{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 107670, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 4.976780904615956e-05, "loss": 1.9554, "step": 500 }, { "epoch": 0.03, "learning_rate": 4.953561809231913e-05, "loss": 1.9046, "step": 1000 }, { "epoch": 0.04, "learning_rate": 4.9303427138478686e-05, "loss": 1.9126, "step": 1500 }, { "epoch": 0.06, "learning_rate": 4.9071236184638245e-05, "loss": 1.8888, "step": 2000 }, { "epoch": 0.07, "learning_rate": 4.883904523079781e-05, "loss": 1.8961, "step": 2500 }, { "epoch": 0.08, "learning_rate": 4.860685427695737e-05, "loss": 1.8829, "step": 3000 }, { "epoch": 0.1, "learning_rate": 4.837466332311693e-05, "loss": 1.8841, "step": 3500 }, { "epoch": 0.11, "learning_rate": 4.81424723692765e-05, "loss": 1.8742, "step": 4000 }, { "epoch": 0.13, "learning_rate": 4.791028141543606e-05, "loss": 1.8831, "step": 4500 }, { "epoch": 0.14, "learning_rate": 4.767809046159562e-05, "loss": 1.8814, "step": 5000 }, { "epoch": 0.15, "learning_rate": 4.7445899507755184e-05, "loss": 1.8688, "step": 5500 }, { "epoch": 0.17, "learning_rate": 4.721370855391474e-05, "loss": 1.8595, "step": 6000 }, { "epoch": 0.18, "learning_rate": 4.69815176000743e-05, "loss": 1.8524, "step": 6500 }, { "epoch": 0.2, "learning_rate": 4.674932664623387e-05, "loss": 1.8648, "step": 7000 }, { "epoch": 0.21, "learning_rate": 4.6517135692393426e-05, "loss": 1.8587, "step": 7500 }, { "epoch": 0.22, "learning_rate": 4.6284944738552985e-05, "loss": 1.8574, "step": 8000 }, { "epoch": 0.24, "learning_rate": 4.605275378471255e-05, "loss": 1.8662, "step": 8500 }, { "epoch": 0.25, "learning_rate": 4.582056283087211e-05, "loss": 1.8661, "step": 9000 }, { "epoch": 0.26, "learning_rate": 4.558837187703167e-05, "loss": 1.831, "step": 9500 }, { "epoch": 0.28, "learning_rate": 4.5356180923191235e-05, "loss": 1.8423, "step": 10000 }, { "epoch": 0.29, "learning_rate": 4.5123989969350793e-05, "loss": 1.8448, "step": 10500 }, { "epoch": 0.31, "learning_rate": 4.489179901551035e-05, "loss": 1.8567, "step": 11000 }, { "epoch": 0.32, "learning_rate": 4.4659608061669925e-05, "loss": 1.8425, "step": 11500 }, { "epoch": 0.33, "learning_rate": 4.4427417107829484e-05, "loss": 1.8516, "step": 12000 }, { "epoch": 0.35, "learning_rate": 4.419522615398904e-05, "loss": 1.8333, "step": 12500 }, { "epoch": 0.36, "learning_rate": 4.396303520014861e-05, "loss": 1.8588, "step": 13000 }, { "epoch": 0.38, "learning_rate": 4.373084424630817e-05, "loss": 1.8571, "step": 13500 }, { "epoch": 0.39, "learning_rate": 4.3498653292467726e-05, "loss": 1.8509, "step": 14000 }, { "epoch": 0.4, "learning_rate": 4.326646233862729e-05, "loss": 1.8604, "step": 14500 }, { "epoch": 0.42, "learning_rate": 4.303427138478685e-05, "loss": 1.8501, "step": 15000 }, { "epoch": 0.43, "learning_rate": 4.280208043094641e-05, "loss": 1.8367, "step": 15500 }, { "epoch": 0.45, "learning_rate": 4.2569889477105975e-05, "loss": 1.8487, "step": 16000 }, { "epoch": 0.46, "learning_rate": 4.2337698523265534e-05, "loss": 1.8472, "step": 16500 }, { "epoch": 0.47, "learning_rate": 4.210550756942509e-05, "loss": 1.8576, "step": 17000 }, { "epoch": 0.49, "learning_rate": 4.187331661558466e-05, "loss": 1.8554, "step": 17500 }, { "epoch": 0.5, "learning_rate": 4.164112566174422e-05, "loss": 1.8348, "step": 18000 }, { "epoch": 0.52, "learning_rate": 4.140893470790378e-05, "loss": 1.8368, "step": 18500 }, { "epoch": 0.53, "learning_rate": 4.117674375406335e-05, "loss": 1.8508, "step": 19000 }, { "epoch": 0.54, "learning_rate": 4.094455280022291e-05, "loss": 1.8349, "step": 19500 }, { "epoch": 0.56, "learning_rate": 4.071236184638247e-05, "loss": 1.8553, "step": 20000 }, { "epoch": 0.57, "learning_rate": 4.048017089254203e-05, "loss": 1.838, "step": 20500 }, { "epoch": 0.59, "learning_rate": 4.024797993870159e-05, "loss": 1.8591, "step": 21000 }, { "epoch": 0.6, "learning_rate": 4.001578898486115e-05, "loss": 1.8444, "step": 21500 }, { "epoch": 0.61, "learning_rate": 3.9783598031020716e-05, "loss": 1.8336, "step": 22000 }, { "epoch": 0.63, "learning_rate": 3.9551407077180275e-05, "loss": 1.8404, "step": 22500 }, { "epoch": 0.64, "learning_rate": 3.9319216123339834e-05, "loss": 1.8406, "step": 23000 }, { "epoch": 0.65, "learning_rate": 3.90870251694994e-05, "loss": 1.8488, "step": 23500 }, { "epoch": 0.67, "learning_rate": 3.885483421565896e-05, "loss": 1.8399, "step": 24000 }, { "epoch": 0.68, "learning_rate": 3.862264326181852e-05, "loss": 1.8331, "step": 24500 }, { "epoch": 0.7, "learning_rate": 3.839045230797808e-05, "loss": 1.8437, "step": 25000 }, { "epoch": 0.71, "learning_rate": 3.815826135413764e-05, "loss": 1.8311, "step": 25500 }, { "epoch": 0.72, "learning_rate": 3.792607040029721e-05, "loss": 1.8397, "step": 26000 }, { "epoch": 0.74, "learning_rate": 3.769387944645677e-05, "loss": 1.8535, "step": 26500 }, { "epoch": 0.75, "learning_rate": 3.746168849261633e-05, "loss": 1.8485, "step": 27000 }, { "epoch": 0.77, "learning_rate": 3.722949753877589e-05, "loss": 1.8113, "step": 27500 }, { "epoch": 0.78, "learning_rate": 3.6997306584935456e-05, "loss": 1.8148, "step": 28000 }, { "epoch": 0.79, "learning_rate": 3.6765115631095015e-05, "loss": 1.8111, "step": 28500 }, { "epoch": 0.81, "learning_rate": 3.6532924677254574e-05, "loss": 1.822, "step": 29000 }, { "epoch": 0.82, "learning_rate": 3.630073372341414e-05, "loss": 1.8385, "step": 29500 }, { "epoch": 0.84, "learning_rate": 3.60685427695737e-05, "loss": 1.8217, "step": 30000 }, { "epoch": 0.85, "learning_rate": 3.583635181573326e-05, "loss": 1.8279, "step": 30500 }, { "epoch": 0.86, "learning_rate": 3.560416086189282e-05, "loss": 1.8443, "step": 31000 }, { "epoch": 0.88, "learning_rate": 3.537196990805238e-05, "loss": 1.848, "step": 31500 }, { "epoch": 0.89, "learning_rate": 3.513977895421194e-05, "loss": 1.8408, "step": 32000 }, { "epoch": 0.91, "learning_rate": 3.490758800037151e-05, "loss": 1.8216, "step": 32500 }, { "epoch": 0.92, "learning_rate": 3.4675397046531066e-05, "loss": 1.8322, "step": 33000 }, { "epoch": 0.93, "learning_rate": 3.444320609269063e-05, "loss": 1.8282, "step": 33500 }, { "epoch": 0.95, "learning_rate": 3.42110151388502e-05, "loss": 1.8157, "step": 34000 }, { "epoch": 0.96, "learning_rate": 3.3978824185009756e-05, "loss": 1.8295, "step": 34500 }, { "epoch": 0.98, "learning_rate": 3.3746633231169315e-05, "loss": 1.82, "step": 35000 }, { "epoch": 0.99, "learning_rate": 3.351444227732888e-05, "loss": 1.8348, "step": 35500 }, { "epoch": 1.0, "learning_rate": 3.328225132348844e-05, "loss": 1.8431, "step": 36000 }, { "epoch": 1.02, "learning_rate": 3.3050060369648e-05, "loss": 1.8095, "step": 36500 }, { "epoch": 1.03, "learning_rate": 3.2817869415807564e-05, "loss": 1.8152, "step": 37000 }, { "epoch": 1.04, "learning_rate": 3.258567846196712e-05, "loss": 1.8002, "step": 37500 }, { "epoch": 1.06, "learning_rate": 3.235348750812668e-05, "loss": 1.8187, "step": 38000 }, { "epoch": 1.07, "learning_rate": 3.212129655428625e-05, "loss": 1.8142, "step": 38500 }, { "epoch": 1.09, "learning_rate": 3.1889105600445806e-05, "loss": 1.8112, "step": 39000 }, { "epoch": 1.1, "learning_rate": 3.1656914646605365e-05, "loss": 1.7863, "step": 39500 }, { "epoch": 1.11, "learning_rate": 3.142472369276493e-05, "loss": 1.8133, "step": 40000 }, { "epoch": 1.13, "learning_rate": 3.119253273892449e-05, "loss": 1.8012, "step": 40500 }, { "epoch": 1.14, "learning_rate": 3.0960341785084055e-05, "loss": 1.812, "step": 41000 }, { "epoch": 1.16, "learning_rate": 3.072815083124362e-05, "loss": 1.8144, "step": 41500 }, { "epoch": 1.17, "learning_rate": 3.049595987740318e-05, "loss": 1.7835, "step": 42000 }, { "epoch": 1.18, "learning_rate": 3.0263768923562742e-05, "loss": 1.826, "step": 42500 }, { "epoch": 1.2, "learning_rate": 3.00315779697223e-05, "loss": 1.8227, "step": 43000 }, { "epoch": 1.21, "learning_rate": 2.9799387015881863e-05, "loss": 1.7956, "step": 43500 }, { "epoch": 1.23, "learning_rate": 2.9567196062041426e-05, "loss": 1.8176, "step": 44000 }, { "epoch": 1.24, "learning_rate": 2.9335005108200985e-05, "loss": 1.8231, "step": 44500 }, { "epoch": 1.25, "learning_rate": 2.9102814154360547e-05, "loss": 1.7996, "step": 45000 }, { "epoch": 1.27, "learning_rate": 2.887062320052011e-05, "loss": 1.8013, "step": 45500 }, { "epoch": 1.28, "learning_rate": 2.8638432246679668e-05, "loss": 1.794, "step": 46000 }, { "epoch": 1.3, "learning_rate": 2.840624129283923e-05, "loss": 1.8123, "step": 46500 }, { "epoch": 1.31, "learning_rate": 2.8174050338998793e-05, "loss": 1.8006, "step": 47000 }, { "epoch": 1.32, "learning_rate": 2.794185938515835e-05, "loss": 1.8032, "step": 47500 }, { "epoch": 1.34, "learning_rate": 2.7709668431317914e-05, "loss": 1.8006, "step": 48000 }, { "epoch": 1.35, "learning_rate": 2.7477477477477483e-05, "loss": 1.8017, "step": 48500 }, { "epoch": 1.37, "learning_rate": 2.7245286523637042e-05, "loss": 1.8057, "step": 49000 }, { "epoch": 1.38, "learning_rate": 2.7013095569796604e-05, "loss": 1.783, "step": 49500 }, { "epoch": 1.39, "learning_rate": 2.6780904615956166e-05, "loss": 1.8021, "step": 50000 }, { "epoch": 1.41, "learning_rate": 2.6548713662115725e-05, "loss": 1.7841, "step": 50500 }, { "epoch": 1.42, "learning_rate": 2.6316522708275288e-05, "loss": 1.8369, "step": 51000 }, { "epoch": 1.43, "learning_rate": 2.608433175443485e-05, "loss": 1.8124, "step": 51500 }, { "epoch": 1.45, "learning_rate": 2.585214080059441e-05, "loss": 1.7967, "step": 52000 }, { "epoch": 1.46, "learning_rate": 2.561994984675397e-05, "loss": 1.8099, "step": 52500 }, { "epoch": 1.48, "learning_rate": 2.5387758892913533e-05, "loss": 1.8068, "step": 53000 }, { "epoch": 1.49, "learning_rate": 2.5155567939073092e-05, "loss": 1.81, "step": 53500 }, { "epoch": 1.5, "learning_rate": 2.4923376985232658e-05, "loss": 1.8087, "step": 54000 }, { "epoch": 1.52, "learning_rate": 2.469118603139222e-05, "loss": 1.8065, "step": 54500 }, { "epoch": 1.53, "learning_rate": 2.445899507755178e-05, "loss": 1.7961, "step": 55000 }, { "epoch": 1.55, "learning_rate": 2.422680412371134e-05, "loss": 1.7958, "step": 55500 }, { "epoch": 1.56, "learning_rate": 2.3994613169870904e-05, "loss": 1.7864, "step": 56000 }, { "epoch": 1.57, "learning_rate": 2.3762422216030463e-05, "loss": 1.8008, "step": 56500 }, { "epoch": 1.59, "learning_rate": 2.3530231262190025e-05, "loss": 1.7768, "step": 57000 }, { "epoch": 1.6, "learning_rate": 2.329804030834959e-05, "loss": 1.8006, "step": 57500 }, { "epoch": 1.62, "learning_rate": 2.306584935450915e-05, "loss": 1.7848, "step": 58000 }, { "epoch": 1.63, "learning_rate": 2.283365840066871e-05, "loss": 1.8123, "step": 58500 }, { "epoch": 1.64, "learning_rate": 2.2601467446828274e-05, "loss": 1.7992, "step": 59000 }, { "epoch": 1.66, "learning_rate": 2.2369276492987833e-05, "loss": 1.8063, "step": 59500 }, { "epoch": 1.67, "learning_rate": 2.2137085539147395e-05, "loss": 1.8029, "step": 60000 }, { "epoch": 1.69, "learning_rate": 2.1904894585306957e-05, "loss": 1.8058, "step": 60500 }, { "epoch": 1.7, "learning_rate": 2.1672703631466516e-05, "loss": 1.7953, "step": 61000 }, { "epoch": 1.71, "learning_rate": 2.1440512677626082e-05, "loss": 1.7968, "step": 61500 }, { "epoch": 1.73, "learning_rate": 2.1208321723785644e-05, "loss": 1.797, "step": 62000 }, { "epoch": 1.74, "learning_rate": 2.0976130769945203e-05, "loss": 1.8062, "step": 62500 }, { "epoch": 1.76, "learning_rate": 2.0743939816104765e-05, "loss": 1.7876, "step": 63000 }, { "epoch": 1.77, "learning_rate": 2.0511748862264328e-05, "loss": 1.817, "step": 63500 }, { "epoch": 1.78, "learning_rate": 2.0279557908423887e-05, "loss": 1.8015, "step": 64000 }, { "epoch": 1.8, "learning_rate": 2.004736695458345e-05, "loss": 1.8002, "step": 64500 }, { "epoch": 1.81, "learning_rate": 1.9815176000743015e-05, "loss": 1.8066, "step": 65000 }, { "epoch": 1.83, "learning_rate": 1.9582985046902573e-05, "loss": 1.814, "step": 65500 }, { "epoch": 1.84, "learning_rate": 1.9350794093062136e-05, "loss": 1.7941, "step": 66000 }, { "epoch": 1.85, "learning_rate": 1.9118603139221698e-05, "loss": 1.7946, "step": 66500 }, { "epoch": 1.87, "learning_rate": 1.8886412185381257e-05, "loss": 1.7813, "step": 67000 }, { "epoch": 1.88, "learning_rate": 1.865422123154082e-05, "loss": 1.7869, "step": 67500 }, { "epoch": 1.89, "learning_rate": 1.842203027770038e-05, "loss": 1.8165, "step": 68000 }, { "epoch": 1.91, "learning_rate": 1.8189839323859944e-05, "loss": 1.7859, "step": 68500 }, { "epoch": 1.92, "learning_rate": 1.7957648370019506e-05, "loss": 1.785, "step": 69000 }, { "epoch": 1.94, "learning_rate": 1.772545741617907e-05, "loss": 1.7911, "step": 69500 }, { "epoch": 1.95, "learning_rate": 1.7493266462338627e-05, "loss": 1.8297, "step": 70000 }, { "epoch": 1.96, "learning_rate": 1.726107550849819e-05, "loss": 1.8001, "step": 70500 }, { "epoch": 1.98, "learning_rate": 1.7028884554657752e-05, "loss": 1.7867, "step": 71000 }, { "epoch": 1.99, "learning_rate": 1.679669360081731e-05, "loss": 1.7958, "step": 71500 }, { "epoch": 2.01, "learning_rate": 1.6564502646976873e-05, "loss": 1.7747, "step": 72000 }, { "epoch": 2.02, "learning_rate": 1.633231169313644e-05, "loss": 1.786, "step": 72500 }, { "epoch": 2.03, "learning_rate": 1.6100120739295998e-05, "loss": 1.7847, "step": 73000 }, { "epoch": 2.05, "learning_rate": 1.586792978545556e-05, "loss": 1.78, "step": 73500 }, { "epoch": 2.06, "learning_rate": 1.5635738831615122e-05, "loss": 1.774, "step": 74000 }, { "epoch": 2.08, "learning_rate": 1.540354787777468e-05, "loss": 1.804, "step": 74500 }, { "epoch": 2.09, "learning_rate": 1.5171356923934243e-05, "loss": 1.7776, "step": 75000 }, { "epoch": 2.1, "learning_rate": 1.4939165970093804e-05, "loss": 1.8007, "step": 75500 }, { "epoch": 2.12, "learning_rate": 1.470697501625337e-05, "loss": 1.7747, "step": 76000 }, { "epoch": 2.13, "learning_rate": 1.447478406241293e-05, "loss": 1.7932, "step": 76500 }, { "epoch": 2.15, "learning_rate": 1.424259310857249e-05, "loss": 1.7718, "step": 77000 }, { "epoch": 2.16, "learning_rate": 1.4010402154732053e-05, "loss": 1.7861, "step": 77500 }, { "epoch": 2.17, "learning_rate": 1.3778211200891614e-05, "loss": 1.783, "step": 78000 }, { "epoch": 2.19, "learning_rate": 1.3546020247051174e-05, "loss": 1.7914, "step": 78500 }, { "epoch": 2.2, "learning_rate": 1.3313829293210736e-05, "loss": 1.7831, "step": 79000 }, { "epoch": 2.22, "learning_rate": 1.30816383393703e-05, "loss": 1.7912, "step": 79500 }, { "epoch": 2.23, "learning_rate": 1.2849447385529861e-05, "loss": 1.7925, "step": 80000 }, { "epoch": 2.24, "learning_rate": 1.2617256431689423e-05, "loss": 1.7918, "step": 80500 }, { "epoch": 2.26, "learning_rate": 1.2385065477848984e-05, "loss": 1.7757, "step": 81000 }, { "epoch": 2.27, "learning_rate": 1.2152874524008545e-05, "loss": 1.7817, "step": 81500 }, { "epoch": 2.28, "learning_rate": 1.1920683570168107e-05, "loss": 1.7793, "step": 82000 }, { "epoch": 2.3, "learning_rate": 1.1688492616327669e-05, "loss": 1.7921, "step": 82500 }, { "epoch": 2.31, "learning_rate": 1.145630166248723e-05, "loss": 1.7963, "step": 83000 }, { "epoch": 2.33, "learning_rate": 1.1224110708646792e-05, "loss": 1.7736, "step": 83500 }, { "epoch": 2.34, "learning_rate": 1.0991919754806353e-05, "loss": 1.7757, "step": 84000 }, { "epoch": 2.35, "learning_rate": 1.0759728800965915e-05, "loss": 1.7496, "step": 84500 }, { "epoch": 2.37, "learning_rate": 1.0527537847125477e-05, "loss": 1.8009, "step": 85000 }, { "epoch": 2.38, "learning_rate": 1.0295346893285038e-05, "loss": 1.781, "step": 85500 }, { "epoch": 2.4, "learning_rate": 1.00631559394446e-05, "loss": 1.794, "step": 86000 }, { "epoch": 2.41, "learning_rate": 9.830964985604162e-06, "loss": 1.7692, "step": 86500 }, { "epoch": 2.42, "learning_rate": 9.598774031763723e-06, "loss": 1.7747, "step": 87000 }, { "epoch": 2.44, "learning_rate": 9.366583077923283e-06, "loss": 1.7775, "step": 87500 }, { "epoch": 2.45, "learning_rate": 9.134392124082847e-06, "loss": 1.7649, "step": 88000 }, { "epoch": 2.47, "learning_rate": 8.902201170242408e-06, "loss": 1.7753, "step": 88500 }, { "epoch": 2.48, "learning_rate": 8.670010216401969e-06, "loss": 1.771, "step": 89000 }, { "epoch": 2.49, "learning_rate": 8.437819262561531e-06, "loss": 1.7666, "step": 89500 }, { "epoch": 2.51, "learning_rate": 8.205628308721093e-06, "loss": 1.7985, "step": 90000 }, { "epoch": 2.52, "learning_rate": 7.973437354880654e-06, "loss": 1.7865, "step": 90500 }, { "epoch": 2.54, "learning_rate": 7.741246401040216e-06, "loss": 1.7887, "step": 91000 }, { "epoch": 2.55, "learning_rate": 7.509055447199777e-06, "loss": 1.7825, "step": 91500 }, { "epoch": 2.56, "learning_rate": 7.27686449335934e-06, "loss": 1.7903, "step": 92000 }, { "epoch": 2.58, "learning_rate": 7.0446735395189e-06, "loss": 1.7918, "step": 92500 }, { "epoch": 2.59, "learning_rate": 6.812482585678462e-06, "loss": 1.7839, "step": 93000 }, { "epoch": 2.61, "learning_rate": 6.580291631838025e-06, "loss": 1.768, "step": 93500 }, { "epoch": 2.62, "learning_rate": 6.3481006779975855e-06, "loss": 1.7795, "step": 94000 }, { "epoch": 2.63, "learning_rate": 6.115909724157147e-06, "loss": 1.7708, "step": 94500 }, { "epoch": 2.65, "learning_rate": 5.883718770316709e-06, "loss": 1.7872, "step": 95000 }, { "epoch": 2.66, "learning_rate": 5.65152781647627e-06, "loss": 1.7933, "step": 95500 }, { "epoch": 2.67, "learning_rate": 5.419336862635832e-06, "loss": 1.7701, "step": 96000 }, { "epoch": 2.69, "learning_rate": 5.1871459087953935e-06, "loss": 1.7807, "step": 96500 }, { "epoch": 2.7, "learning_rate": 4.954954954954955e-06, "loss": 1.7973, "step": 97000 }, { "epoch": 2.72, "learning_rate": 4.722764001114516e-06, "loss": 1.8021, "step": 97500 }, { "epoch": 2.73, "learning_rate": 4.490573047274079e-06, "loss": 1.8002, "step": 98000 }, { "epoch": 2.74, "learning_rate": 4.25838209343364e-06, "loss": 1.7783, "step": 98500 }, { "epoch": 2.76, "learning_rate": 4.0261911395932016e-06, "loss": 1.784, "step": 99000 }, { "epoch": 2.77, "learning_rate": 3.7940001857527634e-06, "loss": 1.7755, "step": 99500 }, { "epoch": 2.79, "learning_rate": 3.561809231912325e-06, "loss": 1.7845, "step": 100000 }, { "epoch": 2.8, "learning_rate": 3.3296182780718867e-06, "loss": 1.791, "step": 100500 }, { "epoch": 2.81, "learning_rate": 3.097427324231448e-06, "loss": 1.7703, "step": 101000 }, { "epoch": 2.83, "learning_rate": 2.86523637039101e-06, "loss": 1.7896, "step": 101500 }, { "epoch": 2.84, "learning_rate": 2.6330454165505714e-06, "loss": 1.7663, "step": 102000 }, { "epoch": 2.86, "learning_rate": 2.400854462710133e-06, "loss": 1.7777, "step": 102500 }, { "epoch": 2.87, "learning_rate": 2.1686635088696947e-06, "loss": 1.7817, "step": 103000 }, { "epoch": 2.88, "learning_rate": 1.936472555029256e-06, "loss": 1.7943, "step": 103500 }, { "epoch": 2.9, "learning_rate": 1.7042816011888178e-06, "loss": 1.7932, "step": 104000 }, { "epoch": 2.91, "learning_rate": 1.4720906473483793e-06, "loss": 1.7895, "step": 104500 }, { "epoch": 2.93, "learning_rate": 1.239899693507941e-06, "loss": 1.772, "step": 105000 }, { "epoch": 2.94, "learning_rate": 1.0077087396675028e-06, "loss": 1.7622, "step": 105500 }, { "epoch": 2.95, "learning_rate": 7.755177858270642e-07, "loss": 1.7828, "step": 106000 }, { "epoch": 2.97, "learning_rate": 5.433268319866259e-07, "loss": 1.7889, "step": 106500 }, { "epoch": 2.98, "learning_rate": 3.1113587814618745e-07, "loss": 1.7694, "step": 107000 }, { "epoch": 3.0, "learning_rate": 7.894492430574905e-08, "loss": 1.7807, "step": 107500 }, { "epoch": 3.0, "step": 107670, "total_flos": 2.32802467336618e+17, "train_loss": 1.6543810425201497, "train_runtime": 42620.4478, "train_samples_per_second": 20.21, "train_steps_per_second": 2.526 } ], "max_steps": 107670, "num_train_epochs": 3, "total_flos": 2.32802467336618e+17, "trial_name": null, "trial_params": null }