{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 239755,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.9895726887864694e-05,
      "loss": 2.0097,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.979145377572939e-05,
      "loss": 1.8204,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.968718066359409e-05,
      "loss": 1.6177,
      "step": 1500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.958290755145878e-05,
      "loss": 1.543,
      "step": 2000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.947863443932348e-05,
      "loss": 1.5073,
      "step": 2500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.937436132718817e-05,
      "loss": 1.4273,
      "step": 3000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9270088215052865e-05,
      "loss": 1.3995,
      "step": 3500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.916581510291756e-05,
      "loss": 1.2902,
      "step": 4000
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.906154199078226e-05,
      "loss": 1.2541,
      "step": 4500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.895726887864695e-05,
      "loss": 1.1981,
      "step": 5000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.885299576651165e-05,
      "loss": 1.1933,
      "step": 5500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8748722654376343e-05,
      "loss": 1.1827,
      "step": 6000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.8644449542241035e-05,
      "loss": 1.1112,
      "step": 6500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.8540176430105734e-05,
      "loss": 1.1385,
      "step": 7000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.843590331797043e-05,
      "loss": 1.0471,
      "step": 7500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.8331630205835124e-05,
      "loss": 1.0287,
      "step": 8000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.822735709369982e-05,
      "loss": 1.0343,
      "step": 8500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.8123083981564514e-05,
      "loss": 1.0191,
      "step": 9000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.801881086942921e-05,
      "loss": 1.0007,
      "step": 9500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.791453775729391e-05,
      "loss": 0.9207,
      "step": 10000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.78102646451586e-05,
      "loss": 0.9356,
      "step": 10500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.7705991533023294e-05,
      "loss": 0.8999,
      "step": 11000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.760171842088799e-05,
      "loss": 0.9197,
      "step": 11500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.7497445308752684e-05,
      "loss": 0.8851,
      "step": 12000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.739317219661738e-05,
      "loss": 0.8467,
      "step": 12500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.728889908448208e-05,
      "loss": 0.8545,
      "step": 13000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.718462597234677e-05,
      "loss": 0.844,
      "step": 13500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.708035286021147e-05,
      "loss": 0.8036,
      "step": 14000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.697607974807616e-05,
      "loss": 0.7726,
      "step": 14500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.6871806635940855e-05,
      "loss": 0.8192,
      "step": 15000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.676753352380555e-05,
      "loss": 0.786,
      "step": 15500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.666326041167025e-05,
      "loss": 0.7673,
      "step": 16000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.6558987299534944e-05,
      "loss": 0.7822,
      "step": 16500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.645471418739964e-05,
      "loss": 0.7499,
      "step": 17000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.6350441075264334e-05,
      "loss": 0.7388,
      "step": 17500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.6246167963129025e-05,
      "loss": 0.7096,
      "step": 18000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.614189485099373e-05,
      "loss": 0.6983,
      "step": 18500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.603762173885842e-05,
      "loss": 0.712,
      "step": 19000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.5933348626723114e-05,
      "loss": 0.6714,
      "step": 19500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.582907551458781e-05,
      "loss": 0.6959,
      "step": 20000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.5724802402452504e-05,
      "loss": 0.6654,
      "step": 20500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.5620529290317196e-05,
      "loss": 0.659,
      "step": 21000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.55162561781819e-05,
      "loss": 0.6296,
      "step": 21500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.541198306604659e-05,
      "loss": 0.6371,
      "step": 22000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.5307709953911285e-05,
      "loss": 0.6505,
      "step": 22500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.520343684177598e-05,
      "loss": 0.621,
      "step": 23000
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.5099163729640675e-05,
      "loss": 0.6013,
      "step": 23500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.4994890617505366e-05,
      "loss": 0.6206,
      "step": 24000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.489061750537007e-05,
      "loss": 0.5886,
      "step": 24500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.478634439323476e-05,
      "loss": 0.5938,
      "step": 25000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.4682071281099455e-05,
      "loss": 0.5512,
      "step": 25500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.4577798168964154e-05,
      "loss": 0.5524,
      "step": 26000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.4473525056828845e-05,
      "loss": 0.5555,
      "step": 26500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.4369251944693544e-05,
      "loss": 0.5736,
      "step": 27000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.426497883255824e-05,
      "loss": 0.5437,
      "step": 27500
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.4160705720422934e-05,
      "loss": 0.5492,
      "step": 28000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.4056432608287626e-05,
      "loss": 0.5497,
      "step": 28500
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.3952159496152324e-05,
      "loss": 0.538,
      "step": 29000
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.3847886384017016e-05,
      "loss": 0.5215,
      "step": 29500
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.3743613271881714e-05,
      "loss": 0.4956,
      "step": 30000
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.363934015974641e-05,
      "loss": 0.5114,
      "step": 30500
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.3535067047611104e-05,
      "loss": 0.51,
      "step": 31000
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.34307939354758e-05,
      "loss": 0.5159,
      "step": 31500
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.3326520823340495e-05,
      "loss": 0.4825,
      "step": 32000
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.3222247711205186e-05,
      "loss": 0.495,
      "step": 32500
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.3117974599069885e-05,
      "loss": 0.4683,
      "step": 33000
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.301370148693458e-05,
      "loss": 0.4663,
      "step": 33500
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.2909428374799275e-05,
      "loss": 0.4545,
      "step": 34000
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.280515526266397e-05,
      "loss": 0.4684,
      "step": 34500
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.2700882150528665e-05,
      "loss": 0.4444,
      "step": 35000
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.2596609038393363e-05,
      "loss": 0.4467,
      "step": 35500
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.249233592625806e-05,
      "loss": 0.4538,
      "step": 36000
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.2388062814122754e-05,
      "loss": 0.4192,
      "step": 36500
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.2283789701987445e-05,
      "loss": 0.4444,
      "step": 37000
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.2179516589852144e-05,
      "loss": 0.4396,
      "step": 37500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.2075243477716836e-05,
      "loss": 0.4456,
      "step": 38000
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.1970970365581534e-05,
      "loss": 0.4337,
      "step": 38500
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.186669725344623e-05,
      "loss": 0.4108,
      "step": 39000
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.1762424141310924e-05,
      "loss": 0.4416,
      "step": 39500
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.1658151029175616e-05,
      "loss": 0.4061,
      "step": 40000
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.1553877917040314e-05,
      "loss": 0.3816,
      "step": 40500
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.1449604804905006e-05,
      "loss": 0.3916,
      "step": 41000
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.1345331692769704e-05,
      "loss": 0.3914,
      "step": 41500
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.12410585806344e-05,
      "loss": 0.4327,
      "step": 42000
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.1136785468499095e-05,
      "loss": 0.3784,
      "step": 42500
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.1032512356363786e-05,
      "loss": 0.4018,
      "step": 43000
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.0928239244228485e-05,
      "loss": 0.3416,
      "step": 43500
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.0823966132093177e-05,
      "loss": 0.3602,
      "step": 44000
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.0719693019957875e-05,
      "loss": 0.3911,
      "step": 44500
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.0615419907822573e-05,
      "loss": 0.4008,
      "step": 45000
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.0511146795687265e-05,
      "loss": 0.3659,
      "step": 45500
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.040687368355196e-05,
      "loss": 0.3536,
      "step": 46000
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.0302600571416655e-05,
      "loss": 0.3782,
      "step": 46500
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.0198327459281354e-05,
      "loss": 0.3624,
      "step": 47000
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.0094054347146045e-05,
      "loss": 0.3284,
      "step": 47500
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.9989781235010744e-05,
      "loss": 0.1845,
      "step": 48000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.9885508122875436e-05,
      "loss": 0.2659,
      "step": 48500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.9781235010740134e-05,
      "loss": 0.2861,
      "step": 49000
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.9676961898604826e-05,
      "loss": 0.2826,
      "step": 49500
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.9572688786469524e-05,
      "loss": 0.2703,
      "step": 50000
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.9468415674334216e-05,
      "loss": 0.2455,
      "step": 50500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.9364142562198914e-05,
      "loss": 0.2555,
      "step": 51000
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.9259869450063606e-05,
      "loss": 0.262,
      "step": 51500
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.9155596337928305e-05,
      "loss": 0.2483,
      "step": 52000
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.9051323225792996e-05,
      "loss": 0.2604,
      "step": 52500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.8947050113657695e-05,
      "loss": 0.2376,
      "step": 53000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.884277700152239e-05,
      "loss": 0.2499,
      "step": 53500
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.8738503889387085e-05,
      "loss": 0.2632,
      "step": 54000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.863423077725178e-05,
      "loss": 0.2506,
      "step": 54500
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.8529957665116475e-05,
      "loss": 0.2541,
      "step": 55000
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.8425684552981174e-05,
      "loss": 0.2272,
      "step": 55500
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.8321411440845865e-05,
      "loss": 0.2688,
      "step": 56000
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.8217138328710564e-05,
      "loss": 0.235,
      "step": 56500
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.8112865216575255e-05,
      "loss": 0.2608,
      "step": 57000
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.800859210443995e-05,
      "loss": 0.2506,
      "step": 57500
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.7904318992304646e-05,
      "loss": 0.2437,
      "step": 58000
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.7800045880169344e-05,
      "loss": 0.2361,
      "step": 58500
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.7695772768034036e-05,
      "loss": 0.2512,
      "step": 59000
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.7591499655898734e-05,
      "loss": 0.2152,
      "step": 59500
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.7487226543763426e-05,
      "loss": 0.2545,
      "step": 60000
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.738295343162812e-05,
      "loss": 0.2286,
      "step": 60500
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.7278680319492816e-05,
      "loss": 0.2297,
      "step": 61000
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.7174407207357515e-05,
      "loss": 0.2478,
      "step": 61500
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.7070134095222206e-05,
      "loss": 0.2344,
      "step": 62000
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.6965860983086905e-05,
      "loss": 0.2134,
      "step": 62500
    },
    {
      "epoch": 1.31,
      "learning_rate": 3.6861587870951596e-05,
      "loss": 0.241,
      "step": 63000
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.675731475881629e-05,
      "loss": 0.2374,
      "step": 63500
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.6653041646680993e-05,
      "loss": 0.216,
      "step": 64000
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.6548768534545685e-05,
      "loss": 0.2258,
      "step": 64500
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.644449542241038e-05,
      "loss": 0.2325,
      "step": 65000
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.6340222310275075e-05,
      "loss": 0.2286,
      "step": 65500
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.623594919813977e-05,
      "loss": 0.2417,
      "step": 66000
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.6131676086004465e-05,
      "loss": 0.2324,
      "step": 66500
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.6027402973869164e-05,
      "loss": 0.2056,
      "step": 67000
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.5923129861733856e-05,
      "loss": 0.231,
      "step": 67500
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.581885674959855e-05,
      "loss": 0.2391,
      "step": 68000
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.5714583637463246e-05,
      "loss": 0.2188,
      "step": 68500
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.561031052532794e-05,
      "loss": 0.213,
      "step": 69000
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.5506037413192636e-05,
      "loss": 0.2096,
      "step": 69500
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.5401764301057334e-05,
      "loss": 0.24,
      "step": 70000
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.5297491188922026e-05,
      "loss": 0.2246,
      "step": 70500
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.5193218076786725e-05,
      "loss": 0.2104,
      "step": 71000
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.5088944964651416e-05,
      "loss": 0.2397,
      "step": 71500
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.498467185251611e-05,
      "loss": 0.2076,
      "step": 72000
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.4880398740380806e-05,
      "loss": 0.1854,
      "step": 72500
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.4776125628245505e-05,
      "loss": 0.2053,
      "step": 73000
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.4671852516110197e-05,
      "loss": 0.1901,
      "step": 73500
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.4567579403974895e-05,
      "loss": 0.2008,
      "step": 74000
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.446330629183959e-05,
      "loss": 0.2115,
      "step": 74500
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.435903317970428e-05,
      "loss": 0.2057,
      "step": 75000
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.4254760067568984e-05,
      "loss": 0.1976,
      "step": 75500
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.4150486955433675e-05,
      "loss": 0.1856,
      "step": 76000
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.404621384329837e-05,
      "loss": 0.2248,
      "step": 76500
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.3941940731163066e-05,
      "loss": 0.1992,
      "step": 77000
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.383766761902776e-05,
      "loss": 0.2046,
      "step": 77500
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.373339450689245e-05,
      "loss": 0.1814,
      "step": 78000
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.3629121394757154e-05,
      "loss": 0.193,
      "step": 78500
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.3524848282621846e-05,
      "loss": 0.1805,
      "step": 79000
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.342057517048654e-05,
      "loss": 0.2065,
      "step": 79500
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.3316302058351236e-05,
      "loss": 0.1871,
      "step": 80000
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.321202894621593e-05,
      "loss": 0.191,
      "step": 80500
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.3107755834080626e-05,
      "loss": 0.1948,
      "step": 81000
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.3003482721945325e-05,
      "loss": 0.1803,
      "step": 81500
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.2899209609810016e-05,
      "loss": 0.1796,
      "step": 82000
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.279493649767471e-05,
      "loss": 0.1847,
      "step": 82500
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.2690663385539407e-05,
      "loss": 0.1512,
      "step": 83000
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.25863902734041e-05,
      "loss": 0.2074,
      "step": 83500
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.24821171612688e-05,
      "loss": 0.1718,
      "step": 84000
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.2377844049133495e-05,
      "loss": 0.1735,
      "step": 84500
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.227357093699819e-05,
      "loss": 0.1928,
      "step": 85000
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.216929782486288e-05,
      "loss": 0.1916,
      "step": 85500
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.206502471272758e-05,
      "loss": 0.1708,
      "step": 86000
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.196075160059227e-05,
      "loss": 0.1976,
      "step": 86500
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.185647848845697e-05,
      "loss": 0.1646,
      "step": 87000
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.1752205376321666e-05,
      "loss": 0.1641,
      "step": 87500
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.164793226418636e-05,
      "loss": 0.1692,
      "step": 88000
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.1543659152051056e-05,
      "loss": 0.1572,
      "step": 88500
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.143938603991575e-05,
      "loss": 0.1892,
      "step": 89000
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1335112927780446e-05,
      "loss": 0.1725,
      "step": 89500
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.123083981564514e-05,
      "loss": 0.1531,
      "step": 90000
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.1126566703509836e-05,
      "loss": 0.1645,
      "step": 90500
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.102229359137453e-05,
      "loss": 0.1838,
      "step": 91000
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.0918020479239226e-05,
      "loss": 0.187,
      "step": 91500
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.081374736710392e-05,
      "loss": 0.1606,
      "step": 92000
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.0709474254968617e-05,
      "loss": 0.1669,
      "step": 92500
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.0605201142833315e-05,
      "loss": 0.1454,
      "step": 93000
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.0500928030698007e-05,
      "loss": 0.1528,
      "step": 93500
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.0396654918562702e-05,
      "loss": 0.1522,
      "step": 94000
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.0292381806427393e-05,
      "loss": 0.1398,
      "step": 94500
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.018810869429209e-05,
      "loss": 0.1848,
      "step": 95000
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.0083835582156787e-05,
      "loss": 0.1568,
      "step": 95500
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.9979562470021482e-05,
      "loss": 0.0976,
      "step": 96000
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.9875289357886177e-05,
      "loss": 0.0937,
      "step": 96500
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.9771016245750872e-05,
      "loss": 0.1108,
      "step": 97000
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.9666743133615564e-05,
      "loss": 0.1167,
      "step": 97500
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.9562470021480266e-05,
      "loss": 0.0964,
      "step": 98000
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.945819690934496e-05,
      "loss": 0.1095,
      "step": 98500
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.9353923797209653e-05,
      "loss": 0.1088,
      "step": 99000
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.9249650685074348e-05,
      "loss": 0.1049,
      "step": 99500
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.9145377572939043e-05,
      "loss": 0.1072,
      "step": 100000
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.9041104460803738e-05,
      "loss": 0.1173,
      "step": 100500
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.8936831348668436e-05,
      "loss": 0.1058,
      "step": 101000
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.883255823653313e-05,
      "loss": 0.1017,
      "step": 101500
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.8728285124397823e-05,
      "loss": 0.1008,
      "step": 102000
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.8624012012262518e-05,
      "loss": 0.1167,
      "step": 102500
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.8519738900127213e-05,
      "loss": 0.0925,
      "step": 103000
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.841546578799191e-05,
      "loss": 0.1107,
      "step": 103500
    },
    {
      "epoch": 2.17,
      "learning_rate": 2.8311192675856607e-05,
      "loss": 0.1131,
      "step": 104000
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.8206919563721302e-05,
      "loss": 0.1059,
      "step": 104500
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.8102646451585997e-05,
      "loss": 0.1073,
      "step": 105000
    },
    {
      "epoch": 2.2,
      "learning_rate": 2.799837333945069e-05,
      "loss": 0.112,
      "step": 105500
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.7894100227315384e-05,
      "loss": 0.1114,
      "step": 106000
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.7789827115180082e-05,
      "loss": 0.1333,
      "step": 106500
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.7685554003044777e-05,
      "loss": 0.1007,
      "step": 107000
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.7581280890909472e-05,
      "loss": 0.1127,
      "step": 107500
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.7477007778774168e-05,
      "loss": 0.1023,
      "step": 108000
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.737273466663886e-05,
      "loss": 0.0971,
      "step": 108500
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.7268461554503554e-05,
      "loss": 0.1018,
      "step": 109000
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.7164188442368256e-05,
      "loss": 0.1137,
      "step": 109500
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.7059915330232948e-05,
      "loss": 0.106,
      "step": 110000
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.6955642218097643e-05,
      "loss": 0.0998,
      "step": 110500
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6851369105962338e-05,
      "loss": 0.0934,
      "step": 111000
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.6747095993827033e-05,
      "loss": 0.1166,
      "step": 111500
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.6642822881691725e-05,
      "loss": 0.1028,
      "step": 112000
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.6538549769556427e-05,
      "loss": 0.1005,
      "step": 112500
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.643427665742112e-05,
      "loss": 0.1105,
      "step": 113000
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.6330003545285813e-05,
      "loss": 0.1114,
      "step": 113500
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.622573043315051e-05,
      "loss": 0.0951,
      "step": 114000
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.6121457321015204e-05,
      "loss": 0.1141,
      "step": 114500
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.6017184208879895e-05,
      "loss": 0.0885,
      "step": 115000
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.5912911096744597e-05,
      "loss": 0.112,
      "step": 115500
    },
    {
      "epoch": 2.42,
      "learning_rate": 2.5808637984609292e-05,
      "loss": 0.1167,
      "step": 116000
    },
    {
      "epoch": 2.43,
      "learning_rate": 2.5704364872473984e-05,
      "loss": 0.0975,
      "step": 116500
    },
    {
      "epoch": 2.44,
      "learning_rate": 2.560009176033868e-05,
      "loss": 0.0982,
      "step": 117000
    },
    {
      "epoch": 2.45,
      "learning_rate": 2.5495818648203374e-05,
      "loss": 0.1061,
      "step": 117500
    },
    {
      "epoch": 2.46,
      "learning_rate": 2.5391545536068073e-05,
      "loss": 0.1037,
      "step": 118000
    },
    {
      "epoch": 2.47,
      "learning_rate": 2.5287272423932768e-05,
      "loss": 0.0899,
      "step": 118500
    },
    {
      "epoch": 2.48,
      "learning_rate": 2.5182999311797463e-05,
      "loss": 0.0965,
      "step": 119000
    },
    {
      "epoch": 2.49,
      "learning_rate": 2.5078726199662154e-05,
      "loss": 0.1122,
      "step": 119500
    },
    {
      "epoch": 2.5,
      "learning_rate": 2.4974453087526853e-05,
      "loss": 0.1096,
      "step": 120000
    },
    {
      "epoch": 2.51,
      "learning_rate": 2.4870179975391548e-05,
      "loss": 0.0948,
      "step": 120500
    },
    {
      "epoch": 2.52,
      "learning_rate": 2.476590686325624e-05,
      "loss": 0.0827,
      "step": 121000
    },
    {
      "epoch": 2.53,
      "learning_rate": 2.4661633751120938e-05,
      "loss": 0.0865,
      "step": 121500
    },
    {
      "epoch": 2.54,
      "learning_rate": 2.4557360638985633e-05,
      "loss": 0.0936,
      "step": 122000
    },
    {
      "epoch": 2.55,
      "learning_rate": 2.445308752685033e-05,
      "loss": 0.1097,
      "step": 122500
    },
    {
      "epoch": 2.57,
      "learning_rate": 2.4348814414715023e-05,
      "loss": 0.1022,
      "step": 123000
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.424454130257972e-05,
      "loss": 0.089,
      "step": 123500
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.4140268190444414e-05,
      "loss": 0.0932,
      "step": 124000
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.403599507830911e-05,
      "loss": 0.0871,
      "step": 124500
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.3931721966173804e-05,
      "loss": 0.1136,
      "step": 125000
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.38274488540385e-05,
      "loss": 0.0951,
      "step": 125500
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.3723175741903194e-05,
      "loss": 0.098,
      "step": 126000
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.361890262976789e-05,
      "loss": 0.0828,
      "step": 126500
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.3514629517632584e-05,
      "loss": 0.1078,
      "step": 127000
    },
    {
      "epoch": 2.66,
      "learning_rate": 2.341035640549728e-05,
      "loss": 0.0896,
      "step": 127500
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.3306083293361974e-05,
      "loss": 0.0901,
      "step": 128000
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.3201810181226673e-05,
      "loss": 0.0887,
      "step": 128500
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.3097537069091364e-05,
      "loss": 0.0915,
      "step": 129000
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.299326395695606e-05,
      "loss": 0.0857,
      "step": 129500
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.2888990844820758e-05,
      "loss": 0.1049,
      "step": 130000
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.278471773268545e-05,
      "loss": 0.1008,
      "step": 130500
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.2680444620550145e-05,
      "loss": 0.0904,
      "step": 131000
    },
    {
      "epoch": 2.74,
      "learning_rate": 2.2576171508414843e-05,
      "loss": 0.0968,
      "step": 131500
    },
    {
      "epoch": 2.75,
      "learning_rate": 2.2471898396279535e-05,
      "loss": 0.0824,
      "step": 132000
    },
    {
      "epoch": 2.76,
      "learning_rate": 2.236762528414423e-05,
      "loss": 0.0866,
      "step": 132500
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.226335217200893e-05,
      "loss": 0.0836,
      "step": 133000
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.2159079059873624e-05,
      "loss": 0.1039,
      "step": 133500
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.2054805947738315e-05,
      "loss": 0.0776,
      "step": 134000
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.1950532835603014e-05,
      "loss": 0.0774,
      "step": 134500
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.184625972346771e-05,
      "loss": 0.0937,
      "step": 135000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.17419866113324e-05,
      "loss": 0.0965,
      "step": 135500
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.16377134991971e-05,
      "loss": 0.0917,
      "step": 136000
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.1533440387061794e-05,
      "loss": 0.0817,
      "step": 136500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.1429167274926486e-05,
      "loss": 0.0769,
      "step": 137000
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.1324894162791184e-05,
      "loss": 0.1112,
      "step": 137500
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.122062105065588e-05,
      "loss": 0.0763,
      "step": 138000
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.1116347938520574e-05,
      "loss": 0.0891,
      "step": 138500
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.101207482638527e-05,
      "loss": 0.0885,
      "step": 139000
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.0907801714249965e-05,
      "loss": 0.0709,
      "step": 139500
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.080352860211466e-05,
      "loss": 0.0844,
      "step": 140000
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.0699255489979355e-05,
      "loss": 0.0868,
      "step": 140500
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.059498237784405e-05,
      "loss": 0.0867,
      "step": 141000
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.0490709265708745e-05,
      "loss": 0.0902,
      "step": 141500
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.038643615357344e-05,
      "loss": 0.0807,
      "step": 142000
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.0282163041438135e-05,
      "loss": 0.0819,
      "step": 142500
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.017788992930283e-05,
      "loss": 0.0979,
      "step": 143000
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.0073616817167525e-05,
      "loss": 0.0796,
      "step": 143500
    },
    {
      "epoch": 3.0,
      "learning_rate": 4.9969343705032224e-05,
      "loss": 0.0773,
      "step": 144000
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.9865070592896916e-05,
      "loss": 0.0796,
      "step": 144500
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.9760797480761615e-05,
      "loss": 0.0831,
      "step": 145000
    },
    {
      "epoch": 3.03,
      "learning_rate": 4.9656524368626306e-05,
      "loss": 0.0935,
      "step": 145500
    },
    {
      "epoch": 3.04,
      "learning_rate": 4.9552251256491005e-05,
      "loss": 0.0973,
      "step": 146000
    },
    {
      "epoch": 3.06,
      "learning_rate": 4.94479781443557e-05,
      "loss": 0.1146,
      "step": 146500
    },
    {
      "epoch": 3.07,
      "learning_rate": 4.9343705032220395e-05,
      "loss": 0.1104,
      "step": 147000
    },
    {
      "epoch": 3.08,
      "learning_rate": 4.923943192008509e-05,
      "loss": 0.109,
      "step": 147500
    },
    {
      "epoch": 3.09,
      "learning_rate": 4.9135158807949785e-05,
      "loss": 0.1118,
      "step": 148000
    },
    {
      "epoch": 3.1,
      "learning_rate": 4.903088569581448e-05,
      "loss": 0.1169,
      "step": 148500
    },
    {
      "epoch": 3.11,
      "learning_rate": 4.8926612583679175e-05,
      "loss": 0.1463,
      "step": 149000
    },
    {
      "epoch": 3.12,
      "learning_rate": 4.8822339471543874e-05,
      "loss": 0.1227,
      "step": 149500
    },
    {
      "epoch": 3.13,
      "learning_rate": 4.8718066359408565e-05,
      "loss": 0.1233,
      "step": 150000
    },
    {
      "epoch": 3.14,
      "learning_rate": 4.861379324727326e-05,
      "loss": 0.1258,
      "step": 150500
    },
    {
      "epoch": 3.15,
      "learning_rate": 4.8509520135137956e-05,
      "loss": 0.1439,
      "step": 151000
    },
    {
      "epoch": 3.16,
      "learning_rate": 4.8405247023002654e-05,
      "loss": 0.1225,
      "step": 151500
    },
    {
      "epoch": 3.17,
      "learning_rate": 4.8300973910867346e-05,
      "loss": 0.1204,
      "step": 152000
    },
    {
      "epoch": 3.18,
      "learning_rate": 4.8196700798732044e-05,
      "loss": 0.1256,
      "step": 152500
    },
    {
      "epoch": 3.19,
      "learning_rate": 4.8092427686596736e-05,
      "loss": 0.1498,
      "step": 153000
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.798815457446143e-05,
      "loss": 0.1416,
      "step": 153500
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.7883881462326126e-05,
      "loss": 0.1263,
      "step": 154000
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.7779608350190825e-05,
      "loss": 0.1285,
      "step": 154500
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.7675335238055516e-05,
      "loss": 0.1294,
      "step": 155000
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.7571062125920215e-05,
      "loss": 0.1491,
      "step": 155500
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.7466789013784906e-05,
      "loss": 0.1495,
      "step": 156000
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.73625159016496e-05,
      "loss": 0.1414,
      "step": 156500
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.7258242789514297e-05,
      "loss": 0.1384,
      "step": 157000
    },
    {
      "epoch": 3.28,
      "learning_rate": 4.7153969677378995e-05,
      "loss": 0.1376,
      "step": 157500
    },
    {
      "epoch": 3.3,
      "learning_rate": 4.704969656524369e-05,
      "loss": 0.1456,
      "step": 158000
    },
    {
      "epoch": 3.31,
      "learning_rate": 4.6945423453108385e-05,
      "loss": 0.1326,
      "step": 158500
    },
    {
      "epoch": 3.32,
      "learning_rate": 4.684115034097308e-05,
      "loss": 0.1676,
      "step": 159000
    },
    {
      "epoch": 3.33,
      "learning_rate": 4.6736877228837775e-05,
      "loss": 0.1485,
      "step": 159500
    },
    {
      "epoch": 3.34,
      "learning_rate": 4.6632604116702474e-05,
      "loss": 0.1319,
      "step": 160000
    },
    {
      "epoch": 3.35,
      "learning_rate": 4.6528331004567166e-05,
      "loss": 0.1349,
      "step": 160500
    },
    {
      "epoch": 3.36,
      "learning_rate": 4.642405789243186e-05,
      "loss": 0.1386,
      "step": 161000
    },
    {
      "epoch": 3.37,
      "learning_rate": 4.6319784780296556e-05,
      "loss": 0.1562,
      "step": 161500
    },
    {
      "epoch": 3.38,
      "learning_rate": 4.621551166816125e-05,
      "loss": 0.1441,
      "step": 162000
    },
    {
      "epoch": 3.39,
      "learning_rate": 4.6111238556025946e-05,
      "loss": 0.1413,
      "step": 162500
    },
    {
      "epoch": 3.4,
      "learning_rate": 4.6006965443890644e-05,
      "loss": 0.1614,
      "step": 163000
    },
    {
      "epoch": 3.41,
      "learning_rate": 4.5902692331755336e-05,
      "loss": 0.1516,
      "step": 163500
    },
    {
      "epoch": 3.42,
      "learning_rate": 4.5798419219620035e-05,
      "loss": 0.1448,
      "step": 164000
    },
    {
      "epoch": 3.43,
      "learning_rate": 4.5694146107484726e-05,
      "loss": 0.1514,
      "step": 164500
    },
    {
      "epoch": 3.44,
      "learning_rate": 4.558987299534942e-05,
      "loss": 0.1558,
      "step": 165000
    },
    {
      "epoch": 3.45,
      "learning_rate": 4.5485599883214116e-05,
      "loss": 0.1487,
      "step": 165500
    },
    {
      "epoch": 3.46,
      "learning_rate": 4.5381326771078815e-05,
      "loss": 0.1487,
      "step": 166000
    },
    {
      "epoch": 3.47,
      "learning_rate": 4.5277053658943507e-05,
      "loss": 0.1447,
      "step": 166500
    },
    {
      "epoch": 3.48,
      "learning_rate": 4.5172780546808205e-05,
      "loss": 0.1729,
      "step": 167000
    },
    {
      "epoch": 3.49,
      "learning_rate": 4.50685074346729e-05,
      "loss": 0.1425,
      "step": 167500
    },
    {
      "epoch": 3.5,
      "learning_rate": 4.496423432253759e-05,
      "loss": 0.1562,
      "step": 168000
    },
    {
      "epoch": 3.51,
      "learning_rate": 4.4859961210402294e-05,
      "loss": 0.1481,
      "step": 168500
    },
    {
      "epoch": 3.52,
      "learning_rate": 4.4755688098266985e-05,
      "loss": 0.1466,
      "step": 169000
    },
    {
      "epoch": 3.53,
      "learning_rate": 4.465141498613168e-05,
      "loss": 0.133,
      "step": 169500
    },
    {
      "epoch": 3.55,
      "learning_rate": 4.4547141873996376e-05,
      "loss": 0.1485,
      "step": 170000
    },
    {
      "epoch": 3.56,
      "learning_rate": 4.444286876186107e-05,
      "loss": 0.1332,
      "step": 170500
    },
    {
      "epoch": 3.57,
      "learning_rate": 4.433859564972576e-05,
      "loss": 0.1463,
      "step": 171000
    },
    {
      "epoch": 3.58,
      "learning_rate": 4.4234322537590464e-05,
      "loss": 0.1279,
      "step": 171500
    },
    {
      "epoch": 3.59,
      "learning_rate": 4.4130049425455156e-05,
      "loss": 0.1312,
      "step": 172000
    },
    {
      "epoch": 3.6,
      "learning_rate": 4.402577631331985e-05,
      "loss": 0.1387,
      "step": 172500
    },
    {
      "epoch": 3.61,
      "learning_rate": 4.3921503201184546e-05,
      "loss": 0.1305,
      "step": 173000
    },
    {
      "epoch": 3.62,
      "learning_rate": 4.381723008904924e-05,
      "loss": 0.1403,
      "step": 173500
    },
    {
      "epoch": 3.63,
      "learning_rate": 4.371295697691393e-05,
      "loss": 0.1462,
      "step": 174000
    },
    {
      "epoch": 3.64,
      "learning_rate": 4.3608683864778635e-05,
      "loss": 0.1404,
      "step": 174500
    },
    {
      "epoch": 3.65,
      "learning_rate": 4.3504410752643326e-05,
      "loss": 0.1384,
      "step": 175000
    },
    {
      "epoch": 3.66,
      "learning_rate": 4.340013764050802e-05,
      "loss": 0.1396,
      "step": 175500
    },
    {
      "epoch": 3.67,
      "learning_rate": 4.3295864528372717e-05,
      "loss": 0.1235,
      "step": 176000
    },
    {
      "epoch": 3.68,
      "learning_rate": 4.319159141623741e-05,
      "loss": 0.1337,
      "step": 176500
    },
    {
      "epoch": 3.69,
      "learning_rate": 4.308731830410211e-05,
      "loss": 0.1382,
      "step": 177000
    },
    {
      "epoch": 3.7,
      "learning_rate": 4.2983045191966805e-05,
      "loss": 0.1376,
      "step": 177500
    },
    {
      "epoch": 3.71,
      "learning_rate": 4.28787720798315e-05,
      "loss": 0.1458,
      "step": 178000
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.277449896769619e-05,
      "loss": 0.1541,
      "step": 178500
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.267022585556089e-05,
      "loss": 0.1327,
      "step": 179000
    },
    {
      "epoch": 3.74,
      "learning_rate": 4.256595274342558e-05,
      "loss": 0.1395,
      "step": 179500
    },
    {
      "epoch": 3.75,
      "learning_rate": 4.246167963129028e-05,
      "loss": 0.1272,
      "step": 180000
    },
    {
      "epoch": 3.76,
      "learning_rate": 4.2357406519154976e-05,
      "loss": 0.1276,
      "step": 180500
    },
    {
      "epoch": 3.77,
      "learning_rate": 4.225313340701967e-05,
      "loss": 0.1519,
      "step": 181000
    },
    {
      "epoch": 3.79,
      "learning_rate": 4.2148860294884366e-05,
      "loss": 0.1236,
      "step": 181500
    },
    {
      "epoch": 3.8,
      "learning_rate": 4.204458718274906e-05,
      "loss": 0.1173,
      "step": 182000
    },
    {
      "epoch": 3.81,
      "learning_rate": 4.194031407061375e-05,
      "loss": 0.144,
      "step": 182500
    },
    {
      "epoch": 3.82,
      "learning_rate": 4.183604095847845e-05,
      "loss": 0.1489,
      "step": 183000
    },
    {
      "epoch": 3.83,
      "learning_rate": 4.1731767846343146e-05,
      "loss": 0.1405,
      "step": 183500
    },
    {
      "epoch": 3.84,
      "learning_rate": 4.162749473420784e-05,
      "loss": 0.12,
      "step": 184000
    },
    {
      "epoch": 3.85,
      "learning_rate": 4.1523221622072536e-05,
      "loss": 0.1278,
      "step": 184500
    },
    {
      "epoch": 3.86,
      "learning_rate": 4.141894850993723e-05,
      "loss": 0.1298,
      "step": 185000
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.1314675397801927e-05,
      "loss": 0.1269,
      "step": 185500
    },
    {
      "epoch": 3.88,
      "learning_rate": 4.1210402285666625e-05,
      "loss": 0.1278,
      "step": 186000
    },
    {
      "epoch": 3.89,
      "learning_rate": 4.110612917353132e-05,
      "loss": 0.129,
      "step": 186500
    },
    {
      "epoch": 3.9,
      "learning_rate": 4.100185606139601e-05,
      "loss": 0.1369,
      "step": 187000
    },
    {
      "epoch": 3.91,
      "learning_rate": 4.089758294926071e-05,
      "loss": 0.1337,
      "step": 187500
    },
    {
      "epoch": 3.92,
      "learning_rate": 4.07933098371254e-05,
      "loss": 0.1352,
      "step": 188000
    },
    {
      "epoch": 3.93,
      "learning_rate": 4.06890367249901e-05,
      "loss": 0.1133,
      "step": 188500
    },
    {
      "epoch": 3.94,
      "learning_rate": 4.0584763612854795e-05,
      "loss": 0.1498,
      "step": 189000
    },
    {
      "epoch": 3.95,
      "learning_rate": 4.048049050071949e-05,
      "loss": 0.1344,
      "step": 189500
    },
    {
      "epoch": 3.96,
      "learning_rate": 4.037621738858418e-05,
      "loss": 0.1467,
      "step": 190000
    },
    {
      "epoch": 3.97,
      "learning_rate": 4.027194427644888e-05,
      "loss": 0.1235,
      "step": 190500
    },
    {
      "epoch": 3.98,
      "learning_rate": 4.016767116431357e-05,
      "loss": 0.1333,
      "step": 191000
    },
    {
      "epoch": 3.99,
      "learning_rate": 4.006339805217827e-05,
      "loss": 0.1198,
      "step": 191500
    },
    {
      "epoch": 4.0,
      "learning_rate": 4.9959124940042965e-05,
      "loss": 0.0767,
      "step": 192000
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.985485182790766e-05,
      "loss": 0.0963,
      "step": 192500
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.9750578715772355e-05,
      "loss": 0.1067,
      "step": 193000
    },
    {
      "epoch": 4.04,
      "learning_rate": 4.964630560363705e-05,
      "loss": 0.1075,
      "step": 193500
    },
    {
      "epoch": 4.05,
      "learning_rate": 4.9542032491501745e-05,
      "loss": 0.1101,
      "step": 194000
    },
    {
      "epoch": 4.06,
      "learning_rate": 4.943775937936644e-05,
      "loss": 0.1055,
      "step": 194500
    },
    {
      "epoch": 4.07,
      "learning_rate": 4.9333486267231136e-05,
      "loss": 0.103,
      "step": 195000
    },
    {
      "epoch": 4.08,
      "learning_rate": 4.922921315509583e-05,
      "loss": 0.114,
      "step": 195500
    },
    {
      "epoch": 4.09,
      "learning_rate": 4.9124940042960526e-05,
      "loss": 0.1182,
      "step": 196000
    },
    {
      "epoch": 4.1,
      "learning_rate": 4.9020666930825224e-05,
      "loss": 0.1224,
      "step": 196500
    },
    {
      "epoch": 4.11,
      "learning_rate": 4.8916393818689916e-05,
      "loss": 0.1168,
      "step": 197000
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.881212070655461e-05,
      "loss": 0.1395,
      "step": 197500
    },
    {
      "epoch": 4.13,
      "learning_rate": 4.8707847594419306e-05,
      "loss": 0.1203,
      "step": 198000
    },
    {
      "epoch": 4.14,
      "learning_rate": 4.8603574482284005e-05,
      "loss": 0.1176,
      "step": 198500
    },
    {
      "epoch": 4.15,
      "learning_rate": 4.8499301370148696e-05,
      "loss": 0.1254,
      "step": 199000
    },
    {
      "epoch": 4.16,
      "learning_rate": 4.8395028258013395e-05,
      "loss": 0.1148,
      "step": 199500
    },
    {
      "epoch": 4.17,
      "learning_rate": 4.8290755145878086e-05,
      "loss": 0.1202,
      "step": 200000
    },
    {
      "epoch": 4.18,
      "learning_rate": 4.818648203374278e-05,
      "loss": 0.1228,
      "step": 200500
    },
    {
      "epoch": 4.19,
      "learning_rate": 4.808220892160748e-05,
      "loss": 0.1249,
      "step": 201000
    },
    {
      "epoch": 4.2,
      "learning_rate": 4.7977935809472175e-05,
      "loss": 0.1413,
      "step": 201500
    },
    {
      "epoch": 4.21,
      "learning_rate": 4.787366269733687e-05,
      "loss": 0.151,
      "step": 202000
    },
    {
      "epoch": 4.22,
      "learning_rate": 4.7769389585201565e-05,
      "loss": 0.1217,
      "step": 202500
    },
    {
      "epoch": 4.23,
      "learning_rate": 4.766511647306626e-05,
      "loss": 0.1251,
      "step": 203000
    },
    {
      "epoch": 4.24,
      "learning_rate": 4.756084336093095e-05,
      "loss": 0.1299,
      "step": 203500
    },
    {
      "epoch": 4.25,
      "learning_rate": 4.745657024879565e-05,
      "loss": 0.1337,
      "step": 204000
    },
    {
      "epoch": 4.26,
      "learning_rate": 4.7352297136660346e-05,
      "loss": 0.1249,
      "step": 204500
    },
    {
      "epoch": 4.28,
      "learning_rate": 4.724802402452504e-05,
      "loss": 0.1273,
      "step": 205000
    },
    {
      "epoch": 4.29,
      "learning_rate": 4.7143750912389736e-05,
      "loss": 0.123,
      "step": 205500
    },
    {
      "epoch": 4.3,
      "learning_rate": 4.703947780025443e-05,
      "loss": 0.1142,
      "step": 206000
    },
    {
      "epoch": 4.31,
      "learning_rate": 4.693520468811912e-05,
      "loss": 0.1485,
      "step": 206500
    },
    {
      "epoch": 4.32,
      "learning_rate": 4.683093157598382e-05,
      "loss": 0.1247,
      "step": 207000
    },
    {
      "epoch": 4.33,
      "learning_rate": 4.6726658463848516e-05,
      "loss": 0.1238,
      "step": 207500
    },
    {
      "epoch": 4.34,
      "learning_rate": 4.662238535171321e-05,
      "loss": 0.1242,
      "step": 208000
    },
    {
      "epoch": 4.35,
      "learning_rate": 4.6518112239577906e-05,
      "loss": 0.1253,
      "step": 208500
    },
    {
      "epoch": 4.36,
      "learning_rate": 4.64138391274426e-05,
      "loss": 0.1325,
      "step": 209000
    },
    {
      "epoch": 4.37,
      "learning_rate": 4.630956601530729e-05,
      "loss": 0.1258,
      "step": 209500
    },
    {
      "epoch": 4.38,
      "learning_rate": 4.6205292903171995e-05,
      "loss": 0.1184,
      "step": 210000
    },
    {
      "epoch": 4.39,
      "learning_rate": 4.6101019791036687e-05,
      "loss": 0.1195,
      "step": 210500
    },
    {
      "epoch": 4.4,
      "learning_rate": 4.599674667890138e-05,
      "loss": 0.1382,
      "step": 211000
    },
    {
      "epoch": 4.41,
      "learning_rate": 4.589247356676608e-05,
      "loss": 0.138,
      "step": 211500
    },
    {
      "epoch": 4.42,
      "learning_rate": 4.578820045463077e-05,
      "loss": 0.1297,
      "step": 212000
    },
    {
      "epoch": 4.43,
      "learning_rate": 4.568392734249547e-05,
      "loss": 0.1161,
      "step": 212500
    },
    {
      "epoch": 4.44,
      "learning_rate": 4.5579654230360165e-05,
      "loss": 0.1273,
      "step": 213000
    },
    {
      "epoch": 4.45,
      "learning_rate": 4.547538111822486e-05,
      "loss": 0.1513,
      "step": 213500
    },
    {
      "epoch": 4.46,
      "learning_rate": 4.537110800608955e-05,
      "loss": 0.1043,
      "step": 214000
    },
    {
      "epoch": 4.47,
      "learning_rate": 4.526683489395425e-05,
      "loss": 0.1333,
      "step": 214500
    },
    {
      "epoch": 4.48,
      "learning_rate": 4.516256178181894e-05,
      "loss": 0.1228,
      "step": 215000
    },
    {
      "epoch": 4.49,
      "learning_rate": 4.505828866968364e-05,
      "loss": 0.1255,
      "step": 215500
    },
    {
      "epoch": 4.5,
      "learning_rate": 4.4954015557548336e-05,
      "loss": 0.1047,
      "step": 216000
    },
    {
      "epoch": 4.52,
      "learning_rate": 4.484974244541303e-05,
      "loss": 0.1412,
      "step": 216500
    },
    {
      "epoch": 4.53,
      "learning_rate": 4.4745469333277726e-05,
      "loss": 0.1399,
      "step": 217000
    },
    {
      "epoch": 4.54,
      "learning_rate": 4.464119622114242e-05,
      "loss": 0.12,
      "step": 217500
    },
    {
      "epoch": 4.55,
      "learning_rate": 4.453692310900711e-05,
      "loss": 0.1273,
      "step": 218000
    },
    {
      "epoch": 4.56,
      "learning_rate": 4.443264999687181e-05,
      "loss": 0.1154,
      "step": 218500
    },
    {
      "epoch": 4.57,
      "learning_rate": 4.4328376884736506e-05,
      "loss": 0.1364,
      "step": 219000
    },
    {
      "epoch": 4.58,
      "learning_rate": 4.42241037726012e-05,
      "loss": 0.1231,
      "step": 219500
    },
    {
      "epoch": 4.59,
      "learning_rate": 4.4119830660465897e-05,
      "loss": 0.1127,
      "step": 220000
    },
    {
      "epoch": 4.6,
      "learning_rate": 4.401555754833059e-05,
      "loss": 0.1183,
      "step": 220500
    },
    {
      "epoch": 4.61,
      "learning_rate": 4.391128443619528e-05,
      "loss": 0.1202,
      "step": 221000
    },
    {
      "epoch": 4.62,
      "learning_rate": 4.3807011324059985e-05,
      "loss": 0.1313,
      "step": 221500
    },
    {
      "epoch": 4.63,
      "learning_rate": 4.370273821192468e-05,
      "loss": 0.1213,
      "step": 222000
    },
    {
      "epoch": 4.64,
      "learning_rate": 4.359846509978937e-05,
      "loss": 0.1199,
      "step": 222500
    },
    {
      "epoch": 4.65,
      "learning_rate": 4.349419198765407e-05,
      "loss": 0.1051,
      "step": 223000
    },
    {
      "epoch": 4.66,
      "learning_rate": 4.338991887551876e-05,
      "loss": 0.1251,
      "step": 223500
    },
    {
      "epoch": 4.67,
      "learning_rate": 4.328564576338345e-05,
      "loss": 0.1292,
      "step": 224000
    },
    {
      "epoch": 4.68,
      "learning_rate": 4.3181372651248156e-05,
      "loss": 0.1228,
      "step": 224500
    },
    {
      "epoch": 4.69,
      "learning_rate": 4.307709953911285e-05,
      "loss": 0.1158,
      "step": 225000
    },
    {
      "epoch": 4.7,
      "learning_rate": 4.297282642697754e-05,
      "loss": 0.1084,
      "step": 225500
    },
    {
      "epoch": 4.71,
      "learning_rate": 4.286855331484224e-05,
      "loss": 0.1183,
      "step": 226000
    },
    {
      "epoch": 4.72,
      "learning_rate": 4.276428020270693e-05,
      "loss": 0.1103,
      "step": 226500
    },
    {
      "epoch": 4.73,
      "learning_rate": 4.266000709057163e-05,
      "loss": 0.1037,
      "step": 227000
    },
    {
      "epoch": 4.74,
      "learning_rate": 4.2555733978436326e-05,
      "loss": 0.1093,
      "step": 227500
    },
    {
      "epoch": 4.75,
      "learning_rate": 4.245146086630102e-05,
      "loss": 0.1005,
      "step": 228000
    },
    {
      "epoch": 4.77,
      "learning_rate": 4.234718775416571e-05,
      "loss": 0.1307,
      "step": 228500
    },
    {
      "epoch": 4.78,
      "learning_rate": 4.224291464203041e-05,
      "loss": 0.125,
      "step": 229000
    },
    {
      "epoch": 4.79,
      "learning_rate": 4.21386415298951e-05,
      "loss": 0.113,
      "step": 229500
    },
    {
      "epoch": 4.8,
      "learning_rate": 4.20343684177598e-05,
      "loss": 0.1093,
      "step": 230000
    },
    {
      "epoch": 4.81,
      "learning_rate": 4.19300953056245e-05,
      "loss": 0.1135,
      "step": 230500
    },
    {
      "epoch": 4.82,
      "learning_rate": 4.182582219348919e-05,
      "loss": 0.1086,
      "step": 231000
    },
    {
      "epoch": 4.83,
      "learning_rate": 4.172154908135388e-05,
      "loss": 0.1282,
      "step": 231500
    },
    {
      "epoch": 4.84,
      "learning_rate": 4.161727596921858e-05,
      "loss": 0.0988,
      "step": 232000
    },
    {
      "epoch": 4.85,
      "learning_rate": 4.151300285708327e-05,
      "loss": 0.1118,
      "step": 232500
    },
    {
      "epoch": 4.86,
      "learning_rate": 4.140872974494797e-05,
      "loss": 0.1233,
      "step": 233000
    },
    {
      "epoch": 4.87,
      "learning_rate": 4.130445663281267e-05,
      "loss": 0.1327,
      "step": 233500
    },
    {
      "epoch": 4.88,
      "learning_rate": 4.120018352067736e-05,
      "loss": 0.1061,
      "step": 234000
    },
    {
      "epoch": 4.89,
      "learning_rate": 4.109591040854206e-05,
      "loss": 0.1085,
      "step": 234500
    },
    {
      "epoch": 4.9,
      "learning_rate": 4.099163729640675e-05,
      "loss": 0.0944,
      "step": 235000
    },
    {
      "epoch": 4.91,
      "learning_rate": 4.088736418427145e-05,
      "loss": 0.1009,
      "step": 235500
    },
    {
      "epoch": 4.92,
      "learning_rate": 4.078309107213614e-05,
      "loss": 0.1104,
      "step": 236000
    },
    {
      "epoch": 4.93,
      "learning_rate": 4.067881796000084e-05,
      "loss": 0.1201,
      "step": 236500
    },
    {
      "epoch": 4.94,
      "learning_rate": 4.057454484786553e-05,
      "loss": 0.1067,
      "step": 237000
    },
    {
      "epoch": 4.95,
      "learning_rate": 4.047027173573023e-05,
      "loss": 0.117,
      "step": 237500
    },
    {
      "epoch": 4.96,
      "learning_rate": 4.036599862359492e-05,
      "loss": 0.1221,
      "step": 238000
    },
    {
      "epoch": 4.97,
      "learning_rate": 4.026172551145962e-05,
      "loss": 0.1082,
      "step": 238500
    },
    {
      "epoch": 4.98,
      "learning_rate": 4.0157452399324317e-05,
      "loss": 0.1087,
      "step": 239000
    },
    {
      "epoch": 4.99,
      "learning_rate": 4.005317928718901e-05,
      "loss": 0.1024,
      "step": 239500
    },
    {
      "epoch": 5.0,
      "step": 239755,
      "total_flos": 3.624187009725235e+16,
      "train_loss": 0.02392078927181249,
      "train_runtime": 8161.1668,
      "train_samples_per_second": 117.509,
      "train_steps_per_second": 29.378
    }
  ],
  "max_steps": 239755,
  "num_train_epochs": 5,
  "total_flos": 3.624187009725235e+16,
  "trial_name": null,
  "trial_params": null
}