{ "best_metric": null, "best_model_checkpoint": null, "epoch": 10.0, "eval_steps": 500, "global_step": 226530, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.02, "learning_rate": 4.98896393413676e-05, "loss": 7.855, "step": 500 }, { "epoch": 0.04, "learning_rate": 4.977927868273518e-05, "loss": 6.9826, "step": 1000 }, { "epoch": 0.07, "learning_rate": 4.9668918024102774e-05, "loss": 6.6355, "step": 1500 }, { "epoch": 0.09, "learning_rate": 4.955855736547036e-05, "loss": 6.3627, "step": 2000 }, { "epoch": 0.11, "learning_rate": 4.944819670683795e-05, "loss": 6.2169, "step": 2500 }, { "epoch": 0.13, "learning_rate": 4.933783604820554e-05, "loss": 6.0647, "step": 3000 }, { "epoch": 0.15, "learning_rate": 4.922747538957313e-05, "loss": 5.9159, "step": 3500 }, { "epoch": 0.18, "learning_rate": 4.9117114730940715e-05, "loss": 5.7711, "step": 4000 }, { "epoch": 0.2, "learning_rate": 4.900675407230831e-05, "loss": 5.6904, "step": 4500 }, { "epoch": 0.22, "learning_rate": 4.88963934136759e-05, "loss": 5.571, "step": 5000 }, { "epoch": 0.24, "learning_rate": 4.8786032755043486e-05, "loss": 5.49, "step": 5500 }, { "epoch": 0.26, "learning_rate": 4.8675672096411075e-05, "loss": 5.3809, "step": 6000 }, { "epoch": 0.29, "learning_rate": 4.856531143777866e-05, "loss": 5.3067, "step": 6500 }, { "epoch": 0.31, "learning_rate": 4.845495077914625e-05, "loss": 5.2368, "step": 7000 }, { "epoch": 0.33, "learning_rate": 4.8344590120513846e-05, "loss": 5.1583, "step": 7500 }, { "epoch": 0.35, "learning_rate": 4.823422946188143e-05, "loss": 5.1042, "step": 8000 }, { "epoch": 0.38, "learning_rate": 4.812386880324902e-05, "loss": 5.0425, "step": 8500 }, { "epoch": 0.4, "learning_rate": 4.801350814461661e-05, "loss": 5.0018, "step": 9000 }, { "epoch": 0.42, "learning_rate": 4.79031474859842e-05, "loss": 4.9537, "step": 9500 }, { "epoch": 0.44, "learning_rate": 4.779278682735179e-05, "loss": 4.9003, "step": 10000 }, { "epoch": 0.46, "learning_rate": 4.768242616871938e-05, "loss": 4.8523, "step": 10500 }, { "epoch": 0.49, "learning_rate": 4.7572065510086963e-05, "loss": 4.8387, "step": 11000 }, { "epoch": 0.51, "learning_rate": 4.746170485145456e-05, "loss": 4.7748, "step": 11500 }, { "epoch": 0.53, "learning_rate": 4.735134419282215e-05, "loss": 4.7472, "step": 12000 }, { "epoch": 0.55, "learning_rate": 4.7240983534189735e-05, "loss": 4.722, "step": 12500 }, { "epoch": 0.57, "learning_rate": 4.713062287555732e-05, "loss": 4.6713, "step": 13000 }, { "epoch": 0.6, "learning_rate": 4.702026221692492e-05, "loss": 4.6363, "step": 13500 }, { "epoch": 0.62, "learning_rate": 4.69099015582925e-05, "loss": 4.6174, "step": 14000 }, { "epoch": 0.64, "learning_rate": 4.6799540899660094e-05, "loss": 4.593, "step": 14500 }, { "epoch": 0.66, "learning_rate": 4.668918024102768e-05, "loss": 4.5752, "step": 15000 }, { "epoch": 0.68, "learning_rate": 4.657881958239527e-05, "loss": 4.5604, "step": 15500 }, { "epoch": 0.71, "learning_rate": 4.646845892376286e-05, "loss": 4.5126, "step": 16000 }, { "epoch": 0.73, "learning_rate": 4.6358098265130454e-05, "loss": 4.4943, "step": 16500 }, { "epoch": 0.75, "learning_rate": 4.6247737606498036e-05, "loss": 4.4695, "step": 17000 }, { "epoch": 0.77, "learning_rate": 4.613737694786563e-05, "loss": 4.4434, "step": 17500 }, { "epoch": 0.79, "learning_rate": 4.602701628923322e-05, "loss": 4.4365, "step": 18000 }, { "epoch": 0.82, "learning_rate": 4.591665563060081e-05, "loss": 4.3896, "step": 18500 }, { "epoch": 0.84, "learning_rate": 4.5806294971968395e-05, "loss": 4.3728, "step": 19000 }, { "epoch": 0.86, "learning_rate": 4.569593431333598e-05, "loss": 4.348, "step": 19500 }, { "epoch": 0.88, "learning_rate": 4.558557365470357e-05, "loss": 4.3362, "step": 20000 }, { "epoch": 0.9, "learning_rate": 4.5475212996071167e-05, "loss": 4.2904, "step": 20500 }, { "epoch": 0.93, "learning_rate": 4.536485233743875e-05, "loss": 4.2728, "step": 21000 }, { "epoch": 0.95, "learning_rate": 4.525449167880634e-05, "loss": 4.2577, "step": 21500 }, { "epoch": 0.97, "learning_rate": 4.514413102017393e-05, "loss": 4.2348, "step": 22000 }, { "epoch": 0.99, "learning_rate": 4.503377036154152e-05, "loss": 4.2174, "step": 22500 }, { "epoch": 1.02, "learning_rate": 4.492340970290911e-05, "loss": 4.1921, "step": 23000 }, { "epoch": 1.04, "learning_rate": 4.48130490442767e-05, "loss": 4.1695, "step": 23500 }, { "epoch": 1.06, "learning_rate": 4.4702688385644284e-05, "loss": 4.1917, "step": 24000 }, { "epoch": 1.08, "learning_rate": 4.459232772701188e-05, "loss": 4.143, "step": 24500 }, { "epoch": 1.1, "learning_rate": 4.448196706837947e-05, "loss": 4.1317, "step": 25000 }, { "epoch": 1.13, "learning_rate": 4.4371606409747055e-05, "loss": 4.1043, "step": 25500 }, { "epoch": 1.15, "learning_rate": 4.4261245751114644e-05, "loss": 4.0883, "step": 26000 }, { "epoch": 1.17, "learning_rate": 4.415088509248224e-05, "loss": 4.0739, "step": 26500 }, { "epoch": 1.19, "learning_rate": 4.404052443384982e-05, "loss": 4.0767, "step": 27000 }, { "epoch": 1.21, "learning_rate": 4.3930163775217415e-05, "loss": 4.0529, "step": 27500 }, { "epoch": 1.24, "learning_rate": 4.3819803116585e-05, "loss": 4.041, "step": 28000 }, { "epoch": 1.26, "learning_rate": 4.370944245795259e-05, "loss": 4.0143, "step": 28500 }, { "epoch": 1.28, "learning_rate": 4.359908179932018e-05, "loss": 3.9921, "step": 29000 }, { "epoch": 1.3, "learning_rate": 4.3488721140687775e-05, "loss": 3.9872, "step": 29500 }, { "epoch": 1.32, "learning_rate": 4.3378360482055356e-05, "loss": 3.9652, "step": 30000 }, { "epoch": 1.35, "learning_rate": 4.326799982342295e-05, "loss": 3.9583, "step": 30500 }, { "epoch": 1.37, "learning_rate": 4.315763916479054e-05, "loss": 3.9451, "step": 31000 }, { "epoch": 1.39, "learning_rate": 4.304727850615813e-05, "loss": 3.9318, "step": 31500 }, { "epoch": 1.41, "learning_rate": 4.2936917847525716e-05, "loss": 3.9023, "step": 32000 }, { "epoch": 1.43, "learning_rate": 4.2826557188893304e-05, "loss": 3.8742, "step": 32500 }, { "epoch": 1.46, "learning_rate": 4.271619653026089e-05, "loss": 3.8979, "step": 33000 }, { "epoch": 1.48, "learning_rate": 4.260583587162849e-05, "loss": 3.8582, "step": 33500 }, { "epoch": 1.5, "learning_rate": 4.2495475212996075e-05, "loss": 3.8363, "step": 34000 }, { "epoch": 1.52, "learning_rate": 4.2385114554363664e-05, "loss": 3.826, "step": 34500 }, { "epoch": 1.55, "learning_rate": 4.227475389573125e-05, "loss": 3.824, "step": 35000 }, { "epoch": 1.57, "learning_rate": 4.216439323709884e-05, "loss": 3.7984, "step": 35500 }, { "epoch": 1.59, "learning_rate": 4.205403257846643e-05, "loss": 3.7817, "step": 36000 }, { "epoch": 1.61, "learning_rate": 4.194367191983402e-05, "loss": 3.7696, "step": 36500 }, { "epoch": 1.63, "learning_rate": 4.1833311261201605e-05, "loss": 3.7553, "step": 37000 }, { "epoch": 1.66, "learning_rate": 4.17229506025692e-05, "loss": 3.7469, "step": 37500 }, { "epoch": 1.68, "learning_rate": 4.161258994393679e-05, "loss": 3.732, "step": 38000 }, { "epoch": 1.7, "learning_rate": 4.1502229285304376e-05, "loss": 3.7407, "step": 38500 }, { "epoch": 1.72, "learning_rate": 4.1391868626671964e-05, "loss": 3.7176, "step": 39000 }, { "epoch": 1.74, "learning_rate": 4.128150796803956e-05, "loss": 3.71, "step": 39500 }, { "epoch": 1.77, "learning_rate": 4.117114730940714e-05, "loss": 3.709, "step": 40000 }, { "epoch": 1.79, "learning_rate": 4.1060786650774736e-05, "loss": 3.677, "step": 40500 }, { "epoch": 1.81, "learning_rate": 4.0950425992142324e-05, "loss": 3.6671, "step": 41000 }, { "epoch": 1.83, "learning_rate": 4.084006533350991e-05, "loss": 3.6457, "step": 41500 }, { "epoch": 1.85, "learning_rate": 4.07297046748775e-05, "loss": 3.6528, "step": 42000 }, { "epoch": 1.88, "learning_rate": 4.0619344016245095e-05, "loss": 3.6485, "step": 42500 }, { "epoch": 1.9, "learning_rate": 4.050898335761268e-05, "loss": 3.6456, "step": 43000 }, { "epoch": 1.92, "learning_rate": 4.039862269898027e-05, "loss": 3.6146, "step": 43500 }, { "epoch": 1.94, "learning_rate": 4.028826204034786e-05, "loss": 3.6269, "step": 44000 }, { "epoch": 1.96, "learning_rate": 4.017790138171545e-05, "loss": 3.614, "step": 44500 }, { "epoch": 1.99, "learning_rate": 4.0067540723083036e-05, "loss": 3.5848, "step": 45000 }, { "epoch": 2.01, "learning_rate": 3.9957180064450625e-05, "loss": 3.5802, "step": 45500 }, { "epoch": 2.03, "learning_rate": 3.984681940581821e-05, "loss": 3.5824, "step": 46000 }, { "epoch": 2.05, "learning_rate": 3.973645874718581e-05, "loss": 3.5543, "step": 46500 }, { "epoch": 2.07, "learning_rate": 3.9626098088553396e-05, "loss": 3.5592, "step": 47000 }, { "epoch": 2.1, "learning_rate": 3.9515737429920984e-05, "loss": 3.5176, "step": 47500 }, { "epoch": 2.12, "learning_rate": 3.940537677128857e-05, "loss": 3.5316, "step": 48000 }, { "epoch": 2.14, "learning_rate": 3.929501611265616e-05, "loss": 3.5072, "step": 48500 }, { "epoch": 2.16, "learning_rate": 3.918465545402375e-05, "loss": 3.5152, "step": 49000 }, { "epoch": 2.19, "learning_rate": 3.9074294795391344e-05, "loss": 3.4943, "step": 49500 }, { "epoch": 2.21, "learning_rate": 3.8963934136758925e-05, "loss": 3.505, "step": 50000 }, { "epoch": 2.23, "learning_rate": 3.885357347812652e-05, "loss": 3.4826, "step": 50500 }, { "epoch": 2.25, "learning_rate": 3.874321281949411e-05, "loss": 3.4639, "step": 51000 }, { "epoch": 2.27, "learning_rate": 3.86328521608617e-05, "loss": 3.4813, "step": 51500 }, { "epoch": 2.3, "learning_rate": 3.8522491502229285e-05, "loss": 3.4717, "step": 52000 }, { "epoch": 2.32, "learning_rate": 3.841213084359688e-05, "loss": 3.4705, "step": 52500 }, { "epoch": 2.34, "learning_rate": 3.830177018496446e-05, "loss": 3.4374, "step": 53000 }, { "epoch": 2.36, "learning_rate": 3.8191409526332056e-05, "loss": 3.4321, "step": 53500 }, { "epoch": 2.38, "learning_rate": 3.8081048867699645e-05, "loss": 3.4489, "step": 54000 }, { "epoch": 2.41, "learning_rate": 3.797068820906723e-05, "loss": 3.4124, "step": 54500 }, { "epoch": 2.43, "learning_rate": 3.786032755043482e-05, "loss": 3.4061, "step": 55000 }, { "epoch": 2.45, "learning_rate": 3.7749966891802416e-05, "loss": 3.4028, "step": 55500 }, { "epoch": 2.47, "learning_rate": 3.763960623317e-05, "loss": 3.3825, "step": 56000 }, { "epoch": 2.49, "learning_rate": 3.752924557453759e-05, "loss": 3.3922, "step": 56500 }, { "epoch": 2.52, "learning_rate": 3.741888491590518e-05, "loss": 3.3839, "step": 57000 }, { "epoch": 2.54, "learning_rate": 3.730852425727277e-05, "loss": 3.3888, "step": 57500 }, { "epoch": 2.56, "learning_rate": 3.719816359864036e-05, "loss": 3.3722, "step": 58000 }, { "epoch": 2.58, "learning_rate": 3.7087802940007945e-05, "loss": 3.3507, "step": 58500 }, { "epoch": 2.6, "learning_rate": 3.6977442281375534e-05, "loss": 3.3475, "step": 59000 }, { "epoch": 2.63, "learning_rate": 3.686708162274313e-05, "loss": 3.3432, "step": 59500 }, { "epoch": 2.65, "learning_rate": 3.675672096411072e-05, "loss": 3.3564, "step": 60000 }, { "epoch": 2.67, "learning_rate": 3.6646360305478305e-05, "loss": 3.331, "step": 60500 }, { "epoch": 2.69, "learning_rate": 3.653599964684589e-05, "loss": 3.3267, "step": 61000 }, { "epoch": 2.71, "learning_rate": 3.642563898821348e-05, "loss": 3.3173, "step": 61500 }, { "epoch": 2.74, "learning_rate": 3.631527832958107e-05, "loss": 3.2926, "step": 62000 }, { "epoch": 2.76, "learning_rate": 3.6204917670948665e-05, "loss": 3.3027, "step": 62500 }, { "epoch": 2.78, "learning_rate": 3.6094557012316246e-05, "loss": 3.2966, "step": 63000 }, { "epoch": 2.8, "learning_rate": 3.598419635368384e-05, "loss": 3.2973, "step": 63500 }, { "epoch": 2.83, "learning_rate": 3.587383569505143e-05, "loss": 3.2974, "step": 64000 }, { "epoch": 2.85, "learning_rate": 3.576347503641902e-05, "loss": 3.3035, "step": 64500 }, { "epoch": 2.87, "learning_rate": 3.5653114377786606e-05, "loss": 3.297, "step": 65000 }, { "epoch": 2.89, "learning_rate": 3.55427537191542e-05, "loss": 3.2761, "step": 65500 }, { "epoch": 2.91, "learning_rate": 3.543239306052178e-05, "loss": 3.2871, "step": 66000 }, { "epoch": 2.94, "learning_rate": 3.532203240188938e-05, "loss": 3.2621, "step": 66500 }, { "epoch": 2.96, "learning_rate": 3.5211671743256965e-05, "loss": 3.2646, "step": 67000 }, { "epoch": 2.98, "learning_rate": 3.5101311084624553e-05, "loss": 3.2538, "step": 67500 }, { "epoch": 3.0, "learning_rate": 3.499095042599214e-05, "loss": 3.247, "step": 68000 }, { "epoch": 3.02, "learning_rate": 3.488058976735974e-05, "loss": 3.2398, "step": 68500 }, { "epoch": 3.05, "learning_rate": 3.477022910872732e-05, "loss": 3.219, "step": 69000 }, { "epoch": 3.07, "learning_rate": 3.465986845009491e-05, "loss": 3.2208, "step": 69500 }, { "epoch": 3.09, "learning_rate": 3.45495077914625e-05, "loss": 3.2071, "step": 70000 }, { "epoch": 3.11, "learning_rate": 3.443914713283009e-05, "loss": 3.2269, "step": 70500 }, { "epoch": 3.13, "learning_rate": 3.432878647419768e-05, "loss": 3.218, "step": 71000 }, { "epoch": 3.16, "learning_rate": 3.4218425815565266e-05, "loss": 3.2206, "step": 71500 }, { "epoch": 3.18, "learning_rate": 3.4108065156932854e-05, "loss": 3.1963, "step": 72000 }, { "epoch": 3.2, "learning_rate": 3.399770449830045e-05, "loss": 3.1893, "step": 72500 }, { "epoch": 3.22, "learning_rate": 3.388734383966804e-05, "loss": 3.1861, "step": 73000 }, { "epoch": 3.24, "learning_rate": 3.3776983181035626e-05, "loss": 3.1908, "step": 73500 }, { "epoch": 3.27, "learning_rate": 3.3666622522403214e-05, "loss": 3.1648, "step": 74000 }, { "epoch": 3.29, "learning_rate": 3.35562618637708e-05, "loss": 3.1668, "step": 74500 }, { "epoch": 3.31, "learning_rate": 3.344590120513839e-05, "loss": 3.1649, "step": 75000 }, { "epoch": 3.33, "learning_rate": 3.3335540546505985e-05, "loss": 3.1564, "step": 75500 }, { "epoch": 3.35, "learning_rate": 3.322517988787357e-05, "loss": 3.1445, "step": 76000 }, { "epoch": 3.38, "learning_rate": 3.311481922924116e-05, "loss": 3.1372, "step": 76500 }, { "epoch": 3.4, "learning_rate": 3.300445857060875e-05, "loss": 3.1581, "step": 77000 }, { "epoch": 3.42, "learning_rate": 3.289409791197634e-05, "loss": 3.1479, "step": 77500 }, { "epoch": 3.44, "learning_rate": 3.2783737253343926e-05, "loss": 3.1489, "step": 78000 }, { "epoch": 3.47, "learning_rate": 3.267337659471152e-05, "loss": 3.1375, "step": 78500 }, { "epoch": 3.49, "learning_rate": 3.25630159360791e-05, "loss": 3.1397, "step": 79000 }, { "epoch": 3.51, "learning_rate": 3.24526552774467e-05, "loss": 3.1281, "step": 79500 }, { "epoch": 3.53, "learning_rate": 3.2342294618814286e-05, "loss": 3.1195, "step": 80000 }, { "epoch": 3.55, "learning_rate": 3.2231933960181874e-05, "loss": 3.1139, "step": 80500 }, { "epoch": 3.58, "learning_rate": 3.212157330154946e-05, "loss": 3.1121, "step": 81000 }, { "epoch": 3.6, "learning_rate": 3.201121264291706e-05, "loss": 3.1138, "step": 81500 }, { "epoch": 3.62, "learning_rate": 3.1900851984284646e-05, "loss": 3.0999, "step": 82000 }, { "epoch": 3.64, "learning_rate": 3.1790491325652234e-05, "loss": 3.1108, "step": 82500 }, { "epoch": 3.66, "learning_rate": 3.168013066701982e-05, "loss": 3.0985, "step": 83000 }, { "epoch": 3.69, "learning_rate": 3.156977000838741e-05, "loss": 3.0951, "step": 83500 }, { "epoch": 3.71, "learning_rate": 3.1459409349755005e-05, "loss": 3.1011, "step": 84000 }, { "epoch": 3.73, "learning_rate": 3.134904869112259e-05, "loss": 3.0798, "step": 84500 }, { "epoch": 3.75, "learning_rate": 3.123868803249018e-05, "loss": 3.0822, "step": 85000 }, { "epoch": 3.77, "learning_rate": 3.112832737385777e-05, "loss": 3.088, "step": 85500 }, { "epoch": 3.8, "learning_rate": 3.101796671522536e-05, "loss": 3.0756, "step": 86000 }, { "epoch": 3.82, "learning_rate": 3.0907606056592946e-05, "loss": 3.0744, "step": 86500 }, { "epoch": 3.84, "learning_rate": 3.079724539796054e-05, "loss": 3.0834, "step": 87000 }, { "epoch": 3.86, "learning_rate": 3.068688473932812e-05, "loss": 3.0687, "step": 87500 }, { "epoch": 3.88, "learning_rate": 3.057652408069572e-05, "loss": 3.0488, "step": 88000 }, { "epoch": 3.91, "learning_rate": 3.0466163422063302e-05, "loss": 3.0629, "step": 88500 }, { "epoch": 3.93, "learning_rate": 3.0355802763430897e-05, "loss": 3.0516, "step": 89000 }, { "epoch": 3.95, "learning_rate": 3.0245442104798482e-05, "loss": 3.0653, "step": 89500 }, { "epoch": 3.97, "learning_rate": 3.0135081446166074e-05, "loss": 3.0395, "step": 90000 }, { "epoch": 4.0, "learning_rate": 3.0024720787533662e-05, "loss": 3.0313, "step": 90500 }, { "epoch": 4.02, "learning_rate": 2.9914360128901254e-05, "loss": 3.024, "step": 91000 }, { "epoch": 4.04, "learning_rate": 2.980399947026884e-05, "loss": 3.0226, "step": 91500 }, { "epoch": 4.06, "learning_rate": 2.969363881163643e-05, "loss": 3.0257, "step": 92000 }, { "epoch": 4.08, "learning_rate": 2.958327815300402e-05, "loss": 3.0177, "step": 92500 }, { "epoch": 4.11, "learning_rate": 2.947291749437161e-05, "loss": 3.011, "step": 93000 }, { "epoch": 4.13, "learning_rate": 2.9362556835739198e-05, "loss": 3.0119, "step": 93500 }, { "epoch": 4.15, "learning_rate": 2.925219617710679e-05, "loss": 2.9969, "step": 94000 }, { "epoch": 4.17, "learning_rate": 2.9141835518474375e-05, "loss": 3.018, "step": 94500 }, { "epoch": 4.19, "learning_rate": 2.9031474859841966e-05, "loss": 3.0119, "step": 95000 }, { "epoch": 4.22, "learning_rate": 2.8921114201209554e-05, "loss": 3.0031, "step": 95500 }, { "epoch": 4.24, "learning_rate": 2.8810753542577146e-05, "loss": 3.0037, "step": 96000 }, { "epoch": 4.26, "learning_rate": 2.870039288394473e-05, "loss": 2.9866, "step": 96500 }, { "epoch": 4.28, "learning_rate": 2.8590032225312326e-05, "loss": 3.0029, "step": 97000 }, { "epoch": 4.3, "learning_rate": 2.847967156667991e-05, "loss": 3.0032, "step": 97500 }, { "epoch": 4.33, "learning_rate": 2.8369310908047502e-05, "loss": 2.9706, "step": 98000 }, { "epoch": 4.35, "learning_rate": 2.825895024941509e-05, "loss": 2.9844, "step": 98500 }, { "epoch": 4.37, "learning_rate": 2.8148589590782682e-05, "loss": 2.9758, "step": 99000 }, { "epoch": 4.39, "learning_rate": 2.8038228932150267e-05, "loss": 2.9609, "step": 99500 }, { "epoch": 4.41, "learning_rate": 2.792786827351786e-05, "loss": 2.9714, "step": 100000 }, { "epoch": 4.44, "learning_rate": 2.7817507614885447e-05, "loss": 2.9702, "step": 100500 }, { "epoch": 4.46, "learning_rate": 2.7707146956253038e-05, "loss": 2.9755, "step": 101000 }, { "epoch": 4.48, "learning_rate": 2.7596786297620626e-05, "loss": 2.9457, "step": 101500 }, { "epoch": 4.5, "learning_rate": 2.7486425638988218e-05, "loss": 2.9575, "step": 102000 }, { "epoch": 4.52, "learning_rate": 2.7376064980355803e-05, "loss": 2.9565, "step": 102500 }, { "epoch": 4.55, "learning_rate": 2.7265704321723395e-05, "loss": 2.9376, "step": 103000 }, { "epoch": 4.57, "learning_rate": 2.7155343663090983e-05, "loss": 2.9508, "step": 103500 }, { "epoch": 4.59, "learning_rate": 2.7044983004458574e-05, "loss": 2.9427, "step": 104000 }, { "epoch": 4.61, "learning_rate": 2.693462234582616e-05, "loss": 2.9427, "step": 104500 }, { "epoch": 4.64, "learning_rate": 2.682426168719375e-05, "loss": 2.9428, "step": 105000 }, { "epoch": 4.66, "learning_rate": 2.671390102856134e-05, "loss": 2.9441, "step": 105500 }, { "epoch": 4.68, "learning_rate": 2.660354036992893e-05, "loss": 2.9409, "step": 106000 }, { "epoch": 4.7, "learning_rate": 2.649317971129652e-05, "loss": 2.9254, "step": 106500 }, { "epoch": 4.72, "learning_rate": 2.638281905266411e-05, "loss": 2.9344, "step": 107000 }, { "epoch": 4.75, "learning_rate": 2.6272458394031695e-05, "loss": 2.9313, "step": 107500 }, { "epoch": 4.77, "learning_rate": 2.6162097735399287e-05, "loss": 2.926, "step": 108000 }, { "epoch": 4.79, "learning_rate": 2.6051737076766875e-05, "loss": 2.9341, "step": 108500 }, { "epoch": 4.81, "learning_rate": 2.5941376418134467e-05, "loss": 2.9209, "step": 109000 }, { "epoch": 4.83, "learning_rate": 2.583101575950205e-05, "loss": 2.9083, "step": 109500 }, { "epoch": 4.86, "learning_rate": 2.5720655100869646e-05, "loss": 2.9081, "step": 110000 }, { "epoch": 4.88, "learning_rate": 2.561029444223723e-05, "loss": 2.8912, "step": 110500 }, { "epoch": 4.9, "learning_rate": 2.5499933783604823e-05, "loss": 2.9015, "step": 111000 }, { "epoch": 4.92, "learning_rate": 2.538957312497241e-05, "loss": 2.8944, "step": 111500 }, { "epoch": 4.94, "learning_rate": 2.5279212466340003e-05, "loss": 2.9002, "step": 112000 }, { "epoch": 4.97, "learning_rate": 2.5168851807707588e-05, "loss": 2.9021, "step": 112500 }, { "epoch": 4.99, "learning_rate": 2.505849114907518e-05, "loss": 2.8858, "step": 113000 }, { "epoch": 5.01, "learning_rate": 2.4948130490442767e-05, "loss": 2.8896, "step": 113500 }, { "epoch": 5.03, "learning_rate": 2.4837769831810356e-05, "loss": 2.872, "step": 114000 }, { "epoch": 5.05, "learning_rate": 2.4727409173177947e-05, "loss": 2.8754, "step": 114500 }, { "epoch": 5.08, "learning_rate": 2.4617048514545535e-05, "loss": 2.8811, "step": 115000 }, { "epoch": 5.1, "learning_rate": 2.4506687855913124e-05, "loss": 2.8895, "step": 115500 }, { "epoch": 5.12, "learning_rate": 2.4396327197280712e-05, "loss": 2.8927, "step": 116000 }, { "epoch": 5.14, "learning_rate": 2.4285966538648303e-05, "loss": 2.8724, "step": 116500 }, { "epoch": 5.16, "learning_rate": 2.417560588001589e-05, "loss": 2.8696, "step": 117000 }, { "epoch": 5.19, "learning_rate": 2.406524522138348e-05, "loss": 2.8622, "step": 117500 }, { "epoch": 5.21, "learning_rate": 2.395488456275107e-05, "loss": 2.8595, "step": 118000 }, { "epoch": 5.23, "learning_rate": 2.384452390411866e-05, "loss": 2.8666, "step": 118500 }, { "epoch": 5.25, "learning_rate": 2.3734163245486248e-05, "loss": 2.8444, "step": 119000 }, { "epoch": 5.28, "learning_rate": 2.362380258685384e-05, "loss": 2.8432, "step": 119500 }, { "epoch": 5.3, "learning_rate": 2.3513441928221428e-05, "loss": 2.8656, "step": 120000 }, { "epoch": 5.32, "learning_rate": 2.3403081269589016e-05, "loss": 2.8539, "step": 120500 }, { "epoch": 5.34, "learning_rate": 2.3292720610956607e-05, "loss": 2.8426, "step": 121000 }, { "epoch": 5.36, "learning_rate": 2.3182359952324196e-05, "loss": 2.856, "step": 121500 }, { "epoch": 5.39, "learning_rate": 2.3071999293691784e-05, "loss": 2.8566, "step": 122000 }, { "epoch": 5.41, "learning_rate": 2.2961638635059375e-05, "loss": 2.8455, "step": 122500 }, { "epoch": 5.43, "learning_rate": 2.2851277976426967e-05, "loss": 2.8429, "step": 123000 }, { "epoch": 5.45, "learning_rate": 2.2740917317794555e-05, "loss": 2.8431, "step": 123500 }, { "epoch": 5.47, "learning_rate": 2.2630556659162144e-05, "loss": 2.8347, "step": 124000 }, { "epoch": 5.5, "learning_rate": 2.2520196000529735e-05, "loss": 2.8189, "step": 124500 }, { "epoch": 5.52, "learning_rate": 2.2409835341897323e-05, "loss": 2.8323, "step": 125000 }, { "epoch": 5.54, "learning_rate": 2.229947468326491e-05, "loss": 2.8366, "step": 125500 }, { "epoch": 5.56, "learning_rate": 2.21891140246325e-05, "loss": 2.8309, "step": 126000 }, { "epoch": 5.58, "learning_rate": 2.207875336600009e-05, "loss": 2.8222, "step": 126500 }, { "epoch": 5.61, "learning_rate": 2.196839270736768e-05, "loss": 2.822, "step": 127000 }, { "epoch": 5.63, "learning_rate": 2.1858032048735268e-05, "loss": 2.831, "step": 127500 }, { "epoch": 5.65, "learning_rate": 2.174767139010286e-05, "loss": 2.8172, "step": 128000 }, { "epoch": 5.67, "learning_rate": 2.1637310731470448e-05, "loss": 2.8112, "step": 128500 }, { "epoch": 5.69, "learning_rate": 2.1526950072838036e-05, "loss": 2.8105, "step": 129000 }, { "epoch": 5.72, "learning_rate": 2.1416589414205627e-05, "loss": 2.8175, "step": 129500 }, { "epoch": 5.74, "learning_rate": 2.1306228755573216e-05, "loss": 2.8036, "step": 130000 }, { "epoch": 5.76, "learning_rate": 2.1195868096940804e-05, "loss": 2.8057, "step": 130500 }, { "epoch": 5.78, "learning_rate": 2.1085507438308395e-05, "loss": 2.8263, "step": 131000 }, { "epoch": 5.8, "learning_rate": 2.0975146779675984e-05, "loss": 2.802, "step": 131500 }, { "epoch": 5.83, "learning_rate": 2.0864786121043572e-05, "loss": 2.8149, "step": 132000 }, { "epoch": 5.85, "learning_rate": 2.075442546241116e-05, "loss": 2.7993, "step": 132500 }, { "epoch": 5.87, "learning_rate": 2.064406480377875e-05, "loss": 2.809, "step": 133000 }, { "epoch": 5.89, "learning_rate": 2.053370414514634e-05, "loss": 2.7923, "step": 133500 }, { "epoch": 5.92, "learning_rate": 2.0423343486513928e-05, "loss": 2.8004, "step": 134000 }, { "epoch": 5.94, "learning_rate": 2.031298282788152e-05, "loss": 2.7943, "step": 134500 }, { "epoch": 5.96, "learning_rate": 2.0202622169249108e-05, "loss": 2.7971, "step": 135000 }, { "epoch": 5.98, "learning_rate": 2.0092261510616696e-05, "loss": 2.7913, "step": 135500 }, { "epoch": 6.0, "learning_rate": 1.9981900851984288e-05, "loss": 2.7839, "step": 136000 }, { "epoch": 6.03, "learning_rate": 1.9871540193351876e-05, "loss": 2.7821, "step": 136500 }, { "epoch": 6.05, "learning_rate": 1.9761179534719464e-05, "loss": 2.7832, "step": 137000 }, { "epoch": 6.07, "learning_rate": 1.9650818876087056e-05, "loss": 2.7768, "step": 137500 }, { "epoch": 6.09, "learning_rate": 1.9540458217454644e-05, "loss": 2.7808, "step": 138000 }, { "epoch": 6.11, "learning_rate": 1.9430097558822232e-05, "loss": 2.7602, "step": 138500 }, { "epoch": 6.14, "learning_rate": 1.931973690018982e-05, "loss": 2.7757, "step": 139000 }, { "epoch": 6.16, "learning_rate": 1.9209376241557412e-05, "loss": 2.7805, "step": 139500 }, { "epoch": 6.18, "learning_rate": 1.9099015582925e-05, "loss": 2.7524, "step": 140000 }, { "epoch": 6.2, "learning_rate": 1.898865492429259e-05, "loss": 2.768, "step": 140500 }, { "epoch": 6.22, "learning_rate": 1.887829426566018e-05, "loss": 2.7736, "step": 141000 }, { "epoch": 6.25, "learning_rate": 1.8767933607027768e-05, "loss": 2.755, "step": 141500 }, { "epoch": 6.27, "learning_rate": 1.8657572948395356e-05, "loss": 2.7659, "step": 142000 }, { "epoch": 6.29, "learning_rate": 1.8547212289762948e-05, "loss": 2.7609, "step": 142500 }, { "epoch": 6.31, "learning_rate": 1.8436851631130536e-05, "loss": 2.7552, "step": 143000 }, { "epoch": 6.33, "learning_rate": 1.8326490972498124e-05, "loss": 2.7453, "step": 143500 }, { "epoch": 6.36, "learning_rate": 1.8216130313865716e-05, "loss": 2.762, "step": 144000 }, { "epoch": 6.38, "learning_rate": 1.8105769655233304e-05, "loss": 2.7626, "step": 144500 }, { "epoch": 6.4, "learning_rate": 1.7995408996600893e-05, "loss": 2.7481, "step": 145000 }, { "epoch": 6.42, "learning_rate": 1.7885048337968484e-05, "loss": 2.745, "step": 145500 }, { "epoch": 6.45, "learning_rate": 1.7774687679336072e-05, "loss": 2.7483, "step": 146000 }, { "epoch": 6.47, "learning_rate": 1.766432702070366e-05, "loss": 2.7454, "step": 146500 }, { "epoch": 6.49, "learning_rate": 1.755396636207125e-05, "loss": 2.7438, "step": 147000 }, { "epoch": 6.51, "learning_rate": 1.744360570343884e-05, "loss": 2.7544, "step": 147500 }, { "epoch": 6.53, "learning_rate": 1.733324504480643e-05, "loss": 2.7412, "step": 148000 }, { "epoch": 6.56, "learning_rate": 1.7222884386174017e-05, "loss": 2.7303, "step": 148500 }, { "epoch": 6.58, "learning_rate": 1.711252372754161e-05, "loss": 2.7464, "step": 149000 }, { "epoch": 6.6, "learning_rate": 1.7002163068909197e-05, "loss": 2.7312, "step": 149500 }, { "epoch": 6.62, "learning_rate": 1.6891802410276785e-05, "loss": 2.7302, "step": 150000 }, { "epoch": 6.64, "learning_rate": 1.6781441751644376e-05, "loss": 2.7202, "step": 150500 }, { "epoch": 6.67, "learning_rate": 1.6671081093011965e-05, "loss": 2.716, "step": 151000 }, { "epoch": 6.69, "learning_rate": 1.6560720434379553e-05, "loss": 2.7209, "step": 151500 }, { "epoch": 6.71, "learning_rate": 1.6450359775747144e-05, "loss": 2.7151, "step": 152000 }, { "epoch": 6.73, "learning_rate": 1.6339999117114733e-05, "loss": 2.7211, "step": 152500 }, { "epoch": 6.75, "learning_rate": 1.622963845848232e-05, "loss": 2.7304, "step": 153000 }, { "epoch": 6.78, "learning_rate": 1.611927779984991e-05, "loss": 2.7166, "step": 153500 }, { "epoch": 6.8, "learning_rate": 1.60089171412175e-05, "loss": 2.704, "step": 154000 }, { "epoch": 6.82, "learning_rate": 1.589855648258509e-05, "loss": 2.7101, "step": 154500 }, { "epoch": 6.84, "learning_rate": 1.5788195823952677e-05, "loss": 2.7131, "step": 155000 }, { "epoch": 6.86, "learning_rate": 1.567783516532027e-05, "loss": 2.7249, "step": 155500 }, { "epoch": 6.89, "learning_rate": 1.5567474506687857e-05, "loss": 2.7081, "step": 156000 }, { "epoch": 6.91, "learning_rate": 1.5457113848055445e-05, "loss": 2.7138, "step": 156500 }, { "epoch": 6.93, "learning_rate": 1.5346753189423037e-05, "loss": 2.7035, "step": 157000 }, { "epoch": 6.95, "learning_rate": 1.5236392530790625e-05, "loss": 2.6958, "step": 157500 }, { "epoch": 6.97, "learning_rate": 1.5126031872158213e-05, "loss": 2.7289, "step": 158000 }, { "epoch": 7.0, "learning_rate": 1.5015671213525803e-05, "loss": 2.7104, "step": 158500 }, { "epoch": 7.02, "learning_rate": 1.4905310554893393e-05, "loss": 2.7047, "step": 159000 }, { "epoch": 7.04, "learning_rate": 1.4794949896260981e-05, "loss": 2.6943, "step": 159500 }, { "epoch": 7.06, "learning_rate": 1.4684589237628571e-05, "loss": 2.6976, "step": 160000 }, { "epoch": 7.09, "learning_rate": 1.4574228578996161e-05, "loss": 2.7057, "step": 160500 }, { "epoch": 7.11, "learning_rate": 1.446386792036375e-05, "loss": 2.7014, "step": 161000 }, { "epoch": 7.13, "learning_rate": 1.4353507261731339e-05, "loss": 2.6955, "step": 161500 }, { "epoch": 7.15, "learning_rate": 1.4243146603098927e-05, "loss": 2.6801, "step": 162000 }, { "epoch": 7.17, "learning_rate": 1.4132785944466517e-05, "loss": 2.6839, "step": 162500 }, { "epoch": 7.2, "learning_rate": 1.4022425285834107e-05, "loss": 2.6768, "step": 163000 }, { "epoch": 7.22, "learning_rate": 1.3912064627201695e-05, "loss": 2.682, "step": 163500 }, { "epoch": 7.24, "learning_rate": 1.3801703968569285e-05, "loss": 2.6826, "step": 164000 }, { "epoch": 7.26, "learning_rate": 1.3691343309936873e-05, "loss": 2.6889, "step": 164500 }, { "epoch": 7.28, "learning_rate": 1.3580982651304463e-05, "loss": 2.6773, "step": 165000 }, { "epoch": 7.31, "learning_rate": 1.3470621992672053e-05, "loss": 2.6864, "step": 165500 }, { "epoch": 7.33, "learning_rate": 1.3360261334039641e-05, "loss": 2.6754, "step": 166000 }, { "epoch": 7.35, "learning_rate": 1.3249900675407231e-05, "loss": 2.6709, "step": 166500 }, { "epoch": 7.37, "learning_rate": 1.3139540016774821e-05, "loss": 2.6883, "step": 167000 }, { "epoch": 7.39, "learning_rate": 1.302917935814241e-05, "loss": 2.6711, "step": 167500 }, { "epoch": 7.42, "learning_rate": 1.291881869951e-05, "loss": 2.6678, "step": 168000 }, { "epoch": 7.44, "learning_rate": 1.2808458040877588e-05, "loss": 2.6892, "step": 168500 }, { "epoch": 7.46, "learning_rate": 1.2698097382245178e-05, "loss": 2.6791, "step": 169000 }, { "epoch": 7.48, "learning_rate": 1.2587736723612767e-05, "loss": 2.6766, "step": 169500 }, { "epoch": 7.5, "learning_rate": 1.2477376064980356e-05, "loss": 2.6463, "step": 170000 }, { "epoch": 7.53, "learning_rate": 1.2367015406347946e-05, "loss": 2.6794, "step": 170500 }, { "epoch": 7.55, "learning_rate": 1.2256654747715535e-05, "loss": 2.6526, "step": 171000 }, { "epoch": 7.57, "learning_rate": 1.2146294089083124e-05, "loss": 2.6555, "step": 171500 }, { "epoch": 7.59, "learning_rate": 1.2035933430450714e-05, "loss": 2.6601, "step": 172000 }, { "epoch": 7.61, "learning_rate": 1.1925572771818302e-05, "loss": 2.6551, "step": 172500 }, { "epoch": 7.64, "learning_rate": 1.1815212113185892e-05, "loss": 2.6652, "step": 173000 }, { "epoch": 7.66, "learning_rate": 1.1704851454553482e-05, "loss": 2.6643, "step": 173500 }, { "epoch": 7.68, "learning_rate": 1.159449079592107e-05, "loss": 2.6539, "step": 174000 }, { "epoch": 7.7, "learning_rate": 1.148413013728866e-05, "loss": 2.6737, "step": 174500 }, { "epoch": 7.73, "learning_rate": 1.1373769478656248e-05, "loss": 2.6559, "step": 175000 }, { "epoch": 7.75, "learning_rate": 1.1263408820023838e-05, "loss": 2.6595, "step": 175500 }, { "epoch": 7.77, "learning_rate": 1.1153048161391428e-05, "loss": 2.6624, "step": 176000 }, { "epoch": 7.79, "learning_rate": 1.1042687502759016e-05, "loss": 2.6431, "step": 176500 }, { "epoch": 7.81, "learning_rate": 1.0932326844126606e-05, "loss": 2.6496, "step": 177000 }, { "epoch": 7.84, "learning_rate": 1.0821966185494196e-05, "loss": 2.6586, "step": 177500 }, { "epoch": 7.86, "learning_rate": 1.0711605526861784e-05, "loss": 2.655, "step": 178000 }, { "epoch": 7.88, "learning_rate": 1.0601244868229374e-05, "loss": 2.6589, "step": 178500 }, { "epoch": 7.9, "learning_rate": 1.0490884209596962e-05, "loss": 2.6389, "step": 179000 }, { "epoch": 7.92, "learning_rate": 1.0380523550964552e-05, "loss": 2.6326, "step": 179500 }, { "epoch": 7.95, "learning_rate": 1.0270162892332142e-05, "loss": 2.6503, "step": 180000 }, { "epoch": 7.97, "learning_rate": 1.015980223369973e-05, "loss": 2.646, "step": 180500 }, { "epoch": 7.99, "learning_rate": 1.004944157506732e-05, "loss": 2.6389, "step": 181000 }, { "epoch": 8.01, "learning_rate": 9.939080916434908e-06, "loss": 2.6488, "step": 181500 }, { "epoch": 8.03, "learning_rate": 9.828720257802498e-06, "loss": 2.6256, "step": 182000 }, { "epoch": 8.06, "learning_rate": 9.718359599170088e-06, "loss": 2.6521, "step": 182500 }, { "epoch": 8.08, "learning_rate": 9.607998940537676e-06, "loss": 2.6273, "step": 183000 }, { "epoch": 8.1, "learning_rate": 9.497638281905268e-06, "loss": 2.6379, "step": 183500 }, { "epoch": 8.12, "learning_rate": 9.387277623272856e-06, "loss": 2.6416, "step": 184000 }, { "epoch": 8.14, "learning_rate": 9.276916964640446e-06, "loss": 2.6296, "step": 184500 }, { "epoch": 8.17, "learning_rate": 9.166556306008036e-06, "loss": 2.6396, "step": 185000 }, { "epoch": 8.19, "learning_rate": 9.056195647375624e-06, "loss": 2.6305, "step": 185500 }, { "epoch": 8.21, "learning_rate": 8.945834988743214e-06, "loss": 2.6318, "step": 186000 }, { "epoch": 8.23, "learning_rate": 8.835474330110802e-06, "loss": 2.6333, "step": 186500 }, { "epoch": 8.25, "learning_rate": 8.725113671478392e-06, "loss": 2.6185, "step": 187000 }, { "epoch": 8.28, "learning_rate": 8.614753012845982e-06, "loss": 2.6275, "step": 187500 }, { "epoch": 8.3, "learning_rate": 8.50439235421357e-06, "loss": 2.6265, "step": 188000 }, { "epoch": 8.32, "learning_rate": 8.39403169558116e-06, "loss": 2.6203, "step": 188500 }, { "epoch": 8.34, "learning_rate": 8.28367103694875e-06, "loss": 2.6228, "step": 189000 }, { "epoch": 8.37, "learning_rate": 8.173310378316338e-06, "loss": 2.6316, "step": 189500 }, { "epoch": 8.39, "learning_rate": 8.062949719683928e-06, "loss": 2.6137, "step": 190000 }, { "epoch": 8.41, "learning_rate": 7.952589061051516e-06, "loss": 2.6243, "step": 190500 }, { "epoch": 8.43, "learning_rate": 7.842228402419106e-06, "loss": 2.6192, "step": 191000 }, { "epoch": 8.45, "learning_rate": 7.731867743786696e-06, "loss": 2.6191, "step": 191500 }, { "epoch": 8.48, "learning_rate": 7.621507085154285e-06, "loss": 2.6222, "step": 192000 }, { "epoch": 8.5, "learning_rate": 7.511146426521874e-06, "loss": 2.6101, "step": 192500 }, { "epoch": 8.52, "learning_rate": 7.4007857678894635e-06, "loss": 2.6214, "step": 193000 }, { "epoch": 8.54, "learning_rate": 7.2904251092570525e-06, "loss": 2.6073, "step": 193500 }, { "epoch": 8.56, "learning_rate": 7.180064450624642e-06, "loss": 2.6024, "step": 194000 }, { "epoch": 8.59, "learning_rate": 7.0697037919922315e-06, "loss": 2.6149, "step": 194500 }, { "epoch": 8.61, "learning_rate": 6.9593431333598205e-06, "loss": 2.6251, "step": 195000 }, { "epoch": 8.63, "learning_rate": 6.84898247472741e-06, "loss": 2.6094, "step": 195500 }, { "epoch": 8.65, "learning_rate": 6.738621816094999e-06, "loss": 2.6137, "step": 196000 }, { "epoch": 8.67, "learning_rate": 6.6282611574625886e-06, "loss": 2.609, "step": 196500 }, { "epoch": 8.7, "learning_rate": 6.517900498830178e-06, "loss": 2.6086, "step": 197000 }, { "epoch": 8.72, "learning_rate": 6.407539840197767e-06, "loss": 2.6021, "step": 197500 }, { "epoch": 8.74, "learning_rate": 6.297179181565356e-06, "loss": 2.6147, "step": 198000 }, { "epoch": 8.76, "learning_rate": 6.186818522932946e-06, "loss": 2.6163, "step": 198500 }, { "epoch": 8.78, "learning_rate": 6.076457864300535e-06, "loss": 2.6073, "step": 199000 }, { "epoch": 8.81, "learning_rate": 5.966097205668124e-06, "loss": 2.6089, "step": 199500 }, { "epoch": 8.83, "learning_rate": 5.855736547035713e-06, "loss": 2.596, "step": 200000 }, { "epoch": 8.85, "learning_rate": 5.745375888403303e-06, "loss": 2.593, "step": 200500 }, { "epoch": 8.87, "learning_rate": 5.635015229770892e-06, "loss": 2.6106, "step": 201000 }, { "epoch": 8.9, "learning_rate": 5.524654571138481e-06, "loss": 2.5894, "step": 201500 }, { "epoch": 8.92, "learning_rate": 5.41429391250607e-06, "loss": 2.5895, "step": 202000 }, { "epoch": 8.94, "learning_rate": 5.30393325387366e-06, "loss": 2.5983, "step": 202500 }, { "epoch": 8.96, "learning_rate": 5.193572595241249e-06, "loss": 2.6066, "step": 203000 }, { "epoch": 8.98, "learning_rate": 5.083211936608838e-06, "loss": 2.5866, "step": 203500 }, { "epoch": 9.01, "learning_rate": 4.972851277976427e-06, "loss": 2.6039, "step": 204000 }, { "epoch": 9.03, "learning_rate": 4.862490619344016e-06, "loss": 2.5904, "step": 204500 }, { "epoch": 9.05, "learning_rate": 4.752129960711606e-06, "loss": 2.5951, "step": 205000 }, { "epoch": 9.07, "learning_rate": 4.641769302079195e-06, "loss": 2.5923, "step": 205500 }, { "epoch": 9.09, "learning_rate": 4.531408643446784e-06, "loss": 2.6, "step": 206000 }, { "epoch": 9.12, "learning_rate": 4.421047984814373e-06, "loss": 2.6014, "step": 206500 }, { "epoch": 9.14, "learning_rate": 4.310687326181963e-06, "loss": 2.5878, "step": 207000 }, { "epoch": 9.16, "learning_rate": 4.200326667549552e-06, "loss": 2.595, "step": 207500 }, { "epoch": 9.18, "learning_rate": 4.089966008917141e-06, "loss": 2.5872, "step": 208000 }, { "epoch": 9.2, "learning_rate": 3.97960535028473e-06, "loss": 2.5912, "step": 208500 }, { "epoch": 9.23, "learning_rate": 3.86924469165232e-06, "loss": 2.5883, "step": 209000 }, { "epoch": 9.25, "learning_rate": 3.758884033019909e-06, "loss": 2.5867, "step": 209500 }, { "epoch": 9.27, "learning_rate": 3.6485233743874983e-06, "loss": 2.5768, "step": 210000 }, { "epoch": 9.29, "learning_rate": 3.5381627157550878e-06, "loss": 2.5746, "step": 210500 }, { "epoch": 9.31, "learning_rate": 3.427802057122677e-06, "loss": 2.5878, "step": 211000 }, { "epoch": 9.34, "learning_rate": 3.317441398490266e-06, "loss": 2.5827, "step": 211500 }, { "epoch": 9.36, "learning_rate": 3.2070807398578554e-06, "loss": 2.5782, "step": 212000 }, { "epoch": 9.38, "learning_rate": 3.096720081225445e-06, "loss": 2.5837, "step": 212500 }, { "epoch": 9.4, "learning_rate": 2.9863594225930343e-06, "loss": 2.58, "step": 213000 }, { "epoch": 9.42, "learning_rate": 2.8759987639606234e-06, "loss": 2.5829, "step": 213500 }, { "epoch": 9.45, "learning_rate": 2.765638105328213e-06, "loss": 2.5873, "step": 214000 }, { "epoch": 9.47, "learning_rate": 2.655277446695802e-06, "loss": 2.5786, "step": 214500 }, { "epoch": 9.49, "learning_rate": 2.5449167880633914e-06, "loss": 2.6002, "step": 215000 }, { "epoch": 9.51, "learning_rate": 2.4345561294309805e-06, "loss": 2.5866, "step": 215500 }, { "epoch": 9.54, "learning_rate": 2.32419547079857e-06, "loss": 2.5861, "step": 216000 }, { "epoch": 9.56, "learning_rate": 2.213834812166159e-06, "loss": 2.5893, "step": 216500 }, { "epoch": 9.58, "learning_rate": 2.103474153533748e-06, "loss": 2.5722, "step": 217000 }, { "epoch": 9.6, "learning_rate": 1.9931134949013376e-06, "loss": 2.5867, "step": 217500 }, { "epoch": 9.62, "learning_rate": 1.8827528362689268e-06, "loss": 2.5773, "step": 218000 }, { "epoch": 9.65, "learning_rate": 1.772392177636516e-06, "loss": 2.5921, "step": 218500 }, { "epoch": 9.67, "learning_rate": 1.6620315190041056e-06, "loss": 2.5922, "step": 219000 }, { "epoch": 9.69, "learning_rate": 1.5516708603716949e-06, "loss": 2.5771, "step": 219500 }, { "epoch": 9.71, "learning_rate": 1.441310201739284e-06, "loss": 2.5658, "step": 220000 }, { "epoch": 9.73, "learning_rate": 1.3309495431068732e-06, "loss": 2.5758, "step": 220500 }, { "epoch": 9.76, "learning_rate": 1.2205888844744625e-06, "loss": 2.5781, "step": 221000 }, { "epoch": 9.78, "learning_rate": 1.1102282258420517e-06, "loss": 2.5685, "step": 221500 }, { "epoch": 9.8, "learning_rate": 9.998675672096412e-07, "loss": 2.578, "step": 222000 }, { "epoch": 9.82, "learning_rate": 8.895069085772305e-07, "loss": 2.5726, "step": 222500 }, { "epoch": 9.84, "learning_rate": 7.791462499448198e-07, "loss": 2.5737, "step": 223000 }, { "epoch": 9.87, "learning_rate": 6.68785591312409e-07, "loss": 2.5841, "step": 223500 }, { "epoch": 9.89, "learning_rate": 5.584249326799983e-07, "loss": 2.5824, "step": 224000 }, { "epoch": 9.91, "learning_rate": 4.480642740475875e-07, "loss": 2.5898, "step": 224500 }, { "epoch": 9.93, "learning_rate": 3.3770361541517685e-07, "loss": 2.5766, "step": 225000 }, { "epoch": 9.95, "learning_rate": 2.273429567827661e-07, "loss": 2.5611, "step": 225500 }, { "epoch": 9.98, "learning_rate": 1.1698229815035537e-07, "loss": 2.575, "step": 226000 }, { "epoch": 10.0, "learning_rate": 6.621639517944644e-09, "loss": 2.5662, "step": 226500 }, { "epoch": 10.0, "step": 226530, "total_flos": 6.757045384548372e+18, "train_loss": 3.1833379993784776, "train_runtime": 282485.8482, "train_samples_per_second": 25.661, "train_steps_per_second": 0.802 } ], "logging_steps": 500, "max_steps": 226530, "num_train_epochs": 10, "save_steps": 100000, "total_flos": 6.757045384548372e+18, "trial_name": null, "trial_params": null }