{
  "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": 3.0407,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9642744862671126e-05,
      "loss": 2.1865,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.946411729400669e-05,
      "loss": 1.9141,
      "step": 1500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9285489725342256e-05,
      "loss": 1.766,
      "step": 2000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.910686215667781e-05,
      "loss": 1.6542,
      "step": 2500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.892823458801338e-05,
      "loss": 1.6047,
      "step": 3000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.874960701934894e-05,
      "loss": 1.53,
      "step": 3500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.85709794506845e-05,
      "loss": 1.4824,
      "step": 4000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8392351882020064e-05,
      "loss": 1.4351,
      "step": 4500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.821372431335563e-05,
      "loss": 1.405,
      "step": 5000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8035096744691194e-05,
      "loss": 1.3847,
      "step": 5500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7856469176026755e-05,
      "loss": 1.3464,
      "step": 6000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.767784160736232e-05,
      "loss": 1.3163,
      "step": 6500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.749921403869788e-05,
      "loss": 1.3101,
      "step": 7000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.732058647003344e-05,
      "loss": 1.3054,
      "step": 7500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.7141958901369e-05,
      "loss": 1.2715,
      "step": 8000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.696333133270457e-05,
      "loss": 1.2606,
      "step": 8500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.6784703764040125e-05,
      "loss": 1.2423,
      "step": 9000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.660607619537569e-05,
      "loss": 1.2302,
      "step": 9500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.6427448626711255e-05,
      "loss": 1.2361,
      "step": 10000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.6248821058046816e-05,
      "loss": 1.2078,
      "step": 10500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.607019348938238e-05,
      "loss": 1.1992,
      "step": 11000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.5891565920717946e-05,
      "loss": 1.1775,
      "step": 11500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.57129383520535e-05,
      "loss": 1.1808,
      "step": 12000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.553431078338907e-05,
      "loss": 1.1644,
      "step": 12500
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.535568321472463e-05,
      "loss": 1.1504,
      "step": 13000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.517705564606019e-05,
      "loss": 1.1445,
      "step": 13500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.4998428077395754e-05,
      "loss": 1.1474,
      "step": 14000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.4819800508731316e-05,
      "loss": 1.1255,
      "step": 14500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.464117294006688e-05,
      "loss": 1.1065,
      "step": 15000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.446254537140244e-05,
      "loss": 1.1128,
      "step": 15500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.428391780273801e-05,
      "loss": 1.1114,
      "step": 16000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.410529023407357e-05,
      "loss": 1.106,
      "step": 16500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.392666266540913e-05,
      "loss": 1.1163,
      "step": 17000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.374803509674469e-05,
      "loss": 1.0843,
      "step": 17500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.356940752808026e-05,
      "loss": 1.099,
      "step": 18000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.3390779959415816e-05,
      "loss": 1.0785,
      "step": 18500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.3212152390751384e-05,
      "loss": 1.0842,
      "step": 19000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.3033524822086946e-05,
      "loss": 1.0623,
      "step": 19500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.285489725342251e-05,
      "loss": 1.0563,
      "step": 20000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.267626968475807e-05,
      "loss": 1.0568,
      "step": 20500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.249764211609363e-05,
      "loss": 1.0363,
      "step": 21000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.231901454742919e-05,
      "loss": 1.0361,
      "step": 21500
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.2140386978764754e-05,
      "loss": 1.0522,
      "step": 22000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.196175941010032e-05,
      "loss": 1.0287,
      "step": 22500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.178313184143588e-05,
      "loss": 1.0361,
      "step": 23000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1604504272771445e-05,
      "loss": 1.036,
      "step": 23500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.142587670410701e-05,
      "loss": 1.0123,
      "step": 24000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.124724913544257e-05,
      "loss": 1.0257,
      "step": 24500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.106862156677813e-05,
      "loss": 1.0405,
      "step": 25000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.08899939981137e-05,
      "loss": 1.0048,
      "step": 25500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.071136642944925e-05,
      "loss": 1.0212,
      "step": 26000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.053273886078482e-05,
      "loss": 1.009,
      "step": 26500
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.035411129212038e-05,
      "loss": 0.9971,
      "step": 27000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.0175483723455945e-05,
      "loss": 0.9901,
      "step": 27500
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.9996856154791506e-05,
      "loss": 1.0051,
      "step": 28000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.981822858612707e-05,
      "loss": 1.0079,
      "step": 28500
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.9639601017462636e-05,
      "loss": 0.9895,
      "step": 29000
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.946097344879819e-05,
      "loss": 0.9962,
      "step": 29500
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.928234588013376e-05,
      "loss": 0.9996,
      "step": 30000
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.910371831146932e-05,
      "loss": 0.9601,
      "step": 30500
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.892509074280488e-05,
      "loss": 0.9753,
      "step": 31000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.8746463174140444e-05,
      "loss": 0.9694,
      "step": 31500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.856783560547601e-05,
      "loss": 0.9772,
      "step": 32000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.838920803681157e-05,
      "loss": 0.9499,
      "step": 32500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.8210580468147136e-05,
      "loss": 0.9681,
      "step": 33000
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.80319528994827e-05,
      "loss": 0.9411,
      "step": 33500
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.785332533081826e-05,
      "loss": 0.9631,
      "step": 34000
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.767469776215382e-05,
      "loss": 0.9602,
      "step": 34500
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.749607019348938e-05,
      "loss": 0.9639,
      "step": 35000
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.7317442624824944e-05,
      "loss": 0.9624,
      "step": 35500
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.7138815056160506e-05,
      "loss": 0.9317,
      "step": 36000
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.6960187487496074e-05,
      "loss": 0.9261,
      "step": 36500
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.6781559918831636e-05,
      "loss": 0.934,
      "step": 37000
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.66029323501672e-05,
      "loss": 0.9347,
      "step": 37500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.642430478150276e-05,
      "loss": 0.9525,
      "step": 38000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.624567721283833e-05,
      "loss": 0.9436,
      "step": 38500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.606704964417388e-05,
      "loss": 0.9286,
      "step": 39000
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.588842207550945e-05,
      "loss": 0.9251,
      "step": 39500
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.570979450684501e-05,
      "loss": 0.9306,
      "step": 40000
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.5531166938180574e-05,
      "loss": 0.9318,
      "step": 40500
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.5352539369516135e-05,
      "loss": 0.9208,
      "step": 41000
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.51739118008517e-05,
      "loss": 0.9259,
      "step": 41500
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.499528423218726e-05,
      "loss": 0.9019,
      "step": 42000
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.481665666352282e-05,
      "loss": 0.9123,
      "step": 42500
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.463802909485839e-05,
      "loss": 0.9135,
      "step": 43000
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.445940152619394e-05,
      "loss": 0.9216,
      "step": 43500
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.428077395752951e-05,
      "loss": 0.9109,
      "step": 44000
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.410214638886507e-05,
      "loss": 0.9165,
      "step": 44500
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.3923518820200635e-05,
      "loss": 0.8947,
      "step": 45000
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.3744891251536196e-05,
      "loss": 0.8987,
      "step": 45500
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.3566263682871765e-05,
      "loss": 0.9046,
      "step": 46000
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.338763611420732e-05,
      "loss": 0.9059,
      "step": 46500
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.320900854554289e-05,
      "loss": 0.8544,
      "step": 47000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.303038097687845e-05,
      "loss": 0.8138,
      "step": 47500
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.285175340821401e-05,
      "loss": 0.8298,
      "step": 48000
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.267312583954957e-05,
      "loss": 0.8221,
      "step": 48500
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.2494498270885134e-05,
      "loss": 0.8253,
      "step": 49000
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.23158707022207e-05,
      "loss": 0.8283,
      "step": 49500
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.213724313355626e-05,
      "loss": 0.8233,
      "step": 50000
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.1958615564891826e-05,
      "loss": 0.8352,
      "step": 50500
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.177998799622739e-05,
      "loss": 0.8091,
      "step": 51000
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.160136042756295e-05,
      "loss": 0.8165,
      "step": 51500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.142273285889851e-05,
      "loss": 0.8138,
      "step": 52000
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.124410529023408e-05,
      "loss": 0.813,
      "step": 52500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.1065477721569634e-05,
      "loss": 0.8255,
      "step": 53000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.08868501529052e-05,
      "loss": 0.8225,
      "step": 53500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.0708222584240764e-05,
      "loss": 0.8126,
      "step": 54000
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.0529595015576326e-05,
      "loss": 0.8222,
      "step": 54500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0350967446911887e-05,
      "loss": 0.8229,
      "step": 55000
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.0172339878247452e-05,
      "loss": 0.8082,
      "step": 55500
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.999371230958301e-05,
      "loss": 0.8134,
      "step": 56000
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.9815084740918575e-05,
      "loss": 0.802,
      "step": 56500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.963645717225414e-05,
      "loss": 0.8113,
      "step": 57000
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.94578296035897e-05,
      "loss": 0.8032,
      "step": 57500
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9279202034925264e-05,
      "loss": 0.815,
      "step": 58000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.910057446626083e-05,
      "loss": 0.8118,
      "step": 58500
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.892194689759639e-05,
      "loss": 0.8076,
      "step": 59000
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8743319328931952e-05,
      "loss": 0.8103,
      "step": 59500
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8564691760267513e-05,
      "loss": 0.8171,
      "step": 60000
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.838606419160308e-05,
      "loss": 0.8126,
      "step": 60500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8207436622938637e-05,
      "loss": 0.8106,
      "step": 61000
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.80288090542742e-05,
      "loss": 0.803,
      "step": 61500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7850181485609767e-05,
      "loss": 0.7868,
      "step": 62000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7671553916945325e-05,
      "loss": 0.8008,
      "step": 62500
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.749292634828089e-05,
      "loss": 0.8002,
      "step": 63000
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7314298779616455e-05,
      "loss": 0.8122,
      "step": 63500
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.7135671210952013e-05,
      "loss": 0.7968,
      "step": 64000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.6957043642287578e-05,
      "loss": 0.7949,
      "step": 64500
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6778416073623143e-05,
      "loss": 0.7983,
      "step": 65000
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.65997885049587e-05,
      "loss": 0.796,
      "step": 65500
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6421160936294266e-05,
      "loss": 0.7916,
      "step": 66000
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6242533367629828e-05,
      "loss": 0.7874,
      "step": 66500
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.606390579896539e-05,
      "loss": 0.79,
      "step": 67000
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.588527823030095e-05,
      "loss": 0.8037,
      "step": 67500
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5706650661636516e-05,
      "loss": 0.8038,
      "step": 68000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5528023092972074e-05,
      "loss": 0.7896,
      "step": 68500
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.534939552430764e-05,
      "loss": 0.7999,
      "step": 69000
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.5170767955643204e-05,
      "loss": 0.7911,
      "step": 69500
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.4992140386978766e-05,
      "loss": 0.7965,
      "step": 70000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4813512818314327e-05,
      "loss": 0.7814,
      "step": 70500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4634885249649892e-05,
      "loss": 0.7759,
      "step": 71000
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4456257680985454e-05,
      "loss": 0.7906,
      "step": 71500
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.427763011232102e-05,
      "loss": 0.7911,
      "step": 72000
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.409900254365658e-05,
      "loss": 0.7933,
      "step": 72500
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.3920374974992142e-05,
      "loss": 0.772,
      "step": 73000
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3741747406327704e-05,
      "loss": 0.7895,
      "step": 73500
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3563119837663265e-05,
      "loss": 0.7757,
      "step": 74000
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3384492268998827e-05,
      "loss": 0.7771,
      "step": 74500
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3205864700334392e-05,
      "loss": 0.7775,
      "step": 75000
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.3027237131669954e-05,
      "loss": 0.7788,
      "step": 75500
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2848609563005515e-05,
      "loss": 0.773,
      "step": 76000
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.266998199434108e-05,
      "loss": 0.7776,
      "step": 76500
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2491354425676642e-05,
      "loss": 0.7982,
      "step": 77000
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2312726857012207e-05,
      "loss": 0.777,
      "step": 77500
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.213409928834777e-05,
      "loss": 0.7965,
      "step": 78000
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.195547171968333e-05,
      "loss": 0.7835,
      "step": 78500
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1776844151018895e-05,
      "loss": 0.767,
      "step": 79000
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.1598216582354457e-05,
      "loss": 0.7714,
      "step": 79500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1419589013690018e-05,
      "loss": 0.7707,
      "step": 80000
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.124096144502558e-05,
      "loss": 0.7763,
      "step": 80500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.106233387636114e-05,
      "loss": 0.7559,
      "step": 81000
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.0883706307696703e-05,
      "loss": 0.7597,
      "step": 81500
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.0705078739032268e-05,
      "loss": 0.7717,
      "step": 82000
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.052645117036783e-05,
      "loss": 0.7637,
      "step": 82500
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0347823601703395e-05,
      "loss": 0.7641,
      "step": 83000
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0169196033038956e-05,
      "loss": 0.7778,
      "step": 83500
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.9990568464374518e-05,
      "loss": 0.7553,
      "step": 84000
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.9811940895710083e-05,
      "loss": 0.7672,
      "step": 84500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9633313327045644e-05,
      "loss": 0.7525,
      "step": 85000
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9454685758381206e-05,
      "loss": 0.7748,
      "step": 85500
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.927605818971677e-05,
      "loss": 0.7632,
      "step": 86000
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9097430621052333e-05,
      "loss": 0.7679,
      "step": 86500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8918803052387894e-05,
      "loss": 0.746,
      "step": 87000
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8740175483723456e-05,
      "loss": 0.7637,
      "step": 87500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8561547915059017e-05,
      "loss": 0.7518,
      "step": 88000
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8382920346394582e-05,
      "loss": 0.7578,
      "step": 88500
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8204292777730144e-05,
      "loss": 0.7611,
      "step": 89000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8025665209065706e-05,
      "loss": 0.7576,
      "step": 89500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.784703764040127e-05,
      "loss": 0.7405,
      "step": 90000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.7668410071736832e-05,
      "loss": 0.7657,
      "step": 90500
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7489782503072394e-05,
      "loss": 0.7539,
      "step": 91000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.731115493440796e-05,
      "loss": 0.7496,
      "step": 91500
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.713252736574352e-05,
      "loss": 0.7447,
      "step": 92000
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.6953899797079085e-05,
      "loss": 0.7582,
      "step": 92500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6775272228414647e-05,
      "loss": 0.747,
      "step": 93000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.659664465975021e-05,
      "loss": 0.7202,
      "step": 93500
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.641801709108577e-05,
      "loss": 0.6645,
      "step": 94000
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6239389522421332e-05,
      "loss": 0.6758,
      "step": 94500
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.6060761953756893e-05,
      "loss": 0.6758,
      "step": 95000
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.588213438509246e-05,
      "loss": 0.6657,
      "step": 95500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.570350681642802e-05,
      "loss": 0.6665,
      "step": 96000
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.552487924776358e-05,
      "loss": 0.6546,
      "step": 96500
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5346251679099147e-05,
      "loss": 0.6688,
      "step": 97000
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5167624110434708e-05,
      "loss": 0.6745,
      "step": 97500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.4988996541770273e-05,
      "loss": 0.6786,
      "step": 98000
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.4810368973105835e-05,
      "loss": 0.6826,
      "step": 98500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4631741404441396e-05,
      "loss": 0.6631,
      "step": 99000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.445311383577696e-05,
      "loss": 0.6707,
      "step": 99500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4274486267112521e-05,
      "loss": 0.665,
      "step": 100000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4095858698448083e-05,
      "loss": 0.6729,
      "step": 100500
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.3917231129783648e-05,
      "loss": 0.6633,
      "step": 101000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.373860356111921e-05,
      "loss": 0.665,
      "step": 101500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3559975992454771e-05,
      "loss": 0.6712,
      "step": 102000
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3381348423790336e-05,
      "loss": 0.665,
      "step": 102500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3202720855125898e-05,
      "loss": 0.671,
      "step": 103000
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.3024093286461461e-05,
      "loss": 0.6733,
      "step": 103500
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2845465717797023e-05,
      "loss": 0.6646,
      "step": 104000
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2666838149132584e-05,
      "loss": 0.6707,
      "step": 104500
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2488210580468147e-05,
      "loss": 0.6591,
      "step": 105000
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.230958301180371e-05,
      "loss": 0.6626,
      "step": 105500
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2130955443139274e-05,
      "loss": 0.6545,
      "step": 106000
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1952327874474836e-05,
      "loss": 0.6652,
      "step": 106500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1773700305810397e-05,
      "loss": 0.6533,
      "step": 107000
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.159507273714596e-05,
      "loss": 0.6613,
      "step": 107500
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1416445168481524e-05,
      "loss": 0.6618,
      "step": 108000
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1237817599817085e-05,
      "loss": 0.659,
      "step": 108500
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1059190031152649e-05,
      "loss": 0.6643,
      "step": 109000
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0880562462488212e-05,
      "loss": 0.6622,
      "step": 109500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0701934893823774e-05,
      "loss": 0.6553,
      "step": 110000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0523307325159335e-05,
      "loss": 0.6566,
      "step": 110500
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0344679756494899e-05,
      "loss": 0.6607,
      "step": 111000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0166052187830462e-05,
      "loss": 0.6465,
      "step": 111500
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.987424619166023e-06,
      "loss": 0.6546,
      "step": 112000
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.808797050501587e-06,
      "loss": 0.6674,
      "step": 112500
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.63016948183715e-06,
      "loss": 0.6641,
      "step": 113000
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.451541913172712e-06,
      "loss": 0.6569,
      "step": 113500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.272914344508273e-06,
      "loss": 0.6515,
      "step": 114000
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.094286775843837e-06,
      "loss": 0.6528,
      "step": 114500
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.9156592071794e-06,
      "loss": 0.6386,
      "step": 115000
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.737031638514963e-06,
      "loss": 0.655,
      "step": 115500
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.558404069850525e-06,
      "loss": 0.6534,
      "step": 116000
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.379776501186088e-06,
      "loss": 0.6557,
      "step": 116500
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.20114893252165e-06,
      "loss": 0.661,
      "step": 117000
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.022521363857213e-06,
      "loss": 0.6463,
      "step": 117500
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.843893795192775e-06,
      "loss": 0.6652,
      "step": 118000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.665266226528338e-06,
      "loss": 0.6587,
      "step": 118500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.486638657863901e-06,
      "loss": 0.6448,
      "step": 119000
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.308011089199463e-06,
      "loss": 0.6624,
      "step": 119500
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.129383520535025e-06,
      "loss": 0.655,
      "step": 120000
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.9507559518705885e-06,
      "loss": 0.6524,
      "step": 120500
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.772128383206151e-06,
      "loss": 0.6535,
      "step": 121000
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.5935008145417126e-06,
      "loss": 0.6509,
      "step": 121500
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.414873245877276e-06,
      "loss": 0.6528,
      "step": 122000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.236245677212839e-06,
      "loss": 0.6625,
      "step": 122500
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.057618108548401e-06,
      "loss": 0.6533,
      "step": 123000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.878990539883964e-06,
      "loss": 0.6405,
      "step": 123500
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.7003629712195265e-06,
      "loss": 0.6562,
      "step": 124000
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.521735402555089e-06,
      "loss": 0.6458,
      "step": 124500
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.343107833890651e-06,
      "loss": 0.6448,
      "step": 125000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.164480265226214e-06,
      "loss": 0.6486,
      "step": 125500
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.985852696561777e-06,
      "loss": 0.6559,
      "step": 126000
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.807225127897339e-06,
      "loss": 0.6329,
      "step": 126500
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.628597559232902e-06,
      "loss": 0.6447,
      "step": 127000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.4499699905684645e-06,
      "loss": 0.6551,
      "step": 127500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.271342421904028e-06,
      "loss": 0.6441,
      "step": 128000
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.092714853239589e-06,
      "loss": 0.6481,
      "step": 128500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.914087284575153e-06,
      "loss": 0.6405,
      "step": 129000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.7354597159107147e-06,
      "loss": 0.6376,
      "step": 129500
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.556832147246278e-06,
      "loss": 0.6509,
      "step": 130000
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.37820457858184e-06,
      "loss": 0.6321,
      "step": 130500
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.1995770099174025e-06,
      "loss": 0.6372,
      "step": 131000
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.0209494412529654e-06,
      "loss": 0.6348,
      "step": 131500
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.8423218725885283e-06,
      "loss": 0.6541,
      "step": 132000
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.6636943039240907e-06,
      "loss": 0.6462,
      "step": 132500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.485066735259653e-06,
      "loss": 0.6451,
      "step": 133000
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.306439166595216e-06,
      "loss": 0.6398,
      "step": 133500
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.127811597930778e-06,
      "loss": 0.6435,
      "step": 134000
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.949184029266341e-06,
      "loss": 0.6371,
      "step": 134500
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.7705564606019034e-06,
      "loss": 0.6291,
      "step": 135000
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.591928891937466e-06,
      "loss": 0.6437,
      "step": 135500
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.4133013232730287e-06,
      "loss": 0.6429,
      "step": 136000
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.2346737546085914e-06,
      "loss": 0.6381,
      "step": 136500
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.056046185944154e-06,
      "loss": 0.6388,
      "step": 137000
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.774186172797166e-07,
      "loss": 0.6388,
      "step": 137500
    },
    {
      "epoch": 2.96,
      "learning_rate": 6.987910486152791e-07,
      "loss": 0.639,
      "step": 138000
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.201634799508417e-07,
      "loss": 0.6275,
      "step": 138500
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.415359112864043e-07,
      "loss": 0.6425,
      "step": 139000
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.629083426219669e-07,
      "loss": 0.6301,
      "step": 139500
    },
    {
      "epoch": 3.0,
      "step": 139956,
      "total_flos": 1.4238860552921088e+17,
      "train_loss": 0.8559671926554289,
      "train_runtime": 48243.0463,
      "train_samples_per_second": 29.01,
      "train_steps_per_second": 2.901
    }
  ],
  "max_steps": 139956,
  "num_train_epochs": 3,
  "total_flos": 1.4238860552921088e+17,
  "trial_name": null,
  "trial_params": null
}