{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 199170,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.98744790882161e-05,
      "loss": 3.7619,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.97489581764322e-05,
      "loss": 2.8947,
      "step": 1000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.962343726464829e-05,
      "loss": 2.5182,
      "step": 1500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.949791635286439e-05,
      "loss": 2.3038,
      "step": 2000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.937239544108049e-05,
      "loss": 2.1453,
      "step": 2500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9246874529296584e-05,
      "loss": 2.0297,
      "step": 3000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.912135361751268e-05,
      "loss": 1.9309,
      "step": 3500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.899583270572878e-05,
      "loss": 1.8563,
      "step": 4000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.8870311793944875e-05,
      "loss": 1.8049,
      "step": 4500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.874479088216097e-05,
      "loss": 1.7541,
      "step": 5000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.861926997037707e-05,
      "loss": 1.7037,
      "step": 5500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8493749058593166e-05,
      "loss": 1.6645,
      "step": 6000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.836822814680926e-05,
      "loss": 1.621,
      "step": 6500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.824270723502536e-05,
      "loss": 1.6029,
      "step": 7000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8117186323241456e-05,
      "loss": 1.5655,
      "step": 7500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.799166541145755e-05,
      "loss": 1.5463,
      "step": 8000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7866144499673646e-05,
      "loss": 1.523,
      "step": 8500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.774062358788974e-05,
      "loss": 1.4924,
      "step": 9000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.761510267610584e-05,
      "loss": 1.4706,
      "step": 9500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7489581764321937e-05,
      "loss": 1.4786,
      "step": 10000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.736406085253803e-05,
      "loss": 1.4326,
      "step": 10500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.723853994075413e-05,
      "loss": 1.4092,
      "step": 11000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.711301902897023e-05,
      "loss": 1.4245,
      "step": 11500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.698749811718632e-05,
      "loss": 1.4063,
      "step": 12000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.6861977205402423e-05,
      "loss": 1.3734,
      "step": 12500
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.673645629361852e-05,
      "loss": 1.3613,
      "step": 13000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.661093538183461e-05,
      "loss": 1.3539,
      "step": 13500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.6485414470050714e-05,
      "loss": 1.3356,
      "step": 14000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.635989355826681e-05,
      "loss": 1.358,
      "step": 14500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.6234372646482904e-05,
      "loss": 1.311,
      "step": 15000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.6108851734699005e-05,
      "loss": 1.3026,
      "step": 15500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.59833308229151e-05,
      "loss": 1.308,
      "step": 16000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.5857809911131194e-05,
      "loss": 1.2758,
      "step": 16500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.5732288999347296e-05,
      "loss": 1.2894,
      "step": 17000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.560676808756339e-05,
      "loss": 1.2631,
      "step": 17500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.5481247175779485e-05,
      "loss": 1.2737,
      "step": 18000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.535572626399559e-05,
      "loss": 1.2568,
      "step": 18500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.523020535221168e-05,
      "loss": 1.2442,
      "step": 19000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.510468444042778e-05,
      "loss": 1.2455,
      "step": 19500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.497916352864388e-05,
      "loss": 1.244,
      "step": 20000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.485364261685997e-05,
      "loss": 1.23,
      "step": 20500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.4728121705076074e-05,
      "loss": 1.2089,
      "step": 21000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.460260079329217e-05,
      "loss": 1.2161,
      "step": 21500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.447707988150826e-05,
      "loss": 1.217,
      "step": 22000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.435155896972436e-05,
      "loss": 1.1856,
      "step": 22500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.422603805794045e-05,
      "loss": 1.1964,
      "step": 23000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.410051714615655e-05,
      "loss": 1.1793,
      "step": 23500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.397499623437265e-05,
      "loss": 1.1949,
      "step": 24000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.384947532258874e-05,
      "loss": 1.1823,
      "step": 24500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.372395441080484e-05,
      "loss": 1.1872,
      "step": 25000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.359843349902094e-05,
      "loss": 1.1712,
      "step": 25500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.3472912587237034e-05,
      "loss": 1.1792,
      "step": 26000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.334739167545313e-05,
      "loss": 1.1628,
      "step": 26500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.322187076366923e-05,
      "loss": 1.1464,
      "step": 27000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.3096349851885325e-05,
      "loss": 1.1668,
      "step": 27500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.297082894010142e-05,
      "loss": 1.1377,
      "step": 28000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.284530802831752e-05,
      "loss": 1.1426,
      "step": 28500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.2719787116533615e-05,
      "loss": 1.1347,
      "step": 29000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.259426620474971e-05,
      "loss": 1.1226,
      "step": 29500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.246874529296581e-05,
      "loss": 1.1094,
      "step": 30000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.2343224381181906e-05,
      "loss": 1.1204,
      "step": 30500
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.2217703469398e-05,
      "loss": 1.1171,
      "step": 31000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.20921825576141e-05,
      "loss": 1.1354,
      "step": 31500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.19666616458302e-05,
      "loss": 1.0998,
      "step": 32000
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.18411407340463e-05,
      "loss": 1.0948,
      "step": 32500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.171561982226239e-05,
      "loss": 1.0967,
      "step": 33000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.159009891047849e-05,
      "loss": 1.0943,
      "step": 33500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.146457799869459e-05,
      "loss": 1.1228,
      "step": 34000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.1339057086910684e-05,
      "loss": 1.0944,
      "step": 34500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.121353617512678e-05,
      "loss": 1.0928,
      "step": 35000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.108801526334288e-05,
      "loss": 1.1025,
      "step": 35500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.0962494351558975e-05,
      "loss": 1.0873,
      "step": 36000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.083697343977507e-05,
      "loss": 1.0977,
      "step": 36500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.0711452527991164e-05,
      "loss": 1.0785,
      "step": 37000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.0585931616207266e-05,
      "loss": 1.0754,
      "step": 37500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.046041070442336e-05,
      "loss": 1.0702,
      "step": 38000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.0334889792639455e-05,
      "loss": 1.0656,
      "step": 38500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.020936888085555e-05,
      "loss": 1.0817,
      "step": 39000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.0083847969071644e-05,
      "loss": 1.0647,
      "step": 39500
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.9958327057287746e-05,
      "loss": 1.0403,
      "step": 40000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.983280614550384e-05,
      "loss": 1.052,
      "step": 40500
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.9707285233719935e-05,
      "loss": 1.0472,
      "step": 41000
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.9581764321936036e-05,
      "loss": 1.0509,
      "step": 41500
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.945624341015213e-05,
      "loss": 1.0534,
      "step": 42000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.9330722498368226e-05,
      "loss": 1.0461,
      "step": 42500
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.920520158658433e-05,
      "loss": 1.0376,
      "step": 43000
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.907968067480042e-05,
      "loss": 1.0298,
      "step": 43500
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.895415976301652e-05,
      "loss": 1.0333,
      "step": 44000
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.882863885123262e-05,
      "loss": 1.0267,
      "step": 44500
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.870311793944871e-05,
      "loss": 1.0263,
      "step": 45000
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.8577597027664814e-05,
      "loss": 1.0276,
      "step": 45500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.845207611588091e-05,
      "loss": 1.0267,
      "step": 46000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.8326555204097004e-05,
      "loss": 1.0242,
      "step": 46500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.8201034292313105e-05,
      "loss": 1.0191,
      "step": 47000
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.80755133805292e-05,
      "loss": 1.0134,
      "step": 47500
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7949992468745294e-05,
      "loss": 1.0165,
      "step": 48000
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.7824471556961396e-05,
      "loss": 1.0201,
      "step": 48500
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.769895064517749e-05,
      "loss": 1.0201,
      "step": 49000
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.7573429733393585e-05,
      "loss": 1.02,
      "step": 49500
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.7447908821609687e-05,
      "loss": 1.0144,
      "step": 50000
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.732238790982578e-05,
      "loss": 0.9911,
      "step": 50500
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.7196866998041876e-05,
      "loss": 0.9944,
      "step": 51000
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.707134608625798e-05,
      "loss": 0.9963,
      "step": 51500
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.694582517447407e-05,
      "loss": 0.9875,
      "step": 52000
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.682030426269017e-05,
      "loss": 1.0059,
      "step": 52500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.669478335090626e-05,
      "loss": 0.9848,
      "step": 53000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.6569262439122356e-05,
      "loss": 0.9947,
      "step": 53500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.644374152733845e-05,
      "loss": 0.9865,
      "step": 54000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.631822061555455e-05,
      "loss": 0.9884,
      "step": 54500
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.619269970377065e-05,
      "loss": 0.9781,
      "step": 55000
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.606717879198674e-05,
      "loss": 0.9934,
      "step": 55500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.594165788020284e-05,
      "loss": 0.9937,
      "step": 56000
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.581613696841894e-05,
      "loss": 0.9714,
      "step": 56500
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.569061605663503e-05,
      "loss": 0.9809,
      "step": 57000
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.5565095144851134e-05,
      "loss": 0.9709,
      "step": 57500
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.543957423306723e-05,
      "loss": 0.9656,
      "step": 58000
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.531405332128333e-05,
      "loss": 0.9632,
      "step": 58500
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.5188532409499425e-05,
      "loss": 0.9767,
      "step": 59000
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.506301149771552e-05,
      "loss": 0.9669,
      "step": 59500
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.493749058593162e-05,
      "loss": 0.9692,
      "step": 60000
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.4811969674147715e-05,
      "loss": 0.97,
      "step": 60500
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.468644876236381e-05,
      "loss": 0.9514,
      "step": 61000
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.456092785057991e-05,
      "loss": 0.9421,
      "step": 61500
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.4435406938796006e-05,
      "loss": 0.9734,
      "step": 62000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.43098860270121e-05,
      "loss": 0.9562,
      "step": 62500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.41843651152282e-05,
      "loss": 0.9625,
      "step": 63000
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.40588442034443e-05,
      "loss": 0.9542,
      "step": 63500
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.393332329166039e-05,
      "loss": 0.9497,
      "step": 64000
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.380780237987649e-05,
      "loss": 0.9431,
      "step": 64500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.368228146809259e-05,
      "loss": 0.9456,
      "step": 65000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.355676055630868e-05,
      "loss": 0.9395,
      "step": 65500
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.3431239644524784e-05,
      "loss": 0.9393,
      "step": 66000
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.330571873274088e-05,
      "loss": 0.9243,
      "step": 66500
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.318019782095697e-05,
      "loss": 0.8449,
      "step": 67000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.305467690917307e-05,
      "loss": 0.8542,
      "step": 67500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.292915599738917e-05,
      "loss": 0.8496,
      "step": 68000
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.2803635085605264e-05,
      "loss": 0.8452,
      "step": 68500
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.267811417382136e-05,
      "loss": 0.8361,
      "step": 69000
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.2552593262037453e-05,
      "loss": 0.8441,
      "step": 69500
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.242707235025355e-05,
      "loss": 0.8536,
      "step": 70000
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.230155143846965e-05,
      "loss": 0.8338,
      "step": 70500
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.2176030526685744e-05,
      "loss": 0.845,
      "step": 71000
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.2050509614901846e-05,
      "loss": 0.8544,
      "step": 71500
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.192498870311794e-05,
      "loss": 0.8487,
      "step": 72000
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.1799467791334035e-05,
      "loss": 0.8273,
      "step": 72500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.1673946879550136e-05,
      "loss": 0.8581,
      "step": 73000
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.154842596776623e-05,
      "loss": 0.8493,
      "step": 73500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.1422905055982326e-05,
      "loss": 0.8321,
      "step": 74000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.129738414419843e-05,
      "loss": 0.8404,
      "step": 74500
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.117186323241452e-05,
      "loss": 0.8207,
      "step": 75000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.104634232063062e-05,
      "loss": 0.8439,
      "step": 75500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.092082140884672e-05,
      "loss": 0.8462,
      "step": 76000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.079530049706281e-05,
      "loss": 0.8441,
      "step": 76500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.066977958527891e-05,
      "loss": 0.8394,
      "step": 77000
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.054425867349501e-05,
      "loss": 0.8441,
      "step": 77500
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.0418737761711104e-05,
      "loss": 0.8462,
      "step": 78000
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0293216849927198e-05,
      "loss": 0.8392,
      "step": 78500
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.0167695938143296e-05,
      "loss": 0.835,
      "step": 79000
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.004217502635939e-05,
      "loss": 0.8544,
      "step": 79500
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.991665411457549e-05,
      "loss": 0.8275,
      "step": 80000
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.9791133202791587e-05,
      "loss": 0.8288,
      "step": 80500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9665612291007682e-05,
      "loss": 0.838,
      "step": 81000
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.9540091379223776e-05,
      "loss": 0.8369,
      "step": 81500
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9414570467439878e-05,
      "loss": 0.8309,
      "step": 82000
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9289049555655973e-05,
      "loss": 0.8313,
      "step": 82500
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9163528643872067e-05,
      "loss": 0.845,
      "step": 83000
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.903800773208817e-05,
      "loss": 0.8348,
      "step": 83500
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.8912486820304263e-05,
      "loss": 0.8231,
      "step": 84000
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.8786965908520365e-05,
      "loss": 0.8286,
      "step": 84500
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.866144499673646e-05,
      "loss": 0.8203,
      "step": 85000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8535924084952554e-05,
      "loss": 0.8264,
      "step": 85500
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8410403173168652e-05,
      "loss": 0.8291,
      "step": 86000
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.828488226138475e-05,
      "loss": 0.8222,
      "step": 86500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8159361349600845e-05,
      "loss": 0.8357,
      "step": 87000
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.8033840437816943e-05,
      "loss": 0.8327,
      "step": 87500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7908319526033038e-05,
      "loss": 0.834,
      "step": 88000
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7782798614249132e-05,
      "loss": 0.8187,
      "step": 88500
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7657277702465234e-05,
      "loss": 0.8252,
      "step": 89000
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.753175679068133e-05,
      "loss": 0.8313,
      "step": 89500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7406235878897423e-05,
      "loss": 0.8255,
      "step": 90000
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7280714967113525e-05,
      "loss": 0.827,
      "step": 90500
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.715519405532962e-05,
      "loss": 0.8251,
      "step": 91000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.7029673143545714e-05,
      "loss": 0.8173,
      "step": 91500
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6904152231761815e-05,
      "loss": 0.82,
      "step": 92000
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.677863131997791e-05,
      "loss": 0.814,
      "step": 92500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6653110408194005e-05,
      "loss": 0.8218,
      "step": 93000
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6527589496410106e-05,
      "loss": 0.8118,
      "step": 93500
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.64020685846262e-05,
      "loss": 0.8189,
      "step": 94000
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.6276547672842296e-05,
      "loss": 0.8215,
      "step": 94500
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6151026761058394e-05,
      "loss": 0.822,
      "step": 95000
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.6025505849274488e-05,
      "loss": 0.8149,
      "step": 95500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5899984937490583e-05,
      "loss": 0.8271,
      "step": 96000
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5774464025706684e-05,
      "loss": 0.823,
      "step": 96500
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.564894311392278e-05,
      "loss": 0.8167,
      "step": 97000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.552342220213888e-05,
      "loss": 0.8127,
      "step": 97500
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.5397901290354975e-05,
      "loss": 0.7968,
      "step": 98000
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.527238037857107e-05,
      "loss": 0.8144,
      "step": 98500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.514685946678717e-05,
      "loss": 0.8111,
      "step": 99000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.5021338555003266e-05,
      "loss": 0.8125,
      "step": 99500
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.489581764321936e-05,
      "loss": 0.8158,
      "step": 100000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.477029673143546e-05,
      "loss": 0.8132,
      "step": 100500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4644775819651557e-05,
      "loss": 0.8119,
      "step": 101000
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.451925490786765e-05,
      "loss": 0.8121,
      "step": 101500
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.439373399608375e-05,
      "loss": 0.7989,
      "step": 102000
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4268213084299844e-05,
      "loss": 0.8099,
      "step": 102500
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4142692172515942e-05,
      "loss": 0.8157,
      "step": 103000
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.4017171260732037e-05,
      "loss": 0.807,
      "step": 103500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3891650348948135e-05,
      "loss": 0.8021,
      "step": 104000
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3766129437164233e-05,
      "loss": 0.7899,
      "step": 104500
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3640608525380328e-05,
      "loss": 0.8051,
      "step": 105000
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3515087613596426e-05,
      "loss": 0.8025,
      "step": 105500
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3389566701812524e-05,
      "loss": 0.7956,
      "step": 106000
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.326404579002862e-05,
      "loss": 0.8237,
      "step": 106500
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3138524878244717e-05,
      "loss": 0.8056,
      "step": 107000
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.3013003966460815e-05,
      "loss": 0.7946,
      "step": 107500
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2887483054676913e-05,
      "loss": 0.7888,
      "step": 108000
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2761962142893007e-05,
      "loss": 0.7957,
      "step": 108500
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.2636441231109105e-05,
      "loss": 0.7968,
      "step": 109000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.25109203193252e-05,
      "loss": 0.7996,
      "step": 109500
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2385399407541298e-05,
      "loss": 0.7899,
      "step": 110000
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2259878495757393e-05,
      "loss": 0.8012,
      "step": 110500
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.213435758397349e-05,
      "loss": 0.7944,
      "step": 111000
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.2008836672189586e-05,
      "loss": 0.7916,
      "step": 111500
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1883315760405684e-05,
      "loss": 0.7932,
      "step": 112000
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1757794848621782e-05,
      "loss": 0.7904,
      "step": 112500
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.1632273936837876e-05,
      "loss": 0.7935,
      "step": 113000
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1506753025053974e-05,
      "loss": 0.7881,
      "step": 113500
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1381232113270073e-05,
      "loss": 0.7754,
      "step": 114000
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.125571120148617e-05,
      "loss": 0.7974,
      "step": 114500
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.1130190289702265e-05,
      "loss": 0.7857,
      "step": 115000
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.1004669377918363e-05,
      "loss": 0.7714,
      "step": 115500
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.087914846613446e-05,
      "loss": 0.7925,
      "step": 116000
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.0753627554350556e-05,
      "loss": 0.7864,
      "step": 116500
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.0628106642566654e-05,
      "loss": 0.7866,
      "step": 117000
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.050258573078275e-05,
      "loss": 0.789,
      "step": 117500
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0377064818998843e-05,
      "loss": 0.7667,
      "step": 118000
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.025154390721494e-05,
      "loss": 0.7824,
      "step": 118500
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.012602299543104e-05,
      "loss": 0.7811,
      "step": 119000
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0000502083647138e-05,
      "loss": 0.7861,
      "step": 119500
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.9874981171863232e-05,
      "loss": 0.779,
      "step": 120000
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.974946026007933e-05,
      "loss": 0.7766,
      "step": 120500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.962393934829543e-05,
      "loss": 0.772,
      "step": 121000
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9498418436511523e-05,
      "loss": 0.7826,
      "step": 121500
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.937289752472762e-05,
      "loss": 0.7771,
      "step": 122000
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.924737661294372e-05,
      "loss": 0.7636,
      "step": 122500
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9121855701159814e-05,
      "loss": 0.7749,
      "step": 123000
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8996334789375912e-05,
      "loss": 0.7696,
      "step": 123500
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.887081387759201e-05,
      "loss": 0.7742,
      "step": 124000
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8745292965808105e-05,
      "loss": 0.7568,
      "step": 124500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.86197720540242e-05,
      "loss": 0.7782,
      "step": 125000
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8494251142240297e-05,
      "loss": 0.7762,
      "step": 125500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8368730230456396e-05,
      "loss": 0.7607,
      "step": 126000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.824320931867249e-05,
      "loss": 0.764,
      "step": 126500
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8117688406888588e-05,
      "loss": 0.7698,
      "step": 127000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.7992167495104686e-05,
      "loss": 0.768,
      "step": 127500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.786664658332078e-05,
      "loss": 0.7729,
      "step": 128000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.774112567153688e-05,
      "loss": 0.7641,
      "step": 128500
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.7615604759752977e-05,
      "loss": 0.7623,
      "step": 129000
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7490083847969072e-05,
      "loss": 0.7729,
      "step": 129500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.736456293618517e-05,
      "loss": 0.7548,
      "step": 130000
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.7239042024401268e-05,
      "loss": 0.7738,
      "step": 130500
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.7113521112617363e-05,
      "loss": 0.7659,
      "step": 131000
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.698800020083346e-05,
      "loss": 0.7525,
      "step": 131500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6862479289049555e-05,
      "loss": 0.7809,
      "step": 132000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6736958377265653e-05,
      "loss": 0.777,
      "step": 132500
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6611437465481748e-05,
      "loss": 0.725,
      "step": 133000
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.6485916553697846e-05,
      "loss": 0.657,
      "step": 133500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.6360395641913944e-05,
      "loss": 0.6593,
      "step": 134000
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.623487473013004e-05,
      "loss": 0.6657,
      "step": 134500
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6109353818346137e-05,
      "loss": 0.6536,
      "step": 135000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.5983832906562235e-05,
      "loss": 0.6721,
      "step": 135500
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.585831199477833e-05,
      "loss": 0.6559,
      "step": 136000
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5732791082994428e-05,
      "loss": 0.6645,
      "step": 136500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5607270171210526e-05,
      "loss": 0.6506,
      "step": 137000
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.548174925942662e-05,
      "loss": 0.673,
      "step": 137500
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.535622834764272e-05,
      "loss": 0.6579,
      "step": 138000
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5230707435858815e-05,
      "loss": 0.6589,
      "step": 138500
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5105186524074913e-05,
      "loss": 0.6698,
      "step": 139000
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.4979665612291008e-05,
      "loss": 0.6613,
      "step": 139500
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.4854144700507106e-05,
      "loss": 0.6528,
      "step": 140000
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4728623788723204e-05,
      "loss": 0.6667,
      "step": 140500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4603102876939298e-05,
      "loss": 0.6694,
      "step": 141000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.4477581965155396e-05,
      "loss": 0.6735,
      "step": 141500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4352061053371493e-05,
      "loss": 0.6559,
      "step": 142000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4226540141587587e-05,
      "loss": 0.6551,
      "step": 142500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4101019229803686e-05,
      "loss": 0.6628,
      "step": 143000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.3975498318019784e-05,
      "loss": 0.6638,
      "step": 143500
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3849977406235878e-05,
      "loss": 0.6602,
      "step": 144000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3724456494451976e-05,
      "loss": 0.6618,
      "step": 144500
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3598935582668074e-05,
      "loss": 0.6663,
      "step": 145000
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.347341467088417e-05,
      "loss": 0.6533,
      "step": 145500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3347893759100265e-05,
      "loss": 0.6451,
      "step": 146000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3222372847316364e-05,
      "loss": 0.6701,
      "step": 146500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3096851935532462e-05,
      "loss": 0.6705,
      "step": 147000
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.2971331023748556e-05,
      "loss": 0.6605,
      "step": 147500
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2845810111964654e-05,
      "loss": 0.6517,
      "step": 148000
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2720289200180752e-05,
      "loss": 0.6596,
      "step": 148500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2594768288396847e-05,
      "loss": 0.6531,
      "step": 149000
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2469247376612943e-05,
      "loss": 0.6542,
      "step": 149500
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.234372646482904e-05,
      "loss": 0.6586,
      "step": 150000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2218205553045138e-05,
      "loss": 0.647,
      "step": 150500
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2092684641261234e-05,
      "loss": 0.6634,
      "step": 151000
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1967163729477332e-05,
      "loss": 0.65,
      "step": 151500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1841642817693429e-05,
      "loss": 0.6467,
      "step": 152000
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1716121905909525e-05,
      "loss": 0.6595,
      "step": 152500
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1590600994125621e-05,
      "loss": 0.6564,
      "step": 153000
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1465080082341718e-05,
      "loss": 0.6583,
      "step": 153500
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1339559170557816e-05,
      "loss": 0.6623,
      "step": 154000
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1214038258773912e-05,
      "loss": 0.6557,
      "step": 154500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1088517346990009e-05,
      "loss": 0.6523,
      "step": 155000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.0962996435206107e-05,
      "loss": 0.6707,
      "step": 155500
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0837475523422203e-05,
      "loss": 0.658,
      "step": 156000
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.07119546116383e-05,
      "loss": 0.6381,
      "step": 156500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0586433699854396e-05,
      "loss": 0.6633,
      "step": 157000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0460912788070492e-05,
      "loss": 0.6568,
      "step": 157500
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.033539187628659e-05,
      "loss": 0.6483,
      "step": 158000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0209870964502687e-05,
      "loss": 0.6478,
      "step": 158500
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0084350052718783e-05,
      "loss": 0.6565,
      "step": 159000
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.958829140934881e-06,
      "loss": 0.6487,
      "step": 159500
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.833308229150977e-06,
      "loss": 0.6547,
      "step": 160000
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.707787317367074e-06,
      "loss": 0.6511,
      "step": 160500
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.58226640558317e-06,
      "loss": 0.6421,
      "step": 161000
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.456745493799266e-06,
      "loss": 0.6417,
      "step": 161500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.331224582015364e-06,
      "loss": 0.6452,
      "step": 162000
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.20570367023146e-06,
      "loss": 0.6525,
      "step": 162500
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.080182758447557e-06,
      "loss": 0.6516,
      "step": 163000
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.954661846663655e-06,
      "loss": 0.646,
      "step": 163500
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.829140934879752e-06,
      "loss": 0.6428,
      "step": 164000
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.703620023095848e-06,
      "loss": 0.6431,
      "step": 164500
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.578099111311944e-06,
      "loss": 0.6477,
      "step": 165000
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.45257819952804e-06,
      "loss": 0.6491,
      "step": 165500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.327057287744139e-06,
      "loss": 0.6482,
      "step": 166000
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.201536375960235e-06,
      "loss": 0.6458,
      "step": 166500
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.076015464176333e-06,
      "loss": 0.6483,
      "step": 167000
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.95049455239243e-06,
      "loss": 0.6491,
      "step": 167500
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.824973640608526e-06,
      "loss": 0.6408,
      "step": 168000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.699452728824622e-06,
      "loss": 0.6491,
      "step": 168500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.5739318170407196e-06,
      "loss": 0.6472,
      "step": 169000
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.448410905256815e-06,
      "loss": 0.6422,
      "step": 169500
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.322889993472913e-06,
      "loss": 0.6471,
      "step": 170000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.1973690816890095e-06,
      "loss": 0.6343,
      "step": 170500
    },
    {
      "epoch": 2.58,
      "learning_rate": 7.071848169905107e-06,
      "loss": 0.6446,
      "step": 171000
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.946327258121203e-06,
      "loss": 0.6444,
      "step": 171500
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.8208063463372995e-06,
      "loss": 0.6455,
      "step": 172000
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.6952854345533975e-06,
      "loss": 0.6448,
      "step": 172500
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.569764522769493e-06,
      "loss": 0.6367,
      "step": 173000
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.444243610985591e-06,
      "loss": 0.6303,
      "step": 173500
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.3187226992016875e-06,
      "loss": 0.6392,
      "step": 174000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.193201787417785e-06,
      "loss": 0.6296,
      "step": 174500
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.06768087563388e-06,
      "loss": 0.6411,
      "step": 175000
    },
    {
      "epoch": 2.64,
      "learning_rate": 5.942159963849977e-06,
      "loss": 0.6432,
      "step": 175500
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.816639052066075e-06,
      "loss": 0.6483,
      "step": 176000
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.691118140282172e-06,
      "loss": 0.6355,
      "step": 176500
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.565597228498268e-06,
      "loss": 0.6426,
      "step": 177000
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.440076316714365e-06,
      "loss": 0.64,
      "step": 177500
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.314555404930462e-06,
      "loss": 0.6434,
      "step": 178000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.189034493146558e-06,
      "loss": 0.6373,
      "step": 178500
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.063513581362655e-06,
      "loss": 0.6281,
      "step": 179000
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.937992669578752e-06,
      "loss": 0.6323,
      "step": 179500
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.812471757794849e-06,
      "loss": 0.6334,
      "step": 180000
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.686950846010945e-06,
      "loss": 0.6316,
      "step": 180500
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.5614299342270425e-06,
      "loss": 0.6442,
      "step": 181000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.435909022443139e-06,
      "loss": 0.6429,
      "step": 181500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.310388110659236e-06,
      "loss": 0.6224,
      "step": 182000
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.1848671988753325e-06,
      "loss": 0.6314,
      "step": 182500
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.05934628709143e-06,
      "loss": 0.6576,
      "step": 183000
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.933825375307527e-06,
      "loss": 0.6348,
      "step": 183500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.808304463523623e-06,
      "loss": 0.635,
      "step": 184000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.68278355173972e-06,
      "loss": 0.6332,
      "step": 184500
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.557262639955817e-06,
      "loss": 0.6282,
      "step": 185000
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.4317417281719137e-06,
      "loss": 0.6333,
      "step": 185500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.30622081638801e-06,
      "loss": 0.6345,
      "step": 186000
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.1806999046041072e-06,
      "loss": 0.6408,
      "step": 186500
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.055178992820204e-06,
      "loss": 0.6284,
      "step": 187000
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.929658081036301e-06,
      "loss": 0.6332,
      "step": 187500
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.8041371692523976e-06,
      "loss": 0.6361,
      "step": 188000
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.6786162574684944e-06,
      "loss": 0.6307,
      "step": 188500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.553095345684591e-06,
      "loss": 0.6373,
      "step": 189000
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.427574433900688e-06,
      "loss": 0.6421,
      "step": 189500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.3020535221167848e-06,
      "loss": 0.635,
      "step": 190000
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.1765326103328816e-06,
      "loss": 0.6273,
      "step": 190500
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.0510116985489784e-06,
      "loss": 0.6374,
      "step": 191000
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.925490786765075e-06,
      "loss": 0.6333,
      "step": 191500
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.799969874981172e-06,
      "loss": 0.6285,
      "step": 192000
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6744489631972687e-06,
      "loss": 0.6266,
      "step": 192500
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.5489280514133655e-06,
      "loss": 0.6358,
      "step": 193000
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.4234071396294623e-06,
      "loss": 0.6263,
      "step": 193500
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.2978862278455591e-06,
      "loss": 0.6428,
      "step": 194000
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.172365316061656e-06,
      "loss": 0.6295,
      "step": 194500
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.0468444042777527e-06,
      "loss": 0.632,
      "step": 195000
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.213234924938495e-07,
      "loss": 0.6203,
      "step": 195500
    },
    {
      "epoch": 2.95,
      "learning_rate": 7.958025807099463e-07,
      "loss": 0.6223,
      "step": 196000
    },
    {
      "epoch": 2.96,
      "learning_rate": 6.702816689260432e-07,
      "loss": 0.6341,
      "step": 196500
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.4476075714214e-07,
      "loss": 0.628,
      "step": 197000
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.192398453582367e-07,
      "loss": 0.6267,
      "step": 197500
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.937189335743335e-07,
      "loss": 0.6196,
      "step": 198000
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.681980217904303e-07,
      "loss": 0.6263,
      "step": 198500
    },
    {
      "epoch": 3.0,
      "learning_rate": 4.267711000652709e-08,
      "loss": 0.6211,
      "step": 199000
    },
    {
      "epoch": 3.0,
      "step": 199170,
      "total_flos": 2.210416975580037e+17,
      "train_loss": 0.564786915779832,
      "train_runtime": 39960.6259,
      "train_samples_per_second": 49.841,
      "train_steps_per_second": 4.984
    }
  ],
  "max_steps": 199170,
  "num_train_epochs": 3,
  "total_flos": 2.210416975580037e+17,
  "trial_name": null,
  "trial_params": null
}