{
  "best_metric": 0.7570347785949707,
  "best_model_checkpoint": "beto-mulanSDS-esp-mchoice/checkpoint-95184",
  "epoch": 2.0,
  "global_step": 95184,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.973735081526307e-05,
      "loss": 1.0962,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.947470163052614e-05,
      "loss": 0.9714,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.921205244578921e-05,
      "loss": 0.9612,
      "step": 1500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.894940326105228e-05,
      "loss": 0.9123,
      "step": 2000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.868675407631535e-05,
      "loss": 0.8874,
      "step": 2500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.842410489157842e-05,
      "loss": 0.8928,
      "step": 3000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.816145570684149e-05,
      "loss": 0.8706,
      "step": 3500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.7898806522104555e-05,
      "loss": 0.8714,
      "step": 4000
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.7636157337367624e-05,
      "loss": 0.886,
      "step": 4500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.7373508152630694e-05,
      "loss": 0.8701,
      "step": 5000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.711085896789377e-05,
      "loss": 0.841,
      "step": 5500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.684820978315684e-05,
      "loss": 0.817,
      "step": 6000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.65855605984199e-05,
      "loss": 0.8312,
      "step": 6500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.632291141368297e-05,
      "loss": 0.8249,
      "step": 7000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.6060262228946046e-05,
      "loss": 0.8095,
      "step": 7500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.5797613044209115e-05,
      "loss": 0.8356,
      "step": 8000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.5534963859472184e-05,
      "loss": 0.82,
      "step": 8500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.527231467473525e-05,
      "loss": 0.8038,
      "step": 9000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.500966548999832e-05,
      "loss": 0.8042,
      "step": 9500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.474701630526139e-05,
      "loss": 0.7926,
      "step": 10000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.448436712052446e-05,
      "loss": 0.8045,
      "step": 10500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.422171793578753e-05,
      "loss": 0.7923,
      "step": 11000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.39590687510506e-05,
      "loss": 0.8196,
      "step": 11500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.369641956631367e-05,
      "loss": 0.8151,
      "step": 12000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.343377038157674e-05,
      "loss": 0.7516,
      "step": 12500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.3171121196839806e-05,
      "loss": 0.7586,
      "step": 13000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.2908472012102875e-05,
      "loss": 0.7782,
      "step": 13500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.2645822827365944e-05,
      "loss": 0.7728,
      "step": 14000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.2383173642629014e-05,
      "loss": 0.782,
      "step": 14500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.212052445789208e-05,
      "loss": 0.7804,
      "step": 15000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.185787527315516e-05,
      "loss": 0.7705,
      "step": 15500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.159522608841822e-05,
      "loss": 0.7687,
      "step": 16000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.133257690368129e-05,
      "loss": 0.772,
      "step": 16500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.106992771894436e-05,
      "loss": 0.7656,
      "step": 17000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.0807278534207435e-05,
      "loss": 0.7603,
      "step": 17500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.0544629349470504e-05,
      "loss": 0.7927,
      "step": 18000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.0281980164733566e-05,
      "loss": 0.7649,
      "step": 18500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.0019330979996636e-05,
      "loss": 0.7714,
      "step": 19000
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.975668179525971e-05,
      "loss": 0.7505,
      "step": 19500
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.949403261052278e-05,
      "loss": 0.7597,
      "step": 20000
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.923138342578585e-05,
      "loss": 0.752,
      "step": 20500
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.896873424104891e-05,
      "loss": 0.744,
      "step": 21000
    },
    {
      "epoch": 0.45,
      "learning_rate": 3.870608505631199e-05,
      "loss": 0.7451,
      "step": 21500
    },
    {
      "epoch": 0.46,
      "learning_rate": 3.844343587157506e-05,
      "loss": 0.7809,
      "step": 22000
    },
    {
      "epoch": 0.47,
      "learning_rate": 3.8180786686838126e-05,
      "loss": 0.7694,
      "step": 22500
    },
    {
      "epoch": 0.48,
      "learning_rate": 3.7918137502101195e-05,
      "loss": 0.7728,
      "step": 23000
    },
    {
      "epoch": 0.49,
      "learning_rate": 3.7655488317364264e-05,
      "loss": 0.7607,
      "step": 23500
    },
    {
      "epoch": 0.5,
      "learning_rate": 3.7392839132627334e-05,
      "loss": 0.7481,
      "step": 24000
    },
    {
      "epoch": 0.51,
      "learning_rate": 3.71301899478904e-05,
      "loss": 0.7438,
      "step": 24500
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.686754076315347e-05,
      "loss": 0.7502,
      "step": 25000
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.660489157841655e-05,
      "loss": 0.7354,
      "step": 25500
    },
    {
      "epoch": 0.55,
      "learning_rate": 3.634224239367961e-05,
      "loss": 0.7971,
      "step": 26000
    },
    {
      "epoch": 0.56,
      "learning_rate": 3.607959320894268e-05,
      "loss": 0.7605,
      "step": 26500
    },
    {
      "epoch": 0.57,
      "learning_rate": 3.581694402420575e-05,
      "loss": 0.7513,
      "step": 27000
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.5554294839468824e-05,
      "loss": 0.7515,
      "step": 27500
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.529164565473189e-05,
      "loss": 0.7228,
      "step": 28000
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.5028996469994956e-05,
      "loss": 0.7596,
      "step": 28500
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.4766347285258025e-05,
      "loss": 0.7645,
      "step": 29000
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.45036981005211e-05,
      "loss": 0.7634,
      "step": 29500
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.424104891578417e-05,
      "loss": 0.7581,
      "step": 30000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.397839973104724e-05,
      "loss": 0.7504,
      "step": 30500
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.37157505463103e-05,
      "loss": 0.7309,
      "step": 31000
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.345310136157338e-05,
      "loss": 0.7705,
      "step": 31500
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.3190452176836446e-05,
      "loss": 0.777,
      "step": 32000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.2927802992099515e-05,
      "loss": 0.753,
      "step": 32500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.2665153807362584e-05,
      "loss": 0.7707,
      "step": 33000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.2402504622625654e-05,
      "loss": 0.7327,
      "step": 33500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.213985543788872e-05,
      "loss": 0.7693,
      "step": 34000
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.187720625315179e-05,
      "loss": 0.7579,
      "step": 34500
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.161455706841486e-05,
      "loss": 0.7581,
      "step": 35000
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.135190788367793e-05,
      "loss": 0.77,
      "step": 35500
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.1089258698941e-05,
      "loss": 0.7263,
      "step": 36000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.082660951420407e-05,
      "loss": 0.7478,
      "step": 36500
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.056396032946714e-05,
      "loss": 0.7646,
      "step": 37000
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.030131114473021e-05,
      "loss": 0.7408,
      "step": 37500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.0038661959993276e-05,
      "loss": 0.7101,
      "step": 38000
    },
    {
      "epoch": 0.81,
      "learning_rate": 2.9776012775256345e-05,
      "loss": 0.7487,
      "step": 38500
    },
    {
      "epoch": 0.82,
      "learning_rate": 2.9513363590519417e-05,
      "loss": 0.732,
      "step": 39000
    },
    {
      "epoch": 0.83,
      "learning_rate": 2.9250714405782486e-05,
      "loss": 0.7522,
      "step": 39500
    },
    {
      "epoch": 0.84,
      "learning_rate": 2.898806522104556e-05,
      "loss": 0.7193,
      "step": 40000
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.872541603630862e-05,
      "loss": 0.7343,
      "step": 40500
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.8462766851571694e-05,
      "loss": 0.7295,
      "step": 41000
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.8200117666834763e-05,
      "loss": 0.7265,
      "step": 41500
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.7937468482097835e-05,
      "loss": 0.7143,
      "step": 42000
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.7674819297360904e-05,
      "loss": 0.6877,
      "step": 42500
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.741217011262397e-05,
      "loss": 0.7043,
      "step": 43000
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.714952092788704e-05,
      "loss": 0.7297,
      "step": 43500
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.6886871743150112e-05,
      "loss": 0.7749,
      "step": 44000
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.662422255841318e-05,
      "loss": 0.7227,
      "step": 44500
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.636157337367625e-05,
      "loss": 0.709,
      "step": 45000
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.6098924188939316e-05,
      "loss": 0.7258,
      "step": 45500
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.5836275004202388e-05,
      "loss": 0.6954,
      "step": 46000
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.5573625819465457e-05,
      "loss": 0.7173,
      "step": 46500
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.531097663472853e-05,
      "loss": 0.713,
      "step": 47000
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.50483274499916e-05,
      "loss": 0.7,
      "step": 47500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7327029705047607,
      "eval_loss": 0.7685355544090271,
      "eval_runtime": 578.0716,
      "eval_samples_per_second": 68.607,
      "eval_steps_per_second": 11.435,
      "step": 47592
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.4785678265254668e-05,
      "loss": 0.6245,
      "step": 48000
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.4523029080517734e-05,
      "loss": 0.6892,
      "step": 48500
    },
    {
      "epoch": 1.03,
      "learning_rate": 2.4260379895780806e-05,
      "loss": 0.628,
      "step": 49000
    },
    {
      "epoch": 1.04,
      "learning_rate": 2.3997730711043875e-05,
      "loss": 0.6394,
      "step": 49500
    },
    {
      "epoch": 1.05,
      "learning_rate": 2.3735081526306945e-05,
      "loss": 0.6367,
      "step": 50000
    },
    {
      "epoch": 1.06,
      "learning_rate": 2.3472432341570014e-05,
      "loss": 0.6092,
      "step": 50500
    },
    {
      "epoch": 1.07,
      "learning_rate": 2.3209783156833083e-05,
      "loss": 0.6286,
      "step": 51000
    },
    {
      "epoch": 1.08,
      "learning_rate": 2.2947133972096152e-05,
      "loss": 0.6809,
      "step": 51500
    },
    {
      "epoch": 1.09,
      "learning_rate": 2.268448478735922e-05,
      "loss": 0.6185,
      "step": 52000
    },
    {
      "epoch": 1.1,
      "learning_rate": 2.242183560262229e-05,
      "loss": 0.578,
      "step": 52500
    },
    {
      "epoch": 1.11,
      "learning_rate": 2.215918641788536e-05,
      "loss": 0.6271,
      "step": 53000
    },
    {
      "epoch": 1.12,
      "learning_rate": 2.189653723314843e-05,
      "loss": 0.654,
      "step": 53500
    },
    {
      "epoch": 1.13,
      "learning_rate": 2.16338880484115e-05,
      "loss": 0.6708,
      "step": 54000
    },
    {
      "epoch": 1.15,
      "learning_rate": 2.1371238863674567e-05,
      "loss": 0.6218,
      "step": 54500
    },
    {
      "epoch": 1.16,
      "learning_rate": 2.110858967893764e-05,
      "loss": 0.6413,
      "step": 55000
    },
    {
      "epoch": 1.17,
      "learning_rate": 2.0845940494200708e-05,
      "loss": 0.6441,
      "step": 55500
    },
    {
      "epoch": 1.18,
      "learning_rate": 2.0583291309463777e-05,
      "loss": 0.6444,
      "step": 56000
    },
    {
      "epoch": 1.19,
      "learning_rate": 2.0320642124726846e-05,
      "loss": 0.6311,
      "step": 56500
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.0057992939989916e-05,
      "loss": 0.5755,
      "step": 57000
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.9795343755252985e-05,
      "loss": 0.6561,
      "step": 57500
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.9532694570516054e-05,
      "loss": 0.6098,
      "step": 58000
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.9270045385779123e-05,
      "loss": 0.6406,
      "step": 58500
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.9007396201042192e-05,
      "loss": 0.6184,
      "step": 59000
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.874474701630526e-05,
      "loss": 0.603,
      "step": 59500
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.848209783156833e-05,
      "loss": 0.5957,
      "step": 60000
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.8219448646831403e-05,
      "loss": 0.5994,
      "step": 60500
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.7956799462094472e-05,
      "loss": 0.6365,
      "step": 61000
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.769415027735754e-05,
      "loss": 0.6169,
      "step": 61500
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.743150109262061e-05,
      "loss": 0.63,
      "step": 62000
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.716885190788368e-05,
      "loss": 0.6202,
      "step": 62500
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.6906202723146748e-05,
      "loss": 0.6016,
      "step": 63000
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.6643553538409817e-05,
      "loss": 0.6026,
      "step": 63500
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.6380904353672887e-05,
      "loss": 0.6037,
      "step": 64000
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.6118255168935956e-05,
      "loss": 0.6282,
      "step": 64500
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.5855605984199025e-05,
      "loss": 0.6208,
      "step": 65000
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.5592956799462094e-05,
      "loss": 0.6094,
      "step": 65500
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.5330307614725163e-05,
      "loss": 0.6494,
      "step": 66000
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.5067658429988235e-05,
      "loss": 0.6306,
      "step": 66500
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.4805009245251303e-05,
      "loss": 0.6054,
      "step": 67000
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.4542360060514374e-05,
      "loss": 0.6126,
      "step": 67500
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.4279710875777441e-05,
      "loss": 0.6283,
      "step": 68000
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.4017061691040512e-05,
      "loss": 0.635,
      "step": 68500
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.3754412506303583e-05,
      "loss": 0.6037,
      "step": 69000
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.349176332156665e-05,
      "loss": 0.6232,
      "step": 69500
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.3229114136829721e-05,
      "loss": 0.6044,
      "step": 70000
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.2966464952092788e-05,
      "loss": 0.6094,
      "step": 70500
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.270381576735586e-05,
      "loss": 0.5856,
      "step": 71000
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.2441166582618928e-05,
      "loss": 0.6061,
      "step": 71500
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.2178517397881997e-05,
      "loss": 0.571,
      "step": 72000
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.1915868213145067e-05,
      "loss": 0.5986,
      "step": 72500
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.1653219028408137e-05,
      "loss": 0.6136,
      "step": 73000
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.1390569843671207e-05,
      "loss": 0.6133,
      "step": 73500
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.1127920658934276e-05,
      "loss": 0.6026,
      "step": 74000
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.0865271474197345e-05,
      "loss": 0.5867,
      "step": 74500
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.0602622289460414e-05,
      "loss": 0.6146,
      "step": 75000
    },
    {
      "epoch": 1.59,
      "learning_rate": 1.0339973104723485e-05,
      "loss": 0.6087,
      "step": 75500
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.0077323919986554e-05,
      "loss": 0.6133,
      "step": 76000
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.814674735249623e-06,
      "loss": 0.5743,
      "step": 76500
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.552025550512692e-06,
      "loss": 0.606,
      "step": 77000
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.289376365775761e-06,
      "loss": 0.6097,
      "step": 77500
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.02672718103883e-06,
      "loss": 0.5951,
      "step": 78000
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.764077996301901e-06,
      "loss": 0.6024,
      "step": 78500
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.50142881156497e-06,
      "loss": 0.5819,
      "step": 79000
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.23877962682804e-06,
      "loss": 0.592,
      "step": 79500
    },
    {
      "epoch": 1.68,
      "learning_rate": 7.976130442091108e-06,
      "loss": 0.5664,
      "step": 80000
    },
    {
      "epoch": 1.69,
      "learning_rate": 7.713481257354178e-06,
      "loss": 0.6218,
      "step": 80500
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.4508320726172475e-06,
      "loss": 0.5798,
      "step": 81000
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.188182887880317e-06,
      "loss": 0.589,
      "step": 81500
    },
    {
      "epoch": 1.72,
      "learning_rate": 6.925533703143386e-06,
      "loss": 0.5907,
      "step": 82000
    },
    {
      "epoch": 1.73,
      "learning_rate": 6.662884518406455e-06,
      "loss": 0.593,
      "step": 82500
    },
    {
      "epoch": 1.74,
      "learning_rate": 6.400235333669524e-06,
      "loss": 0.5453,
      "step": 83000
    },
    {
      "epoch": 1.75,
      "learning_rate": 6.137586148932594e-06,
      "loss": 0.5838,
      "step": 83500
    },
    {
      "epoch": 1.77,
      "learning_rate": 5.874936964195663e-06,
      "loss": 0.585,
      "step": 84000
    },
    {
      "epoch": 1.78,
      "learning_rate": 5.612287779458733e-06,
      "loss": 0.6179,
      "step": 84500
    },
    {
      "epoch": 1.79,
      "learning_rate": 5.349638594721802e-06,
      "loss": 0.6065,
      "step": 85000
    },
    {
      "epoch": 1.8,
      "learning_rate": 5.086989409984871e-06,
      "loss": 0.55,
      "step": 85500
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.824340225247941e-06,
      "loss": 0.6015,
      "step": 86000
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.56169104051101e-06,
      "loss": 0.5786,
      "step": 86500
    },
    {
      "epoch": 1.83,
      "learning_rate": 4.299041855774079e-06,
      "loss": 0.5576,
      "step": 87000
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.036392671037149e-06,
      "loss": 0.569,
      "step": 87500
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.7737434863002185e-06,
      "loss": 0.5692,
      "step": 88000
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.5110943015632885e-06,
      "loss": 0.5528,
      "step": 88500
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.2484451168263576e-06,
      "loss": 0.5592,
      "step": 89000
    },
    {
      "epoch": 1.88,
      "learning_rate": 2.985795932089427e-06,
      "loss": 0.5756,
      "step": 89500
    },
    {
      "epoch": 1.89,
      "learning_rate": 2.7231467473524962e-06,
      "loss": 0.5845,
      "step": 90000
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.4604975626155658e-06,
      "loss": 0.5619,
      "step": 90500
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.1978483778786353e-06,
      "loss": 0.5945,
      "step": 91000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.935199193141705e-06,
      "loss": 0.5672,
      "step": 91500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.672550008404774e-06,
      "loss": 0.5588,
      "step": 92000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.4099008236678435e-06,
      "loss": 0.5619,
      "step": 92500
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.1472516389309128e-06,
      "loss": 0.5772,
      "step": 93000
    },
    {
      "epoch": 1.96,
      "learning_rate": 8.846024541939823e-07,
      "loss": 0.5774,
      "step": 93500
    },
    {
      "epoch": 1.98,
      "learning_rate": 6.219532694570517e-07,
      "loss": 0.5765,
      "step": 94000
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.5930408472012103e-07,
      "loss": 0.5612,
      "step": 94500
    },
    {
      "epoch": 2.0,
      "learning_rate": 9.665489998319047e-08,
      "loss": 0.6085,
      "step": 95000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7570347785949707,
      "eval_loss": 0.6955135464668274,
      "eval_runtime": 589.2402,
      "eval_samples_per_second": 67.307,
      "eval_steps_per_second": 11.218,
      "step": 95184
    }
  ],
  "max_steps": 95184,
  "num_train_epochs": 2,
  "total_flos": 1.2628160717770704e+17,
  "trial_name": null,
  "trial_params": null
}