{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.8957528957528957,
  "global_step": 30000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 2.9903474903474904e-05,
      "loss": 4.7662,
      "step": 100
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9806949806949808e-05,
      "loss": 5.8747,
      "step": 200
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.9710424710424712e-05,
      "loss": 6.2395,
      "step": 300
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.9613899613899616e-05,
      "loss": 6.2412,
      "step": 400
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.9517374517374516e-05,
      "loss": 6.2405,
      "step": 500
    },
    {
      "epoch": 0.06,
      "learning_rate": 2.9420849420849423e-05,
      "loss": 3.4061,
      "step": 600
    },
    {
      "epoch": 0.07,
      "learning_rate": 2.9324324324324327e-05,
      "loss": 2.1295,
      "step": 700
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.9227799227799227e-05,
      "loss": 2.0931,
      "step": 800
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.913127413127413e-05,
      "loss": 1.9645,
      "step": 900
    },
    {
      "epoch": 0.1,
      "learning_rate": 2.9034749034749038e-05,
      "loss": 1.8975,
      "step": 1000
    },
    {
      "epoch": 0.11,
      "learning_rate": 2.8938223938223938e-05,
      "loss": 1.8587,
      "step": 1100
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.8841698841698842e-05,
      "loss": 1.7731,
      "step": 1200
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.874517374517375e-05,
      "loss": 1.7233,
      "step": 1300
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.864864864864865e-05,
      "loss": 1.6743,
      "step": 1400
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.8552123552123553e-05,
      "loss": 1.6522,
      "step": 1500
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.8455598455598453e-05,
      "loss": 1.579,
      "step": 1600
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.835907335907336e-05,
      "loss": 1.6699,
      "step": 1700
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.8262548262548264e-05,
      "loss": 2.5233,
      "step": 1800
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.8166023166023164e-05,
      "loss": 1.8225,
      "step": 1900
    },
    {
      "epoch": 0.19,
      "learning_rate": 2.806949806949807e-05,
      "loss": 1.5727,
      "step": 2000
    },
    {
      "epoch": 0.2,
      "learning_rate": 2.7972972972972975e-05,
      "loss": 1.5446,
      "step": 2100
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.7876447876447876e-05,
      "loss": 1.5248,
      "step": 2200
    },
    {
      "epoch": 0.22,
      "learning_rate": 2.777992277992278e-05,
      "loss": 1.5045,
      "step": 2300
    },
    {
      "epoch": 0.23,
      "learning_rate": 2.7683397683397686e-05,
      "loss": 1.4864,
      "step": 2400
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.7586872586872587e-05,
      "loss": 1.4444,
      "step": 2500
    },
    {
      "epoch": 0.25,
      "learning_rate": 2.749034749034749e-05,
      "loss": 1.4808,
      "step": 2600
    },
    {
      "epoch": 0.26,
      "learning_rate": 2.7393822393822394e-05,
      "loss": 1.4867,
      "step": 2700
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.7297297297297298e-05,
      "loss": 1.4764,
      "step": 2800
    },
    {
      "epoch": 0.28,
      "learning_rate": 2.72007722007722e-05,
      "loss": 1.4469,
      "step": 2900
    },
    {
      "epoch": 0.29,
      "learning_rate": 2.7104247104247105e-05,
      "loss": 1.4145,
      "step": 3000
    },
    {
      "epoch": 0.3,
      "learning_rate": 2.700772200772201e-05,
      "loss": 1.4232,
      "step": 3100
    },
    {
      "epoch": 0.31,
      "learning_rate": 2.6911196911196913e-05,
      "loss": 1.3993,
      "step": 3200
    },
    {
      "epoch": 0.32,
      "learning_rate": 2.6814671814671816e-05,
      "loss": 1.4332,
      "step": 3300
    },
    {
      "epoch": 0.33,
      "learning_rate": 2.6718146718146717e-05,
      "loss": 1.4679,
      "step": 3400
    },
    {
      "epoch": 0.34,
      "learning_rate": 2.6621621621621624e-05,
      "loss": 1.3816,
      "step": 3500
    },
    {
      "epoch": 0.35,
      "learning_rate": 2.6525096525096527e-05,
      "loss": 1.4694,
      "step": 3600
    },
    {
      "epoch": 0.36,
      "learning_rate": 2.6428571428571428e-05,
      "loss": 1.4303,
      "step": 3700
    },
    {
      "epoch": 0.37,
      "learning_rate": 2.633204633204633e-05,
      "loss": 1.3515,
      "step": 3800
    },
    {
      "epoch": 0.38,
      "learning_rate": 2.623552123552124e-05,
      "loss": 1.4127,
      "step": 3900
    },
    {
      "epoch": 0.39,
      "learning_rate": 2.613899613899614e-05,
      "loss": 1.3385,
      "step": 4000
    },
    {
      "epoch": 0.4,
      "learning_rate": 2.6042471042471043e-05,
      "loss": 1.2839,
      "step": 4100
    },
    {
      "epoch": 0.41,
      "learning_rate": 2.594594594594595e-05,
      "loss": 1.35,
      "step": 4200
    },
    {
      "epoch": 0.42,
      "learning_rate": 2.584942084942085e-05,
      "loss": 1.3536,
      "step": 4300
    },
    {
      "epoch": 0.42,
      "learning_rate": 2.5752895752895754e-05,
      "loss": 1.3542,
      "step": 4400
    },
    {
      "epoch": 0.43,
      "learning_rate": 2.5656370656370657e-05,
      "loss": 1.3096,
      "step": 4500
    },
    {
      "epoch": 0.44,
      "learning_rate": 2.555984555984556e-05,
      "loss": 1.4777,
      "step": 4600
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.5463320463320465e-05,
      "loss": 1.2712,
      "step": 4700
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.5366795366795365e-05,
      "loss": 1.3084,
      "step": 4800
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.527027027027027e-05,
      "loss": 1.3296,
      "step": 4900
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.5173745173745176e-05,
      "loss": 1.5364,
      "step": 5000
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.5077220077220076e-05,
      "loss": 2.753,
      "step": 5100
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.498069498069498e-05,
      "loss": 1.742,
      "step": 5200
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.4884169884169887e-05,
      "loss": 1.6347,
      "step": 5300
    },
    {
      "epoch": 0.52,
      "learning_rate": 2.4787644787644787e-05,
      "loss": 1.6604,
      "step": 5400
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.469111969111969e-05,
      "loss": 1.5615,
      "step": 5500
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.4594594594594595e-05,
      "loss": 1.3292,
      "step": 5600
    },
    {
      "epoch": 0.55,
      "learning_rate": 2.44980694980695e-05,
      "loss": 1.2548,
      "step": 5700
    },
    {
      "epoch": 0.56,
      "learning_rate": 2.4401544401544402e-05,
      "loss": 1.2897,
      "step": 5800
    },
    {
      "epoch": 0.57,
      "learning_rate": 2.4305019305019306e-05,
      "loss": 1.2414,
      "step": 5900
    },
    {
      "epoch": 0.58,
      "learning_rate": 2.420849420849421e-05,
      "loss": 1.2883,
      "step": 6000
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.4111969111969113e-05,
      "loss": 1.3017,
      "step": 6100
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.4015444015444017e-05,
      "loss": 1.2273,
      "step": 6200
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.3918918918918917e-05,
      "loss": 1.2292,
      "step": 6300
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.3822393822393824e-05,
      "loss": 1.3544,
      "step": 6400
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.3725868725868728e-05,
      "loss": 1.2666,
      "step": 6500
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.362934362934363e-05,
      "loss": 1.2791,
      "step": 6600
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.3532818532818532e-05,
      "loss": 1.3095,
      "step": 6700
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.343629343629344e-05,
      "loss": 1.2471,
      "step": 6800
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.333976833976834e-05,
      "loss": 1.3152,
      "step": 6900
    },
    {
      "epoch": 0.68,
      "learning_rate": 2.3243243243243243e-05,
      "loss": 1.1917,
      "step": 7000
    },
    {
      "epoch": 0.69,
      "learning_rate": 2.314671814671815e-05,
      "loss": 1.2383,
      "step": 7100
    },
    {
      "epoch": 0.69,
      "learning_rate": 2.305019305019305e-05,
      "loss": 1.2902,
      "step": 7200
    },
    {
      "epoch": 0.7,
      "learning_rate": 2.2953667953667954e-05,
      "loss": 1.2329,
      "step": 7300
    },
    {
      "epoch": 0.71,
      "learning_rate": 2.2857142857142858e-05,
      "loss": 1.1985,
      "step": 7400
    },
    {
      "epoch": 0.72,
      "learning_rate": 2.2760617760617762e-05,
      "loss": 1.1914,
      "step": 7500
    },
    {
      "epoch": 0.73,
      "learning_rate": 2.2664092664092665e-05,
      "loss": 1.2317,
      "step": 7600
    },
    {
      "epoch": 0.74,
      "learning_rate": 2.2567567567567566e-05,
      "loss": 1.2853,
      "step": 7700
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.247104247104247e-05,
      "loss": 1.2564,
      "step": 7800
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.2374517374517377e-05,
      "loss": 1.2579,
      "step": 7900
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.2277992277992277e-05,
      "loss": 1.1445,
      "step": 8000
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.218146718146718e-05,
      "loss": 1.1992,
      "step": 8100
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.2084942084942088e-05,
      "loss": 1.1943,
      "step": 8200
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.1988416988416988e-05,
      "loss": 1.1685,
      "step": 8300
    },
    {
      "epoch": 0.81,
      "learning_rate": 2.1891891891891892e-05,
      "loss": 1.1765,
      "step": 8400
    },
    {
      "epoch": 0.82,
      "learning_rate": 2.1795366795366795e-05,
      "loss": 1.168,
      "step": 8500
    },
    {
      "epoch": 0.83,
      "learning_rate": 2.16988416988417e-05,
      "loss": 1.2075,
      "step": 8600
    },
    {
      "epoch": 0.84,
      "learning_rate": 2.1602316602316603e-05,
      "loss": 1.1779,
      "step": 8700
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.1505791505791507e-05,
      "loss": 1.1858,
      "step": 8800
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.140926640926641e-05,
      "loss": 1.1828,
      "step": 8900
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.1312741312741314e-05,
      "loss": 1.2031,
      "step": 9000
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.1216216216216218e-05,
      "loss": 1.1677,
      "step": 9100
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.1119691119691118e-05,
      "loss": 1.1525,
      "step": 9200
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.1023166023166025e-05,
      "loss": 1.1876,
      "step": 9300
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.092664092664093e-05,
      "loss": 1.1632,
      "step": 9400
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.083011583011583e-05,
      "loss": 1.1599,
      "step": 9500
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.0733590733590733e-05,
      "loss": 1.1763,
      "step": 9600
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.063706563706564e-05,
      "loss": 1.1605,
      "step": 9700
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.054054054054054e-05,
      "loss": 1.3013,
      "step": 9800
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.0444015444015444e-05,
      "loss": 1.1579,
      "step": 9900
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.034749034749035e-05,
      "loss": 1.2543,
      "step": 10000
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.025096525096525e-05,
      "loss": 1.1719,
      "step": 10100
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.0154440154440155e-05,
      "loss": 1.1845,
      "step": 10200
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.005791505791506e-05,
      "loss": 1.2079,
      "step": 10300
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.9961389961389962e-05,
      "loss": 1.0919,
      "step": 10400
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.9864864864864866e-05,
      "loss": 1.0697,
      "step": 10500
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.976833976833977e-05,
      "loss": 1.009,
      "step": 10600
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.967181467181467e-05,
      "loss": 1.0711,
      "step": 10700
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.9575289575289577e-05,
      "loss": 0.9108,
      "step": 10800
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.9478764478764478e-05,
      "loss": 0.9708,
      "step": 10900
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.938223938223938e-05,
      "loss": 0.9146,
      "step": 11000
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.928571428571429e-05,
      "loss": 0.9609,
      "step": 11100
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.918918918918919e-05,
      "loss": 1.0282,
      "step": 11200
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.9092664092664092e-05,
      "loss": 0.9474,
      "step": 11300
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.8996138996138996e-05,
      "loss": 0.9342,
      "step": 11400
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.88996138996139e-05,
      "loss": 0.9938,
      "step": 11500
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.8803088803088804e-05,
      "loss": 0.9652,
      "step": 11600
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.8706563706563707e-05,
      "loss": 0.9364,
      "step": 11700
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.8610038610038608e-05,
      "loss": 0.9736,
      "step": 11800
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.8513513513513515e-05,
      "loss": 1.0112,
      "step": 11900
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.841698841698842e-05,
      "loss": 1.0339,
      "step": 12000
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.832046332046332e-05,
      "loss": 0.9825,
      "step": 12100
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.8223938223938226e-05,
      "loss": 0.9595,
      "step": 12200
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.812741312741313e-05,
      "loss": 0.9858,
      "step": 12300
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.803088803088803e-05,
      "loss": 0.9754,
      "step": 12400
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.7934362934362933e-05,
      "loss": 1.0461,
      "step": 12500
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.783783783783784e-05,
      "loss": 0.9761,
      "step": 12600
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.774131274131274e-05,
      "loss": 0.999,
      "step": 12700
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.7644787644787645e-05,
      "loss": 0.9608,
      "step": 12800
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.7548262548262552e-05,
      "loss": 0.9852,
      "step": 12900
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.7451737451737452e-05,
      "loss": 0.9841,
      "step": 13000
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.7355212355212356e-05,
      "loss": 0.9992,
      "step": 13100
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.725868725868726e-05,
      "loss": 0.9912,
      "step": 13200
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.7162162162162163e-05,
      "loss": 0.998,
      "step": 13300
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.7065637065637067e-05,
      "loss": 0.9671,
      "step": 13400
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.696911196911197e-05,
      "loss": 0.9372,
      "step": 13500
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.687258687258687e-05,
      "loss": 1.0567,
      "step": 13600
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.6776061776061778e-05,
      "loss": 0.9795,
      "step": 13700
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.6679536679536678e-05,
      "loss": 0.955,
      "step": 13800
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.6583011583011582e-05,
      "loss": 1.0127,
      "step": 13900
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.648648648648649e-05,
      "loss": 0.9543,
      "step": 14000
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.638996138996139e-05,
      "loss": 0.9838,
      "step": 14100
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.6293436293436293e-05,
      "loss": 0.9507,
      "step": 14200
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.6196911196911197e-05,
      "loss": 0.9387,
      "step": 14300
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.61003861003861e-05,
      "loss": 0.9229,
      "step": 14400
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.6003861003861004e-05,
      "loss": 0.99,
      "step": 14500
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.5907335907335908e-05,
      "loss": 0.9069,
      "step": 14600
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.5810810810810808e-05,
      "loss": 0.9713,
      "step": 14700
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.5714285714285715e-05,
      "loss": 0.9303,
      "step": 14800
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.561776061776062e-05,
      "loss": 0.9915,
      "step": 14900
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.552123552123552e-05,
      "loss": 0.9553,
      "step": 15000
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.5424710424710426e-05,
      "loss": 0.9928,
      "step": 15100
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.532818532818533e-05,
      "loss": 0.9777,
      "step": 15200
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.5231660231660232e-05,
      "loss": 0.9512,
      "step": 15300
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.5135135135135134e-05,
      "loss": 0.9259,
      "step": 15400
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.503861003861004e-05,
      "loss": 0.8975,
      "step": 15500
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.4942084942084943e-05,
      "loss": 1.0091,
      "step": 15600
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.4845559845559845e-05,
      "loss": 1.0214,
      "step": 15700
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.4749034749034749e-05,
      "loss": 0.9119,
      "step": 15800
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.4652509652509654e-05,
      "loss": 0.9976,
      "step": 15900
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.4555984555984556e-05,
      "loss": 0.9755,
      "step": 16000
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.445945945945946e-05,
      "loss": 0.9176,
      "step": 16100
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.4362934362934362e-05,
      "loss": 0.9487,
      "step": 16200
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.4266409266409268e-05,
      "loss": 0.947,
      "step": 16300
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.416988416988417e-05,
      "loss": 1.0008,
      "step": 16400
    },
    {
      "epoch": 1.59,
      "learning_rate": 1.4073359073359073e-05,
      "loss": 0.9436,
      "step": 16500
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.3976833976833977e-05,
      "loss": 0.9633,
      "step": 16600
    },
    {
      "epoch": 1.61,
      "learning_rate": 1.388030888030888e-05,
      "loss": 0.9685,
      "step": 16700
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.3783783783783784e-05,
      "loss": 0.9424,
      "step": 16800
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.3687258687258688e-05,
      "loss": 0.8993,
      "step": 16900
    },
    {
      "epoch": 1.64,
      "learning_rate": 1.3590733590733592e-05,
      "loss": 0.9622,
      "step": 17000
    },
    {
      "epoch": 1.65,
      "learning_rate": 1.3494208494208494e-05,
      "loss": 0.9735,
      "step": 17100
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.33976833976834e-05,
      "loss": 0.9564,
      "step": 17200
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.3301158301158301e-05,
      "loss": 0.9575,
      "step": 17300
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.3204633204633205e-05,
      "loss": 0.8914,
      "step": 17400
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.3108108108108109e-05,
      "loss": 0.9441,
      "step": 17500
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.3011583011583012e-05,
      "loss": 1.0149,
      "step": 17600
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.2915057915057914e-05,
      "loss": 0.9439,
      "step": 17700
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.2818532818532818e-05,
      "loss": 0.9569,
      "step": 17800
    },
    {
      "epoch": 1.73,
      "learning_rate": 1.2722007722007723e-05,
      "loss": 0.8933,
      "step": 17900
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.2625482625482625e-05,
      "loss": 0.9097,
      "step": 18000
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.2528957528957529e-05,
      "loss": 0.8836,
      "step": 18100
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.2432432432432433e-05,
      "loss": 0.9698,
      "step": 18200
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.2335907335907337e-05,
      "loss": 0.9441,
      "step": 18300
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.2239382239382239e-05,
      "loss": 0.9384,
      "step": 18400
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.2142857142857144e-05,
      "loss": 0.9416,
      "step": 18500
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.2046332046332046e-05,
      "loss": 0.9299,
      "step": 18600
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.194980694980695e-05,
      "loss": 0.967,
      "step": 18700
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.1853281853281855e-05,
      "loss": 0.8969,
      "step": 18800
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.1756756756756757e-05,
      "loss": 0.925,
      "step": 18900
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.166023166023166e-05,
      "loss": 0.942,
      "step": 19000
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.1563706563706563e-05,
      "loss": 0.982,
      "step": 19100
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.1467181467181468e-05,
      "loss": 0.9224,
      "step": 19200
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.137065637065637e-05,
      "loss": 0.9527,
      "step": 19300
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.1274131274131274e-05,
      "loss": 0.9188,
      "step": 19400
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.1177606177606178e-05,
      "loss": 0.9603,
      "step": 19500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.1081081081081081e-05,
      "loss": 0.9158,
      "step": 19600
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.0984555984555985e-05,
      "loss": 0.9615,
      "step": 19700
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.0888030888030889e-05,
      "loss": 0.9307,
      "step": 19800
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.0791505791505792e-05,
      "loss": 0.8881,
      "step": 19900
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.0694980694980694e-05,
      "loss": 0.8812,
      "step": 20000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.05984555984556e-05,
      "loss": 0.9301,
      "step": 20100
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.0501930501930502e-05,
      "loss": 0.9501,
      "step": 20200
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.0405405405405406e-05,
      "loss": 0.9434,
      "step": 20300
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.030888030888031e-05,
      "loss": 0.9466,
      "step": 20400
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.0212355212355213e-05,
      "loss": 0.8856,
      "step": 20500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.0115830115830115e-05,
      "loss": 0.9773,
      "step": 20600
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.0019305019305019e-05,
      "loss": 0.8732,
      "step": 20700
    },
    {
      "epoch": 2.01,
      "learning_rate": 9.922779922779924e-06,
      "loss": 0.7991,
      "step": 20800
    },
    {
      "epoch": 2.02,
      "learning_rate": 9.826254826254826e-06,
      "loss": 0.7911,
      "step": 20900
    },
    {
      "epoch": 2.03,
      "learning_rate": 9.72972972972973e-06,
      "loss": 0.7547,
      "step": 21000
    },
    {
      "epoch": 2.04,
      "learning_rate": 9.633204633204634e-06,
      "loss": 0.7243,
      "step": 21100
    },
    {
      "epoch": 2.05,
      "learning_rate": 9.536679536679537e-06,
      "loss": 0.7338,
      "step": 21200
    },
    {
      "epoch": 2.06,
      "learning_rate": 9.44015444015444e-06,
      "loss": 0.7552,
      "step": 21300
    },
    {
      "epoch": 2.07,
      "learning_rate": 9.343629343629345e-06,
      "loss": 0.7639,
      "step": 21400
    },
    {
      "epoch": 2.08,
      "learning_rate": 9.247104247104247e-06,
      "loss": 0.7705,
      "step": 21500
    },
    {
      "epoch": 2.08,
      "learning_rate": 9.15057915057915e-06,
      "loss": 0.7605,
      "step": 21600
    },
    {
      "epoch": 2.09,
      "learning_rate": 9.054054054054056e-06,
      "loss": 0.7734,
      "step": 21700
    },
    {
      "epoch": 2.1,
      "learning_rate": 8.957528957528958e-06,
      "loss": 0.747,
      "step": 21800
    },
    {
      "epoch": 2.11,
      "learning_rate": 8.861003861003861e-06,
      "loss": 0.7564,
      "step": 21900
    },
    {
      "epoch": 2.12,
      "learning_rate": 8.764478764478765e-06,
      "loss": 0.7362,
      "step": 22000
    },
    {
      "epoch": 2.13,
      "learning_rate": 8.667953667953669e-06,
      "loss": 0.7052,
      "step": 22100
    },
    {
      "epoch": 2.14,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.7525,
      "step": 22200
    },
    {
      "epoch": 2.15,
      "learning_rate": 8.474903474903475e-06,
      "loss": 0.754,
      "step": 22300
    },
    {
      "epoch": 2.16,
      "learning_rate": 8.378378378378378e-06,
      "loss": 0.7815,
      "step": 22400
    },
    {
      "epoch": 2.17,
      "learning_rate": 8.281853281853282e-06,
      "loss": 0.7356,
      "step": 22500
    },
    {
      "epoch": 2.18,
      "learning_rate": 8.185328185328184e-06,
      "loss": 0.7302,
      "step": 22600
    },
    {
      "epoch": 2.19,
      "learning_rate": 8.08880308880309e-06,
      "loss": 0.7673,
      "step": 22700
    },
    {
      "epoch": 2.2,
      "learning_rate": 7.992277992277993e-06,
      "loss": 0.7557,
      "step": 22800
    },
    {
      "epoch": 2.21,
      "learning_rate": 7.895752895752895e-06,
      "loss": 0.7523,
      "step": 22900
    },
    {
      "epoch": 2.22,
      "learning_rate": 7.7992277992278e-06,
      "loss": 0.7761,
      "step": 23000
    },
    {
      "epoch": 2.23,
      "learning_rate": 7.702702702702703e-06,
      "loss": 0.7202,
      "step": 23100
    },
    {
      "epoch": 2.24,
      "learning_rate": 7.606177606177607e-06,
      "loss": 0.7508,
      "step": 23200
    },
    {
      "epoch": 2.25,
      "learning_rate": 7.509652509652509e-06,
      "loss": 0.7161,
      "step": 23300
    },
    {
      "epoch": 2.26,
      "learning_rate": 7.413127413127414e-06,
      "loss": 0.8187,
      "step": 23400
    },
    {
      "epoch": 2.27,
      "learning_rate": 7.3166023166023166e-06,
      "loss": 0.7501,
      "step": 23500
    },
    {
      "epoch": 2.28,
      "learning_rate": 7.22007722007722e-06,
      "loss": 0.7494,
      "step": 23600
    },
    {
      "epoch": 2.29,
      "learning_rate": 7.123552123552124e-06,
      "loss": 0.7742,
      "step": 23700
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.027027027027027e-06,
      "loss": 0.7279,
      "step": 23800
    },
    {
      "epoch": 2.31,
      "learning_rate": 6.9305019305019305e-06,
      "loss": 0.7401,
      "step": 23900
    },
    {
      "epoch": 2.32,
      "learning_rate": 6.833976833976834e-06,
      "loss": 0.7151,
      "step": 24000
    },
    {
      "epoch": 2.33,
      "learning_rate": 6.737451737451738e-06,
      "loss": 0.7656,
      "step": 24100
    },
    {
      "epoch": 2.34,
      "learning_rate": 6.640926640926642e-06,
      "loss": 0.7402,
      "step": 24200
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.5444015444015445e-06,
      "loss": 0.7264,
      "step": 24300
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.447876447876448e-06,
      "loss": 0.7397,
      "step": 24400
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.351351351351352e-06,
      "loss": 0.7495,
      "step": 24500
    },
    {
      "epoch": 2.37,
      "learning_rate": 6.254826254826255e-06,
      "loss": 0.7814,
      "step": 24600
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.1583011583011585e-06,
      "loss": 0.7627,
      "step": 24700
    },
    {
      "epoch": 2.39,
      "learning_rate": 6.061776061776061e-06,
      "loss": 0.6907,
      "step": 24800
    },
    {
      "epoch": 2.4,
      "learning_rate": 5.965250965250965e-06,
      "loss": 0.7289,
      "step": 24900
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.868725868725869e-06,
      "loss": 0.7333,
      "step": 25000
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.7722007722007725e-06,
      "loss": 0.7544,
      "step": 25100
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.675675675675676e-06,
      "loss": 0.753,
      "step": 25200
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.57915057915058e-06,
      "loss": 0.7658,
      "step": 25300
    },
    {
      "epoch": 2.45,
      "learning_rate": 5.482625482625483e-06,
      "loss": 0.7222,
      "step": 25400
    },
    {
      "epoch": 2.46,
      "learning_rate": 5.3861003861003864e-06,
      "loss": 0.726,
      "step": 25500
    },
    {
      "epoch": 2.47,
      "learning_rate": 5.289575289575289e-06,
      "loss": 0.7128,
      "step": 25600
    },
    {
      "epoch": 2.48,
      "learning_rate": 5.193050193050193e-06,
      "loss": 0.7381,
      "step": 25700
    },
    {
      "epoch": 2.49,
      "learning_rate": 5.096525096525097e-06,
      "loss": 0.7035,
      "step": 25800
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.9999999999999996e-06,
      "loss": 0.681,
      "step": 25900
    },
    {
      "epoch": 2.51,
      "learning_rate": 4.903474903474903e-06,
      "loss": 0.7638,
      "step": 26000
    },
    {
      "epoch": 2.52,
      "learning_rate": 4.806949806949808e-06,
      "loss": 0.7013,
      "step": 26100
    },
    {
      "epoch": 2.53,
      "learning_rate": 4.710424710424711e-06,
      "loss": 0.6841,
      "step": 26200
    },
    {
      "epoch": 2.54,
      "learning_rate": 4.613899613899614e-06,
      "loss": 0.753,
      "step": 26300
    },
    {
      "epoch": 2.55,
      "learning_rate": 4.517374517374517e-06,
      "loss": 0.7082,
      "step": 26400
    },
    {
      "epoch": 2.56,
      "learning_rate": 4.420849420849421e-06,
      "loss": 0.7213,
      "step": 26500
    },
    {
      "epoch": 2.57,
      "learning_rate": 4.324324324324325e-06,
      "loss": 0.7049,
      "step": 26600
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.2277992277992275e-06,
      "loss": 0.6806,
      "step": 26700
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.131274131274131e-06,
      "loss": 0.716,
      "step": 26800
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.034749034749035e-06,
      "loss": 0.7481,
      "step": 26900
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.938223938223938e-06,
      "loss": 0.7112,
      "step": 27000
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.841698841698842e-06,
      "loss": 0.7514,
      "step": 27100
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.745173745173745e-06,
      "loss": 0.6975,
      "step": 27200
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.648648648648649e-06,
      "loss": 0.7166,
      "step": 27300
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.552123552123552e-06,
      "loss": 0.7702,
      "step": 27400
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.455598455598456e-06,
      "loss": 0.713,
      "step": 27500
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.359073359073359e-06,
      "loss": 0.7266,
      "step": 27600
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.2625482625482625e-06,
      "loss": 0.7089,
      "step": 27700
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.166023166023166e-06,
      "loss": 0.6851,
      "step": 27800
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.06949806949807e-06,
      "loss": 0.7234,
      "step": 27900
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.972972972972973e-06,
      "loss": 0.6683,
      "step": 28000
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.8764478764478764e-06,
      "loss": 0.7244,
      "step": 28100
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.7799227799227797e-06,
      "loss": 0.7578,
      "step": 28200
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.683397683397684e-06,
      "loss": 0.7604,
      "step": 28300
    },
    {
      "epoch": 2.74,
      "learning_rate": 2.586872586872587e-06,
      "loss": 0.7131,
      "step": 28400
    },
    {
      "epoch": 2.75,
      "learning_rate": 2.4903474903474904e-06,
      "loss": 0.7049,
      "step": 28500
    },
    {
      "epoch": 2.76,
      "learning_rate": 2.3938223938223937e-06,
      "loss": 0.7552,
      "step": 28600
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.2972972972972974e-06,
      "loss": 0.6907,
      "step": 28700
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.200772200772201e-06,
      "loss": 0.7238,
      "step": 28800
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.1042471042471044e-06,
      "loss": 0.7431,
      "step": 28900
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.0077220077220077e-06,
      "loss": 0.7096,
      "step": 29000
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.9111969111969114e-06,
      "loss": 0.6985,
      "step": 29100
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.8146718146718149e-06,
      "loss": 0.7541,
      "step": 29200
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.7181467181467181e-06,
      "loss": 0.762,
      "step": 29300
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.6216216216216219e-06,
      "loss": 0.6741,
      "step": 29400
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.5250965250965251e-06,
      "loss": 0.6974,
      "step": 29500
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.7249,
      "step": 29600
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.3320463320463321e-06,
      "loss": 0.6886,
      "step": 29700
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.2355212355212356e-06,
      "loss": 0.6499,
      "step": 29800
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.1389961389961391e-06,
      "loss": 0.7713,
      "step": 29900
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.0424710424710424e-06,
      "loss": 0.711,
      "step": 30000
    }
  ],
  "max_steps": 31080,
  "num_train_epochs": 3,
  "total_flos": 6.270808605555917e+16,
  "trial_name": null,
  "trial_params": null
}