{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.4804804804804805, "eval_steps": 500.0, "global_step": 3000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 5.339028296849973e-08, "loss": 1.1519, "step": 1 }, { "epoch": 0.0, "learning_rate": 1.0678056593699946e-07, "loss": 1.131, "step": 2 }, { "epoch": 0.0, "learning_rate": 1.601708489054992e-07, "loss": 1.1885, "step": 3 }, { "epoch": 0.0, "learning_rate": 2.1356113187399892e-07, "loss": 1.2035, "step": 4 }, { "epoch": 0.0, "learning_rate": 2.6695141484249866e-07, "loss": 1.1678, "step": 5 }, { "epoch": 0.0, "learning_rate": 3.203416978109984e-07, "loss": 1.2134, "step": 6 }, { "epoch": 0.0, "learning_rate": 3.7373198077949814e-07, "loss": 1.2847, "step": 7 }, { "epoch": 0.0, "learning_rate": 4.2712226374799783e-07, "loss": 1.2305, "step": 8 }, { "epoch": 0.0, "learning_rate": 4.805125467164976e-07, "loss": 1.1712, "step": 9 }, { "epoch": 0.0, "learning_rate": 5.339028296849973e-07, "loss": 1.1829, "step": 10 }, { "epoch": 0.0, "learning_rate": 5.872931126534971e-07, "loss": 1.2316, "step": 11 }, { "epoch": 0.0, "learning_rate": 6.406833956219968e-07, "loss": 1.1653, "step": 12 }, { "epoch": 0.0, "learning_rate": 6.940736785904966e-07, "loss": 1.3393, "step": 13 }, { "epoch": 0.0, "learning_rate": 7.474639615589963e-07, "loss": 1.1699, "step": 14 }, { "epoch": 0.0, "learning_rate": 8.00854244527496e-07, "loss": 1.1665, "step": 15 }, { "epoch": 0.0, "learning_rate": 8.542445274959957e-07, "loss": 1.1754, "step": 16 }, { "epoch": 0.0, "learning_rate": 9.076348104644955e-07, "loss": 1.2994, "step": 17 }, { "epoch": 0.0, "learning_rate": 9.610250934329953e-07, "loss": 1.2245, "step": 18 }, { "epoch": 0.0, "learning_rate": 1.014415376401495e-06, "loss": 1.2045, "step": 19 }, { "epoch": 0.0, "learning_rate": 1.0678056593699946e-06, "loss": 1.1242, "step": 20 }, { "epoch": 0.0, "learning_rate": 1.1211959423384945e-06, "loss": 1.1493, "step": 21 }, { "epoch": 0.0, "learning_rate": 1.1745862253069942e-06, "loss": 1.1814, "step": 22 }, { "epoch": 0.0, "learning_rate": 1.227976508275494e-06, "loss": 1.1887, "step": 23 }, { "epoch": 0.0, "learning_rate": 1.2813667912439936e-06, "loss": 1.1558, "step": 24 }, { "epoch": 0.0, "learning_rate": 1.3347570742124935e-06, "loss": 1.1393, "step": 25 }, { "epoch": 0.0, "learning_rate": 1.3881473571809932e-06, "loss": 1.2196, "step": 26 }, { "epoch": 0.0, "learning_rate": 1.4415376401494929e-06, "loss": 1.1657, "step": 27 }, { "epoch": 0.0, "learning_rate": 1.4949279231179926e-06, "loss": 1.199, "step": 28 }, { "epoch": 0.0, "learning_rate": 1.5483182060864923e-06, "loss": 1.226, "step": 29 }, { "epoch": 0.0, "learning_rate": 1.601708489054992e-06, "loss": 1.0933, "step": 30 }, { "epoch": 0.0, "learning_rate": 1.6550987720234916e-06, "loss": 1.2596, "step": 31 }, { "epoch": 0.01, "learning_rate": 1.7084890549919913e-06, "loss": 1.1734, "step": 32 }, { "epoch": 0.01, "learning_rate": 1.7618793379604912e-06, "loss": 1.1658, "step": 33 }, { "epoch": 0.01, "learning_rate": 1.815269620928991e-06, "loss": 1.1203, "step": 34 }, { "epoch": 0.01, "learning_rate": 1.8686599038974908e-06, "loss": 1.1814, "step": 35 }, { "epoch": 0.01, "learning_rate": 1.9220501868659905e-06, "loss": 1.2229, "step": 36 }, { "epoch": 0.01, "learning_rate": 1.9754404698344904e-06, "loss": 1.1853, "step": 37 }, { "epoch": 0.01, "learning_rate": 2.02883075280299e-06, "loss": 1.1213, "step": 38 }, { "epoch": 0.01, "learning_rate": 2.08222103577149e-06, "loss": 1.1162, "step": 39 }, { "epoch": 0.01, "learning_rate": 2.1356113187399893e-06, "loss": 1.2338, "step": 40 }, { "epoch": 0.01, "learning_rate": 2.189001601708489e-06, "loss": 1.2004, "step": 41 }, { "epoch": 0.01, "learning_rate": 2.242391884676989e-06, "loss": 1.1598, "step": 42 }, { "epoch": 0.01, "learning_rate": 2.2957821676454886e-06, "loss": 1.1818, "step": 43 }, { "epoch": 0.01, "learning_rate": 2.3491724506139885e-06, "loss": 1.1863, "step": 44 }, { "epoch": 0.01, "learning_rate": 2.402562733582488e-06, "loss": 1.1761, "step": 45 }, { "epoch": 0.01, "learning_rate": 2.455953016550988e-06, "loss": 1.1758, "step": 46 }, { "epoch": 0.01, "learning_rate": 2.5093432995194873e-06, "loss": 1.2511, "step": 47 }, { "epoch": 0.01, "learning_rate": 2.5627335824879872e-06, "loss": 1.2365, "step": 48 }, { "epoch": 0.01, "learning_rate": 2.616123865456487e-06, "loss": 1.2522, "step": 49 }, { "epoch": 0.01, "learning_rate": 2.669514148424987e-06, "loss": 1.1519, "step": 50 }, { "epoch": 0.01, "learning_rate": 2.7229044313934865e-06, "loss": 1.1373, "step": 51 }, { "epoch": 0.01, "learning_rate": 2.7762947143619864e-06, "loss": 1.2664, "step": 52 }, { "epoch": 0.01, "learning_rate": 2.829684997330486e-06, "loss": 1.2756, "step": 53 }, { "epoch": 0.01, "learning_rate": 2.8830752802989858e-06, "loss": 1.2247, "step": 54 }, { "epoch": 0.01, "learning_rate": 2.9364655632674857e-06, "loss": 1.2131, "step": 55 }, { "epoch": 0.01, "learning_rate": 2.989855846235985e-06, "loss": 1.0988, "step": 56 }, { "epoch": 0.01, "learning_rate": 3.043246129204485e-06, "loss": 1.1998, "step": 57 }, { "epoch": 0.01, "learning_rate": 3.0966364121729845e-06, "loss": 1.1489, "step": 58 }, { "epoch": 0.01, "learning_rate": 3.1500266951414844e-06, "loss": 1.1841, "step": 59 }, { "epoch": 0.01, "learning_rate": 3.203416978109984e-06, "loss": 1.2035, "step": 60 }, { "epoch": 0.01, "learning_rate": 3.256807261078484e-06, "loss": 1.211, "step": 61 }, { "epoch": 0.01, "learning_rate": 3.3101975440469833e-06, "loss": 1.1498, "step": 62 }, { "epoch": 0.01, "learning_rate": 3.363587827015483e-06, "loss": 1.2441, "step": 63 }, { "epoch": 0.01, "learning_rate": 3.4169781099839827e-06, "loss": 1.1729, "step": 64 }, { "epoch": 0.01, "learning_rate": 3.4703683929524826e-06, "loss": 1.1566, "step": 65 }, { "epoch": 0.01, "learning_rate": 3.5237586759209825e-06, "loss": 1.1664, "step": 66 }, { "epoch": 0.01, "learning_rate": 3.577148958889482e-06, "loss": 1.2512, "step": 67 }, { "epoch": 0.01, "learning_rate": 3.630539241857982e-06, "loss": 1.2457, "step": 68 }, { "epoch": 0.01, "learning_rate": 3.6839295248264813e-06, "loss": 1.2225, "step": 69 }, { "epoch": 0.01, "learning_rate": 3.7373198077949817e-06, "loss": 1.1345, "step": 70 }, { "epoch": 0.01, "learning_rate": 3.7907100907634816e-06, "loss": 1.2028, "step": 71 }, { "epoch": 0.01, "learning_rate": 3.844100373731981e-06, "loss": 1.2052, "step": 72 }, { "epoch": 0.01, "learning_rate": 3.897490656700481e-06, "loss": 1.2091, "step": 73 }, { "epoch": 0.01, "learning_rate": 3.950880939668981e-06, "loss": 1.0868, "step": 74 }, { "epoch": 0.01, "learning_rate": 4.00427122263748e-06, "loss": 1.2526, "step": 75 }, { "epoch": 0.01, "learning_rate": 4.05766150560598e-06, "loss": 1.1422, "step": 76 }, { "epoch": 0.01, "learning_rate": 4.11105178857448e-06, "loss": 1.2049, "step": 77 }, { "epoch": 0.01, "learning_rate": 4.16444207154298e-06, "loss": 1.1802, "step": 78 }, { "epoch": 0.01, "learning_rate": 4.217832354511479e-06, "loss": 1.1971, "step": 79 }, { "epoch": 0.01, "learning_rate": 4.2712226374799786e-06, "loss": 1.1843, "step": 80 }, { "epoch": 0.01, "learning_rate": 4.324612920448479e-06, "loss": 1.1816, "step": 81 }, { "epoch": 0.01, "learning_rate": 4.378003203416978e-06, "loss": 1.1293, "step": 82 }, { "epoch": 0.01, "learning_rate": 4.431393486385478e-06, "loss": 1.1142, "step": 83 }, { "epoch": 0.01, "learning_rate": 4.484783769353978e-06, "loss": 1.1774, "step": 84 }, { "epoch": 0.01, "learning_rate": 4.538174052322478e-06, "loss": 1.2019, "step": 85 }, { "epoch": 0.01, "learning_rate": 4.591564335290977e-06, "loss": 1.148, "step": 86 }, { "epoch": 0.01, "learning_rate": 4.644954618259477e-06, "loss": 1.172, "step": 87 }, { "epoch": 0.01, "learning_rate": 4.698344901227977e-06, "loss": 1.1457, "step": 88 }, { "epoch": 0.01, "learning_rate": 4.751735184196476e-06, "loss": 1.1821, "step": 89 }, { "epoch": 0.01, "learning_rate": 4.805125467164976e-06, "loss": 1.1379, "step": 90 }, { "epoch": 0.01, "learning_rate": 4.858515750133475e-06, "loss": 1.1764, "step": 91 }, { "epoch": 0.01, "learning_rate": 4.911906033101976e-06, "loss": 1.0792, "step": 92 }, { "epoch": 0.01, "learning_rate": 4.965296316070475e-06, "loss": 1.1235, "step": 93 }, { "epoch": 0.02, "learning_rate": 5.018686599038975e-06, "loss": 1.1348, "step": 94 }, { "epoch": 0.02, "learning_rate": 5.072076882007475e-06, "loss": 1.1725, "step": 95 }, { "epoch": 0.02, "learning_rate": 5.1254671649759744e-06, "loss": 1.2299, "step": 96 }, { "epoch": 0.02, "learning_rate": 5.178857447944474e-06, "loss": 1.1146, "step": 97 }, { "epoch": 0.02, "learning_rate": 5.232247730912974e-06, "loss": 1.1673, "step": 98 }, { "epoch": 0.02, "learning_rate": 5.2856380138814746e-06, "loss": 1.1528, "step": 99 }, { "epoch": 0.02, "learning_rate": 5.339028296849974e-06, "loss": 1.1818, "step": 100 }, { "epoch": 0.02, "learning_rate": 5.3924185798184735e-06, "loss": 1.2084, "step": 101 }, { "epoch": 0.02, "learning_rate": 5.445808862786973e-06, "loss": 1.2052, "step": 102 }, { "epoch": 0.02, "learning_rate": 5.499199145755473e-06, "loss": 1.2343, "step": 103 }, { "epoch": 0.02, "learning_rate": 5.552589428723973e-06, "loss": 1.1516, "step": 104 }, { "epoch": 0.02, "learning_rate": 5.605979711692472e-06, "loss": 1.1097, "step": 105 }, { "epoch": 0.02, "learning_rate": 5.659369994660972e-06, "loss": 1.1854, "step": 106 }, { "epoch": 0.02, "learning_rate": 5.712760277629472e-06, "loss": 1.0893, "step": 107 }, { "epoch": 0.02, "learning_rate": 5.7661505605979716e-06, "loss": 1.1322, "step": 108 }, { "epoch": 0.02, "learning_rate": 5.819540843566471e-06, "loss": 1.062, "step": 109 }, { "epoch": 0.02, "learning_rate": 5.872931126534971e-06, "loss": 1.1864, "step": 110 }, { "epoch": 0.02, "learning_rate": 5.926321409503471e-06, "loss": 1.1158, "step": 111 }, { "epoch": 0.02, "learning_rate": 5.97971169247197e-06, "loss": 1.2044, "step": 112 }, { "epoch": 0.02, "learning_rate": 6.03310197544047e-06, "loss": 1.1237, "step": 113 }, { "epoch": 0.02, "learning_rate": 6.08649225840897e-06, "loss": 1.1003, "step": 114 }, { "epoch": 0.02, "learning_rate": 6.13988254137747e-06, "loss": 1.1282, "step": 115 }, { "epoch": 0.02, "learning_rate": 6.193272824345969e-06, "loss": 1.3289, "step": 116 }, { "epoch": 0.02, "learning_rate": 6.2466631073144686e-06, "loss": 1.1743, "step": 117 }, { "epoch": 0.02, "learning_rate": 6.300053390282969e-06, "loss": 1.1749, "step": 118 }, { "epoch": 0.02, "learning_rate": 6.353443673251469e-06, "loss": 1.2537, "step": 119 }, { "epoch": 0.02, "learning_rate": 6.406833956219968e-06, "loss": 1.2066, "step": 120 }, { "epoch": 0.02, "learning_rate": 6.460224239188468e-06, "loss": 1.2335, "step": 121 }, { "epoch": 0.02, "learning_rate": 6.513614522156968e-06, "loss": 1.1238, "step": 122 }, { "epoch": 0.02, "learning_rate": 6.567004805125468e-06, "loss": 1.1625, "step": 123 }, { "epoch": 0.02, "learning_rate": 6.620395088093967e-06, "loss": 1.1897, "step": 124 }, { "epoch": 0.02, "learning_rate": 6.673785371062467e-06, "loss": 1.1744, "step": 125 }, { "epoch": 0.02, "learning_rate": 6.727175654030966e-06, "loss": 1.1743, "step": 126 }, { "epoch": 0.02, "learning_rate": 6.780565936999467e-06, "loss": 1.1517, "step": 127 }, { "epoch": 0.02, "learning_rate": 6.833956219967965e-06, "loss": 1.1954, "step": 128 }, { "epoch": 0.02, "learning_rate": 6.8873465029364665e-06, "loss": 1.1778, "step": 129 }, { "epoch": 0.02, "learning_rate": 6.940736785904965e-06, "loss": 1.0821, "step": 130 }, { "epoch": 0.02, "learning_rate": 6.9941270688734655e-06, "loss": 1.0958, "step": 131 }, { "epoch": 0.02, "learning_rate": 7.047517351841965e-06, "loss": 1.1217, "step": 132 }, { "epoch": 0.02, "learning_rate": 7.100907634810465e-06, "loss": 1.1351, "step": 133 }, { "epoch": 0.02, "learning_rate": 7.154297917778964e-06, "loss": 1.1304, "step": 134 }, { "epoch": 0.02, "learning_rate": 7.207688200747464e-06, "loss": 1.1677, "step": 135 }, { "epoch": 0.02, "learning_rate": 7.261078483715964e-06, "loss": 1.1134, "step": 136 }, { "epoch": 0.02, "learning_rate": 7.314468766684464e-06, "loss": 1.1907, "step": 137 }, { "epoch": 0.02, "learning_rate": 7.367859049652963e-06, "loss": 1.0656, "step": 138 }, { "epoch": 0.02, "learning_rate": 7.421249332621463e-06, "loss": 1.0599, "step": 139 }, { "epoch": 0.02, "learning_rate": 7.474639615589963e-06, "loss": 1.1851, "step": 140 }, { "epoch": 0.02, "learning_rate": 7.528029898558463e-06, "loss": 1.0748, "step": 141 }, { "epoch": 0.02, "learning_rate": 7.581420181526963e-06, "loss": 1.1282, "step": 142 }, { "epoch": 0.02, "learning_rate": 7.634810464495463e-06, "loss": 1.1742, "step": 143 }, { "epoch": 0.02, "learning_rate": 7.688200747463962e-06, "loss": 1.091, "step": 144 }, { "epoch": 0.02, "learning_rate": 7.741591030432462e-06, "loss": 1.0685, "step": 145 }, { "epoch": 0.02, "learning_rate": 7.794981313400963e-06, "loss": 1.0517, "step": 146 }, { "epoch": 0.02, "learning_rate": 7.84837159636946e-06, "loss": 1.0753, "step": 147 }, { "epoch": 0.02, "learning_rate": 7.901761879337962e-06, "loss": 1.1392, "step": 148 }, { "epoch": 0.02, "learning_rate": 7.955152162306461e-06, "loss": 1.1704, "step": 149 }, { "epoch": 0.02, "learning_rate": 8.00854244527496e-06, "loss": 1.2283, "step": 150 }, { "epoch": 0.02, "learning_rate": 8.06193272824346e-06, "loss": 1.0746, "step": 151 }, { "epoch": 0.02, "learning_rate": 8.11532301121196e-06, "loss": 1.1594, "step": 152 }, { "epoch": 0.02, "learning_rate": 8.168713294180459e-06, "loss": 1.0428, "step": 153 }, { "epoch": 0.02, "learning_rate": 8.22210357714896e-06, "loss": 1.1732, "step": 154 }, { "epoch": 0.02, "learning_rate": 8.275493860117458e-06, "loss": 1.0759, "step": 155 }, { "epoch": 0.02, "learning_rate": 8.32888414308596e-06, "loss": 1.1562, "step": 156 }, { "epoch": 0.03, "learning_rate": 8.382274426054459e-06, "loss": 1.1555, "step": 157 }, { "epoch": 0.03, "learning_rate": 8.435664709022958e-06, "loss": 1.0729, "step": 158 }, { "epoch": 0.03, "learning_rate": 8.489054991991458e-06, "loss": 1.0673, "step": 159 }, { "epoch": 0.03, "learning_rate": 8.542445274959957e-06, "loss": 1.1738, "step": 160 }, { "epoch": 0.03, "learning_rate": 8.595835557928457e-06, "loss": 1.2082, "step": 161 }, { "epoch": 0.03, "learning_rate": 8.649225840896958e-06, "loss": 1.0557, "step": 162 }, { "epoch": 0.03, "learning_rate": 8.702616123865456e-06, "loss": 1.1255, "step": 163 }, { "epoch": 0.03, "learning_rate": 8.756006406833957e-06, "loss": 1.0937, "step": 164 }, { "epoch": 0.03, "learning_rate": 8.809396689802456e-06, "loss": 1.1094, "step": 165 }, { "epoch": 0.03, "learning_rate": 8.862786972770956e-06, "loss": 1.1105, "step": 166 }, { "epoch": 0.03, "learning_rate": 8.916177255739455e-06, "loss": 1.1031, "step": 167 }, { "epoch": 0.03, "learning_rate": 8.969567538707956e-06, "loss": 1.1348, "step": 168 }, { "epoch": 0.03, "learning_rate": 9.022957821676456e-06, "loss": 1.1108, "step": 169 }, { "epoch": 0.03, "learning_rate": 9.076348104644955e-06, "loss": 1.0686, "step": 170 }, { "epoch": 0.03, "learning_rate": 9.129738387613455e-06, "loss": 1.1079, "step": 171 }, { "epoch": 0.03, "learning_rate": 9.183128670581954e-06, "loss": 1.1015, "step": 172 }, { "epoch": 0.03, "learning_rate": 9.236518953550455e-06, "loss": 1.0701, "step": 173 }, { "epoch": 0.03, "learning_rate": 9.289909236518953e-06, "loss": 1.0625, "step": 174 }, { "epoch": 0.03, "learning_rate": 9.343299519487454e-06, "loss": 1.1531, "step": 175 }, { "epoch": 0.03, "learning_rate": 9.396689802455954e-06, "loss": 1.0979, "step": 176 }, { "epoch": 0.03, "learning_rate": 9.450080085424453e-06, "loss": 1.1268, "step": 177 }, { "epoch": 0.03, "learning_rate": 9.503470368392953e-06, "loss": 1.0879, "step": 178 }, { "epoch": 0.03, "learning_rate": 9.556860651361452e-06, "loss": 1.1643, "step": 179 }, { "epoch": 0.03, "learning_rate": 9.610250934329952e-06, "loss": 1.0697, "step": 180 }, { "epoch": 0.03, "learning_rate": 9.663641217298453e-06, "loss": 1.1041, "step": 181 }, { "epoch": 0.03, "learning_rate": 9.71703150026695e-06, "loss": 1.108, "step": 182 }, { "epoch": 0.03, "learning_rate": 9.770421783235452e-06, "loss": 1.1606, "step": 183 }, { "epoch": 0.03, "learning_rate": 9.823812066203951e-06, "loss": 1.0562, "step": 184 }, { "epoch": 0.03, "learning_rate": 9.87720234917245e-06, "loss": 1.0988, "step": 185 }, { "epoch": 0.03, "learning_rate": 9.93059263214095e-06, "loss": 1.0946, "step": 186 }, { "epoch": 0.03, "learning_rate": 9.983982915109451e-06, "loss": 1.1646, "step": 187 }, { "epoch": 0.03, "learning_rate": 1.003737319807795e-05, "loss": 1.1548, "step": 188 }, { "epoch": 0.03, "learning_rate": 1.009076348104645e-05, "loss": 1.1023, "step": 189 }, { "epoch": 0.03, "learning_rate": 1.014415376401495e-05, "loss": 1.1327, "step": 190 }, { "epoch": 0.03, "learning_rate": 1.019754404698345e-05, "loss": 1.0766, "step": 191 }, { "epoch": 0.03, "learning_rate": 1.0250934329951949e-05, "loss": 1.0923, "step": 192 }, { "epoch": 0.03, "learning_rate": 1.0304324612920448e-05, "loss": 1.1047, "step": 193 }, { "epoch": 0.03, "learning_rate": 1.0357714895888948e-05, "loss": 1.0593, "step": 194 }, { "epoch": 0.03, "learning_rate": 1.0411105178857449e-05, "loss": 1.1411, "step": 195 }, { "epoch": 0.03, "learning_rate": 1.0464495461825948e-05, "loss": 1.1183, "step": 196 }, { "epoch": 0.03, "learning_rate": 1.0517885744794448e-05, "loss": 1.1017, "step": 197 }, { "epoch": 0.03, "learning_rate": 1.0571276027762949e-05, "loss": 1.023, "step": 198 }, { "epoch": 0.03, "learning_rate": 1.0624666310731447e-05, "loss": 1.0716, "step": 199 }, { "epoch": 0.03, "learning_rate": 1.0678056593699948e-05, "loss": 1.151, "step": 200 }, { "epoch": 0.03, "learning_rate": 1.0731446876668448e-05, "loss": 1.1029, "step": 201 }, { "epoch": 0.03, "learning_rate": 1.0784837159636947e-05, "loss": 1.0716, "step": 202 }, { "epoch": 0.03, "learning_rate": 1.0838227442605447e-05, "loss": 1.1097, "step": 203 }, { "epoch": 0.03, "learning_rate": 1.0891617725573946e-05, "loss": 1.076, "step": 204 }, { "epoch": 0.03, "learning_rate": 1.0945008008542445e-05, "loss": 1.1012, "step": 205 }, { "epoch": 0.03, "learning_rate": 1.0998398291510947e-05, "loss": 1.1681, "step": 206 }, { "epoch": 0.03, "learning_rate": 1.1051788574479444e-05, "loss": 1.0501, "step": 207 }, { "epoch": 0.03, "learning_rate": 1.1105178857447946e-05, "loss": 1.1424, "step": 208 }, { "epoch": 0.03, "learning_rate": 1.1158569140416445e-05, "loss": 1.02, "step": 209 }, { "epoch": 0.03, "learning_rate": 1.1211959423384945e-05, "loss": 1.0938, "step": 210 }, { "epoch": 0.03, "learning_rate": 1.1265349706353444e-05, "loss": 1.0524, "step": 211 }, { "epoch": 0.03, "learning_rate": 1.1318739989321944e-05, "loss": 1.1265, "step": 212 }, { "epoch": 0.03, "learning_rate": 1.1372130272290443e-05, "loss": 1.1095, "step": 213 }, { "epoch": 0.03, "learning_rate": 1.1425520555258944e-05, "loss": 1.1274, "step": 214 }, { "epoch": 0.03, "learning_rate": 1.1478910838227442e-05, "loss": 1.0425, "step": 215 }, { "epoch": 0.03, "learning_rate": 1.1532301121195943e-05, "loss": 1.1197, "step": 216 }, { "epoch": 0.03, "learning_rate": 1.1585691404164443e-05, "loss": 1.0442, "step": 217 }, { "epoch": 0.03, "learning_rate": 1.1639081687132942e-05, "loss": 1.0514, "step": 218 }, { "epoch": 0.04, "learning_rate": 1.1692471970101442e-05, "loss": 1.0894, "step": 219 }, { "epoch": 0.04, "learning_rate": 1.1745862253069943e-05, "loss": 1.1156, "step": 220 }, { "epoch": 0.04, "learning_rate": 1.179925253603844e-05, "loss": 1.0904, "step": 221 }, { "epoch": 0.04, "learning_rate": 1.1852642819006942e-05, "loss": 1.0571, "step": 222 }, { "epoch": 0.04, "learning_rate": 1.1906033101975441e-05, "loss": 1.13, "step": 223 }, { "epoch": 0.04, "learning_rate": 1.195942338494394e-05, "loss": 1.0712, "step": 224 }, { "epoch": 0.04, "learning_rate": 1.2012813667912442e-05, "loss": 1.0788, "step": 225 }, { "epoch": 0.04, "learning_rate": 1.206620395088094e-05, "loss": 0.9675, "step": 226 }, { "epoch": 0.04, "learning_rate": 1.211959423384944e-05, "loss": 1.0643, "step": 227 }, { "epoch": 0.04, "learning_rate": 1.217298451681794e-05, "loss": 1.0711, "step": 228 }, { "epoch": 0.04, "learning_rate": 1.222637479978644e-05, "loss": 1.0582, "step": 229 }, { "epoch": 0.04, "learning_rate": 1.227976508275494e-05, "loss": 1.0112, "step": 230 }, { "epoch": 0.04, "learning_rate": 1.2333155365723439e-05, "loss": 1.0713, "step": 231 }, { "epoch": 0.04, "learning_rate": 1.2386545648691938e-05, "loss": 1.07, "step": 232 }, { "epoch": 0.04, "learning_rate": 1.243993593166044e-05, "loss": 1.0944, "step": 233 }, { "epoch": 0.04, "learning_rate": 1.2493326214628937e-05, "loss": 1.0853, "step": 234 }, { "epoch": 0.04, "learning_rate": 1.2546716497597438e-05, "loss": 1.0051, "step": 235 }, { "epoch": 0.04, "learning_rate": 1.2600106780565938e-05, "loss": 1.0493, "step": 236 }, { "epoch": 0.04, "learning_rate": 1.2653497063534436e-05, "loss": 1.0778, "step": 237 }, { "epoch": 0.04, "learning_rate": 1.2706887346502938e-05, "loss": 1.1361, "step": 238 }, { "epoch": 0.04, "learning_rate": 1.2760277629471438e-05, "loss": 1.0754, "step": 239 }, { "epoch": 0.04, "learning_rate": 1.2813667912439936e-05, "loss": 1.04, "step": 240 }, { "epoch": 0.04, "learning_rate": 1.2867058195408435e-05, "loss": 1.1126, "step": 241 }, { "epoch": 0.04, "learning_rate": 1.2920448478376936e-05, "loss": 1.0009, "step": 242 }, { "epoch": 0.04, "learning_rate": 1.2973838761345436e-05, "loss": 1.18, "step": 243 }, { "epoch": 0.04, "learning_rate": 1.3027229044313935e-05, "loss": 1.0052, "step": 244 }, { "epoch": 0.04, "learning_rate": 1.3080619327282435e-05, "loss": 0.9884, "step": 245 }, { "epoch": 0.04, "learning_rate": 1.3134009610250936e-05, "loss": 1.0609, "step": 246 }, { "epoch": 0.04, "learning_rate": 1.3187399893219435e-05, "loss": 1.1313, "step": 247 }, { "epoch": 0.04, "learning_rate": 1.3240790176187933e-05, "loss": 1.0079, "step": 248 }, { "epoch": 0.04, "learning_rate": 1.3294180459156433e-05, "loss": 1.1126, "step": 249 }, { "epoch": 0.04, "learning_rate": 1.3347570742124934e-05, "loss": 1.0631, "step": 250 }, { "epoch": 0.04, "learning_rate": 1.3400961025093433e-05, "loss": 0.9945, "step": 251 }, { "epoch": 0.04, "learning_rate": 1.3454351308061933e-05, "loss": 1.0249, "step": 252 }, { "epoch": 0.04, "learning_rate": 1.3507741591030434e-05, "loss": 1.024, "step": 253 }, { "epoch": 0.04, "learning_rate": 1.3561131873998933e-05, "loss": 1.0913, "step": 254 }, { "epoch": 0.04, "learning_rate": 1.3614522156967433e-05, "loss": 1.0676, "step": 255 }, { "epoch": 0.04, "learning_rate": 1.366791243993593e-05, "loss": 1.0162, "step": 256 }, { "epoch": 0.04, "learning_rate": 1.3721302722904434e-05, "loss": 1.0917, "step": 257 }, { "epoch": 0.04, "learning_rate": 1.3774693005872933e-05, "loss": 1.2169, "step": 258 }, { "epoch": 0.04, "learning_rate": 1.382808328884143e-05, "loss": 1.0521, "step": 259 }, { "epoch": 0.04, "learning_rate": 1.388147357180993e-05, "loss": 1.0908, "step": 260 }, { "epoch": 0.04, "learning_rate": 1.3934863854778431e-05, "loss": 0.9955, "step": 261 }, { "epoch": 0.04, "learning_rate": 1.3988254137746931e-05, "loss": 1.0699, "step": 262 }, { "epoch": 0.04, "learning_rate": 1.404164442071543e-05, "loss": 1.0566, "step": 263 }, { "epoch": 0.04, "learning_rate": 1.409503470368393e-05, "loss": 1.1868, "step": 264 }, { "epoch": 0.04, "learning_rate": 1.4148424986652431e-05, "loss": 1.0764, "step": 265 }, { "epoch": 0.04, "learning_rate": 1.420181526962093e-05, "loss": 0.9607, "step": 266 }, { "epoch": 0.04, "learning_rate": 1.4255205552589428e-05, "loss": 1.1271, "step": 267 }, { "epoch": 0.04, "learning_rate": 1.4308595835557928e-05, "loss": 1.0939, "step": 268 }, { "epoch": 0.04, "learning_rate": 1.4361986118526429e-05, "loss": 1.0101, "step": 269 }, { "epoch": 0.04, "learning_rate": 1.4415376401494928e-05, "loss": 0.9672, "step": 270 }, { "epoch": 0.04, "learning_rate": 1.4468766684463428e-05, "loss": 1.0908, "step": 271 }, { "epoch": 0.04, "learning_rate": 1.4522156967431927e-05, "loss": 1.0368, "step": 272 }, { "epoch": 0.04, "learning_rate": 1.4575547250400429e-05, "loss": 1.0456, "step": 273 }, { "epoch": 0.04, "learning_rate": 1.4628937533368928e-05, "loss": 1.0987, "step": 274 }, { "epoch": 0.04, "learning_rate": 1.4682327816337428e-05, "loss": 0.9878, "step": 275 }, { "epoch": 0.04, "learning_rate": 1.4735718099305925e-05, "loss": 1.0616, "step": 276 }, { "epoch": 0.04, "learning_rate": 1.4789108382274428e-05, "loss": 1.0383, "step": 277 }, { "epoch": 0.04, "learning_rate": 1.4842498665242926e-05, "loss": 1.0865, "step": 278 }, { "epoch": 0.04, "learning_rate": 1.4895888948211425e-05, "loss": 1.0082, "step": 279 }, { "epoch": 0.04, "learning_rate": 1.4949279231179927e-05, "loss": 1.098, "step": 280 }, { "epoch": 0.05, "learning_rate": 1.5002669514148426e-05, "loss": 1.0613, "step": 281 }, { "epoch": 0.05, "learning_rate": 1.5056059797116926e-05, "loss": 1.1496, "step": 282 }, { "epoch": 0.05, "learning_rate": 1.5109450080085425e-05, "loss": 1.0308, "step": 283 }, { "epoch": 0.05, "learning_rate": 1.5162840363053926e-05, "loss": 1.0124, "step": 284 }, { "epoch": 0.05, "learning_rate": 1.5216230646022426e-05, "loss": 1.0436, "step": 285 }, { "epoch": 0.05, "learning_rate": 1.5269620928990925e-05, "loss": 1.0246, "step": 286 }, { "epoch": 0.05, "learning_rate": 1.5323011211959423e-05, "loss": 1.0441, "step": 287 }, { "epoch": 0.05, "learning_rate": 1.5376401494927924e-05, "loss": 1.0614, "step": 288 }, { "epoch": 0.05, "learning_rate": 1.5429791777896425e-05, "loss": 1.1014, "step": 289 }, { "epoch": 0.05, "learning_rate": 1.5483182060864923e-05, "loss": 1.0538, "step": 290 }, { "epoch": 0.05, "learning_rate": 1.553657234383342e-05, "loss": 1.1507, "step": 291 }, { "epoch": 0.05, "learning_rate": 1.5589962626801925e-05, "loss": 0.9903, "step": 292 }, { "epoch": 0.05, "learning_rate": 1.5643352909770423e-05, "loss": 1.0656, "step": 293 }, { "epoch": 0.05, "learning_rate": 1.569674319273892e-05, "loss": 0.987, "step": 294 }, { "epoch": 0.05, "learning_rate": 1.5750133475707422e-05, "loss": 1.0226, "step": 295 }, { "epoch": 0.05, "learning_rate": 1.5803523758675923e-05, "loss": 1.1493, "step": 296 }, { "epoch": 0.05, "learning_rate": 1.585691404164442e-05, "loss": 1.0324, "step": 297 }, { "epoch": 0.05, "learning_rate": 1.5910304324612922e-05, "loss": 1.0519, "step": 298 }, { "epoch": 0.05, "learning_rate": 1.596369460758142e-05, "loss": 1.1041, "step": 299 }, { "epoch": 0.05, "learning_rate": 1.601708489054992e-05, "loss": 1.0577, "step": 300 }, { "epoch": 0.05, "learning_rate": 1.607047517351842e-05, "loss": 1.2002, "step": 301 }, { "epoch": 0.05, "learning_rate": 1.612386545648692e-05, "loss": 1.0158, "step": 302 }, { "epoch": 0.05, "learning_rate": 1.6177255739455418e-05, "loss": 1.0336, "step": 303 }, { "epoch": 0.05, "learning_rate": 1.623064602242392e-05, "loss": 1.0575, "step": 304 }, { "epoch": 0.05, "learning_rate": 1.628403630539242e-05, "loss": 1.1031, "step": 305 }, { "epoch": 0.05, "learning_rate": 1.6337426588360918e-05, "loss": 1.0925, "step": 306 }, { "epoch": 0.05, "learning_rate": 1.6390816871329416e-05, "loss": 1.1075, "step": 307 }, { "epoch": 0.05, "learning_rate": 1.644420715429792e-05, "loss": 1.1262, "step": 308 }, { "epoch": 0.05, "learning_rate": 1.6497597437266418e-05, "loss": 1.0076, "step": 309 }, { "epoch": 0.05, "learning_rate": 1.6550987720234916e-05, "loss": 1.0407, "step": 310 }, { "epoch": 0.05, "learning_rate": 1.660437800320342e-05, "loss": 0.9923, "step": 311 }, { "epoch": 0.05, "learning_rate": 1.665776828617192e-05, "loss": 1.0172, "step": 312 }, { "epoch": 0.05, "learning_rate": 1.6711158569140416e-05, "loss": 1.0671, "step": 313 }, { "epoch": 0.05, "learning_rate": 1.6764548852108917e-05, "loss": 1.0814, "step": 314 }, { "epoch": 0.05, "learning_rate": 1.681793913507742e-05, "loss": 1.0052, "step": 315 }, { "epoch": 0.05, "learning_rate": 1.6871329418045916e-05, "loss": 1.1241, "step": 316 }, { "epoch": 0.05, "learning_rate": 1.6924719701014417e-05, "loss": 1.0465, "step": 317 }, { "epoch": 0.05, "learning_rate": 1.6978109983982915e-05, "loss": 1.0566, "step": 318 }, { "epoch": 0.05, "learning_rate": 1.7031500266951416e-05, "loss": 1.0607, "step": 319 }, { "epoch": 0.05, "learning_rate": 1.7084890549919914e-05, "loss": 1.0419, "step": 320 }, { "epoch": 0.05, "learning_rate": 1.7138280832888415e-05, "loss": 1.0354, "step": 321 }, { "epoch": 0.05, "learning_rate": 1.7191671115856913e-05, "loss": 1.0066, "step": 322 }, { "epoch": 0.05, "learning_rate": 1.7245061398825414e-05, "loss": 1.0625, "step": 323 }, { "epoch": 0.05, "learning_rate": 1.7298451681793916e-05, "loss": 1.0686, "step": 324 }, { "epoch": 0.05, "learning_rate": 1.7351841964762413e-05, "loss": 1.0389, "step": 325 }, { "epoch": 0.05, "learning_rate": 1.740523224773091e-05, "loss": 0.9976, "step": 326 }, { "epoch": 0.05, "learning_rate": 1.7458622530699416e-05, "loss": 1.081, "step": 327 }, { "epoch": 0.05, "learning_rate": 1.7512012813667913e-05, "loss": 0.9951, "step": 328 }, { "epoch": 0.05, "learning_rate": 1.756540309663641e-05, "loss": 1.0764, "step": 329 }, { "epoch": 0.05, "learning_rate": 1.7618793379604912e-05, "loss": 0.9625, "step": 330 }, { "epoch": 0.05, "learning_rate": 1.7672183662573414e-05, "loss": 1.1659, "step": 331 }, { "epoch": 0.05, "learning_rate": 1.772557394554191e-05, "loss": 1.0652, "step": 332 }, { "epoch": 0.05, "learning_rate": 1.7778964228510413e-05, "loss": 0.9364, "step": 333 }, { "epoch": 0.05, "learning_rate": 1.783235451147891e-05, "loss": 1.0629, "step": 334 }, { "epoch": 0.05, "learning_rate": 1.788574479444741e-05, "loss": 0.9194, "step": 335 }, { "epoch": 0.05, "learning_rate": 1.7939135077415913e-05, "loss": 1.0329, "step": 336 }, { "epoch": 0.05, "learning_rate": 1.799252536038441e-05, "loss": 1.078, "step": 337 }, { "epoch": 0.05, "learning_rate": 1.804591564335291e-05, "loss": 1.0876, "step": 338 }, { "epoch": 0.05, "learning_rate": 1.809930592632141e-05, "loss": 1.1503, "step": 339 }, { "epoch": 0.05, "learning_rate": 1.815269620928991e-05, "loss": 1.115, "step": 340 }, { "epoch": 0.05, "learning_rate": 1.820608649225841e-05, "loss": 1.059, "step": 341 }, { "epoch": 0.05, "learning_rate": 1.825947677522691e-05, "loss": 0.995, "step": 342 }, { "epoch": 0.05, "learning_rate": 1.831286705819541e-05, "loss": 0.9741, "step": 343 }, { "epoch": 0.06, "learning_rate": 1.836625734116391e-05, "loss": 1.042, "step": 344 }, { "epoch": 0.06, "learning_rate": 1.8419647624132406e-05, "loss": 1.0188, "step": 345 }, { "epoch": 0.06, "learning_rate": 1.847303790710091e-05, "loss": 1.0981, "step": 346 }, { "epoch": 0.06, "learning_rate": 1.852642819006941e-05, "loss": 1.0003, "step": 347 }, { "epoch": 0.06, "learning_rate": 1.8579818473037906e-05, "loss": 0.9777, "step": 348 }, { "epoch": 0.06, "learning_rate": 1.8633208756006408e-05, "loss": 0.967, "step": 349 }, { "epoch": 0.06, "learning_rate": 1.868659903897491e-05, "loss": 1.0018, "step": 350 }, { "epoch": 0.06, "learning_rate": 1.8739989321943407e-05, "loss": 1.0136, "step": 351 }, { "epoch": 0.06, "learning_rate": 1.8793379604911908e-05, "loss": 0.9827, "step": 352 }, { "epoch": 0.06, "learning_rate": 1.8846769887880405e-05, "loss": 1.0308, "step": 353 }, { "epoch": 0.06, "learning_rate": 1.8900160170848907e-05, "loss": 1.1211, "step": 354 }, { "epoch": 0.06, "learning_rate": 1.8953550453817408e-05, "loss": 0.9842, "step": 355 }, { "epoch": 0.06, "learning_rate": 1.9006940736785906e-05, "loss": 1.0412, "step": 356 }, { "epoch": 0.06, "learning_rate": 1.9060331019754403e-05, "loss": 1.0049, "step": 357 }, { "epoch": 0.06, "learning_rate": 1.9113721302722905e-05, "loss": 1.0469, "step": 358 }, { "epoch": 0.06, "learning_rate": 1.9167111585691406e-05, "loss": 1.0774, "step": 359 }, { "epoch": 0.06, "learning_rate": 1.9220501868659903e-05, "loss": 0.9836, "step": 360 }, { "epoch": 0.06, "learning_rate": 1.9273892151628405e-05, "loss": 1.0846, "step": 361 }, { "epoch": 0.06, "learning_rate": 1.9327282434596906e-05, "loss": 1.018, "step": 362 }, { "epoch": 0.06, "learning_rate": 1.9380672717565404e-05, "loss": 0.9799, "step": 363 }, { "epoch": 0.06, "learning_rate": 1.94340630005339e-05, "loss": 1.0674, "step": 364 }, { "epoch": 0.06, "learning_rate": 1.9487453283502406e-05, "loss": 0.9978, "step": 365 }, { "epoch": 0.06, "learning_rate": 1.9540843566470904e-05, "loss": 1.0238, "step": 366 }, { "epoch": 0.06, "learning_rate": 1.95942338494394e-05, "loss": 1.0183, "step": 367 }, { "epoch": 0.06, "learning_rate": 1.9647624132407903e-05, "loss": 1.0607, "step": 368 }, { "epoch": 0.06, "learning_rate": 1.9701014415376404e-05, "loss": 0.9441, "step": 369 }, { "epoch": 0.06, "learning_rate": 1.97544046983449e-05, "loss": 1.0906, "step": 370 }, { "epoch": 0.06, "learning_rate": 1.9807794981313403e-05, "loss": 0.9779, "step": 371 }, { "epoch": 0.06, "learning_rate": 1.98611852642819e-05, "loss": 0.9763, "step": 372 }, { "epoch": 0.06, "learning_rate": 1.9914575547250402e-05, "loss": 1.0098, "step": 373 }, { "epoch": 0.06, "learning_rate": 1.9967965830218903e-05, "loss": 0.9915, "step": 374 }, { "epoch": 0.06, "learning_rate": 2.00213561131874e-05, "loss": 1.0945, "step": 375 }, { "epoch": 0.06, "learning_rate": 2.00747463961559e-05, "loss": 1.0579, "step": 376 }, { "epoch": 0.06, "learning_rate": 2.01281366791244e-05, "loss": 0.9866, "step": 377 }, { "epoch": 0.06, "learning_rate": 2.01815269620929e-05, "loss": 0.9981, "step": 378 }, { "epoch": 0.06, "learning_rate": 2.02349172450614e-05, "loss": 0.9991, "step": 379 }, { "epoch": 0.06, "learning_rate": 2.02883075280299e-05, "loss": 1.0389, "step": 380 }, { "epoch": 0.06, "learning_rate": 2.03416978109984e-05, "loss": 1.0497, "step": 381 }, { "epoch": 0.06, "learning_rate": 2.03950880939669e-05, "loss": 1.0852, "step": 382 }, { "epoch": 0.06, "learning_rate": 2.0448478376935397e-05, "loss": 1.0107, "step": 383 }, { "epoch": 0.06, "learning_rate": 2.0501868659903898e-05, "loss": 0.9849, "step": 384 }, { "epoch": 0.06, "learning_rate": 2.05552589428724e-05, "loss": 1.0955, "step": 385 }, { "epoch": 0.06, "learning_rate": 2.0608649225840897e-05, "loss": 1.0022, "step": 386 }, { "epoch": 0.06, "learning_rate": 2.0662039508809398e-05, "loss": 1.0476, "step": 387 }, { "epoch": 0.06, "learning_rate": 2.0715429791777896e-05, "loss": 0.9823, "step": 388 }, { "epoch": 0.06, "learning_rate": 2.0768820074746397e-05, "loss": 1.0244, "step": 389 }, { "epoch": 0.06, "learning_rate": 2.0822210357714898e-05, "loss": 1.0278, "step": 390 }, { "epoch": 0.06, "learning_rate": 2.0875600640683396e-05, "loss": 1.0045, "step": 391 }, { "epoch": 0.06, "learning_rate": 2.0928990923651897e-05, "loss": 1.0968, "step": 392 }, { "epoch": 0.06, "learning_rate": 2.0982381206620398e-05, "loss": 1.0239, "step": 393 }, { "epoch": 0.06, "learning_rate": 2.1035771489588896e-05, "loss": 1.0429, "step": 394 }, { "epoch": 0.06, "learning_rate": 2.1089161772557394e-05, "loss": 0.9409, "step": 395 }, { "epoch": 0.06, "learning_rate": 2.1142552055525898e-05, "loss": 0.9477, "step": 396 }, { "epoch": 0.06, "learning_rate": 2.1195942338494396e-05, "loss": 0.9529, "step": 397 }, { "epoch": 0.06, "learning_rate": 2.1249332621462894e-05, "loss": 1.0308, "step": 398 }, { "epoch": 0.06, "learning_rate": 2.1302722904431395e-05, "loss": 1.0132, "step": 399 }, { "epoch": 0.06, "learning_rate": 2.1356113187399896e-05, "loss": 1.0835, "step": 400 }, { "epoch": 0.06, "learning_rate": 2.1409503470368394e-05, "loss": 1.0512, "step": 401 }, { "epoch": 0.06, "learning_rate": 2.1462893753336895e-05, "loss": 0.9652, "step": 402 }, { "epoch": 0.06, "learning_rate": 2.1516284036305393e-05, "loss": 1.0123, "step": 403 }, { "epoch": 0.06, "learning_rate": 2.1569674319273894e-05, "loss": 1.0023, "step": 404 }, { "epoch": 0.06, "learning_rate": 2.1623064602242392e-05, "loss": 1.0651, "step": 405 }, { "epoch": 0.07, "learning_rate": 2.1676454885210893e-05, "loss": 0.9424, "step": 406 }, { "epoch": 0.07, "learning_rate": 2.172984516817939e-05, "loss": 1.043, "step": 407 }, { "epoch": 0.07, "learning_rate": 2.1783235451147892e-05, "loss": 0.966, "step": 408 }, { "epoch": 0.07, "learning_rate": 2.1836625734116393e-05, "loss": 1.0288, "step": 409 }, { "epoch": 0.07, "learning_rate": 2.189001601708489e-05, "loss": 0.9396, "step": 410 }, { "epoch": 0.07, "learning_rate": 2.194340630005339e-05, "loss": 0.979, "step": 411 }, { "epoch": 0.07, "learning_rate": 2.1996796583021893e-05, "loss": 1.0011, "step": 412 }, { "epoch": 0.07, "learning_rate": 2.205018686599039e-05, "loss": 0.9889, "step": 413 }, { "epoch": 0.07, "learning_rate": 2.210357714895889e-05, "loss": 0.9923, "step": 414 }, { "epoch": 0.07, "learning_rate": 2.215696743192739e-05, "loss": 0.9919, "step": 415 }, { "epoch": 0.07, "learning_rate": 2.221035771489589e-05, "loss": 1.0288, "step": 416 }, { "epoch": 0.07, "learning_rate": 2.226374799786439e-05, "loss": 0.9898, "step": 417 }, { "epoch": 0.07, "learning_rate": 2.231713828083289e-05, "loss": 1.044, "step": 418 }, { "epoch": 0.07, "learning_rate": 2.2370528563801388e-05, "loss": 0.9967, "step": 419 }, { "epoch": 0.07, "learning_rate": 2.242391884676989e-05, "loss": 1.059, "step": 420 }, { "epoch": 0.07, "learning_rate": 2.247730912973839e-05, "loss": 0.9283, "step": 421 }, { "epoch": 0.07, "learning_rate": 2.2530699412706888e-05, "loss": 1.0363, "step": 422 }, { "epoch": 0.07, "learning_rate": 2.258408969567539e-05, "loss": 1.0245, "step": 423 }, { "epoch": 0.07, "learning_rate": 2.2637479978643887e-05, "loss": 0.9384, "step": 424 }, { "epoch": 0.07, "learning_rate": 2.2690870261612388e-05, "loss": 1.0581, "step": 425 }, { "epoch": 0.07, "learning_rate": 2.2744260544580886e-05, "loss": 1.131, "step": 426 }, { "epoch": 0.07, "learning_rate": 2.2797650827549387e-05, "loss": 1.0637, "step": 427 }, { "epoch": 0.07, "learning_rate": 2.285104111051789e-05, "loss": 1.0231, "step": 428 }, { "epoch": 0.07, "learning_rate": 2.2904431393486386e-05, "loss": 1.004, "step": 429 }, { "epoch": 0.07, "learning_rate": 2.2957821676454884e-05, "loss": 0.9352, "step": 430 }, { "epoch": 0.07, "learning_rate": 2.301121195942339e-05, "loss": 0.9402, "step": 431 }, { "epoch": 0.07, "learning_rate": 2.3064602242391886e-05, "loss": 0.9782, "step": 432 }, { "epoch": 0.07, "learning_rate": 2.3117992525360384e-05, "loss": 1.055, "step": 433 }, { "epoch": 0.07, "learning_rate": 2.3171382808328885e-05, "loss": 0.9743, "step": 434 }, { "epoch": 0.07, "learning_rate": 2.3224773091297386e-05, "loss": 0.9446, "step": 435 }, { "epoch": 0.07, "learning_rate": 2.3278163374265884e-05, "loss": 1.0707, "step": 436 }, { "epoch": 0.07, "learning_rate": 2.3331553657234385e-05, "loss": 1.0723, "step": 437 }, { "epoch": 0.07, "learning_rate": 2.3384943940202883e-05, "loss": 0.9433, "step": 438 }, { "epoch": 0.07, "learning_rate": 2.3438334223171384e-05, "loss": 0.9362, "step": 439 }, { "epoch": 0.07, "learning_rate": 2.3491724506139885e-05, "loss": 0.9979, "step": 440 }, { "epoch": 0.07, "learning_rate": 2.3545114789108383e-05, "loss": 0.9595, "step": 441 }, { "epoch": 0.07, "learning_rate": 2.359850507207688e-05, "loss": 1.0579, "step": 442 }, { "epoch": 0.07, "learning_rate": 2.3651895355045382e-05, "loss": 0.9879, "step": 443 }, { "epoch": 0.07, "learning_rate": 2.3705285638013883e-05, "loss": 0.9259, "step": 444 }, { "epoch": 0.07, "learning_rate": 2.375867592098238e-05, "loss": 1.0557, "step": 445 }, { "epoch": 0.07, "learning_rate": 2.3812066203950882e-05, "loss": 1.0147, "step": 446 }, { "epoch": 0.07, "learning_rate": 2.3865456486919383e-05, "loss": 1.045, "step": 447 }, { "epoch": 0.07, "learning_rate": 2.391884676988788e-05, "loss": 1.0295, "step": 448 }, { "epoch": 0.07, "learning_rate": 2.397223705285638e-05, "loss": 1.0099, "step": 449 }, { "epoch": 0.07, "learning_rate": 2.4025627335824884e-05, "loss": 0.9608, "step": 450 }, { "epoch": 0.07, "learning_rate": 2.407901761879338e-05, "loss": 1.0677, "step": 451 }, { "epoch": 0.07, "learning_rate": 2.413240790176188e-05, "loss": 0.9676, "step": 452 }, { "epoch": 0.07, "learning_rate": 2.418579818473038e-05, "loss": 1.0612, "step": 453 }, { "epoch": 0.07, "learning_rate": 2.423918846769888e-05, "loss": 0.9387, "step": 454 }, { "epoch": 0.07, "learning_rate": 2.429257875066738e-05, "loss": 1.0215, "step": 455 }, { "epoch": 0.07, "learning_rate": 2.434596903363588e-05, "loss": 1.0495, "step": 456 }, { "epoch": 0.07, "learning_rate": 2.4399359316604378e-05, "loss": 1.0048, "step": 457 }, { "epoch": 0.07, "learning_rate": 2.445274959957288e-05, "loss": 1.0405, "step": 458 }, { "epoch": 0.07, "learning_rate": 2.450613988254138e-05, "loss": 1.0578, "step": 459 }, { "epoch": 0.07, "learning_rate": 2.455953016550988e-05, "loss": 0.9904, "step": 460 }, { "epoch": 0.07, "learning_rate": 2.4612920448478376e-05, "loss": 0.8538, "step": 461 }, { "epoch": 0.07, "learning_rate": 2.4666310731446877e-05, "loss": 1.0561, "step": 462 }, { "epoch": 0.07, "learning_rate": 2.471970101441538e-05, "loss": 0.9915, "step": 463 }, { "epoch": 0.07, "learning_rate": 2.4773091297383876e-05, "loss": 0.9869, "step": 464 }, { "epoch": 0.07, "learning_rate": 2.4826481580352377e-05, "loss": 0.9399, "step": 465 }, { "epoch": 0.07, "learning_rate": 2.487987186332088e-05, "loss": 1.1012, "step": 466 }, { "epoch": 0.07, "learning_rate": 2.4933262146289376e-05, "loss": 0.9271, "step": 467 }, { "epoch": 0.07, "learning_rate": 2.4986652429257874e-05, "loss": 0.9935, "step": 468 }, { "epoch": 0.08, "learning_rate": 2.5040042712226375e-05, "loss": 0.9909, "step": 469 }, { "epoch": 0.08, "learning_rate": 2.5093432995194877e-05, "loss": 1.0379, "step": 470 }, { "epoch": 0.08, "learning_rate": 2.5146823278163374e-05, "loss": 0.9358, "step": 471 }, { "epoch": 0.08, "learning_rate": 2.5200213561131876e-05, "loss": 1.0726, "step": 472 }, { "epoch": 0.08, "learning_rate": 2.5253603844100377e-05, "loss": 1.0247, "step": 473 }, { "epoch": 0.08, "learning_rate": 2.530699412706887e-05, "loss": 1.023, "step": 474 }, { "epoch": 0.08, "learning_rate": 2.5360384410037376e-05, "loss": 0.9002, "step": 475 }, { "epoch": 0.08, "learning_rate": 2.5413774693005877e-05, "loss": 1.0014, "step": 476 }, { "epoch": 0.08, "learning_rate": 2.546716497597437e-05, "loss": 1.0119, "step": 477 }, { "epoch": 0.08, "learning_rate": 2.5520555258942876e-05, "loss": 0.9875, "step": 478 }, { "epoch": 0.08, "learning_rate": 2.5573945541911377e-05, "loss": 1.0277, "step": 479 }, { "epoch": 0.08, "learning_rate": 2.562733582487987e-05, "loss": 1.0462, "step": 480 }, { "epoch": 0.08, "learning_rate": 2.5680726107848373e-05, "loss": 0.9701, "step": 481 }, { "epoch": 0.08, "learning_rate": 2.573411639081687e-05, "loss": 1.1017, "step": 482 }, { "epoch": 0.08, "learning_rate": 2.578750667378537e-05, "loss": 1.0511, "step": 483 }, { "epoch": 0.08, "learning_rate": 2.5840896956753873e-05, "loss": 1.097, "step": 484 }, { "epoch": 0.08, "learning_rate": 2.589428723972237e-05, "loss": 0.9989, "step": 485 }, { "epoch": 0.08, "learning_rate": 2.594767752269087e-05, "loss": 0.8801, "step": 486 }, { "epoch": 0.08, "learning_rate": 2.6001067805659373e-05, "loss": 1.0247, "step": 487 }, { "epoch": 0.08, "learning_rate": 2.605445808862787e-05, "loss": 1.002, "step": 488 }, { "epoch": 0.08, "learning_rate": 2.6107848371596372e-05, "loss": 0.9389, "step": 489 }, { "epoch": 0.08, "learning_rate": 2.616123865456487e-05, "loss": 0.9997, "step": 490 }, { "epoch": 0.08, "learning_rate": 2.621462893753337e-05, "loss": 0.9874, "step": 491 }, { "epoch": 0.08, "learning_rate": 2.6268019220501872e-05, "loss": 1.034, "step": 492 }, { "epoch": 0.08, "learning_rate": 2.6321409503470366e-05, "loss": 1.0375, "step": 493 }, { "epoch": 0.08, "learning_rate": 2.637479978643887e-05, "loss": 0.9875, "step": 494 }, { "epoch": 0.08, "learning_rate": 2.6428190069407372e-05, "loss": 1.0245, "step": 495 }, { "epoch": 0.08, "learning_rate": 2.6481580352375866e-05, "loss": 1.071, "step": 496 }, { "epoch": 0.08, "learning_rate": 2.653497063534437e-05, "loss": 1.0071, "step": 497 }, { "epoch": 0.08, "learning_rate": 2.6588360918312865e-05, "loss": 0.929, "step": 498 }, { "epoch": 0.08, "learning_rate": 2.6641751201281366e-05, "loss": 0.9934, "step": 499 }, { "epoch": 0.08, "learning_rate": 2.6695141484249868e-05, "loss": 1.0693, "step": 500 }, { "epoch": 0.08, "learning_rate": 2.6748531767218365e-05, "loss": 0.9838, "step": 501 }, { "epoch": 0.08, "learning_rate": 2.6801922050186867e-05, "loss": 0.981, "step": 502 }, { "epoch": 0.08, "learning_rate": 2.6855312333155368e-05, "loss": 0.9439, "step": 503 }, { "epoch": 0.08, "learning_rate": 2.6908702616123866e-05, "loss": 1.0213, "step": 504 }, { "epoch": 0.08, "learning_rate": 2.6962092899092367e-05, "loss": 0.9804, "step": 505 }, { "epoch": 0.08, "learning_rate": 2.7015483182060868e-05, "loss": 1.016, "step": 506 }, { "epoch": 0.08, "learning_rate": 2.7068873465029366e-05, "loss": 0.9172, "step": 507 }, { "epoch": 0.08, "learning_rate": 2.7122263747997867e-05, "loss": 0.8566, "step": 508 }, { "epoch": 0.08, "learning_rate": 2.7175654030966365e-05, "loss": 1.0037, "step": 509 }, { "epoch": 0.08, "learning_rate": 2.7229044313934866e-05, "loss": 0.968, "step": 510 }, { "epoch": 0.08, "learning_rate": 2.7282434596903367e-05, "loss": 0.9892, "step": 511 }, { "epoch": 0.08, "learning_rate": 2.733582487987186e-05, "loss": 0.9924, "step": 512 }, { "epoch": 0.08, "learning_rate": 2.7389215162840366e-05, "loss": 1.053, "step": 513 }, { "epoch": 0.08, "learning_rate": 2.7442605445808867e-05, "loss": 1.035, "step": 514 }, { "epoch": 0.08, "learning_rate": 2.749599572877736e-05, "loss": 0.9516, "step": 515 }, { "epoch": 0.08, "learning_rate": 2.7549386011745866e-05, "loss": 0.9142, "step": 516 }, { "epoch": 0.08, "learning_rate": 2.760277629471436e-05, "loss": 0.9729, "step": 517 }, { "epoch": 0.08, "learning_rate": 2.765616657768286e-05, "loss": 1.0747, "step": 518 }, { "epoch": 0.08, "learning_rate": 2.7709556860651363e-05, "loss": 1.0343, "step": 519 }, { "epoch": 0.08, "learning_rate": 2.776294714361986e-05, "loss": 0.9445, "step": 520 }, { "epoch": 0.08, "learning_rate": 2.7816337426588362e-05, "loss": 1.063, "step": 521 }, { "epoch": 0.08, "learning_rate": 2.7869727709556863e-05, "loss": 0.9257, "step": 522 }, { "epoch": 0.08, "learning_rate": 2.792311799252536e-05, "loss": 1.0501, "step": 523 }, { "epoch": 0.08, "learning_rate": 2.7976508275493862e-05, "loss": 1.0185, "step": 524 }, { "epoch": 0.08, "learning_rate": 2.802989855846236e-05, "loss": 0.9675, "step": 525 }, { "epoch": 0.08, "learning_rate": 2.808328884143086e-05, "loss": 1.0637, "step": 526 }, { "epoch": 0.08, "learning_rate": 2.8136679124399362e-05, "loss": 0.9372, "step": 527 }, { "epoch": 0.08, "learning_rate": 2.819006940736786e-05, "loss": 0.933, "step": 528 }, { "epoch": 0.08, "learning_rate": 2.824345969033636e-05, "loss": 0.9605, "step": 529 }, { "epoch": 0.08, "learning_rate": 2.8296849973304862e-05, "loss": 1.029, "step": 530 }, { "epoch": 0.09, "learning_rate": 2.8350240256273357e-05, "loss": 0.9839, "step": 531 }, { "epoch": 0.09, "learning_rate": 2.840363053924186e-05, "loss": 1.0616, "step": 532 }, { "epoch": 0.09, "learning_rate": 2.8457020822210362e-05, "loss": 1.044, "step": 533 }, { "epoch": 0.09, "learning_rate": 2.8510411105178857e-05, "loss": 0.9752, "step": 534 }, { "epoch": 0.09, "learning_rate": 2.856380138814736e-05, "loss": 0.9103, "step": 535 }, { "epoch": 0.09, "learning_rate": 2.8617191671115856e-05, "loss": 0.9771, "step": 536 }, { "epoch": 0.09, "learning_rate": 2.8670581954084357e-05, "loss": 0.9663, "step": 537 }, { "epoch": 0.09, "learning_rate": 2.8723972237052858e-05, "loss": 0.9067, "step": 538 }, { "epoch": 0.09, "learning_rate": 2.8777362520021356e-05, "loss": 0.9152, "step": 539 }, { "epoch": 0.09, "learning_rate": 2.8830752802989857e-05, "loss": 0.9933, "step": 540 }, { "epoch": 0.09, "learning_rate": 2.8884143085958358e-05, "loss": 0.9518, "step": 541 }, { "epoch": 0.09, "learning_rate": 2.8937533368926856e-05, "loss": 0.9851, "step": 542 }, { "epoch": 0.09, "learning_rate": 2.8990923651895357e-05, "loss": 1.061, "step": 543 }, { "epoch": 0.09, "learning_rate": 2.9044313934863855e-05, "loss": 1.0664, "step": 544 }, { "epoch": 0.09, "learning_rate": 2.9097704217832356e-05, "loss": 1.0818, "step": 545 }, { "epoch": 0.09, "learning_rate": 2.9151094500800857e-05, "loss": 1.0569, "step": 546 }, { "epoch": 0.09, "learning_rate": 2.9204484783769355e-05, "loss": 1.0385, "step": 547 }, { "epoch": 0.09, "learning_rate": 2.9257875066737856e-05, "loss": 1.0313, "step": 548 }, { "epoch": 0.09, "learning_rate": 2.9311265349706357e-05, "loss": 0.9262, "step": 549 }, { "epoch": 0.09, "learning_rate": 2.9364655632674855e-05, "loss": 0.9578, "step": 550 }, { "epoch": 0.09, "learning_rate": 2.9418045915643356e-05, "loss": 1.0034, "step": 551 }, { "epoch": 0.09, "learning_rate": 2.947143619861185e-05, "loss": 0.9434, "step": 552 }, { "epoch": 0.09, "learning_rate": 2.9524826481580352e-05, "loss": 0.9338, "step": 553 }, { "epoch": 0.09, "learning_rate": 2.9578216764548856e-05, "loss": 1.0074, "step": 554 }, { "epoch": 0.09, "learning_rate": 2.963160704751735e-05, "loss": 0.975, "step": 555 }, { "epoch": 0.09, "learning_rate": 2.9684997330485852e-05, "loss": 0.9275, "step": 556 }, { "epoch": 0.09, "learning_rate": 2.9738387613454357e-05, "loss": 0.947, "step": 557 }, { "epoch": 0.09, "learning_rate": 2.979177789642285e-05, "loss": 1.0284, "step": 558 }, { "epoch": 0.09, "learning_rate": 2.9845168179391352e-05, "loss": 0.9675, "step": 559 }, { "epoch": 0.09, "learning_rate": 2.9898558462359853e-05, "loss": 0.9691, "step": 560 }, { "epoch": 0.09, "learning_rate": 2.995194874532835e-05, "loss": 1.0428, "step": 561 }, { "epoch": 0.09, "learning_rate": 3.0005339028296852e-05, "loss": 0.9198, "step": 562 }, { "epoch": 0.09, "learning_rate": 3.005872931126535e-05, "loss": 1.0094, "step": 563 }, { "epoch": 0.09, "learning_rate": 3.011211959423385e-05, "loss": 0.9981, "step": 564 }, { "epoch": 0.09, "learning_rate": 3.0165509877202352e-05, "loss": 0.9258, "step": 565 }, { "epoch": 0.09, "learning_rate": 3.021890016017085e-05, "loss": 0.9734, "step": 566 }, { "epoch": 0.09, "learning_rate": 3.027229044313935e-05, "loss": 1.0757, "step": 567 }, { "epoch": 0.09, "learning_rate": 3.0325680726107852e-05, "loss": 1.0308, "step": 568 }, { "epoch": 0.09, "learning_rate": 3.037907100907635e-05, "loss": 1.0368, "step": 569 }, { "epoch": 0.09, "learning_rate": 3.043246129204485e-05, "loss": 1.0567, "step": 570 }, { "epoch": 0.09, "learning_rate": 3.0485851575013346e-05, "loss": 0.9996, "step": 571 }, { "epoch": 0.09, "learning_rate": 3.053924185798185e-05, "loss": 0.9498, "step": 572 }, { "epoch": 0.09, "learning_rate": 3.059263214095035e-05, "loss": 0.9957, "step": 573 }, { "epoch": 0.09, "learning_rate": 3.0646022423918846e-05, "loss": 0.9993, "step": 574 }, { "epoch": 0.09, "learning_rate": 3.069941270688735e-05, "loss": 1.1335, "step": 575 }, { "epoch": 0.09, "learning_rate": 3.075280298985585e-05, "loss": 1.0311, "step": 576 }, { "epoch": 0.09, "learning_rate": 3.080619327282435e-05, "loss": 0.9749, "step": 577 }, { "epoch": 0.09, "learning_rate": 3.085958355579285e-05, "loss": 0.9368, "step": 578 }, { "epoch": 0.09, "learning_rate": 3.0912973838761345e-05, "loss": 0.9227, "step": 579 }, { "epoch": 0.09, "learning_rate": 3.0966364121729846e-05, "loss": 1.0203, "step": 580 }, { "epoch": 0.09, "learning_rate": 3.101975440469835e-05, "loss": 0.9876, "step": 581 }, { "epoch": 0.09, "learning_rate": 3.107314468766684e-05, "loss": 0.9182, "step": 582 }, { "epoch": 0.09, "learning_rate": 3.112653497063534e-05, "loss": 0.8816, "step": 583 }, { "epoch": 0.09, "learning_rate": 3.117992525360385e-05, "loss": 1.0131, "step": 584 }, { "epoch": 0.09, "learning_rate": 3.1233315536572345e-05, "loss": 0.9775, "step": 585 }, { "epoch": 0.09, "learning_rate": 3.1286705819540846e-05, "loss": 1.0039, "step": 586 }, { "epoch": 0.09, "learning_rate": 3.134009610250934e-05, "loss": 1.0131, "step": 587 }, { "epoch": 0.09, "learning_rate": 3.139348638547784e-05, "loss": 0.9407, "step": 588 }, { "epoch": 0.09, "learning_rate": 3.144687666844634e-05, "loss": 0.9634, "step": 589 }, { "epoch": 0.09, "learning_rate": 3.1500266951414844e-05, "loss": 0.9645, "step": 590 }, { "epoch": 0.09, "learning_rate": 3.1553657234383346e-05, "loss": 1.0293, "step": 591 }, { "epoch": 0.09, "learning_rate": 3.160704751735185e-05, "loss": 1.0629, "step": 592 }, { "epoch": 0.09, "learning_rate": 3.166043780032034e-05, "loss": 0.9456, "step": 593 }, { "epoch": 0.1, "learning_rate": 3.171382808328884e-05, "loss": 0.9446, "step": 594 }, { "epoch": 0.1, "learning_rate": 3.1767218366257343e-05, "loss": 0.9204, "step": 595 }, { "epoch": 0.1, "learning_rate": 3.1820608649225845e-05, "loss": 0.9762, "step": 596 }, { "epoch": 0.1, "learning_rate": 3.1873998932194346e-05, "loss": 0.9, "step": 597 }, { "epoch": 0.1, "learning_rate": 3.192738921516284e-05, "loss": 0.9754, "step": 598 }, { "epoch": 0.1, "learning_rate": 3.198077949813134e-05, "loss": 0.9616, "step": 599 }, { "epoch": 0.1, "learning_rate": 3.203416978109984e-05, "loss": 0.9367, "step": 600 }, { "epoch": 0.1, "learning_rate": 3.208756006406834e-05, "loss": 0.99, "step": 601 }, { "epoch": 0.1, "learning_rate": 3.214095034703684e-05, "loss": 0.8983, "step": 602 }, { "epoch": 0.1, "learning_rate": 3.2194340630005346e-05, "loss": 0.9543, "step": 603 }, { "epoch": 0.1, "learning_rate": 3.224773091297384e-05, "loss": 0.9471, "step": 604 }, { "epoch": 0.1, "learning_rate": 3.230112119594234e-05, "loss": 1.047, "step": 605 }, { "epoch": 0.1, "learning_rate": 3.2354511478910836e-05, "loss": 0.968, "step": 606 }, { "epoch": 0.1, "learning_rate": 3.240790176187934e-05, "loss": 0.9964, "step": 607 }, { "epoch": 0.1, "learning_rate": 3.246129204484784e-05, "loss": 0.9936, "step": 608 }, { "epoch": 0.1, "learning_rate": 3.251468232781634e-05, "loss": 1.0218, "step": 609 }, { "epoch": 0.1, "learning_rate": 3.256807261078484e-05, "loss": 0.8523, "step": 610 }, { "epoch": 0.1, "learning_rate": 3.262146289375334e-05, "loss": 1.0856, "step": 611 }, { "epoch": 0.1, "learning_rate": 3.2674853176721836e-05, "loss": 1.028, "step": 612 }, { "epoch": 0.1, "learning_rate": 3.272824345969034e-05, "loss": 0.9507, "step": 613 }, { "epoch": 0.1, "learning_rate": 3.278163374265883e-05, "loss": 0.9941, "step": 614 }, { "epoch": 0.1, "learning_rate": 3.283502402562734e-05, "loss": 0.9557, "step": 615 }, { "epoch": 0.1, "learning_rate": 3.288841430859584e-05, "loss": 0.8858, "step": 616 }, { "epoch": 0.1, "learning_rate": 3.2941804591564335e-05, "loss": 1.0158, "step": 617 }, { "epoch": 0.1, "learning_rate": 3.2995194874532837e-05, "loss": 0.8908, "step": 618 }, { "epoch": 0.1, "learning_rate": 3.304858515750134e-05, "loss": 1.0373, "step": 619 }, { "epoch": 0.1, "learning_rate": 3.310197544046983e-05, "loss": 0.9695, "step": 620 }, { "epoch": 0.1, "learning_rate": 3.315536572343833e-05, "loss": 1.021, "step": 621 }, { "epoch": 0.1, "learning_rate": 3.320875600640684e-05, "loss": 0.9215, "step": 622 }, { "epoch": 0.1, "learning_rate": 3.3262146289375336e-05, "loss": 1.0052, "step": 623 }, { "epoch": 0.1, "learning_rate": 3.331553657234384e-05, "loss": 1.0172, "step": 624 }, { "epoch": 0.1, "learning_rate": 3.336892685531233e-05, "loss": 0.9293, "step": 625 }, { "epoch": 0.1, "learning_rate": 3.342231713828083e-05, "loss": 0.9828, "step": 626 }, { "epoch": 0.1, "learning_rate": 3.3475707421249334e-05, "loss": 0.9518, "step": 627 }, { "epoch": 0.1, "learning_rate": 3.3529097704217835e-05, "loss": 0.9951, "step": 628 }, { "epoch": 0.1, "learning_rate": 3.3582487987186336e-05, "loss": 0.9469, "step": 629 }, { "epoch": 0.1, "learning_rate": 3.363587827015484e-05, "loss": 0.8836, "step": 630 }, { "epoch": 0.1, "learning_rate": 3.368926855312333e-05, "loss": 0.9368, "step": 631 }, { "epoch": 0.1, "learning_rate": 3.374265883609183e-05, "loss": 0.9265, "step": 632 }, { "epoch": 0.1, "learning_rate": 3.379604911906033e-05, "loss": 1.0142, "step": 633 }, { "epoch": 0.1, "learning_rate": 3.3849439402028835e-05, "loss": 0.9654, "step": 634 }, { "epoch": 0.1, "learning_rate": 3.3902829684997336e-05, "loss": 1.0002, "step": 635 }, { "epoch": 0.1, "learning_rate": 3.395621996796583e-05, "loss": 1.0039, "step": 636 }, { "epoch": 0.1, "learning_rate": 3.400961025093433e-05, "loss": 0.9982, "step": 637 }, { "epoch": 0.1, "learning_rate": 3.406300053390283e-05, "loss": 0.9182, "step": 638 }, { "epoch": 0.1, "learning_rate": 3.411639081687133e-05, "loss": 1.0396, "step": 639 }, { "epoch": 0.1, "learning_rate": 3.416978109983983e-05, "loss": 1.0303, "step": 640 }, { "epoch": 0.1, "learning_rate": 3.422317138280833e-05, "loss": 1.075, "step": 641 }, { "epoch": 0.1, "learning_rate": 3.427656166577683e-05, "loss": 1.0544, "step": 642 }, { "epoch": 0.1, "learning_rate": 3.432995194874533e-05, "loss": 0.9209, "step": 643 }, { "epoch": 0.1, "learning_rate": 3.4383342231713826e-05, "loss": 0.9995, "step": 644 }, { "epoch": 0.1, "learning_rate": 3.443673251468233e-05, "loss": 0.989, "step": 645 }, { "epoch": 0.1, "learning_rate": 3.449012279765083e-05, "loss": 0.9313, "step": 646 }, { "epoch": 0.1, "learning_rate": 3.454351308061933e-05, "loss": 0.9195, "step": 647 }, { "epoch": 0.1, "learning_rate": 3.459690336358783e-05, "loss": 1.1007, "step": 648 }, { "epoch": 0.1, "learning_rate": 3.465029364655633e-05, "loss": 0.9513, "step": 649 }, { "epoch": 0.1, "learning_rate": 3.4703683929524827e-05, "loss": 0.9831, "step": 650 }, { "epoch": 0.1, "learning_rate": 3.475707421249333e-05, "loss": 0.9461, "step": 651 }, { "epoch": 0.1, "learning_rate": 3.481046449546182e-05, "loss": 0.961, "step": 652 }, { "epoch": 0.1, "learning_rate": 3.486385477843033e-05, "loss": 0.9846, "step": 653 }, { "epoch": 0.1, "learning_rate": 3.491724506139883e-05, "loss": 0.8905, "step": 654 }, { "epoch": 0.1, "learning_rate": 3.4970635344367326e-05, "loss": 1.0135, "step": 655 }, { "epoch": 0.11, "learning_rate": 3.502402562733583e-05, "loss": 0.9153, "step": 656 }, { "epoch": 0.11, "learning_rate": 3.507741591030433e-05, "loss": 1.0074, "step": 657 }, { "epoch": 0.11, "learning_rate": 3.513080619327282e-05, "loss": 0.9987, "step": 658 }, { "epoch": 0.11, "learning_rate": 3.5184196476241324e-05, "loss": 0.9209, "step": 659 }, { "epoch": 0.11, "learning_rate": 3.5237586759209825e-05, "loss": 0.9777, "step": 660 }, { "epoch": 0.11, "learning_rate": 3.5290977042178326e-05, "loss": 0.9598, "step": 661 }, { "epoch": 0.11, "learning_rate": 3.534436732514683e-05, "loss": 0.9491, "step": 662 }, { "epoch": 0.11, "learning_rate": 3.539775760811532e-05, "loss": 1.0367, "step": 663 }, { "epoch": 0.11, "learning_rate": 3.545114789108382e-05, "loss": 0.9286, "step": 664 }, { "epoch": 0.11, "learning_rate": 3.5504538174052324e-05, "loss": 1.03, "step": 665 }, { "epoch": 0.11, "learning_rate": 3.5557928457020825e-05, "loss": 1.0125, "step": 666 }, { "epoch": 0.11, "learning_rate": 3.5611318739989326e-05, "loss": 0.8942, "step": 667 }, { "epoch": 0.11, "learning_rate": 3.566470902295782e-05, "loss": 0.8717, "step": 668 }, { "epoch": 0.11, "learning_rate": 3.571809930592632e-05, "loss": 0.996, "step": 669 }, { "epoch": 0.11, "learning_rate": 3.577148958889482e-05, "loss": 0.9156, "step": 670 }, { "epoch": 0.11, "learning_rate": 3.582487987186332e-05, "loss": 0.9659, "step": 671 }, { "epoch": 0.11, "learning_rate": 3.5878270154831825e-05, "loss": 0.9831, "step": 672 }, { "epoch": 0.11, "learning_rate": 3.5931660437800326e-05, "loss": 0.8611, "step": 673 }, { "epoch": 0.11, "learning_rate": 3.598505072076882e-05, "loss": 0.9905, "step": 674 }, { "epoch": 0.11, "learning_rate": 3.603844100373732e-05, "loss": 0.9577, "step": 675 }, { "epoch": 0.11, "learning_rate": 3.609183128670582e-05, "loss": 0.9353, "step": 676 }, { "epoch": 0.11, "learning_rate": 3.614522156967432e-05, "loss": 0.955, "step": 677 }, { "epoch": 0.11, "learning_rate": 3.619861185264282e-05, "loss": 0.9486, "step": 678 }, { "epoch": 0.11, "learning_rate": 3.625200213561132e-05, "loss": 0.8409, "step": 679 }, { "epoch": 0.11, "learning_rate": 3.630539241857982e-05, "loss": 0.9268, "step": 680 }, { "epoch": 0.11, "learning_rate": 3.635878270154832e-05, "loss": 0.9514, "step": 681 }, { "epoch": 0.11, "learning_rate": 3.641217298451682e-05, "loss": 0.8697, "step": 682 }, { "epoch": 0.11, "learning_rate": 3.646556326748532e-05, "loss": 0.9783, "step": 683 }, { "epoch": 0.11, "learning_rate": 3.651895355045382e-05, "loss": 0.9263, "step": 684 }, { "epoch": 0.11, "learning_rate": 3.657234383342232e-05, "loss": 1.0298, "step": 685 }, { "epoch": 0.11, "learning_rate": 3.662573411639082e-05, "loss": 0.9565, "step": 686 }, { "epoch": 0.11, "learning_rate": 3.6679124399359316e-05, "loss": 0.8875, "step": 687 }, { "epoch": 0.11, "learning_rate": 3.673251468232782e-05, "loss": 1.0049, "step": 688 }, { "epoch": 0.11, "learning_rate": 3.678590496529632e-05, "loss": 0.9899, "step": 689 }, { "epoch": 0.11, "learning_rate": 3.683929524826481e-05, "loss": 0.9658, "step": 690 }, { "epoch": 0.11, "learning_rate": 3.689268553123332e-05, "loss": 0.9777, "step": 691 }, { "epoch": 0.11, "learning_rate": 3.694607581420182e-05, "loss": 0.9829, "step": 692 }, { "epoch": 0.11, "learning_rate": 3.6999466097170316e-05, "loss": 1.0698, "step": 693 }, { "epoch": 0.11, "learning_rate": 3.705285638013882e-05, "loss": 0.9383, "step": 694 }, { "epoch": 0.11, "learning_rate": 3.710624666310731e-05, "loss": 0.9618, "step": 695 }, { "epoch": 0.11, "learning_rate": 3.715963694607581e-05, "loss": 0.8965, "step": 696 }, { "epoch": 0.11, "learning_rate": 3.7213027229044314e-05, "loss": 0.9043, "step": 697 }, { "epoch": 0.11, "learning_rate": 3.7266417512012815e-05, "loss": 1.0027, "step": 698 }, { "epoch": 0.11, "learning_rate": 3.7319807794981316e-05, "loss": 1.0209, "step": 699 }, { "epoch": 0.11, "learning_rate": 3.737319807794982e-05, "loss": 0.9522, "step": 700 }, { "epoch": 0.11, "learning_rate": 3.742658836091831e-05, "loss": 1.0743, "step": 701 }, { "epoch": 0.11, "learning_rate": 3.747997864388681e-05, "loss": 0.9294, "step": 702 }, { "epoch": 0.11, "learning_rate": 3.7533368926855314e-05, "loss": 0.9394, "step": 703 }, { "epoch": 0.11, "learning_rate": 3.7586759209823815e-05, "loss": 0.9216, "step": 704 }, { "epoch": 0.11, "learning_rate": 3.7640149492792317e-05, "loss": 0.9668, "step": 705 }, { "epoch": 0.11, "learning_rate": 3.769353977576081e-05, "loss": 0.9845, "step": 706 }, { "epoch": 0.11, "learning_rate": 3.774693005872931e-05, "loss": 0.8878, "step": 707 }, { "epoch": 0.11, "learning_rate": 3.780032034169781e-05, "loss": 0.999, "step": 708 }, { "epoch": 0.11, "learning_rate": 3.785371062466631e-05, "loss": 0.8852, "step": 709 }, { "epoch": 0.11, "learning_rate": 3.7907100907634816e-05, "loss": 0.9752, "step": 710 }, { "epoch": 0.11, "learning_rate": 3.796049119060332e-05, "loss": 1.0355, "step": 711 }, { "epoch": 0.11, "learning_rate": 3.801388147357181e-05, "loss": 0.8024, "step": 712 }, { "epoch": 0.11, "learning_rate": 3.806727175654031e-05, "loss": 0.9751, "step": 713 }, { "epoch": 0.11, "learning_rate": 3.812066203950881e-05, "loss": 0.8969, "step": 714 }, { "epoch": 0.11, "learning_rate": 3.817405232247731e-05, "loss": 0.986, "step": 715 }, { "epoch": 0.11, "learning_rate": 3.822744260544581e-05, "loss": 0.9205, "step": 716 }, { "epoch": 0.11, "learning_rate": 3.828083288841431e-05, "loss": 0.9689, "step": 717 }, { "epoch": 0.11, "learning_rate": 3.833422317138281e-05, "loss": 0.9645, "step": 718 }, { "epoch": 0.12, "learning_rate": 3.838761345435131e-05, "loss": 0.9232, "step": 719 }, { "epoch": 0.12, "learning_rate": 3.844100373731981e-05, "loss": 1.0538, "step": 720 }, { "epoch": 0.12, "learning_rate": 3.849439402028831e-05, "loss": 0.9743, "step": 721 }, { "epoch": 0.12, "learning_rate": 3.854778430325681e-05, "loss": 0.8964, "step": 722 }, { "epoch": 0.12, "learning_rate": 3.860117458622531e-05, "loss": 0.9353, "step": 723 }, { "epoch": 0.12, "learning_rate": 3.865456486919381e-05, "loss": 0.9954, "step": 724 }, { "epoch": 0.12, "learning_rate": 3.8707955152162306e-05, "loss": 0.9608, "step": 725 }, { "epoch": 0.12, "learning_rate": 3.876134543513081e-05, "loss": 0.9159, "step": 726 }, { "epoch": 0.12, "learning_rate": 3.881473571809931e-05, "loss": 0.9544, "step": 727 }, { "epoch": 0.12, "learning_rate": 3.88681260010678e-05, "loss": 0.887, "step": 728 }, { "epoch": 0.12, "learning_rate": 3.892151628403631e-05, "loss": 1.0047, "step": 729 }, { "epoch": 0.12, "learning_rate": 3.897490656700481e-05, "loss": 0.9486, "step": 730 }, { "epoch": 0.12, "learning_rate": 3.9028296849973306e-05, "loss": 0.8553, "step": 731 }, { "epoch": 0.12, "learning_rate": 3.908168713294181e-05, "loss": 0.9719, "step": 732 }, { "epoch": 0.12, "learning_rate": 3.91350774159103e-05, "loss": 0.9297, "step": 733 }, { "epoch": 0.12, "learning_rate": 3.91884676988788e-05, "loss": 1.0412, "step": 734 }, { "epoch": 0.12, "learning_rate": 3.9241857981847304e-05, "loss": 1.0635, "step": 735 }, { "epoch": 0.12, "learning_rate": 3.9295248264815805e-05, "loss": 0.9565, "step": 736 }, { "epoch": 0.12, "learning_rate": 3.9348638547784307e-05, "loss": 0.9018, "step": 737 }, { "epoch": 0.12, "learning_rate": 3.940202883075281e-05, "loss": 0.9155, "step": 738 }, { "epoch": 0.12, "learning_rate": 3.94554191137213e-05, "loss": 0.8562, "step": 739 }, { "epoch": 0.12, "learning_rate": 3.95088093966898e-05, "loss": 0.9492, "step": 740 }, { "epoch": 0.12, "learning_rate": 3.9562199679658305e-05, "loss": 0.946, "step": 741 }, { "epoch": 0.12, "learning_rate": 3.9615589962626806e-05, "loss": 0.9443, "step": 742 }, { "epoch": 0.12, "learning_rate": 3.966898024559531e-05, "loss": 0.9233, "step": 743 }, { "epoch": 0.12, "learning_rate": 3.97223705285638e-05, "loss": 0.931, "step": 744 }, { "epoch": 0.12, "learning_rate": 3.97757608115323e-05, "loss": 0.9076, "step": 745 }, { "epoch": 0.12, "learning_rate": 3.9829151094500804e-05, "loss": 0.8753, "step": 746 }, { "epoch": 0.12, "learning_rate": 3.98825413774693e-05, "loss": 0.956, "step": 747 }, { "epoch": 0.12, "learning_rate": 3.9935931660437806e-05, "loss": 0.9535, "step": 748 }, { "epoch": 0.12, "learning_rate": 3.99893219434063e-05, "loss": 0.9985, "step": 749 }, { "epoch": 0.12, "learning_rate": 4.00427122263748e-05, "loss": 1.0047, "step": 750 }, { "epoch": 0.12, "learning_rate": 4.00961025093433e-05, "loss": 0.9556, "step": 751 }, { "epoch": 0.12, "learning_rate": 4.01494927923118e-05, "loss": 1.0561, "step": 752 }, { "epoch": 0.12, "learning_rate": 4.02028830752803e-05, "loss": 0.9698, "step": 753 }, { "epoch": 0.12, "learning_rate": 4.02562733582488e-05, "loss": 1.01, "step": 754 }, { "epoch": 0.12, "learning_rate": 4.03096636412173e-05, "loss": 0.8668, "step": 755 }, { "epoch": 0.12, "learning_rate": 4.03630539241858e-05, "loss": 0.9296, "step": 756 }, { "epoch": 0.12, "learning_rate": 4.04164442071543e-05, "loss": 0.8836, "step": 757 }, { "epoch": 0.12, "learning_rate": 4.04698344901228e-05, "loss": 0.9895, "step": 758 }, { "epoch": 0.12, "learning_rate": 4.05232247730913e-05, "loss": 0.9457, "step": 759 }, { "epoch": 0.12, "learning_rate": 4.05766150560598e-05, "loss": 0.8793, "step": 760 }, { "epoch": 0.12, "learning_rate": 4.06300053390283e-05, "loss": 0.944, "step": 761 }, { "epoch": 0.12, "learning_rate": 4.06833956219968e-05, "loss": 0.9401, "step": 762 }, { "epoch": 0.12, "learning_rate": 4.0736785904965296e-05, "loss": 1.0217, "step": 763 }, { "epoch": 0.12, "learning_rate": 4.07901761879338e-05, "loss": 1.0389, "step": 764 }, { "epoch": 0.12, "learning_rate": 4.08435664709023e-05, "loss": 0.9621, "step": 765 }, { "epoch": 0.12, "learning_rate": 4.089695675387079e-05, "loss": 0.9218, "step": 766 }, { "epoch": 0.12, "learning_rate": 4.09503470368393e-05, "loss": 1.0195, "step": 767 }, { "epoch": 0.12, "learning_rate": 4.1003737319807795e-05, "loss": 0.883, "step": 768 }, { "epoch": 0.12, "learning_rate": 4.10571276027763e-05, "loss": 0.9068, "step": 769 }, { "epoch": 0.12, "learning_rate": 4.11105178857448e-05, "loss": 0.9657, "step": 770 }, { "epoch": 0.12, "learning_rate": 4.116390816871329e-05, "loss": 1.0571, "step": 771 }, { "epoch": 0.12, "learning_rate": 4.1217298451681793e-05, "loss": 0.9567, "step": 772 }, { "epoch": 0.12, "learning_rate": 4.1270688734650295e-05, "loss": 0.9148, "step": 773 }, { "epoch": 0.12, "learning_rate": 4.1324079017618796e-05, "loss": 0.9295, "step": 774 }, { "epoch": 0.12, "learning_rate": 4.13774693005873e-05, "loss": 0.9581, "step": 775 }, { "epoch": 0.12, "learning_rate": 4.143085958355579e-05, "loss": 0.924, "step": 776 }, { "epoch": 0.12, "learning_rate": 4.148424986652429e-05, "loss": 0.9566, "step": 777 }, { "epoch": 0.12, "learning_rate": 4.1537640149492794e-05, "loss": 0.8998, "step": 778 }, { "epoch": 0.12, "learning_rate": 4.1591030432461295e-05, "loss": 1.0397, "step": 779 }, { "epoch": 0.12, "learning_rate": 4.1644420715429796e-05, "loss": 0.9098, "step": 780 }, { "epoch": 0.13, "learning_rate": 4.16978109983983e-05, "loss": 0.8756, "step": 781 }, { "epoch": 0.13, "learning_rate": 4.175120128136679e-05, "loss": 0.9625, "step": 782 }, { "epoch": 0.13, "learning_rate": 4.180459156433529e-05, "loss": 0.9188, "step": 783 }, { "epoch": 0.13, "learning_rate": 4.1857981847303794e-05, "loss": 0.9587, "step": 784 }, { "epoch": 0.13, "learning_rate": 4.1911372130272295e-05, "loss": 0.9669, "step": 785 }, { "epoch": 0.13, "learning_rate": 4.1964762413240796e-05, "loss": 0.9261, "step": 786 }, { "epoch": 0.13, "learning_rate": 4.201815269620929e-05, "loss": 0.9093, "step": 787 }, { "epoch": 0.13, "learning_rate": 4.207154297917779e-05, "loss": 0.9631, "step": 788 }, { "epoch": 0.13, "learning_rate": 4.212493326214629e-05, "loss": 0.9891, "step": 789 }, { "epoch": 0.13, "learning_rate": 4.217832354511479e-05, "loss": 1.0083, "step": 790 }, { "epoch": 0.13, "learning_rate": 4.223171382808329e-05, "loss": 0.8909, "step": 791 }, { "epoch": 0.13, "learning_rate": 4.2285104111051797e-05, "loss": 0.9984, "step": 792 }, { "epoch": 0.13, "learning_rate": 4.233849439402029e-05, "loss": 0.9405, "step": 793 }, { "epoch": 0.13, "learning_rate": 4.239188467698879e-05, "loss": 0.9713, "step": 794 }, { "epoch": 0.13, "learning_rate": 4.2445274959957286e-05, "loss": 1.0142, "step": 795 }, { "epoch": 0.13, "learning_rate": 4.249866524292579e-05, "loss": 0.8503, "step": 796 }, { "epoch": 0.13, "learning_rate": 4.255205552589429e-05, "loss": 1.0305, "step": 797 }, { "epoch": 0.13, "learning_rate": 4.260544580886279e-05, "loss": 0.9229, "step": 798 }, { "epoch": 0.13, "learning_rate": 4.265883609183129e-05, "loss": 0.9627, "step": 799 }, { "epoch": 0.13, "learning_rate": 4.271222637479979e-05, "loss": 0.992, "step": 800 }, { "epoch": 0.13, "learning_rate": 4.276561665776829e-05, "loss": 0.9263, "step": 801 }, { "epoch": 0.13, "learning_rate": 4.281900694073679e-05, "loss": 0.8738, "step": 802 }, { "epoch": 0.13, "learning_rate": 4.287239722370528e-05, "loss": 0.9525, "step": 803 }, { "epoch": 0.13, "learning_rate": 4.292578750667379e-05, "loss": 0.9702, "step": 804 }, { "epoch": 0.13, "learning_rate": 4.297917778964229e-05, "loss": 0.9396, "step": 805 }, { "epoch": 0.13, "learning_rate": 4.3032568072610786e-05, "loss": 0.9507, "step": 806 }, { "epoch": 0.13, "learning_rate": 4.308595835557929e-05, "loss": 1.0126, "step": 807 }, { "epoch": 0.13, "learning_rate": 4.313934863854779e-05, "loss": 0.8001, "step": 808 }, { "epoch": 0.13, "learning_rate": 4.319273892151628e-05, "loss": 0.9551, "step": 809 }, { "epoch": 0.13, "learning_rate": 4.3246129204484784e-05, "loss": 0.9445, "step": 810 }, { "epoch": 0.13, "learning_rate": 4.329951948745329e-05, "loss": 0.8943, "step": 811 }, { "epoch": 0.13, "learning_rate": 4.3352909770421786e-05, "loss": 1.0033, "step": 812 }, { "epoch": 0.13, "learning_rate": 4.340630005339029e-05, "loss": 0.954, "step": 813 }, { "epoch": 0.13, "learning_rate": 4.345969033635878e-05, "loss": 0.9113, "step": 814 }, { "epoch": 0.13, "learning_rate": 4.351308061932728e-05, "loss": 0.9328, "step": 815 }, { "epoch": 0.13, "learning_rate": 4.3566470902295784e-05, "loss": 0.9207, "step": 816 }, { "epoch": 0.13, "learning_rate": 4.3619861185264285e-05, "loss": 0.9193, "step": 817 }, { "epoch": 0.13, "learning_rate": 4.3673251468232786e-05, "loss": 0.9364, "step": 818 }, { "epoch": 0.13, "learning_rate": 4.372664175120129e-05, "loss": 0.8975, "step": 819 }, { "epoch": 0.13, "learning_rate": 4.378003203416978e-05, "loss": 0.9224, "step": 820 }, { "epoch": 0.13, "learning_rate": 4.383342231713828e-05, "loss": 0.9429, "step": 821 }, { "epoch": 0.13, "learning_rate": 4.388681260010678e-05, "loss": 0.9805, "step": 822 }, { "epoch": 0.13, "learning_rate": 4.3940202883075285e-05, "loss": 0.9637, "step": 823 }, { "epoch": 0.13, "learning_rate": 4.3993593166043787e-05, "loss": 0.8926, "step": 824 }, { "epoch": 0.13, "learning_rate": 4.404698344901228e-05, "loss": 0.9031, "step": 825 }, { "epoch": 0.13, "learning_rate": 4.410037373198078e-05, "loss": 0.9228, "step": 826 }, { "epoch": 0.13, "learning_rate": 4.415376401494928e-05, "loss": 0.9657, "step": 827 }, { "epoch": 0.13, "learning_rate": 4.420715429791778e-05, "loss": 1.0295, "step": 828 }, { "epoch": 0.13, "learning_rate": 4.426054458088628e-05, "loss": 0.9487, "step": 829 }, { "epoch": 0.13, "learning_rate": 4.431393486385478e-05, "loss": 0.8766, "step": 830 }, { "epoch": 0.13, "learning_rate": 4.436732514682328e-05, "loss": 0.9612, "step": 831 }, { "epoch": 0.13, "learning_rate": 4.442071542979178e-05, "loss": 0.9142, "step": 832 }, { "epoch": 0.13, "learning_rate": 4.447410571276028e-05, "loss": 1.1025, "step": 833 }, { "epoch": 0.13, "learning_rate": 4.452749599572878e-05, "loss": 1.0437, "step": 834 }, { "epoch": 0.13, "learning_rate": 4.458088627869728e-05, "loss": 0.9137, "step": 835 }, { "epoch": 0.13, "learning_rate": 4.463427656166578e-05, "loss": 0.9484, "step": 836 }, { "epoch": 0.13, "learning_rate": 4.468766684463428e-05, "loss": 0.9105, "step": 837 }, { "epoch": 0.13, "learning_rate": 4.4741057127602776e-05, "loss": 0.9908, "step": 838 }, { "epoch": 0.13, "learning_rate": 4.479444741057128e-05, "loss": 0.9819, "step": 839 }, { "epoch": 0.13, "learning_rate": 4.484783769353978e-05, "loss": 0.9508, "step": 840 }, { "epoch": 0.13, "learning_rate": 4.490122797650827e-05, "loss": 0.9607, "step": 841 }, { "epoch": 0.13, "learning_rate": 4.495461825947678e-05, "loss": 0.9287, "step": 842 }, { "epoch": 0.14, "learning_rate": 4.500800854244528e-05, "loss": 0.916, "step": 843 }, { "epoch": 0.14, "learning_rate": 4.5061398825413776e-05, "loss": 0.9159, "step": 844 }, { "epoch": 0.14, "learning_rate": 4.511478910838228e-05, "loss": 0.9864, "step": 845 }, { "epoch": 0.14, "learning_rate": 4.516817939135078e-05, "loss": 0.9466, "step": 846 }, { "epoch": 0.14, "learning_rate": 4.522156967431927e-05, "loss": 0.8827, "step": 847 }, { "epoch": 0.14, "learning_rate": 4.5274959957287774e-05, "loss": 0.969, "step": 848 }, { "epoch": 0.14, "learning_rate": 4.5328350240256275e-05, "loss": 0.9072, "step": 849 }, { "epoch": 0.14, "learning_rate": 4.5381740523224776e-05, "loss": 0.8245, "step": 850 }, { "epoch": 0.14, "learning_rate": 4.543513080619328e-05, "loss": 0.9698, "step": 851 }, { "epoch": 0.14, "learning_rate": 4.548852108916177e-05, "loss": 0.823, "step": 852 }, { "epoch": 0.14, "learning_rate": 4.554191137213027e-05, "loss": 1.0087, "step": 853 }, { "epoch": 0.14, "learning_rate": 4.5595301655098774e-05, "loss": 1.0223, "step": 854 }, { "epoch": 0.14, "learning_rate": 4.5648691938067275e-05, "loss": 0.9561, "step": 855 }, { "epoch": 0.14, "learning_rate": 4.570208222103578e-05, "loss": 0.9364, "step": 856 }, { "epoch": 0.14, "learning_rate": 4.575547250400427e-05, "loss": 0.977, "step": 857 }, { "epoch": 0.14, "learning_rate": 4.580886278697277e-05, "loss": 0.8975, "step": 858 }, { "epoch": 0.14, "learning_rate": 4.586225306994127e-05, "loss": 0.8898, "step": 859 }, { "epoch": 0.14, "learning_rate": 4.591564335290977e-05, "loss": 1.006, "step": 860 }, { "epoch": 0.14, "learning_rate": 4.5969033635878276e-05, "loss": 0.8953, "step": 861 }, { "epoch": 0.14, "learning_rate": 4.602242391884678e-05, "loss": 0.8903, "step": 862 }, { "epoch": 0.14, "learning_rate": 4.607581420181527e-05, "loss": 0.9575, "step": 863 }, { "epoch": 0.14, "learning_rate": 4.612920448478377e-05, "loss": 0.9931, "step": 864 }, { "epoch": 0.14, "learning_rate": 4.618259476775227e-05, "loss": 0.838, "step": 865 }, { "epoch": 0.14, "learning_rate": 4.623598505072077e-05, "loss": 0.99, "step": 866 }, { "epoch": 0.14, "learning_rate": 4.628937533368927e-05, "loss": 0.9372, "step": 867 }, { "epoch": 0.14, "learning_rate": 4.634276561665777e-05, "loss": 0.9597, "step": 868 }, { "epoch": 0.14, "learning_rate": 4.639615589962627e-05, "loss": 1.0027, "step": 869 }, { "epoch": 0.14, "learning_rate": 4.644954618259477e-05, "loss": 0.8938, "step": 870 }, { "epoch": 0.14, "learning_rate": 4.650293646556327e-05, "loss": 0.9054, "step": 871 }, { "epoch": 0.14, "learning_rate": 4.655632674853177e-05, "loss": 0.8855, "step": 872 }, { "epoch": 0.14, "learning_rate": 4.660971703150027e-05, "loss": 0.9635, "step": 873 }, { "epoch": 0.14, "learning_rate": 4.666310731446877e-05, "loss": 0.8933, "step": 874 }, { "epoch": 0.14, "learning_rate": 4.671649759743727e-05, "loss": 0.8802, "step": 875 }, { "epoch": 0.14, "learning_rate": 4.6769887880405766e-05, "loss": 0.9195, "step": 876 }, { "epoch": 0.14, "learning_rate": 4.682327816337427e-05, "loss": 0.8607, "step": 877 }, { "epoch": 0.14, "learning_rate": 4.687666844634277e-05, "loss": 0.9505, "step": 878 }, { "epoch": 0.14, "learning_rate": 4.693005872931126e-05, "loss": 1.0288, "step": 879 }, { "epoch": 0.14, "learning_rate": 4.698344901227977e-05, "loss": 0.9939, "step": 880 }, { "epoch": 0.14, "learning_rate": 4.703683929524827e-05, "loss": 0.9252, "step": 881 }, { "epoch": 0.14, "learning_rate": 4.7090229578216766e-05, "loss": 0.9283, "step": 882 }, { "epoch": 0.14, "learning_rate": 4.714361986118527e-05, "loss": 0.9624, "step": 883 }, { "epoch": 0.14, "learning_rate": 4.719701014415376e-05, "loss": 1.0646, "step": 884 }, { "epoch": 0.14, "learning_rate": 4.725040042712226e-05, "loss": 0.9139, "step": 885 }, { "epoch": 0.14, "learning_rate": 4.7303790710090764e-05, "loss": 0.92, "step": 886 }, { "epoch": 0.14, "learning_rate": 4.7357180993059266e-05, "loss": 0.8895, "step": 887 }, { "epoch": 0.14, "learning_rate": 4.741057127602777e-05, "loss": 1.0019, "step": 888 }, { "epoch": 0.14, "learning_rate": 4.746396155899627e-05, "loss": 0.9734, "step": 889 }, { "epoch": 0.14, "learning_rate": 4.751735184196476e-05, "loss": 0.9363, "step": 890 }, { "epoch": 0.14, "learning_rate": 4.7570742124933263e-05, "loss": 0.917, "step": 891 }, { "epoch": 0.14, "learning_rate": 4.7624132407901765e-05, "loss": 0.9629, "step": 892 }, { "epoch": 0.14, "learning_rate": 4.7677522690870266e-05, "loss": 0.8799, "step": 893 }, { "epoch": 0.14, "learning_rate": 4.773091297383877e-05, "loss": 0.9132, "step": 894 }, { "epoch": 0.14, "learning_rate": 4.778430325680726e-05, "loss": 0.933, "step": 895 }, { "epoch": 0.14, "learning_rate": 4.783769353977576e-05, "loss": 0.8414, "step": 896 }, { "epoch": 0.14, "learning_rate": 4.7891083822744264e-05, "loss": 0.971, "step": 897 }, { "epoch": 0.14, "learning_rate": 4.794447410571276e-05, "loss": 0.9184, "step": 898 }, { "epoch": 0.14, "learning_rate": 4.7997864388681266e-05, "loss": 0.9864, "step": 899 }, { "epoch": 0.14, "learning_rate": 4.805125467164977e-05, "loss": 0.9143, "step": 900 }, { "epoch": 0.14, "learning_rate": 4.810464495461826e-05, "loss": 0.9637, "step": 901 }, { "epoch": 0.14, "learning_rate": 4.815803523758676e-05, "loss": 1.0159, "step": 902 }, { "epoch": 0.14, "learning_rate": 4.821142552055526e-05, "loss": 0.9408, "step": 903 }, { "epoch": 0.14, "learning_rate": 4.826481580352376e-05, "loss": 0.8541, "step": 904 }, { "epoch": 0.14, "learning_rate": 4.831820608649226e-05, "loss": 0.9367, "step": 905 }, { "epoch": 0.15, "learning_rate": 4.837159636946076e-05, "loss": 0.9604, "step": 906 }, { "epoch": 0.15, "learning_rate": 4.842498665242926e-05, "loss": 0.8933, "step": 907 }, { "epoch": 0.15, "learning_rate": 4.847837693539776e-05, "loss": 0.9778, "step": 908 }, { "epoch": 0.15, "learning_rate": 4.853176721836626e-05, "loss": 1.0157, "step": 909 }, { "epoch": 0.15, "learning_rate": 4.858515750133476e-05, "loss": 0.8589, "step": 910 }, { "epoch": 0.15, "learning_rate": 4.863854778430326e-05, "loss": 0.901, "step": 911 }, { "epoch": 0.15, "learning_rate": 4.869193806727176e-05, "loss": 0.9056, "step": 912 }, { "epoch": 0.15, "learning_rate": 4.874532835024026e-05, "loss": 0.8447, "step": 913 }, { "epoch": 0.15, "learning_rate": 4.8798718633208757e-05, "loss": 0.8914, "step": 914 }, { "epoch": 0.15, "learning_rate": 4.885210891617726e-05, "loss": 0.9932, "step": 915 }, { "epoch": 0.15, "learning_rate": 4.890549919914576e-05, "loss": 0.9661, "step": 916 }, { "epoch": 0.15, "learning_rate": 4.895888948211425e-05, "loss": 0.9225, "step": 917 }, { "epoch": 0.15, "learning_rate": 4.901227976508276e-05, "loss": 0.8972, "step": 918 }, { "epoch": 0.15, "learning_rate": 4.9065670048051256e-05, "loss": 0.9744, "step": 919 }, { "epoch": 0.15, "learning_rate": 4.911906033101976e-05, "loss": 1.0188, "step": 920 }, { "epoch": 0.15, "learning_rate": 4.917245061398826e-05, "loss": 0.9548, "step": 921 }, { "epoch": 0.15, "learning_rate": 4.922584089695675e-05, "loss": 0.9802, "step": 922 }, { "epoch": 0.15, "learning_rate": 4.9279231179925254e-05, "loss": 0.9461, "step": 923 }, { "epoch": 0.15, "learning_rate": 4.9332621462893755e-05, "loss": 0.9216, "step": 924 }, { "epoch": 0.15, "learning_rate": 4.9386011745862256e-05, "loss": 0.8925, "step": 925 }, { "epoch": 0.15, "learning_rate": 4.943940202883076e-05, "loss": 0.8602, "step": 926 }, { "epoch": 0.15, "learning_rate": 4.949279231179926e-05, "loss": 0.8733, "step": 927 }, { "epoch": 0.15, "learning_rate": 4.954618259476775e-05, "loss": 0.9072, "step": 928 }, { "epoch": 0.15, "learning_rate": 4.9599572877736254e-05, "loss": 0.9503, "step": 929 }, { "epoch": 0.15, "learning_rate": 4.9652963160704755e-05, "loss": 0.9248, "step": 930 }, { "epoch": 0.15, "learning_rate": 4.9706353443673256e-05, "loss": 0.8548, "step": 931 }, { "epoch": 0.15, "learning_rate": 4.975974372664176e-05, "loss": 1.002, "step": 932 }, { "epoch": 0.15, "learning_rate": 4.981313400961025e-05, "loss": 0.9393, "step": 933 }, { "epoch": 0.15, "learning_rate": 4.986652429257875e-05, "loss": 0.9231, "step": 934 }, { "epoch": 0.15, "learning_rate": 4.9919914575547254e-05, "loss": 0.9554, "step": 935 }, { "epoch": 0.15, "learning_rate": 4.997330485851575e-05, "loss": 0.8637, "step": 936 }, { "epoch": 0.15, "learning_rate": 5.002669514148425e-05, "loss": 0.9109, "step": 937 }, { "epoch": 0.15, "learning_rate": 5.008008542445275e-05, "loss": 0.8683, "step": 938 }, { "epoch": 0.15, "learning_rate": 5.013347570742125e-05, "loss": 0.9424, "step": 939 }, { "epoch": 0.15, "learning_rate": 5.018686599038975e-05, "loss": 0.9946, "step": 940 }, { "epoch": 0.15, "learning_rate": 5.0240256273358254e-05, "loss": 0.8472, "step": 941 }, { "epoch": 0.15, "learning_rate": 5.029364655632675e-05, "loss": 1.0024, "step": 942 }, { "epoch": 0.15, "learning_rate": 5.034703683929525e-05, "loss": 1.0243, "step": 943 }, { "epoch": 0.15, "learning_rate": 5.040042712226375e-05, "loss": 0.9003, "step": 944 }, { "epoch": 0.15, "learning_rate": 5.045381740523225e-05, "loss": 1.0056, "step": 945 }, { "epoch": 0.15, "learning_rate": 5.050720768820075e-05, "loss": 0.9275, "step": 946 }, { "epoch": 0.15, "learning_rate": 5.0560597971169255e-05, "loss": 0.8602, "step": 947 }, { "epoch": 0.15, "learning_rate": 5.061398825413774e-05, "loss": 0.8654, "step": 948 }, { "epoch": 0.15, "learning_rate": 5.066737853710625e-05, "loss": 0.8655, "step": 949 }, { "epoch": 0.15, "learning_rate": 5.072076882007475e-05, "loss": 0.8634, "step": 950 }, { "epoch": 0.15, "learning_rate": 5.077415910304325e-05, "loss": 0.9527, "step": 951 }, { "epoch": 0.15, "learning_rate": 5.0827549386011754e-05, "loss": 0.8788, "step": 952 }, { "epoch": 0.15, "learning_rate": 5.088093966898024e-05, "loss": 0.897, "step": 953 }, { "epoch": 0.15, "learning_rate": 5.093432995194874e-05, "loss": 0.9563, "step": 954 }, { "epoch": 0.15, "learning_rate": 5.0987720234917244e-05, "loss": 0.9153, "step": 955 }, { "epoch": 0.15, "learning_rate": 5.104111051788575e-05, "loss": 0.9431, "step": 956 }, { "epoch": 0.15, "learning_rate": 5.109450080085425e-05, "loss": 0.9169, "step": 957 }, { "epoch": 0.15, "learning_rate": 5.1147891083822754e-05, "loss": 0.8708, "step": 958 }, { "epoch": 0.15, "learning_rate": 5.120128136679124e-05, "loss": 0.9959, "step": 959 }, { "epoch": 0.15, "learning_rate": 5.125467164975974e-05, "loss": 0.8575, "step": 960 }, { "epoch": 0.15, "learning_rate": 5.1308061932728244e-05, "loss": 0.9322, "step": 961 }, { "epoch": 0.15, "learning_rate": 5.1361452215696745e-05, "loss": 0.9872, "step": 962 }, { "epoch": 0.15, "learning_rate": 5.141484249866525e-05, "loss": 1.0395, "step": 963 }, { "epoch": 0.15, "learning_rate": 5.146823278163374e-05, "loss": 0.94, "step": 964 }, { "epoch": 0.15, "learning_rate": 5.152162306460224e-05, "loss": 0.9349, "step": 965 }, { "epoch": 0.15, "learning_rate": 5.157501334757074e-05, "loss": 0.934, "step": 966 }, { "epoch": 0.15, "learning_rate": 5.1628403630539244e-05, "loss": 0.901, "step": 967 }, { "epoch": 0.16, "learning_rate": 5.1681793913507745e-05, "loss": 0.9933, "step": 968 }, { "epoch": 0.16, "learning_rate": 5.173518419647624e-05, "loss": 0.9197, "step": 969 }, { "epoch": 0.16, "learning_rate": 5.178857447944474e-05, "loss": 0.8931, "step": 970 }, { "epoch": 0.16, "learning_rate": 5.184196476241324e-05, "loss": 0.9787, "step": 971 }, { "epoch": 0.16, "learning_rate": 5.189535504538174e-05, "loss": 0.9657, "step": 972 }, { "epoch": 0.16, "learning_rate": 5.1948745328350244e-05, "loss": 0.9558, "step": 973 }, { "epoch": 0.16, "learning_rate": 5.2002135611318746e-05, "loss": 0.9693, "step": 974 }, { "epoch": 0.16, "learning_rate": 5.205552589428724e-05, "loss": 0.9223, "step": 975 }, { "epoch": 0.16, "learning_rate": 5.210891617725574e-05, "loss": 0.9305, "step": 976 }, { "epoch": 0.16, "learning_rate": 5.216230646022424e-05, "loss": 0.9222, "step": 977 }, { "epoch": 0.16, "learning_rate": 5.2215696743192743e-05, "loss": 0.8687, "step": 978 }, { "epoch": 0.16, "learning_rate": 5.2269087026161245e-05, "loss": 0.9074, "step": 979 }, { "epoch": 0.16, "learning_rate": 5.232247730912974e-05, "loss": 0.9011, "step": 980 }, { "epoch": 0.16, "learning_rate": 5.237586759209824e-05, "loss": 0.9228, "step": 981 }, { "epoch": 0.16, "learning_rate": 5.242925787506674e-05, "loss": 0.9639, "step": 982 }, { "epoch": 0.16, "learning_rate": 5.248264815803524e-05, "loss": 0.8892, "step": 983 }, { "epoch": 0.16, "learning_rate": 5.2536038441003744e-05, "loss": 0.904, "step": 984 }, { "epoch": 0.16, "learning_rate": 5.2589428723972245e-05, "loss": 0.9065, "step": 985 }, { "epoch": 0.16, "learning_rate": 5.264281900694073e-05, "loss": 0.9126, "step": 986 }, { "epoch": 0.16, "learning_rate": 5.269620928990924e-05, "loss": 0.9506, "step": 987 }, { "epoch": 0.16, "learning_rate": 5.274959957287774e-05, "loss": 0.874, "step": 988 }, { "epoch": 0.16, "learning_rate": 5.280298985584624e-05, "loss": 0.9255, "step": 989 }, { "epoch": 0.16, "learning_rate": 5.2856380138814744e-05, "loss": 0.8924, "step": 990 }, { "epoch": 0.16, "learning_rate": 5.290977042178323e-05, "loss": 0.8669, "step": 991 }, { "epoch": 0.16, "learning_rate": 5.296316070475173e-05, "loss": 0.99, "step": 992 }, { "epoch": 0.16, "learning_rate": 5.3016550987720234e-05, "loss": 0.9679, "step": 993 }, { "epoch": 0.16, "learning_rate": 5.306994127068874e-05, "loss": 0.9536, "step": 994 }, { "epoch": 0.16, "learning_rate": 5.312333155365724e-05, "loss": 0.9243, "step": 995 }, { "epoch": 0.16, "learning_rate": 5.317672183662573e-05, "loss": 0.863, "step": 996 }, { "epoch": 0.16, "learning_rate": 5.323011211959423e-05, "loss": 0.936, "step": 997 }, { "epoch": 0.16, "learning_rate": 5.328350240256273e-05, "loss": 0.8945, "step": 998 }, { "epoch": 0.16, "learning_rate": 5.3336892685531234e-05, "loss": 0.9709, "step": 999 }, { "epoch": 0.16, "learning_rate": 5.3390282968499735e-05, "loss": 1.0121, "step": 1000 }, { "epoch": 0.16, "learning_rate": 5.344367325146824e-05, "loss": 0.9834, "step": 1001 }, { "epoch": 0.16, "learning_rate": 5.349706353443673e-05, "loss": 0.9245, "step": 1002 }, { "epoch": 0.16, "learning_rate": 5.355045381740523e-05, "loss": 0.8707, "step": 1003 }, { "epoch": 0.16, "learning_rate": 5.360384410037373e-05, "loss": 0.8788, "step": 1004 }, { "epoch": 0.16, "learning_rate": 5.3657234383342234e-05, "loss": 0.8398, "step": 1005 }, { "epoch": 0.16, "learning_rate": 5.3710624666310736e-05, "loss": 0.9927, "step": 1006 }, { "epoch": 0.16, "learning_rate": 5.376401494927923e-05, "loss": 1.0283, "step": 1007 }, { "epoch": 0.16, "learning_rate": 5.381740523224773e-05, "loss": 0.9076, "step": 1008 }, { "epoch": 0.16, "learning_rate": 5.387079551521623e-05, "loss": 0.9191, "step": 1009 }, { "epoch": 0.16, "learning_rate": 5.3924185798184733e-05, "loss": 0.8842, "step": 1010 }, { "epoch": 0.16, "learning_rate": 5.3977576081153235e-05, "loss": 0.9712, "step": 1011 }, { "epoch": 0.16, "learning_rate": 5.4030966364121736e-05, "loss": 0.8923, "step": 1012 }, { "epoch": 0.16, "learning_rate": 5.408435664709023e-05, "loss": 0.917, "step": 1013 }, { "epoch": 0.16, "learning_rate": 5.413774693005873e-05, "loss": 0.9257, "step": 1014 }, { "epoch": 0.16, "learning_rate": 5.419113721302723e-05, "loss": 0.9218, "step": 1015 }, { "epoch": 0.16, "learning_rate": 5.4244527495995734e-05, "loss": 0.8879, "step": 1016 }, { "epoch": 0.16, "learning_rate": 5.4297917778964235e-05, "loss": 0.8885, "step": 1017 }, { "epoch": 0.16, "learning_rate": 5.435130806193273e-05, "loss": 0.9342, "step": 1018 }, { "epoch": 0.16, "learning_rate": 5.440469834490123e-05, "loss": 0.8413, "step": 1019 }, { "epoch": 0.16, "learning_rate": 5.445808862786973e-05, "loss": 0.8647, "step": 1020 }, { "epoch": 0.16, "learning_rate": 5.451147891083823e-05, "loss": 0.8869, "step": 1021 }, { "epoch": 0.16, "learning_rate": 5.4564869193806734e-05, "loss": 0.9446, "step": 1022 }, { "epoch": 0.16, "learning_rate": 5.461825947677522e-05, "loss": 0.9359, "step": 1023 }, { "epoch": 0.16, "learning_rate": 5.467164975974372e-05, "loss": 0.9663, "step": 1024 }, { "epoch": 0.16, "learning_rate": 5.472504004271223e-05, "loss": 0.8769, "step": 1025 }, { "epoch": 0.16, "learning_rate": 5.477843032568073e-05, "loss": 1.0309, "step": 1026 }, { "epoch": 0.16, "learning_rate": 5.483182060864923e-05, "loss": 0.9851, "step": 1027 }, { "epoch": 0.16, "learning_rate": 5.4885210891617734e-05, "loss": 0.8891, "step": 1028 }, { "epoch": 0.16, "learning_rate": 5.493860117458622e-05, "loss": 0.8986, "step": 1029 }, { "epoch": 0.16, "learning_rate": 5.499199145755472e-05, "loss": 0.8957, "step": 1030 }, { "epoch": 0.17, "learning_rate": 5.5045381740523224e-05, "loss": 0.9819, "step": 1031 }, { "epoch": 0.17, "learning_rate": 5.509877202349173e-05, "loss": 0.8133, "step": 1032 }, { "epoch": 0.17, "learning_rate": 5.515216230646023e-05, "loss": 0.9502, "step": 1033 }, { "epoch": 0.17, "learning_rate": 5.520555258942872e-05, "loss": 0.9164, "step": 1034 }, { "epoch": 0.17, "learning_rate": 5.525894287239722e-05, "loss": 0.9338, "step": 1035 }, { "epoch": 0.17, "learning_rate": 5.531233315536572e-05, "loss": 0.9865, "step": 1036 }, { "epoch": 0.17, "learning_rate": 5.5365723438334224e-05, "loss": 0.8958, "step": 1037 }, { "epoch": 0.17, "learning_rate": 5.5419113721302726e-05, "loss": 0.9655, "step": 1038 }, { "epoch": 0.17, "learning_rate": 5.5472504004271234e-05, "loss": 0.9732, "step": 1039 }, { "epoch": 0.17, "learning_rate": 5.552589428723972e-05, "loss": 0.946, "step": 1040 }, { "epoch": 0.17, "learning_rate": 5.557928457020822e-05, "loss": 0.8843, "step": 1041 }, { "epoch": 0.17, "learning_rate": 5.5632674853176724e-05, "loss": 0.9143, "step": 1042 }, { "epoch": 0.17, "learning_rate": 5.5686065136145225e-05, "loss": 0.9969, "step": 1043 }, { "epoch": 0.17, "learning_rate": 5.5739455419113726e-05, "loss": 0.9166, "step": 1044 }, { "epoch": 0.17, "learning_rate": 5.579284570208222e-05, "loss": 0.8999, "step": 1045 }, { "epoch": 0.17, "learning_rate": 5.584623598505072e-05, "loss": 0.9461, "step": 1046 }, { "epoch": 0.17, "learning_rate": 5.589962626801922e-05, "loss": 0.9424, "step": 1047 }, { "epoch": 0.17, "learning_rate": 5.5953016550987724e-05, "loss": 0.9453, "step": 1048 }, { "epoch": 0.17, "learning_rate": 5.6006406833956225e-05, "loss": 0.9146, "step": 1049 }, { "epoch": 0.17, "learning_rate": 5.605979711692472e-05, "loss": 0.8507, "step": 1050 }, { "epoch": 0.17, "learning_rate": 5.611318739989322e-05, "loss": 0.8735, "step": 1051 }, { "epoch": 0.17, "learning_rate": 5.616657768286172e-05, "loss": 0.9325, "step": 1052 }, { "epoch": 0.17, "learning_rate": 5.621996796583022e-05, "loss": 0.9739, "step": 1053 }, { "epoch": 0.17, "learning_rate": 5.6273358248798724e-05, "loss": 0.9019, "step": 1054 }, { "epoch": 0.17, "learning_rate": 5.6326748531767225e-05, "loss": 0.9544, "step": 1055 }, { "epoch": 0.17, "learning_rate": 5.638013881473572e-05, "loss": 0.8832, "step": 1056 }, { "epoch": 0.17, "learning_rate": 5.643352909770422e-05, "loss": 0.9707, "step": 1057 }, { "epoch": 0.17, "learning_rate": 5.648691938067272e-05, "loss": 1.0006, "step": 1058 }, { "epoch": 0.17, "learning_rate": 5.654030966364122e-05, "loss": 0.937, "step": 1059 }, { "epoch": 0.17, "learning_rate": 5.6593699946609724e-05, "loss": 0.9387, "step": 1060 }, { "epoch": 0.17, "learning_rate": 5.664709022957821e-05, "loss": 0.9532, "step": 1061 }, { "epoch": 0.17, "learning_rate": 5.670048051254671e-05, "loss": 0.9049, "step": 1062 }, { "epoch": 0.17, "learning_rate": 5.675387079551522e-05, "loss": 0.9397, "step": 1063 }, { "epoch": 0.17, "learning_rate": 5.680726107848372e-05, "loss": 0.8399, "step": 1064 }, { "epoch": 0.17, "learning_rate": 5.6860651361452223e-05, "loss": 0.9044, "step": 1065 }, { "epoch": 0.17, "learning_rate": 5.6914041644420725e-05, "loss": 0.9115, "step": 1066 }, { "epoch": 0.17, "learning_rate": 5.696743192738921e-05, "loss": 0.9211, "step": 1067 }, { "epoch": 0.17, "learning_rate": 5.702082221035771e-05, "loss": 0.9522, "step": 1068 }, { "epoch": 0.17, "learning_rate": 5.7074212493326215e-05, "loss": 0.9381, "step": 1069 }, { "epoch": 0.17, "learning_rate": 5.712760277629472e-05, "loss": 0.8777, "step": 1070 }, { "epoch": 0.17, "learning_rate": 5.7180993059263224e-05, "loss": 0.8798, "step": 1071 }, { "epoch": 0.17, "learning_rate": 5.723438334223171e-05, "loss": 0.9032, "step": 1072 }, { "epoch": 0.17, "learning_rate": 5.728777362520021e-05, "loss": 0.8731, "step": 1073 }, { "epoch": 0.17, "learning_rate": 5.7341163908168714e-05, "loss": 0.8669, "step": 1074 }, { "epoch": 0.17, "learning_rate": 5.7394554191137215e-05, "loss": 0.8302, "step": 1075 }, { "epoch": 0.17, "learning_rate": 5.7447944474105716e-05, "loss": 1.0601, "step": 1076 }, { "epoch": 0.17, "learning_rate": 5.750133475707421e-05, "loss": 0.8658, "step": 1077 }, { "epoch": 0.17, "learning_rate": 5.755472504004271e-05, "loss": 0.8648, "step": 1078 }, { "epoch": 0.17, "learning_rate": 5.760811532301121e-05, "loss": 0.9048, "step": 1079 }, { "epoch": 0.17, "learning_rate": 5.7661505605979714e-05, "loss": 0.9161, "step": 1080 }, { "epoch": 0.17, "learning_rate": 5.7714895888948215e-05, "loss": 0.9327, "step": 1081 }, { "epoch": 0.17, "learning_rate": 5.7768286171916716e-05, "loss": 1.0074, "step": 1082 }, { "epoch": 0.17, "learning_rate": 5.782167645488521e-05, "loss": 0.8105, "step": 1083 }, { "epoch": 0.17, "learning_rate": 5.787506673785371e-05, "loss": 0.9889, "step": 1084 }, { "epoch": 0.17, "learning_rate": 5.792845702082221e-05, "loss": 0.9604, "step": 1085 }, { "epoch": 0.17, "learning_rate": 5.7981847303790714e-05, "loss": 0.9047, "step": 1086 }, { "epoch": 0.17, "learning_rate": 5.8035237586759215e-05, "loss": 0.902, "step": 1087 }, { "epoch": 0.17, "learning_rate": 5.808862786972771e-05, "loss": 0.8981, "step": 1088 }, { "epoch": 0.17, "learning_rate": 5.814201815269621e-05, "loss": 0.869, "step": 1089 }, { "epoch": 0.17, "learning_rate": 5.819540843566471e-05, "loss": 0.9099, "step": 1090 }, { "epoch": 0.17, "learning_rate": 5.824879871863321e-05, "loss": 0.927, "step": 1091 }, { "epoch": 0.17, "learning_rate": 5.8302189001601714e-05, "loss": 0.9693, "step": 1092 }, { "epoch": 0.18, "learning_rate": 5.8355579284570216e-05, "loss": 0.97, "step": 1093 }, { "epoch": 0.18, "learning_rate": 5.840896956753871e-05, "loss": 0.9281, "step": 1094 }, { "epoch": 0.18, "learning_rate": 5.846235985050721e-05, "loss": 0.9763, "step": 1095 }, { "epoch": 0.18, "learning_rate": 5.851575013347571e-05, "loss": 0.9987, "step": 1096 }, { "epoch": 0.18, "learning_rate": 5.8569140416444213e-05, "loss": 0.8638, "step": 1097 }, { "epoch": 0.18, "learning_rate": 5.8622530699412715e-05, "loss": 0.8477, "step": 1098 }, { "epoch": 0.18, "learning_rate": 5.86759209823812e-05, "loss": 0.9098, "step": 1099 }, { "epoch": 0.18, "learning_rate": 5.872931126534971e-05, "loss": 0.9676, "step": 1100 }, { "epoch": 0.18, "learning_rate": 5.878270154831821e-05, "loss": 0.9158, "step": 1101 }, { "epoch": 0.18, "learning_rate": 5.883609183128671e-05, "loss": 0.8121, "step": 1102 }, { "epoch": 0.18, "learning_rate": 5.8889482114255214e-05, "loss": 0.893, "step": 1103 }, { "epoch": 0.18, "learning_rate": 5.89428723972237e-05, "loss": 0.8988, "step": 1104 }, { "epoch": 0.18, "learning_rate": 5.89962626801922e-05, "loss": 0.8415, "step": 1105 }, { "epoch": 0.18, "learning_rate": 5.9049652963160704e-05, "loss": 0.934, "step": 1106 }, { "epoch": 0.18, "learning_rate": 5.910304324612921e-05, "loss": 0.8543, "step": 1107 }, { "epoch": 0.18, "learning_rate": 5.915643352909771e-05, "loss": 0.8616, "step": 1108 }, { "epoch": 0.18, "learning_rate": 5.9209823812066214e-05, "loss": 0.8998, "step": 1109 }, { "epoch": 0.18, "learning_rate": 5.92632140950347e-05, "loss": 1.0685, "step": 1110 }, { "epoch": 0.18, "learning_rate": 5.93166043780032e-05, "loss": 0.8824, "step": 1111 }, { "epoch": 0.18, "learning_rate": 5.9369994660971704e-05, "loss": 0.9549, "step": 1112 }, { "epoch": 0.18, "learning_rate": 5.9423384943940205e-05, "loss": 0.8697, "step": 1113 }, { "epoch": 0.18, "learning_rate": 5.947677522690871e-05, "loss": 1.0121, "step": 1114 }, { "epoch": 0.18, "learning_rate": 5.95301655098772e-05, "loss": 0.9389, "step": 1115 }, { "epoch": 0.18, "learning_rate": 5.95835557928457e-05, "loss": 0.9114, "step": 1116 }, { "epoch": 0.18, "learning_rate": 5.96369460758142e-05, "loss": 0.9574, "step": 1117 }, { "epoch": 0.18, "learning_rate": 5.9690336358782704e-05, "loss": 0.8282, "step": 1118 }, { "epoch": 0.18, "learning_rate": 5.9743726641751205e-05, "loss": 0.9076, "step": 1119 }, { "epoch": 0.18, "learning_rate": 5.9797116924719707e-05, "loss": 0.9902, "step": 1120 }, { "epoch": 0.18, "learning_rate": 5.98505072076882e-05, "loss": 0.89, "step": 1121 }, { "epoch": 0.18, "learning_rate": 5.99038974906567e-05, "loss": 1.0033, "step": 1122 }, { "epoch": 0.18, "learning_rate": 5.99572877736252e-05, "loss": 0.9813, "step": 1123 }, { "epoch": 0.18, "learning_rate": 6.0010678056593704e-05, "loss": 0.818, "step": 1124 }, { "epoch": 0.18, "learning_rate": 6.0064068339562206e-05, "loss": 0.9318, "step": 1125 }, { "epoch": 0.18, "learning_rate": 6.01174586225307e-05, "loss": 0.9175, "step": 1126 }, { "epoch": 0.18, "learning_rate": 6.01708489054992e-05, "loss": 0.883, "step": 1127 }, { "epoch": 0.18, "learning_rate": 6.02242391884677e-05, "loss": 0.9147, "step": 1128 }, { "epoch": 0.18, "learning_rate": 6.0277629471436204e-05, "loss": 0.9034, "step": 1129 }, { "epoch": 0.18, "learning_rate": 6.0331019754404705e-05, "loss": 0.941, "step": 1130 }, { "epoch": 0.18, "learning_rate": 6.03844100373732e-05, "loss": 0.8958, "step": 1131 }, { "epoch": 0.18, "learning_rate": 6.04378003203417e-05, "loss": 0.9412, "step": 1132 }, { "epoch": 0.18, "learning_rate": 6.04911906033102e-05, "loss": 0.9052, "step": 1133 }, { "epoch": 0.18, "learning_rate": 6.05445808862787e-05, "loss": 0.9717, "step": 1134 }, { "epoch": 0.18, "learning_rate": 6.0597971169247204e-05, "loss": 0.9275, "step": 1135 }, { "epoch": 0.18, "learning_rate": 6.0651361452215705e-05, "loss": 0.9315, "step": 1136 }, { "epoch": 0.18, "learning_rate": 6.070475173518419e-05, "loss": 0.8984, "step": 1137 }, { "epoch": 0.18, "learning_rate": 6.07581420181527e-05, "loss": 0.8888, "step": 1138 }, { "epoch": 0.18, "learning_rate": 6.08115323011212e-05, "loss": 0.8673, "step": 1139 }, { "epoch": 0.18, "learning_rate": 6.08649225840897e-05, "loss": 0.8089, "step": 1140 }, { "epoch": 0.18, "learning_rate": 6.0918312867058204e-05, "loss": 0.9689, "step": 1141 }, { "epoch": 0.18, "learning_rate": 6.097170315002669e-05, "loss": 0.9863, "step": 1142 }, { "epoch": 0.18, "learning_rate": 6.102509343299519e-05, "loss": 0.8704, "step": 1143 }, { "epoch": 0.18, "learning_rate": 6.10784837159637e-05, "loss": 0.9, "step": 1144 }, { "epoch": 0.18, "learning_rate": 6.11318739989322e-05, "loss": 0.9445, "step": 1145 }, { "epoch": 0.18, "learning_rate": 6.11852642819007e-05, "loss": 0.8677, "step": 1146 }, { "epoch": 0.18, "learning_rate": 6.12386545648692e-05, "loss": 0.8494, "step": 1147 }, { "epoch": 0.18, "learning_rate": 6.129204484783769e-05, "loss": 0.8693, "step": 1148 }, { "epoch": 0.18, "learning_rate": 6.13454351308062e-05, "loss": 0.8204, "step": 1149 }, { "epoch": 0.18, "learning_rate": 6.13988254137747e-05, "loss": 0.9019, "step": 1150 }, { "epoch": 0.18, "learning_rate": 6.14522156967432e-05, "loss": 1.0226, "step": 1151 }, { "epoch": 0.18, "learning_rate": 6.15056059797117e-05, "loss": 0.9893, "step": 1152 }, { "epoch": 0.18, "learning_rate": 6.155899626268019e-05, "loss": 0.9549, "step": 1153 }, { "epoch": 0.18, "learning_rate": 6.16123865456487e-05, "loss": 0.9373, "step": 1154 }, { "epoch": 0.18, "learning_rate": 6.16657768286172e-05, "loss": 0.9162, "step": 1155 }, { "epoch": 0.19, "learning_rate": 6.17191671115857e-05, "loss": 0.9631, "step": 1156 }, { "epoch": 0.19, "learning_rate": 6.17725573945542e-05, "loss": 0.9009, "step": 1157 }, { "epoch": 0.19, "learning_rate": 6.182594767752269e-05, "loss": 0.8709, "step": 1158 }, { "epoch": 0.19, "learning_rate": 6.187933796049118e-05, "loss": 0.8624, "step": 1159 }, { "epoch": 0.19, "learning_rate": 6.193272824345969e-05, "loss": 0.9453, "step": 1160 }, { "epoch": 0.19, "learning_rate": 6.19861185264282e-05, "loss": 0.9892, "step": 1161 }, { "epoch": 0.19, "learning_rate": 6.20395088093967e-05, "loss": 0.9007, "step": 1162 }, { "epoch": 0.19, "learning_rate": 6.20928990923652e-05, "loss": 0.9035, "step": 1163 }, { "epoch": 0.19, "learning_rate": 6.214628937533368e-05, "loss": 0.9432, "step": 1164 }, { "epoch": 0.19, "learning_rate": 6.219967965830219e-05, "loss": 0.9259, "step": 1165 }, { "epoch": 0.19, "learning_rate": 6.225306994127069e-05, "loss": 0.9206, "step": 1166 }, { "epoch": 0.19, "learning_rate": 6.23064602242392e-05, "loss": 0.9571, "step": 1167 }, { "epoch": 0.19, "learning_rate": 6.23598505072077e-05, "loss": 0.8462, "step": 1168 }, { "epoch": 0.19, "learning_rate": 6.241324079017618e-05, "loss": 0.9078, "step": 1169 }, { "epoch": 0.19, "learning_rate": 6.246663107314469e-05, "loss": 0.9075, "step": 1170 }, { "epoch": 0.19, "learning_rate": 6.252002135611319e-05, "loss": 0.9518, "step": 1171 }, { "epoch": 0.19, "learning_rate": 6.257341163908169e-05, "loss": 0.8839, "step": 1172 }, { "epoch": 0.19, "learning_rate": 6.262680192205019e-05, "loss": 0.9018, "step": 1173 }, { "epoch": 0.19, "learning_rate": 6.268019220501868e-05, "loss": 0.926, "step": 1174 }, { "epoch": 0.19, "learning_rate": 6.273358248798719e-05, "loss": 0.8443, "step": 1175 }, { "epoch": 0.19, "learning_rate": 6.278697277095568e-05, "loss": 0.9463, "step": 1176 }, { "epoch": 0.19, "learning_rate": 6.284036305392419e-05, "loss": 1.0064, "step": 1177 }, { "epoch": 0.19, "learning_rate": 6.289375333689269e-05, "loss": 0.9295, "step": 1178 }, { "epoch": 0.19, "learning_rate": 6.29471436198612e-05, "loss": 0.9553, "step": 1179 }, { "epoch": 0.19, "learning_rate": 6.300053390282969e-05, "loss": 0.8895, "step": 1180 }, { "epoch": 0.19, "learning_rate": 6.305392418579818e-05, "loss": 0.9462, "step": 1181 }, { "epoch": 0.19, "learning_rate": 6.310731446876669e-05, "loss": 0.9663, "step": 1182 }, { "epoch": 0.19, "learning_rate": 6.316070475173519e-05, "loss": 0.9546, "step": 1183 }, { "epoch": 0.19, "learning_rate": 6.32140950347037e-05, "loss": 0.9086, "step": 1184 }, { "epoch": 0.19, "learning_rate": 6.326748531767219e-05, "loss": 0.9395, "step": 1185 }, { "epoch": 0.19, "learning_rate": 6.332087560064068e-05, "loss": 0.9147, "step": 1186 }, { "epoch": 0.19, "learning_rate": 6.337426588360919e-05, "loss": 0.9235, "step": 1187 }, { "epoch": 0.19, "learning_rate": 6.342765616657768e-05, "loss": 0.8472, "step": 1188 }, { "epoch": 0.19, "learning_rate": 6.348104644954619e-05, "loss": 1.0364, "step": 1189 }, { "epoch": 0.19, "learning_rate": 6.353443673251469e-05, "loss": 0.9346, "step": 1190 }, { "epoch": 0.19, "learning_rate": 6.358782701548318e-05, "loss": 0.9301, "step": 1191 }, { "epoch": 0.19, "learning_rate": 6.364121729845169e-05, "loss": 0.9486, "step": 1192 }, { "epoch": 0.19, "learning_rate": 6.369460758142018e-05, "loss": 0.8825, "step": 1193 }, { "epoch": 0.19, "learning_rate": 6.374799786438869e-05, "loss": 0.8253, "step": 1194 }, { "epoch": 0.19, "learning_rate": 6.380138814735719e-05, "loss": 0.8611, "step": 1195 }, { "epoch": 0.19, "learning_rate": 6.385477843032568e-05, "loss": 1.0273, "step": 1196 }, { "epoch": 0.19, "learning_rate": 6.390816871329417e-05, "loss": 0.9665, "step": 1197 }, { "epoch": 0.19, "learning_rate": 6.396155899626268e-05, "loss": 0.9929, "step": 1198 }, { "epoch": 0.19, "learning_rate": 6.401494927923119e-05, "loss": 0.9068, "step": 1199 }, { "epoch": 0.19, "learning_rate": 6.406833956219969e-05, "loss": 0.9227, "step": 1200 }, { "epoch": 0.19, "learning_rate": 6.412172984516818e-05, "loss": 0.9418, "step": 1201 }, { "epoch": 0.19, "learning_rate": 6.417512012813667e-05, "loss": 0.9268, "step": 1202 }, { "epoch": 0.19, "learning_rate": 6.422851041110518e-05, "loss": 0.9017, "step": 1203 }, { "epoch": 0.19, "learning_rate": 6.428190069407368e-05, "loss": 0.8231, "step": 1204 }, { "epoch": 0.19, "learning_rate": 6.433529097704218e-05, "loss": 0.9349, "step": 1205 }, { "epoch": 0.19, "learning_rate": 6.438868126001069e-05, "loss": 0.8773, "step": 1206 }, { "epoch": 0.19, "learning_rate": 6.444207154297917e-05, "loss": 0.924, "step": 1207 }, { "epoch": 0.19, "learning_rate": 6.449546182594768e-05, "loss": 0.8924, "step": 1208 }, { "epoch": 0.19, "learning_rate": 6.454885210891618e-05, "loss": 0.8913, "step": 1209 }, { "epoch": 0.19, "learning_rate": 6.460224239188468e-05, "loss": 0.8729, "step": 1210 }, { "epoch": 0.19, "learning_rate": 6.465563267485318e-05, "loss": 0.8717, "step": 1211 }, { "epoch": 0.19, "learning_rate": 6.470902295782167e-05, "loss": 0.8891, "step": 1212 }, { "epoch": 0.19, "learning_rate": 6.476241324079018e-05, "loss": 0.8603, "step": 1213 }, { "epoch": 0.19, "learning_rate": 6.481580352375867e-05, "loss": 0.8654, "step": 1214 }, { "epoch": 0.19, "learning_rate": 6.486919380672718e-05, "loss": 0.886, "step": 1215 }, { "epoch": 0.19, "learning_rate": 6.492258408969568e-05, "loss": 0.9277, "step": 1216 }, { "epoch": 0.19, "learning_rate": 6.497597437266418e-05, "loss": 0.8655, "step": 1217 }, { "epoch": 0.2, "learning_rate": 6.502936465563268e-05, "loss": 0.8219, "step": 1218 }, { "epoch": 0.2, "learning_rate": 6.508275493860117e-05, "loss": 0.8388, "step": 1219 }, { "epoch": 0.2, "learning_rate": 6.513614522156968e-05, "loss": 0.8832, "step": 1220 }, { "epoch": 0.2, "learning_rate": 6.518953550453818e-05, "loss": 0.8219, "step": 1221 }, { "epoch": 0.2, "learning_rate": 6.524292578750668e-05, "loss": 0.886, "step": 1222 }, { "epoch": 0.2, "learning_rate": 6.529631607047518e-05, "loss": 0.8956, "step": 1223 }, { "epoch": 0.2, "learning_rate": 6.534970635344367e-05, "loss": 0.8716, "step": 1224 }, { "epoch": 0.2, "learning_rate": 6.540309663641218e-05, "loss": 0.8974, "step": 1225 }, { "epoch": 0.2, "learning_rate": 6.545648691938067e-05, "loss": 0.9102, "step": 1226 }, { "epoch": 0.2, "learning_rate": 6.550987720234918e-05, "loss": 0.908, "step": 1227 }, { "epoch": 0.2, "learning_rate": 6.556326748531766e-05, "loss": 0.8737, "step": 1228 }, { "epoch": 0.2, "learning_rate": 6.561665776828617e-05, "loss": 0.8208, "step": 1229 }, { "epoch": 0.2, "learning_rate": 6.567004805125468e-05, "loss": 0.8666, "step": 1230 }, { "epoch": 0.2, "learning_rate": 6.572343833422317e-05, "loss": 0.9196, "step": 1231 }, { "epoch": 0.2, "learning_rate": 6.577682861719168e-05, "loss": 0.9293, "step": 1232 }, { "epoch": 0.2, "learning_rate": 6.583021890016018e-05, "loss": 0.9485, "step": 1233 }, { "epoch": 0.2, "learning_rate": 6.588360918312867e-05, "loss": 0.9342, "step": 1234 }, { "epoch": 0.2, "learning_rate": 6.593699946609717e-05, "loss": 1.0187, "step": 1235 }, { "epoch": 0.2, "learning_rate": 6.599038974906567e-05, "loss": 0.9642, "step": 1236 }, { "epoch": 0.2, "learning_rate": 6.604378003203418e-05, "loss": 0.8919, "step": 1237 }, { "epoch": 0.2, "learning_rate": 6.609717031500268e-05, "loss": 0.9028, "step": 1238 }, { "epoch": 0.2, "learning_rate": 6.615056059797117e-05, "loss": 0.8503, "step": 1239 }, { "epoch": 0.2, "learning_rate": 6.620395088093966e-05, "loss": 0.8978, "step": 1240 }, { "epoch": 0.2, "learning_rate": 6.625734116390817e-05, "loss": 0.8714, "step": 1241 }, { "epoch": 0.2, "learning_rate": 6.631073144687667e-05, "loss": 0.9138, "step": 1242 }, { "epoch": 0.2, "learning_rate": 6.636412172984517e-05, "loss": 0.8918, "step": 1243 }, { "epoch": 0.2, "learning_rate": 6.641751201281368e-05, "loss": 1.0415, "step": 1244 }, { "epoch": 0.2, "learning_rate": 6.647090229578216e-05, "loss": 0.8428, "step": 1245 }, { "epoch": 0.2, "learning_rate": 6.652429257875067e-05, "loss": 0.9113, "step": 1246 }, { "epoch": 0.2, "learning_rate": 6.657768286171917e-05, "loss": 0.9433, "step": 1247 }, { "epoch": 0.2, "learning_rate": 6.663107314468767e-05, "loss": 0.9067, "step": 1248 }, { "epoch": 0.2, "learning_rate": 6.668446342765617e-05, "loss": 0.9105, "step": 1249 }, { "epoch": 0.2, "learning_rate": 6.673785371062466e-05, "loss": 0.8677, "step": 1250 }, { "epoch": 0.2, "learning_rate": 6.679124399359317e-05, "loss": 0.873, "step": 1251 }, { "epoch": 0.2, "learning_rate": 6.684463427656166e-05, "loss": 0.8229, "step": 1252 }, { "epoch": 0.2, "learning_rate": 6.689802455953017e-05, "loss": 0.9975, "step": 1253 }, { "epoch": 0.2, "learning_rate": 6.695141484249867e-05, "loss": 0.9032, "step": 1254 }, { "epoch": 0.2, "learning_rate": 6.700480512546716e-05, "loss": 0.8547, "step": 1255 }, { "epoch": 0.2, "learning_rate": 6.705819540843567e-05, "loss": 0.864, "step": 1256 }, { "epoch": 0.2, "learning_rate": 6.711158569140416e-05, "loss": 0.9589, "step": 1257 }, { "epoch": 0.2, "learning_rate": 6.716497597437267e-05, "loss": 0.9844, "step": 1258 }, { "epoch": 0.2, "learning_rate": 6.721836625734117e-05, "loss": 0.8763, "step": 1259 }, { "epoch": 0.2, "learning_rate": 6.727175654030967e-05, "loss": 0.9617, "step": 1260 }, { "epoch": 0.2, "learning_rate": 6.732514682327817e-05, "loss": 0.9602, "step": 1261 }, { "epoch": 0.2, "learning_rate": 6.737853710624666e-05, "loss": 0.9225, "step": 1262 }, { "epoch": 0.2, "learning_rate": 6.743192738921517e-05, "loss": 0.8524, "step": 1263 }, { "epoch": 0.2, "learning_rate": 6.748531767218367e-05, "loss": 0.9324, "step": 1264 }, { "epoch": 0.2, "learning_rate": 6.753870795515217e-05, "loss": 0.8952, "step": 1265 }, { "epoch": 0.2, "learning_rate": 6.759209823812065e-05, "loss": 0.9472, "step": 1266 }, { "epoch": 0.2, "learning_rate": 6.764548852108916e-05, "loss": 0.7954, "step": 1267 }, { "epoch": 0.2, "learning_rate": 6.769887880405767e-05, "loss": 0.9414, "step": 1268 }, { "epoch": 0.2, "learning_rate": 6.775226908702616e-05, "loss": 0.9365, "step": 1269 }, { "epoch": 0.2, "learning_rate": 6.780565936999467e-05, "loss": 0.907, "step": 1270 }, { "epoch": 0.2, "learning_rate": 6.785904965296317e-05, "loss": 0.8935, "step": 1271 }, { "epoch": 0.2, "learning_rate": 6.791243993593166e-05, "loss": 0.9478, "step": 1272 }, { "epoch": 0.2, "learning_rate": 6.796583021890016e-05, "loss": 0.9091, "step": 1273 }, { "epoch": 0.2, "learning_rate": 6.801922050186866e-05, "loss": 0.9287, "step": 1274 }, { "epoch": 0.2, "learning_rate": 6.807261078483717e-05, "loss": 0.9264, "step": 1275 }, { "epoch": 0.2, "learning_rate": 6.812600106780567e-05, "loss": 0.8507, "step": 1276 }, { "epoch": 0.2, "learning_rate": 6.817939135077416e-05, "loss": 0.8773, "step": 1277 }, { "epoch": 0.2, "learning_rate": 6.823278163374265e-05, "loss": 0.8684, "step": 1278 }, { "epoch": 0.2, "learning_rate": 6.828617191671116e-05, "loss": 0.8758, "step": 1279 }, { "epoch": 0.21, "learning_rate": 6.833956219967966e-05, "loss": 0.8403, "step": 1280 }, { "epoch": 0.21, "learning_rate": 6.839295248264816e-05, "loss": 0.7819, "step": 1281 }, { "epoch": 0.21, "learning_rate": 6.844634276561666e-05, "loss": 0.8716, "step": 1282 }, { "epoch": 0.21, "learning_rate": 6.849973304858515e-05, "loss": 0.8238, "step": 1283 }, { "epoch": 0.21, "learning_rate": 6.855312333155366e-05, "loss": 0.9572, "step": 1284 }, { "epoch": 0.21, "learning_rate": 6.860651361452216e-05, "loss": 0.9211, "step": 1285 }, { "epoch": 0.21, "learning_rate": 6.865990389749066e-05, "loss": 0.9095, "step": 1286 }, { "epoch": 0.21, "learning_rate": 6.871329418045916e-05, "loss": 0.9145, "step": 1287 }, { "epoch": 0.21, "learning_rate": 6.876668446342765e-05, "loss": 0.876, "step": 1288 }, { "epoch": 0.21, "learning_rate": 6.882007474639616e-05, "loss": 0.9101, "step": 1289 }, { "epoch": 0.21, "learning_rate": 6.887346502936466e-05, "loss": 0.8469, "step": 1290 }, { "epoch": 0.21, "learning_rate": 6.892685531233316e-05, "loss": 0.9589, "step": 1291 }, { "epoch": 0.21, "learning_rate": 6.898024559530166e-05, "loss": 0.8757, "step": 1292 }, { "epoch": 0.21, "learning_rate": 6.903363587827015e-05, "loss": 0.8839, "step": 1293 }, { "epoch": 0.21, "learning_rate": 6.908702616123866e-05, "loss": 0.8288, "step": 1294 }, { "epoch": 0.21, "learning_rate": 6.914041644420715e-05, "loss": 0.9512, "step": 1295 }, { "epoch": 0.21, "learning_rate": 6.919380672717566e-05, "loss": 0.7996, "step": 1296 }, { "epoch": 0.21, "learning_rate": 6.924719701014416e-05, "loss": 1.0084, "step": 1297 }, { "epoch": 0.21, "learning_rate": 6.930058729311266e-05, "loss": 0.9008, "step": 1298 }, { "epoch": 0.21, "learning_rate": 6.935397757608116e-05, "loss": 0.8817, "step": 1299 }, { "epoch": 0.21, "learning_rate": 6.940736785904965e-05, "loss": 0.8224, "step": 1300 }, { "epoch": 0.21, "learning_rate": 6.946075814201816e-05, "loss": 0.9009, "step": 1301 }, { "epoch": 0.21, "learning_rate": 6.951414842498666e-05, "loss": 0.8801, "step": 1302 }, { "epoch": 0.21, "learning_rate": 6.956753870795516e-05, "loss": 0.9162, "step": 1303 }, { "epoch": 0.21, "learning_rate": 6.962092899092364e-05, "loss": 0.8031, "step": 1304 }, { "epoch": 0.21, "learning_rate": 6.967431927389215e-05, "loss": 0.8807, "step": 1305 }, { "epoch": 0.21, "learning_rate": 6.972770955686066e-05, "loss": 0.8398, "step": 1306 }, { "epoch": 0.21, "learning_rate": 6.978109983982915e-05, "loss": 0.8778, "step": 1307 }, { "epoch": 0.21, "learning_rate": 6.983449012279766e-05, "loss": 0.9986, "step": 1308 }, { "epoch": 0.21, "learning_rate": 6.988788040576614e-05, "loss": 0.9362, "step": 1309 }, { "epoch": 0.21, "learning_rate": 6.994127068873465e-05, "loss": 0.789, "step": 1310 }, { "epoch": 0.21, "learning_rate": 6.999466097170315e-05, "loss": 0.8281, "step": 1311 }, { "epoch": 0.21, "learning_rate": 7.004805125467165e-05, "loss": 0.9217, "step": 1312 }, { "epoch": 0.21, "learning_rate": 7.010144153764016e-05, "loss": 0.9394, "step": 1313 }, { "epoch": 0.21, "learning_rate": 7.015483182060866e-05, "loss": 0.8371, "step": 1314 }, { "epoch": 0.21, "learning_rate": 7.020822210357715e-05, "loss": 0.8969, "step": 1315 }, { "epoch": 0.21, "learning_rate": 7.026161238654564e-05, "loss": 0.905, "step": 1316 }, { "epoch": 0.21, "learning_rate": 7.031500266951415e-05, "loss": 0.9328, "step": 1317 }, { "epoch": 0.21, "learning_rate": 7.036839295248265e-05, "loss": 0.8682, "step": 1318 }, { "epoch": 0.21, "learning_rate": 7.042178323545116e-05, "loss": 0.9279, "step": 1319 }, { "epoch": 0.21, "learning_rate": 7.047517351841965e-05, "loss": 0.9384, "step": 1320 }, { "epoch": 0.21, "learning_rate": 7.052856380138814e-05, "loss": 0.8907, "step": 1321 }, { "epoch": 0.21, "learning_rate": 7.058195408435665e-05, "loss": 0.9216, "step": 1322 }, { "epoch": 0.21, "learning_rate": 7.063534436732515e-05, "loss": 0.8603, "step": 1323 }, { "epoch": 0.21, "learning_rate": 7.068873465029365e-05, "loss": 0.8464, "step": 1324 }, { "epoch": 0.21, "learning_rate": 7.074212493326215e-05, "loss": 0.8653, "step": 1325 }, { "epoch": 0.21, "learning_rate": 7.079551521623064e-05, "loss": 0.8113, "step": 1326 }, { "epoch": 0.21, "learning_rate": 7.084890549919915e-05, "loss": 0.9124, "step": 1327 }, { "epoch": 0.21, "learning_rate": 7.090229578216765e-05, "loss": 0.8042, "step": 1328 }, { "epoch": 0.21, "learning_rate": 7.095568606513615e-05, "loss": 0.9387, "step": 1329 }, { "epoch": 0.21, "learning_rate": 7.100907634810465e-05, "loss": 0.9877, "step": 1330 }, { "epoch": 0.21, "learning_rate": 7.106246663107314e-05, "loss": 0.9571, "step": 1331 }, { "epoch": 0.21, "learning_rate": 7.111585691404165e-05, "loss": 0.8685, "step": 1332 }, { "epoch": 0.21, "learning_rate": 7.116924719701014e-05, "loss": 0.925, "step": 1333 }, { "epoch": 0.21, "learning_rate": 7.122263747997865e-05, "loss": 0.9153, "step": 1334 }, { "epoch": 0.21, "learning_rate": 7.127602776294715e-05, "loss": 0.9478, "step": 1335 }, { "epoch": 0.21, "learning_rate": 7.132941804591564e-05, "loss": 0.8373, "step": 1336 }, { "epoch": 0.21, "learning_rate": 7.138280832888415e-05, "loss": 0.8942, "step": 1337 }, { "epoch": 0.21, "learning_rate": 7.143619861185264e-05, "loss": 0.894, "step": 1338 }, { "epoch": 0.21, "learning_rate": 7.148958889482115e-05, "loss": 0.8025, "step": 1339 }, { "epoch": 0.21, "learning_rate": 7.154297917778965e-05, "loss": 0.8658, "step": 1340 }, { "epoch": 0.21, "learning_rate": 7.159636946075815e-05, "loss": 1.0315, "step": 1341 }, { "epoch": 0.21, "learning_rate": 7.164975974372663e-05, "loss": 0.8811, "step": 1342 }, { "epoch": 0.22, "learning_rate": 7.170315002669514e-05, "loss": 0.9893, "step": 1343 }, { "epoch": 0.22, "learning_rate": 7.175654030966365e-05, "loss": 0.8728, "step": 1344 }, { "epoch": 0.22, "learning_rate": 7.180993059263214e-05, "loss": 0.88, "step": 1345 }, { "epoch": 0.22, "learning_rate": 7.186332087560065e-05, "loss": 0.9441, "step": 1346 }, { "epoch": 0.22, "learning_rate": 7.191671115856913e-05, "loss": 0.8617, "step": 1347 }, { "epoch": 0.22, "learning_rate": 7.197010144153764e-05, "loss": 0.883, "step": 1348 }, { "epoch": 0.22, "learning_rate": 7.202349172450614e-05, "loss": 0.9154, "step": 1349 }, { "epoch": 0.22, "learning_rate": 7.207688200747464e-05, "loss": 0.8474, "step": 1350 }, { "epoch": 0.22, "learning_rate": 7.213027229044315e-05, "loss": 0.8797, "step": 1351 }, { "epoch": 0.22, "learning_rate": 7.218366257341165e-05, "loss": 0.8851, "step": 1352 }, { "epoch": 0.22, "learning_rate": 7.223705285638014e-05, "loss": 0.9236, "step": 1353 }, { "epoch": 0.22, "learning_rate": 7.229044313934864e-05, "loss": 0.9766, "step": 1354 }, { "epoch": 0.22, "learning_rate": 7.234383342231714e-05, "loss": 0.7983, "step": 1355 }, { "epoch": 0.22, "learning_rate": 7.239722370528564e-05, "loss": 0.8142, "step": 1356 }, { "epoch": 0.22, "learning_rate": 7.245061398825415e-05, "loss": 0.8182, "step": 1357 }, { "epoch": 0.22, "learning_rate": 7.250400427122264e-05, "loss": 0.8334, "step": 1358 }, { "epoch": 0.22, "learning_rate": 7.255739455419113e-05, "loss": 0.8039, "step": 1359 }, { "epoch": 0.22, "learning_rate": 7.261078483715964e-05, "loss": 0.8159, "step": 1360 }, { "epoch": 0.22, "learning_rate": 7.266417512012814e-05, "loss": 0.9494, "step": 1361 }, { "epoch": 0.22, "learning_rate": 7.271756540309664e-05, "loss": 0.8358, "step": 1362 }, { "epoch": 0.22, "learning_rate": 7.277095568606514e-05, "loss": 0.8466, "step": 1363 }, { "epoch": 0.22, "learning_rate": 7.282434596903363e-05, "loss": 0.8691, "step": 1364 }, { "epoch": 0.22, "learning_rate": 7.287773625200214e-05, "loss": 0.8756, "step": 1365 }, { "epoch": 0.22, "learning_rate": 7.293112653497064e-05, "loss": 0.9289, "step": 1366 }, { "epoch": 0.22, "learning_rate": 7.298451681793914e-05, "loss": 0.9222, "step": 1367 }, { "epoch": 0.22, "learning_rate": 7.303790710090764e-05, "loss": 0.9822, "step": 1368 }, { "epoch": 0.22, "learning_rate": 7.309129738387613e-05, "loss": 0.8227, "step": 1369 }, { "epoch": 0.22, "learning_rate": 7.314468766684464e-05, "loss": 0.7719, "step": 1370 }, { "epoch": 0.22, "learning_rate": 7.319807794981313e-05, "loss": 0.9624, "step": 1371 }, { "epoch": 0.22, "learning_rate": 7.325146823278164e-05, "loss": 0.9175, "step": 1372 }, { "epoch": 0.22, "learning_rate": 7.330485851575014e-05, "loss": 0.8719, "step": 1373 }, { "epoch": 0.22, "learning_rate": 7.335824879871863e-05, "loss": 0.9663, "step": 1374 }, { "epoch": 0.22, "learning_rate": 7.341163908168714e-05, "loss": 0.8276, "step": 1375 }, { "epoch": 0.22, "learning_rate": 7.346502936465563e-05, "loss": 0.8192, "step": 1376 }, { "epoch": 0.22, "learning_rate": 7.351841964762414e-05, "loss": 0.8544, "step": 1377 }, { "epoch": 0.22, "learning_rate": 7.357180993059264e-05, "loss": 0.8247, "step": 1378 }, { "epoch": 0.22, "learning_rate": 7.362520021356114e-05, "loss": 0.9235, "step": 1379 }, { "epoch": 0.22, "learning_rate": 7.367859049652963e-05, "loss": 0.8735, "step": 1380 }, { "epoch": 0.22, "learning_rate": 7.373198077949813e-05, "loss": 0.9211, "step": 1381 }, { "epoch": 0.22, "learning_rate": 7.378537106246664e-05, "loss": 0.9341, "step": 1382 }, { "epoch": 0.22, "learning_rate": 7.383876134543514e-05, "loss": 0.8375, "step": 1383 }, { "epoch": 0.22, "learning_rate": 7.389215162840364e-05, "loss": 0.8517, "step": 1384 }, { "epoch": 0.22, "learning_rate": 7.394554191137212e-05, "loss": 0.8424, "step": 1385 }, { "epoch": 0.22, "learning_rate": 7.399893219434063e-05, "loss": 0.9248, "step": 1386 }, { "epoch": 0.22, "learning_rate": 7.405232247730913e-05, "loss": 0.9546, "step": 1387 }, { "epoch": 0.22, "learning_rate": 7.410571276027763e-05, "loss": 0.9176, "step": 1388 }, { "epoch": 0.22, "learning_rate": 7.415910304324614e-05, "loss": 0.8784, "step": 1389 }, { "epoch": 0.22, "learning_rate": 7.421249332621462e-05, "loss": 0.8658, "step": 1390 }, { "epoch": 0.22, "learning_rate": 7.426588360918313e-05, "loss": 0.8871, "step": 1391 }, { "epoch": 0.22, "learning_rate": 7.431927389215163e-05, "loss": 0.8839, "step": 1392 }, { "epoch": 0.22, "learning_rate": 7.437266417512013e-05, "loss": 0.8434, "step": 1393 }, { "epoch": 0.22, "learning_rate": 7.442605445808863e-05, "loss": 0.8728, "step": 1394 }, { "epoch": 0.22, "learning_rate": 7.447944474105714e-05, "loss": 0.8803, "step": 1395 }, { "epoch": 0.22, "learning_rate": 7.453283502402563e-05, "loss": 0.9148, "step": 1396 }, { "epoch": 0.22, "learning_rate": 7.458622530699412e-05, "loss": 0.8199, "step": 1397 }, { "epoch": 0.22, "learning_rate": 7.463961558996263e-05, "loss": 0.8858, "step": 1398 }, { "epoch": 0.22, "learning_rate": 7.469300587293113e-05, "loss": 0.8644, "step": 1399 }, { "epoch": 0.22, "learning_rate": 7.474639615589963e-05, "loss": 0.871, "step": 1400 }, { "epoch": 0.22, "learning_rate": 7.479978643886813e-05, "loss": 0.87, "step": 1401 }, { "epoch": 0.22, "learning_rate": 7.485317672183662e-05, "loss": 0.8647, "step": 1402 }, { "epoch": 0.22, "learning_rate": 7.490656700480513e-05, "loss": 0.8918, "step": 1403 }, { "epoch": 0.22, "learning_rate": 7.495995728777363e-05, "loss": 0.8168, "step": 1404 }, { "epoch": 0.23, "learning_rate": 7.501334757074213e-05, "loss": 0.845, "step": 1405 }, { "epoch": 0.23, "learning_rate": 7.506673785371063e-05, "loss": 0.8649, "step": 1406 }, { "epoch": 0.23, "learning_rate": 7.512012813667912e-05, "loss": 0.8258, "step": 1407 }, { "epoch": 0.23, "learning_rate": 7.517351841964763e-05, "loss": 0.9606, "step": 1408 }, { "epoch": 0.23, "learning_rate": 7.522690870261613e-05, "loss": 0.8963, "step": 1409 }, { "epoch": 0.23, "learning_rate": 7.528029898558463e-05, "loss": 0.8441, "step": 1410 }, { "epoch": 0.23, "learning_rate": 7.533368926855313e-05, "loss": 0.9086, "step": 1411 }, { "epoch": 0.23, "learning_rate": 7.538707955152162e-05, "loss": 0.8645, "step": 1412 }, { "epoch": 0.23, "learning_rate": 7.544046983449013e-05, "loss": 0.8664, "step": 1413 }, { "epoch": 0.23, "learning_rate": 7.549386011745862e-05, "loss": 0.8226, "step": 1414 }, { "epoch": 0.23, "learning_rate": 7.554725040042713e-05, "loss": 0.8822, "step": 1415 }, { "epoch": 0.23, "learning_rate": 7.560064068339563e-05, "loss": 0.883, "step": 1416 }, { "epoch": 0.23, "learning_rate": 7.565403096636412e-05, "loss": 0.8974, "step": 1417 }, { "epoch": 0.23, "learning_rate": 7.570742124933262e-05, "loss": 0.8683, "step": 1418 }, { "epoch": 0.23, "learning_rate": 7.576081153230112e-05, "loss": 0.876, "step": 1419 }, { "epoch": 0.23, "learning_rate": 7.581420181526963e-05, "loss": 0.8607, "step": 1420 }, { "epoch": 0.23, "learning_rate": 7.586759209823813e-05, "loss": 0.846, "step": 1421 }, { "epoch": 0.23, "learning_rate": 7.592098238120663e-05, "loss": 0.876, "step": 1422 }, { "epoch": 0.23, "learning_rate": 7.597437266417511e-05, "loss": 0.9551, "step": 1423 }, { "epoch": 0.23, "learning_rate": 7.602776294714362e-05, "loss": 0.9301, "step": 1424 }, { "epoch": 0.23, "learning_rate": 7.608115323011212e-05, "loss": 0.798, "step": 1425 }, { "epoch": 0.23, "learning_rate": 7.613454351308062e-05, "loss": 0.8583, "step": 1426 }, { "epoch": 0.23, "learning_rate": 7.618793379604913e-05, "loss": 0.8839, "step": 1427 }, { "epoch": 0.23, "learning_rate": 7.624132407901761e-05, "loss": 0.9154, "step": 1428 }, { "epoch": 0.23, "learning_rate": 7.629471436198612e-05, "loss": 0.9229, "step": 1429 }, { "epoch": 0.23, "learning_rate": 7.634810464495462e-05, "loss": 0.8636, "step": 1430 }, { "epoch": 0.23, "learning_rate": 7.640149492792312e-05, "loss": 0.8633, "step": 1431 }, { "epoch": 0.23, "learning_rate": 7.645488521089162e-05, "loss": 0.8651, "step": 1432 }, { "epoch": 0.23, "learning_rate": 7.650827549386013e-05, "loss": 0.9009, "step": 1433 }, { "epoch": 0.23, "learning_rate": 7.656166577682862e-05, "loss": 0.859, "step": 1434 }, { "epoch": 0.23, "learning_rate": 7.661505605979711e-05, "loss": 0.8872, "step": 1435 }, { "epoch": 0.23, "learning_rate": 7.666844634276562e-05, "loss": 0.8771, "step": 1436 }, { "epoch": 0.23, "learning_rate": 7.672183662573412e-05, "loss": 0.8685, "step": 1437 }, { "epoch": 0.23, "learning_rate": 7.677522690870263e-05, "loss": 0.9429, "step": 1438 }, { "epoch": 0.23, "learning_rate": 7.682861719167112e-05, "loss": 0.8614, "step": 1439 }, { "epoch": 0.23, "learning_rate": 7.688200747463961e-05, "loss": 0.8566, "step": 1440 }, { "epoch": 0.23, "learning_rate": 7.693539775760812e-05, "loss": 0.9015, "step": 1441 }, { "epoch": 0.23, "learning_rate": 7.698878804057662e-05, "loss": 0.9319, "step": 1442 }, { "epoch": 0.23, "learning_rate": 7.704217832354512e-05, "loss": 0.872, "step": 1443 }, { "epoch": 0.23, "learning_rate": 7.709556860651362e-05, "loss": 0.8443, "step": 1444 }, { "epoch": 0.23, "learning_rate": 7.714895888948211e-05, "loss": 0.8337, "step": 1445 }, { "epoch": 0.23, "learning_rate": 7.720234917245062e-05, "loss": 0.9533, "step": 1446 }, { "epoch": 0.23, "learning_rate": 7.725573945541912e-05, "loss": 0.9238, "step": 1447 }, { "epoch": 0.23, "learning_rate": 7.730912973838762e-05, "loss": 0.9027, "step": 1448 }, { "epoch": 0.23, "learning_rate": 7.736252002135612e-05, "loss": 0.8897, "step": 1449 }, { "epoch": 0.23, "learning_rate": 7.741591030432461e-05, "loss": 0.8807, "step": 1450 }, { "epoch": 0.23, "learning_rate": 7.746930058729312e-05, "loss": 0.8172, "step": 1451 }, { "epoch": 0.23, "learning_rate": 7.752269087026161e-05, "loss": 0.92, "step": 1452 }, { "epoch": 0.23, "learning_rate": 7.757608115323012e-05, "loss": 0.8861, "step": 1453 }, { "epoch": 0.23, "learning_rate": 7.762947143619862e-05, "loss": 0.8345, "step": 1454 }, { "epoch": 0.23, "learning_rate": 7.768286171916711e-05, "loss": 0.931, "step": 1455 }, { "epoch": 0.23, "learning_rate": 7.77362520021356e-05, "loss": 0.8762, "step": 1456 }, { "epoch": 0.23, "learning_rate": 7.778964228510411e-05, "loss": 0.8913, "step": 1457 }, { "epoch": 0.23, "learning_rate": 7.784303256807262e-05, "loss": 0.8859, "step": 1458 }, { "epoch": 0.23, "learning_rate": 7.789642285104112e-05, "loss": 0.8713, "step": 1459 }, { "epoch": 0.23, "learning_rate": 7.794981313400962e-05, "loss": 0.8921, "step": 1460 }, { "epoch": 0.23, "learning_rate": 7.80032034169781e-05, "loss": 0.8688, "step": 1461 }, { "epoch": 0.23, "learning_rate": 7.805659369994661e-05, "loss": 0.852, "step": 1462 }, { "epoch": 0.23, "learning_rate": 7.810998398291511e-05, "loss": 0.8587, "step": 1463 }, { "epoch": 0.23, "learning_rate": 7.816337426588362e-05, "loss": 0.8519, "step": 1464 }, { "epoch": 0.23, "learning_rate": 7.821676454885212e-05, "loss": 0.9636, "step": 1465 }, { "epoch": 0.23, "learning_rate": 7.82701548318206e-05, "loss": 0.863, "step": 1466 }, { "epoch": 0.23, "learning_rate": 7.832354511478911e-05, "loss": 0.8361, "step": 1467 }, { "epoch": 0.24, "learning_rate": 7.83769353977576e-05, "loss": 0.858, "step": 1468 }, { "epoch": 0.24, "learning_rate": 7.843032568072611e-05, "loss": 0.8606, "step": 1469 }, { "epoch": 0.24, "learning_rate": 7.848371596369461e-05, "loss": 0.8981, "step": 1470 }, { "epoch": 0.24, "learning_rate": 7.85371062466631e-05, "loss": 0.9566, "step": 1471 }, { "epoch": 0.24, "learning_rate": 7.859049652963161e-05, "loss": 0.8813, "step": 1472 }, { "epoch": 0.24, "learning_rate": 7.86438868126001e-05, "loss": 0.887, "step": 1473 }, { "epoch": 0.24, "learning_rate": 7.869727709556861e-05, "loss": 0.8207, "step": 1474 }, { "epoch": 0.24, "learning_rate": 7.875066737853711e-05, "loss": 0.792, "step": 1475 }, { "epoch": 0.24, "learning_rate": 7.880405766150562e-05, "loss": 0.9149, "step": 1476 }, { "epoch": 0.24, "learning_rate": 7.885744794447411e-05, "loss": 0.9184, "step": 1477 }, { "epoch": 0.24, "learning_rate": 7.89108382274426e-05, "loss": 0.889, "step": 1478 }, { "epoch": 0.24, "learning_rate": 7.896422851041111e-05, "loss": 0.8999, "step": 1479 }, { "epoch": 0.24, "learning_rate": 7.90176187933796e-05, "loss": 0.9185, "step": 1480 }, { "epoch": 0.24, "learning_rate": 7.907100907634811e-05, "loss": 0.9865, "step": 1481 }, { "epoch": 0.24, "learning_rate": 7.912439935931661e-05, "loss": 0.7923, "step": 1482 }, { "epoch": 0.24, "learning_rate": 7.91777896422851e-05, "loss": 0.8685, "step": 1483 }, { "epoch": 0.24, "learning_rate": 7.923117992525361e-05, "loss": 0.8835, "step": 1484 }, { "epoch": 0.24, "learning_rate": 7.92845702082221e-05, "loss": 0.8413, "step": 1485 }, { "epoch": 0.24, "learning_rate": 7.933796049119061e-05, "loss": 0.8819, "step": 1486 }, { "epoch": 0.24, "learning_rate": 7.939135077415911e-05, "loss": 0.855, "step": 1487 }, { "epoch": 0.24, "learning_rate": 7.94447410571276e-05, "loss": 0.8657, "step": 1488 }, { "epoch": 0.24, "learning_rate": 7.949813134009611e-05, "loss": 0.8322, "step": 1489 }, { "epoch": 0.24, "learning_rate": 7.95515216230646e-05, "loss": 0.9126, "step": 1490 }, { "epoch": 0.24, "learning_rate": 7.960491190603311e-05, "loss": 0.8364, "step": 1491 }, { "epoch": 0.24, "learning_rate": 7.965830218900161e-05, "loss": 0.8929, "step": 1492 }, { "epoch": 0.24, "learning_rate": 7.97116924719701e-05, "loss": 0.9238, "step": 1493 }, { "epoch": 0.24, "learning_rate": 7.97650827549386e-05, "loss": 0.8856, "step": 1494 }, { "epoch": 0.24, "learning_rate": 7.98184730379071e-05, "loss": 0.8671, "step": 1495 }, { "epoch": 0.24, "learning_rate": 7.987186332087561e-05, "loss": 0.8808, "step": 1496 }, { "epoch": 0.24, "learning_rate": 7.99252536038441e-05, "loss": 0.9396, "step": 1497 }, { "epoch": 0.24, "learning_rate": 7.99786438868126e-05, "loss": 0.9872, "step": 1498 }, { "epoch": 0.24, "learning_rate": 8.00320341697811e-05, "loss": 0.9229, "step": 1499 }, { "epoch": 0.24, "learning_rate": 8.00854244527496e-05, "loss": 0.8129, "step": 1500 }, { "epoch": 0.24, "learning_rate": 8.01388147357181e-05, "loss": 0.879, "step": 1501 }, { "epoch": 0.24, "learning_rate": 8.01922050186866e-05, "loss": 0.8118, "step": 1502 }, { "epoch": 0.24, "learning_rate": 8.024559530165511e-05, "loss": 1.0146, "step": 1503 }, { "epoch": 0.24, "learning_rate": 8.02989855846236e-05, "loss": 0.8915, "step": 1504 }, { "epoch": 0.24, "learning_rate": 8.03523758675921e-05, "loss": 0.8515, "step": 1505 }, { "epoch": 0.24, "learning_rate": 8.04057661505606e-05, "loss": 0.9174, "step": 1506 }, { "epoch": 0.24, "learning_rate": 8.04591564335291e-05, "loss": 0.9001, "step": 1507 }, { "epoch": 0.24, "learning_rate": 8.05125467164976e-05, "loss": 0.8241, "step": 1508 }, { "epoch": 0.24, "learning_rate": 8.056593699946609e-05, "loss": 0.9126, "step": 1509 }, { "epoch": 0.24, "learning_rate": 8.06193272824346e-05, "loss": 0.8647, "step": 1510 }, { "epoch": 0.24, "learning_rate": 8.06727175654031e-05, "loss": 0.9017, "step": 1511 }, { "epoch": 0.24, "learning_rate": 8.07261078483716e-05, "loss": 0.8378, "step": 1512 }, { "epoch": 0.24, "learning_rate": 8.07794981313401e-05, "loss": 0.87, "step": 1513 }, { "epoch": 0.24, "learning_rate": 8.08328884143086e-05, "loss": 0.9794, "step": 1514 }, { "epoch": 0.24, "learning_rate": 8.08862786972771e-05, "loss": 0.9226, "step": 1515 }, { "epoch": 0.24, "learning_rate": 8.09396689802456e-05, "loss": 0.8823, "step": 1516 }, { "epoch": 0.24, "learning_rate": 8.09930592632141e-05, "loss": 0.9586, "step": 1517 }, { "epoch": 0.24, "learning_rate": 8.10464495461826e-05, "loss": 0.8726, "step": 1518 }, { "epoch": 0.24, "learning_rate": 8.10998398291511e-05, "loss": 0.8525, "step": 1519 }, { "epoch": 0.24, "learning_rate": 8.11532301121196e-05, "loss": 0.9699, "step": 1520 }, { "epoch": 0.24, "learning_rate": 8.12066203950881e-05, "loss": 0.9412, "step": 1521 }, { "epoch": 0.24, "learning_rate": 8.12600106780566e-05, "loss": 0.8397, "step": 1522 }, { "epoch": 0.24, "learning_rate": 8.13134009610251e-05, "loss": 0.858, "step": 1523 }, { "epoch": 0.24, "learning_rate": 8.13667912439936e-05, "loss": 0.8953, "step": 1524 }, { "epoch": 0.24, "learning_rate": 8.142018152696208e-05, "loss": 0.9806, "step": 1525 }, { "epoch": 0.24, "learning_rate": 8.147357180993059e-05, "loss": 0.9182, "step": 1526 }, { "epoch": 0.24, "learning_rate": 8.15269620928991e-05, "loss": 0.8829, "step": 1527 }, { "epoch": 0.24, "learning_rate": 8.15803523758676e-05, "loss": 0.7862, "step": 1528 }, { "epoch": 0.24, "learning_rate": 8.16337426588361e-05, "loss": 0.8267, "step": 1529 }, { "epoch": 0.25, "learning_rate": 8.16871329418046e-05, "loss": 0.918, "step": 1530 }, { "epoch": 0.25, "learning_rate": 8.174052322477309e-05, "loss": 0.9086, "step": 1531 }, { "epoch": 0.25, "learning_rate": 8.179391350774159e-05, "loss": 0.7801, "step": 1532 }, { "epoch": 0.25, "learning_rate": 8.18473037907101e-05, "loss": 0.87, "step": 1533 }, { "epoch": 0.25, "learning_rate": 8.19006940736786e-05, "loss": 0.9293, "step": 1534 }, { "epoch": 0.25, "learning_rate": 8.19540843566471e-05, "loss": 0.9454, "step": 1535 }, { "epoch": 0.25, "learning_rate": 8.200747463961559e-05, "loss": 0.7332, "step": 1536 }, { "epoch": 0.25, "learning_rate": 8.206086492258409e-05, "loss": 0.9068, "step": 1537 }, { "epoch": 0.25, "learning_rate": 8.21142552055526e-05, "loss": 0.8492, "step": 1538 }, { "epoch": 0.25, "learning_rate": 8.216764548852109e-05, "loss": 0.9176, "step": 1539 }, { "epoch": 0.25, "learning_rate": 8.22210357714896e-05, "loss": 0.8643, "step": 1540 }, { "epoch": 0.25, "learning_rate": 8.22744260544581e-05, "loss": 0.8561, "step": 1541 }, { "epoch": 0.25, "learning_rate": 8.232781633742658e-05, "loss": 0.8304, "step": 1542 }, { "epoch": 0.25, "learning_rate": 8.238120662039509e-05, "loss": 0.9997, "step": 1543 }, { "epoch": 0.25, "learning_rate": 8.243459690336359e-05, "loss": 0.9471, "step": 1544 }, { "epoch": 0.25, "learning_rate": 8.24879871863321e-05, "loss": 0.8643, "step": 1545 }, { "epoch": 0.25, "learning_rate": 8.254137746930059e-05, "loss": 0.9098, "step": 1546 }, { "epoch": 0.25, "learning_rate": 8.259476775226908e-05, "loss": 0.8049, "step": 1547 }, { "epoch": 0.25, "learning_rate": 8.264815803523759e-05, "loss": 0.8741, "step": 1548 }, { "epoch": 0.25, "learning_rate": 8.270154831820609e-05, "loss": 0.9772, "step": 1549 }, { "epoch": 0.25, "learning_rate": 8.27549386011746e-05, "loss": 0.8846, "step": 1550 }, { "epoch": 0.25, "learning_rate": 8.280832888414309e-05, "loss": 0.9021, "step": 1551 }, { "epoch": 0.25, "learning_rate": 8.286171916711158e-05, "loss": 0.8817, "step": 1552 }, { "epoch": 0.25, "learning_rate": 8.291510945008009e-05, "loss": 0.9388, "step": 1553 }, { "epoch": 0.25, "learning_rate": 8.296849973304858e-05, "loss": 0.8097, "step": 1554 }, { "epoch": 0.25, "learning_rate": 8.302189001601709e-05, "loss": 0.9018, "step": 1555 }, { "epoch": 0.25, "learning_rate": 8.307528029898559e-05, "loss": 0.9223, "step": 1556 }, { "epoch": 0.25, "learning_rate": 8.31286705819541e-05, "loss": 0.8278, "step": 1557 }, { "epoch": 0.25, "learning_rate": 8.318206086492259e-05, "loss": 0.8742, "step": 1558 }, { "epoch": 0.25, "learning_rate": 8.323545114789108e-05, "loss": 0.815, "step": 1559 }, { "epoch": 0.25, "learning_rate": 8.328884143085959e-05, "loss": 0.8237, "step": 1560 }, { "epoch": 0.25, "learning_rate": 8.334223171382809e-05, "loss": 0.9192, "step": 1561 }, { "epoch": 0.25, "learning_rate": 8.33956219967966e-05, "loss": 0.8641, "step": 1562 }, { "epoch": 0.25, "learning_rate": 8.344901227976508e-05, "loss": 0.8376, "step": 1563 }, { "epoch": 0.25, "learning_rate": 8.350240256273358e-05, "loss": 0.9238, "step": 1564 }, { "epoch": 0.25, "learning_rate": 8.355579284570209e-05, "loss": 0.9011, "step": 1565 }, { "epoch": 0.25, "learning_rate": 8.360918312867059e-05, "loss": 0.8868, "step": 1566 }, { "epoch": 0.25, "learning_rate": 8.36625734116391e-05, "loss": 0.8449, "step": 1567 }, { "epoch": 0.25, "learning_rate": 8.371596369460759e-05, "loss": 0.9427, "step": 1568 }, { "epoch": 0.25, "learning_rate": 8.376935397757608e-05, "loss": 0.9063, "step": 1569 }, { "epoch": 0.25, "learning_rate": 8.382274426054459e-05, "loss": 0.8104, "step": 1570 }, { "epoch": 0.25, "learning_rate": 8.387613454351308e-05, "loss": 0.7926, "step": 1571 }, { "epoch": 0.25, "learning_rate": 8.392952482648159e-05, "loss": 0.867, "step": 1572 }, { "epoch": 0.25, "learning_rate": 8.398291510945009e-05, "loss": 0.8953, "step": 1573 }, { "epoch": 0.25, "learning_rate": 8.403630539241858e-05, "loss": 0.8561, "step": 1574 }, { "epoch": 0.25, "learning_rate": 8.408969567538708e-05, "loss": 0.8821, "step": 1575 }, { "epoch": 0.25, "learning_rate": 8.414308595835558e-05, "loss": 0.9073, "step": 1576 }, { "epoch": 0.25, "learning_rate": 8.419647624132409e-05, "loss": 0.8996, "step": 1577 }, { "epoch": 0.25, "learning_rate": 8.424986652429259e-05, "loss": 0.8656, "step": 1578 }, { "epoch": 0.25, "learning_rate": 8.430325680726108e-05, "loss": 0.8542, "step": 1579 }, { "epoch": 0.25, "learning_rate": 8.435664709022957e-05, "loss": 0.8889, "step": 1580 }, { "epoch": 0.25, "learning_rate": 8.441003737319808e-05, "loss": 0.8584, "step": 1581 }, { "epoch": 0.25, "learning_rate": 8.446342765616658e-05, "loss": 0.9033, "step": 1582 }, { "epoch": 0.25, "learning_rate": 8.451681793913509e-05, "loss": 0.8085, "step": 1583 }, { "epoch": 0.25, "learning_rate": 8.457020822210359e-05, "loss": 0.906, "step": 1584 }, { "epoch": 0.25, "learning_rate": 8.462359850507207e-05, "loss": 0.8373, "step": 1585 }, { "epoch": 0.25, "learning_rate": 8.467698878804058e-05, "loss": 0.8849, "step": 1586 }, { "epoch": 0.25, "learning_rate": 8.473037907100908e-05, "loss": 0.8299, "step": 1587 }, { "epoch": 0.25, "learning_rate": 8.478376935397758e-05, "loss": 0.8435, "step": 1588 }, { "epoch": 0.25, "learning_rate": 8.483715963694608e-05, "loss": 0.8584, "step": 1589 }, { "epoch": 0.25, "learning_rate": 8.489054991991457e-05, "loss": 0.8958, "step": 1590 }, { "epoch": 0.25, "learning_rate": 8.494394020288308e-05, "loss": 0.8936, "step": 1591 }, { "epoch": 0.25, "learning_rate": 8.499733048585158e-05, "loss": 0.8582, "step": 1592 }, { "epoch": 0.26, "learning_rate": 8.505072076882008e-05, "loss": 0.8857, "step": 1593 }, { "epoch": 0.26, "learning_rate": 8.510411105178858e-05, "loss": 0.8469, "step": 1594 }, { "epoch": 0.26, "learning_rate": 8.515750133475709e-05, "loss": 0.8416, "step": 1595 }, { "epoch": 0.26, "learning_rate": 8.521089161772558e-05, "loss": 0.8353, "step": 1596 }, { "epoch": 0.26, "learning_rate": 8.526428190069407e-05, "loss": 0.886, "step": 1597 }, { "epoch": 0.26, "learning_rate": 8.531767218366258e-05, "loss": 0.8992, "step": 1598 }, { "epoch": 0.26, "learning_rate": 8.537106246663108e-05, "loss": 0.8528, "step": 1599 }, { "epoch": 0.26, "learning_rate": 8.542445274959958e-05, "loss": 0.9512, "step": 1600 }, { "epoch": 0.26, "learning_rate": 8.547784303256808e-05, "loss": 0.9177, "step": 1601 }, { "epoch": 0.26, "learning_rate": 8.553123331553657e-05, "loss": 0.8921, "step": 1602 }, { "epoch": 0.26, "learning_rate": 8.558462359850508e-05, "loss": 0.8703, "step": 1603 }, { "epoch": 0.26, "learning_rate": 8.563801388147358e-05, "loss": 0.8306, "step": 1604 }, { "epoch": 0.26, "learning_rate": 8.569140416444208e-05, "loss": 0.8856, "step": 1605 }, { "epoch": 0.26, "learning_rate": 8.574479444741056e-05, "loss": 0.9, "step": 1606 }, { "epoch": 0.26, "learning_rate": 8.579818473037907e-05, "loss": 0.8589, "step": 1607 }, { "epoch": 0.26, "learning_rate": 8.585157501334758e-05, "loss": 0.8604, "step": 1608 }, { "epoch": 0.26, "learning_rate": 8.590496529631607e-05, "loss": 0.7521, "step": 1609 }, { "epoch": 0.26, "learning_rate": 8.595835557928458e-05, "loss": 0.8758, "step": 1610 }, { "epoch": 0.26, "learning_rate": 8.601174586225308e-05, "loss": 0.813, "step": 1611 }, { "epoch": 0.26, "learning_rate": 8.606513614522157e-05, "loss": 1.02, "step": 1612 }, { "epoch": 0.26, "learning_rate": 8.611852642819007e-05, "loss": 0.8667, "step": 1613 }, { "epoch": 0.26, "learning_rate": 8.617191671115857e-05, "loss": 0.8749, "step": 1614 }, { "epoch": 0.26, "learning_rate": 8.622530699412708e-05, "loss": 0.8824, "step": 1615 }, { "epoch": 0.26, "learning_rate": 8.627869727709558e-05, "loss": 0.8085, "step": 1616 }, { "epoch": 0.26, "learning_rate": 8.633208756006407e-05, "loss": 0.7946, "step": 1617 }, { "epoch": 0.26, "learning_rate": 8.638547784303257e-05, "loss": 0.8117, "step": 1618 }, { "epoch": 0.26, "learning_rate": 8.643886812600107e-05, "loss": 0.8983, "step": 1619 }, { "epoch": 0.26, "learning_rate": 8.649225840896957e-05, "loss": 0.8955, "step": 1620 }, { "epoch": 0.26, "learning_rate": 8.654564869193808e-05, "loss": 0.902, "step": 1621 }, { "epoch": 0.26, "learning_rate": 8.659903897490658e-05, "loss": 0.9235, "step": 1622 }, { "epoch": 0.26, "learning_rate": 8.665242925787506e-05, "loss": 0.9893, "step": 1623 }, { "epoch": 0.26, "learning_rate": 8.670581954084357e-05, "loss": 0.8114, "step": 1624 }, { "epoch": 0.26, "learning_rate": 8.675920982381207e-05, "loss": 0.811, "step": 1625 }, { "epoch": 0.26, "learning_rate": 8.681260010678057e-05, "loss": 0.816, "step": 1626 }, { "epoch": 0.26, "learning_rate": 8.686599038974907e-05, "loss": 0.8719, "step": 1627 }, { "epoch": 0.26, "learning_rate": 8.691938067271756e-05, "loss": 0.92, "step": 1628 }, { "epoch": 0.26, "learning_rate": 8.697277095568607e-05, "loss": 0.8838, "step": 1629 }, { "epoch": 0.26, "learning_rate": 8.702616123865457e-05, "loss": 0.9615, "step": 1630 }, { "epoch": 0.26, "learning_rate": 8.707955152162307e-05, "loss": 0.8874, "step": 1631 }, { "epoch": 0.26, "learning_rate": 8.713294180459157e-05, "loss": 0.9521, "step": 1632 }, { "epoch": 0.26, "learning_rate": 8.718633208756006e-05, "loss": 0.8621, "step": 1633 }, { "epoch": 0.26, "learning_rate": 8.723972237052857e-05, "loss": 0.867, "step": 1634 }, { "epoch": 0.26, "learning_rate": 8.729311265349706e-05, "loss": 0.8571, "step": 1635 }, { "epoch": 0.26, "learning_rate": 8.734650293646557e-05, "loss": 0.8448, "step": 1636 }, { "epoch": 0.26, "learning_rate": 8.739989321943407e-05, "loss": 0.8351, "step": 1637 }, { "epoch": 0.26, "learning_rate": 8.745328350240257e-05, "loss": 0.8202, "step": 1638 }, { "epoch": 0.26, "learning_rate": 8.750667378537107e-05, "loss": 0.8572, "step": 1639 }, { "epoch": 0.26, "learning_rate": 8.756006406833956e-05, "loss": 0.8716, "step": 1640 }, { "epoch": 0.26, "learning_rate": 8.761345435130807e-05, "loss": 0.7947, "step": 1641 }, { "epoch": 0.26, "learning_rate": 8.766684463427657e-05, "loss": 0.7408, "step": 1642 }, { "epoch": 0.26, "learning_rate": 8.772023491724507e-05, "loss": 0.8593, "step": 1643 }, { "epoch": 0.26, "learning_rate": 8.777362520021355e-05, "loss": 0.9091, "step": 1644 }, { "epoch": 0.26, "learning_rate": 8.782701548318206e-05, "loss": 0.9048, "step": 1645 }, { "epoch": 0.26, "learning_rate": 8.788040576615057e-05, "loss": 0.8734, "step": 1646 }, { "epoch": 0.26, "learning_rate": 8.793379604911907e-05, "loss": 0.8238, "step": 1647 }, { "epoch": 0.26, "learning_rate": 8.798718633208757e-05, "loss": 0.833, "step": 1648 }, { "epoch": 0.26, "learning_rate": 8.804057661505605e-05, "loss": 0.829, "step": 1649 }, { "epoch": 0.26, "learning_rate": 8.809396689802456e-05, "loss": 0.8298, "step": 1650 }, { "epoch": 0.26, "learning_rate": 8.814735718099306e-05, "loss": 0.9232, "step": 1651 }, { "epoch": 0.26, "learning_rate": 8.820074746396156e-05, "loss": 0.8432, "step": 1652 }, { "epoch": 0.26, "learning_rate": 8.825413774693007e-05, "loss": 0.8821, "step": 1653 }, { "epoch": 0.26, "learning_rate": 8.830752802989857e-05, "loss": 0.8359, "step": 1654 }, { "epoch": 0.27, "learning_rate": 8.836091831286706e-05, "loss": 0.8559, "step": 1655 }, { "epoch": 0.27, "learning_rate": 8.841430859583556e-05, "loss": 0.9262, "step": 1656 }, { "epoch": 0.27, "learning_rate": 8.846769887880406e-05, "loss": 0.8796, "step": 1657 }, { "epoch": 0.27, "learning_rate": 8.852108916177256e-05, "loss": 0.8281, "step": 1658 }, { "epoch": 0.27, "learning_rate": 8.857447944474107e-05, "loss": 0.8146, "step": 1659 }, { "epoch": 0.27, "learning_rate": 8.862786972770956e-05, "loss": 0.9069, "step": 1660 }, { "epoch": 0.27, "learning_rate": 8.868126001067805e-05, "loss": 0.9164, "step": 1661 }, { "epoch": 0.27, "learning_rate": 8.873465029364656e-05, "loss": 0.931, "step": 1662 }, { "epoch": 0.27, "learning_rate": 8.878804057661506e-05, "loss": 0.8474, "step": 1663 }, { "epoch": 0.27, "learning_rate": 8.884143085958356e-05, "loss": 0.8259, "step": 1664 }, { "epoch": 0.27, "learning_rate": 8.889482114255206e-05, "loss": 0.9156, "step": 1665 }, { "epoch": 0.27, "learning_rate": 8.894821142552055e-05, "loss": 0.8877, "step": 1666 }, { "epoch": 0.27, "learning_rate": 8.900160170848906e-05, "loss": 0.852, "step": 1667 }, { "epoch": 0.27, "learning_rate": 8.905499199145756e-05, "loss": 0.8777, "step": 1668 }, { "epoch": 0.27, "learning_rate": 8.910838227442606e-05, "loss": 0.8403, "step": 1669 }, { "epoch": 0.27, "learning_rate": 8.916177255739456e-05, "loss": 0.938, "step": 1670 }, { "epoch": 0.27, "learning_rate": 8.921516284036305e-05, "loss": 0.8874, "step": 1671 }, { "epoch": 0.27, "learning_rate": 8.926855312333156e-05, "loss": 0.9276, "step": 1672 }, { "epoch": 0.27, "learning_rate": 8.932194340630006e-05, "loss": 0.9614, "step": 1673 }, { "epoch": 0.27, "learning_rate": 8.937533368926856e-05, "loss": 0.9415, "step": 1674 }, { "epoch": 0.27, "learning_rate": 8.942872397223706e-05, "loss": 0.8473, "step": 1675 }, { "epoch": 0.27, "learning_rate": 8.948211425520555e-05, "loss": 0.8712, "step": 1676 }, { "epoch": 0.27, "learning_rate": 8.953550453817406e-05, "loss": 0.8949, "step": 1677 }, { "epoch": 0.27, "learning_rate": 8.958889482114255e-05, "loss": 0.8814, "step": 1678 }, { "epoch": 0.27, "learning_rate": 8.964228510411106e-05, "loss": 0.8557, "step": 1679 }, { "epoch": 0.27, "learning_rate": 8.969567538707956e-05, "loss": 0.8863, "step": 1680 }, { "epoch": 0.27, "learning_rate": 8.974906567004806e-05, "loss": 0.8726, "step": 1681 }, { "epoch": 0.27, "learning_rate": 8.980245595301655e-05, "loss": 0.8666, "step": 1682 }, { "epoch": 0.27, "learning_rate": 8.985584623598505e-05, "loss": 0.7747, "step": 1683 }, { "epoch": 0.27, "learning_rate": 8.990923651895356e-05, "loss": 0.8767, "step": 1684 }, { "epoch": 0.27, "learning_rate": 8.996262680192206e-05, "loss": 0.945, "step": 1685 }, { "epoch": 0.27, "learning_rate": 9.001601708489056e-05, "loss": 0.7753, "step": 1686 }, { "epoch": 0.27, "learning_rate": 9.006940736785904e-05, "loss": 0.8042, "step": 1687 }, { "epoch": 0.27, "learning_rate": 9.012279765082755e-05, "loss": 0.8651, "step": 1688 }, { "epoch": 0.27, "learning_rate": 9.017618793379605e-05, "loss": 0.821, "step": 1689 }, { "epoch": 0.27, "learning_rate": 9.022957821676455e-05, "loss": 0.9036, "step": 1690 }, { "epoch": 0.27, "learning_rate": 9.028296849973306e-05, "loss": 0.8241, "step": 1691 }, { "epoch": 0.27, "learning_rate": 9.033635878270156e-05, "loss": 0.9347, "step": 1692 }, { "epoch": 0.27, "learning_rate": 9.038974906567005e-05, "loss": 0.8677, "step": 1693 }, { "epoch": 0.27, "learning_rate": 9.044313934863855e-05, "loss": 0.9031, "step": 1694 }, { "epoch": 0.27, "learning_rate": 9.049652963160705e-05, "loss": 0.8279, "step": 1695 }, { "epoch": 0.27, "learning_rate": 9.054991991457555e-05, "loss": 0.8972, "step": 1696 }, { "epoch": 0.27, "learning_rate": 9.060331019754406e-05, "loss": 0.8764, "step": 1697 }, { "epoch": 0.27, "learning_rate": 9.065670048051255e-05, "loss": 0.9527, "step": 1698 }, { "epoch": 0.27, "learning_rate": 9.071009076348104e-05, "loss": 0.8448, "step": 1699 }, { "epoch": 0.27, "learning_rate": 9.076348104644955e-05, "loss": 0.9453, "step": 1700 }, { "epoch": 0.27, "learning_rate": 9.081687132941805e-05, "loss": 0.8678, "step": 1701 }, { "epoch": 0.27, "learning_rate": 9.087026161238656e-05, "loss": 0.8891, "step": 1702 }, { "epoch": 0.27, "learning_rate": 9.092365189535505e-05, "loss": 0.8274, "step": 1703 }, { "epoch": 0.27, "learning_rate": 9.097704217832354e-05, "loss": 0.8984, "step": 1704 }, { "epoch": 0.27, "learning_rate": 9.103043246129205e-05, "loss": 0.9585, "step": 1705 }, { "epoch": 0.27, "learning_rate": 9.108382274426055e-05, "loss": 0.9144, "step": 1706 }, { "epoch": 0.27, "learning_rate": 9.113721302722905e-05, "loss": 0.7861, "step": 1707 }, { "epoch": 0.27, "learning_rate": 9.119060331019755e-05, "loss": 0.8767, "step": 1708 }, { "epoch": 0.27, "learning_rate": 9.124399359316604e-05, "loss": 0.8475, "step": 1709 }, { "epoch": 0.27, "learning_rate": 9.129738387613455e-05, "loss": 0.8196, "step": 1710 }, { "epoch": 0.27, "learning_rate": 9.135077415910305e-05, "loss": 0.8895, "step": 1711 }, { "epoch": 0.27, "learning_rate": 9.140416444207155e-05, "loss": 0.8951, "step": 1712 }, { "epoch": 0.27, "learning_rate": 9.145755472504005e-05, "loss": 0.8747, "step": 1713 }, { "epoch": 0.27, "learning_rate": 9.151094500800854e-05, "loss": 0.8989, "step": 1714 }, { "epoch": 0.27, "learning_rate": 9.156433529097705e-05, "loss": 0.8894, "step": 1715 }, { "epoch": 0.27, "learning_rate": 9.161772557394554e-05, "loss": 0.8474, "step": 1716 }, { "epoch": 0.27, "learning_rate": 9.167111585691405e-05, "loss": 0.8299, "step": 1717 }, { "epoch": 0.28, "learning_rate": 9.172450613988255e-05, "loss": 0.8777, "step": 1718 }, { "epoch": 0.28, "learning_rate": 9.177789642285105e-05, "loss": 0.8133, "step": 1719 }, { "epoch": 0.28, "learning_rate": 9.183128670581954e-05, "loss": 0.8907, "step": 1720 }, { "epoch": 0.28, "learning_rate": 9.188467698878804e-05, "loss": 0.9283, "step": 1721 }, { "epoch": 0.28, "learning_rate": 9.193806727175655e-05, "loss": 0.8022, "step": 1722 }, { "epoch": 0.28, "learning_rate": 9.199145755472505e-05, "loss": 0.9325, "step": 1723 }, { "epoch": 0.28, "learning_rate": 9.204484783769355e-05, "loss": 0.8754, "step": 1724 }, { "epoch": 0.28, "learning_rate": 9.209823812066203e-05, "loss": 0.8225, "step": 1725 }, { "epoch": 0.28, "learning_rate": 9.215162840363054e-05, "loss": 0.8673, "step": 1726 }, { "epoch": 0.28, "learning_rate": 9.220501868659904e-05, "loss": 0.8744, "step": 1727 }, { "epoch": 0.28, "learning_rate": 9.225840896956754e-05, "loss": 0.8217, "step": 1728 }, { "epoch": 0.28, "learning_rate": 9.231179925253605e-05, "loss": 0.9149, "step": 1729 }, { "epoch": 0.28, "learning_rate": 9.236518953550453e-05, "loss": 0.8941, "step": 1730 }, { "epoch": 0.28, "learning_rate": 9.241857981847304e-05, "loss": 0.8521, "step": 1731 }, { "epoch": 0.28, "learning_rate": 9.247197010144154e-05, "loss": 0.8409, "step": 1732 }, { "epoch": 0.28, "learning_rate": 9.252536038441004e-05, "loss": 0.8758, "step": 1733 }, { "epoch": 0.28, "learning_rate": 9.257875066737854e-05, "loss": 0.8089, "step": 1734 }, { "epoch": 0.28, "learning_rate": 9.263214095034705e-05, "loss": 0.8785, "step": 1735 }, { "epoch": 0.28, "learning_rate": 9.268553123331554e-05, "loss": 0.8857, "step": 1736 }, { "epoch": 0.28, "learning_rate": 9.273892151628404e-05, "loss": 0.8508, "step": 1737 }, { "epoch": 0.28, "learning_rate": 9.279231179925254e-05, "loss": 0.7851, "step": 1738 }, { "epoch": 0.28, "learning_rate": 9.284570208222104e-05, "loss": 0.9077, "step": 1739 }, { "epoch": 0.28, "learning_rate": 9.289909236518955e-05, "loss": 0.9187, "step": 1740 }, { "epoch": 0.28, "learning_rate": 9.295248264815804e-05, "loss": 0.7736, "step": 1741 }, { "epoch": 0.28, "learning_rate": 9.300587293112653e-05, "loss": 0.8706, "step": 1742 }, { "epoch": 0.28, "learning_rate": 9.305926321409504e-05, "loss": 0.8546, "step": 1743 }, { "epoch": 0.28, "learning_rate": 9.311265349706354e-05, "loss": 0.8786, "step": 1744 }, { "epoch": 0.28, "learning_rate": 9.316604378003204e-05, "loss": 0.8378, "step": 1745 }, { "epoch": 0.28, "learning_rate": 9.321943406300054e-05, "loss": 0.8588, "step": 1746 }, { "epoch": 0.28, "learning_rate": 9.327282434596903e-05, "loss": 0.7748, "step": 1747 }, { "epoch": 0.28, "learning_rate": 9.332621462893754e-05, "loss": 0.7989, "step": 1748 }, { "epoch": 0.28, "learning_rate": 9.337960491190604e-05, "loss": 0.8719, "step": 1749 }, { "epoch": 0.28, "learning_rate": 9.343299519487454e-05, "loss": 0.8186, "step": 1750 }, { "epoch": 0.28, "learning_rate": 9.348638547784304e-05, "loss": 0.8819, "step": 1751 }, { "epoch": 0.28, "learning_rate": 9.353977576081153e-05, "loss": 0.8477, "step": 1752 }, { "epoch": 0.28, "learning_rate": 9.359316604378004e-05, "loss": 0.8841, "step": 1753 }, { "epoch": 0.28, "learning_rate": 9.364655632674853e-05, "loss": 0.9288, "step": 1754 }, { "epoch": 0.28, "learning_rate": 9.369994660971704e-05, "loss": 0.9265, "step": 1755 }, { "epoch": 0.28, "learning_rate": 9.375333689268554e-05, "loss": 0.8884, "step": 1756 }, { "epoch": 0.28, "learning_rate": 9.380672717565403e-05, "loss": 0.843, "step": 1757 }, { "epoch": 0.28, "learning_rate": 9.386011745862253e-05, "loss": 0.8339, "step": 1758 }, { "epoch": 0.28, "learning_rate": 9.391350774159103e-05, "loss": 0.8459, "step": 1759 }, { "epoch": 0.28, "learning_rate": 9.396689802455954e-05, "loss": 0.9223, "step": 1760 }, { "epoch": 0.28, "learning_rate": 9.402028830752804e-05, "loss": 0.8527, "step": 1761 }, { "epoch": 0.28, "learning_rate": 9.407367859049654e-05, "loss": 0.7889, "step": 1762 }, { "epoch": 0.28, "learning_rate": 9.412706887346502e-05, "loss": 0.8862, "step": 1763 }, { "epoch": 0.28, "learning_rate": 9.418045915643353e-05, "loss": 0.8537, "step": 1764 }, { "epoch": 0.28, "learning_rate": 9.423384943940203e-05, "loss": 0.7744, "step": 1765 }, { "epoch": 0.28, "learning_rate": 9.428723972237054e-05, "loss": 0.8276, "step": 1766 }, { "epoch": 0.28, "learning_rate": 9.434063000533904e-05, "loss": 0.9221, "step": 1767 }, { "epoch": 0.28, "learning_rate": 9.439402028830752e-05, "loss": 0.8187, "step": 1768 }, { "epoch": 0.28, "learning_rate": 9.444741057127603e-05, "loss": 0.8761, "step": 1769 }, { "epoch": 0.28, "learning_rate": 9.450080085424453e-05, "loss": 0.764, "step": 1770 }, { "epoch": 0.28, "learning_rate": 9.455419113721303e-05, "loss": 0.8358, "step": 1771 }, { "epoch": 0.28, "learning_rate": 9.460758142018153e-05, "loss": 0.8902, "step": 1772 }, { "epoch": 0.28, "learning_rate": 9.466097170315004e-05, "loss": 0.882, "step": 1773 }, { "epoch": 0.28, "learning_rate": 9.471436198611853e-05, "loss": 0.8434, "step": 1774 }, { "epoch": 0.28, "learning_rate": 9.476775226908703e-05, "loss": 0.8986, "step": 1775 }, { "epoch": 0.28, "learning_rate": 9.482114255205553e-05, "loss": 0.8056, "step": 1776 }, { "epoch": 0.28, "learning_rate": 9.487453283502403e-05, "loss": 0.8894, "step": 1777 }, { "epoch": 0.28, "learning_rate": 9.492792311799254e-05, "loss": 0.8388, "step": 1778 }, { "epoch": 0.28, "learning_rate": 9.498131340096103e-05, "loss": 0.9607, "step": 1779 }, { "epoch": 0.29, "learning_rate": 9.503470368392952e-05, "loss": 0.816, "step": 1780 }, { "epoch": 0.29, "learning_rate": 9.508809396689803e-05, "loss": 0.8719, "step": 1781 }, { "epoch": 0.29, "learning_rate": 9.514148424986653e-05, "loss": 0.8561, "step": 1782 }, { "epoch": 0.29, "learning_rate": 9.519487453283503e-05, "loss": 0.8271, "step": 1783 }, { "epoch": 0.29, "learning_rate": 9.524826481580353e-05, "loss": 0.9256, "step": 1784 }, { "epoch": 0.29, "learning_rate": 9.530165509877202e-05, "loss": 0.8364, "step": 1785 }, { "epoch": 0.29, "learning_rate": 9.535504538174053e-05, "loss": 0.83, "step": 1786 }, { "epoch": 0.29, "learning_rate": 9.540843566470903e-05, "loss": 0.9353, "step": 1787 }, { "epoch": 0.29, "learning_rate": 9.546182594767753e-05, "loss": 0.9182, "step": 1788 }, { "epoch": 0.29, "learning_rate": 9.551521623064603e-05, "loss": 0.8126, "step": 1789 }, { "epoch": 0.29, "learning_rate": 9.556860651361452e-05, "loss": 0.8375, "step": 1790 }, { "epoch": 0.29, "learning_rate": 9.562199679658303e-05, "loss": 0.7981, "step": 1791 }, { "epoch": 0.29, "learning_rate": 9.567538707955153e-05, "loss": 0.849, "step": 1792 }, { "epoch": 0.29, "learning_rate": 9.572877736252003e-05, "loss": 0.8086, "step": 1793 }, { "epoch": 0.29, "learning_rate": 9.578216764548853e-05, "loss": 0.8594, "step": 1794 }, { "epoch": 0.29, "learning_rate": 9.583555792845702e-05, "loss": 0.8089, "step": 1795 }, { "epoch": 0.29, "learning_rate": 9.588894821142552e-05, "loss": 0.8355, "step": 1796 }, { "epoch": 0.29, "learning_rate": 9.594233849439402e-05, "loss": 0.8839, "step": 1797 }, { "epoch": 0.29, "learning_rate": 9.599572877736253e-05, "loss": 0.8344, "step": 1798 }, { "epoch": 0.29, "learning_rate": 9.604911906033103e-05, "loss": 0.8602, "step": 1799 }, { "epoch": 0.29, "learning_rate": 9.610250934329953e-05, "loss": 0.9165, "step": 1800 }, { "epoch": 0.29, "learning_rate": 9.615589962626802e-05, "loss": 0.8623, "step": 1801 }, { "epoch": 0.29, "learning_rate": 9.620928990923652e-05, "loss": 0.8979, "step": 1802 }, { "epoch": 0.29, "learning_rate": 9.626268019220502e-05, "loss": 0.8524, "step": 1803 }, { "epoch": 0.29, "learning_rate": 9.631607047517353e-05, "loss": 0.9195, "step": 1804 }, { "epoch": 0.29, "learning_rate": 9.636946075814203e-05, "loss": 0.9594, "step": 1805 }, { "epoch": 0.29, "learning_rate": 9.642285104111051e-05, "loss": 0.8043, "step": 1806 }, { "epoch": 0.29, "learning_rate": 9.647624132407902e-05, "loss": 0.8075, "step": 1807 }, { "epoch": 0.29, "learning_rate": 9.652963160704752e-05, "loss": 0.9001, "step": 1808 }, { "epoch": 0.29, "learning_rate": 9.658302189001602e-05, "loss": 0.9229, "step": 1809 }, { "epoch": 0.29, "learning_rate": 9.663641217298452e-05, "loss": 0.8038, "step": 1810 }, { "epoch": 0.29, "learning_rate": 9.668980245595301e-05, "loss": 0.8623, "step": 1811 }, { "epoch": 0.29, "learning_rate": 9.674319273892152e-05, "loss": 0.8245, "step": 1812 }, { "epoch": 0.29, "learning_rate": 9.679658302189002e-05, "loss": 0.8539, "step": 1813 }, { "epoch": 0.29, "learning_rate": 9.684997330485852e-05, "loss": 0.9009, "step": 1814 }, { "epoch": 0.29, "learning_rate": 9.690336358782702e-05, "loss": 0.8302, "step": 1815 }, { "epoch": 0.29, "learning_rate": 9.695675387079553e-05, "loss": 0.8255, "step": 1816 }, { "epoch": 0.29, "learning_rate": 9.701014415376402e-05, "loss": 0.9014, "step": 1817 }, { "epoch": 0.29, "learning_rate": 9.706353443673251e-05, "loss": 0.859, "step": 1818 }, { "epoch": 0.29, "learning_rate": 9.711692471970102e-05, "loss": 0.7705, "step": 1819 }, { "epoch": 0.29, "learning_rate": 9.717031500266952e-05, "loss": 0.8394, "step": 1820 }, { "epoch": 0.29, "learning_rate": 9.722370528563803e-05, "loss": 0.8815, "step": 1821 }, { "epoch": 0.29, "learning_rate": 9.727709556860652e-05, "loss": 0.9386, "step": 1822 }, { "epoch": 0.29, "learning_rate": 9.733048585157501e-05, "loss": 0.831, "step": 1823 }, { "epoch": 0.29, "learning_rate": 9.738387613454352e-05, "loss": 0.8457, "step": 1824 }, { "epoch": 0.29, "learning_rate": 9.743726641751202e-05, "loss": 0.9313, "step": 1825 }, { "epoch": 0.29, "learning_rate": 9.749065670048052e-05, "loss": 0.9132, "step": 1826 }, { "epoch": 0.29, "learning_rate": 9.754404698344902e-05, "loss": 0.8248, "step": 1827 }, { "epoch": 0.29, "learning_rate": 9.759743726641751e-05, "loss": 0.879, "step": 1828 }, { "epoch": 0.29, "learning_rate": 9.765082754938602e-05, "loss": 0.9405, "step": 1829 }, { "epoch": 0.29, "learning_rate": 9.770421783235452e-05, "loss": 0.8665, "step": 1830 }, { "epoch": 0.29, "learning_rate": 9.775760811532302e-05, "loss": 0.8601, "step": 1831 }, { "epoch": 0.29, "learning_rate": 9.781099839829152e-05, "loss": 0.8415, "step": 1832 }, { "epoch": 0.29, "learning_rate": 9.786438868126001e-05, "loss": 0.9015, "step": 1833 }, { "epoch": 0.29, "learning_rate": 9.79177789642285e-05, "loss": 0.8458, "step": 1834 }, { "epoch": 0.29, "learning_rate": 9.797116924719701e-05, "loss": 0.9128, "step": 1835 }, { "epoch": 0.29, "learning_rate": 9.802455953016552e-05, "loss": 0.8081, "step": 1836 }, { "epoch": 0.29, "learning_rate": 9.807794981313402e-05, "loss": 0.7926, "step": 1837 }, { "epoch": 0.29, "learning_rate": 9.813134009610251e-05, "loss": 0.7914, "step": 1838 }, { "epoch": 0.29, "learning_rate": 9.8184730379071e-05, "loss": 0.9181, "step": 1839 }, { "epoch": 0.29, "learning_rate": 9.823812066203951e-05, "loss": 0.7981, "step": 1840 }, { "epoch": 0.29, "learning_rate": 9.829151094500801e-05, "loss": 0.8679, "step": 1841 }, { "epoch": 0.3, "learning_rate": 9.834490122797652e-05, "loss": 0.8234, "step": 1842 }, { "epoch": 0.3, "learning_rate": 9.839829151094502e-05, "loss": 0.8831, "step": 1843 }, { "epoch": 0.3, "learning_rate": 9.84516817939135e-05, "loss": 0.8455, "step": 1844 }, { "epoch": 0.3, "learning_rate": 9.850507207688201e-05, "loss": 0.815, "step": 1845 }, { "epoch": 0.3, "learning_rate": 9.855846235985051e-05, "loss": 0.9047, "step": 1846 }, { "epoch": 0.3, "learning_rate": 9.861185264281901e-05, "loss": 0.8146, "step": 1847 }, { "epoch": 0.3, "learning_rate": 9.866524292578751e-05, "loss": 0.9126, "step": 1848 }, { "epoch": 0.3, "learning_rate": 9.8718633208756e-05, "loss": 0.8721, "step": 1849 }, { "epoch": 0.3, "learning_rate": 9.877202349172451e-05, "loss": 0.9536, "step": 1850 }, { "epoch": 0.3, "learning_rate": 9.8825413774693e-05, "loss": 0.8883, "step": 1851 }, { "epoch": 0.3, "learning_rate": 9.887880405766151e-05, "loss": 0.8666, "step": 1852 }, { "epoch": 0.3, "learning_rate": 9.893219434063001e-05, "loss": 0.9119, "step": 1853 }, { "epoch": 0.3, "learning_rate": 9.898558462359852e-05, "loss": 0.8966, "step": 1854 }, { "epoch": 0.3, "learning_rate": 9.903897490656701e-05, "loss": 0.7708, "step": 1855 }, { "epoch": 0.3, "learning_rate": 9.90923651895355e-05, "loss": 0.827, "step": 1856 }, { "epoch": 0.3, "learning_rate": 9.914575547250401e-05, "loss": 0.8305, "step": 1857 }, { "epoch": 0.3, "learning_rate": 9.919914575547251e-05, "loss": 0.9048, "step": 1858 }, { "epoch": 0.3, "learning_rate": 9.925253603844102e-05, "loss": 0.916, "step": 1859 }, { "epoch": 0.3, "learning_rate": 9.930592632140951e-05, "loss": 0.8864, "step": 1860 }, { "epoch": 0.3, "learning_rate": 9.9359316604378e-05, "loss": 0.9653, "step": 1861 }, { "epoch": 0.3, "learning_rate": 9.941270688734651e-05, "loss": 0.8652, "step": 1862 }, { "epoch": 0.3, "learning_rate": 9.9466097170315e-05, "loss": 0.8449, "step": 1863 }, { "epoch": 0.3, "learning_rate": 9.951948745328351e-05, "loss": 0.8807, "step": 1864 }, { "epoch": 0.3, "learning_rate": 9.9572877736252e-05, "loss": 0.8891, "step": 1865 }, { "epoch": 0.3, "learning_rate": 9.96262680192205e-05, "loss": 0.8282, "step": 1866 }, { "epoch": 0.3, "learning_rate": 9.967965830218901e-05, "loss": 0.8783, "step": 1867 }, { "epoch": 0.3, "learning_rate": 9.97330485851575e-05, "loss": 0.8722, "step": 1868 }, { "epoch": 0.3, "learning_rate": 9.978643886812601e-05, "loss": 0.8179, "step": 1869 }, { "epoch": 0.3, "learning_rate": 9.983982915109451e-05, "loss": 0.8847, "step": 1870 }, { "epoch": 0.3, "learning_rate": 9.9893219434063e-05, "loss": 0.825, "step": 1871 }, { "epoch": 0.3, "learning_rate": 9.99466097170315e-05, "loss": 0.8338, "step": 1872 }, { "epoch": 0.3, "learning_rate": 0.0001, "loss": 0.7527, "step": 1873 }, { "epoch": 0.3, "learning_rate": 9.999999913157823e-05, "loss": 0.9221, "step": 1874 }, { "epoch": 0.3, "learning_rate": 9.999999652631291e-05, "loss": 0.8337, "step": 1875 }, { "epoch": 0.3, "learning_rate": 9.999999218420416e-05, "loss": 0.9184, "step": 1876 }, { "epoch": 0.3, "learning_rate": 9.99999861052521e-05, "loss": 0.8249, "step": 1877 }, { "epoch": 0.3, "learning_rate": 9.999997828945699e-05, "loss": 0.8068, "step": 1878 }, { "epoch": 0.3, "learning_rate": 9.999996873681905e-05, "loss": 0.9081, "step": 1879 }, { "epoch": 0.3, "learning_rate": 9.999995744733863e-05, "loss": 0.866, "step": 1880 }, { "epoch": 0.3, "learning_rate": 9.999994442101614e-05, "loss": 0.879, "step": 1881 }, { "epoch": 0.3, "learning_rate": 9.999992965785202e-05, "loss": 0.8812, "step": 1882 }, { "epoch": 0.3, "learning_rate": 9.999991315784679e-05, "loss": 0.7397, "step": 1883 }, { "epoch": 0.3, "learning_rate": 9.9999894921001e-05, "loss": 0.8775, "step": 1884 }, { "epoch": 0.3, "learning_rate": 9.999987494731529e-05, "loss": 0.8612, "step": 1885 }, { "epoch": 0.3, "learning_rate": 9.999985323679035e-05, "loss": 0.833, "step": 1886 }, { "epoch": 0.3, "learning_rate": 9.999982978942697e-05, "loss": 0.8431, "step": 1887 }, { "epoch": 0.3, "learning_rate": 9.999980460522593e-05, "loss": 0.8694, "step": 1888 }, { "epoch": 0.3, "learning_rate": 9.999977768418814e-05, "loss": 0.884, "step": 1889 }, { "epoch": 0.3, "learning_rate": 9.999974902631449e-05, "loss": 0.8805, "step": 1890 }, { "epoch": 0.3, "learning_rate": 9.999971863160598e-05, "loss": 0.834, "step": 1891 }, { "epoch": 0.3, "learning_rate": 9.999968650006371e-05, "loss": 0.96, "step": 1892 }, { "epoch": 0.3, "learning_rate": 9.999965263168874e-05, "loss": 0.8078, "step": 1893 }, { "epoch": 0.3, "learning_rate": 9.999961702648229e-05, "loss": 0.8735, "step": 1894 }, { "epoch": 0.3, "learning_rate": 9.999957968444558e-05, "loss": 0.8314, "step": 1895 }, { "epoch": 0.3, "learning_rate": 9.999954060557992e-05, "loss": 0.8225, "step": 1896 }, { "epoch": 0.3, "learning_rate": 9.999949978988663e-05, "loss": 0.8236, "step": 1897 }, { "epoch": 0.3, "learning_rate": 9.999945723736717e-05, "loss": 0.9477, "step": 1898 }, { "epoch": 0.3, "learning_rate": 9.9999412948023e-05, "loss": 0.8263, "step": 1899 }, { "epoch": 0.3, "learning_rate": 9.999936692185565e-05, "loss": 0.8138, "step": 1900 }, { "epoch": 0.3, "learning_rate": 9.999931915886675e-05, "loss": 0.8651, "step": 1901 }, { "epoch": 0.3, "learning_rate": 9.999926965905793e-05, "loss": 0.9148, "step": 1902 }, { "epoch": 0.3, "learning_rate": 9.99992184224309e-05, "loss": 0.9608, "step": 1903 }, { "epoch": 0.3, "learning_rate": 9.999916544898748e-05, "loss": 0.9818, "step": 1904 }, { "epoch": 0.31, "learning_rate": 9.999911073872947e-05, "loss": 0.8924, "step": 1905 }, { "epoch": 0.31, "learning_rate": 9.999905429165879e-05, "loss": 0.8454, "step": 1906 }, { "epoch": 0.31, "learning_rate": 9.999899610777741e-05, "loss": 0.8197, "step": 1907 }, { "epoch": 0.31, "learning_rate": 9.999893618708734e-05, "loss": 0.7915, "step": 1908 }, { "epoch": 0.31, "learning_rate": 9.999887452959064e-05, "loss": 0.8589, "step": 1909 }, { "epoch": 0.31, "learning_rate": 9.99988111352895e-05, "loss": 0.8616, "step": 1910 }, { "epoch": 0.31, "learning_rate": 9.999874600418608e-05, "loss": 0.8632, "step": 1911 }, { "epoch": 0.31, "learning_rate": 9.999867913628265e-05, "loss": 0.8949, "step": 1912 }, { "epoch": 0.31, "learning_rate": 9.999861053158156e-05, "loss": 0.8407, "step": 1913 }, { "epoch": 0.31, "learning_rate": 9.999854019008517e-05, "loss": 0.9048, "step": 1914 }, { "epoch": 0.31, "learning_rate": 9.999846811179592e-05, "loss": 0.8511, "step": 1915 }, { "epoch": 0.31, "learning_rate": 9.999839429671632e-05, "loss": 0.8066, "step": 1916 }, { "epoch": 0.31, "learning_rate": 9.999831874484893e-05, "loss": 0.8604, "step": 1917 }, { "epoch": 0.31, "learning_rate": 9.999824145619639e-05, "loss": 0.8816, "step": 1918 }, { "epoch": 0.31, "learning_rate": 9.999816243076136e-05, "loss": 0.8038, "step": 1919 }, { "epoch": 0.31, "learning_rate": 9.99980816685466e-05, "loss": 0.8876, "step": 1920 }, { "epoch": 0.31, "learning_rate": 9.999799916955494e-05, "loss": 0.8527, "step": 1921 }, { "epoch": 0.31, "learning_rate": 9.999791493378921e-05, "loss": 0.9824, "step": 1922 }, { "epoch": 0.31, "learning_rate": 9.999782896125233e-05, "loss": 0.888, "step": 1923 }, { "epoch": 0.31, "learning_rate": 9.999774125194731e-05, "loss": 0.9098, "step": 1924 }, { "epoch": 0.31, "learning_rate": 9.999765180587719e-05, "loss": 0.7849, "step": 1925 }, { "epoch": 0.31, "learning_rate": 9.999756062304508e-05, "loss": 0.8654, "step": 1926 }, { "epoch": 0.31, "learning_rate": 9.999746770345416e-05, "loss": 0.8463, "step": 1927 }, { "epoch": 0.31, "learning_rate": 9.999737304710761e-05, "loss": 0.9597, "step": 1928 }, { "epoch": 0.31, "learning_rate": 9.999727665400875e-05, "loss": 0.8497, "step": 1929 }, { "epoch": 0.31, "learning_rate": 9.999717852416095e-05, "loss": 0.8433, "step": 1930 }, { "epoch": 0.31, "learning_rate": 9.999707865756759e-05, "loss": 0.8616, "step": 1931 }, { "epoch": 0.31, "learning_rate": 9.999697705423214e-05, "loss": 0.8447, "step": 1932 }, { "epoch": 0.31, "learning_rate": 9.999687371415815e-05, "loss": 0.8896, "step": 1933 }, { "epoch": 0.31, "learning_rate": 9.999676863734919e-05, "loss": 0.9009, "step": 1934 }, { "epoch": 0.31, "learning_rate": 9.99966618238089e-05, "loss": 0.8302, "step": 1935 }, { "epoch": 0.31, "learning_rate": 9.999655327354102e-05, "loss": 0.8558, "step": 1936 }, { "epoch": 0.31, "learning_rate": 9.99964429865493e-05, "loss": 0.9253, "step": 1937 }, { "epoch": 0.31, "learning_rate": 9.999633096283758e-05, "loss": 0.8151, "step": 1938 }, { "epoch": 0.31, "learning_rate": 9.999621720240975e-05, "loss": 0.7794, "step": 1939 }, { "epoch": 0.31, "learning_rate": 9.999610170526974e-05, "loss": 0.9153, "step": 1940 }, { "epoch": 0.31, "learning_rate": 9.999598447142161e-05, "loss": 0.9172, "step": 1941 }, { "epoch": 0.31, "learning_rate": 9.999586550086939e-05, "loss": 0.8052, "step": 1942 }, { "epoch": 0.31, "learning_rate": 9.999574479361724e-05, "loss": 0.8268, "step": 1943 }, { "epoch": 0.31, "learning_rate": 9.999562234966932e-05, "loss": 0.8303, "step": 1944 }, { "epoch": 0.31, "learning_rate": 9.999549816902992e-05, "loss": 0.7712, "step": 1945 }, { "epoch": 0.31, "learning_rate": 9.999537225170333e-05, "loss": 0.8592, "step": 1946 }, { "epoch": 0.31, "learning_rate": 9.999524459769394e-05, "loss": 0.8305, "step": 1947 }, { "epoch": 0.31, "learning_rate": 9.999511520700617e-05, "loss": 0.9389, "step": 1948 }, { "epoch": 0.31, "learning_rate": 9.999498407964452e-05, "loss": 0.8929, "step": 1949 }, { "epoch": 0.31, "learning_rate": 9.999485121561354e-05, "loss": 0.8865, "step": 1950 }, { "epoch": 0.31, "learning_rate": 9.999471661491786e-05, "loss": 0.8435, "step": 1951 }, { "epoch": 0.31, "learning_rate": 9.999458027756214e-05, "loss": 0.7746, "step": 1952 }, { "epoch": 0.31, "learning_rate": 9.999444220355113e-05, "loss": 0.8541, "step": 1953 }, { "epoch": 0.31, "learning_rate": 9.999430239288961e-05, "loss": 0.8788, "step": 1954 }, { "epoch": 0.31, "learning_rate": 9.999416084558244e-05, "loss": 0.9061, "step": 1955 }, { "epoch": 0.31, "learning_rate": 9.999401756163454e-05, "loss": 0.8229, "step": 1956 }, { "epoch": 0.31, "learning_rate": 9.99938725410509e-05, "loss": 0.8353, "step": 1957 }, { "epoch": 0.31, "learning_rate": 9.999372578383654e-05, "loss": 0.8225, "step": 1958 }, { "epoch": 0.31, "learning_rate": 9.999357728999657e-05, "loss": 0.852, "step": 1959 }, { "epoch": 0.31, "learning_rate": 9.999342705953613e-05, "loss": 0.79, "step": 1960 }, { "epoch": 0.31, "learning_rate": 9.999327509246047e-05, "loss": 0.8861, "step": 1961 }, { "epoch": 0.31, "learning_rate": 9.999312138877484e-05, "loss": 0.8002, "step": 1962 }, { "epoch": 0.31, "learning_rate": 9.999296594848457e-05, "loss": 0.858, "step": 1963 }, { "epoch": 0.31, "learning_rate": 9.999280877159512e-05, "loss": 0.8167, "step": 1964 }, { "epoch": 0.31, "learning_rate": 9.999264985811188e-05, "loss": 0.8326, "step": 1965 }, { "epoch": 0.31, "learning_rate": 9.999248920804041e-05, "loss": 0.8971, "step": 1966 }, { "epoch": 0.32, "learning_rate": 9.999232682138627e-05, "loss": 0.8005, "step": 1967 }, { "epoch": 0.32, "learning_rate": 9.99921626981551e-05, "loss": 0.7948, "step": 1968 }, { "epoch": 0.32, "learning_rate": 9.999199683835262e-05, "loss": 0.8201, "step": 1969 }, { "epoch": 0.32, "learning_rate": 9.999182924198461e-05, "loss": 0.8504, "step": 1970 }, { "epoch": 0.32, "learning_rate": 9.999165990905683e-05, "loss": 0.9165, "step": 1971 }, { "epoch": 0.32, "learning_rate": 9.999148883957521e-05, "loss": 0.9248, "step": 1972 }, { "epoch": 0.32, "learning_rate": 9.999131603354567e-05, "loss": 0.8247, "step": 1973 }, { "epoch": 0.32, "learning_rate": 9.999114149097422e-05, "loss": 0.8353, "step": 1974 }, { "epoch": 0.32, "learning_rate": 9.999096521186694e-05, "loss": 0.8213, "step": 1975 }, { "epoch": 0.32, "learning_rate": 9.999078719622992e-05, "loss": 0.8959, "step": 1976 }, { "epoch": 0.32, "learning_rate": 9.999060744406938e-05, "loss": 0.8314, "step": 1977 }, { "epoch": 0.32, "learning_rate": 9.999042595539155e-05, "loss": 0.8525, "step": 1978 }, { "epoch": 0.32, "learning_rate": 9.99902427302027e-05, "loss": 0.7908, "step": 1979 }, { "epoch": 0.32, "learning_rate": 9.999005776850925e-05, "loss": 0.8141, "step": 1980 }, { "epoch": 0.32, "learning_rate": 9.998987107031758e-05, "loss": 0.7772, "step": 1981 }, { "epoch": 0.32, "learning_rate": 9.998968263563423e-05, "loss": 0.8721, "step": 1982 }, { "epoch": 0.32, "learning_rate": 9.998949246446567e-05, "loss": 0.8196, "step": 1983 }, { "epoch": 0.32, "learning_rate": 9.998930055681856e-05, "loss": 0.8463, "step": 1984 }, { "epoch": 0.32, "learning_rate": 9.998910691269955e-05, "loss": 0.869, "step": 1985 }, { "epoch": 0.32, "learning_rate": 9.998891153211538e-05, "loss": 0.877, "step": 1986 }, { "epoch": 0.32, "learning_rate": 9.998871441507283e-05, "loss": 0.9004, "step": 1987 }, { "epoch": 0.32, "learning_rate": 9.998851556157873e-05, "loss": 0.8386, "step": 1988 }, { "epoch": 0.32, "learning_rate": 9.998831497164e-05, "loss": 0.811, "step": 1989 }, { "epoch": 0.32, "learning_rate": 9.998811264526363e-05, "loss": 0.8486, "step": 1990 }, { "epoch": 0.32, "learning_rate": 9.998790858245662e-05, "loss": 0.8326, "step": 1991 }, { "epoch": 0.32, "learning_rate": 9.998770278322604e-05, "loss": 0.8796, "step": 1992 }, { "epoch": 0.32, "learning_rate": 9.998749524757909e-05, "loss": 0.86, "step": 1993 }, { "epoch": 0.32, "learning_rate": 9.998728597552294e-05, "loss": 0.8319, "step": 1994 }, { "epoch": 0.32, "learning_rate": 9.99870749670649e-05, "loss": 0.8532, "step": 1995 }, { "epoch": 0.32, "learning_rate": 9.998686222221224e-05, "loss": 0.8141, "step": 1996 }, { "epoch": 0.32, "learning_rate": 9.998664774097239e-05, "loss": 0.8911, "step": 1997 }, { "epoch": 0.32, "learning_rate": 9.99864315233528e-05, "loss": 0.7875, "step": 1998 }, { "epoch": 0.32, "learning_rate": 9.998621356936098e-05, "loss": 0.8265, "step": 1999 }, { "epoch": 0.32, "learning_rate": 9.998599387900449e-05, "loss": 0.8158, "step": 2000 }, { "epoch": 0.32, "learning_rate": 9.998577245229097e-05, "loss": 0.8281, "step": 2001 }, { "epoch": 0.32, "learning_rate": 9.99855492892281e-05, "loss": 0.7661, "step": 2002 }, { "epoch": 0.32, "learning_rate": 9.998532438982364e-05, "loss": 0.8691, "step": 2003 }, { "epoch": 0.32, "learning_rate": 9.99850977540854e-05, "loss": 0.785, "step": 2004 }, { "epoch": 0.32, "learning_rate": 9.998486938202127e-05, "loss": 0.8768, "step": 2005 }, { "epoch": 0.32, "learning_rate": 9.998463927363915e-05, "loss": 0.7654, "step": 2006 }, { "epoch": 0.32, "learning_rate": 9.998440742894706e-05, "loss": 0.8557, "step": 2007 }, { "epoch": 0.32, "learning_rate": 9.998417384795304e-05, "loss": 0.7672, "step": 2008 }, { "epoch": 0.32, "learning_rate": 9.998393853066522e-05, "loss": 0.8468, "step": 2009 }, { "epoch": 0.32, "learning_rate": 9.998370147709175e-05, "loss": 0.8035, "step": 2010 }, { "epoch": 0.32, "learning_rate": 9.998346268724089e-05, "loss": 0.8743, "step": 2011 }, { "epoch": 0.32, "learning_rate": 9.998322216112092e-05, "loss": 0.8355, "step": 2012 }, { "epoch": 0.32, "learning_rate": 9.998297989874019e-05, "loss": 0.823, "step": 2013 }, { "epoch": 0.32, "learning_rate": 9.998273590010712e-05, "loss": 0.8717, "step": 2014 }, { "epoch": 0.32, "learning_rate": 9.99824901652302e-05, "loss": 0.9096, "step": 2015 }, { "epoch": 0.32, "learning_rate": 9.998224269411794e-05, "loss": 0.844, "step": 2016 }, { "epoch": 0.32, "learning_rate": 9.998199348677897e-05, "loss": 0.8528, "step": 2017 }, { "epoch": 0.32, "learning_rate": 9.998174254322191e-05, "loss": 0.823, "step": 2018 }, { "epoch": 0.32, "learning_rate": 9.998148986345551e-05, "loss": 0.7388, "step": 2019 }, { "epoch": 0.32, "learning_rate": 9.998123544748852e-05, "loss": 0.8813, "step": 2020 }, { "epoch": 0.32, "learning_rate": 9.99809792953298e-05, "loss": 0.8034, "step": 2021 }, { "epoch": 0.32, "learning_rate": 9.998072140698823e-05, "loss": 0.8397, "step": 2022 }, { "epoch": 0.32, "learning_rate": 9.998046178247276e-05, "loss": 0.8222, "step": 2023 }, { "epoch": 0.32, "learning_rate": 9.998020042179245e-05, "loss": 0.7914, "step": 2024 }, { "epoch": 0.32, "learning_rate": 9.997993732495634e-05, "loss": 0.7979, "step": 2025 }, { "epoch": 0.32, "learning_rate": 9.997967249197359e-05, "loss": 0.8701, "step": 2026 }, { "epoch": 0.32, "learning_rate": 9.997940592285338e-05, "loss": 0.8141, "step": 2027 }, { "epoch": 0.32, "learning_rate": 9.997913761760499e-05, "loss": 0.7864, "step": 2028 }, { "epoch": 0.32, "learning_rate": 9.997886757623774e-05, "loss": 0.8737, "step": 2029 }, { "epoch": 0.33, "learning_rate": 9.997859579876099e-05, "loss": 0.8855, "step": 2030 }, { "epoch": 0.33, "learning_rate": 9.997832228518419e-05, "loss": 0.7844, "step": 2031 }, { "epoch": 0.33, "learning_rate": 9.997804703551685e-05, "loss": 0.7913, "step": 2032 }, { "epoch": 0.33, "learning_rate": 9.997777004976852e-05, "loss": 0.8179, "step": 2033 }, { "epoch": 0.33, "learning_rate": 9.997749132794882e-05, "loss": 0.8402, "step": 2034 }, { "epoch": 0.33, "learning_rate": 9.997721087006746e-05, "loss": 0.8483, "step": 2035 }, { "epoch": 0.33, "learning_rate": 9.997692867613415e-05, "loss": 0.8481, "step": 2036 }, { "epoch": 0.33, "learning_rate": 9.99766447461587e-05, "loss": 0.7957, "step": 2037 }, { "epoch": 0.33, "learning_rate": 9.997635908015097e-05, "loss": 0.8752, "step": 2038 }, { "epoch": 0.33, "learning_rate": 9.997607167812091e-05, "loss": 0.831, "step": 2039 }, { "epoch": 0.33, "learning_rate": 9.997578254007846e-05, "loss": 0.8656, "step": 2040 }, { "epoch": 0.33, "learning_rate": 9.997549166603371e-05, "loss": 0.8017, "step": 2041 }, { "epoch": 0.33, "learning_rate": 9.997519905599673e-05, "loss": 0.8263, "step": 2042 }, { "epoch": 0.33, "learning_rate": 9.997490470997769e-05, "loss": 0.8307, "step": 2043 }, { "epoch": 0.33, "learning_rate": 9.997460862798681e-05, "loss": 0.8829, "step": 2044 }, { "epoch": 0.33, "learning_rate": 9.99743108100344e-05, "loss": 0.8148, "step": 2045 }, { "epoch": 0.33, "learning_rate": 9.997401125613078e-05, "loss": 0.8016, "step": 2046 }, { "epoch": 0.33, "learning_rate": 9.997370996628638e-05, "loss": 0.8058, "step": 2047 }, { "epoch": 0.33, "learning_rate": 9.997340694051164e-05, "loss": 0.7106, "step": 2048 }, { "epoch": 0.33, "learning_rate": 9.99731021788171e-05, "loss": 0.8797, "step": 2049 }, { "epoch": 0.33, "learning_rate": 9.997279568121334e-05, "loss": 0.8299, "step": 2050 }, { "epoch": 0.33, "learning_rate": 9.997248744771102e-05, "loss": 0.8501, "step": 2051 }, { "epoch": 0.33, "learning_rate": 9.997217747832083e-05, "loss": 0.8787, "step": 2052 }, { "epoch": 0.33, "learning_rate": 9.997186577305355e-05, "loss": 0.7664, "step": 2053 }, { "epoch": 0.33, "learning_rate": 9.997155233192e-05, "loss": 0.8119, "step": 2054 }, { "epoch": 0.33, "learning_rate": 9.997123715493106e-05, "loss": 0.883, "step": 2055 }, { "epoch": 0.33, "learning_rate": 9.99709202420977e-05, "loss": 0.9351, "step": 2056 }, { "epoch": 0.33, "learning_rate": 9.997060159343092e-05, "loss": 0.8854, "step": 2057 }, { "epoch": 0.33, "learning_rate": 9.997028120894179e-05, "loss": 0.8319, "step": 2058 }, { "epoch": 0.33, "learning_rate": 9.996995908864142e-05, "loss": 0.8328, "step": 2059 }, { "epoch": 0.33, "learning_rate": 9.996963523254101e-05, "loss": 0.8293, "step": 2060 }, { "epoch": 0.33, "learning_rate": 9.996930964065183e-05, "loss": 0.855, "step": 2061 }, { "epoch": 0.33, "learning_rate": 9.996898231298519e-05, "loss": 0.793, "step": 2062 }, { "epoch": 0.33, "learning_rate": 9.996865324955242e-05, "loss": 0.8108, "step": 2063 }, { "epoch": 0.33, "learning_rate": 9.996832245036498e-05, "loss": 0.9099, "step": 2064 }, { "epoch": 0.33, "learning_rate": 9.996798991543437e-05, "loss": 0.8672, "step": 2065 }, { "epoch": 0.33, "learning_rate": 9.99676556447721e-05, "loss": 0.8146, "step": 2066 }, { "epoch": 0.33, "learning_rate": 9.996731963838984e-05, "loss": 0.8575, "step": 2067 }, { "epoch": 0.33, "learning_rate": 9.996698189629921e-05, "loss": 0.7978, "step": 2068 }, { "epoch": 0.33, "learning_rate": 9.996664241851197e-05, "loss": 0.7912, "step": 2069 }, { "epoch": 0.33, "learning_rate": 9.996630120503991e-05, "loss": 0.8505, "step": 2070 }, { "epoch": 0.33, "learning_rate": 9.996595825589488e-05, "loss": 0.8731, "step": 2071 }, { "epoch": 0.33, "learning_rate": 9.996561357108878e-05, "loss": 0.8857, "step": 2072 }, { "epoch": 0.33, "learning_rate": 9.99652671506336e-05, "loss": 0.7934, "step": 2073 }, { "epoch": 0.33, "learning_rate": 9.996491899454136e-05, "loss": 0.8461, "step": 2074 }, { "epoch": 0.33, "learning_rate": 9.996456910282419e-05, "loss": 0.883, "step": 2075 }, { "epoch": 0.33, "learning_rate": 9.996421747549419e-05, "loss": 0.8384, "step": 2076 }, { "epoch": 0.33, "learning_rate": 9.996386411256361e-05, "loss": 0.7766, "step": 2077 }, { "epoch": 0.33, "learning_rate": 9.996350901404472e-05, "loss": 0.9088, "step": 2078 }, { "epoch": 0.33, "learning_rate": 9.996315217994984e-05, "loss": 0.8901, "step": 2079 }, { "epoch": 0.33, "learning_rate": 9.996279361029138e-05, "loss": 0.9015, "step": 2080 }, { "epoch": 0.33, "learning_rate": 9.996243330508179e-05, "loss": 0.8259, "step": 2081 }, { "epoch": 0.33, "learning_rate": 9.996207126433359e-05, "loss": 0.8037, "step": 2082 }, { "epoch": 0.33, "learning_rate": 9.996170748805935e-05, "loss": 0.8817, "step": 2083 }, { "epoch": 0.33, "learning_rate": 9.996134197627171e-05, "loss": 0.9644, "step": 2084 }, { "epoch": 0.33, "learning_rate": 9.996097472898336e-05, "loss": 0.8355, "step": 2085 }, { "epoch": 0.33, "learning_rate": 9.996060574620707e-05, "loss": 0.8051, "step": 2086 }, { "epoch": 0.33, "learning_rate": 9.996023502795566e-05, "loss": 0.8913, "step": 2087 }, { "epoch": 0.33, "learning_rate": 9.9959862574242e-05, "loss": 0.8254, "step": 2088 }, { "epoch": 0.33, "learning_rate": 9.9959488385079e-05, "loss": 0.8211, "step": 2089 }, { "epoch": 0.33, "learning_rate": 9.995911246047971e-05, "loss": 0.7724, "step": 2090 }, { "epoch": 0.33, "learning_rate": 9.995873480045714e-05, "loss": 0.899, "step": 2091 }, { "epoch": 0.34, "learning_rate": 9.995835540502446e-05, "loss": 0.8513, "step": 2092 }, { "epoch": 0.34, "learning_rate": 9.99579742741948e-05, "loss": 0.9031, "step": 2093 }, { "epoch": 0.34, "learning_rate": 9.995759140798142e-05, "loss": 0.7135, "step": 2094 }, { "epoch": 0.34, "learning_rate": 9.995720680639761e-05, "loss": 0.877, "step": 2095 }, { "epoch": 0.34, "learning_rate": 9.995682046945677e-05, "loss": 0.7642, "step": 2096 }, { "epoch": 0.34, "learning_rate": 9.995643239717227e-05, "loss": 0.9076, "step": 2097 }, { "epoch": 0.34, "learning_rate": 9.995604258955762e-05, "loss": 0.8145, "step": 2098 }, { "epoch": 0.34, "learning_rate": 9.995565104662635e-05, "loss": 0.8664, "step": 2099 }, { "epoch": 0.34, "learning_rate": 9.995525776839205e-05, "loss": 0.8713, "step": 2100 }, { "epoch": 0.34, "learning_rate": 9.99548627548684e-05, "loss": 0.7731, "step": 2101 }, { "epoch": 0.34, "learning_rate": 9.99544660060691e-05, "loss": 0.7918, "step": 2102 }, { "epoch": 0.34, "learning_rate": 9.995406752200796e-05, "loss": 0.7504, "step": 2103 }, { "epoch": 0.34, "learning_rate": 9.995366730269881e-05, "loss": 0.7782, "step": 2104 }, { "epoch": 0.34, "learning_rate": 9.995326534815554e-05, "loss": 0.8167, "step": 2105 }, { "epoch": 0.34, "learning_rate": 9.995286165839213e-05, "loss": 0.8409, "step": 2106 }, { "epoch": 0.34, "learning_rate": 9.99524562334226e-05, "loss": 0.8665, "step": 2107 }, { "epoch": 0.34, "learning_rate": 9.995204907326103e-05, "loss": 0.8012, "step": 2108 }, { "epoch": 0.34, "learning_rate": 9.995164017792156e-05, "loss": 0.8695, "step": 2109 }, { "epoch": 0.34, "learning_rate": 9.99512295474184e-05, "loss": 0.777, "step": 2110 }, { "epoch": 0.34, "learning_rate": 9.99508171817658e-05, "loss": 0.8105, "step": 2111 }, { "epoch": 0.34, "learning_rate": 9.99504030809781e-05, "loss": 0.7725, "step": 2112 }, { "epoch": 0.34, "learning_rate": 9.994998724506969e-05, "loss": 0.8545, "step": 2113 }, { "epoch": 0.34, "learning_rate": 9.9949569674055e-05, "loss": 0.8063, "step": 2114 }, { "epoch": 0.34, "learning_rate": 9.994915036794852e-05, "loss": 0.7454, "step": 2115 }, { "epoch": 0.34, "learning_rate": 9.994872932676486e-05, "loss": 0.7228, "step": 2116 }, { "epoch": 0.34, "learning_rate": 9.994830655051861e-05, "loss": 0.8459, "step": 2117 }, { "epoch": 0.34, "learning_rate": 9.994788203922447e-05, "loss": 0.8479, "step": 2118 }, { "epoch": 0.34, "learning_rate": 9.994745579289717e-05, "loss": 0.8625, "step": 2119 }, { "epoch": 0.34, "learning_rate": 9.994702781155155e-05, "loss": 0.9242, "step": 2120 }, { "epoch": 0.34, "learning_rate": 9.994659809520243e-05, "loss": 0.8496, "step": 2121 }, { "epoch": 0.34, "learning_rate": 9.994616664386479e-05, "loss": 0.7799, "step": 2122 }, { "epoch": 0.34, "learning_rate": 9.994573345755356e-05, "loss": 0.9055, "step": 2123 }, { "epoch": 0.34, "learning_rate": 9.994529853628382e-05, "loss": 0.7998, "step": 2124 }, { "epoch": 0.34, "learning_rate": 9.994486188007071e-05, "loss": 0.8267, "step": 2125 }, { "epoch": 0.34, "learning_rate": 9.994442348892934e-05, "loss": 0.9677, "step": 2126 }, { "epoch": 0.34, "learning_rate": 9.994398336287496e-05, "loss": 0.7756, "step": 2127 }, { "epoch": 0.34, "learning_rate": 9.994354150192287e-05, "loss": 0.7581, "step": 2128 }, { "epoch": 0.34, "learning_rate": 9.99430979060884e-05, "loss": 0.8142, "step": 2129 }, { "epoch": 0.34, "learning_rate": 9.994265257538697e-05, "loss": 0.8174, "step": 2130 }, { "epoch": 0.34, "learning_rate": 9.994220550983404e-05, "loss": 0.8373, "step": 2131 }, { "epoch": 0.34, "learning_rate": 9.994175670944517e-05, "loss": 0.8423, "step": 2132 }, { "epoch": 0.34, "learning_rate": 9.994130617423592e-05, "loss": 0.9281, "step": 2133 }, { "epoch": 0.34, "learning_rate": 9.994085390422193e-05, "loss": 0.8866, "step": 2134 }, { "epoch": 0.34, "learning_rate": 9.994039989941894e-05, "loss": 0.8135, "step": 2135 }, { "epoch": 0.34, "learning_rate": 9.993994415984271e-05, "loss": 0.8084, "step": 2136 }, { "epoch": 0.34, "learning_rate": 9.993948668550907e-05, "loss": 0.8583, "step": 2137 }, { "epoch": 0.34, "learning_rate": 9.993902747643391e-05, "loss": 0.8963, "step": 2138 }, { "epoch": 0.34, "learning_rate": 9.993856653263319e-05, "loss": 0.8549, "step": 2139 }, { "epoch": 0.34, "learning_rate": 9.99381038541229e-05, "loss": 0.7939, "step": 2140 }, { "epoch": 0.34, "learning_rate": 9.993763944091912e-05, "loss": 0.8125, "step": 2141 }, { "epoch": 0.34, "learning_rate": 9.9937173293038e-05, "loss": 0.879, "step": 2142 }, { "epoch": 0.34, "learning_rate": 9.993670541049572e-05, "loss": 0.9181, "step": 2143 }, { "epoch": 0.34, "learning_rate": 9.993623579330854e-05, "loss": 0.8393, "step": 2144 }, { "epoch": 0.34, "learning_rate": 9.993576444149276e-05, "loss": 0.7317, "step": 2145 }, { "epoch": 0.34, "learning_rate": 9.993529135506476e-05, "loss": 0.9317, "step": 2146 }, { "epoch": 0.34, "learning_rate": 9.993481653404098e-05, "loss": 0.8973, "step": 2147 }, { "epoch": 0.34, "learning_rate": 9.99343399784379e-05, "loss": 0.8186, "step": 2148 }, { "epoch": 0.34, "learning_rate": 9.993386168827208e-05, "loss": 0.8264, "step": 2149 }, { "epoch": 0.34, "learning_rate": 9.993338166356014e-05, "loss": 0.8547, "step": 2150 }, { "epoch": 0.34, "learning_rate": 9.993289990431875e-05, "loss": 0.7708, "step": 2151 }, { "epoch": 0.34, "learning_rate": 9.993241641056464e-05, "loss": 0.8186, "step": 2152 }, { "epoch": 0.34, "learning_rate": 9.993193118231462e-05, "loss": 0.8619, "step": 2153 }, { "epoch": 0.34, "learning_rate": 9.993144421958553e-05, "loss": 0.815, "step": 2154 }, { "epoch": 0.35, "learning_rate": 9.993095552239428e-05, "loss": 0.7949, "step": 2155 }, { "epoch": 0.35, "learning_rate": 9.993046509075787e-05, "loss": 0.92, "step": 2156 }, { "epoch": 0.35, "learning_rate": 9.992997292469332e-05, "loss": 0.8334, "step": 2157 }, { "epoch": 0.35, "learning_rate": 9.992947902421774e-05, "loss": 0.8611, "step": 2158 }, { "epoch": 0.35, "learning_rate": 9.992898338934826e-05, "loss": 0.8397, "step": 2159 }, { "epoch": 0.35, "learning_rate": 9.992848602010212e-05, "loss": 0.9366, "step": 2160 }, { "epoch": 0.35, "learning_rate": 9.992798691649659e-05, "loss": 0.86, "step": 2161 }, { "epoch": 0.35, "learning_rate": 9.9927486078549e-05, "loss": 0.7918, "step": 2162 }, { "epoch": 0.35, "learning_rate": 9.992698350627676e-05, "loss": 0.8279, "step": 2163 }, { "epoch": 0.35, "learning_rate": 9.992647919969732e-05, "loss": 0.8147, "step": 2164 }, { "epoch": 0.35, "learning_rate": 9.99259731588282e-05, "loss": 0.8096, "step": 2165 }, { "epoch": 0.35, "learning_rate": 9.992546538368698e-05, "loss": 0.8593, "step": 2166 }, { "epoch": 0.35, "learning_rate": 9.992495587429129e-05, "loss": 0.7621, "step": 2167 }, { "epoch": 0.35, "learning_rate": 9.992444463065884e-05, "loss": 0.7654, "step": 2168 }, { "epoch": 0.35, "learning_rate": 9.99239316528074e-05, "loss": 0.8461, "step": 2169 }, { "epoch": 0.35, "learning_rate": 9.992341694075476e-05, "loss": 0.7726, "step": 2170 }, { "epoch": 0.35, "learning_rate": 9.99229004945188e-05, "loss": 0.8025, "step": 2171 }, { "epoch": 0.35, "learning_rate": 9.992238231411751e-05, "loss": 0.8769, "step": 2172 }, { "epoch": 0.35, "learning_rate": 9.992186239956883e-05, "loss": 0.8312, "step": 2173 }, { "epoch": 0.35, "learning_rate": 9.992134075089084e-05, "loss": 0.8031, "step": 2174 }, { "epoch": 0.35, "learning_rate": 9.992081736810166e-05, "loss": 0.8597, "step": 2175 }, { "epoch": 0.35, "learning_rate": 9.992029225121949e-05, "loss": 0.8263, "step": 2176 }, { "epoch": 0.35, "learning_rate": 9.991976540026254e-05, "loss": 0.9011, "step": 2177 }, { "epoch": 0.35, "learning_rate": 9.991923681524914e-05, "loss": 0.8669, "step": 2178 }, { "epoch": 0.35, "learning_rate": 9.991870649619763e-05, "loss": 0.8229, "step": 2179 }, { "epoch": 0.35, "learning_rate": 9.991817444312645e-05, "loss": 0.8609, "step": 2180 }, { "epoch": 0.35, "learning_rate": 9.991764065605406e-05, "loss": 0.8546, "step": 2181 }, { "epoch": 0.35, "learning_rate": 9.991710513499901e-05, "loss": 0.9265, "step": 2182 }, { "epoch": 0.35, "learning_rate": 9.991656787997992e-05, "loss": 0.7647, "step": 2183 }, { "epoch": 0.35, "learning_rate": 9.991602889101543e-05, "loss": 0.9718, "step": 2184 }, { "epoch": 0.35, "learning_rate": 9.991548816812427e-05, "loss": 0.8975, "step": 2185 }, { "epoch": 0.35, "learning_rate": 9.991494571132523e-05, "loss": 0.8361, "step": 2186 }, { "epoch": 0.35, "learning_rate": 9.991440152063715e-05, "loss": 0.8256, "step": 2187 }, { "epoch": 0.35, "learning_rate": 9.991385559607892e-05, "loss": 0.8442, "step": 2188 }, { "epoch": 0.35, "learning_rate": 9.991330793766952e-05, "loss": 0.807, "step": 2189 }, { "epoch": 0.35, "learning_rate": 9.991275854542797e-05, "loss": 0.8595, "step": 2190 }, { "epoch": 0.35, "learning_rate": 9.991220741937335e-05, "loss": 0.8023, "step": 2191 }, { "epoch": 0.35, "learning_rate": 9.991165455952481e-05, "loss": 0.802, "step": 2192 }, { "epoch": 0.35, "learning_rate": 9.991109996590155e-05, "loss": 0.7871, "step": 2193 }, { "epoch": 0.35, "learning_rate": 9.991054363852284e-05, "loss": 0.8489, "step": 2194 }, { "epoch": 0.35, "learning_rate": 9.990998557740801e-05, "loss": 0.8893, "step": 2195 }, { "epoch": 0.35, "learning_rate": 9.990942578257641e-05, "loss": 0.8552, "step": 2196 }, { "epoch": 0.35, "learning_rate": 9.990886425404753e-05, "loss": 0.8059, "step": 2197 }, { "epoch": 0.35, "learning_rate": 9.990830099184086e-05, "loss": 0.8197, "step": 2198 }, { "epoch": 0.35, "learning_rate": 9.990773599597596e-05, "loss": 0.7944, "step": 2199 }, { "epoch": 0.35, "learning_rate": 9.990716926647248e-05, "loss": 0.8559, "step": 2200 }, { "epoch": 0.35, "learning_rate": 9.990660080335007e-05, "loss": 0.8078, "step": 2201 }, { "epoch": 0.35, "learning_rate": 9.990603060662848e-05, "loss": 0.7864, "step": 2202 }, { "epoch": 0.35, "learning_rate": 9.990545867632755e-05, "loss": 0.9162, "step": 2203 }, { "epoch": 0.35, "learning_rate": 9.99048850124671e-05, "loss": 0.8785, "step": 2204 }, { "epoch": 0.35, "learning_rate": 9.990430961506712e-05, "loss": 0.8502, "step": 2205 }, { "epoch": 0.35, "learning_rate": 9.990373248414755e-05, "loss": 0.8409, "step": 2206 }, { "epoch": 0.35, "learning_rate": 9.990315361972844e-05, "loss": 0.7742, "step": 2207 }, { "epoch": 0.35, "learning_rate": 9.990257302182993e-05, "loss": 0.7969, "step": 2208 }, { "epoch": 0.35, "learning_rate": 9.990199069047214e-05, "loss": 0.8108, "step": 2209 }, { "epoch": 0.35, "learning_rate": 9.990140662567533e-05, "loss": 0.7738, "step": 2210 }, { "epoch": 0.35, "learning_rate": 9.99008208274598e-05, "loss": 0.9146, "step": 2211 }, { "epoch": 0.35, "learning_rate": 9.990023329584587e-05, "loss": 0.8536, "step": 2212 }, { "epoch": 0.35, "learning_rate": 9.989964403085397e-05, "loss": 0.9099, "step": 2213 }, { "epoch": 0.35, "learning_rate": 9.989905303250454e-05, "loss": 0.9396, "step": 2214 }, { "epoch": 0.35, "learning_rate": 9.989846030081814e-05, "loss": 0.8494, "step": 2215 }, { "epoch": 0.35, "learning_rate": 9.989786583581535e-05, "loss": 0.8012, "step": 2216 }, { "epoch": 0.36, "learning_rate": 9.989726963751682e-05, "loss": 0.7665, "step": 2217 }, { "epoch": 0.36, "learning_rate": 9.989667170594326e-05, "loss": 0.8296, "step": 2218 }, { "epoch": 0.36, "learning_rate": 9.989607204111544e-05, "loss": 0.7929, "step": 2219 }, { "epoch": 0.36, "learning_rate": 9.989547064305418e-05, "loss": 0.7919, "step": 2220 }, { "epoch": 0.36, "learning_rate": 9.989486751178039e-05, "loss": 0.7475, "step": 2221 }, { "epoch": 0.36, "learning_rate": 9.9894262647315e-05, "loss": 0.822, "step": 2222 }, { "epoch": 0.36, "learning_rate": 9.989365604967905e-05, "loss": 0.9314, "step": 2223 }, { "epoch": 0.36, "learning_rate": 9.989304771889359e-05, "loss": 0.7469, "step": 2224 }, { "epoch": 0.36, "learning_rate": 9.989243765497976e-05, "loss": 0.8805, "step": 2225 }, { "epoch": 0.36, "learning_rate": 9.989182585795873e-05, "loss": 0.8504, "step": 2226 }, { "epoch": 0.36, "learning_rate": 9.989121232785176e-05, "loss": 0.7833, "step": 2227 }, { "epoch": 0.36, "learning_rate": 9.98905970646802e-05, "loss": 0.864, "step": 2228 }, { "epoch": 0.36, "learning_rate": 9.988998006846537e-05, "loss": 0.8168, "step": 2229 }, { "epoch": 0.36, "learning_rate": 9.988936133922875e-05, "loss": 0.8147, "step": 2230 }, { "epoch": 0.36, "learning_rate": 9.98887408769918e-05, "loss": 0.8427, "step": 2231 }, { "epoch": 0.36, "learning_rate": 9.988811868177607e-05, "loss": 0.9181, "step": 2232 }, { "epoch": 0.36, "learning_rate": 9.98874947536032e-05, "loss": 0.9245, "step": 2233 }, { "epoch": 0.36, "learning_rate": 9.988686909249483e-05, "loss": 0.8528, "step": 2234 }, { "epoch": 0.36, "learning_rate": 9.988624169847272e-05, "loss": 0.7552, "step": 2235 }, { "epoch": 0.36, "learning_rate": 9.988561257155866e-05, "loss": 0.9027, "step": 2236 }, { "epoch": 0.36, "learning_rate": 9.988498171177449e-05, "loss": 0.7336, "step": 2237 }, { "epoch": 0.36, "learning_rate": 9.988434911914215e-05, "loss": 0.8321, "step": 2238 }, { "epoch": 0.36, "learning_rate": 9.988371479368357e-05, "loss": 0.7758, "step": 2239 }, { "epoch": 0.36, "learning_rate": 9.988307873542083e-05, "loss": 0.7371, "step": 2240 }, { "epoch": 0.36, "learning_rate": 9.9882440944376e-05, "loss": 0.8301, "step": 2241 }, { "epoch": 0.36, "learning_rate": 9.988180142057124e-05, "loss": 0.8404, "step": 2242 }, { "epoch": 0.36, "learning_rate": 9.988116016402877e-05, "loss": 0.8198, "step": 2243 }, { "epoch": 0.36, "learning_rate": 9.988051717477088e-05, "loss": 0.8389, "step": 2244 }, { "epoch": 0.36, "learning_rate": 9.987987245281987e-05, "loss": 0.8339, "step": 2245 }, { "epoch": 0.36, "learning_rate": 9.987922599819814e-05, "loss": 0.8982, "step": 2246 }, { "epoch": 0.36, "learning_rate": 9.987857781092818e-05, "loss": 0.8597, "step": 2247 }, { "epoch": 0.36, "learning_rate": 9.987792789103248e-05, "loss": 0.7817, "step": 2248 }, { "epoch": 0.36, "learning_rate": 9.987727623853361e-05, "loss": 0.859, "step": 2249 }, { "epoch": 0.36, "learning_rate": 9.987662285345423e-05, "loss": 0.8471, "step": 2250 }, { "epoch": 0.36, "learning_rate": 9.987596773581702e-05, "loss": 0.8554, "step": 2251 }, { "epoch": 0.36, "learning_rate": 9.987531088564473e-05, "loss": 0.8203, "step": 2252 }, { "epoch": 0.36, "learning_rate": 9.987465230296021e-05, "loss": 0.7801, "step": 2253 }, { "epoch": 0.36, "learning_rate": 9.987399198778629e-05, "loss": 0.8018, "step": 2254 }, { "epoch": 0.36, "learning_rate": 9.987332994014595e-05, "loss": 0.8843, "step": 2255 }, { "epoch": 0.36, "learning_rate": 9.987266616006214e-05, "loss": 0.9373, "step": 2256 }, { "epoch": 0.36, "learning_rate": 9.987200064755797e-05, "loss": 0.8254, "step": 2257 }, { "epoch": 0.36, "learning_rate": 9.987133340265654e-05, "loss": 0.9072, "step": 2258 }, { "epoch": 0.36, "learning_rate": 9.987066442538102e-05, "loss": 0.8426, "step": 2259 }, { "epoch": 0.36, "learning_rate": 9.986999371575465e-05, "loss": 0.9163, "step": 2260 }, { "epoch": 0.36, "learning_rate": 9.986932127380072e-05, "loss": 0.8469, "step": 2261 }, { "epoch": 0.36, "learning_rate": 9.98686470995426e-05, "loss": 0.8326, "step": 2262 }, { "epoch": 0.36, "learning_rate": 9.98679711930037e-05, "loss": 0.885, "step": 2263 }, { "epoch": 0.36, "learning_rate": 9.986729355420752e-05, "loss": 0.8478, "step": 2264 }, { "epoch": 0.36, "learning_rate": 9.986661418317759e-05, "loss": 0.8317, "step": 2265 }, { "epoch": 0.36, "learning_rate": 9.986593307993749e-05, "loss": 0.7681, "step": 2266 }, { "epoch": 0.36, "learning_rate": 9.98652502445109e-05, "loss": 0.7891, "step": 2267 }, { "epoch": 0.36, "learning_rate": 9.986456567692153e-05, "loss": 0.7426, "step": 2268 }, { "epoch": 0.36, "learning_rate": 9.986387937719316e-05, "loss": 0.8045, "step": 2269 }, { "epoch": 0.36, "learning_rate": 9.986319134534963e-05, "loss": 0.8408, "step": 2270 }, { "epoch": 0.36, "learning_rate": 9.986250158141484e-05, "loss": 0.8241, "step": 2271 }, { "epoch": 0.36, "learning_rate": 9.986181008541277e-05, "loss": 0.8545, "step": 2272 }, { "epoch": 0.36, "learning_rate": 9.986111685736741e-05, "loss": 0.7871, "step": 2273 }, { "epoch": 0.36, "learning_rate": 9.986042189730284e-05, "loss": 0.9034, "step": 2274 }, { "epoch": 0.36, "learning_rate": 9.985972520524324e-05, "loss": 0.8113, "step": 2275 }, { "epoch": 0.36, "learning_rate": 9.985902678121277e-05, "loss": 0.8729, "step": 2276 }, { "epoch": 0.36, "learning_rate": 9.98583266252357e-05, "loss": 0.921, "step": 2277 }, { "epoch": 0.36, "learning_rate": 9.985762473733637e-05, "loss": 0.8539, "step": 2278 }, { "epoch": 0.37, "learning_rate": 9.985692111753915e-05, "loss": 0.8375, "step": 2279 }, { "epoch": 0.37, "learning_rate": 9.985621576586848e-05, "loss": 0.9056, "step": 2280 }, { "epoch": 0.37, "learning_rate": 9.985550868234886e-05, "loss": 0.8119, "step": 2281 }, { "epoch": 0.37, "learning_rate": 9.985479986700486e-05, "loss": 0.8404, "step": 2282 }, { "epoch": 0.37, "learning_rate": 9.985408931986107e-05, "loss": 0.7685, "step": 2283 }, { "epoch": 0.37, "learning_rate": 9.985337704094222e-05, "loss": 0.885, "step": 2284 }, { "epoch": 0.37, "learning_rate": 9.985266303027302e-05, "loss": 0.7503, "step": 2285 }, { "epoch": 0.37, "learning_rate": 9.98519472878783e-05, "loss": 0.7905, "step": 2286 }, { "epoch": 0.37, "learning_rate": 9.98512298137829e-05, "loss": 0.8403, "step": 2287 }, { "epoch": 0.37, "learning_rate": 9.985051060801174e-05, "loss": 0.8056, "step": 2288 }, { "epoch": 0.37, "learning_rate": 9.984978967058982e-05, "loss": 0.832, "step": 2289 }, { "epoch": 0.37, "learning_rate": 9.984906700154217e-05, "loss": 0.8669, "step": 2290 }, { "epoch": 0.37, "learning_rate": 9.98483426008939e-05, "loss": 0.783, "step": 2291 }, { "epoch": 0.37, "learning_rate": 9.984761646867016e-05, "loss": 0.8048, "step": 2292 }, { "epoch": 0.37, "learning_rate": 9.984688860489619e-05, "loss": 0.8681, "step": 2293 }, { "epoch": 0.37, "learning_rate": 9.984615900959728e-05, "loss": 0.9621, "step": 2294 }, { "epoch": 0.37, "learning_rate": 9.984542768279877e-05, "loss": 0.83, "step": 2295 }, { "epoch": 0.37, "learning_rate": 9.984469462452603e-05, "loss": 0.8713, "step": 2296 }, { "epoch": 0.37, "learning_rate": 9.984395983480456e-05, "loss": 0.8303, "step": 2297 }, { "epoch": 0.37, "learning_rate": 9.984322331365987e-05, "loss": 0.8559, "step": 2298 }, { "epoch": 0.37, "learning_rate": 9.984248506111756e-05, "loss": 0.8953, "step": 2299 }, { "epoch": 0.37, "learning_rate": 9.984174507720326e-05, "loss": 0.8946, "step": 2300 }, { "epoch": 0.37, "learning_rate": 9.984100336194268e-05, "loss": 0.8699, "step": 2301 }, { "epoch": 0.37, "learning_rate": 9.984025991536158e-05, "loss": 0.7861, "step": 2302 }, { "epoch": 0.37, "learning_rate": 9.983951473748578e-05, "loss": 0.8165, "step": 2303 }, { "epoch": 0.37, "learning_rate": 9.983876782834118e-05, "loss": 0.904, "step": 2304 }, { "epoch": 0.37, "learning_rate": 9.983801918795373e-05, "loss": 0.8561, "step": 2305 }, { "epoch": 0.37, "learning_rate": 9.983726881634943e-05, "loss": 0.848, "step": 2306 }, { "epoch": 0.37, "learning_rate": 9.983651671355432e-05, "loss": 0.8391, "step": 2307 }, { "epoch": 0.37, "learning_rate": 9.983576287959455e-05, "loss": 0.749, "step": 2308 }, { "epoch": 0.37, "learning_rate": 9.98350073144963e-05, "loss": 0.8654, "step": 2309 }, { "epoch": 0.37, "learning_rate": 9.983425001828581e-05, "loss": 0.9472, "step": 2310 }, { "epoch": 0.37, "learning_rate": 9.983349099098942e-05, "loss": 0.7552, "step": 2311 }, { "epoch": 0.37, "learning_rate": 9.983273023263344e-05, "loss": 0.7591, "step": 2312 }, { "epoch": 0.37, "learning_rate": 9.983196774324435e-05, "loss": 0.9038, "step": 2313 }, { "epoch": 0.37, "learning_rate": 9.98312035228486e-05, "loss": 0.8333, "step": 2314 }, { "epoch": 0.37, "learning_rate": 9.983043757147275e-05, "loss": 0.8911, "step": 2315 }, { "epoch": 0.37, "learning_rate": 9.98296698891434e-05, "loss": 0.7357, "step": 2316 }, { "epoch": 0.37, "learning_rate": 9.982890047588725e-05, "loss": 0.8809, "step": 2317 }, { "epoch": 0.37, "learning_rate": 9.982812933173097e-05, "loss": 0.7795, "step": 2318 }, { "epoch": 0.37, "learning_rate": 9.982735645670139e-05, "loss": 0.9119, "step": 2319 }, { "epoch": 0.37, "learning_rate": 9.982658185082534e-05, "loss": 0.8846, "step": 2320 }, { "epoch": 0.37, "learning_rate": 9.982580551412972e-05, "loss": 0.8625, "step": 2321 }, { "epoch": 0.37, "learning_rate": 9.982502744664153e-05, "loss": 0.9252, "step": 2322 }, { "epoch": 0.37, "learning_rate": 9.982424764838776e-05, "loss": 0.8517, "step": 2323 }, { "epoch": 0.37, "learning_rate": 9.982346611939552e-05, "loss": 0.8197, "step": 2324 }, { "epoch": 0.37, "learning_rate": 9.982268285969196e-05, "loss": 0.8774, "step": 2325 }, { "epoch": 0.37, "learning_rate": 9.982189786930427e-05, "loss": 0.7738, "step": 2326 }, { "epoch": 0.37, "learning_rate": 9.982111114825974e-05, "loss": 0.8105, "step": 2327 }, { "epoch": 0.37, "learning_rate": 9.982032269658568e-05, "loss": 0.8097, "step": 2328 }, { "epoch": 0.37, "learning_rate": 9.981953251430949e-05, "loss": 0.7502, "step": 2329 }, { "epoch": 0.37, "learning_rate": 9.98187406014586e-05, "loss": 0.8674, "step": 2330 }, { "epoch": 0.37, "learning_rate": 9.981794695806055e-05, "loss": 0.8614, "step": 2331 }, { "epoch": 0.37, "learning_rate": 9.981715158414287e-05, "loss": 0.8659, "step": 2332 }, { "epoch": 0.37, "learning_rate": 9.981635447973323e-05, "loss": 0.9052, "step": 2333 }, { "epoch": 0.37, "learning_rate": 9.98155556448593e-05, "loss": 0.805, "step": 2334 }, { "epoch": 0.37, "learning_rate": 9.98147550795488e-05, "loss": 0.8684, "step": 2335 }, { "epoch": 0.37, "learning_rate": 9.981395278382959e-05, "loss": 0.8204, "step": 2336 }, { "epoch": 0.37, "learning_rate": 9.981314875772951e-05, "loss": 0.8597, "step": 2337 }, { "epoch": 0.37, "learning_rate": 9.981234300127648e-05, "loss": 0.8929, "step": 2338 }, { "epoch": 0.37, "learning_rate": 9.981153551449853e-05, "loss": 0.8369, "step": 2339 }, { "epoch": 0.37, "learning_rate": 9.981072629742367e-05, "loss": 0.8436, "step": 2340 }, { "epoch": 0.37, "learning_rate": 9.980991535008003e-05, "loss": 0.8532, "step": 2341 }, { "epoch": 0.38, "learning_rate": 9.980910267249577e-05, "loss": 0.6922, "step": 2342 }, { "epoch": 0.38, "learning_rate": 9.980828826469913e-05, "loss": 0.7531, "step": 2343 }, { "epoch": 0.38, "learning_rate": 9.980747212671839e-05, "loss": 0.813, "step": 2344 }, { "epoch": 0.38, "learning_rate": 9.98066542585819e-05, "loss": 0.7857, "step": 2345 }, { "epoch": 0.38, "learning_rate": 9.980583466031807e-05, "loss": 0.8843, "step": 2346 }, { "epoch": 0.38, "learning_rate": 9.980501333195538e-05, "loss": 0.8587, "step": 2347 }, { "epoch": 0.38, "learning_rate": 9.980419027352236e-05, "loss": 0.8795, "step": 2348 }, { "epoch": 0.38, "learning_rate": 9.980336548504759e-05, "loss": 0.7967, "step": 2349 }, { "epoch": 0.38, "learning_rate": 9.980253896655971e-05, "loss": 0.9035, "step": 2350 }, { "epoch": 0.38, "learning_rate": 9.980171071808747e-05, "loss": 0.878, "step": 2351 }, { "epoch": 0.38, "learning_rate": 9.98008807396596e-05, "loss": 0.9006, "step": 2352 }, { "epoch": 0.38, "learning_rate": 9.980004903130495e-05, "loss": 0.857, "step": 2353 }, { "epoch": 0.38, "learning_rate": 9.979921559305243e-05, "loss": 0.7781, "step": 2354 }, { "epoch": 0.38, "learning_rate": 9.979838042493095e-05, "loss": 0.8016, "step": 2355 }, { "epoch": 0.38, "learning_rate": 9.979754352696955e-05, "loss": 0.7843, "step": 2356 }, { "epoch": 0.38, "learning_rate": 9.979670489919729e-05, "loss": 0.8841, "step": 2357 }, { "epoch": 0.38, "learning_rate": 9.979586454164329e-05, "loss": 0.9093, "step": 2358 }, { "epoch": 0.38, "learning_rate": 9.979502245433676e-05, "loss": 0.7791, "step": 2359 }, { "epoch": 0.38, "learning_rate": 9.979417863730696e-05, "loss": 0.8226, "step": 2360 }, { "epoch": 0.38, "learning_rate": 9.979333309058318e-05, "loss": 0.8364, "step": 2361 }, { "epoch": 0.38, "learning_rate": 9.979248581419479e-05, "loss": 0.885, "step": 2362 }, { "epoch": 0.38, "learning_rate": 9.979163680817124e-05, "loss": 0.8367, "step": 2363 }, { "epoch": 0.38, "learning_rate": 9.979078607254202e-05, "loss": 0.7538, "step": 2364 }, { "epoch": 0.38, "learning_rate": 9.978993360733667e-05, "loss": 0.9061, "step": 2365 }, { "epoch": 0.38, "learning_rate": 9.97890794125848e-05, "loss": 0.8247, "step": 2366 }, { "epoch": 0.38, "learning_rate": 9.978822348831608e-05, "loss": 0.7901, "step": 2367 }, { "epoch": 0.38, "learning_rate": 9.978736583456028e-05, "loss": 0.776, "step": 2368 }, { "epoch": 0.38, "learning_rate": 9.978650645134714e-05, "loss": 0.879, "step": 2369 }, { "epoch": 0.38, "learning_rate": 9.978564533870654e-05, "loss": 0.7452, "step": 2370 }, { "epoch": 0.38, "learning_rate": 9.978478249666839e-05, "loss": 0.9529, "step": 2371 }, { "epoch": 0.38, "learning_rate": 9.978391792526267e-05, "loss": 0.8219, "step": 2372 }, { "epoch": 0.38, "learning_rate": 9.97830516245194e-05, "loss": 0.844, "step": 2373 }, { "epoch": 0.38, "learning_rate": 9.978218359446866e-05, "loss": 0.8157, "step": 2374 }, { "epoch": 0.38, "learning_rate": 9.978131383514063e-05, "loss": 0.9487, "step": 2375 }, { "epoch": 0.38, "learning_rate": 9.97804423465655e-05, "loss": 0.8369, "step": 2376 }, { "epoch": 0.38, "learning_rate": 9.977956912877356e-05, "loss": 0.8969, "step": 2377 }, { "epoch": 0.38, "learning_rate": 9.977869418179516e-05, "loss": 0.8588, "step": 2378 }, { "epoch": 0.38, "learning_rate": 9.977781750566064e-05, "loss": 0.7841, "step": 2379 }, { "epoch": 0.38, "learning_rate": 9.977693910040049e-05, "loss": 0.8756, "step": 2380 }, { "epoch": 0.38, "learning_rate": 9.977605896604521e-05, "loss": 0.7745, "step": 2381 }, { "epoch": 0.38, "learning_rate": 9.977517710262538e-05, "loss": 0.8692, "step": 2382 }, { "epoch": 0.38, "learning_rate": 9.977429351017165e-05, "loss": 0.8938, "step": 2383 }, { "epoch": 0.38, "learning_rate": 9.977340818871466e-05, "loss": 0.9045, "step": 2384 }, { "epoch": 0.38, "learning_rate": 9.977252113828524e-05, "loss": 0.8445, "step": 2385 }, { "epoch": 0.38, "learning_rate": 9.977163235891414e-05, "loss": 0.7931, "step": 2386 }, { "epoch": 0.38, "learning_rate": 9.977074185063225e-05, "loss": 0.7568, "step": 2387 }, { "epoch": 0.38, "learning_rate": 9.97698496134705e-05, "loss": 0.8021, "step": 2388 }, { "epoch": 0.38, "learning_rate": 9.976895564745991e-05, "loss": 0.7929, "step": 2389 }, { "epoch": 0.38, "learning_rate": 9.976805995263152e-05, "loss": 0.7728, "step": 2390 }, { "epoch": 0.38, "learning_rate": 9.976716252901643e-05, "loss": 0.9501, "step": 2391 }, { "epoch": 0.38, "learning_rate": 9.976626337664582e-05, "loss": 0.7923, "step": 2392 }, { "epoch": 0.38, "learning_rate": 9.976536249555093e-05, "loss": 0.9025, "step": 2393 }, { "epoch": 0.38, "learning_rate": 9.976445988576305e-05, "loss": 0.8676, "step": 2394 }, { "epoch": 0.38, "learning_rate": 9.976355554731354e-05, "loss": 0.8572, "step": 2395 }, { "epoch": 0.38, "learning_rate": 9.976264948023379e-05, "loss": 0.837, "step": 2396 }, { "epoch": 0.38, "learning_rate": 9.976174168455532e-05, "loss": 0.8322, "step": 2397 }, { "epoch": 0.38, "learning_rate": 9.976083216030962e-05, "loss": 0.8447, "step": 2398 }, { "epoch": 0.38, "learning_rate": 9.97599209075283e-05, "loss": 0.8567, "step": 2399 }, { "epoch": 0.38, "learning_rate": 9.975900792624301e-05, "loss": 0.8158, "step": 2400 }, { "epoch": 0.38, "learning_rate": 9.975809321648548e-05, "loss": 0.7896, "step": 2401 }, { "epoch": 0.38, "learning_rate": 9.975717677828747e-05, "loss": 0.7688, "step": 2402 }, { "epoch": 0.38, "learning_rate": 9.975625861168082e-05, "loss": 0.8537, "step": 2403 }, { "epoch": 0.39, "learning_rate": 9.97553387166974e-05, "loss": 0.8556, "step": 2404 }, { "epoch": 0.39, "learning_rate": 9.97544170933692e-05, "loss": 0.8872, "step": 2405 }, { "epoch": 0.39, "learning_rate": 9.975349374172823e-05, "loss": 0.9227, "step": 2406 }, { "epoch": 0.39, "learning_rate": 9.975256866180653e-05, "loss": 0.9232, "step": 2407 }, { "epoch": 0.39, "learning_rate": 9.975164185363628e-05, "loss": 0.8156, "step": 2408 }, { "epoch": 0.39, "learning_rate": 9.975071331724963e-05, "loss": 0.8603, "step": 2409 }, { "epoch": 0.39, "learning_rate": 9.974978305267887e-05, "loss": 0.7895, "step": 2410 }, { "epoch": 0.39, "learning_rate": 9.974885105995631e-05, "loss": 0.8816, "step": 2411 }, { "epoch": 0.39, "learning_rate": 9.97479173391143e-05, "loss": 0.8435, "step": 2412 }, { "epoch": 0.39, "learning_rate": 9.974698189018529e-05, "loss": 0.8607, "step": 2413 }, { "epoch": 0.39, "learning_rate": 9.974604471320179e-05, "loss": 0.8291, "step": 2414 }, { "epoch": 0.39, "learning_rate": 9.974510580819634e-05, "loss": 0.8026, "step": 2415 }, { "epoch": 0.39, "learning_rate": 9.974416517520156e-05, "loss": 0.8443, "step": 2416 }, { "epoch": 0.39, "learning_rate": 9.974322281425009e-05, "loss": 0.7889, "step": 2417 }, { "epoch": 0.39, "learning_rate": 9.974227872537471e-05, "loss": 0.7825, "step": 2418 }, { "epoch": 0.39, "learning_rate": 9.974133290860821e-05, "loss": 0.8728, "step": 2419 }, { "epoch": 0.39, "learning_rate": 9.974038536398342e-05, "loss": 0.9335, "step": 2420 }, { "epoch": 0.39, "learning_rate": 9.973943609153326e-05, "loss": 0.8971, "step": 2421 }, { "epoch": 0.39, "learning_rate": 9.973848509129073e-05, "loss": 0.8555, "step": 2422 }, { "epoch": 0.39, "learning_rate": 9.973753236328884e-05, "loss": 0.8762, "step": 2423 }, { "epoch": 0.39, "learning_rate": 9.97365779075607e-05, "loss": 0.8502, "step": 2424 }, { "epoch": 0.39, "learning_rate": 9.973562172413944e-05, "loss": 0.8079, "step": 2425 }, { "epoch": 0.39, "learning_rate": 9.973466381305829e-05, "loss": 0.8233, "step": 2426 }, { "epoch": 0.39, "learning_rate": 9.973370417435055e-05, "loss": 0.7753, "step": 2427 }, { "epoch": 0.39, "learning_rate": 9.973274280804951e-05, "loss": 0.8648, "step": 2428 }, { "epoch": 0.39, "learning_rate": 9.97317797141886e-05, "loss": 0.7923, "step": 2429 }, { "epoch": 0.39, "learning_rate": 9.973081489280125e-05, "loss": 0.8706, "step": 2430 }, { "epoch": 0.39, "learning_rate": 9.9729848343921e-05, "loss": 0.7939, "step": 2431 }, { "epoch": 0.39, "learning_rate": 9.97288800675814e-05, "loss": 0.8221, "step": 2432 }, { "epoch": 0.39, "learning_rate": 9.97279100638161e-05, "loss": 0.8553, "step": 2433 }, { "epoch": 0.39, "learning_rate": 9.972693833265878e-05, "loss": 0.8687, "step": 2434 }, { "epoch": 0.39, "learning_rate": 9.972596487414323e-05, "loss": 0.8048, "step": 2435 }, { "epoch": 0.39, "learning_rate": 9.972498968830322e-05, "loss": 0.838, "step": 2436 }, { "epoch": 0.39, "learning_rate": 9.972401277517265e-05, "loss": 0.8907, "step": 2437 }, { "epoch": 0.39, "learning_rate": 9.972303413478546e-05, "loss": 0.8303, "step": 2438 }, { "epoch": 0.39, "learning_rate": 9.972205376717563e-05, "loss": 0.791, "step": 2439 }, { "epoch": 0.39, "learning_rate": 9.972107167237721e-05, "loss": 0.8107, "step": 2440 }, { "epoch": 0.39, "learning_rate": 9.972008785042432e-05, "loss": 0.8303, "step": 2441 }, { "epoch": 0.39, "learning_rate": 9.971910230135116e-05, "loss": 0.905, "step": 2442 }, { "epoch": 0.39, "learning_rate": 9.971811502519194e-05, "loss": 0.7963, "step": 2443 }, { "epoch": 0.39, "learning_rate": 9.971712602198096e-05, "loss": 0.8349, "step": 2444 }, { "epoch": 0.39, "learning_rate": 9.971613529175257e-05, "loss": 0.8717, "step": 2445 }, { "epoch": 0.39, "learning_rate": 9.971514283454118e-05, "loss": 0.8668, "step": 2446 }, { "epoch": 0.39, "learning_rate": 9.971414865038128e-05, "loss": 0.7913, "step": 2447 }, { "epoch": 0.39, "learning_rate": 9.971315273930742e-05, "loss": 0.8901, "step": 2448 }, { "epoch": 0.39, "learning_rate": 9.971215510135416e-05, "loss": 0.8896, "step": 2449 }, { "epoch": 0.39, "learning_rate": 9.971115573655616e-05, "loss": 0.7852, "step": 2450 }, { "epoch": 0.39, "learning_rate": 9.971015464494815e-05, "loss": 0.8429, "step": 2451 }, { "epoch": 0.39, "learning_rate": 9.97091518265649e-05, "loss": 0.8335, "step": 2452 }, { "epoch": 0.39, "learning_rate": 9.970814728144125e-05, "loss": 0.8949, "step": 2453 }, { "epoch": 0.39, "learning_rate": 9.970714100961207e-05, "loss": 0.8045, "step": 2454 }, { "epoch": 0.39, "learning_rate": 9.970613301111235e-05, "loss": 0.7615, "step": 2455 }, { "epoch": 0.39, "learning_rate": 9.970512328597707e-05, "loss": 0.8921, "step": 2456 }, { "epoch": 0.39, "learning_rate": 9.970411183424134e-05, "loss": 0.7864, "step": 2457 }, { "epoch": 0.39, "learning_rate": 9.970309865594026e-05, "loss": 0.8828, "step": 2458 }, { "epoch": 0.39, "learning_rate": 9.970208375110905e-05, "loss": 0.8194, "step": 2459 }, { "epoch": 0.39, "learning_rate": 9.970106711978297e-05, "loss": 0.7989, "step": 2460 }, { "epoch": 0.39, "learning_rate": 9.97000487619973e-05, "loss": 0.8237, "step": 2461 }, { "epoch": 0.39, "learning_rate": 9.969902867778745e-05, "loss": 0.8803, "step": 2462 }, { "epoch": 0.39, "learning_rate": 9.969800686718882e-05, "loss": 0.8145, "step": 2463 }, { "epoch": 0.39, "learning_rate": 9.969698333023694e-05, "loss": 0.8088, "step": 2464 }, { "epoch": 0.39, "learning_rate": 9.969595806696734e-05, "loss": 0.7766, "step": 2465 }, { "epoch": 0.39, "learning_rate": 9.969493107741566e-05, "loss": 0.8156, "step": 2466 }, { "epoch": 0.4, "learning_rate": 9.969390236161754e-05, "loss": 0.7883, "step": 2467 }, { "epoch": 0.4, "learning_rate": 9.969287191960875e-05, "loss": 0.8774, "step": 2468 }, { "epoch": 0.4, "learning_rate": 9.969183975142505e-05, "loss": 0.9079, "step": 2469 }, { "epoch": 0.4, "learning_rate": 9.969080585710231e-05, "loss": 0.7634, "step": 2470 }, { "epoch": 0.4, "learning_rate": 9.968977023667646e-05, "loss": 0.8178, "step": 2471 }, { "epoch": 0.4, "learning_rate": 9.968873289018344e-05, "loss": 0.8766, "step": 2472 }, { "epoch": 0.4, "learning_rate": 9.96876938176593e-05, "loss": 0.8157, "step": 2473 }, { "epoch": 0.4, "learning_rate": 9.968665301914015e-05, "loss": 0.8347, "step": 2474 }, { "epoch": 0.4, "learning_rate": 9.968561049466214e-05, "loss": 0.8426, "step": 2475 }, { "epoch": 0.4, "learning_rate": 9.968456624426146e-05, "loss": 0.8835, "step": 2476 }, { "epoch": 0.4, "learning_rate": 9.96835202679744e-05, "loss": 0.759, "step": 2477 }, { "epoch": 0.4, "learning_rate": 9.96824725658373e-05, "loss": 0.7815, "step": 2478 }, { "epoch": 0.4, "learning_rate": 9.968142313788656e-05, "loss": 0.7385, "step": 2479 }, { "epoch": 0.4, "learning_rate": 9.96803719841586e-05, "loss": 0.8222, "step": 2480 }, { "epoch": 0.4, "learning_rate": 9.967931910468994e-05, "loss": 0.7974, "step": 2481 }, { "epoch": 0.4, "learning_rate": 9.967826449951721e-05, "loss": 0.8087, "step": 2482 }, { "epoch": 0.4, "learning_rate": 9.967720816867699e-05, "loss": 0.8638, "step": 2483 }, { "epoch": 0.4, "learning_rate": 9.967615011220596e-05, "loss": 0.9059, "step": 2484 }, { "epoch": 0.4, "learning_rate": 9.967509033014093e-05, "loss": 0.7689, "step": 2485 }, { "epoch": 0.4, "learning_rate": 9.967402882251867e-05, "loss": 0.7643, "step": 2486 }, { "epoch": 0.4, "learning_rate": 9.967296558937607e-05, "loss": 0.8771, "step": 2487 }, { "epoch": 0.4, "learning_rate": 9.967190063075005e-05, "loss": 0.8542, "step": 2488 }, { "epoch": 0.4, "learning_rate": 9.967083394667762e-05, "loss": 0.831, "step": 2489 }, { "epoch": 0.4, "learning_rate": 9.966976553719582e-05, "loss": 0.7582, "step": 2490 }, { "epoch": 0.4, "learning_rate": 9.966869540234178e-05, "loss": 0.8679, "step": 2491 }, { "epoch": 0.4, "learning_rate": 9.966762354215266e-05, "loss": 0.8408, "step": 2492 }, { "epoch": 0.4, "learning_rate": 9.96665499566657e-05, "loss": 0.9159, "step": 2493 }, { "epoch": 0.4, "learning_rate": 9.966547464591818e-05, "loss": 0.783, "step": 2494 }, { "epoch": 0.4, "learning_rate": 9.966439760994746e-05, "loss": 0.8458, "step": 2495 }, { "epoch": 0.4, "learning_rate": 9.966331884879097e-05, "loss": 0.8606, "step": 2496 }, { "epoch": 0.4, "learning_rate": 9.966223836248616e-05, "loss": 0.8681, "step": 2497 }, { "epoch": 0.4, "learning_rate": 9.966115615107057e-05, "loss": 0.8732, "step": 2498 }, { "epoch": 0.4, "learning_rate": 9.966007221458179e-05, "loss": 0.842, "step": 2499 }, { "epoch": 0.4, "learning_rate": 9.965898655305748e-05, "loss": 0.8035, "step": 2500 }, { "epoch": 0.4, "learning_rate": 9.965789916653533e-05, "loss": 0.7324, "step": 2501 }, { "epoch": 0.4, "learning_rate": 9.965681005505316e-05, "loss": 0.8232, "step": 2502 }, { "epoch": 0.4, "learning_rate": 9.965571921864874e-05, "loss": 0.8276, "step": 2503 }, { "epoch": 0.4, "learning_rate": 9.965462665736001e-05, "loss": 0.8244, "step": 2504 }, { "epoch": 0.4, "learning_rate": 9.965353237122492e-05, "loss": 0.8662, "step": 2505 }, { "epoch": 0.4, "learning_rate": 9.965243636028146e-05, "loss": 0.8465, "step": 2506 }, { "epoch": 0.4, "learning_rate": 9.96513386245677e-05, "loss": 0.7936, "step": 2507 }, { "epoch": 0.4, "learning_rate": 9.96502391641218e-05, "loss": 0.7184, "step": 2508 }, { "epoch": 0.4, "learning_rate": 9.964913797898193e-05, "loss": 0.9299, "step": 2509 }, { "epoch": 0.4, "learning_rate": 9.964803506918634e-05, "loss": 0.737, "step": 2510 }, { "epoch": 0.4, "learning_rate": 9.964693043477335e-05, "loss": 0.8292, "step": 2511 }, { "epoch": 0.4, "learning_rate": 9.964582407578134e-05, "loss": 0.8526, "step": 2512 }, { "epoch": 0.4, "learning_rate": 9.964471599224872e-05, "loss": 0.8071, "step": 2513 }, { "epoch": 0.4, "learning_rate": 9.964360618421397e-05, "loss": 0.8438, "step": 2514 }, { "epoch": 0.4, "learning_rate": 9.96424946517157e-05, "loss": 0.8141, "step": 2515 }, { "epoch": 0.4, "learning_rate": 9.964138139479247e-05, "loss": 0.792, "step": 2516 }, { "epoch": 0.4, "learning_rate": 9.964026641348298e-05, "loss": 0.8849, "step": 2517 }, { "epoch": 0.4, "learning_rate": 9.963914970782593e-05, "loss": 0.8392, "step": 2518 }, { "epoch": 0.4, "learning_rate": 9.963803127786013e-05, "loss": 0.7801, "step": 2519 }, { "epoch": 0.4, "learning_rate": 9.963691112362443e-05, "loss": 0.7571, "step": 2520 }, { "epoch": 0.4, "learning_rate": 9.963578924515776e-05, "loss": 0.8268, "step": 2521 }, { "epoch": 0.4, "learning_rate": 9.963466564249904e-05, "loss": 0.731, "step": 2522 }, { "epoch": 0.4, "learning_rate": 9.963354031568735e-05, "loss": 0.8376, "step": 2523 }, { "epoch": 0.4, "learning_rate": 9.963241326476174e-05, "loss": 0.846, "step": 2524 }, { "epoch": 0.4, "learning_rate": 9.96312844897614e-05, "loss": 0.8321, "step": 2525 }, { "epoch": 0.4, "learning_rate": 9.96301539907255e-05, "loss": 0.8441, "step": 2526 }, { "epoch": 0.4, "learning_rate": 9.962902176769335e-05, "loss": 0.8384, "step": 2527 }, { "epoch": 0.4, "learning_rate": 9.962788782070426e-05, "loss": 0.7713, "step": 2528 }, { "epoch": 0.41, "learning_rate": 9.962675214979762e-05, "loss": 0.8553, "step": 2529 }, { "epoch": 0.41, "learning_rate": 9.962561475501287e-05, "loss": 0.9902, "step": 2530 }, { "epoch": 0.41, "learning_rate": 9.962447563638953e-05, "loss": 0.7323, "step": 2531 }, { "epoch": 0.41, "learning_rate": 9.962333479396719e-05, "loss": 0.8191, "step": 2532 }, { "epoch": 0.41, "learning_rate": 9.962219222778544e-05, "loss": 0.7428, "step": 2533 }, { "epoch": 0.41, "learning_rate": 9.962104793788398e-05, "loss": 0.9158, "step": 2534 }, { "epoch": 0.41, "learning_rate": 9.961990192430259e-05, "loss": 0.8696, "step": 2535 }, { "epoch": 0.41, "learning_rate": 9.961875418708104e-05, "loss": 0.8414, "step": 2536 }, { "epoch": 0.41, "learning_rate": 9.961760472625921e-05, "loss": 0.9187, "step": 2537 }, { "epoch": 0.41, "learning_rate": 9.961645354187706e-05, "loss": 0.8379, "step": 2538 }, { "epoch": 0.41, "learning_rate": 9.961530063397452e-05, "loss": 0.7628, "step": 2539 }, { "epoch": 0.41, "learning_rate": 9.961414600259167e-05, "loss": 0.7846, "step": 2540 }, { "epoch": 0.41, "learning_rate": 9.961298964776864e-05, "loss": 0.8553, "step": 2541 }, { "epoch": 0.41, "learning_rate": 9.961183156954557e-05, "loss": 0.8679, "step": 2542 }, { "epoch": 0.41, "learning_rate": 9.961067176796268e-05, "loss": 0.8154, "step": 2543 }, { "epoch": 0.41, "learning_rate": 9.960951024306029e-05, "loss": 0.8272, "step": 2544 }, { "epoch": 0.41, "learning_rate": 9.960834699487873e-05, "loss": 0.7706, "step": 2545 }, { "epoch": 0.41, "learning_rate": 9.96071820234584e-05, "loss": 0.8119, "step": 2546 }, { "epoch": 0.41, "learning_rate": 9.960601532883977e-05, "loss": 0.9405, "step": 2547 }, { "epoch": 0.41, "learning_rate": 9.960484691106338e-05, "loss": 0.8716, "step": 2548 }, { "epoch": 0.41, "learning_rate": 9.96036767701698e-05, "loss": 0.7775, "step": 2549 }, { "epoch": 0.41, "learning_rate": 9.96025049061997e-05, "loss": 0.7878, "step": 2550 }, { "epoch": 0.41, "learning_rate": 9.960133131919377e-05, "loss": 0.8641, "step": 2551 }, { "epoch": 0.41, "learning_rate": 9.960015600919278e-05, "loss": 0.9356, "step": 2552 }, { "epoch": 0.41, "learning_rate": 9.959897897623755e-05, "loss": 0.8575, "step": 2553 }, { "epoch": 0.41, "learning_rate": 9.959780022036899e-05, "loss": 0.7712, "step": 2554 }, { "epoch": 0.41, "learning_rate": 9.959661974162801e-05, "loss": 0.8481, "step": 2555 }, { "epoch": 0.41, "learning_rate": 9.959543754005564e-05, "loss": 0.7635, "step": 2556 }, { "epoch": 0.41, "learning_rate": 9.959425361569295e-05, "loss": 0.8218, "step": 2557 }, { "epoch": 0.41, "learning_rate": 9.959306796858106e-05, "loss": 0.8055, "step": 2558 }, { "epoch": 0.41, "learning_rate": 9.959188059876115e-05, "loss": 0.7734, "step": 2559 }, { "epoch": 0.41, "learning_rate": 9.959069150627446e-05, "loss": 0.8202, "step": 2560 }, { "epoch": 0.41, "learning_rate": 9.95895006911623e-05, "loss": 0.749, "step": 2561 }, { "epoch": 0.41, "learning_rate": 9.958830815346606e-05, "loss": 0.819, "step": 2562 }, { "epoch": 0.41, "learning_rate": 9.958711389322713e-05, "loss": 0.8482, "step": 2563 }, { "epoch": 0.41, "learning_rate": 9.958591791048702e-05, "loss": 0.8554, "step": 2564 }, { "epoch": 0.41, "learning_rate": 9.958472020528725e-05, "loss": 0.7964, "step": 2565 }, { "epoch": 0.41, "learning_rate": 9.958352077766945e-05, "loss": 0.8752, "step": 2566 }, { "epoch": 0.41, "learning_rate": 9.958231962767526e-05, "loss": 0.8635, "step": 2567 }, { "epoch": 0.41, "learning_rate": 9.958111675534642e-05, "loss": 0.7866, "step": 2568 }, { "epoch": 0.41, "learning_rate": 9.957991216072474e-05, "loss": 0.8642, "step": 2569 }, { "epoch": 0.41, "learning_rate": 9.9578705843852e-05, "loss": 0.8037, "step": 2570 }, { "epoch": 0.41, "learning_rate": 9.957749780477014e-05, "loss": 0.8484, "step": 2571 }, { "epoch": 0.41, "learning_rate": 9.957628804352112e-05, "loss": 0.801, "step": 2572 }, { "epoch": 0.41, "learning_rate": 9.9575076560147e-05, "loss": 0.8627, "step": 2573 }, { "epoch": 0.41, "learning_rate": 9.957386335468979e-05, "loss": 0.8854, "step": 2574 }, { "epoch": 0.41, "learning_rate": 9.957264842719169e-05, "loss": 0.7272, "step": 2575 }, { "epoch": 0.41, "learning_rate": 9.957143177769488e-05, "loss": 0.9092, "step": 2576 }, { "epoch": 0.41, "learning_rate": 9.957021340624164e-05, "loss": 0.754, "step": 2577 }, { "epoch": 0.41, "learning_rate": 9.956899331287426e-05, "loss": 0.8172, "step": 2578 }, { "epoch": 0.41, "learning_rate": 9.956777149763515e-05, "loss": 0.7655, "step": 2579 }, { "epoch": 0.41, "learning_rate": 9.956654796056673e-05, "loss": 0.756, "step": 2580 }, { "epoch": 0.41, "learning_rate": 9.956532270171153e-05, "loss": 0.7882, "step": 2581 }, { "epoch": 0.41, "learning_rate": 9.956409572111209e-05, "loss": 0.8193, "step": 2582 }, { "epoch": 0.41, "learning_rate": 9.956286701881104e-05, "loss": 0.8076, "step": 2583 }, { "epoch": 0.41, "learning_rate": 9.956163659485107e-05, "loss": 0.8476, "step": 2584 }, { "epoch": 0.41, "learning_rate": 9.956040444927491e-05, "loss": 0.8648, "step": 2585 }, { "epoch": 0.41, "learning_rate": 9.955917058212534e-05, "loss": 0.8534, "step": 2586 }, { "epoch": 0.41, "learning_rate": 9.955793499344525e-05, "loss": 0.7631, "step": 2587 }, { "epoch": 0.41, "learning_rate": 9.955669768327757e-05, "loss": 0.8219, "step": 2588 }, { "epoch": 0.41, "learning_rate": 9.955545865166526e-05, "loss": 0.818, "step": 2589 }, { "epoch": 0.41, "learning_rate": 9.955421789865135e-05, "loss": 0.8351, "step": 2590 }, { "epoch": 0.41, "learning_rate": 9.955297542427896e-05, "loss": 0.8768, "step": 2591 }, { "epoch": 0.42, "learning_rate": 9.955173122859125e-05, "loss": 0.8068, "step": 2592 }, { "epoch": 0.42, "learning_rate": 9.955048531163141e-05, "loss": 0.9644, "step": 2593 }, { "epoch": 0.42, "learning_rate": 9.954923767344277e-05, "loss": 0.7652, "step": 2594 }, { "epoch": 0.42, "learning_rate": 9.954798831406862e-05, "loss": 0.8051, "step": 2595 }, { "epoch": 0.42, "learning_rate": 9.954673723355239e-05, "loss": 0.7989, "step": 2596 }, { "epoch": 0.42, "learning_rate": 9.954548443193751e-05, "loss": 0.8328, "step": 2597 }, { "epoch": 0.42, "learning_rate": 9.954422990926754e-05, "loss": 0.8684, "step": 2598 }, { "epoch": 0.42, "learning_rate": 9.954297366558603e-05, "loss": 0.8064, "step": 2599 }, { "epoch": 0.42, "learning_rate": 9.954171570093661e-05, "loss": 0.8631, "step": 2600 }, { "epoch": 0.42, "learning_rate": 9.954045601536299e-05, "loss": 0.8093, "step": 2601 }, { "epoch": 0.42, "learning_rate": 9.953919460890895e-05, "loss": 0.8395, "step": 2602 }, { "epoch": 0.42, "learning_rate": 9.953793148161826e-05, "loss": 0.839, "step": 2603 }, { "epoch": 0.42, "learning_rate": 9.953666663353484e-05, "loss": 0.8418, "step": 2604 }, { "epoch": 0.42, "learning_rate": 9.95354000647026e-05, "loss": 0.7438, "step": 2605 }, { "epoch": 0.42, "learning_rate": 9.953413177516555e-05, "loss": 0.9166, "step": 2606 }, { "epoch": 0.42, "learning_rate": 9.953286176496773e-05, "loss": 0.8182, "step": 2607 }, { "epoch": 0.42, "learning_rate": 9.953159003415328e-05, "loss": 0.8478, "step": 2608 }, { "epoch": 0.42, "learning_rate": 9.953031658276637e-05, "loss": 0.8513, "step": 2609 }, { "epoch": 0.42, "learning_rate": 9.952904141085122e-05, "loss": 0.8216, "step": 2610 }, { "epoch": 0.42, "learning_rate": 9.952776451845214e-05, "loss": 0.851, "step": 2611 }, { "epoch": 0.42, "learning_rate": 9.952648590561347e-05, "loss": 0.8252, "step": 2612 }, { "epoch": 0.42, "learning_rate": 9.952520557237966e-05, "loss": 0.793, "step": 2613 }, { "epoch": 0.42, "learning_rate": 9.952392351879513e-05, "loss": 0.8308, "step": 2614 }, { "epoch": 0.42, "learning_rate": 9.952263974490447e-05, "loss": 0.8157, "step": 2615 }, { "epoch": 0.42, "learning_rate": 9.952135425075224e-05, "loss": 0.8882, "step": 2616 }, { "epoch": 0.42, "learning_rate": 9.95200670363831e-05, "loss": 0.889, "step": 2617 }, { "epoch": 0.42, "learning_rate": 9.951877810184177e-05, "loss": 0.9591, "step": 2618 }, { "epoch": 0.42, "learning_rate": 9.951748744717301e-05, "loss": 0.7985, "step": 2619 }, { "epoch": 0.42, "learning_rate": 9.951619507242169e-05, "loss": 0.8473, "step": 2620 }, { "epoch": 0.42, "learning_rate": 9.951490097763265e-05, "loss": 0.8437, "step": 2621 }, { "epoch": 0.42, "learning_rate": 9.951360516285088e-05, "loss": 0.7672, "step": 2622 }, { "epoch": 0.42, "learning_rate": 9.95123076281214e-05, "loss": 0.7603, "step": 2623 }, { "epoch": 0.42, "learning_rate": 9.951100837348924e-05, "loss": 0.8249, "step": 2624 }, { "epoch": 0.42, "learning_rate": 9.950970739899958e-05, "loss": 0.7876, "step": 2625 }, { "epoch": 0.42, "learning_rate": 9.950840470469757e-05, "loss": 0.8644, "step": 2626 }, { "epoch": 0.42, "learning_rate": 9.950710029062848e-05, "loss": 0.757, "step": 2627 }, { "epoch": 0.42, "learning_rate": 9.950579415683763e-05, "loss": 0.7669, "step": 2628 }, { "epoch": 0.42, "learning_rate": 9.950448630337036e-05, "loss": 0.7823, "step": 2629 }, { "epoch": 0.42, "learning_rate": 9.950317673027217e-05, "loss": 0.8732, "step": 2630 }, { "epoch": 0.42, "learning_rate": 9.950186543758846e-05, "loss": 0.816, "step": 2631 }, { "epoch": 0.42, "learning_rate": 9.950055242536484e-05, "loss": 0.7932, "step": 2632 }, { "epoch": 0.42, "learning_rate": 9.94992376936469e-05, "loss": 0.7907, "step": 2633 }, { "epoch": 0.42, "learning_rate": 9.949792124248032e-05, "loss": 0.8469, "step": 2634 }, { "epoch": 0.42, "learning_rate": 9.949660307191082e-05, "loss": 0.903, "step": 2635 }, { "epoch": 0.42, "learning_rate": 9.94952831819842e-05, "loss": 0.8445, "step": 2636 }, { "epoch": 0.42, "learning_rate": 9.949396157274629e-05, "loss": 0.7933, "step": 2637 }, { "epoch": 0.42, "learning_rate": 9.949263824424301e-05, "loss": 0.7655, "step": 2638 }, { "epoch": 0.42, "learning_rate": 9.949131319652033e-05, "loss": 0.8572, "step": 2639 }, { "epoch": 0.42, "learning_rate": 9.948998642962427e-05, "loss": 0.8822, "step": 2640 }, { "epoch": 0.42, "learning_rate": 9.948865794360093e-05, "loss": 0.8111, "step": 2641 }, { "epoch": 0.42, "learning_rate": 9.948732773849645e-05, "loss": 0.8378, "step": 2642 }, { "epoch": 0.42, "learning_rate": 9.948599581435704e-05, "loss": 0.8432, "step": 2643 }, { "epoch": 0.42, "learning_rate": 9.948466217122895e-05, "loss": 0.8287, "step": 2644 }, { "epoch": 0.42, "learning_rate": 9.948332680915855e-05, "loss": 0.8596, "step": 2645 }, { "epoch": 0.42, "learning_rate": 9.948198972819218e-05, "loss": 0.7736, "step": 2646 }, { "epoch": 0.42, "learning_rate": 9.948065092837631e-05, "loss": 0.8561, "step": 2647 }, { "epoch": 0.42, "learning_rate": 9.947931040975741e-05, "loss": 0.8983, "step": 2648 }, { "epoch": 0.42, "learning_rate": 9.94779681723821e-05, "loss": 0.7696, "step": 2649 }, { "epoch": 0.42, "learning_rate": 9.9476624216297e-05, "loss": 0.9349, "step": 2650 }, { "epoch": 0.42, "learning_rate": 9.947527854154875e-05, "loss": 0.7596, "step": 2651 }, { "epoch": 0.42, "learning_rate": 9.947393114818411e-05, "loss": 0.878, "step": 2652 }, { "epoch": 0.42, "learning_rate": 9.947258203624991e-05, "loss": 0.9014, "step": 2653 }, { "epoch": 0.43, "learning_rate": 9.9471231205793e-05, "loss": 0.7899, "step": 2654 }, { "epoch": 0.43, "learning_rate": 9.94698786568603e-05, "loss": 0.8959, "step": 2655 }, { "epoch": 0.43, "learning_rate": 9.946852438949879e-05, "loss": 0.9461, "step": 2656 }, { "epoch": 0.43, "learning_rate": 9.946716840375551e-05, "loss": 0.8281, "step": 2657 }, { "epoch": 0.43, "learning_rate": 9.946581069967758e-05, "loss": 0.7859, "step": 2658 }, { "epoch": 0.43, "learning_rate": 9.946445127731216e-05, "loss": 0.9692, "step": 2659 }, { "epoch": 0.43, "learning_rate": 9.946309013670645e-05, "loss": 0.8785, "step": 2660 }, { "epoch": 0.43, "learning_rate": 9.946172727790774e-05, "loss": 0.8976, "step": 2661 }, { "epoch": 0.43, "learning_rate": 9.946036270096339e-05, "loss": 0.8464, "step": 2662 }, { "epoch": 0.43, "learning_rate": 9.945899640592081e-05, "loss": 0.7384, "step": 2663 }, { "epoch": 0.43, "learning_rate": 9.94576283928274e-05, "loss": 0.7453, "step": 2664 }, { "epoch": 0.43, "learning_rate": 9.945625866173074e-05, "loss": 0.8131, "step": 2665 }, { "epoch": 0.43, "learning_rate": 9.945488721267839e-05, "loss": 0.7092, "step": 2666 }, { "epoch": 0.43, "learning_rate": 9.945351404571799e-05, "loss": 0.8165, "step": 2667 }, { "epoch": 0.43, "learning_rate": 9.945213916089724e-05, "loss": 0.8492, "step": 2668 }, { "epoch": 0.43, "learning_rate": 9.945076255826391e-05, "loss": 0.8155, "step": 2669 }, { "epoch": 0.43, "learning_rate": 9.944938423786578e-05, "loss": 0.886, "step": 2670 }, { "epoch": 0.43, "learning_rate": 9.944800419975078e-05, "loss": 0.88, "step": 2671 }, { "epoch": 0.43, "learning_rate": 9.944662244396682e-05, "loss": 0.905, "step": 2672 }, { "epoch": 0.43, "learning_rate": 9.94452389705619e-05, "loss": 0.8295, "step": 2673 }, { "epoch": 0.43, "learning_rate": 9.944385377958409e-05, "loss": 0.7863, "step": 2674 }, { "epoch": 0.43, "learning_rate": 9.944246687108147e-05, "loss": 0.9269, "step": 2675 }, { "epoch": 0.43, "learning_rate": 9.944107824510228e-05, "loss": 0.7488, "step": 2676 }, { "epoch": 0.43, "learning_rate": 9.94396879016947e-05, "loss": 0.8978, "step": 2677 }, { "epoch": 0.43, "learning_rate": 9.943829584090705e-05, "loss": 0.8803, "step": 2678 }, { "epoch": 0.43, "learning_rate": 9.943690206278767e-05, "loss": 0.8362, "step": 2679 }, { "epoch": 0.43, "learning_rate": 9.9435506567385e-05, "loss": 0.7771, "step": 2680 }, { "epoch": 0.43, "learning_rate": 9.94341093547475e-05, "loss": 0.8483, "step": 2681 }, { "epoch": 0.43, "learning_rate": 9.943271042492373e-05, "loss": 0.8371, "step": 2682 }, { "epoch": 0.43, "learning_rate": 9.943130977796224e-05, "loss": 0.8757, "step": 2683 }, { "epoch": 0.43, "learning_rate": 9.942990741391171e-05, "loss": 0.7119, "step": 2684 }, { "epoch": 0.43, "learning_rate": 9.942850333282085e-05, "loss": 0.9243, "step": 2685 }, { "epoch": 0.43, "learning_rate": 9.942709753473843e-05, "loss": 0.8891, "step": 2686 }, { "epoch": 0.43, "learning_rate": 9.942569001971329e-05, "loss": 0.7965, "step": 2687 }, { "epoch": 0.43, "learning_rate": 9.942428078779433e-05, "loss": 0.8092, "step": 2688 }, { "epoch": 0.43, "learning_rate": 9.942286983903049e-05, "loss": 0.8949, "step": 2689 }, { "epoch": 0.43, "learning_rate": 9.942145717347076e-05, "loss": 0.7596, "step": 2690 }, { "epoch": 0.43, "learning_rate": 9.942004279116426e-05, "loss": 0.7681, "step": 2691 }, { "epoch": 0.43, "learning_rate": 9.941862669216008e-05, "loss": 0.7989, "step": 2692 }, { "epoch": 0.43, "learning_rate": 9.941720887650745e-05, "loss": 0.837, "step": 2693 }, { "epoch": 0.43, "learning_rate": 9.941578934425558e-05, "loss": 0.8658, "step": 2694 }, { "epoch": 0.43, "learning_rate": 9.94143680954538e-05, "loss": 0.8347, "step": 2695 }, { "epoch": 0.43, "learning_rate": 9.941294513015149e-05, "loss": 0.7866, "step": 2696 }, { "epoch": 0.43, "learning_rate": 9.941152044839806e-05, "loss": 0.8473, "step": 2697 }, { "epoch": 0.43, "learning_rate": 9.9410094050243e-05, "loss": 0.77, "step": 2698 }, { "epoch": 0.43, "learning_rate": 9.940866593573586e-05, "loss": 0.9243, "step": 2699 }, { "epoch": 0.43, "learning_rate": 9.940723610492626e-05, "loss": 0.7333, "step": 2700 }, { "epoch": 0.43, "learning_rate": 9.940580455786385e-05, "loss": 0.7878, "step": 2701 }, { "epoch": 0.43, "learning_rate": 9.940437129459839e-05, "loss": 0.8348, "step": 2702 }, { "epoch": 0.43, "learning_rate": 9.940293631517964e-05, "loss": 0.7565, "step": 2703 }, { "epoch": 0.43, "learning_rate": 9.940149961965745e-05, "loss": 0.778, "step": 2704 }, { "epoch": 0.43, "learning_rate": 9.940006120808172e-05, "loss": 0.8672, "step": 2705 }, { "epoch": 0.43, "learning_rate": 9.939862108050243e-05, "loss": 0.8845, "step": 2706 }, { "epoch": 0.43, "learning_rate": 9.939717923696961e-05, "loss": 0.8448, "step": 2707 }, { "epoch": 0.43, "learning_rate": 9.93957356775333e-05, "loss": 0.7547, "step": 2708 }, { "epoch": 0.43, "learning_rate": 9.939429040224372e-05, "loss": 0.8637, "step": 2709 }, { "epoch": 0.43, "learning_rate": 9.939284341115102e-05, "loss": 0.7487, "step": 2710 }, { "epoch": 0.43, "learning_rate": 9.939139470430548e-05, "loss": 0.8386, "step": 2711 }, { "epoch": 0.43, "learning_rate": 9.938994428175741e-05, "loss": 0.8138, "step": 2712 }, { "epoch": 0.43, "learning_rate": 9.938849214355721e-05, "loss": 0.8125, "step": 2713 }, { "epoch": 0.43, "learning_rate": 9.938703828975532e-05, "loss": 0.8492, "step": 2714 }, { "epoch": 0.43, "learning_rate": 9.938558272040224e-05, "loss": 0.8437, "step": 2715 }, { "epoch": 0.43, "learning_rate": 9.938412543554853e-05, "loss": 0.7316, "step": 2716 }, { "epoch": 0.44, "learning_rate": 9.93826664352448e-05, "loss": 0.7869, "step": 2717 }, { "epoch": 0.44, "learning_rate": 9.938120571954177e-05, "loss": 0.8262, "step": 2718 }, { "epoch": 0.44, "learning_rate": 9.937974328849014e-05, "loss": 0.782, "step": 2719 }, { "epoch": 0.44, "learning_rate": 9.937827914214073e-05, "loss": 0.8017, "step": 2720 }, { "epoch": 0.44, "learning_rate": 9.93768132805444e-05, "loss": 0.8009, "step": 2721 }, { "epoch": 0.44, "learning_rate": 9.937534570375205e-05, "loss": 0.7657, "step": 2722 }, { "epoch": 0.44, "learning_rate": 9.937387641181469e-05, "loss": 0.7651, "step": 2723 }, { "epoch": 0.44, "learning_rate": 9.937240540478332e-05, "loss": 0.8564, "step": 2724 }, { "epoch": 0.44, "learning_rate": 9.937093268270908e-05, "loss": 0.8604, "step": 2725 }, { "epoch": 0.44, "learning_rate": 9.93694582456431e-05, "loss": 0.8413, "step": 2726 }, { "epoch": 0.44, "learning_rate": 9.93679820936366e-05, "loss": 0.7581, "step": 2727 }, { "epoch": 0.44, "learning_rate": 9.936650422674087e-05, "loss": 0.7683, "step": 2728 }, { "epoch": 0.44, "learning_rate": 9.936502464500723e-05, "loss": 0.8571, "step": 2729 }, { "epoch": 0.44, "learning_rate": 9.93635433484871e-05, "loss": 0.7715, "step": 2730 }, { "epoch": 0.44, "learning_rate": 9.936206033723192e-05, "loss": 0.8874, "step": 2731 }, { "epoch": 0.44, "learning_rate": 9.936057561129319e-05, "loss": 0.8714, "step": 2732 }, { "epoch": 0.44, "learning_rate": 9.935908917072252e-05, "loss": 0.8566, "step": 2733 }, { "epoch": 0.44, "learning_rate": 9.93576010155715e-05, "loss": 0.7612, "step": 2734 }, { "epoch": 0.44, "learning_rate": 9.935611114589188e-05, "loss": 0.8447, "step": 2735 }, { "epoch": 0.44, "learning_rate": 9.935461956173536e-05, "loss": 0.7272, "step": 2736 }, { "epoch": 0.44, "learning_rate": 9.935312626315376e-05, "loss": 0.8654, "step": 2737 }, { "epoch": 0.44, "learning_rate": 9.9351631250199e-05, "loss": 0.8022, "step": 2738 }, { "epoch": 0.44, "learning_rate": 9.935013452292297e-05, "loss": 0.8301, "step": 2739 }, { "epoch": 0.44, "learning_rate": 9.934863608137766e-05, "loss": 0.7559, "step": 2740 }, { "epoch": 0.44, "learning_rate": 9.934713592561513e-05, "loss": 0.8879, "step": 2741 }, { "epoch": 0.44, "learning_rate": 9.93456340556875e-05, "loss": 0.7399, "step": 2742 }, { "epoch": 0.44, "learning_rate": 9.934413047164694e-05, "loss": 0.8456, "step": 2743 }, { "epoch": 0.44, "learning_rate": 9.934262517354566e-05, "loss": 0.7633, "step": 2744 }, { "epoch": 0.44, "learning_rate": 9.934111816143598e-05, "loss": 0.7676, "step": 2745 }, { "epoch": 0.44, "learning_rate": 9.933960943537022e-05, "loss": 0.7641, "step": 2746 }, { "epoch": 0.44, "learning_rate": 9.93380989954008e-05, "loss": 0.8048, "step": 2747 }, { "epoch": 0.44, "learning_rate": 9.933658684158018e-05, "loss": 0.811, "step": 2748 }, { "epoch": 0.44, "learning_rate": 9.933507297396092e-05, "loss": 0.8066, "step": 2749 }, { "epoch": 0.44, "learning_rate": 9.933355739259556e-05, "loss": 0.7789, "step": 2750 }, { "epoch": 0.44, "learning_rate": 9.933204009753677e-05, "loss": 0.7827, "step": 2751 }, { "epoch": 0.44, "learning_rate": 9.933052108883727e-05, "loss": 0.7817, "step": 2752 }, { "epoch": 0.44, "learning_rate": 9.932900036654979e-05, "loss": 0.8077, "step": 2753 }, { "epoch": 0.44, "learning_rate": 9.932747793072719e-05, "loss": 0.8102, "step": 2754 }, { "epoch": 0.44, "learning_rate": 9.932595378142233e-05, "loss": 0.8057, "step": 2755 }, { "epoch": 0.44, "learning_rate": 9.932442791868817e-05, "loss": 0.8264, "step": 2756 }, { "epoch": 0.44, "learning_rate": 9.932290034257772e-05, "loss": 0.8268, "step": 2757 }, { "epoch": 0.44, "learning_rate": 9.932137105314401e-05, "loss": 0.836, "step": 2758 }, { "epoch": 0.44, "learning_rate": 9.93198400504402e-05, "loss": 0.8825, "step": 2759 }, { "epoch": 0.44, "learning_rate": 9.931830733451945e-05, "loss": 0.9561, "step": 2760 }, { "epoch": 0.44, "learning_rate": 9.9316772905435e-05, "loss": 0.8564, "step": 2761 }, { "epoch": 0.44, "learning_rate": 9.931523676324016e-05, "loss": 0.7641, "step": 2762 }, { "epoch": 0.44, "learning_rate": 9.931369890798829e-05, "loss": 0.8372, "step": 2763 }, { "epoch": 0.44, "learning_rate": 9.931215933973284e-05, "loss": 0.8929, "step": 2764 }, { "epoch": 0.44, "learning_rate": 9.931061805852723e-05, "loss": 0.8451, "step": 2765 }, { "epoch": 0.44, "learning_rate": 9.930907506442503e-05, "loss": 0.8359, "step": 2766 }, { "epoch": 0.44, "learning_rate": 9.930753035747985e-05, "loss": 0.7678, "step": 2767 }, { "epoch": 0.44, "learning_rate": 9.930598393774533e-05, "loss": 0.809, "step": 2768 }, { "epoch": 0.44, "learning_rate": 9.930443580527519e-05, "loss": 0.7923, "step": 2769 }, { "epoch": 0.44, "learning_rate": 9.930288596012322e-05, "loss": 0.8183, "step": 2770 }, { "epoch": 0.44, "learning_rate": 9.930133440234325e-05, "loss": 0.8067, "step": 2771 }, { "epoch": 0.44, "learning_rate": 9.929978113198917e-05, "loss": 0.7797, "step": 2772 }, { "epoch": 0.44, "learning_rate": 9.929822614911493e-05, "loss": 0.9066, "step": 2773 }, { "epoch": 0.44, "learning_rate": 9.929666945377457e-05, "loss": 0.8895, "step": 2774 }, { "epoch": 0.44, "learning_rate": 9.929511104602215e-05, "loss": 0.7955, "step": 2775 }, { "epoch": 0.44, "learning_rate": 9.92935509259118e-05, "loss": 0.7723, "step": 2776 }, { "epoch": 0.44, "learning_rate": 9.929198909349772e-05, "loss": 0.8161, "step": 2777 }, { "epoch": 0.44, "learning_rate": 9.929042554883416e-05, "loss": 0.8307, "step": 2778 }, { "epoch": 0.45, "learning_rate": 9.928886029197544e-05, "loss": 0.7965, "step": 2779 }, { "epoch": 0.45, "learning_rate": 9.928729332297591e-05, "loss": 0.7736, "step": 2780 }, { "epoch": 0.45, "learning_rate": 9.928572464189003e-05, "loss": 0.8854, "step": 2781 }, { "epoch": 0.45, "learning_rate": 9.928415424877228e-05, "loss": 0.7703, "step": 2782 }, { "epoch": 0.45, "learning_rate": 9.92825821436772e-05, "loss": 0.9248, "step": 2783 }, { "epoch": 0.45, "learning_rate": 9.92810083266594e-05, "loss": 0.8449, "step": 2784 }, { "epoch": 0.45, "learning_rate": 9.927943279777358e-05, "loss": 0.8289, "step": 2785 }, { "epoch": 0.45, "learning_rate": 9.927785555707445e-05, "loss": 0.775, "step": 2786 }, { "epoch": 0.45, "learning_rate": 9.927627660461679e-05, "loss": 0.8291, "step": 2787 }, { "epoch": 0.45, "learning_rate": 9.927469594045545e-05, "loss": 0.8151, "step": 2788 }, { "epoch": 0.45, "learning_rate": 9.927311356464534e-05, "loss": 0.8431, "step": 2789 }, { "epoch": 0.45, "learning_rate": 9.927152947724143e-05, "loss": 0.896, "step": 2790 }, { "epoch": 0.45, "learning_rate": 9.926994367829875e-05, "loss": 0.8365, "step": 2791 }, { "epoch": 0.45, "learning_rate": 9.926835616787238e-05, "loss": 0.7915, "step": 2792 }, { "epoch": 0.45, "learning_rate": 9.926676694601746e-05, "loss": 0.7815, "step": 2793 }, { "epoch": 0.45, "learning_rate": 9.92651760127892e-05, "loss": 0.7724, "step": 2794 }, { "epoch": 0.45, "learning_rate": 9.926358336824286e-05, "loss": 0.8227, "step": 2795 }, { "epoch": 0.45, "learning_rate": 9.926198901243378e-05, "loss": 0.7468, "step": 2796 }, { "epoch": 0.45, "learning_rate": 9.926039294541733e-05, "loss": 0.9431, "step": 2797 }, { "epoch": 0.45, "learning_rate": 9.925879516724894e-05, "loss": 0.7607, "step": 2798 }, { "epoch": 0.45, "learning_rate": 9.925719567798413e-05, "loss": 0.7528, "step": 2799 }, { "epoch": 0.45, "learning_rate": 9.925559447767845e-05, "loss": 0.9014, "step": 2800 }, { "epoch": 0.45, "learning_rate": 9.925399156638754e-05, "loss": 0.8479, "step": 2801 }, { "epoch": 0.45, "learning_rate": 9.925238694416705e-05, "loss": 0.813, "step": 2802 }, { "epoch": 0.45, "learning_rate": 9.925078061107276e-05, "loss": 0.8025, "step": 2803 }, { "epoch": 0.45, "learning_rate": 9.924917256716042e-05, "loss": 0.8619, "step": 2804 }, { "epoch": 0.45, "learning_rate": 9.924756281248593e-05, "loss": 0.8667, "step": 2805 }, { "epoch": 0.45, "learning_rate": 9.924595134710517e-05, "loss": 0.8541, "step": 2806 }, { "epoch": 0.45, "learning_rate": 9.924433817107417e-05, "loss": 0.8534, "step": 2807 }, { "epoch": 0.45, "learning_rate": 9.924272328444892e-05, "loss": 0.7496, "step": 2808 }, { "epoch": 0.45, "learning_rate": 9.924110668728552e-05, "loss": 0.8141, "step": 2809 }, { "epoch": 0.45, "learning_rate": 9.923948837964014e-05, "loss": 0.8517, "step": 2810 }, { "epoch": 0.45, "learning_rate": 9.923786836156901e-05, "loss": 0.8972, "step": 2811 }, { "epoch": 0.45, "learning_rate": 9.923624663312837e-05, "loss": 0.836, "step": 2812 }, { "epoch": 0.45, "learning_rate": 9.923462319437458e-05, "loss": 0.8411, "step": 2813 }, { "epoch": 0.45, "learning_rate": 9.9232998045364e-05, "loss": 0.8344, "step": 2814 }, { "epoch": 0.45, "learning_rate": 9.923137118615313e-05, "loss": 0.817, "step": 2815 }, { "epoch": 0.45, "learning_rate": 9.922974261679847e-05, "loss": 0.8802, "step": 2816 }, { "epoch": 0.45, "learning_rate": 9.922811233735656e-05, "loss": 0.8628, "step": 2817 }, { "epoch": 0.45, "learning_rate": 9.922648034788404e-05, "loss": 0.8233, "step": 2818 }, { "epoch": 0.45, "learning_rate": 9.922484664843763e-05, "loss": 0.8614, "step": 2819 }, { "epoch": 0.45, "learning_rate": 9.922321123907405e-05, "loss": 0.8147, "step": 2820 }, { "epoch": 0.45, "learning_rate": 9.922157411985011e-05, "loss": 0.8141, "step": 2821 }, { "epoch": 0.45, "learning_rate": 9.92199352908227e-05, "loss": 0.8529, "step": 2822 }, { "epoch": 0.45, "learning_rate": 9.921829475204873e-05, "loss": 0.7996, "step": 2823 }, { "epoch": 0.45, "learning_rate": 9.92166525035852e-05, "loss": 0.7722, "step": 2824 }, { "epoch": 0.45, "learning_rate": 9.921500854548915e-05, "loss": 0.8068, "step": 2825 }, { "epoch": 0.45, "learning_rate": 9.921336287781768e-05, "loss": 0.8256, "step": 2826 }, { "epoch": 0.45, "learning_rate": 9.921171550062795e-05, "loss": 0.8288, "step": 2827 }, { "epoch": 0.45, "learning_rate": 9.92100664139772e-05, "loss": 0.8603, "step": 2828 }, { "epoch": 0.45, "learning_rate": 9.920841561792272e-05, "loss": 0.7951, "step": 2829 }, { "epoch": 0.45, "learning_rate": 9.920676311252183e-05, "loss": 0.7637, "step": 2830 }, { "epoch": 0.45, "learning_rate": 9.920510889783195e-05, "loss": 0.8324, "step": 2831 }, { "epoch": 0.45, "learning_rate": 9.920345297391054e-05, "loss": 0.7572, "step": 2832 }, { "epoch": 0.45, "learning_rate": 9.920179534081511e-05, "loss": 0.7969, "step": 2833 }, { "epoch": 0.45, "learning_rate": 9.920013599860326e-05, "loss": 0.8153, "step": 2834 }, { "epoch": 0.45, "learning_rate": 9.919847494733261e-05, "loss": 0.8619, "step": 2835 }, { "epoch": 0.45, "learning_rate": 9.919681218706088e-05, "loss": 0.7846, "step": 2836 }, { "epoch": 0.45, "learning_rate": 9.919514771784581e-05, "loss": 0.7601, "step": 2837 }, { "epoch": 0.45, "learning_rate": 9.919348153974524e-05, "loss": 0.753, "step": 2838 }, { "epoch": 0.45, "learning_rate": 9.919181365281703e-05, "loss": 0.8689, "step": 2839 }, { "epoch": 0.45, "learning_rate": 9.919014405711913e-05, "loss": 0.8844, "step": 2840 }, { "epoch": 0.46, "learning_rate": 9.918847275270951e-05, "loss": 0.7811, "step": 2841 }, { "epoch": 0.46, "learning_rate": 9.918679973964626e-05, "loss": 0.7818, "step": 2842 }, { "epoch": 0.46, "learning_rate": 9.918512501798748e-05, "loss": 0.7409, "step": 2843 }, { "epoch": 0.46, "learning_rate": 9.918344858779135e-05, "loss": 0.8744, "step": 2844 }, { "epoch": 0.46, "learning_rate": 9.918177044911608e-05, "loss": 0.8072, "step": 2845 }, { "epoch": 0.46, "learning_rate": 9.918009060202001e-05, "loss": 0.8249, "step": 2846 }, { "epoch": 0.46, "learning_rate": 9.917840904656144e-05, "loss": 0.8391, "step": 2847 }, { "epoch": 0.46, "learning_rate": 9.917672578279882e-05, "loss": 0.7434, "step": 2848 }, { "epoch": 0.46, "learning_rate": 9.917504081079059e-05, "loss": 0.8205, "step": 2849 }, { "epoch": 0.46, "learning_rate": 9.917335413059533e-05, "loss": 0.8701, "step": 2850 }, { "epoch": 0.46, "learning_rate": 9.917166574227158e-05, "loss": 0.7967, "step": 2851 }, { "epoch": 0.46, "learning_rate": 9.9169975645878e-05, "loss": 0.8893, "step": 2852 }, { "epoch": 0.46, "learning_rate": 9.916828384147331e-05, "loss": 0.8205, "step": 2853 }, { "epoch": 0.46, "learning_rate": 9.916659032911628e-05, "loss": 0.8318, "step": 2854 }, { "epoch": 0.46, "learning_rate": 9.916489510886573e-05, "loss": 0.8225, "step": 2855 }, { "epoch": 0.46, "learning_rate": 9.916319818078054e-05, "loss": 0.8814, "step": 2856 }, { "epoch": 0.46, "learning_rate": 9.916149954491969e-05, "loss": 0.8357, "step": 2857 }, { "epoch": 0.46, "learning_rate": 9.915979920134215e-05, "loss": 0.8507, "step": 2858 }, { "epoch": 0.46, "learning_rate": 9.915809715010698e-05, "loss": 0.804, "step": 2859 }, { "epoch": 0.46, "learning_rate": 9.915639339127334e-05, "loss": 0.8571, "step": 2860 }, { "epoch": 0.46, "learning_rate": 9.915468792490037e-05, "loss": 0.9346, "step": 2861 }, { "epoch": 0.46, "learning_rate": 9.915298075104735e-05, "loss": 0.8422, "step": 2862 }, { "epoch": 0.46, "learning_rate": 9.915127186977356e-05, "loss": 0.7478, "step": 2863 }, { "epoch": 0.46, "learning_rate": 9.914956128113837e-05, "loss": 0.7753, "step": 2864 }, { "epoch": 0.46, "learning_rate": 9.91478489852012e-05, "loss": 0.8519, "step": 2865 }, { "epoch": 0.46, "learning_rate": 9.914613498202152e-05, "loss": 0.7397, "step": 2866 }, { "epoch": 0.46, "learning_rate": 9.914441927165888e-05, "loss": 0.7828, "step": 2867 }, { "epoch": 0.46, "learning_rate": 9.914270185417288e-05, "loss": 0.7905, "step": 2868 }, { "epoch": 0.46, "learning_rate": 9.914098272962317e-05, "loss": 0.7782, "step": 2869 }, { "epoch": 0.46, "learning_rate": 9.913926189806948e-05, "loss": 0.7947, "step": 2870 }, { "epoch": 0.46, "learning_rate": 9.913753935957156e-05, "loss": 0.8, "step": 2871 }, { "epoch": 0.46, "learning_rate": 9.913581511418926e-05, "loss": 0.8096, "step": 2872 }, { "epoch": 0.46, "learning_rate": 9.913408916198247e-05, "loss": 0.8527, "step": 2873 }, { "epoch": 0.46, "learning_rate": 9.913236150301119e-05, "loss": 0.8273, "step": 2874 }, { "epoch": 0.46, "learning_rate": 9.913063213733535e-05, "loss": 0.7872, "step": 2875 }, { "epoch": 0.46, "learning_rate": 9.91289010650151e-05, "loss": 0.813, "step": 2876 }, { "epoch": 0.46, "learning_rate": 9.912716828611053e-05, "loss": 0.7307, "step": 2877 }, { "epoch": 0.46, "learning_rate": 9.912543380068183e-05, "loss": 0.7891, "step": 2878 }, { "epoch": 0.46, "learning_rate": 9.912369760878928e-05, "loss": 0.8466, "step": 2879 }, { "epoch": 0.46, "learning_rate": 9.912195971049315e-05, "loss": 0.7598, "step": 2880 }, { "epoch": 0.46, "learning_rate": 9.912022010585384e-05, "loss": 0.8042, "step": 2881 }, { "epoch": 0.46, "learning_rate": 9.911847879493177e-05, "loss": 0.7688, "step": 2882 }, { "epoch": 0.46, "learning_rate": 9.911673577778742e-05, "loss": 0.8117, "step": 2883 }, { "epoch": 0.46, "learning_rate": 9.911499105448135e-05, "loss": 0.8682, "step": 2884 }, { "epoch": 0.46, "learning_rate": 9.911324462507416e-05, "loss": 0.816, "step": 2885 }, { "epoch": 0.46, "learning_rate": 9.91114964896265e-05, "loss": 0.7801, "step": 2886 }, { "epoch": 0.46, "learning_rate": 9.910974664819912e-05, "loss": 0.926, "step": 2887 }, { "epoch": 0.46, "learning_rate": 9.91079951008528e-05, "loss": 0.7511, "step": 2888 }, { "epoch": 0.46, "learning_rate": 9.910624184764838e-05, "loss": 0.793, "step": 2889 }, { "epoch": 0.46, "learning_rate": 9.910448688864674e-05, "loss": 0.6859, "step": 2890 }, { "epoch": 0.46, "learning_rate": 9.910273022390887e-05, "loss": 0.8035, "step": 2891 }, { "epoch": 0.46, "learning_rate": 9.910097185349579e-05, "loss": 0.9176, "step": 2892 }, { "epoch": 0.46, "learning_rate": 9.909921177746858e-05, "loss": 0.8043, "step": 2893 }, { "epoch": 0.46, "learning_rate": 9.909744999588835e-05, "loss": 0.8123, "step": 2894 }, { "epoch": 0.46, "learning_rate": 9.909568650881633e-05, "loss": 0.8418, "step": 2895 }, { "epoch": 0.46, "learning_rate": 9.909392131631376e-05, "loss": 0.6845, "step": 2896 }, { "epoch": 0.46, "learning_rate": 9.909215441844199e-05, "loss": 0.8142, "step": 2897 }, { "epoch": 0.46, "learning_rate": 9.909038581526235e-05, "loss": 0.8184, "step": 2898 }, { "epoch": 0.46, "learning_rate": 9.90886155068363e-05, "loss": 0.8173, "step": 2899 }, { "epoch": 0.46, "learning_rate": 9.908684349322535e-05, "loss": 0.7291, "step": 2900 }, { "epoch": 0.46, "learning_rate": 9.908506977449104e-05, "loss": 0.8036, "step": 2901 }, { "epoch": 0.46, "learning_rate": 9.908329435069495e-05, "loss": 0.8025, "step": 2902 }, { "epoch": 0.46, "learning_rate": 9.90815172218988e-05, "loss": 0.7555, "step": 2903 }, { "epoch": 0.47, "learning_rate": 9.907973838816431e-05, "loss": 0.7506, "step": 2904 }, { "epoch": 0.47, "learning_rate": 9.907795784955327e-05, "loss": 0.8052, "step": 2905 }, { "epoch": 0.47, "learning_rate": 9.907617560612752e-05, "loss": 0.7732, "step": 2906 }, { "epoch": 0.47, "learning_rate": 9.907439165794899e-05, "loss": 0.8479, "step": 2907 }, { "epoch": 0.47, "learning_rate": 9.907260600507962e-05, "loss": 0.8409, "step": 2908 }, { "epoch": 0.47, "learning_rate": 9.907081864758146e-05, "loss": 0.7647, "step": 2909 }, { "epoch": 0.47, "learning_rate": 9.906902958551658e-05, "loss": 0.7896, "step": 2910 }, { "epoch": 0.47, "learning_rate": 9.906723881894715e-05, "loss": 0.8171, "step": 2911 }, { "epoch": 0.47, "learning_rate": 9.906544634793535e-05, "loss": 0.8232, "step": 2912 }, { "epoch": 0.47, "learning_rate": 9.906365217254348e-05, "loss": 0.8309, "step": 2913 }, { "epoch": 0.47, "learning_rate": 9.906185629283382e-05, "loss": 0.8229, "step": 2914 }, { "epoch": 0.47, "learning_rate": 9.906005870886878e-05, "loss": 0.8204, "step": 2915 }, { "epoch": 0.47, "learning_rate": 9.905825942071081e-05, "loss": 0.8759, "step": 2916 }, { "epoch": 0.47, "learning_rate": 9.905645842842239e-05, "loss": 0.8173, "step": 2917 }, { "epoch": 0.47, "learning_rate": 9.905465573206608e-05, "loss": 0.8516, "step": 2918 }, { "epoch": 0.47, "learning_rate": 9.905285133170453e-05, "loss": 0.8062, "step": 2919 }, { "epoch": 0.47, "learning_rate": 9.905104522740037e-05, "loss": 0.8306, "step": 2920 }, { "epoch": 0.47, "learning_rate": 9.904923741921639e-05, "loss": 0.8038, "step": 2921 }, { "epoch": 0.47, "learning_rate": 9.904742790721536e-05, "loss": 0.8799, "step": 2922 }, { "epoch": 0.47, "learning_rate": 9.904561669146014e-05, "loss": 0.7287, "step": 2923 }, { "epoch": 0.47, "learning_rate": 9.904380377201365e-05, "loss": 0.8367, "step": 2924 }, { "epoch": 0.47, "learning_rate": 9.904198914893887e-05, "loss": 0.7073, "step": 2925 }, { "epoch": 0.47, "learning_rate": 9.90401728222988e-05, "loss": 0.7741, "step": 2926 }, { "epoch": 0.47, "learning_rate": 9.90383547921566e-05, "loss": 0.8217, "step": 2927 }, { "epoch": 0.47, "learning_rate": 9.903653505857536e-05, "loss": 0.7804, "step": 2928 }, { "epoch": 0.47, "learning_rate": 9.903471362161831e-05, "loss": 0.8895, "step": 2929 }, { "epoch": 0.47, "learning_rate": 9.903289048134874e-05, "loss": 0.8308, "step": 2930 }, { "epoch": 0.47, "learning_rate": 9.903106563782995e-05, "loss": 0.7752, "step": 2931 }, { "epoch": 0.47, "learning_rate": 9.902923909112536e-05, "loss": 0.9649, "step": 2932 }, { "epoch": 0.47, "learning_rate": 9.902741084129841e-05, "loss": 0.8192, "step": 2933 }, { "epoch": 0.47, "learning_rate": 9.902558088841259e-05, "loss": 0.7506, "step": 2934 }, { "epoch": 0.47, "learning_rate": 9.902374923253148e-05, "loss": 0.7629, "step": 2935 }, { "epoch": 0.47, "learning_rate": 9.902191587371872e-05, "loss": 0.7884, "step": 2936 }, { "epoch": 0.47, "learning_rate": 9.902008081203795e-05, "loss": 0.8171, "step": 2937 }, { "epoch": 0.47, "learning_rate": 9.901824404755298e-05, "loss": 0.8128, "step": 2938 }, { "epoch": 0.47, "learning_rate": 9.901640558032754e-05, "loss": 0.8171, "step": 2939 }, { "epoch": 0.47, "learning_rate": 9.901456541042555e-05, "loss": 0.8162, "step": 2940 }, { "epoch": 0.47, "learning_rate": 9.901272353791092e-05, "loss": 0.7502, "step": 2941 }, { "epoch": 0.47, "learning_rate": 9.901087996284761e-05, "loss": 0.7821, "step": 2942 }, { "epoch": 0.47, "learning_rate": 9.900903468529969e-05, "loss": 0.8174, "step": 2943 }, { "epoch": 0.47, "learning_rate": 9.900718770533121e-05, "loss": 0.8074, "step": 2944 }, { "epoch": 0.47, "learning_rate": 9.90053390230064e-05, "loss": 0.8331, "step": 2945 }, { "epoch": 0.47, "learning_rate": 9.900348863838942e-05, "loss": 0.7077, "step": 2946 }, { "epoch": 0.47, "learning_rate": 9.900163655154456e-05, "loss": 0.8679, "step": 2947 }, { "epoch": 0.47, "learning_rate": 9.899978276253616e-05, "loss": 0.841, "step": 2948 }, { "epoch": 0.47, "learning_rate": 9.899792727142863e-05, "loss": 0.8604, "step": 2949 }, { "epoch": 0.47, "learning_rate": 9.899607007828638e-05, "loss": 0.8326, "step": 2950 }, { "epoch": 0.47, "learning_rate": 9.899421118317398e-05, "loss": 0.8341, "step": 2951 }, { "epoch": 0.47, "learning_rate": 9.899235058615596e-05, "loss": 0.7674, "step": 2952 }, { "epoch": 0.47, "learning_rate": 9.899048828729696e-05, "loss": 0.742, "step": 2953 }, { "epoch": 0.47, "learning_rate": 9.89886242866617e-05, "loss": 0.8979, "step": 2954 }, { "epoch": 0.47, "learning_rate": 9.898675858431487e-05, "loss": 0.8168, "step": 2955 }, { "epoch": 0.47, "learning_rate": 9.898489118032135e-05, "loss": 0.8128, "step": 2956 }, { "epoch": 0.47, "learning_rate": 9.898302207474596e-05, "loss": 0.7333, "step": 2957 }, { "epoch": 0.47, "learning_rate": 9.898115126765364e-05, "loss": 0.8678, "step": 2958 }, { "epoch": 0.47, "learning_rate": 9.897927875910936e-05, "loss": 0.8474, "step": 2959 }, { "epoch": 0.47, "learning_rate": 9.897740454917819e-05, "loss": 0.8583, "step": 2960 }, { "epoch": 0.47, "learning_rate": 9.897552863792523e-05, "loss": 0.7867, "step": 2961 }, { "epoch": 0.47, "learning_rate": 9.897365102541563e-05, "loss": 0.7863, "step": 2962 }, { "epoch": 0.47, "learning_rate": 9.897177171171462e-05, "loss": 0.8178, "step": 2963 }, { "epoch": 0.47, "learning_rate": 9.89698906968875e-05, "loss": 0.8202, "step": 2964 }, { "epoch": 0.47, "learning_rate": 9.896800798099956e-05, "loss": 0.7627, "step": 2965 }, { "epoch": 0.48, "learning_rate": 9.896612356411626e-05, "loss": 0.8742, "step": 2966 }, { "epoch": 0.48, "learning_rate": 9.896423744630301e-05, "loss": 0.842, "step": 2967 }, { "epoch": 0.48, "learning_rate": 9.896234962762536e-05, "loss": 0.8045, "step": 2968 }, { "epoch": 0.48, "learning_rate": 9.896046010814886e-05, "loss": 0.8343, "step": 2969 }, { "epoch": 0.48, "learning_rate": 9.895856888793917e-05, "loss": 0.801, "step": 2970 }, { "epoch": 0.48, "learning_rate": 9.895667596706198e-05, "loss": 0.8227, "step": 2971 }, { "epoch": 0.48, "learning_rate": 9.895478134558304e-05, "loss": 0.8914, "step": 2972 }, { "epoch": 0.48, "learning_rate": 9.895288502356817e-05, "loss": 0.8039, "step": 2973 }, { "epoch": 0.48, "learning_rate": 9.895098700108322e-05, "loss": 0.803, "step": 2974 }, { "epoch": 0.48, "learning_rate": 9.894908727819414e-05, "loss": 0.8308, "step": 2975 }, { "epoch": 0.48, "learning_rate": 9.894718585496693e-05, "loss": 0.7977, "step": 2976 }, { "epoch": 0.48, "learning_rate": 9.894528273146761e-05, "loss": 0.7702, "step": 2977 }, { "epoch": 0.48, "learning_rate": 9.894337790776231e-05, "loss": 0.7998, "step": 2978 }, { "epoch": 0.48, "learning_rate": 9.894147138391721e-05, "loss": 0.8153, "step": 2979 }, { "epoch": 0.48, "learning_rate": 9.89395631599985e-05, "loss": 0.844, "step": 2980 }, { "epoch": 0.48, "learning_rate": 9.89376532360725e-05, "loss": 0.8752, "step": 2981 }, { "epoch": 0.48, "learning_rate": 9.893574161220554e-05, "loss": 0.8851, "step": 2982 }, { "epoch": 0.48, "learning_rate": 9.8933828288464e-05, "loss": 0.7758, "step": 2983 }, { "epoch": 0.48, "learning_rate": 9.89319132649144e-05, "loss": 0.8568, "step": 2984 }, { "epoch": 0.48, "learning_rate": 9.89299965416232e-05, "loss": 0.9182, "step": 2985 }, { "epoch": 0.48, "learning_rate": 9.892807811865704e-05, "loss": 0.8024, "step": 2986 }, { "epoch": 0.48, "learning_rate": 9.892615799608252e-05, "loss": 0.7431, "step": 2987 }, { "epoch": 0.48, "learning_rate": 9.892423617396633e-05, "loss": 0.7869, "step": 2988 }, { "epoch": 0.48, "learning_rate": 9.892231265237526e-05, "loss": 0.8154, "step": 2989 }, { "epoch": 0.48, "learning_rate": 9.892038743137612e-05, "loss": 0.8131, "step": 2990 }, { "epoch": 0.48, "learning_rate": 9.891846051103578e-05, "loss": 0.7072, "step": 2991 }, { "epoch": 0.48, "learning_rate": 9.891653189142117e-05, "loss": 0.9724, "step": 2992 }, { "epoch": 0.48, "learning_rate": 9.891460157259928e-05, "loss": 0.7796, "step": 2993 }, { "epoch": 0.48, "learning_rate": 9.891266955463719e-05, "loss": 0.8069, "step": 2994 }, { "epoch": 0.48, "learning_rate": 9.891073583760199e-05, "loss": 0.7393, "step": 2995 }, { "epoch": 0.48, "learning_rate": 9.890880042156085e-05, "loss": 0.8964, "step": 2996 }, { "epoch": 0.48, "learning_rate": 9.890686330658101e-05, "loss": 0.7906, "step": 2997 }, { "epoch": 0.48, "learning_rate": 9.890492449272977e-05, "loss": 0.8924, "step": 2998 }, { "epoch": 0.48, "learning_rate": 9.890298398007445e-05, "loss": 0.8484, "step": 2999 }, { "epoch": 0.48, "learning_rate": 9.890104176868247e-05, "loss": 0.7745, "step": 3000 } ], "logging_steps": 1.0, "max_steps": 18729, "num_train_epochs": 3, "save_steps": 1000, "total_flos": 1.9664764839936e+16, "trial_name": null, "trial_params": null }