{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 22854, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 4.978121991773869e-05, "loss": 6.7838, "step": 100 }, { "epoch": 0.03, "learning_rate": 4.956243983547738e-05, "loss": 6.1059, "step": 200 }, { "epoch": 0.04, "learning_rate": 4.9343659753216074e-05, "loss": 5.9324, "step": 300 }, { "epoch": 0.05, "learning_rate": 4.912487967095476e-05, "loss": 5.7856, "step": 400 }, { "epoch": 0.07, "learning_rate": 4.890609958869345e-05, "loss": 5.7135, "step": 500 }, { "epoch": 0.08, "learning_rate": 4.868731950643214e-05, "loss": 5.6541, "step": 600 }, { "epoch": 0.09, "learning_rate": 4.846853942417083e-05, "loss": 5.5545, "step": 700 }, { "epoch": 0.11, "learning_rate": 4.824975934190952e-05, "loss": 5.5722, "step": 800 }, { "epoch": 0.12, "learning_rate": 4.803097925964821e-05, "loss": 5.4627, "step": 900 }, { "epoch": 0.13, "learning_rate": 4.7812199177386893e-05, "loss": 5.4158, "step": 1000 }, { "epoch": 0.14, "learning_rate": 4.759341909512558e-05, "loss": 5.3269, "step": 1100 }, { "epoch": 0.16, "learning_rate": 4.737463901286427e-05, "loss": 5.3289, "step": 1200 }, { "epoch": 0.17, "learning_rate": 4.715585893060296e-05, "loss": 5.3057, "step": 1300 }, { "epoch": 0.18, "learning_rate": 4.693707884834165e-05, "loss": 5.2426, "step": 1400 }, { "epoch": 0.2, "learning_rate": 4.671829876608034e-05, "loss": 5.1834, "step": 1500 }, { "epoch": 0.21, "learning_rate": 4.649951868381903e-05, "loss": 5.1129, "step": 1600 }, { "epoch": 0.22, "learning_rate": 4.628073860155771e-05, "loss": 5.1133, "step": 1700 }, { "epoch": 0.24, "learning_rate": 4.6061958519296404e-05, "loss": 4.9882, "step": 1800 }, { "epoch": 0.25, "learning_rate": 4.5843178437035094e-05, "loss": 5.0315, "step": 1900 }, { "epoch": 0.26, "learning_rate": 4.5624398354773784e-05, "loss": 4.9776, "step": 2000 }, { "epoch": 0.28, "learning_rate": 4.5405618272512475e-05, "loss": 4.9737, "step": 2100 }, { "epoch": 0.29, "learning_rate": 4.5186838190251165e-05, "loss": 4.8419, "step": 2200 }, { "epoch": 0.3, "learning_rate": 4.496805810798985e-05, "loss": 4.8256, "step": 2300 }, { "epoch": 0.32, "learning_rate": 4.474927802572854e-05, "loss": 4.8925, "step": 2400 }, { "epoch": 0.33, "learning_rate": 4.453049794346723e-05, "loss": 4.7332, "step": 2500 }, { "epoch": 0.34, "learning_rate": 4.431171786120592e-05, "loss": 4.7318, "step": 2600 }, { "epoch": 0.35, "learning_rate": 4.409293777894461e-05, "loss": 4.69, "step": 2700 }, { "epoch": 0.37, "learning_rate": 4.3874157696683295e-05, "loss": 4.6975, "step": 2800 }, { "epoch": 0.38, "learning_rate": 4.3655377614421985e-05, "loss": 4.7073, "step": 2900 }, { "epoch": 0.39, "learning_rate": 4.3436597532160675e-05, "loss": 4.5963, "step": 3000 }, { "epoch": 0.41, "learning_rate": 4.321781744989936e-05, "loss": 4.5638, "step": 3100 }, { "epoch": 0.42, "learning_rate": 4.299903736763805e-05, "loss": 4.5798, "step": 3200 }, { "epoch": 0.43, "learning_rate": 4.278025728537674e-05, "loss": 4.4876, "step": 3300 }, { "epoch": 0.45, "learning_rate": 4.256147720311543e-05, "loss": 4.5954, "step": 3400 }, { "epoch": 0.46, "learning_rate": 4.234269712085412e-05, "loss": 4.4998, "step": 3500 }, { "epoch": 0.47, "learning_rate": 4.212391703859281e-05, "loss": 4.495, "step": 3600 }, { "epoch": 0.49, "learning_rate": 4.1905136956331495e-05, "loss": 4.537, "step": 3700 }, { "epoch": 0.5, "learning_rate": 4.1686356874070186e-05, "loss": 4.4669, "step": 3800 }, { "epoch": 0.51, "learning_rate": 4.1467576791808876e-05, "loss": 4.5145, "step": 3900 }, { "epoch": 0.53, "learning_rate": 4.1248796709547566e-05, "loss": 4.3922, "step": 4000 }, { "epoch": 0.54, "learning_rate": 4.103001662728626e-05, "loss": 4.4273, "step": 4100 }, { "epoch": 0.55, "learning_rate": 4.081123654502495e-05, "loss": 4.3919, "step": 4200 }, { "epoch": 0.56, "learning_rate": 4.059245646276363e-05, "loss": 4.3121, "step": 4300 }, { "epoch": 0.58, "learning_rate": 4.037367638050232e-05, "loss": 4.419, "step": 4400 }, { "epoch": 0.59, "learning_rate": 4.015489629824101e-05, "loss": 4.3461, "step": 4500 }, { "epoch": 0.6, "learning_rate": 3.9936116215979696e-05, "loss": 4.2377, "step": 4600 }, { "epoch": 0.62, "learning_rate": 3.9717336133718386e-05, "loss": 4.2837, "step": 4700 }, { "epoch": 0.63, "learning_rate": 3.9498556051457077e-05, "loss": 4.2739, "step": 4800 }, { "epoch": 0.64, "learning_rate": 3.927977596919577e-05, "loss": 4.2504, "step": 4900 }, { "epoch": 0.66, "learning_rate": 3.906099588693445e-05, "loss": 4.2794, "step": 5000 }, { "epoch": 0.67, "learning_rate": 3.884221580467314e-05, "loss": 4.2125, "step": 5100 }, { "epoch": 0.68, "learning_rate": 3.862343572241183e-05, "loss": 4.2143, "step": 5200 }, { "epoch": 0.7, "learning_rate": 3.840465564015052e-05, "loss": 4.1628, "step": 5300 }, { "epoch": 0.71, "learning_rate": 3.818587555788921e-05, "loss": 4.1445, "step": 5400 }, { "epoch": 0.72, "learning_rate": 3.79670954756279e-05, "loss": 4.237, "step": 5500 }, { "epoch": 0.74, "learning_rate": 3.774831539336659e-05, "loss": 4.2284, "step": 5600 }, { "epoch": 0.75, "learning_rate": 3.752953531110528e-05, "loss": 4.1451, "step": 5700 }, { "epoch": 0.76, "learning_rate": 3.731075522884397e-05, "loss": 4.1644, "step": 5800 }, { "epoch": 0.77, "learning_rate": 3.709197514658266e-05, "loss": 4.1351, "step": 5900 }, { "epoch": 0.79, "learning_rate": 3.687319506432135e-05, "loss": 4.1229, "step": 6000 }, { "epoch": 0.8, "learning_rate": 3.665441498206004e-05, "loss": 4.1166, "step": 6100 }, { "epoch": 0.81, "learning_rate": 3.643563489979873e-05, "loss": 4.1468, "step": 6200 }, { "epoch": 0.83, "learning_rate": 3.621685481753741e-05, "loss": 4.1475, "step": 6300 }, { "epoch": 0.84, "learning_rate": 3.59980747352761e-05, "loss": 4.1488, "step": 6400 }, { "epoch": 0.85, "learning_rate": 3.577929465301479e-05, "loss": 4.0431, "step": 6500 }, { "epoch": 0.87, "learning_rate": 3.556051457075348e-05, "loss": 3.9611, "step": 6600 }, { "epoch": 0.88, "learning_rate": 3.534173448849217e-05, "loss": 4.1072, "step": 6700 }, { "epoch": 0.89, "learning_rate": 3.512295440623086e-05, "loss": 4.0247, "step": 6800 }, { "epoch": 0.91, "learning_rate": 3.490417432396955e-05, "loss": 3.9853, "step": 6900 }, { "epoch": 0.92, "learning_rate": 3.468539424170823e-05, "loss": 3.9586, "step": 7000 }, { "epoch": 0.93, "learning_rate": 3.446661415944692e-05, "loss": 4.0029, "step": 7100 }, { "epoch": 0.95, "learning_rate": 3.4247834077185614e-05, "loss": 4.0219, "step": 7200 }, { "epoch": 0.96, "learning_rate": 3.4029053994924304e-05, "loss": 3.9708, "step": 7300 }, { "epoch": 0.97, "learning_rate": 3.3810273912662995e-05, "loss": 3.9856, "step": 7400 }, { "epoch": 0.98, "learning_rate": 3.3591493830401685e-05, "loss": 3.9354, "step": 7500 }, { "epoch": 1.0, "learning_rate": 3.337271374814037e-05, "loss": 3.98, "step": 7600 }, { "epoch": 1.01, "learning_rate": 3.315393366587906e-05, "loss": 3.9338, "step": 7700 }, { "epoch": 1.02, "learning_rate": 3.293515358361775e-05, "loss": 3.9415, "step": 7800 }, { "epoch": 1.04, "learning_rate": 3.271637350135644e-05, "loss": 3.9926, "step": 7900 }, { "epoch": 1.05, "learning_rate": 3.249759341909513e-05, "loss": 3.9005, "step": 8000 }, { "epoch": 1.06, "learning_rate": 3.227881333683382e-05, "loss": 3.9296, "step": 8100 }, { "epoch": 1.08, "learning_rate": 3.2060033254572505e-05, "loss": 3.9252, "step": 8200 }, { "epoch": 1.09, "learning_rate": 3.184125317231119e-05, "loss": 3.8741, "step": 8300 }, { "epoch": 1.1, "learning_rate": 3.162247309004988e-05, "loss": 3.9018, "step": 8400 }, { "epoch": 1.12, "learning_rate": 3.140369300778857e-05, "loss": 3.8671, "step": 8500 }, { "epoch": 1.13, "learning_rate": 3.118491292552726e-05, "loss": 3.9379, "step": 8600 }, { "epoch": 1.14, "learning_rate": 3.096613284326595e-05, "loss": 3.8542, "step": 8700 }, { "epoch": 1.16, "learning_rate": 3.074735276100464e-05, "loss": 3.7767, "step": 8800 }, { "epoch": 1.17, "learning_rate": 3.052857267874333e-05, "loss": 3.7247, "step": 8900 }, { "epoch": 1.18, "learning_rate": 3.0309792596482018e-05, "loss": 3.7847, "step": 9000 }, { "epoch": 1.19, "learning_rate": 3.0091012514220705e-05, "loss": 3.7639, "step": 9100 }, { "epoch": 1.21, "learning_rate": 2.9872232431959396e-05, "loss": 3.7938, "step": 9200 }, { "epoch": 1.22, "learning_rate": 2.9653452349698086e-05, "loss": 3.7549, "step": 9300 }, { "epoch": 1.23, "learning_rate": 2.9434672267436773e-05, "loss": 3.774, "step": 9400 }, { "epoch": 1.25, "learning_rate": 2.9215892185175464e-05, "loss": 3.7728, "step": 9500 }, { "epoch": 1.26, "learning_rate": 2.8997112102914154e-05, "loss": 3.7448, "step": 9600 }, { "epoch": 1.27, "learning_rate": 2.877833202065284e-05, "loss": 3.7276, "step": 9700 }, { "epoch": 1.29, "learning_rate": 2.8559551938391532e-05, "loss": 3.7897, "step": 9800 }, { "epoch": 1.3, "learning_rate": 2.8340771856130222e-05, "loss": 3.7273, "step": 9900 }, { "epoch": 1.31, "learning_rate": 2.812199177386891e-05, "loss": 3.745, "step": 10000 }, { "epoch": 1.33, "learning_rate": 2.7903211691607596e-05, "loss": 3.7491, "step": 10100 }, { "epoch": 1.34, "learning_rate": 2.7684431609346283e-05, "loss": 3.8055, "step": 10200 }, { "epoch": 1.35, "learning_rate": 2.7465651527084974e-05, "loss": 3.6932, "step": 10300 }, { "epoch": 1.37, "learning_rate": 2.7246871444823664e-05, "loss": 3.7565, "step": 10400 }, { "epoch": 1.38, "learning_rate": 2.702809136256235e-05, "loss": 3.6891, "step": 10500 }, { "epoch": 1.39, "learning_rate": 2.6809311280301042e-05, "loss": 3.7535, "step": 10600 }, { "epoch": 1.4, "learning_rate": 2.6590531198039732e-05, "loss": 3.5967, "step": 10700 }, { "epoch": 1.42, "learning_rate": 2.637175111577842e-05, "loss": 3.637, "step": 10800 }, { "epoch": 1.43, "learning_rate": 2.615297103351711e-05, "loss": 3.6867, "step": 10900 }, { "epoch": 1.44, "learning_rate": 2.59341909512558e-05, "loss": 3.6911, "step": 11000 }, { "epoch": 1.46, "learning_rate": 2.5715410868994487e-05, "loss": 3.6094, "step": 11100 }, { "epoch": 1.47, "learning_rate": 2.5496630786733178e-05, "loss": 3.6662, "step": 11200 }, { "epoch": 1.48, "learning_rate": 2.5277850704471868e-05, "loss": 3.6195, "step": 11300 }, { "epoch": 1.5, "learning_rate": 2.5059070622210555e-05, "loss": 3.6519, "step": 11400 }, { "epoch": 1.51, "learning_rate": 2.4840290539949242e-05, "loss": 3.6181, "step": 11500 }, { "epoch": 1.52, "learning_rate": 2.4621510457687933e-05, "loss": 3.7262, "step": 11600 }, { "epoch": 1.54, "learning_rate": 2.4402730375426623e-05, "loss": 3.6216, "step": 11700 }, { "epoch": 1.55, "learning_rate": 2.418395029316531e-05, "loss": 3.613, "step": 11800 }, { "epoch": 1.56, "learning_rate": 2.3965170210904e-05, "loss": 3.6809, "step": 11900 }, { "epoch": 1.58, "learning_rate": 2.374639012864269e-05, "loss": 3.6462, "step": 12000 }, { "epoch": 1.59, "learning_rate": 2.352761004638138e-05, "loss": 3.6063, "step": 12100 }, { "epoch": 1.6, "learning_rate": 2.330882996412007e-05, "loss": 3.6304, "step": 12200 }, { "epoch": 1.61, "learning_rate": 2.309004988185876e-05, "loss": 3.6105, "step": 12300 }, { "epoch": 1.63, "learning_rate": 2.2871269799597443e-05, "loss": 3.5885, "step": 12400 }, { "epoch": 1.64, "learning_rate": 2.2652489717336133e-05, "loss": 3.6111, "step": 12500 }, { "epoch": 1.65, "learning_rate": 2.2433709635074824e-05, "loss": 3.572, "step": 12600 }, { "epoch": 1.67, "learning_rate": 2.221492955281351e-05, "loss": 3.6038, "step": 12700 }, { "epoch": 1.68, "learning_rate": 2.19961494705522e-05, "loss": 3.6314, "step": 12800 }, { "epoch": 1.69, "learning_rate": 2.1777369388290892e-05, "loss": 3.5771, "step": 12900 }, { "epoch": 1.71, "learning_rate": 2.1558589306029582e-05, "loss": 3.5381, "step": 13000 }, { "epoch": 1.72, "learning_rate": 2.133980922376827e-05, "loss": 3.5831, "step": 13100 }, { "epoch": 1.73, "learning_rate": 2.112102914150696e-05, "loss": 3.5452, "step": 13200 }, { "epoch": 1.75, "learning_rate": 2.0902249059245647e-05, "loss": 3.6384, "step": 13300 }, { "epoch": 1.76, "learning_rate": 2.0683468976984334e-05, "loss": 3.5737, "step": 13400 }, { "epoch": 1.77, "learning_rate": 2.0464688894723024e-05, "loss": 3.684, "step": 13500 }, { "epoch": 1.79, "learning_rate": 2.0245908812461715e-05, "loss": 3.6435, "step": 13600 }, { "epoch": 1.8, "learning_rate": 2.0027128730200402e-05, "loss": 3.5359, "step": 13700 }, { "epoch": 1.81, "learning_rate": 1.9808348647939092e-05, "loss": 3.5537, "step": 13800 }, { "epoch": 1.82, "learning_rate": 1.9589568565677783e-05, "loss": 3.5483, "step": 13900 }, { "epoch": 1.84, "learning_rate": 1.937078848341647e-05, "loss": 3.4914, "step": 14000 }, { "epoch": 1.85, "learning_rate": 1.915200840115516e-05, "loss": 3.4733, "step": 14100 }, { "epoch": 1.86, "learning_rate": 1.8933228318893847e-05, "loss": 3.5044, "step": 14200 }, { "epoch": 1.88, "learning_rate": 1.8714448236632538e-05, "loss": 3.5383, "step": 14300 }, { "epoch": 1.89, "learning_rate": 1.8495668154371225e-05, "loss": 3.5945, "step": 14400 }, { "epoch": 1.9, "learning_rate": 1.8276888072109915e-05, "loss": 3.5568, "step": 14500 }, { "epoch": 1.92, "learning_rate": 1.8058107989848606e-05, "loss": 3.4929, "step": 14600 }, { "epoch": 1.93, "learning_rate": 1.7839327907587293e-05, "loss": 3.5268, "step": 14700 }, { "epoch": 1.94, "learning_rate": 1.7620547825325983e-05, "loss": 3.5304, "step": 14800 }, { "epoch": 1.96, "learning_rate": 1.7401767743064674e-05, "loss": 3.4343, "step": 14900 }, { "epoch": 1.97, "learning_rate": 1.718298766080336e-05, "loss": 3.5578, "step": 15000 }, { "epoch": 1.98, "learning_rate": 1.6964207578542048e-05, "loss": 3.4089, "step": 15100 }, { "epoch": 2.0, "learning_rate": 1.674542749628074e-05, "loss": 3.4965, "step": 15200 }, { "epoch": 2.01, "learning_rate": 1.652664741401943e-05, "loss": 3.4984, "step": 15300 }, { "epoch": 2.02, "learning_rate": 1.6307867331758116e-05, "loss": 3.5046, "step": 15400 }, { "epoch": 2.03, "learning_rate": 1.6089087249496806e-05, "loss": 3.5016, "step": 15500 }, { "epoch": 2.05, "learning_rate": 1.5870307167235497e-05, "loss": 3.5667, "step": 15600 }, { "epoch": 2.06, "learning_rate": 1.5651527084974184e-05, "loss": 3.5435, "step": 15700 }, { "epoch": 2.07, "learning_rate": 1.5432747002712874e-05, "loss": 3.5086, "step": 15800 }, { "epoch": 2.09, "learning_rate": 1.5213966920451563e-05, "loss": 3.4831, "step": 15900 }, { "epoch": 2.1, "learning_rate": 1.499518683819025e-05, "loss": 3.3926, "step": 16000 }, { "epoch": 2.11, "learning_rate": 1.4776406755928939e-05, "loss": 3.4086, "step": 16100 }, { "epoch": 2.13, "learning_rate": 1.455762667366763e-05, "loss": 3.4987, "step": 16200 }, { "epoch": 2.14, "learning_rate": 1.4338846591406318e-05, "loss": 3.4697, "step": 16300 }, { "epoch": 2.15, "learning_rate": 1.4120066509145009e-05, "loss": 3.5231, "step": 16400 }, { "epoch": 2.17, "learning_rate": 1.3901286426883698e-05, "loss": 3.4806, "step": 16500 }, { "epoch": 2.18, "learning_rate": 1.3682506344622386e-05, "loss": 3.5263, "step": 16600 }, { "epoch": 2.19, "learning_rate": 1.3463726262361077e-05, "loss": 3.3846, "step": 16700 }, { "epoch": 2.21, "learning_rate": 1.3244946180099765e-05, "loss": 3.4415, "step": 16800 }, { "epoch": 2.22, "learning_rate": 1.3026166097838454e-05, "loss": 3.3833, "step": 16900 }, { "epoch": 2.23, "learning_rate": 1.2807386015577141e-05, "loss": 3.4666, "step": 17000 }, { "epoch": 2.24, "learning_rate": 1.258860593331583e-05, "loss": 3.4884, "step": 17100 }, { "epoch": 2.26, "learning_rate": 1.236982585105452e-05, "loss": 3.3259, "step": 17200 }, { "epoch": 2.27, "learning_rate": 1.215104576879321e-05, "loss": 3.4231, "step": 17300 }, { "epoch": 2.28, "learning_rate": 1.1932265686531898e-05, "loss": 3.4166, "step": 17400 }, { "epoch": 2.3, "learning_rate": 1.1713485604270589e-05, "loss": 3.4279, "step": 17500 }, { "epoch": 2.31, "learning_rate": 1.1494705522009276e-05, "loss": 3.4174, "step": 17600 }, { "epoch": 2.32, "learning_rate": 1.1275925439747966e-05, "loss": 3.5166, "step": 17700 }, { "epoch": 2.34, "learning_rate": 1.1057145357486655e-05, "loss": 3.4309, "step": 17800 }, { "epoch": 2.35, "learning_rate": 1.0838365275225344e-05, "loss": 3.4038, "step": 17900 }, { "epoch": 2.36, "learning_rate": 1.0619585192964034e-05, "loss": 3.3785, "step": 18000 }, { "epoch": 2.38, "learning_rate": 1.0400805110702721e-05, "loss": 3.4292, "step": 18100 }, { "epoch": 2.39, "learning_rate": 1.0182025028441412e-05, "loss": 3.4647, "step": 18200 }, { "epoch": 2.4, "learning_rate": 9.9632449461801e-06, "loss": 3.4627, "step": 18300 }, { "epoch": 2.42, "learning_rate": 9.744464863918789e-06, "loss": 3.3596, "step": 18400 }, { "epoch": 2.43, "learning_rate": 9.52568478165748e-06, "loss": 3.3967, "step": 18500 }, { "epoch": 2.44, "learning_rate": 9.306904699396167e-06, "loss": 3.4934, "step": 18600 }, { "epoch": 2.45, "learning_rate": 9.088124617134857e-06, "loss": 3.4232, "step": 18700 }, { "epoch": 2.47, "learning_rate": 8.869344534873546e-06, "loss": 3.3957, "step": 18800 }, { "epoch": 2.48, "learning_rate": 8.650564452612235e-06, "loss": 3.396, "step": 18900 }, { "epoch": 2.49, "learning_rate": 8.431784370350923e-06, "loss": 3.4012, "step": 19000 }, { "epoch": 2.51, "learning_rate": 8.213004288089612e-06, "loss": 3.3845, "step": 19100 }, { "epoch": 2.52, "learning_rate": 7.994224205828303e-06, "loss": 3.3873, "step": 19200 }, { "epoch": 2.53, "learning_rate": 7.775444123566991e-06, "loss": 3.4209, "step": 19300 }, { "epoch": 2.55, "learning_rate": 7.55666404130568e-06, "loss": 3.3871, "step": 19400 }, { "epoch": 2.56, "learning_rate": 7.337883959044369e-06, "loss": 3.4716, "step": 19500 }, { "epoch": 2.57, "learning_rate": 7.119103876783058e-06, "loss": 3.4281, "step": 19600 }, { "epoch": 2.59, "learning_rate": 6.900323794521747e-06, "loss": 3.3927, "step": 19700 }, { "epoch": 2.6, "learning_rate": 6.681543712260437e-06, "loss": 3.374, "step": 19800 }, { "epoch": 2.61, "learning_rate": 6.462763629999125e-06, "loss": 3.3668, "step": 19900 }, { "epoch": 2.63, "learning_rate": 6.2439835477378135e-06, "loss": 3.3673, "step": 20000 }, { "epoch": 2.64, "learning_rate": 6.025203465476503e-06, "loss": 3.3633, "step": 20100 }, { "epoch": 2.65, "learning_rate": 5.806423383215193e-06, "loss": 3.3235, "step": 20200 }, { "epoch": 2.66, "learning_rate": 5.5876433009538815e-06, "loss": 3.3816, "step": 20300 }, { "epoch": 2.68, "learning_rate": 5.36886321869257e-06, "loss": 3.3943, "step": 20400 }, { "epoch": 2.69, "learning_rate": 5.150083136431259e-06, "loss": 3.3534, "step": 20500 }, { "epoch": 2.7, "learning_rate": 4.931303054169949e-06, "loss": 3.4587, "step": 20600 }, { "epoch": 2.72, "learning_rate": 4.7125229719086374e-06, "loss": 3.3572, "step": 20700 }, { "epoch": 2.73, "learning_rate": 4.493742889647327e-06, "loss": 3.3056, "step": 20800 }, { "epoch": 2.74, "learning_rate": 4.274962807386016e-06, "loss": 3.3853, "step": 20900 }, { "epoch": 2.76, "learning_rate": 4.0561827251247046e-06, "loss": 3.4141, "step": 21000 }, { "epoch": 2.77, "learning_rate": 3.837402642863394e-06, "loss": 3.3369, "step": 21100 }, { "epoch": 2.78, "learning_rate": 3.618622560602083e-06, "loss": 3.366, "step": 21200 }, { "epoch": 2.8, "learning_rate": 3.399842478340772e-06, "loss": 3.4209, "step": 21300 }, { "epoch": 2.81, "learning_rate": 3.181062396079461e-06, "loss": 3.3659, "step": 21400 }, { "epoch": 2.82, "learning_rate": 2.96228231381815e-06, "loss": 3.3432, "step": 21500 }, { "epoch": 2.84, "learning_rate": 2.7435022315568393e-06, "loss": 3.4514, "step": 21600 }, { "epoch": 2.85, "learning_rate": 2.524722149295528e-06, "loss": 3.3064, "step": 21700 }, { "epoch": 2.86, "learning_rate": 2.3059420670342172e-06, "loss": 3.4154, "step": 21800 }, { "epoch": 2.87, "learning_rate": 2.0871619847729064e-06, "loss": 3.3934, "step": 21900 }, { "epoch": 2.89, "learning_rate": 1.8683819025115952e-06, "loss": 3.3657, "step": 22000 }, { "epoch": 2.9, "learning_rate": 1.6496018202502846e-06, "loss": 3.3707, "step": 22100 }, { "epoch": 2.91, "learning_rate": 1.4308217379889735e-06, "loss": 3.3213, "step": 22200 }, { "epoch": 2.93, "learning_rate": 1.2120416557276627e-06, "loss": 3.351, "step": 22300 }, { "epoch": 2.94, "learning_rate": 9.932615734663517e-07, "loss": 3.3376, "step": 22400 }, { "epoch": 2.95, "learning_rate": 7.744814912050407e-07, "loss": 3.2788, "step": 22500 }, { "epoch": 2.97, "learning_rate": 5.557014089437298e-07, "loss": 3.3773, "step": 22600 }, { "epoch": 2.98, "learning_rate": 3.3692132668241884e-07, "loss": 3.3724, "step": 22700 }, { "epoch": 2.99, "learning_rate": 1.1814124442110792e-07, "loss": 3.3345, "step": 22800 }, { "epoch": 3.0, "step": 22854, "total_flos": 1.2032870585175245e+17, "train_loss": 3.9046004770976137, "train_runtime": 22926.8282, "train_samples_per_second": 19.935, "train_steps_per_second": 0.997 } ], "max_steps": 22854, "num_train_epochs": 3, "total_flos": 1.2032870585175245e+17, "trial_name": null, "trial_params": null }