1337 }, { "epoch": 0.75, "learning_rate": 0.0004766355140186916, "loss": 0.0, "step": 1338 }, { "epoch": 0.75, "learning_rate": 0.0004758566978193147, "loss": 0.0, "step": 1339 }, { "epoch": 0.75, "learning_rate": 0.0004750778816199377, "loss": 0.0, "step": 1340 }, { "epoch": 0.75, "learning_rate": 0.0004742990654205608, "loss": 0.0, "step": 1341 }, { "epoch": 0.75, "learning_rate": 0.0004735202492211838, "loss": 0.0, "step": 1342 }, { "epoch": 0.75, "learning_rate": 0.0004727414330218069, "loss": 0.0, "step": 1343 }, { "epoch": 0.75, "learning_rate": 0.0004719626168224299, "loss": 0.0, "step": 1344 }, { "epoch": 0.75, "learning_rate": 0.000471183800623053, "loss": 0.0, "step": 1345 }, { "epoch": 0.75, "learning_rate": 0.00047040498442367604, "loss": 0.0, "step": 1346 }, { "epoch": 0.76, "learning_rate": 0.0004696261682242991, "loss": 0.0, "step": 1347 }, { "epoch": 0.76, "learning_rate": 0.0004688473520249221, "loss": 0.0, "step": 1348 }, { "epoch": 0.76, "learning_rate": 0.0004680685358255452, "loss": 0.0, "step": 1349 }, { "epoch": 0.76, "learning_rate": 0.00046728971962616824, "loss": 0.0, "step": 1350 }, { "epoch": 0.76, "learning_rate": 0.00046651090342679133, "loss": 0.0, "step": 1351 }, { "epoch": 0.76, "learning_rate": 0.0004657320872274143, "loss": 0.0, "step": 1352 }, { "epoch": 0.76, "learning_rate": 0.0004649532710280374, "loss": 0.0, "step": 1353 }, { "epoch": 0.76, "learning_rate": 0.00046417445482866044, "loss": 0.0, "step": 1354 }, { "epoch": 0.76, "learning_rate": 0.00046339563862928353, "loss": 0.0, "step": 1355 }, { "epoch": 0.76, "learning_rate": 0.00046261682242990657, "loss": 0.0, "step": 1356 }, { "epoch": 0.76, "learning_rate": 0.0004618380062305296, "loss": 0.0, "step": 1357 }, { "epoch": 0.76, "learning_rate": 0.00046105919003115264, "loss": 0.0, "step": 1358 }, { "epoch": 0.76, "learning_rate": 0.00046028037383177573, "loss": 0.0, "step": 1359 }, { "epoch": 0.76, "learning_rate": 0.00045950155763239877, "loss": 0.0, "step": 1360 }, { "epoch": 0.76, "learning_rate": 0.00045872274143302186, "loss": 0.0, "step": 1361 }, { "epoch": 0.76, "learning_rate": 0.00045794392523364484, "loss": 0.0, "step": 1362 }, { "epoch": 0.76, "learning_rate": 0.00045716510903426793, "loss": 0.0, "step": 1363 }, { "epoch": 0.76, "learning_rate": 0.00045638629283489097, "loss": 0.0, "step": 1364 }, { "epoch": 0.77, "learning_rate": 0.00045560747663551406, "loss": 0.0, "step": 1365 }, { "epoch": 0.77, "learning_rate": 0.0004548286604361371, "loss": 0.0, "step": 1366 }, { "epoch": 0.77, "learning_rate": 0.00045404984423676013, "loss": 0.0, "step": 1367 }, { "epoch": 0.77, "learning_rate": 0.00045327102803738317, "loss": 0.0, "step": 1368 }, { "epoch": 0.77, "learning_rate": 0.00045249221183800626, "loss": 0.0, "step": 1369 }, { "epoch": 0.77, "learning_rate": 0.0004517133956386293, "loss": 0.0, "step": 1370 }, { "epoch": 0.77, "learning_rate": 0.0004509345794392524, "loss": 0.0, "step": 1371 }, { "epoch": 0.77, "learning_rate": 0.00045015576323987537, "loss": 0.0, "step": 1372 }, { "epoch": 0.77, "learning_rate": 0.00044937694704049846, "loss": 0.0, "step": 1373 }, { "epoch": 0.77, "learning_rate": 0.0004485981308411215, "loss": 0.0, "step": 1374 }, { "epoch": 0.77, "learning_rate": 0.0004478193146417446, "loss": 0.0, "step": 1375 }, { "epoch": 0.77, "learning_rate": 0.0004470404984423676, "loss": 0.0, "step": 1376 }, { "epoch": 0.77, "learning_rate": 0.00044626168224299066, "loss": 0.0, "step": 1377 }, { "epoch": 0.77, "learning_rate": 0.0004454828660436137, "loss": 0.0, "step": 1378 }, { "epoch": 0.77, "learning_rate": 0.0004447040498442368, "loss": 0.0, "step": 1379 }, { "epoch": 0.77, "learning_rate": 0.0004439252336448598, "loss": 0.0, "step": 1380 }, { "epoch": 0.77, "learning_rate": 0.0004431464174454829, "loss": 0.0, "step": 1381 }, { "epoch": 0.77, "learning_rate": 0.0004423676012461059, "loss": 0.0, "step": 1382 }, { "epoch": 0.78, "learning_rate": 0.000441588785046729, "loss": 0.0, "step": 1383 }, { "epoch": 0.78, "learning_rate": 0.000440809968847352, "loss": 0.0, "step": 1384 }, { "epoch": 0.78, "learning_rate": 0.0004400311526479751, "loss": 0.0, "step": 1385 }, { "epoch": 0.78, "learning_rate": 0.00043925233644859815, "loss": 0.0, "step": 1386 }, { "epoch": 0.78, "learning_rate": 0.00043847352024922124, "loss": 0.0, "step": 1387 }, { "epoch": 0.78, "learning_rate": 0.0004376947040498442, "loss": 0.0, "step": 1388 }, { "epoch": 0.78, "learning_rate": 0.0004369158878504673, "loss": 0.0, "step": 1389 }, { "epoch": 0.78, "learning_rate": 0.00043613707165109035, "loss": 0.0, "step": 1390 }, { "epoch": 0.78, "learning_rate": 0.00043535825545171344, "loss": 0.0, "step": 1391 }, { "epoch": 0.78, "learning_rate": 0.0004345794392523365, "loss": 0.0, "step": 1392 }, { "epoch": 0.78, "learning_rate": 0.0004338006230529595, "loss": 0.0, "step": 1393 }, { "epoch": 0.78, "learning_rate": 0.00043302180685358255, "loss": 0.0, "step": 1394 }, { "epoch": 0.78, "learning_rate": 0.00043224299065420564, "loss": 0.0, "step": 1395 }, { "epoch": 0.78, "learning_rate": 0.0004314641744548287, "loss": 0.0, "step": 1396 }, { "epoch": 0.78, "learning_rate": 0.00043068535825545177, "loss": 0.0, "step": 1397 }, { "epoch": 0.78, "learning_rate": 0.00042990654205607475, "loss": 0.0, "step": 1398 }, { "epoch": 0.78, "learning_rate": 0.00042912772585669784, "loss": 0.0, "step": 1399 }, { "epoch": 0.78, "learning_rate": 0.0004283489096573209, "loss": 0.0, "step": 1400 }, { "epoch": 0.79, "learning_rate": 0.00042757009345794397, "loss": 0.0, "step": 1401 }, { "epoch": 0.79, "learning_rate": 0.000426791277258567, "loss": 0.0, "step": 1402 }, { "epoch": 0.79, "learning_rate": 0.00042601246105919004, "loss": 0.0, "step": 1403 }, { "epoch": 0.79, "learning_rate": 0.0004252336448598131, "loss": 0.0, "step": 1404 }, { "epoch": 0.79, "learning_rate": 0.00042445482866043617, "loss": 0.0, "step": 1405 }, { "epoch": 0.79, "learning_rate": 0.0004236760124610592, "loss": 0.0, "step": 1406 }, { "epoch": 0.79, "learning_rate": 0.0004228971962616823, "loss": 0.0, "step": 1407 }, { "epoch": 0.79, "learning_rate": 0.0004221183800623053, "loss": 0.0, "step": 1408 }, { "epoch": 0.79, "learning_rate": 0.00042133956386292837, "loss": 0.0, "step": 1409 }, { "epoch": 0.79, "learning_rate": 0.0004205607476635514, "loss": 0.0, "step": 1410 }, { "epoch": 0.79, "learning_rate": 0.0004197819314641745, "loss": 0.0, "step": 1411 }, { "epoch": 0.79, "learning_rate": 0.00041900311526479753, "loss": 0.0, "step": 1412 }, { "epoch": 0.79, "learning_rate": 0.00041822429906542057, "loss": 0.0, "step": 1413 }, { "epoch": 0.79, "learning_rate": 0.0004174454828660436, "loss": 0.0, "step": 1414 }, { "epoch": 0.79, "learning_rate": 0.0004166666666666667, "loss": 0.0, "step": 1415 }, { "epoch": 0.79, "learning_rate": 0.00041588785046728973, "loss": 0.0, "step": 1416 }, { "epoch": 0.79, "learning_rate": 0.0004151090342679128, "loss": 0.0, "step": 1417 }, { "epoch": 0.79, "learning_rate": 0.0004143302180685358, "loss": 0.0, "step": 1418 }, { "epoch": 0.8, "learning_rate": 0.0004135514018691589, "loss": 0.0, "step": 1419 }, { "epoch": 0.8, "learning_rate": 0.00041277258566978193, "loss": 0.0, "step": 1420 }, { "epoch": 0.8, "learning_rate": 0.000411993769470405, "loss": 0.0, "step": 1421 }, { "epoch": 0.8, "learning_rate": 0.00041121495327102806, "loss": 0.0, "step": 1422 }, { "epoch": 0.8, "learning_rate": 0.0004104361370716511, "loss": 0.0, "step": 1423 }, { "epoch": 0.8, "learning_rate": 0.00040965732087227413, "loss": 0.0, "step": 1424 }, { "epoch": 0.8, "learning_rate": 0.0004088785046728972, "loss": 0.0, "step": 1425 }, { "epoch": 0.8, "learning_rate": 0.00040809968847352026, "loss": 0.0, "step": 1426 }, { "epoch": 0.8, "learning_rate": 0.00040732087227414335, "loss": 0.0, "step": 1427 }, { "epoch": 0.8, "learning_rate": 0.00040654205607476633, "loss": 0.0, "step": 1428 }, { "epoch": 0.8, "learning_rate": 0.0004057632398753894, "loss": 0.0, "step": 1429 }, { "epoch": 0.8, "learning_rate": 0.00040498442367601246, "loss": 0.0, "step": 1430 }, { "epoch": 0.8, "learning_rate": 0.00040420560747663555, "loss": 0.0, "step": 1431 }, { "epoch": 0.8, "learning_rate": 0.0004034267912772586, "loss": 0.0, "step": 1432 }, { "epoch": 0.8, "learning_rate": 0.0004026479750778816, "loss": 0.0, "step": 1433 }, { "epoch": 0.8, "learning_rate": 0.00040186915887850466, "loss": 0.0, "step": 1434 }, { "epoch": 0.8, "learning_rate": 0.00040109034267912775, "loss": 0.0, "step": 1435 }, { "epoch": 0.8, "learning_rate": 0.0004003115264797508, "loss": 0.0, "step": 1436 }, { "epoch": 0.81, "learning_rate": 0.0003995327102803739, "loss": 0.0, "step": 1437 }, { "epoch": 0.81, "learning_rate": 0.00039875389408099686, "loss": 0.0, "step": 1438 }, { "epoch": 0.81, "learning_rate": 0.00039797507788161995, "loss": 0.0, "step": 1439 }, { "epoch": 0.81, "learning_rate": 0.000397196261682243, "loss": 0.0, "step": 1440 }, { "epoch": 0.81, "learning_rate": 0.0003964174454828661, "loss": 0.0, "step": 1441 }, { "epoch": 0.81, "learning_rate": 0.0003956386292834891, "loss": 0.0, "step": 1442 }, { "epoch": 0.81, "learning_rate": 0.0003948598130841122, "loss": 0.0, "step": 1443 }, { "epoch": 0.81, "learning_rate": 0.0003940809968847352, "loss": 0.0, "step": 1444 }, { "epoch": 0.81, "learning_rate": 0.0003933021806853583, "loss": 0.0, "step": 1445 }, { "epoch": 0.81, "learning_rate": 0.0003925233644859813, "loss": 0.0, "step": 1446 }, { "epoch": 0.81, "learning_rate": 0.0003917445482866044, "loss": 0.0, "step": 1447 }, { "epoch": 0.81, "learning_rate": 0.00039096573208722744, "loss": 0.0, "step": 1448 }, { "epoch": 0.81, "learning_rate": 0.00039018691588785047, "loss": 0.0, "step": 1449 }, { "epoch": 0.81, "learning_rate": 0.0003894080996884735, "loss": 0.0, "step": 1450 }, { "epoch": 0.81, "learning_rate": 0.0003886292834890966, "loss": 0.0, "step": 1451 }, { "epoch": 0.81, "learning_rate": 0.00038785046728971964, "loss": 0.0, "step": 1452 }, { "epoch": 0.81, "learning_rate": 0.0003870716510903427, "loss": 0.0, "step": 1453 }, { "epoch": 0.82, "learning_rate": 0.0003862928348909657, "loss": 0.0, "step": 1454 }, { "epoch": 0.82, "learning_rate": 0.0003855140186915888, "loss": 0.0, "step": 1455 }, { "epoch": 0.82, "learning_rate": 0.00038473520249221184, "loss": 0.0, "step": 1456 }, { "epoch": 0.82, "learning_rate": 0.0003839563862928349, "loss": 0.0, "step": 1457 }, { "epoch": 0.82, "learning_rate": 0.00038317757009345796, "loss": 0.0, "step": 1458 }, { "epoch": 0.82, "learning_rate": 0.000382398753894081, "loss": 0.0, "step": 1459 }, { "epoch": 0.82, "learning_rate": 0.00038161993769470404, "loss": 0.0, "step": 1460 }, { "epoch": 0.82, "learning_rate": 0.0003808411214953271, "loss": 0.0, "step": 1461 }, { "epoch": 0.82, "learning_rate": 0.00038006230529595016, "loss": 0.0, "step": 1462 }, { "epoch": 0.82, "learning_rate": 0.00037928348909657325, "loss": 0.0, "step": 1463 }, { "epoch": 0.82, "learning_rate": 0.00037850467289719624, "loss": 0.0, "step": 1464 }, { "epoch": 0.82, "learning_rate": 0.0003777258566978193, "loss": 0.0, "step": 1465 }, { "epoch": 0.82, "learning_rate": 0.00037694704049844236, "loss": 0.0, "step": 1466 }, { "epoch": 0.82, "learning_rate": 0.00037616822429906545, "loss": 0.0, "step": 1467 }, { "epoch": 0.82, "learning_rate": 0.0003753894080996885, "loss": 0.0, "step": 1468 }, { "epoch": 0.82, "learning_rate": 0.0003746105919003115, "loss": 0.0, "step": 1469 }, { "epoch": 0.82, "learning_rate": 0.00037383177570093456, "loss": 0.0, "step": 1470 }, { "epoch": 0.82, "learning_rate": 0.00037305295950155765, "loss": 0.0, "step": 1471 }, { "epoch": 0.83, "learning_rate": 0.0003722741433021807, "loss": 0.0, "step": 1472 }, { "epoch": 0.83, "learning_rate": 0.0003714953271028038, "loss": 0.0, "step": 1473 }, { "epoch": 0.83, "learning_rate": 0.00037071651090342676, "loss": 0.0, "step": 1474 }, { "epoch": 0.83, "learning_rate": 0.00036993769470404985, "loss": 0.0, "step": 1475 }, { "epoch": 0.83, "learning_rate": 0.0003691588785046729, "loss": 0.0, "step": 1476 }, { "epoch": 0.83, "learning_rate": 0.000368380062305296, "loss": 0.0, "step": 1477 }, { "epoch": 0.83, "learning_rate": 0.000367601246105919, "loss": 0.0, "step": 1478 }, { "epoch": 0.83, "learning_rate": 0.00036682242990654205, "loss": 0.0, "step": 1479 }, { "epoch": 0.83, "learning_rate": 0.0003660436137071651, "loss": 0.0, "step": 1480 }, { "epoch": 0.83, "learning_rate": 0.0003652647975077882, "loss": 0.0, "step": 1481 }, { "epoch": 0.83, "learning_rate": 0.0003644859813084112, "loss": 0.0, "step": 1482 }, { "epoch": 0.83, "learning_rate": 0.0003637071651090343, "loss": 0.0, "step": 1483 }, { "epoch": 0.83, "learning_rate": 0.0003629283489096573, "loss": 0.0, "step": 1484 }, { "epoch": 0.83, "learning_rate": 0.0003621495327102804, "loss": 0.0, "step": 1485 }, { "epoch": 0.83, "learning_rate": 0.0003613707165109034, "loss": 0.0, "step": 1486 }, { "epoch": 0.83, "learning_rate": 0.0003605919003115265, "loss": 0.0, "step": 1487 }, { "epoch": 0.83, "learning_rate": 0.00035981308411214954, "loss": 0.0, "step": 1488 }, { "epoch": 0.83, "learning_rate": 0.0003590342679127726, "loss": 0.0, "step": 1489 }, { "epoch": 0.84, "learning_rate": 0.0003582554517133956, "loss": 0.0, "step": 1490 }, { "epoch": 0.84, "learning_rate": 0.0003574766355140187, "loss": 0.0, "step": 1491 }, { "epoch": 0.84, "learning_rate": 0.00035669781931464174, "loss": 0.0, "step": 1492 }, { "epoch": 0.84, "learning_rate": 0.00035591900311526483, "loss": 0.0, "step": 1493 }, { "epoch": 0.84, "learning_rate": 0.0003551401869158878, "loss": 0.0, "step": 1494 }, { "epoch": 0.84, "learning_rate": 0.0003543613707165109, "loss": 0.0, "step": 1495 }, { "epoch": 0.84, "learning_rate": 0.00035358255451713394, "loss": 0.0, "step": 1496 }, { "epoch": 0.84, "learning_rate": 0.00035280373831775703, "loss": 0.0, "step": 1497 }, { "epoch": 0.84, "learning_rate": 0.00035202492211838007, "loss": 0.0, "step": 1498 }, { "epoch": 0.84, "learning_rate": 0.00035124610591900316, "loss": 0.0, "step": 1499 }, { "epoch": 0.84, "learning_rate": 0.00035046728971962614, "loss": 0.0, "step": 1500 }, { "epoch": 0.84, "eval_loss": NaN, "eval_runtime": 906.0585, "eval_samples_per_second": 2.916, "eval_steps_per_second": 0.365, "eval_wer": 2.0097029585094353, "step": 1500 }, { "epoch": 0.84, "learning_rate": 0.00034968847352024923, "loss": 0.0, "step": 1501 }, { "epoch": 0.84, "learning_rate": 0.00034890965732087227, "loss": 0.0, "step": 1502 }, { "epoch": 0.84, "learning_rate": 0.00034813084112149536, "loss": 0.0, "step": 1503 }, { "epoch": 0.84, "learning_rate": 0.0003473520249221184, "loss": 0.0, "step": 1504 }, { "epoch": 0.84, "learning_rate": 0.00034657320872274143, "loss": 0.0, "step": 1505 }, { "epoch": 0.84, "learning_rate": 0.00034579439252336447, "loss": 0.0, "step": 1506 }, { "epoch": 0.84, "learning_rate": 0.00034501557632398756, "loss": 0.0, "step": 1507 }, { "epoch": 0.85, "learning_rate": 0.0003442367601246106, "loss": 0.0, "step": 1508 }, { "epoch": 0.85, "learning_rate": 0.0003434579439252337, "loss": 0.0, "step": 1509 }, { "epoch": 0.85, "learning_rate": 0.00034267912772585667, "loss": 0.0, "step": 1510 }, { "epoch": 0.85, "learning_rate": 0.00034190031152647976, "loss": 0.0, "step": 1511 }, { "epoch": 0.85, "learning_rate": 0.0003411214953271028, "loss": 0.0, "step": 1512 }, { "epoch": 0.85, "learning_rate": 0.0003403426791277259, "loss": 0.0, "step": 1513 }, { "epoch": 0.85, "learning_rate": 0.0003395638629283489, "loss": 0.0, "step": 1514 }, { "epoch": 0.85, "learning_rate": 0.00033878504672897196, "loss": 0.0, "step": 1515 }, { "epoch": 0.85, "learning_rate": 0.000338006230529595, "loss": 0.0, "step": 1516 }, { "epoch": 0.85, "learning_rate": 0.0003372274143302181, "loss": 0.0, "step": 1517 }, { "epoch": 0.85, "learning_rate": 0.0003364485981308411, "loss": 0.0, "step": 1518 }, { "epoch": 0.85, "learning_rate": 0.0003356697819314642, "loss": 0.0, "step": 1519 }, { "epoch": 0.85, "learning_rate": 0.0003348909657320872, "loss": 0.0, "step": 1520 }, { "epoch": 0.85, "learning_rate": 0.0003341121495327103, "loss": 0.0, "step": 1521 }, { "epoch": 0.85, "learning_rate": 0.0003333333333333333, "loss": 0.0, "step": 1522 }, { "epoch": 0.85, "learning_rate": 0.0003325545171339564, "loss": 0.0, "step": 1523 }, { "epoch": 0.85, "learning_rate": 0.00033177570093457945, "loss": 0.0, "step": 1524 }, { "epoch": 0.85, "learning_rate": 0.0003309968847352025, "loss": 0.0, "step": 1525 }, { "epoch": 0.86, "learning_rate": 0.0003302180685358255, "loss": 0.0, "step": 1526 }, { "epoch": 0.86, "learning_rate": 0.0003294392523364486, "loss": 0.0, "step": 1527 }, { "epoch": 0.86, "learning_rate": 0.00032866043613707165, "loss": 0.0, "step": 1528 }, { "epoch": 0.86, "learning_rate": 0.00032788161993769474, "loss": 0.0, "step": 1529 }, { "epoch": 0.86, "learning_rate": 0.0003271028037383177, "loss": 0.0, "step": 1530 }, { "epoch": 0.86, "learning_rate": 0.0003263239875389408, "loss": 0.0, "step": 1531 }, { "epoch": 0.86, "learning_rate": 0.00032554517133956385, "loss": 0.0, "step": 1532 }, { "epoch": 0.86, "learning_rate": 0.00032476635514018694, "loss": 0.0, "step": 1533 }, { "epoch": 0.86, "learning_rate": 0.00032398753894081, "loss": 0.0, "step": 1534 }, { "epoch": 0.86, "learning_rate": 0.000323208722741433, "loss": 0.0, "step": 1535 }, { "epoch": 0.86, "learning_rate": 0.00032242990654205605, "loss": 0.0, "step": 1536 }, { "epoch": 0.86, "learning_rate": 0.00032165109034267914, "loss": 0.0, "step": 1537 }, { "epoch": 0.86, "learning_rate": 0.0003208722741433022, "loss": 0.0, "step": 1538 }, { "epoch": 0.86, "learning_rate": 0.00032009345794392527, "loss": 0.0, "step": 1539 }, { "epoch": 0.86, "learning_rate": 0.00031931464174454825, "loss": 0.0, "step": 1540 }, { "epoch": 0.86, "learning_rate": 0.00031853582554517134, "loss": 0.0, "step": 1541 }, { "epoch": 0.86, "learning_rate": 0.0003177570093457944, "loss": 0.0, "step": 1542 }, { "epoch": 0.86, "learning_rate": 0.00031697819314641747, "loss": 0.0, "step": 1543 }, { "epoch": 0.87, "learning_rate": 0.0003161993769470405, "loss": 0.0, "step": 1544 }, { "epoch": 0.87, "learning_rate": 0.00031542056074766354, "loss": 0.0, "step": 1545 }, { "epoch": 0.87, "learning_rate": 0.0003146417445482866, "loss": 0.0, "step": 1546 }, { "epoch": 0.87, "learning_rate": 0.00031386292834890967, "loss": 0.0, "step": 1547 }, { "epoch": 0.87, "learning_rate": 0.0003130841121495327, "loss": 0.0, "step": 1548 }, { "epoch": 0.87, "learning_rate": 0.0003123052959501558, "loss": 0.0, "step": 1549 }, { "epoch": 0.87, "learning_rate": 0.0003115264797507788, "loss": 0.0, "step": 1550 }, { "epoch": 0.87, "learning_rate": 0.00031074766355140187, "loss": 0.0, "step": 1551 }, { "epoch": 0.87, "learning_rate": 0.0003099688473520249, "loss": 0.0, "step": 1552 }, { "epoch": 0.87, "learning_rate": 0.000309190031152648, "loss": 0.0, "step": 1553 }, { "epoch": 0.87, "learning_rate": 0.00030841121495327103, "loss": 0.0, "step": 1554 }, { "epoch": 0.87, "learning_rate": 0.0003076323987538941, "loss": 0.0, "step": 1555 }, { "epoch": 0.87, "learning_rate": 0.0003068535825545171, "loss": 0.0, "step": 1556 }, { "epoch": 0.87, "learning_rate": 0.0003060747663551402, "loss": 0.0, "step": 1557 }, { "epoch": 0.87, "learning_rate": 0.00030529595015576323, "loss": 0.0, "step": 1558 }, { "epoch": 0.87, "learning_rate": 0.0003045171339563863, "loss": 0.0, "step": 1559 }, { "epoch": 0.87, "learning_rate": 0.00030373831775700936, "loss": 0.0, "step": 1560 }, { "epoch": 0.88, "learning_rate": 0.0003029595015576324, "loss": 0.0, "step": 1561 }, { "epoch": 0.88, "learning_rate": 0.00030218068535825543, "loss": 0.0, "step": 1562 }, { "epoch": 0.88, "learning_rate": 0.0003014018691588785, "loss": 0.0, "step": 1563 }, { "epoch": 0.88, "learning_rate": 0.00030062305295950156, "loss": 0.0, "step": 1564 }, { "epoch": 0.88, "learning_rate": 0.00029984423676012465, "loss": 0.0, "step": 1565 }, { "epoch": 0.88, "learning_rate": 0.00029906542056074763, "loss": 0.0, "step": 1566 }, { "epoch": 0.88, "learning_rate": 0.0002982866043613707, "loss": 0.0, "step": 1567 }, { "epoch": 0.88, "learning_rate": 0.00029750778816199376, "loss": 0.0, "step": 1568 }, { "epoch": 0.88, "learning_rate": 0.00029672897196261685, "loss": 0.0, "step": 1569 }, { "epoch": 0.88, "learning_rate": 0.0002959501557632399, "loss": 0.0, "step": 1570 }, { "epoch": 0.88, "learning_rate": 0.0002951713395638629, "loss": 0.0, "step": 1571 }, { "epoch": 0.88, "learning_rate": 0.00029439252336448596, "loss": 0.0, "step": 1572 }, { "epoch": 0.88, "learning_rate": 0.00029361370716510905, "loss": 0.0, "step": 1573 }, { "epoch": 0.88, "learning_rate": 0.0002928348909657321, "loss": 0.0, "step": 1574 }, { "epoch": 0.88, "learning_rate": 0.00029205607476635517, "loss": 0.0, "step": 1575 }, { "epoch": 0.88, "learning_rate": 0.00029127725856697815, "loss": 0.0, "step": 1576 }, { "epoch": 0.88, "learning_rate": 0.00029049844236760125, "loss": 0.0, "step": 1577 }, { "epoch": 0.88, "learning_rate": 0.0002897196261682243, "loss": 0.0, "step": 1578 }, { "epoch": 0.89, "learning_rate": 0.00028894080996884737, "loss": 0.0, "step": 1579 }, { "epoch": 0.89, "learning_rate": 0.0002881619937694704, "loss": 0.0, "step": 1580 }, { "epoch": 0.89, "learning_rate": 0.00028738317757009345, "loss": 0.0, "step": 1581 }, { "epoch": 0.89, "learning_rate": 0.0002866043613707165, "loss": 0.0, "step": 1582 }, { "epoch": 0.89, "learning_rate": 0.00028582554517133957, "loss": 0.0, "step": 1583 }, { "epoch": 0.89, "learning_rate": 0.0002850467289719626, "loss": 0.0, "step": 1584 }, { "epoch": 0.89, "learning_rate": 0.0002842679127725857, "loss": 0.0, "step": 1585 }, { "epoch": 0.89, "learning_rate": 0.0002834890965732087, "loss": 0.0, "step": 1586 }, { "epoch": 0.89, "learning_rate": 0.00028271028037383177, "loss": 0.0, "step": 1587 }, { "epoch": 0.89, "learning_rate": 0.0002819314641744548, "loss": 0.0, "step": 1588 }, { "epoch": 0.89, "learning_rate": 0.0002811526479750779, "loss": 0.0, "step": 1589 }, { "epoch": 0.89, "learning_rate": 0.00028037383177570094, "loss": 0.0, "step": 1590 }, { "epoch": 0.89, "learning_rate": 0.00027959501557632397, "loss": 0.0, "step": 1591 }, { "epoch": 0.89, "learning_rate": 0.000278816199376947, "loss": 0.0, "step": 1592 }, { "epoch": 0.89, "learning_rate": 0.0002780373831775701, "loss": 0.0, "step": 1593 }, { "epoch": 0.89, "learning_rate": 0.00027725856697819314, "loss": 0.0, "step": 1594 }, { "epoch": 0.89, "learning_rate": 0.0002764797507788162, "loss": 0.0, "step": 1595 }, { "epoch": 0.89, "learning_rate": 0.0002757009345794392, "loss": 0.0, "step": 1596 }, { "epoch": 0.9, "learning_rate": 0.0002749221183800623, "loss": 0.0, "step": 1597 }, { "epoch": 0.9, "learning_rate": 0.00027414330218068533, "loss": 0.0, "step": 1598 }, { "epoch": 0.9, "learning_rate": 0.0002733644859813084, "loss": 0.0, "step": 1599 }, { "epoch": 0.9, "learning_rate": 0.00027258566978193146, "loss": 0.0, "step": 1600 }, { "epoch": 0.9, "learning_rate": 0.0002718068535825545, "loss": 0.0, "step": 1601 }, { "epoch": 0.9, "learning_rate": 0.00027102803738317753, "loss": 0.0, "step": 1602 }, { "epoch": 0.9, "learning_rate": 0.0002702492211838006, "loss": 0.0, "step": 1603 }, { "epoch": 0.9, "learning_rate": 0.00026947040498442366, "loss": 0.0, "step": 1604 }, { "epoch": 0.9, "learning_rate": 0.00026869158878504675, "loss": 0.0, "step": 1605 }, { "epoch": 0.9, "learning_rate": 0.0002679127725856698, "loss": 0.0, "step": 1606 }, { "epoch": 0.9, "learning_rate": 0.0002671339563862928, "loss": 0.0, "step": 1607 }, { "epoch": 0.9, "learning_rate": 0.00026635514018691586, "loss": 0.0, "step": 1608 }, { "epoch": 0.9, "learning_rate": 0.00026557632398753895, "loss": 0.0, "step": 1609 }, { "epoch": 0.9, "learning_rate": 0.000264797507788162, "loss": 0.0, "step": 1610 }, { "epoch": 0.9, "learning_rate": 0.0002640186915887851, "loss": 0.0, "step": 1611 }, { "epoch": 0.9, "learning_rate": 0.00026323987538940806, "loss": 0.0, "step": 1612 }, { "epoch": 0.9, "learning_rate": 0.00026246105919003115, "loss": 0.0, "step": 1613 }, { "epoch": 0.9, "learning_rate": 0.0002616822429906542, "loss": 0.0, "step": 1614 }, { "epoch": 0.91, "learning_rate": 0.0002609034267912773, "loss": 0.0, "step": 1615 }, { "epoch": 0.91, "learning_rate": 0.0002601246105919003, "loss": 0.0, "step": 1616 }, { "epoch": 0.91, "learning_rate": 0.00025934579439252335, "loss": 0.0, "step": 1617 }, { "epoch": 0.91, "learning_rate": 0.0002585669781931464, "loss": 0.0, "step": 1618 }, { "epoch": 0.91, "learning_rate": 0.0002577881619937695, "loss": 0.0, "step": 1619 }, { "epoch": 0.91, "learning_rate": 0.0002570093457943925, "loss": 0.0, "step": 1620 }, { "epoch": 0.91, "learning_rate": 0.0002562305295950156, "loss": 0.0, "step": 1621 }, { "epoch": 0.91, "learning_rate": 0.0002554517133956386, "loss": 0.0, "step": 1622 }, { "epoch": 0.91, "learning_rate": 0.0002546728971962617, "loss": 0.0, "step": 1623 }, { "epoch": 0.91, "learning_rate": 0.0002538940809968847, "loss": 0.0, "step": 1624 }, { "epoch": 0.91, "learning_rate": 0.0002531152647975078, "loss": 0.0, "step": 1625 }, { "epoch": 0.91, "learning_rate": 0.00025233644859813084, "loss": 0.0, "step": 1626 }, { "epoch": 0.91, "learning_rate": 0.0002515576323987539, "loss": 0.0, "step": 1627 }, { "epoch": 0.91, "learning_rate": 0.0002507788161993769, "loss": 0.0, "step": 1628 }, { "epoch": 0.91, "learning_rate": 0.00025, "loss": 0.0, "step": 1629 }, { "epoch": 0.91, "learning_rate": 0.00024922118380062304, "loss": 0.0, "step": 1630 }, { "epoch": 0.91, "learning_rate": 0.00024844236760124613, "loss": 0.0, "step": 1631 }, { "epoch": 0.91, "learning_rate": 0.00024766355140186917, "loss": 0.0, "step": 1632 }, { "epoch": 0.92, "learning_rate": 0.0002468847352024922, "loss": 0.0, "step": 1633 }, { "epoch": 0.92, "learning_rate": 0.0002461059190031153, "loss": 0.0, "step": 1634 }, { "epoch": 0.92, "learning_rate": 0.00024532710280373833, "loss": 0.0, "step": 1635 }, { "epoch": 0.92, "learning_rate": 0.00024454828660436137, "loss": 0.0, "step": 1636 }, { "epoch": 0.92, "learning_rate": 0.00024376947040498443, "loss": 0.0, "step": 1637 }, { "epoch": 0.92, "learning_rate": 0.00024299065420560747, "loss": 0.0, "step": 1638 }, { "epoch": 0.92, "learning_rate": 0.00024221183800623053, "loss": 0.0, "step": 1639 }, { "epoch": 0.92, "learning_rate": 0.0002414330218068536, "loss": 0.0, "step": 1640 }, { "epoch": 0.92, "learning_rate": 0.00024065420560747663, "loss": 0.0, "step": 1641 }, { "epoch": 0.92, "learning_rate": 0.0002398753894080997, "loss": 0.0, "step": 1642 }, { "epoch": 0.92, "learning_rate": 0.00023909657320872276, "loss": 0.0, "step": 1643 }, { "epoch": 0.92, "learning_rate": 0.0002383177570093458, "loss": 0.0, "step": 1644 }, { "epoch": 0.92, "learning_rate": 0.00023753894080996886, "loss": 0.0, "step": 1645 }, { "epoch": 0.92, "learning_rate": 0.0002367601246105919, "loss": 0.0, "step": 1646 }, { "epoch": 0.92, "learning_rate": 0.00023598130841121496, "loss": 0.0, "step": 1647 }, { "epoch": 0.92, "learning_rate": 0.00023520249221183802, "loss": 0.0, "step": 1648 }, { "epoch": 0.92, "learning_rate": 0.00023442367601246106, "loss": 0.0, "step": 1649 }, { "epoch": 0.92, "learning_rate": 0.00023364485981308412, "loss": 0.0, "step": 1650 }, { "epoch": 0.93, "learning_rate": 0.00023286604361370716, "loss": 0.0, "step": 1651 }, { "epoch": 0.93, "learning_rate": 0.00023208722741433022, "loss": 0.0, "step": 1652 }, { "epoch": 0.93, "learning_rate": 0.00023130841121495329, "loss": 0.0, "step": 1653 }, { "epoch": 0.93, "learning_rate": 0.00023052959501557632, "loss": 0.0, "step": 1654 }, { "epoch": 0.93, "learning_rate": 0.00022975077881619939, "loss": 0.0, "step": 1655 }, { "epoch": 0.93, "learning_rate": 0.00022897196261682242, "loss": 0.0, "step": 1656 }, { "epoch": 0.93, "learning_rate": 0.00022819314641744548, "loss": 0.0, "step": 1657 }, { "epoch": 0.93, "learning_rate": 0.00022741433021806855, "loss": 0.0, "step": 1658 }, { "epoch": 0.93, "learning_rate": 0.00022663551401869158, "loss": 0.0, "step": 1659 }, { "epoch": 0.93, "learning_rate": 0.00022585669781931465, "loss": 0.0, "step": 1660 }, { "epoch": 0.93, "learning_rate": 0.00022507788161993768, "loss": 0.0, "step": 1661 }, { "epoch": 0.93, "learning_rate": 0.00022429906542056075, "loss": 0.0, "step": 1662 }, { "epoch": 0.93, "learning_rate": 0.0002235202492211838, "loss": 0.0, "step": 1663 }, { "epoch": 0.93, "learning_rate": 0.00022274143302180685, "loss": 0.0, "step": 1664 }, { "epoch": 0.93, "learning_rate": 0.0002219626168224299, "loss": 0.0, "step": 1665 }, { "epoch": 0.93, "learning_rate": 0.00022118380062305295, "loss": 0.0, "step": 1666 }, { "epoch": 0.93, "learning_rate": 0.000220404984423676, "loss": 0.0, "step": 1667 }, { "epoch": 0.93, "learning_rate": 0.00021962616822429908, "loss": 0.0, "step": 1668 }, { "epoch": 0.94, "learning_rate": 0.0002188473520249221, "loss": 0.0, "step": 1669 }, { "epoch": 0.94, "learning_rate": 0.00021806853582554517, "loss": 0.0, "step": 1670 }, { "epoch": 0.94, "learning_rate": 0.00021728971962616824, "loss": 0.0, "step": 1671 }, { "epoch": 0.94, "learning_rate": 0.00021651090342679127, "loss": 0.0, "step": 1672 }, { "epoch": 0.94, "learning_rate": 0.00021573208722741434, "loss": 0.0, "step": 1673 }, { "epoch": 0.94, "learning_rate": 0.00021495327102803737, "loss": 0.0, "step": 1674 }, { "epoch": 0.94, "learning_rate": 0.00021417445482866044, "loss": 0.0, "step": 1675 }, { "epoch": 0.94, "learning_rate": 0.0002133956386292835, "loss": 0.0, "step": 1676 }, { "epoch": 0.94, "learning_rate": 0.00021261682242990654, "loss": 0.0, "step": 1677 }, { "epoch": 0.94, "learning_rate": 0.0002118380062305296, "loss": 0.0, "step": 1678 }, { "epoch": 0.94, "learning_rate": 0.00021105919003115264, "loss": 0.0, "step": 1679 }, { "epoch": 0.94, "learning_rate": 0.0002102803738317757, "loss": 0.0, "step": 1680 }, { "epoch": 0.94, "learning_rate": 0.00020950155763239877, "loss": 0.0, "step": 1681 }, { "epoch": 0.94, "learning_rate": 0.0002087227414330218, "loss": 0.0, "step": 1682 }, { "epoch": 0.94, "learning_rate": 0.00020794392523364486, "loss": 0.0, "step": 1683 }, { "epoch": 0.94, "learning_rate": 0.0002071651090342679, "loss": 0.0, "step": 1684 }, { "epoch": 0.94, "learning_rate": 0.00020638629283489096, "loss": 0.0, "step": 1685 }, { "epoch": 0.95, "learning_rate": 0.00020560747663551403, "loss": 0.0, "step": 1686 }, { "epoch": 0.95, "learning_rate": 0.00020482866043613706, "loss": 0.0, "step": 1687 }, { "epoch": 0.95, "learning_rate": 0.00020404984423676013, "loss": 0.0, "step": 1688 }, { "epoch": 0.95, "learning_rate": 0.00020327102803738316, "loss": 0.0, "step": 1689 }, { "epoch": 0.95, "learning_rate": 0.00020249221183800623, "loss": 0.0, "step": 1690 }, { "epoch": 0.95, "learning_rate": 0.0002017133956386293, "loss": 0.0, "step": 1691 }, { "epoch": 0.95, "learning_rate": 0.00020093457943925233, "loss": 0.0, "step": 1692 }, { "epoch": 0.95, "learning_rate": 0.0002001557632398754, "loss": 0.0, "step": 1693 }, { "epoch": 0.95, "learning_rate": 0.00019937694704049843, "loss": 0.0, "step": 1694 }, { "epoch": 0.95, "learning_rate": 0.0001985981308411215, "loss": 0.0, "step": 1695 }, { "epoch": 0.95, "learning_rate": 0.00019781931464174455, "loss": 0.0, "step": 1696 }, { "epoch": 0.95, "learning_rate": 0.0001970404984423676, "loss": 0.0, "step": 1697 }, { "epoch": 0.95, "learning_rate": 0.00019626168224299065, "loss": 0.0, "step": 1698 }, { "epoch": 0.95, "learning_rate": 0.00019548286604361372, "loss": 0.0, "step": 1699 }, { "epoch": 0.95, "learning_rate": 0.00019470404984423675, "loss": 0.0, "step": 1700 }, { "epoch": 0.95, "learning_rate": 0.00019392523364485982, "loss": 0.0, "step": 1701 }, { "epoch": 0.95, "learning_rate": 0.00019314641744548285, "loss": 0.0, "step": 1702 }, { "epoch": 0.95, "learning_rate": 0.00019236760124610592, "loss": 0.0, "step": 1703 }, { "epoch": 0.96, "learning_rate": 0.00019158878504672898, "loss": 0.0, "step": 1704 }, { "epoch": 0.96, "learning_rate": 0.00019080996884735202, "loss": 0.0, "step": 1705 }, { "epoch": 0.96, "learning_rate": 0.00019003115264797508, "loss": 0.0, "step": 1706 }, { "epoch": 0.96, "learning_rate": 0.00018925233644859812, "loss": 0.0, "step": 1707 }, { "epoch": 0.96, "learning_rate": 0.00018847352024922118, "loss": 0.0, "step": 1708 }, { "epoch": 0.96, "learning_rate": 0.00018769470404984424, "loss": 0.0, "step": 1709 }, { "epoch": 0.96, "learning_rate": 0.00018691588785046728, "loss": 0.0, "step": 1710 }, { "epoch": 0.96, "learning_rate": 0.00018613707165109034, "loss": 0.0, "step": 1711 }, { "epoch": 0.96, "learning_rate": 0.00018535825545171338, "loss": 0.0, "step": 1712 }, { "epoch": 0.96, "learning_rate": 0.00018457943925233644, "loss": 0.0, "step": 1713 }, { "epoch": 0.96, "learning_rate": 0.0001838006230529595, "loss": 0.0, "step": 1714 }, { "epoch": 0.96, "learning_rate": 0.00018302180685358254, "loss": 0.0, "step": 1715 }, { "epoch": 0.96, "learning_rate": 0.0001822429906542056, "loss": 0.0, "step": 1716 }, { "epoch": 0.96, "learning_rate": 0.00018146417445482864, "loss": 0.0, "step": 1717 }, { "epoch": 0.96, "learning_rate": 0.0001806853582554517, "loss": 0.0, "step": 1718 }, { "epoch": 0.96, "learning_rate": 0.00017990654205607477, "loss": 0.0, "step": 1719 }, { "epoch": 0.96, "learning_rate": 0.0001791277258566978, "loss": 0.0, "step": 1720 }, { "epoch": 0.96, "learning_rate": 0.00017834890965732087, "loss": 0.0, "step": 1721 }, { "epoch": 0.97, "learning_rate": 0.0001775700934579439, "loss": 0.0, "step": 1722 }, { "epoch": 0.97, "learning_rate": 0.00017679127725856697, "loss": 0.0, "step": 1723 }, { "epoch": 0.97, "learning_rate": 0.00017601246105919003, "loss": 0.0, "step": 1724 }, { "epoch": 0.97, "learning_rate": 0.00017523364485981307, "loss": 0.0, "step": 1725 }, { "epoch": 0.97, "learning_rate": 0.00017445482866043613, "loss": 0.0, "step": 1726 }, { "epoch": 0.97, "learning_rate": 0.0001736760124610592, "loss": 0.0, "step": 1727 }, { "epoch": 0.97, "learning_rate": 0.00017289719626168223, "loss": 0.0, "step": 1728 }, { "epoch": 0.97, "learning_rate": 0.0001721183800623053, "loss": 0.0, "step": 1729 }, { "epoch": 0.97, "learning_rate": 0.00017133956386292833, "loss": 0.0, "step": 1730 }, { "epoch": 0.97, "learning_rate": 0.0001705607476635514, "loss": 0.0, "step": 1731 }, { "epoch": 0.97, "learning_rate": 0.00016978193146417446, "loss": 0.0, "step": 1732 }, { "epoch": 0.97, "learning_rate": 0.0001690031152647975, "loss": 0.0, "step": 1733 }, { "epoch": 0.97, "learning_rate": 0.00016822429906542056, "loss": 0.0, "step": 1734 }, { "epoch": 0.97, "learning_rate": 0.0001674454828660436, "loss": 0.0, "step": 1735 }, { "epoch": 0.97, "learning_rate": 0.00016666666666666666, "loss": 0.0, "step": 1736 }, { "epoch": 0.97, "learning_rate": 0.00016588785046728972, "loss": 0.0, "step": 1737 }, { "epoch": 0.97, "learning_rate": 0.00016510903426791276, "loss": 0.0, "step": 1738 }, { "epoch": 0.97, "learning_rate": 0.00016433021806853582, "loss": 0.0, "step": 1739 }, { "epoch": 0.98, "learning_rate": 0.00016355140186915886, "loss": 0.0, "step": 1740 }, { "epoch": 0.98, "learning_rate": 0.00016277258566978192, "loss": 0.0, "step": 1741 }, { "epoch": 0.98, "learning_rate": 0.000161993769470405, "loss": 0.0, "step": 1742 }, { "epoch": 0.98, "learning_rate": 0.00016121495327102802, "loss": 0.0, "step": 1743 }, { "epoch": 0.98, "learning_rate": 0.0001604361370716511, "loss": 0.0, "step": 1744 }, { "epoch": 0.98, "learning_rate": 0.00015965732087227412, "loss": 0.0, "step": 1745 }, { "epoch": 0.98, "learning_rate": 0.0001588785046728972, "loss": 0.0, "step": 1746 }, { "epoch": 0.98, "learning_rate": 0.00015809968847352025, "loss": 0.0, "step": 1747 }, { "epoch": 0.98, "learning_rate": 0.0001573208722741433, "loss": 0.0, "step": 1748 }, { "epoch": 0.98, "learning_rate": 0.00015654205607476635, "loss": 0.0, "step": 1749 }, { "epoch": 0.98, "learning_rate": 0.0001557632398753894, "loss": 0.0, "step": 1750 }, { "epoch": 0.98, "learning_rate": 0.00015498442367601245, "loss": 0.0, "step": 1751 }, { "epoch": 0.98, "learning_rate": 0.00015420560747663551, "loss": 0.0, "step": 1752 }, { "epoch": 0.98, "learning_rate": 0.00015342679127725855, "loss": 0.0, "step": 1753 }, { "epoch": 0.98, "learning_rate": 0.00015264797507788161, "loss": 0.0, "step": 1754 }, { "epoch": 0.98, "learning_rate": 0.00015186915887850468, "loss": 0.0, "step": 1755 }, { "epoch": 0.98, "learning_rate": 0.00015109034267912771, "loss": 0.0, "step": 1756 }, { "epoch": 0.98, "learning_rate": 0.00015031152647975078, "loss": 0.0, "step": 1757 }, { "epoch": 0.99, "learning_rate": 0.00014953271028037381, "loss": 0.0, "step": 1758 }, { "epoch": 0.99, "learning_rate": 0.00014875389408099688, "loss": 0.0, "step": 1759 }, { "epoch": 0.99, "learning_rate": 0.00014797507788161994, "loss": 0.0, "step": 1760 }, { "epoch": 0.99, "learning_rate": 0.00014719626168224298, "loss": 0.0, "step": 1761 }, { "epoch": 0.99, "learning_rate": 0.00014641744548286604, "loss": 0.0, "step": 1762 }, { "epoch": 0.99, "learning_rate": 0.00014563862928348908, "loss": 0.0, "step": 1763 }, { "epoch": 0.99, "learning_rate": 0.00014485981308411214, "loss": 0.0, "step": 1764 }, { "epoch": 0.99, "learning_rate": 0.0001440809968847352, "loss": 0.0, "step": 1765 }, { "epoch": 0.99, "learning_rate": 0.00014330218068535824, "loss": 0.0, "step": 1766 }, { "epoch": 0.99, "learning_rate": 0.0001425233644859813, "loss": 0.0, "step": 1767 }, { "epoch": 0.99, "learning_rate": 0.00014174454828660434, "loss": 0.0, "step": 1768 }, { "epoch": 0.99, "learning_rate": 0.0001409657320872274, "loss": 0.0, "step": 1769 }, { "epoch": 0.99, "learning_rate": 0.00014018691588785047, "loss": 0.0, "step": 1770 }, { "epoch": 0.99, "learning_rate": 0.0001394080996884735, "loss": 0.0, "step": 1771 }, { "epoch": 0.99, "learning_rate": 0.00013862928348909657, "loss": 0.0, "step": 1772 }, { "epoch": 0.99, "learning_rate": 0.0001378504672897196, "loss": 0.0, "step": 1773 }, { "epoch": 0.99, "learning_rate": 0.00013707165109034267, "loss": 0.0, "step": 1774 }, { "epoch": 0.99, "learning_rate": 0.00013629283489096573, "loss": 0.0, "step": 1775 }, { "epoch": 1.0, "learning_rate": 0.00013551401869158877, "loss": 0.0, "step": 1776 }, { "epoch": 1.0, "learning_rate": 0.00013473520249221183, "loss": 0.0, "step": 1777 }, { "epoch": 1.0, "learning_rate": 0.0001339563862928349, "loss": 0.0, "step": 1778 }, { "epoch": 1.0, "learning_rate": 0.00013317757009345793, "loss": 0.0, "step": 1779 }, { "epoch": 1.0, "learning_rate": 0.000132398753894081, "loss": 0.0, "step": 1780 }, { "epoch": 1.0, "learning_rate": 0.00013161993769470403, "loss": 0.0, "step": 1781 }, { "epoch": 1.0, "learning_rate": 0.0001308411214953271, "loss": 0.0, "step": 1782 }, { "epoch": 1.0, "learning_rate": 0.00013006230529595016, "loss": 0.0, "step": 1783 }, { "epoch": 1.0, "learning_rate": 0.0001292834890965732, "loss": 0.0, "step": 1784 }, { "epoch": 1.0, "step": 1784, "total_flos": 0.0, "train_loss": 2.2983257342881687, "train_runtime": 7464.978, "train_samples_per_second": 3.823, "train_steps_per_second": 0.239 } ], "max_steps": 1784, "num_train_epochs": 1, "total_flos": 0.0, "trial_name": null, "trial_params": null }