{ "best_metric": null, "best_model_checkpoint": null, "epoch": 5.0, "global_step": 113705, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.02, "learning_rate": 1.991240490743591e-05, "loss": 0.4716, "step": 500 }, { "epoch": 0.04, "learning_rate": 1.982445802735148e-05, "loss": 0.4173, "step": 1000 }, { "epoch": 0.07, "learning_rate": 1.9736511147267052e-05, "loss": 0.3948, "step": 1500 }, { "epoch": 0.09, "learning_rate": 1.9648564267182623e-05, "loss": 0.3768, "step": 2000 }, { "epoch": 0.11, "learning_rate": 1.9560617387098194e-05, "loss": 0.3575, "step": 2500 }, { "epoch": 0.13, "learning_rate": 1.9472670507013765e-05, "loss": 0.349, "step": 3000 }, { "epoch": 0.15, "learning_rate": 1.9384723626929336e-05, "loss": 0.3422, "step": 3500 }, { "epoch": 0.18, "learning_rate": 1.9296776746844907e-05, "loss": 0.3388, "step": 4000 }, { "epoch": 0.2, "learning_rate": 1.9209181654280816e-05, "loss": 0.3321, "step": 4500 }, { "epoch": 0.22, "learning_rate": 1.9121234774196386e-05, "loss": 0.3223, "step": 5000 }, { "epoch": 0.24, "learning_rate": 1.9033463787872126e-05, "loss": 0.3277, "step": 5500 }, { "epoch": 0.26, "learning_rate": 1.8945516907787697e-05, "loss": 0.3259, "step": 6000 }, { "epoch": 0.29, "learning_rate": 1.8857570027703268e-05, "loss": 0.3294, "step": 6500 }, { "epoch": 0.31, "learning_rate": 1.8769623147618842e-05, "loss": 0.3122, "step": 7000 }, { "epoch": 0.33, "learning_rate": 1.868167626753441e-05, "loss": 0.3043, "step": 7500 }, { "epoch": 0.35, "learning_rate": 1.859372938744998e-05, "loss": 0.3159, "step": 8000 }, { "epoch": 0.37, "learning_rate": 1.850578250736555e-05, "loss": 0.3036, "step": 8500 }, { "epoch": 0.4, "learning_rate": 1.8417835627281126e-05, "loss": 0.3022, "step": 9000 }, { "epoch": 0.42, "learning_rate": 1.8329888747196697e-05, "loss": 0.3058, "step": 9500 }, { "epoch": 0.44, "learning_rate": 1.8242117760872437e-05, "loss": 0.2956, "step": 10000 }, { "epoch": 0.46, "learning_rate": 1.8154170880788004e-05, "loss": 0.2984, "step": 10500 }, { "epoch": 0.48, "learning_rate": 1.8066224000703575e-05, "loss": 0.2883, "step": 11000 }, { "epoch": 0.51, "learning_rate": 1.797827712061915e-05, "loss": 0.2856, "step": 11500 }, { "epoch": 0.53, "learning_rate": 1.7890506134294886e-05, "loss": 0.2973, "step": 12000 }, { "epoch": 0.55, "learning_rate": 1.780255925421046e-05, "loss": 0.2863, "step": 12500 }, { "epoch": 0.57, "learning_rate": 1.771461237412603e-05, "loss": 0.2859, "step": 13000 }, { "epoch": 0.59, "learning_rate": 1.76266654940416e-05, "loss": 0.2876, "step": 13500 }, { "epoch": 0.62, "learning_rate": 1.753889450771734e-05, "loss": 0.2999, "step": 14000 }, { "epoch": 0.64, "learning_rate": 1.745094762763291e-05, "loss": 0.2954, "step": 14500 }, { "epoch": 0.66, "learning_rate": 1.7363000747548483e-05, "loss": 0.2903, "step": 15000 }, { "epoch": 0.68, "learning_rate": 1.7275053867464054e-05, "loss": 0.2799, "step": 15500 }, { "epoch": 0.7, "learning_rate": 1.7187106987379625e-05, "loss": 0.2932, "step": 16000 }, { "epoch": 0.73, "learning_rate": 1.7099160107295193e-05, "loss": 0.2887, "step": 16500 }, { "epoch": 0.75, "learning_rate": 1.7011389120970936e-05, "loss": 0.2759, "step": 17000 }, { "epoch": 0.77, "learning_rate": 1.6923442240886507e-05, "loss": 0.2766, "step": 17500 }, { "epoch": 0.79, "learning_rate": 1.6835495360802078e-05, "loss": 0.2823, "step": 18000 }, { "epoch": 0.81, "learning_rate": 1.674754848071765e-05, "loss": 0.2801, "step": 18500 }, { "epoch": 0.84, "learning_rate": 1.665960160063322e-05, "loss": 0.2702, "step": 19000 }, { "epoch": 0.86, "learning_rate": 1.657165472054879e-05, "loss": 0.2805, "step": 19500 }, { "epoch": 0.88, "learning_rate": 1.648370784046436e-05, "loss": 0.2771, "step": 20000 }, { "epoch": 0.9, "learning_rate": 1.6395760960379932e-05, "loss": 0.2814, "step": 20500 }, { "epoch": 0.92, "learning_rate": 1.6307814080295503e-05, "loss": 0.2675, "step": 21000 }, { "epoch": 0.95, "learning_rate": 1.6220218987731412e-05, "loss": 0.2651, "step": 21500 }, { "epoch": 0.97, "learning_rate": 1.6132272107646983e-05, "loss": 0.2738, "step": 22000 }, { "epoch": 0.99, "learning_rate": 1.6044325227562554e-05, "loss": 0.2714, "step": 22500 }, { "epoch": 1.01, "learning_rate": 1.5956378347478124e-05, "loss": 0.2232, "step": 23000 }, { "epoch": 1.03, "learning_rate": 1.5868431467393695e-05, "loss": 0.201, "step": 23500 }, { "epoch": 1.06, "learning_rate": 1.5780484587309266e-05, "loss": 0.2112, "step": 24000 }, { "epoch": 1.08, "learning_rate": 1.5692537707224837e-05, "loss": 0.207, "step": 24500 }, { "epoch": 1.1, "learning_rate": 1.5604590827140408e-05, "loss": 0.2049, "step": 25000 }, { "epoch": 1.12, "learning_rate": 1.5516819840816148e-05, "loss": 0.2047, "step": 25500 }, { "epoch": 1.14, "learning_rate": 1.5429048854491888e-05, "loss": 0.2095, "step": 26000 }, { "epoch": 1.17, "learning_rate": 1.534110197440746e-05, "loss": 0.2149, "step": 26500 }, { "epoch": 1.19, "learning_rate": 1.5253155094323031e-05, "loss": 0.2081, "step": 27000 }, { "epoch": 1.21, "learning_rate": 1.51652082142386e-05, "loss": 0.2139, "step": 27500 }, { "epoch": 1.23, "learning_rate": 1.5077261334154171e-05, "loss": 0.2089, "step": 28000 }, { "epoch": 1.25, "learning_rate": 1.4989314454069744e-05, "loss": 0.2032, "step": 28500 }, { "epoch": 1.28, "learning_rate": 1.4901367573985315e-05, "loss": 0.2083, "step": 29000 }, { "epoch": 1.3, "learning_rate": 1.4813420693900884e-05, "loss": 0.2077, "step": 29500 }, { "epoch": 1.32, "learning_rate": 1.4725473813816455e-05, "loss": 0.208, "step": 30000 }, { "epoch": 1.34, "learning_rate": 1.4638054615012534e-05, "loss": 0.2135, "step": 30500 }, { "epoch": 1.36, "learning_rate": 1.4550107734928105e-05, "loss": 0.2071, "step": 31000 }, { "epoch": 1.39, "learning_rate": 1.4462160854843676e-05, "loss": 0.1954, "step": 31500 }, { "epoch": 1.41, "learning_rate": 1.4374213974759245e-05, "loss": 0.2193, "step": 32000 }, { "epoch": 1.43, "learning_rate": 1.4286267094674818e-05, "loss": 0.2108, "step": 32500 }, { "epoch": 1.45, "learning_rate": 1.4198320214590389e-05, "loss": 0.2041, "step": 33000 }, { "epoch": 1.47, "learning_rate": 1.411037333450596e-05, "loss": 0.2165, "step": 33500 }, { "epoch": 1.5, "learning_rate": 1.4022426454421532e-05, "loss": 0.1961, "step": 34000 }, { "epoch": 1.52, "learning_rate": 1.3934479574337101e-05, "loss": 0.2114, "step": 34500 }, { "epoch": 1.54, "learning_rate": 1.3846532694252672e-05, "loss": 0.2047, "step": 35000 }, { "epoch": 1.56, "learning_rate": 1.3758585814168243e-05, "loss": 0.2006, "step": 35500 }, { "epoch": 1.58, "learning_rate": 1.3670638934083816e-05, "loss": 0.1974, "step": 36000 }, { "epoch": 1.61, "learning_rate": 1.3582867947759556e-05, "loss": 0.2085, "step": 36500 }, { "epoch": 1.63, "learning_rate": 1.3494921067675126e-05, "loss": 0.2022, "step": 37000 }, { "epoch": 1.65, "learning_rate": 1.3407150081350866e-05, "loss": 0.2057, "step": 37500 }, { "epoch": 1.67, "learning_rate": 1.3319203201266435e-05, "loss": 0.2023, "step": 38000 }, { "epoch": 1.69, "learning_rate": 1.3231256321182006e-05, "loss": 0.204, "step": 38500 }, { "epoch": 1.71, "learning_rate": 1.3143309441097579e-05, "loss": 0.1929, "step": 39000 }, { "epoch": 1.74, "learning_rate": 1.305536256101315e-05, "loss": 0.205, "step": 39500 }, { "epoch": 1.76, "learning_rate": 1.296741568092872e-05, "loss": 0.2078, "step": 40000 }, { "epoch": 1.78, "learning_rate": 1.287946880084429e-05, "loss": 0.2026, "step": 40500 }, { "epoch": 1.8, "learning_rate": 1.279169781452003e-05, "loss": 0.196, "step": 41000 }, { "epoch": 1.82, "learning_rate": 1.27037509344356e-05, "loss": 0.214, "step": 41500 }, { "epoch": 1.85, "learning_rate": 1.2615804054351173e-05, "loss": 0.2011, "step": 42000 }, { "epoch": 1.87, "learning_rate": 1.2527857174266744e-05, "loss": 0.2083, "step": 42500 }, { "epoch": 1.89, "learning_rate": 1.2439910294182317e-05, "loss": 0.2045, "step": 43000 }, { "epoch": 1.91, "learning_rate": 1.2351963414097886e-05, "loss": 0.2051, "step": 43500 }, { "epoch": 1.93, "learning_rate": 1.2264016534013457e-05, "loss": 0.2017, "step": 44000 }, { "epoch": 1.96, "learning_rate": 1.2176069653929028e-05, "loss": 0.2079, "step": 44500 }, { "epoch": 1.98, "learning_rate": 1.2088298667604767e-05, "loss": 0.189, "step": 45000 }, { "epoch": 2.0, "learning_rate": 1.2000351787520338e-05, "loss": 0.1927, "step": 45500 }, { "epoch": 2.02, "learning_rate": 1.1912580801196078e-05, "loss": 0.1394, "step": 46000 }, { "epoch": 2.04, "learning_rate": 1.1824809814871818e-05, "loss": 0.1416, "step": 46500 }, { "epoch": 2.07, "learning_rate": 1.1736862934787389e-05, "loss": 0.1475, "step": 47000 }, { "epoch": 2.09, "learning_rate": 1.1648916054702962e-05, "loss": 0.1571, "step": 47500 }, { "epoch": 2.11, "learning_rate": 1.156096917461853e-05, "loss": 0.1504, "step": 48000 }, { "epoch": 2.13, "learning_rate": 1.1473198188294272e-05, "loss": 0.1459, "step": 48500 }, { "epoch": 2.15, "learning_rate": 1.1385251308209841e-05, "loss": 0.1565, "step": 49000 }, { "epoch": 2.18, "learning_rate": 1.1297304428125412e-05, "loss": 0.1527, "step": 49500 }, { "epoch": 2.2, "learning_rate": 1.1209357548040985e-05, "loss": 0.151, "step": 50000 }, { "epoch": 2.22, "learning_rate": 1.1121410667956556e-05, "loss": 0.1642, "step": 50500 }, { "epoch": 2.24, "learning_rate": 1.1033463787872125e-05, "loss": 0.1572, "step": 51000 }, { "epoch": 2.26, "learning_rate": 1.0945516907787696e-05, "loss": 0.162, "step": 51500 }, { "epoch": 2.29, "learning_rate": 1.0857570027703268e-05, "loss": 0.1654, "step": 52000 }, { "epoch": 2.31, "learning_rate": 1.0769799041379008e-05, "loss": 0.1534, "step": 52500 }, { "epoch": 2.33, "learning_rate": 1.0682028055054746e-05, "loss": 0.1652, "step": 53000 }, { "epoch": 2.35, "learning_rate": 1.0594081174970319e-05, "loss": 0.1593, "step": 53500 }, { "epoch": 2.37, "learning_rate": 1.050613429488589e-05, "loss": 0.1609, "step": 54000 }, { "epoch": 2.4, "learning_rate": 1.0418187414801462e-05, "loss": 0.1438, "step": 54500 }, { "epoch": 2.42, "learning_rate": 1.0330240534717032e-05, "loss": 0.1567, "step": 55000 }, { "epoch": 2.44, "learning_rate": 1.0242293654632603e-05, "loss": 0.1484, "step": 55500 }, { "epoch": 2.46, "learning_rate": 1.0154346774548173e-05, "loss": 0.1524, "step": 56000 }, { "epoch": 2.48, "learning_rate": 1.0066575788223913e-05, "loss": 0.1686, "step": 56500 }, { "epoch": 2.51, "learning_rate": 9.978628908139484e-06, "loss": 0.1551, "step": 57000 }, { "epoch": 2.53, "learning_rate": 9.890682028055055e-06, "loss": 0.1617, "step": 57500 }, { "epoch": 2.55, "learning_rate": 9.802735147970626e-06, "loss": 0.1543, "step": 58000 }, { "epoch": 2.57, "learning_rate": 9.714788267886197e-06, "loss": 0.153, "step": 58500 }, { "epoch": 2.59, "learning_rate": 9.62684138780177e-06, "loss": 0.1624, "step": 59000 }, { "epoch": 2.62, "learning_rate": 9.53889450771734e-06, "loss": 0.1649, "step": 59500 }, { "epoch": 2.64, "learning_rate": 9.450947627632911e-06, "loss": 0.1589, "step": 60000 }, { "epoch": 2.66, "learning_rate": 9.363000747548482e-06, "loss": 0.1534, "step": 60500 }, { "epoch": 2.68, "learning_rate": 9.275053867464053e-06, "loss": 0.1554, "step": 61000 }, { "epoch": 2.7, "learning_rate": 9.187282881139791e-06, "loss": 0.1536, "step": 61500 }, { "epoch": 2.73, "learning_rate": 9.099336001055364e-06, "loss": 0.1542, "step": 62000 }, { "epoch": 2.75, "learning_rate": 9.011389120970935e-06, "loss": 0.1552, "step": 62500 }, { "epoch": 2.77, "learning_rate": 8.923442240886505e-06, "loss": 0.1569, "step": 63000 }, { "epoch": 2.79, "learning_rate": 8.835495360802076e-06, "loss": 0.1461, "step": 63500 }, { "epoch": 2.81, "learning_rate": 8.747548480717647e-06, "loss": 0.1493, "step": 64000 }, { "epoch": 2.84, "learning_rate": 8.659777494393387e-06, "loss": 0.1537, "step": 64500 }, { "epoch": 2.86, "learning_rate": 8.571830614308958e-06, "loss": 0.1528, "step": 65000 }, { "epoch": 2.88, "learning_rate": 8.483883734224529e-06, "loss": 0.1575, "step": 65500 }, { "epoch": 2.9, "learning_rate": 8.3959368541401e-06, "loss": 0.158, "step": 66000 }, { "epoch": 2.92, "learning_rate": 8.30798997405567e-06, "loss": 0.1491, "step": 66500 }, { "epoch": 2.95, "learning_rate": 8.220043093971241e-06, "loss": 0.1547, "step": 67000 }, { "epoch": 2.97, "learning_rate": 8.132272107646983e-06, "loss": 0.1482, "step": 67500 }, { "epoch": 2.99, "learning_rate": 8.044325227562552e-06, "loss": 0.161, "step": 68000 }, { "epoch": 3.01, "learning_rate": 7.956378347478125e-06, "loss": 0.1248, "step": 68500 }, { "epoch": 3.03, "learning_rate": 7.868431467393694e-06, "loss": 0.1013, "step": 69000 }, { "epoch": 3.06, "learning_rate": 7.780484587309267e-06, "loss": 0.1053, "step": 69500 }, { "epoch": 3.08, "learning_rate": 7.692537707224836e-06, "loss": 0.1022, "step": 70000 }, { "epoch": 3.1, "learning_rate": 7.604590827140408e-06, "loss": 0.0991, "step": 70500 }, { "epoch": 3.12, "learning_rate": 7.516643947055979e-06, "loss": 0.0933, "step": 71000 }, { "epoch": 3.14, "learning_rate": 7.428872960731719e-06, "loss": 0.1031, "step": 71500 }, { "epoch": 3.17, "learning_rate": 7.340926080647289e-06, "loss": 0.1057, "step": 72000 }, { "epoch": 3.19, "learning_rate": 7.252979200562861e-06, "loss": 0.1003, "step": 72500 }, { "epoch": 3.21, "learning_rate": 7.165032320478431e-06, "loss": 0.1092, "step": 73000 }, { "epoch": 3.23, "learning_rate": 7.0772613341541716e-06, "loss": 0.1145, "step": 73500 }, { "epoch": 3.25, "learning_rate": 6.9893144540697424e-06, "loss": 0.1005, "step": 74000 }, { "epoch": 3.28, "learning_rate": 6.901367573985313e-06, "loss": 0.1068, "step": 74500 }, { "epoch": 3.3, "learning_rate": 6.813420693900884e-06, "loss": 0.1049, "step": 75000 }, { "epoch": 3.32, "learning_rate": 6.725473813816455e-06, "loss": 0.1119, "step": 75500 }, { "epoch": 3.34, "learning_rate": 6.637526933732027e-06, "loss": 0.1119, "step": 76000 }, { "epoch": 3.36, "learning_rate": 6.549755947407767e-06, "loss": 0.1131, "step": 76500 }, { "epoch": 3.39, "learning_rate": 6.461809067323337e-06, "loss": 0.1153, "step": 77000 }, { "epoch": 3.41, "learning_rate": 6.373862187238908e-06, "loss": 0.1036, "step": 77500 }, { "epoch": 3.43, "learning_rate": 6.2859153071544785e-06, "loss": 0.1174, "step": 78000 }, { "epoch": 3.45, "learning_rate": 6.198144320830219e-06, "loss": 0.1013, "step": 78500 }, { "epoch": 3.47, "learning_rate": 6.11019744074579e-06, "loss": 0.101, "step": 79000 }, { "epoch": 3.5, "learning_rate": 6.022250560661361e-06, "loss": 0.1129, "step": 79500 }, { "epoch": 3.52, "learning_rate": 5.934303680576932e-06, "loss": 0.1065, "step": 80000 }, { "epoch": 3.54, "learning_rate": 5.846708588012841e-06, "loss": 0.1031, "step": 80500 }, { "epoch": 3.56, "learning_rate": 5.7587617079284116e-06, "loss": 0.1075, "step": 81000 }, { "epoch": 3.58, "learning_rate": 5.6708148278439824e-06, "loss": 0.1118, "step": 81500 }, { "epoch": 3.61, "learning_rate": 5.582867947759554e-06, "loss": 0.1002, "step": 82000 }, { "epoch": 3.63, "learning_rate": 5.494921067675124e-06, "loss": 0.0996, "step": 82500 }, { "epoch": 3.65, "learning_rate": 5.406974187590696e-06, "loss": 0.112, "step": 83000 }, { "epoch": 3.67, "learning_rate": 5.319027307506267e-06, "loss": 0.1023, "step": 83500 }, { "epoch": 3.69, "learning_rate": 5.231080427421838e-06, "loss": 0.1017, "step": 84000 }, { "epoch": 3.72, "learning_rate": 5.1431335473374086e-06, "loss": 0.105, "step": 84500 }, { "epoch": 3.74, "learning_rate": 5.0551866672529794e-06, "loss": 0.0955, "step": 85000 }, { "epoch": 3.76, "learning_rate": 4.967415680928719e-06, "loss": 0.0985, "step": 85500 }, { "epoch": 3.78, "learning_rate": 4.879644694604459e-06, "loss": 0.1082, "step": 86000 }, { "epoch": 3.8, "learning_rate": 4.79169781452003e-06, "loss": 0.1019, "step": 86500 }, { "epoch": 3.83, "learning_rate": 4.703750934435602e-06, "loss": 0.1103, "step": 87000 }, { "epoch": 3.85, "learning_rate": 4.615804054351173e-06, "loss": 0.1064, "step": 87500 }, { "epoch": 3.87, "learning_rate": 4.5278571742667436e-06, "loss": 0.1095, "step": 88000 }, { "epoch": 3.89, "learning_rate": 4.4399102941823144e-06, "loss": 0.1097, "step": 88500 }, { "epoch": 3.91, "learning_rate": 4.351963414097885e-06, "loss": 0.103, "step": 89000 }, { "epoch": 3.94, "learning_rate": 4.264016534013456e-06, "loss": 0.107, "step": 89500 }, { "epoch": 3.96, "learning_rate": 4.176069653929027e-06, "loss": 0.1079, "step": 90000 }, { "epoch": 3.98, "learning_rate": 4.088122773844598e-06, "loss": 0.1057, "step": 90500 }, { "epoch": 4.0, "learning_rate": 4.000527681280507e-06, "loss": 0.0991, "step": 91000 }, { "epoch": 4.02, "learning_rate": 3.912580801196078e-06, "loss": 0.061, "step": 91500 }, { "epoch": 4.05, "learning_rate": 3.8246339211116486e-06, "loss": 0.0657, "step": 92000 }, { "epoch": 4.07, "learning_rate": 3.73668704102722e-06, "loss": 0.0637, "step": 92500 }, { "epoch": 4.09, "learning_rate": 3.6487401609427908e-06, "loss": 0.0817, "step": 93000 }, { "epoch": 4.11, "learning_rate": 3.5607932808583616e-06, "loss": 0.0646, "step": 93500 }, { "epoch": 4.13, "learning_rate": 3.4728464007739325e-06, "loss": 0.0731, "step": 94000 }, { "epoch": 4.16, "learning_rate": 3.3848995206895034e-06, "loss": 0.0577, "step": 94500 }, { "epoch": 4.18, "learning_rate": 3.296952640605075e-06, "loss": 0.0554, "step": 95000 }, { "epoch": 4.2, "learning_rate": 3.2091816542808146e-06, "loss": 0.0668, "step": 95500 }, { "epoch": 4.22, "learning_rate": 3.1212347741963854e-06, "loss": 0.0678, "step": 96000 }, { "epoch": 4.24, "learning_rate": 3.0334637878721257e-06, "loss": 0.0655, "step": 96500 }, { "epoch": 4.27, "learning_rate": 2.9455169077876966e-06, "loss": 0.0695, "step": 97000 }, { "epoch": 4.29, "learning_rate": 2.8575700277032675e-06, "loss": 0.0626, "step": 97500 }, { "epoch": 4.31, "learning_rate": 2.7696231476188384e-06, "loss": 0.0634, "step": 98000 }, { "epoch": 4.33, "learning_rate": 2.6816762675344093e-06, "loss": 0.0752, "step": 98500 }, { "epoch": 4.35, "learning_rate": 2.59372938744998e-06, "loss": 0.0654, "step": 99000 }, { "epoch": 4.38, "learning_rate": 2.505782507365551e-06, "loss": 0.0648, "step": 99500 }, { "epoch": 4.4, "learning_rate": 2.4178356272811223e-06, "loss": 0.0636, "step": 100000 }, { "epoch": 4.42, "learning_rate": 2.330064640956862e-06, "loss": 0.0712, "step": 100500 }, { "epoch": 4.44, "learning_rate": 2.242117760872433e-06, "loss": 0.0667, "step": 101000 }, { "epoch": 4.46, "learning_rate": 2.1541708807880044e-06, "loss": 0.0633, "step": 101500 }, { "epoch": 4.49, "learning_rate": 2.0662240007035753e-06, "loss": 0.0663, "step": 102000 }, { "epoch": 4.51, "learning_rate": 1.978277120619146e-06, "loss": 0.0681, "step": 102500 }, { "epoch": 4.53, "learning_rate": 1.890330240534717e-06, "loss": 0.0631, "step": 103000 }, { "epoch": 4.55, "learning_rate": 1.8023833604502883e-06, "loss": 0.0726, "step": 103500 }, { "epoch": 4.57, "learning_rate": 1.7144364803658592e-06, "loss": 0.0608, "step": 104000 }, { "epoch": 4.6, "learning_rate": 1.626665494041599e-06, "loss": 0.056, "step": 104500 }, { "epoch": 4.62, "learning_rate": 1.53871861395717e-06, "loss": 0.0661, "step": 105000 }, { "epoch": 4.64, "learning_rate": 1.4507717338727408e-06, "loss": 0.0649, "step": 105500 }, { "epoch": 4.66, "learning_rate": 1.3628248537883121e-06, "loss": 0.0733, "step": 106000 }, { "epoch": 4.68, "learning_rate": 1.274877973703883e-06, "loss": 0.068, "step": 106500 }, { "epoch": 4.71, "learning_rate": 1.1871069873796227e-06, "loss": 0.0644, "step": 107000 }, { "epoch": 4.73, "learning_rate": 1.0991601072951938e-06, "loss": 0.0646, "step": 107500 }, { "epoch": 4.75, "learning_rate": 1.0112132272107648e-06, "loss": 0.0634, "step": 108000 }, { "epoch": 4.77, "learning_rate": 9.232663471263357e-07, "loss": 0.0677, "step": 108500 }, { "epoch": 4.79, "learning_rate": 8.354953608020756e-07, "loss": 0.0749, "step": 109000 }, { "epoch": 4.82, "learning_rate": 7.475484807176466e-07, "loss": 0.0604, "step": 109500 }, { "epoch": 4.84, "learning_rate": 6.596016006332176e-07, "loss": 0.0687, "step": 110000 }, { "epoch": 4.86, "learning_rate": 5.716547205487886e-07, "loss": 0.0717, "step": 110500 }, { "epoch": 4.88, "learning_rate": 4.838837342245284e-07, "loss": 0.058, "step": 111000 }, { "epoch": 4.9, "learning_rate": 3.959368541400994e-07, "loss": 0.0712, "step": 111500 }, { "epoch": 4.93, "learning_rate": 3.079899740556704e-07, "loss": 0.0739, "step": 112000 }, { "epoch": 4.95, "learning_rate": 2.2004309397124138e-07, "loss": 0.0655, "step": 112500 }, { "epoch": 4.97, "learning_rate": 1.3227210764698123e-07, "loss": 0.0697, "step": 113000 }, { "epoch": 4.99, "learning_rate": 4.4325227562552225e-08, "loss": 0.0694, "step": 113500 }, { "epoch": 5.0, "step": 113705, "total_flos": 1.196648813106432e+17, "train_loss": 0.16809054645142282, "train_runtime": 5346.9121, "train_samples_per_second": 340.239, "train_steps_per_second": 21.266 } ], "max_steps": 113705, "num_train_epochs": 5, "total_flos": 1.196648813106432e+17, "trial_name": null, "trial_params": null }