{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 139956, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 4.9821372431335564e-05, "loss": 3.0261, "step": 500 }, { "epoch": 0.02, "learning_rate": 4.9642744862671126e-05, "loss": 2.5495, "step": 1000 }, { "epoch": 0.03, "learning_rate": 4.946411729400669e-05, "loss": 2.3681, "step": 1500 }, { "epoch": 0.04, "learning_rate": 4.9285489725342256e-05, "loss": 2.271, "step": 2000 }, { "epoch": 0.05, "learning_rate": 4.910686215667781e-05, "loss": 2.2006, "step": 2500 }, { "epoch": 0.06, "learning_rate": 4.892823458801338e-05, "loss": 2.1712, "step": 3000 }, { "epoch": 0.08, "learning_rate": 4.874960701934894e-05, "loss": 2.1011, "step": 3500 }, { "epoch": 0.09, "learning_rate": 4.85709794506845e-05, "loss": 2.0535, "step": 4000 }, { "epoch": 0.1, "learning_rate": 4.8392351882020064e-05, "loss": 2.0397, "step": 4500 }, { "epoch": 0.11, "learning_rate": 4.821372431335563e-05, "loss": 1.9944, "step": 5000 }, { "epoch": 0.12, "learning_rate": 4.8035096744691194e-05, "loss": 1.976, "step": 5500 }, { "epoch": 0.13, "learning_rate": 4.7856469176026755e-05, "loss": 1.946, "step": 6000 }, { "epoch": 0.14, "learning_rate": 4.767784160736232e-05, "loss": 1.9083, "step": 6500 }, { "epoch": 0.15, "learning_rate": 4.749921403869788e-05, "loss": 1.9224, "step": 7000 }, { "epoch": 0.16, "learning_rate": 4.732058647003344e-05, "loss": 1.907, "step": 7500 }, { "epoch": 0.17, "learning_rate": 4.7141958901369e-05, "loss": 1.8866, "step": 8000 }, { "epoch": 0.18, "learning_rate": 4.696333133270457e-05, "loss": 1.8647, "step": 8500 }, { "epoch": 0.19, "learning_rate": 4.6784703764040125e-05, "loss": 1.8733, "step": 9000 }, { "epoch": 0.2, "learning_rate": 4.660607619537569e-05, "loss": 1.8536, "step": 9500 }, { "epoch": 0.21, "learning_rate": 4.6427448626711255e-05, "loss": 1.8566, "step": 10000 }, { "epoch": 0.23, "learning_rate": 4.6248821058046816e-05, "loss": 1.8182, "step": 10500 }, { "epoch": 0.24, "learning_rate": 4.607019348938238e-05, "loss": 1.8015, "step": 11000 }, { "epoch": 0.25, "learning_rate": 4.5891565920717946e-05, "loss": 1.7759, "step": 11500 }, { "epoch": 0.26, "learning_rate": 4.57129383520535e-05, "loss": 1.796, "step": 12000 }, { "epoch": 0.27, "learning_rate": 4.553431078338907e-05, "loss": 1.7604, "step": 12500 }, { "epoch": 0.28, "learning_rate": 4.535568321472463e-05, "loss": 1.7672, "step": 13000 }, { "epoch": 0.29, "learning_rate": 4.517705564606019e-05, "loss": 1.74, "step": 13500 }, { "epoch": 0.3, "learning_rate": 4.4998428077395754e-05, "loss": 1.7683, "step": 14000 }, { "epoch": 0.31, "learning_rate": 4.4819800508731316e-05, "loss": 1.7463, "step": 14500 }, { "epoch": 0.32, "learning_rate": 4.464117294006688e-05, "loss": 1.7083, "step": 15000 }, { "epoch": 0.33, "learning_rate": 4.446254537140244e-05, "loss": 1.7211, "step": 15500 }, { "epoch": 0.34, "learning_rate": 4.428391780273801e-05, "loss": 1.7345, "step": 16000 }, { "epoch": 0.35, "learning_rate": 4.410529023407357e-05, "loss": 1.7328, "step": 16500 }, { "epoch": 0.36, "learning_rate": 4.392666266540913e-05, "loss": 1.7239, "step": 17000 }, { "epoch": 0.38, "learning_rate": 4.374803509674469e-05, "loss": 1.6754, "step": 17500 }, { "epoch": 0.39, "learning_rate": 4.356940752808026e-05, "loss": 1.7116, "step": 18000 }, { "epoch": 0.4, "learning_rate": 4.3390779959415816e-05, "loss": 1.6832, "step": 18500 }, { "epoch": 0.41, "learning_rate": 4.3212152390751384e-05, "loss": 1.692, "step": 19000 }, { "epoch": 0.42, "learning_rate": 4.3033524822086946e-05, "loss": 1.6703, "step": 19500 }, { "epoch": 0.43, "learning_rate": 4.285489725342251e-05, "loss": 1.6583, "step": 20000 }, { "epoch": 0.44, "learning_rate": 4.267626968475807e-05, "loss": 1.6815, "step": 20500 }, { "epoch": 0.45, "learning_rate": 4.249764211609363e-05, "loss": 1.6257, "step": 21000 }, { "epoch": 0.46, "learning_rate": 4.231901454742919e-05, "loss": 1.6287, "step": 21500 }, { "epoch": 0.47, "learning_rate": 4.2140386978764754e-05, "loss": 1.676, "step": 22000 }, { "epoch": 0.48, "learning_rate": 4.196175941010032e-05, "loss": 1.6403, "step": 22500 }, { "epoch": 0.49, "learning_rate": 4.178313184143588e-05, "loss": 1.644, "step": 23000 }, { "epoch": 0.5, "learning_rate": 4.1604504272771445e-05, "loss": 1.6315, "step": 23500 }, { "epoch": 0.51, "learning_rate": 4.142587670410701e-05, "loss": 1.6059, "step": 24000 }, { "epoch": 0.53, "learning_rate": 4.124724913544257e-05, "loss": 1.6353, "step": 24500 }, { "epoch": 0.54, "learning_rate": 4.106862156677813e-05, "loss": 1.6485, "step": 25000 }, { "epoch": 0.55, "learning_rate": 4.08899939981137e-05, "loss": 1.6016, "step": 25500 }, { "epoch": 0.56, "learning_rate": 4.071136642944925e-05, "loss": 1.6316, "step": 26000 }, { "epoch": 0.57, "learning_rate": 4.053273886078482e-05, "loss": 1.611, "step": 26500 }, { "epoch": 0.58, "learning_rate": 4.035411129212038e-05, "loss": 1.6171, "step": 27000 }, { "epoch": 0.59, "learning_rate": 4.0175483723455945e-05, "loss": 1.5844, "step": 27500 }, { "epoch": 0.6, "learning_rate": 3.9996856154791506e-05, "loss": 1.591, "step": 28000 }, { "epoch": 0.61, "learning_rate": 3.981822858612707e-05, "loss": 1.6129, "step": 28500 }, { "epoch": 0.62, "learning_rate": 3.9639601017462636e-05, "loss": 1.5688, "step": 29000 }, { "epoch": 0.63, "learning_rate": 3.946097344879819e-05, "loss": 1.6083, "step": 29500 }, { "epoch": 0.64, "learning_rate": 3.928234588013376e-05, "loss": 1.5859, "step": 30000 }, { "epoch": 0.65, "learning_rate": 3.910371831146932e-05, "loss": 1.5663, "step": 30500 }, { "epoch": 0.66, "learning_rate": 3.892509074280488e-05, "loss": 1.566, "step": 31000 }, { "epoch": 0.68, "learning_rate": 3.8746463174140444e-05, "loss": 1.5472, "step": 31500 }, { "epoch": 0.69, "learning_rate": 3.856783560547601e-05, "loss": 1.5689, "step": 32000 }, { "epoch": 0.7, "learning_rate": 3.838920803681157e-05, "loss": 1.5373, "step": 32500 }, { "epoch": 0.71, "learning_rate": 3.8210580468147136e-05, "loss": 1.5555, "step": 33000 }, { "epoch": 0.72, "learning_rate": 3.80319528994827e-05, "loss": 1.5219, "step": 33500 }, { "epoch": 0.73, "learning_rate": 3.785332533081826e-05, "loss": 1.5541, "step": 34000 }, { "epoch": 0.74, "learning_rate": 3.767469776215382e-05, "loss": 1.5465, "step": 34500 }, { "epoch": 0.75, "learning_rate": 3.749607019348938e-05, "loss": 1.5577, "step": 35000 }, { "epoch": 0.76, "learning_rate": 3.7317442624824944e-05, "loss": 1.5503, "step": 35500 }, { "epoch": 0.77, "learning_rate": 3.7138815056160506e-05, "loss": 1.5281, "step": 36000 }, { "epoch": 0.78, "learning_rate": 3.6960187487496074e-05, "loss": 1.5012, "step": 36500 }, { "epoch": 0.79, "learning_rate": 3.6781559918831636e-05, "loss": 1.5179, "step": 37000 }, { "epoch": 0.8, "learning_rate": 3.66029323501672e-05, "loss": 1.5161, "step": 37500 }, { "epoch": 0.81, "learning_rate": 3.642430478150276e-05, "loss": 1.5483, "step": 38000 }, { "epoch": 0.83, "learning_rate": 3.624567721283833e-05, "loss": 1.5334, "step": 38500 }, { "epoch": 0.84, "learning_rate": 3.606704964417388e-05, "loss": 1.5063, "step": 39000 }, { "epoch": 0.85, "learning_rate": 3.588842207550945e-05, "loss": 1.5023, "step": 39500 }, { "epoch": 0.86, "learning_rate": 3.570979450684501e-05, "loss": 1.5186, "step": 40000 }, { "epoch": 0.87, "learning_rate": 3.5531166938180574e-05, "loss": 1.5198, "step": 40500 }, { "epoch": 0.88, "learning_rate": 3.5352539369516135e-05, "loss": 1.5, "step": 41000 }, { "epoch": 0.89, "learning_rate": 3.51739118008517e-05, "loss": 1.5168, "step": 41500 }, { "epoch": 0.9, "learning_rate": 3.499528423218726e-05, "loss": 1.4813, "step": 42000 }, { "epoch": 0.91, "learning_rate": 3.481665666352282e-05, "loss": 1.4846, "step": 42500 }, { "epoch": 0.92, "learning_rate": 3.463802909485839e-05, "loss": 1.5029, "step": 43000 }, { "epoch": 0.93, "learning_rate": 3.445940152619394e-05, "loss": 1.4963, "step": 43500 }, { "epoch": 0.94, "learning_rate": 3.428077395752951e-05, "loss": 1.4963, "step": 44000 }, { "epoch": 0.95, "learning_rate": 3.410214638886507e-05, "loss": 1.4972, "step": 44500 }, { "epoch": 0.96, "learning_rate": 3.3923518820200635e-05, "loss": 1.4607, "step": 45000 }, { "epoch": 0.98, "learning_rate": 3.3744891251536196e-05, "loss": 1.4626, "step": 45500 }, { "epoch": 0.99, "learning_rate": 3.3566263682871765e-05, "loss": 1.4846, "step": 46000 }, { "epoch": 1.0, "learning_rate": 3.338763611420732e-05, "loss": 1.4885, "step": 46500 }, { "epoch": 1.01, "learning_rate": 3.320900854554289e-05, "loss": 1.3792, "step": 47000 }, { "epoch": 1.02, "learning_rate": 3.303038097687845e-05, "loss": 1.2978, "step": 47500 }, { "epoch": 1.03, "learning_rate": 3.285175340821401e-05, "loss": 1.3567, "step": 48000 }, { "epoch": 1.04, "learning_rate": 3.267312583954957e-05, "loss": 1.3151, "step": 48500 }, { "epoch": 1.05, "learning_rate": 3.2494498270885134e-05, "loss": 1.3377, "step": 49000 }, { "epoch": 1.06, "learning_rate": 3.23158707022207e-05, "loss": 1.3275, "step": 49500 }, { "epoch": 1.07, "learning_rate": 3.213724313355626e-05, "loss": 1.302, "step": 50000 }, { "epoch": 1.08, "learning_rate": 3.1958615564891826e-05, "loss": 1.3463, "step": 50500 }, { "epoch": 1.09, "learning_rate": 3.177998799622739e-05, "loss": 1.3089, "step": 51000 }, { "epoch": 1.1, "learning_rate": 3.160136042756295e-05, "loss": 1.3273, "step": 51500 }, { "epoch": 1.11, "learning_rate": 3.142273285889851e-05, "loss": 1.3207, "step": 52000 }, { "epoch": 1.13, "learning_rate": 3.124410529023408e-05, "loss": 1.3121, "step": 52500 }, { "epoch": 1.14, "learning_rate": 3.1065477721569634e-05, "loss": 1.3191, "step": 53000 }, { "epoch": 1.15, "learning_rate": 3.08868501529052e-05, "loss": 1.3268, "step": 53500 }, { "epoch": 1.16, "learning_rate": 3.0708222584240764e-05, "loss": 1.3132, "step": 54000 }, { "epoch": 1.17, "learning_rate": 3.0529595015576326e-05, "loss": 1.3385, "step": 54500 }, { "epoch": 1.18, "learning_rate": 3.0350967446911887e-05, "loss": 1.3278, "step": 55000 }, { "epoch": 1.19, "learning_rate": 3.0172339878247452e-05, "loss": 1.3036, "step": 55500 }, { "epoch": 1.2, "learning_rate": 2.999371230958301e-05, "loss": 1.3253, "step": 56000 }, { "epoch": 1.21, "learning_rate": 2.9815084740918575e-05, "loss": 1.3114, "step": 56500 }, { "epoch": 1.22, "learning_rate": 2.963645717225414e-05, "loss": 1.325, "step": 57000 }, { "epoch": 1.23, "learning_rate": 2.94578296035897e-05, "loss": 1.3047, "step": 57500 }, { "epoch": 1.24, "learning_rate": 2.9279202034925264e-05, "loss": 1.3318, "step": 58000 }, { "epoch": 1.25, "learning_rate": 2.910057446626083e-05, "loss": 1.3265, "step": 58500 }, { "epoch": 1.26, "learning_rate": 2.892194689759639e-05, "loss": 1.3102, "step": 59000 }, { "epoch": 1.28, "learning_rate": 2.8743319328931952e-05, "loss": 1.3188, "step": 59500 }, { "epoch": 1.29, "learning_rate": 2.8564691760267513e-05, "loss": 1.3312, "step": 60000 }, { "epoch": 1.3, "learning_rate": 2.838606419160308e-05, "loss": 1.3176, "step": 60500 }, { "epoch": 1.31, "learning_rate": 2.8207436622938637e-05, "loss": 1.3077, "step": 61000 }, { "epoch": 1.32, "learning_rate": 2.80288090542742e-05, "loss": 1.2991, "step": 61500 }, { "epoch": 1.33, "learning_rate": 2.7850181485609767e-05, "loss": 1.2972, "step": 62000 }, { "epoch": 1.34, "learning_rate": 2.7671553916945325e-05, "loss": 1.3018, "step": 62500 }, { "epoch": 1.35, "learning_rate": 2.749292634828089e-05, "loss": 1.2887, "step": 63000 }, { "epoch": 1.36, "learning_rate": 2.7314298779616455e-05, "loss": 1.3238, "step": 63500 }, { "epoch": 1.37, "learning_rate": 2.7135671210952013e-05, "loss": 1.3005, "step": 64000 }, { "epoch": 1.38, "learning_rate": 2.6957043642287578e-05, "loss": 1.3014, "step": 64500 }, { "epoch": 1.39, "learning_rate": 2.6778416073623143e-05, "loss": 1.3141, "step": 65000 }, { "epoch": 1.4, "learning_rate": 2.65997885049587e-05, "loss": 1.2871, "step": 65500 }, { "epoch": 1.41, "learning_rate": 2.6421160936294266e-05, "loss": 1.2847, "step": 66000 }, { "epoch": 1.43, "learning_rate": 2.6242533367629828e-05, "loss": 1.2816, "step": 66500 }, { "epoch": 1.44, "learning_rate": 2.606390579896539e-05, "loss": 1.2852, "step": 67000 }, { "epoch": 1.45, "learning_rate": 2.588527823030095e-05, "loss": 1.303, "step": 67500 }, { "epoch": 1.46, "learning_rate": 2.5706650661636516e-05, "loss": 1.3148, "step": 68000 }, { "epoch": 1.47, "learning_rate": 2.5528023092972074e-05, "loss": 1.2834, "step": 68500 }, { "epoch": 1.48, "learning_rate": 2.534939552430764e-05, "loss": 1.3065, "step": 69000 }, { "epoch": 1.49, "learning_rate": 2.5170767955643204e-05, "loss": 1.2981, "step": 69500 }, { "epoch": 1.5, "learning_rate": 2.4992140386978766e-05, "loss": 1.3008, "step": 70000 }, { "epoch": 1.51, "learning_rate": 2.4813512818314327e-05, "loss": 1.2784, "step": 70500 }, { "epoch": 1.52, "learning_rate": 2.4634885249649892e-05, "loss": 1.2789, "step": 71000 }, { "epoch": 1.53, "learning_rate": 2.4456257680985454e-05, "loss": 1.2885, "step": 71500 }, { "epoch": 1.54, "learning_rate": 2.427763011232102e-05, "loss": 1.2976, "step": 72000 }, { "epoch": 1.55, "learning_rate": 2.409900254365658e-05, "loss": 1.3013, "step": 72500 }, { "epoch": 1.56, "learning_rate": 2.3920374974992142e-05, "loss": 1.2607, "step": 73000 }, { "epoch": 1.58, "learning_rate": 2.3741747406327704e-05, "loss": 1.2783, "step": 73500 }, { "epoch": 1.59, "learning_rate": 2.3563119837663265e-05, "loss": 1.2787, "step": 74000 }, { "epoch": 1.6, "learning_rate": 2.3384492268998827e-05, "loss": 1.2649, "step": 74500 }, { "epoch": 1.61, "learning_rate": 2.3205864700334392e-05, "loss": 1.2551, "step": 75000 }, { "epoch": 1.62, "learning_rate": 2.3027237131669954e-05, "loss": 1.2707, "step": 75500 }, { "epoch": 1.63, "learning_rate": 2.2848609563005515e-05, "loss": 1.2562, "step": 76000 }, { "epoch": 1.64, "learning_rate": 2.266998199434108e-05, "loss": 1.2677, "step": 76500 }, { "epoch": 1.65, "learning_rate": 2.2491354425676642e-05, "loss": 1.3022, "step": 77000 }, { "epoch": 1.66, "learning_rate": 2.2312726857012207e-05, "loss": 1.2621, "step": 77500 }, { "epoch": 1.67, "learning_rate": 2.213409928834777e-05, "loss": 1.3236, "step": 78000 }, { "epoch": 1.68, "learning_rate": 2.195547171968333e-05, "loss": 1.2718, "step": 78500 }, { "epoch": 1.69, "learning_rate": 2.1776844151018895e-05, "loss": 1.2561, "step": 79000 }, { "epoch": 1.7, "learning_rate": 2.1598216582354457e-05, "loss": 1.2687, "step": 79500 }, { "epoch": 1.71, "learning_rate": 2.1419589013690018e-05, "loss": 1.266, "step": 80000 }, { "epoch": 1.73, "learning_rate": 2.124096144502558e-05, "loss": 1.2539, "step": 80500 }, { "epoch": 1.74, "learning_rate": 2.106233387636114e-05, "loss": 1.2381, "step": 81000 }, { "epoch": 1.75, "learning_rate": 2.0883706307696703e-05, "loss": 1.251, "step": 81500 }, { "epoch": 1.76, "learning_rate": 2.0705078739032268e-05, "loss": 1.2571, "step": 82000 }, { "epoch": 1.77, "learning_rate": 2.052645117036783e-05, "loss": 1.2317, "step": 82500 }, { "epoch": 1.78, "learning_rate": 2.0347823601703395e-05, "loss": 1.2557, "step": 83000 }, { "epoch": 1.79, "learning_rate": 2.0169196033038956e-05, "loss": 1.2681, "step": 83500 }, { "epoch": 1.8, "learning_rate": 1.9990568464374518e-05, "loss": 1.2353, "step": 84000 }, { "epoch": 1.81, "learning_rate": 1.9811940895710083e-05, "loss": 1.2607, "step": 84500 }, { "epoch": 1.82, "learning_rate": 1.9633313327045644e-05, "loss": 1.2359, "step": 85000 }, { "epoch": 1.83, "learning_rate": 1.9454685758381206e-05, "loss": 1.2755, "step": 85500 }, { "epoch": 1.84, "learning_rate": 1.927605818971677e-05, "loss": 1.2528, "step": 86000 }, { "epoch": 1.85, "learning_rate": 1.9097430621052333e-05, "loss": 1.262, "step": 86500 }, { "epoch": 1.86, "learning_rate": 1.8918803052387894e-05, "loss": 1.2371, "step": 87000 }, { "epoch": 1.88, "learning_rate": 1.8740175483723456e-05, "loss": 1.2462, "step": 87500 }, { "epoch": 1.89, "learning_rate": 1.8561547915059017e-05, "loss": 1.2346, "step": 88000 }, { "epoch": 1.9, "learning_rate": 1.8382920346394582e-05, "loss": 1.2477, "step": 88500 }, { "epoch": 1.91, "learning_rate": 1.8204292777730144e-05, "loss": 1.2425, "step": 89000 }, { "epoch": 1.92, "learning_rate": 1.8025665209065706e-05, "loss": 1.2411, "step": 89500 }, { "epoch": 1.93, "learning_rate": 1.784703764040127e-05, "loss": 1.2254, "step": 90000 }, { "epoch": 1.94, "learning_rate": 1.7668410071736832e-05, "loss": 1.256, "step": 90500 }, { "epoch": 1.95, "learning_rate": 1.7489782503072394e-05, "loss": 1.2275, "step": 91000 }, { "epoch": 1.96, "learning_rate": 1.731115493440796e-05, "loss": 1.2344, "step": 91500 }, { "epoch": 1.97, "learning_rate": 1.713252736574352e-05, "loss": 1.2323, "step": 92000 }, { "epoch": 1.98, "learning_rate": 1.6953899797079085e-05, "loss": 1.2328, "step": 92500 }, { "epoch": 1.99, "learning_rate": 1.6775272228414647e-05, "loss": 1.2317, "step": 93000 }, { "epoch": 2.0, "learning_rate": 1.659664465975021e-05, "loss": 1.1778, "step": 93500 }, { "epoch": 2.01, "learning_rate": 1.641801709108577e-05, "loss": 1.0866, "step": 94000 }, { "epoch": 2.03, "learning_rate": 1.6239389522421332e-05, "loss": 1.1002, "step": 94500 }, { "epoch": 2.04, "learning_rate": 1.6060761953756893e-05, "loss": 1.0784, "step": 95000 }, { "epoch": 2.05, "learning_rate": 1.588213438509246e-05, "loss": 1.0747, "step": 95500 }, { "epoch": 2.06, "learning_rate": 1.570350681642802e-05, "loss": 1.0686, "step": 96000 }, { "epoch": 2.07, "learning_rate": 1.552487924776358e-05, "loss": 1.0706, "step": 96500 }, { "epoch": 2.08, "learning_rate": 1.5346251679099147e-05, "loss": 1.0922, "step": 97000 }, { "epoch": 2.09, "learning_rate": 1.5167624110434708e-05, "loss": 1.0904, "step": 97500 }, { "epoch": 2.1, "learning_rate": 1.4988996541770273e-05, "loss": 1.095, "step": 98000 }, { "epoch": 2.11, "learning_rate": 1.4810368973105835e-05, "loss": 1.1084, "step": 98500 }, { "epoch": 2.12, "learning_rate": 1.4631741404441396e-05, "loss": 1.0883, "step": 99000 }, { "epoch": 2.13, "learning_rate": 1.445311383577696e-05, "loss": 1.0834, "step": 99500 }, { "epoch": 2.14, "learning_rate": 1.4274486267112521e-05, "loss": 1.081, "step": 100000 }, { "epoch": 2.15, "learning_rate": 1.4095858698448083e-05, "loss": 1.0732, "step": 100500 }, { "epoch": 2.16, "learning_rate": 1.3917231129783648e-05, "loss": 1.0743, "step": 101000 }, { "epoch": 2.18, "learning_rate": 1.373860356111921e-05, "loss": 1.0873, "step": 101500 }, { "epoch": 2.19, "learning_rate": 1.3559975992454771e-05, "loss": 1.0733, "step": 102000 }, { "epoch": 2.2, "learning_rate": 1.3381348423790336e-05, "loss": 1.0872, "step": 102500 }, { "epoch": 2.21, "learning_rate": 1.3202720855125898e-05, "loss": 1.082, "step": 103000 }, { "epoch": 2.22, "learning_rate": 1.3024093286461461e-05, "loss": 1.1001, "step": 103500 }, { "epoch": 2.23, "learning_rate": 1.2845465717797023e-05, "loss": 1.0692, "step": 104000 }, { "epoch": 2.24, "learning_rate": 1.2666838149132584e-05, "loss": 1.0768, "step": 104500 }, { "epoch": 2.25, "learning_rate": 1.2488210580468147e-05, "loss": 1.0801, "step": 105000 }, { "epoch": 2.26, "learning_rate": 1.230958301180371e-05, "loss": 1.0833, "step": 105500 }, { "epoch": 2.27, "learning_rate": 1.2130955443139274e-05, "loss": 1.0527, "step": 106000 }, { "epoch": 2.28, "learning_rate": 1.1952327874474836e-05, "loss": 1.0759, "step": 106500 }, { "epoch": 2.29, "learning_rate": 1.1773700305810397e-05, "loss": 1.0649, "step": 107000 }, { "epoch": 2.3, "learning_rate": 1.159507273714596e-05, "loss": 1.0795, "step": 107500 }, { "epoch": 2.32, "learning_rate": 1.1416445168481524e-05, "loss": 1.07, "step": 108000 }, { "epoch": 2.33, "learning_rate": 1.1237817599817085e-05, "loss": 1.0656, "step": 108500 }, { "epoch": 2.34, "learning_rate": 1.1059190031152649e-05, "loss": 1.0862, "step": 109000 }, { "epoch": 2.35, "learning_rate": 1.0880562462488212e-05, "loss": 1.0656, "step": 109500 }, { "epoch": 2.36, "learning_rate": 1.0701934893823774e-05, "loss": 1.0504, "step": 110000 }, { "epoch": 2.37, "learning_rate": 1.0523307325159335e-05, "loss": 1.0759, "step": 110500 }, { "epoch": 2.38, "learning_rate": 1.0344679756494899e-05, "loss": 1.0773, "step": 111000 }, { "epoch": 2.39, "learning_rate": 1.0166052187830462e-05, "loss": 1.0403, "step": 111500 }, { "epoch": 2.4, "learning_rate": 9.987424619166023e-06, "loss": 1.0669, "step": 112000 }, { "epoch": 2.41, "learning_rate": 9.808797050501587e-06, "loss": 1.0712, "step": 112500 }, { "epoch": 2.42, "learning_rate": 9.63016948183715e-06, "loss": 1.0814, "step": 113000 }, { "epoch": 2.43, "learning_rate": 9.451541913172712e-06, "loss": 1.0658, "step": 113500 }, { "epoch": 2.44, "learning_rate": 9.272914344508273e-06, "loss": 1.0647, "step": 114000 }, { "epoch": 2.45, "learning_rate": 9.094286775843837e-06, "loss": 1.0753, "step": 114500 }, { "epoch": 2.47, "learning_rate": 8.9156592071794e-06, "loss": 1.0402, "step": 115000 }, { "epoch": 2.48, "learning_rate": 8.737031638514963e-06, "loss": 1.0489, "step": 115500 }, { "epoch": 2.49, "learning_rate": 8.558404069850525e-06, "loss": 1.0638, "step": 116000 }, { "epoch": 2.5, "learning_rate": 8.379776501186088e-06, "loss": 1.0598, "step": 116500 }, { "epoch": 2.51, "learning_rate": 8.20114893252165e-06, "loss": 1.0777, "step": 117000 }, { "epoch": 2.52, "learning_rate": 8.022521363857213e-06, "loss": 1.0479, "step": 117500 }, { "epoch": 2.53, "learning_rate": 7.843893795192775e-06, "loss": 1.072, "step": 118000 }, { "epoch": 2.54, "learning_rate": 7.665266226528338e-06, "loss": 1.0677, "step": 118500 }, { "epoch": 2.55, "learning_rate": 7.486638657863901e-06, "loss": 1.0523, "step": 119000 }, { "epoch": 2.56, "learning_rate": 7.308011089199463e-06, "loss": 1.0766, "step": 119500 }, { "epoch": 2.57, "learning_rate": 7.129383520535025e-06, "loss": 1.0679, "step": 120000 }, { "epoch": 2.58, "learning_rate": 6.9507559518705885e-06, "loss": 1.0589, "step": 120500 }, { "epoch": 2.59, "learning_rate": 6.772128383206151e-06, "loss": 1.062, "step": 121000 }, { "epoch": 2.6, "learning_rate": 6.5935008145417126e-06, "loss": 1.0595, "step": 121500 }, { "epoch": 2.62, "learning_rate": 6.414873245877276e-06, "loss": 1.061, "step": 122000 }, { "epoch": 2.63, "learning_rate": 6.236245677212839e-06, "loss": 1.0784, "step": 122500 }, { "epoch": 2.64, "learning_rate": 6.057618108548401e-06, "loss": 1.0673, "step": 123000 }, { "epoch": 2.65, "learning_rate": 5.878990539883964e-06, "loss": 1.0332, "step": 123500 }, { "epoch": 2.66, "learning_rate": 5.7003629712195265e-06, "loss": 1.0613, "step": 124000 }, { "epoch": 2.67, "learning_rate": 5.521735402555089e-06, "loss": 1.0486, "step": 124500 }, { "epoch": 2.68, "learning_rate": 5.343107833890651e-06, "loss": 1.0545, "step": 125000 }, { "epoch": 2.69, "learning_rate": 5.164480265226214e-06, "loss": 1.0729, "step": 125500 }, { "epoch": 2.7, "learning_rate": 4.985852696561777e-06, "loss": 1.065, "step": 126000 }, { "epoch": 2.71, "learning_rate": 4.807225127897339e-06, "loss": 1.0383, "step": 126500 }, { "epoch": 2.72, "learning_rate": 4.628597559232902e-06, "loss": 1.0382, "step": 127000 }, { "epoch": 2.73, "learning_rate": 4.4499699905684645e-06, "loss": 1.0669, "step": 127500 }, { "epoch": 2.74, "learning_rate": 4.271342421904028e-06, "loss": 1.047, "step": 128000 }, { "epoch": 2.75, "learning_rate": 4.092714853239589e-06, "loss": 1.0383, "step": 128500 }, { "epoch": 2.77, "learning_rate": 3.914087284575153e-06, "loss": 1.0462, "step": 129000 }, { "epoch": 2.78, "learning_rate": 3.7354597159107147e-06, "loss": 1.0378, "step": 129500 }, { "epoch": 2.79, "learning_rate": 3.556832147246278e-06, "loss": 1.0451, "step": 130000 }, { "epoch": 2.8, "learning_rate": 3.37820457858184e-06, "loss": 1.0346, "step": 130500 }, { "epoch": 2.81, "learning_rate": 3.1995770099174025e-06, "loss": 1.0372, "step": 131000 }, { "epoch": 2.82, "learning_rate": 3.0209494412529654e-06, "loss": 1.0318, "step": 131500 }, { "epoch": 2.83, "learning_rate": 2.8423218725885283e-06, "loss": 1.0778, "step": 132000 }, { "epoch": 2.84, "learning_rate": 2.6636943039240907e-06, "loss": 1.0489, "step": 132500 }, { "epoch": 2.85, "learning_rate": 2.485066735259653e-06, "loss": 1.047, "step": 133000 }, { "epoch": 2.86, "learning_rate": 2.306439166595216e-06, "loss": 1.0426, "step": 133500 }, { "epoch": 2.87, "learning_rate": 2.127811597930778e-06, "loss": 1.0594, "step": 134000 }, { "epoch": 2.88, "learning_rate": 1.949184029266341e-06, "loss": 1.0276, "step": 134500 }, { "epoch": 2.89, "learning_rate": 1.7705564606019034e-06, "loss": 1.0146, "step": 135000 }, { "epoch": 2.9, "learning_rate": 1.591928891937466e-06, "loss": 1.0492, "step": 135500 }, { "epoch": 2.92, "learning_rate": 1.4133013232730287e-06, "loss": 1.0447, "step": 136000 }, { "epoch": 2.93, "learning_rate": 1.2346737546085914e-06, "loss": 1.0454, "step": 136500 }, { "epoch": 2.94, "learning_rate": 1.056046185944154e-06, "loss": 1.037, "step": 137000 }, { "epoch": 2.95, "learning_rate": 8.774186172797166e-07, "loss": 1.0339, "step": 137500 }, { "epoch": 2.96, "learning_rate": 6.987910486152791e-07, "loss": 1.0408, "step": 138000 }, { "epoch": 2.97, "learning_rate": 5.201634799508417e-07, "loss": 1.028, "step": 138500 }, { "epoch": 2.98, "learning_rate": 3.415359112864043e-07, "loss": 1.0336, "step": 139000 }, { "epoch": 2.99, "learning_rate": 1.629083426219669e-07, "loss": 1.0233, "step": 139500 }, { "epoch": 3.0, "step": 139956, "total_flos": 2.7345432343830528e+17, "train_loss": 1.3522833067220121, "train_runtime": 43996.1722, "train_samples_per_second": 31.811, "train_steps_per_second": 3.181 } ], "max_steps": 139956, "num_train_epochs": 3, "total_flos": 2.7345432343830528e+17, "trial_name": null, "trial_params": null }