{ "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": 1.991258080119608e-05, "loss": 0.4639, "step": 500 }, { "epoch": 0.04, "learning_rate": 1.982480981487182e-05, "loss": 0.4082, "step": 1000 }, { "epoch": 0.07, "learning_rate": 1.973686293478739e-05, "loss": 0.3774, "step": 1500 }, { "epoch": 0.09, "learning_rate": 1.964891605470296e-05, "loss": 0.3657, "step": 2000 }, { "epoch": 0.11, "learning_rate": 1.9560969174618532e-05, "loss": 0.3592, "step": 2500 }, { "epoch": 0.13, "learning_rate": 1.9473022294534103e-05, "loss": 0.3387, "step": 3000 }, { "epoch": 0.15, "learning_rate": 1.9385075414449674e-05, "loss": 0.33, "step": 3500 }, { "epoch": 0.18, "learning_rate": 1.9297128534365245e-05, "loss": 0.3314, "step": 4000 }, { "epoch": 0.2, "learning_rate": 1.9209181654280816e-05, "loss": 0.3259, "step": 4500 }, { "epoch": 0.22, "learning_rate": 1.9121410667956555e-05, "loss": 0.3168, "step": 5000 }, { "epoch": 0.24, "learning_rate": 1.9033815575392464e-05, "loss": 0.3161, "step": 5500 }, { "epoch": 0.26, "learning_rate": 1.8945868695308035e-05, "loss": 0.3235, "step": 6000 }, { "epoch": 0.29, "learning_rate": 1.8857921815223606e-05, "loss": 0.3226, "step": 6500 }, { "epoch": 0.31, "learning_rate": 1.8769974935139177e-05, "loss": 0.3136, "step": 7000 }, { "epoch": 0.33, "learning_rate": 1.8682028055054748e-05, "loss": 0.3053, "step": 7500 }, { "epoch": 0.35, "learning_rate": 1.8594257068730488e-05, "loss": 0.3108, "step": 8000 }, { "epoch": 0.37, "learning_rate": 1.8506486082406228e-05, "loss": 0.3164, "step": 8500 }, { "epoch": 0.4, "learning_rate": 1.84185392023218e-05, "loss": 0.3093, "step": 9000 }, { "epoch": 0.42, "learning_rate": 1.833059232223737e-05, "loss": 0.3174, "step": 9500 }, { "epoch": 0.44, "learning_rate": 1.8242645442152944e-05, "loss": 0.3108, "step": 10000 }, { "epoch": 0.46, "learning_rate": 1.815469856206851e-05, "loss": 0.3191, "step": 10500 }, { "epoch": 0.48, "learning_rate": 1.8066751681984082e-05, "loss": 0.3196, "step": 11000 }, { "epoch": 0.51, "learning_rate": 1.7978804801899653e-05, "loss": 0.3, "step": 11500 }, { "epoch": 0.53, "learning_rate": 1.7890857921815227e-05, "loss": 0.3108, "step": 12000 }, { "epoch": 0.55, "learning_rate": 1.7802911041730798e-05, "loss": 0.2959, "step": 12500 }, { "epoch": 0.57, "learning_rate": 1.7715140055406538e-05, "loss": 0.316, "step": 13000 }, { "epoch": 0.59, "learning_rate": 1.7627193175322105e-05, "loss": 0.3083, "step": 13500 }, { "epoch": 0.62, "learning_rate": 1.753942218899785e-05, "loss": 0.3106, "step": 14000 }, { "epoch": 0.64, "learning_rate": 1.7451475308913416e-05, "loss": 0.3212, "step": 14500 }, { "epoch": 0.66, "learning_rate": 1.7363528428828987e-05, "loss": 0.317, "step": 15000 }, { "epoch": 0.68, "learning_rate": 1.727558154874456e-05, "loss": 0.294, "step": 15500 }, { "epoch": 0.7, "learning_rate": 1.7187634668660132e-05, "loss": 0.3095, "step": 16000 }, { "epoch": 0.73, "learning_rate": 1.70996877885757e-05, "loss": 0.301, "step": 16500 }, { "epoch": 0.75, "learning_rate": 1.701174090849127e-05, "loss": 0.3031, "step": 17000 }, { "epoch": 0.77, "learning_rate": 1.692396992216701e-05, "loss": 0.3063, "step": 17500 }, { "epoch": 0.79, "learning_rate": 1.683619893584275e-05, "loss": 0.3068, "step": 18000 }, { "epoch": 0.81, "learning_rate": 1.674825205575832e-05, "loss": 0.3033, "step": 18500 }, { "epoch": 0.84, "learning_rate": 1.6660305175673895e-05, "loss": 0.2883, "step": 19000 }, { "epoch": 0.86, "learning_rate": 1.6572358295589466e-05, "loss": 0.3163, "step": 19500 }, { "epoch": 0.88, "learning_rate": 1.6484411415505037e-05, "loss": 0.3099, "step": 20000 }, { "epoch": 0.9, "learning_rate": 1.6396464535420608e-05, "loss": 0.3067, "step": 20500 }, { "epoch": 0.92, "learning_rate": 1.630851765533618e-05, "loss": 0.2882, "step": 21000 }, { "epoch": 0.95, "learning_rate": 1.622057077525175e-05, "loss": 0.2885, "step": 21500 }, { "epoch": 0.97, "learning_rate": 1.613279978892749e-05, "loss": 0.2953, "step": 22000 }, { "epoch": 0.99, "learning_rate": 1.604485290884306e-05, "loss": 0.2925, "step": 22500 }, { "epoch": 1.01, "learning_rate": 1.595690602875863e-05, "loss": 0.2615, "step": 23000 }, { "epoch": 1.03, "learning_rate": 1.5868959148674202e-05, "loss": 0.24, "step": 23500 }, { "epoch": 1.06, "learning_rate": 1.5781188162349942e-05, "loss": 0.2509, "step": 24000 }, { "epoch": 1.08, "learning_rate": 1.5693241282265513e-05, "loss": 0.2507, "step": 24500 }, { "epoch": 1.1, "learning_rate": 1.5605294402181084e-05, "loss": 0.252, "step": 25000 }, { "epoch": 1.12, "learning_rate": 1.5517523415856824e-05, "loss": 0.2545, "step": 25500 }, { "epoch": 1.14, "learning_rate": 1.5429576535772395e-05, "loss": 0.2599, "step": 26000 }, { "epoch": 1.17, "learning_rate": 1.5341629655687966e-05, "loss": 0.2566, "step": 26500 }, { "epoch": 1.19, "learning_rate": 1.5253682775603538e-05, "loss": 0.2484, "step": 27000 }, { "epoch": 1.21, "learning_rate": 1.5165735895519107e-05, "loss": 0.2432, "step": 27500 }, { "epoch": 1.23, "learning_rate": 1.5077789015434678e-05, "loss": 0.2415, "step": 28000 }, { "epoch": 1.25, "learning_rate": 1.4989842135350249e-05, "loss": 0.2489, "step": 28500 }, { "epoch": 1.28, "learning_rate": 1.4902071149025989e-05, "loss": 0.2566, "step": 29000 }, { "epoch": 1.3, "learning_rate": 1.4814124268941562e-05, "loss": 0.2382, "step": 29500 }, { "epoch": 1.32, "learning_rate": 1.4726177388857132e-05, "loss": 0.2458, "step": 30000 }, { "epoch": 1.34, "learning_rate": 1.4638230508772702e-05, "loss": 0.2559, "step": 30500 }, { "epoch": 1.36, "learning_rate": 1.4550283628688272e-05, "loss": 0.2402, "step": 31000 }, { "epoch": 1.39, "learning_rate": 1.4462336748603845e-05, "loss": 0.2447, "step": 31500 }, { "epoch": 1.41, "learning_rate": 1.4374565762279585e-05, "loss": 0.2605, "step": 32000 }, { "epoch": 1.43, "learning_rate": 1.4286618882195156e-05, "loss": 0.2605, "step": 32500 }, { "epoch": 1.45, "learning_rate": 1.4198672002110727e-05, "loss": 0.2499, "step": 33000 }, { "epoch": 1.47, "learning_rate": 1.4110725122026296e-05, "loss": 0.2409, "step": 33500 }, { "epoch": 1.5, "learning_rate": 1.4022778241941868e-05, "loss": 0.2371, "step": 34000 }, { "epoch": 1.52, "learning_rate": 1.393483136185744e-05, "loss": 0.2496, "step": 34500 }, { "epoch": 1.54, "learning_rate": 1.384688448177301e-05, "loss": 0.2351, "step": 35000 }, { "epoch": 1.56, "learning_rate": 1.3758937601688583e-05, "loss": 0.2348, "step": 35500 }, { "epoch": 1.58, "learning_rate": 1.3670990721604152e-05, "loss": 0.2343, "step": 36000 }, { "epoch": 1.61, "learning_rate": 1.3583043841519723e-05, "loss": 0.2425, "step": 36500 }, { "epoch": 1.63, "learning_rate": 1.3495096961435294e-05, "loss": 0.2331, "step": 37000 }, { "epoch": 1.65, "learning_rate": 1.3407150081350866e-05, "loss": 0.2353, "step": 37500 }, { "epoch": 1.67, "learning_rate": 1.3319203201266435e-05, "loss": 0.2353, "step": 38000 }, { "epoch": 1.69, "learning_rate": 1.3231432214942177e-05, "loss": 0.2304, "step": 38500 }, { "epoch": 1.71, "learning_rate": 1.3143485334857746e-05, "loss": 0.2303, "step": 39000 }, { "epoch": 1.74, "learning_rate": 1.3055714348533486e-05, "loss": 0.2327, "step": 39500 }, { "epoch": 1.76, "learning_rate": 1.2967767468449057e-05, "loss": 0.2417, "step": 40000 }, { "epoch": 1.78, "learning_rate": 1.287982058836463e-05, "loss": 0.2443, "step": 40500 }, { "epoch": 1.8, "learning_rate": 1.27918737082802e-05, "loss": 0.2372, "step": 41000 }, { "epoch": 1.82, "learning_rate": 1.2703926828195771e-05, "loss": 0.2461, "step": 41500 }, { "epoch": 1.85, "learning_rate": 1.2616155841871511e-05, "loss": 0.232, "step": 42000 }, { "epoch": 1.87, "learning_rate": 1.2528384855547251e-05, "loss": 0.2371, "step": 42500 }, { "epoch": 1.89, "learning_rate": 1.2440437975462822e-05, "loss": 0.2399, "step": 43000 }, { "epoch": 1.91, "learning_rate": 1.2352491095378391e-05, "loss": 0.2299, "step": 43500 }, { "epoch": 1.93, "learning_rate": 1.2264544215293964e-05, "loss": 0.2376, "step": 44000 }, { "epoch": 1.96, "learning_rate": 1.2176597335209535e-05, "loss": 0.2448, "step": 44500 }, { "epoch": 1.98, "learning_rate": 1.2088650455125105e-05, "loss": 0.2233, "step": 45000 }, { "epoch": 2.0, "learning_rate": 1.2000703575040678e-05, "loss": 0.2241, "step": 45500 }, { "epoch": 2.02, "learning_rate": 1.1912756694956247e-05, "loss": 0.1942, "step": 46000 }, { "epoch": 2.04, "learning_rate": 1.1824985708631987e-05, "loss": 0.1831, "step": 46500 }, { "epoch": 2.07, "learning_rate": 1.1737038828547558e-05, "loss": 0.1868, "step": 47000 }, { "epoch": 2.09, "learning_rate": 1.1649091948463129e-05, "loss": 0.2034, "step": 47500 }, { "epoch": 2.11, "learning_rate": 1.1561145068378701e-05, "loss": 0.1946, "step": 48000 }, { "epoch": 2.13, "learning_rate": 1.1473198188294272e-05, "loss": 0.194, "step": 48500 }, { "epoch": 2.15, "learning_rate": 1.1385251308209841e-05, "loss": 0.2068, "step": 49000 }, { "epoch": 2.18, "learning_rate": 1.1297304428125412e-05, "loss": 0.1832, "step": 49500 }, { "epoch": 2.2, "learning_rate": 1.1209357548040985e-05, "loss": 0.2023, "step": 50000 }, { "epoch": 2.22, "learning_rate": 1.1121586561716725e-05, "loss": 0.1985, "step": 50500 }, { "epoch": 2.24, "learning_rate": 1.1033639681632296e-05, "loss": 0.191, "step": 51000 }, { "epoch": 2.26, "learning_rate": 1.0945692801547867e-05, "loss": 0.208, "step": 51500 }, { "epoch": 2.29, "learning_rate": 1.0857745921463436e-05, "loss": 0.1998, "step": 52000 }, { "epoch": 2.31, "learning_rate": 1.0769799041379008e-05, "loss": 0.1927, "step": 52500 }, { "epoch": 2.33, "learning_rate": 1.068185216129458e-05, "loss": 0.2054, "step": 53000 }, { "epoch": 2.35, "learning_rate": 1.059390528121015e-05, "loss": 0.2038, "step": 53500 }, { "epoch": 2.37, "learning_rate": 1.050613429488589e-05, "loss": 0.193, "step": 54000 }, { "epoch": 2.4, "learning_rate": 1.0418187414801462e-05, "loss": 0.1929, "step": 54500 }, { "epoch": 2.42, "learning_rate": 1.0330240534717032e-05, "loss": 0.1936, "step": 55000 }, { "epoch": 2.44, "learning_rate": 1.024246954839277e-05, "loss": 0.1928, "step": 55500 }, { "epoch": 2.46, "learning_rate": 1.0154522668308342e-05, "loss": 0.1948, "step": 56000 }, { "epoch": 2.48, "learning_rate": 1.0066575788223913e-05, "loss": 0.2021, "step": 56500 }, { "epoch": 2.51, "learning_rate": 9.978628908139484e-06, "loss": 0.193, "step": 57000 }, { "epoch": 2.53, "learning_rate": 9.890682028055055e-06, "loss": 0.2027, "step": 57500 }, { "epoch": 2.55, "learning_rate": 9.802911041730795e-06, "loss": 0.2008, "step": 58000 }, { "epoch": 2.57, "learning_rate": 9.714964161646366e-06, "loss": 0.2006, "step": 58500 }, { "epoch": 2.59, "learning_rate": 9.627017281561938e-06, "loss": 0.1887, "step": 59000 }, { "epoch": 2.62, "learning_rate": 9.539070401477508e-06, "loss": 0.1946, "step": 59500 }, { "epoch": 2.64, "learning_rate": 9.45112352139308e-06, "loss": 0.1936, "step": 60000 }, { "epoch": 2.66, "learning_rate": 9.36317664130865e-06, "loss": 0.1924, "step": 60500 }, { "epoch": 2.68, "learning_rate": 9.275229761224222e-06, "loss": 0.184, "step": 61000 }, { "epoch": 2.7, "learning_rate": 9.187282881139791e-06, "loss": 0.1768, "step": 61500 }, { "epoch": 2.73, "learning_rate": 9.099336001055364e-06, "loss": 0.1845, "step": 62000 }, { "epoch": 2.75, "learning_rate": 9.011565014731104e-06, "loss": 0.1965, "step": 62500 }, { "epoch": 2.77, "learning_rate": 8.923618134646674e-06, "loss": 0.2004, "step": 63000 }, { "epoch": 2.79, "learning_rate": 8.835671254562245e-06, "loss": 0.1911, "step": 63500 }, { "epoch": 2.81, "learning_rate": 8.747724374477816e-06, "loss": 0.1842, "step": 64000 }, { "epoch": 2.84, "learning_rate": 8.659777494393387e-06, "loss": 0.1907, "step": 64500 }, { "epoch": 2.86, "learning_rate": 8.571830614308958e-06, "loss": 0.1883, "step": 65000 }, { "epoch": 2.88, "learning_rate": 8.484059627984698e-06, "loss": 0.1876, "step": 65500 }, { "epoch": 2.9, "learning_rate": 8.396112747900269e-06, "loss": 0.1846, "step": 66000 }, { "epoch": 2.92, "learning_rate": 8.308341761576009e-06, "loss": 0.1939, "step": 66500 }, { "epoch": 2.95, "learning_rate": 8.22039488149158e-06, "loss": 0.1896, "step": 67000 }, { "epoch": 2.97, "learning_rate": 8.13244800140715e-06, "loss": 0.1957, "step": 67500 }, { "epoch": 2.99, "learning_rate": 8.044501121322721e-06, "loss": 0.1881, "step": 68000 }, { "epoch": 3.01, "learning_rate": 7.956554241238292e-06, "loss": 0.1611, "step": 68500 }, { "epoch": 3.03, "learning_rate": 7.868607361153865e-06, "loss": 0.1423, "step": 69000 }, { "epoch": 3.06, "learning_rate": 7.780660481069434e-06, "loss": 0.1419, "step": 69500 }, { "epoch": 3.08, "learning_rate": 7.692889494745175e-06, "loss": 0.1384, "step": 70000 }, { "epoch": 3.1, "learning_rate": 7.6051185084209144e-06, "loss": 0.1436, "step": 70500 }, { "epoch": 3.12, "learning_rate": 7.5171716283364845e-06, "loss": 0.1357, "step": 71000 }, { "epoch": 3.14, "learning_rate": 7.429224748252056e-06, "loss": 0.1481, "step": 71500 }, { "epoch": 3.17, "learning_rate": 7.341277868167628e-06, "loss": 0.1402, "step": 72000 }, { "epoch": 3.19, "learning_rate": 7.253330988083198e-06, "loss": 0.142, "step": 72500 }, { "epoch": 3.21, "learning_rate": 7.16538410799877e-06, "loss": 0.1518, "step": 73000 }, { "epoch": 3.23, "learning_rate": 7.07743722791434e-06, "loss": 0.1507, "step": 73500 }, { "epoch": 3.25, "learning_rate": 6.9894903478299114e-06, "loss": 0.1481, "step": 74000 }, { "epoch": 3.28, "learning_rate": 6.9015434677454815e-06, "loss": 0.1415, "step": 74500 }, { "epoch": 3.3, "learning_rate": 6.813596587661053e-06, "loss": 0.1392, "step": 75000 }, { "epoch": 3.32, "learning_rate": 6.725649707576625e-06, "loss": 0.1374, "step": 75500 }, { "epoch": 3.34, "learning_rate": 6.637878721252364e-06, "loss": 0.1434, "step": 76000 }, { "epoch": 3.36, "learning_rate": 6.549931841167935e-06, "loss": 0.1423, "step": 76500 }, { "epoch": 3.39, "learning_rate": 6.4619849610835065e-06, "loss": 0.1455, "step": 77000 }, { "epoch": 3.41, "learning_rate": 6.374038080999077e-06, "loss": 0.1464, "step": 77500 }, { "epoch": 3.43, "learning_rate": 6.286091200914648e-06, "loss": 0.1461, "step": 78000 }, { "epoch": 3.45, "learning_rate": 6.198144320830219e-06, "loss": 0.1385, "step": 78500 }, { "epoch": 3.47, "learning_rate": 6.11019744074579e-06, "loss": 0.1474, "step": 79000 }, { "epoch": 3.5, "learning_rate": 6.022250560661361e-06, "loss": 0.1498, "step": 79500 }, { "epoch": 3.52, "learning_rate": 5.934303680576932e-06, "loss": 0.1375, "step": 80000 }, { "epoch": 3.54, "learning_rate": 5.8465326942526725e-06, "loss": 0.1407, "step": 80500 }, { "epoch": 3.56, "learning_rate": 5.7585858141682426e-06, "loss": 0.1389, "step": 81000 }, { "epoch": 3.58, "learning_rate": 5.670638934083814e-06, "loss": 0.139, "step": 81500 }, { "epoch": 3.61, "learning_rate": 5.582692053999384e-06, "loss": 0.142, "step": 82000 }, { "epoch": 3.63, "learning_rate": 5.494921067675124e-06, "loss": 0.1346, "step": 82500 }, { "epoch": 3.65, "learning_rate": 5.406974187590696e-06, "loss": 0.1462, "step": 83000 }, { "epoch": 3.67, "learning_rate": 5.319027307506267e-06, "loss": 0.131, "step": 83500 }, { "epoch": 3.69, "learning_rate": 5.231080427421838e-06, "loss": 0.1475, "step": 84000 }, { "epoch": 3.72, "learning_rate": 5.1431335473374086e-06, "loss": 0.137, "step": 84500 }, { "epoch": 3.74, "learning_rate": 5.0551866672529794e-06, "loss": 0.1337, "step": 85000 }, { "epoch": 3.76, "learning_rate": 4.96723978716855e-06, "loss": 0.1399, "step": 85500 }, { "epoch": 3.78, "learning_rate": 4.879292907084122e-06, "loss": 0.1404, "step": 86000 }, { "epoch": 3.8, "learning_rate": 4.791346026999693e-06, "loss": 0.1428, "step": 86500 }, { "epoch": 3.83, "learning_rate": 4.703399146915264e-06, "loss": 0.1446, "step": 87000 }, { "epoch": 3.85, "learning_rate": 4.615452266830835e-06, "loss": 0.1384, "step": 87500 }, { "epoch": 3.87, "learning_rate": 4.5275053867464056e-06, "loss": 0.1483, "step": 88000 }, { "epoch": 3.89, "learning_rate": 4.4397344004221454e-06, "loss": 0.1473, "step": 88500 }, { "epoch": 3.91, "learning_rate": 4.351787520337716e-06, "loss": 0.1446, "step": 89000 }, { "epoch": 3.94, "learning_rate": 4.263840640253287e-06, "loss": 0.1419, "step": 89500 }, { "epoch": 3.96, "learning_rate": 4.175893760168858e-06, "loss": 0.133, "step": 90000 }, { "epoch": 3.98, "learning_rate": 4.087946880084429e-06, "loss": 0.144, "step": 90500 }, { "epoch": 4.0, "learning_rate": 4.000000000000001e-06, "loss": 0.1252, "step": 91000 }, { "epoch": 4.02, "learning_rate": 3.9122290136757405e-06, "loss": 0.1003, "step": 91500 }, { "epoch": 4.05, "learning_rate": 3.8244580273514796e-06, "loss": 0.0883, "step": 92000 }, { "epoch": 4.07, "learning_rate": 3.7365111472670513e-06, "loss": 0.0914, "step": 92500 }, { "epoch": 4.09, "learning_rate": 3.648564267182622e-06, "loss": 0.1104, "step": 93000 }, { "epoch": 4.11, "learning_rate": 3.560617387098193e-06, "loss": 0.0859, "step": 93500 }, { "epoch": 4.13, "learning_rate": 3.472670507013764e-06, "loss": 0.0992, "step": 94000 }, { "epoch": 4.16, "learning_rate": 3.384723626929335e-06, "loss": 0.088, "step": 94500 }, { "epoch": 4.18, "learning_rate": 3.296952640605075e-06, "loss": 0.0973, "step": 95000 }, { "epoch": 4.2, "learning_rate": 3.209005760520646e-06, "loss": 0.0972, "step": 95500 }, { "epoch": 4.22, "learning_rate": 3.121058880436217e-06, "loss": 0.0971, "step": 96000 }, { "epoch": 4.24, "learning_rate": 3.0331120003517877e-06, "loss": 0.096, "step": 96500 }, { "epoch": 4.27, "learning_rate": 2.9451651202673586e-06, "loss": 0.1011, "step": 97000 }, { "epoch": 4.29, "learning_rate": 2.857394133943099e-06, "loss": 0.094, "step": 97500 }, { "epoch": 4.31, "learning_rate": 2.76944725385867e-06, "loss": 0.0998, "step": 98000 }, { "epoch": 4.33, "learning_rate": 2.6815003737742407e-06, "loss": 0.0833, "step": 98500 }, { "epoch": 4.35, "learning_rate": 2.5935534936898116e-06, "loss": 0.103, "step": 99000 }, { "epoch": 4.38, "learning_rate": 2.5056066136053824e-06, "loss": 0.0897, "step": 99500 }, { "epoch": 4.4, "learning_rate": 2.4176597335209537e-06, "loss": 0.1012, "step": 100000 }, { "epoch": 4.42, "learning_rate": 2.3297128534365246e-06, "loss": 0.0923, "step": 100500 }, { "epoch": 4.44, "learning_rate": 2.2419418671122645e-06, "loss": 0.0899, "step": 101000 }, { "epoch": 4.46, "learning_rate": 2.1539949870278354e-06, "loss": 0.0842, "step": 101500 }, { "epoch": 4.49, "learning_rate": 2.0660481069434063e-06, "loss": 0.0961, "step": 102000 }, { "epoch": 4.51, "learning_rate": 1.978101226858977e-06, "loss": 0.0979, "step": 102500 }, { "epoch": 4.53, "learning_rate": 1.890330240534717e-06, "loss": 0.0957, "step": 103000 }, { "epoch": 4.55, "learning_rate": 1.8025592542104569e-06, "loss": 0.0897, "step": 103500 }, { "epoch": 4.57, "learning_rate": 1.7146123741260282e-06, "loss": 0.0739, "step": 104000 }, { "epoch": 4.6, "learning_rate": 1.626665494041599e-06, "loss": 0.094, "step": 104500 }, { "epoch": 4.62, "learning_rate": 1.53871861395717e-06, "loss": 0.0981, "step": 105000 }, { "epoch": 4.64, "learning_rate": 1.4507717338727408e-06, "loss": 0.0828, "step": 105500 }, { "epoch": 4.66, "learning_rate": 1.3628248537883121e-06, "loss": 0.0993, "step": 106000 }, { "epoch": 4.68, "learning_rate": 1.274877973703883e-06, "loss": 0.0916, "step": 106500 }, { "epoch": 4.71, "learning_rate": 1.1869310936194539e-06, "loss": 0.1046, "step": 107000 }, { "epoch": 4.73, "learning_rate": 1.0991601072951938e-06, "loss": 0.0912, "step": 107500 }, { "epoch": 4.75, "learning_rate": 1.0112132272107648e-06, "loss": 0.0969, "step": 108000 }, { "epoch": 4.77, "learning_rate": 9.232663471263357e-07, "loss": 0.0879, "step": 108500 }, { "epoch": 4.79, "learning_rate": 8.353194670419068e-07, "loss": 0.0982, "step": 109000 }, { "epoch": 4.82, "learning_rate": 7.473725869574777e-07, "loss": 0.0853, "step": 109500 }, { "epoch": 4.84, "learning_rate": 6.594257068730488e-07, "loss": 0.0958, "step": 110000 }, { "epoch": 4.86, "learning_rate": 5.714788267886198e-07, "loss": 0.1026, "step": 110500 }, { "epoch": 4.88, "learning_rate": 4.835319467041907e-07, "loss": 0.0867, "step": 111000 }, { "epoch": 4.9, "learning_rate": 3.9558506661976173e-07, "loss": 0.0976, "step": 111500 }, { "epoch": 4.93, "learning_rate": 3.076381865353327e-07, "loss": 0.0966, "step": 112000 }, { "epoch": 4.95, "learning_rate": 2.1986720021107254e-07, "loss": 0.0911, "step": 112500 }, { "epoch": 4.97, "learning_rate": 1.3209621388681237e-07, "loss": 0.0931, "step": 113000 }, { "epoch": 4.99, "learning_rate": 4.414933380238336e-08, "loss": 0.0943, "step": 113500 }, { "epoch": 5.0, "step": 113705, "total_flos": 4.2384937352679936e+17, "train_loss": 0.19827925553449116, "train_runtime": 10419.4099, "train_samples_per_second": 174.6, "train_steps_per_second": 10.913 } ], "max_steps": 113705, "num_train_epochs": 5, "total_flos": 4.2384937352679936e+17, "trial_name": null, "trial_params": null }