{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.0,
  "global_step": 257705,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.99029898527386e-05,
      "loss": 1.2612,
      "step": 500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9805979705477194e-05,
      "loss": 1.139,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.970896955821579e-05,
      "loss": 1.1087,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.961195941095439e-05,
      "loss": 1.1514,
      "step": 2000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.951494926369298e-05,
      "loss": 1.1921,
      "step": 2500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9417939116431584e-05,
      "loss": 1.0896,
      "step": 3000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.932092896917018e-05,
      "loss": 1.1259,
      "step": 3500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9223918821908775e-05,
      "loss": 1.1104,
      "step": 4000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.912690867464737e-05,
      "loss": 1.0653,
      "step": 4500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9029898527385967e-05,
      "loss": 1.1177,
      "step": 5000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.893288838012456e-05,
      "loss": 1.1028,
      "step": 5500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.883587823286316e-05,
      "loss": 1.1255,
      "step": 6000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.873886808560176e-05,
      "loss": 1.0962,
      "step": 6500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.864185793834035e-05,
      "loss": 1.1071,
      "step": 7000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.854484779107895e-05,
      "loss": 1.0509,
      "step": 7500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.844783764381755e-05,
      "loss": 1.0337,
      "step": 8000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.8350827496556143e-05,
      "loss": 1.0519,
      "step": 8500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.825381734929474e-05,
      "loss": 1.0439,
      "step": 9000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.8156807202033335e-05,
      "loss": 1.0587,
      "step": 9500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.805979705477193e-05,
      "loss": 1.0551,
      "step": 10000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.7962786907510526e-05,
      "loss": 1.0638,
      "step": 10500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.786577676024913e-05,
      "loss": 1.0538,
      "step": 11000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.776876661298772e-05,
      "loss": 1.0391,
      "step": 11500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.767175646572632e-05,
      "loss": 1.0361,
      "step": 12000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.7574746318464916e-05,
      "loss": 1.0181,
      "step": 12500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.747773617120351e-05,
      "loss": 1.0409,
      "step": 13000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.738072602394211e-05,
      "loss": 1.0662,
      "step": 13500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.72837158766807e-05,
      "loss": 1.0226,
      "step": 14000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.71867057294193e-05,
      "loss": 1.0217,
      "step": 14500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.7089695582157895e-05,
      "loss": 1.0462,
      "step": 15000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.69926854348965e-05,
      "loss": 1.0233,
      "step": 15500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.6895675287635086e-05,
      "loss": 1.0627,
      "step": 16000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.679866514037369e-05,
      "loss": 0.9915,
      "step": 16500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.6701654993112285e-05,
      "loss": 1.0299,
      "step": 17000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.6604644845850873e-05,
      "loss": 1.0203,
      "step": 17500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.6507634698589476e-05,
      "loss": 1.0109,
      "step": 18000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.641062455132807e-05,
      "loss": 1.0375,
      "step": 18500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.631361440406667e-05,
      "loss": 1.0194,
      "step": 19000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.621660425680526e-05,
      "loss": 1.0495,
      "step": 19500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.611959410954386e-05,
      "loss": 1.0627,
      "step": 20000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.6022583962282455e-05,
      "loss": 1.019,
      "step": 20500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.592557381502106e-05,
      "loss": 1.0007,
      "step": 21000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.582856366775965e-05,
      "loss": 1.0078,
      "step": 21500
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.573155352049824e-05,
      "loss": 1.0453,
      "step": 22000
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.5634543373236844e-05,
      "loss": 1.0154,
      "step": 22500
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.553753322597544e-05,
      "loss": 0.9463,
      "step": 23000
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.5440523078714036e-05,
      "loss": 0.9709,
      "step": 23500
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.534351293145263e-05,
      "loss": 0.9679,
      "step": 24000
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.524650278419123e-05,
      "loss": 0.9654,
      "step": 24500
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.514949263692982e-05,
      "loss": 0.9937,
      "step": 25000
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.5052482489668426e-05,
      "loss": 1.0232,
      "step": 25500
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.4955472342407015e-05,
      "loss": 1.0295,
      "step": 26000
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.485846219514561e-05,
      "loss": 1.0066,
      "step": 26500
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.476145204788421e-05,
      "loss": 1.0571,
      "step": 27000
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.466444190062281e-05,
      "loss": 0.9869,
      "step": 27500
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.4567431753361404e-05,
      "loss": 0.9895,
      "step": 28000
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.44704216061e-05,
      "loss": 1.0054,
      "step": 28500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.4373411458838596e-05,
      "loss": 1.0221,
      "step": 29000
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.427640131157719e-05,
      "loss": 0.989,
      "step": 29500
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.4179391164315794e-05,
      "loss": 1.0258,
      "step": 30000
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.408238101705438e-05,
      "loss": 0.982,
      "step": 30500
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.398537086979298e-05,
      "loss": 0.9852,
      "step": 31000
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.388836072253158e-05,
      "loss": 1.0117,
      "step": 31500
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.379135057527018e-05,
      "loss": 0.9938,
      "step": 32000
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.369434042800877e-05,
      "loss": 0.9978,
      "step": 32500
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.359733028074737e-05,
      "loss": 1.0073,
      "step": 33000
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.3500320133485964e-05,
      "loss": 1.0023,
      "step": 33500
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.340330998622456e-05,
      "loss": 1.0114,
      "step": 34000
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.330629983896316e-05,
      "loss": 0.9857,
      "step": 34500
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.320928969170175e-05,
      "loss": 1.0113,
      "step": 35000
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.311227954444035e-05,
      "loss": 0.9811,
      "step": 35500
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.301526939717895e-05,
      "loss": 0.9571,
      "step": 36000
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.291825924991754e-05,
      "loss": 0.9752,
      "step": 36500
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.282124910265614e-05,
      "loss": 1.0078,
      "step": 37000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.272423895539474e-05,
      "loss": 0.9821,
      "step": 37500
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.262722880813333e-05,
      "loss": 0.9266,
      "step": 38000
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.253021866087193e-05,
      "loss": 1.0045,
      "step": 38500
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.243320851361053e-05,
      "loss": 0.9769,
      "step": 39000
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.233619836634912e-05,
      "loss": 0.9674,
      "step": 39500
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.2239188219087715e-05,
      "loss": 0.9553,
      "step": 40000
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.214217807182632e-05,
      "loss": 0.9824,
      "step": 40500
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.204516792456491e-05,
      "loss": 0.9804,
      "step": 41000
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.194815777730351e-05,
      "loss": 1.0219,
      "step": 41500
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.1851147630042105e-05,
      "loss": 0.9894,
      "step": 42000
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.17541374827807e-05,
      "loss": 0.9952,
      "step": 42500
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.16571273355193e-05,
      "loss": 0.9626,
      "step": 43000
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.156011718825789e-05,
      "loss": 0.9586,
      "step": 43500
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.146310704099649e-05,
      "loss": 0.969,
      "step": 44000
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.1366096893735084e-05,
      "loss": 0.9156,
      "step": 44500
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.1269086746473686e-05,
      "loss": 0.9673,
      "step": 45000
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.1172076599212275e-05,
      "loss": 0.932,
      "step": 45500
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.107506645195088e-05,
      "loss": 0.9648,
      "step": 46000
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.0978056304689474e-05,
      "loss": 0.955,
      "step": 46500
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.088104615742807e-05,
      "loss": 0.9495,
      "step": 47000
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.0784036010166665e-05,
      "loss": 0.9578,
      "step": 47500
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.068702586290526e-05,
      "loss": 0.9528,
      "step": 48000
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.0590015715643857e-05,
      "loss": 0.967,
      "step": 48500
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.049300556838245e-05,
      "loss": 0.9599,
      "step": 49000
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.0395995421121055e-05,
      "loss": 0.9778,
      "step": 49500
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.0298985273859644e-05,
      "loss": 0.9708,
      "step": 50000
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.0201975126598246e-05,
      "loss": 0.9941,
      "step": 50500
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.010496497933684e-05,
      "loss": 0.9813,
      "step": 51000
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.000795483207544e-05,
      "loss": 0.9228,
      "step": 51500
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.9910944684814033e-05,
      "loss": 0.9789,
      "step": 52000
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.981393453755263e-05,
      "loss": 0.9485,
      "step": 52500
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.9716924390291225e-05,
      "loss": 0.9644,
      "step": 53000
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.961991424302982e-05,
      "loss": 0.9815,
      "step": 53500
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.952290409576842e-05,
      "loss": 0.9632,
      "step": 54000
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.942589394850701e-05,
      "loss": 0.9507,
      "step": 54500
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.9328883801245615e-05,
      "loss": 0.9417,
      "step": 55000
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.923187365398421e-05,
      "loss": 0.9866,
      "step": 55500
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.9134863506722806e-05,
      "loss": 0.9535,
      "step": 56000
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.90378533594614e-05,
      "loss": 0.924,
      "step": 56500
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.89408432122e-05,
      "loss": 0.9352,
      "step": 57000
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.884383306493859e-05,
      "loss": 0.9545,
      "step": 57500
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.874682291767719e-05,
      "loss": 0.9883,
      "step": 58000
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.864981277041579e-05,
      "loss": 0.942,
      "step": 58500
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.855280262315438e-05,
      "loss": 0.9676,
      "step": 59000
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.845579247589298e-05,
      "loss": 0.9722,
      "step": 59500
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.835878232863158e-05,
      "loss": 0.982,
      "step": 60000
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.8261772181370175e-05,
      "loss": 0.9846,
      "step": 60500
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.816476203410877e-05,
      "loss": 0.9393,
      "step": 61000
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.8067751886847366e-05,
      "loss": 0.9487,
      "step": 61500
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.797074173958596e-05,
      "loss": 0.9654,
      "step": 62000
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.787373159232456e-05,
      "loss": 0.9586,
      "step": 62500
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.777672144506316e-05,
      "loss": 0.9276,
      "step": 63000
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.767971129780175e-05,
      "loss": 1.0214,
      "step": 63500
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.758270115054035e-05,
      "loss": 0.9482,
      "step": 64000
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.748569100327895e-05,
      "loss": 0.9753,
      "step": 64500
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.738868085601754e-05,
      "loss": 0.9303,
      "step": 65000
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.729167070875614e-05,
      "loss": 0.9104,
      "step": 65500
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.7194660561494734e-05,
      "loss": 0.9294,
      "step": 66000
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.709765041423333e-05,
      "loss": 0.9913,
      "step": 66500
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.7000640266971926e-05,
      "loss": 0.9638,
      "step": 67000
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.690363011971052e-05,
      "loss": 0.9196,
      "step": 67500
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.680661997244912e-05,
      "loss": 0.9547,
      "step": 68000
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.670960982518772e-05,
      "loss": 0.9714,
      "step": 68500
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.6612599677926316e-05,
      "loss": 0.9233,
      "step": 69000
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.651558953066491e-05,
      "loss": 0.9667,
      "step": 69500
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.641857938340351e-05,
      "loss": 0.9559,
      "step": 70000
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.63215692361421e-05,
      "loss": 0.92,
      "step": 70500
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.62245590888807e-05,
      "loss": 0.9526,
      "step": 71000
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.6127548941619294e-05,
      "loss": 0.962,
      "step": 71500
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.603053879435789e-05,
      "loss": 0.9544,
      "step": 72000
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.5933528647096486e-05,
      "loss": 0.9465,
      "step": 72500
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.583651849983509e-05,
      "loss": 0.9693,
      "step": 73000
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.5739508352573684e-05,
      "loss": 0.9634,
      "step": 73500
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.564249820531227e-05,
      "loss": 0.957,
      "step": 74000
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.5545488058050875e-05,
      "loss": 0.9429,
      "step": 74500
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.544847791078947e-05,
      "loss": 0.9594,
      "step": 75000
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.535146776352807e-05,
      "loss": 0.9277,
      "step": 75500
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.525445761626666e-05,
      "loss": 0.9265,
      "step": 76000
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.515744746900526e-05,
      "loss": 0.9294,
      "step": 76500
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.5060437321743854e-05,
      "loss": 0.9848,
      "step": 77000
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.496342717448246e-05,
      "loss": 0.9388,
      "step": 77500
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.4866417027221046e-05,
      "loss": 0.9454,
      "step": 78000
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.476940687995964e-05,
      "loss": 0.8826,
      "step": 78500
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.4672396732698244e-05,
      "loss": 0.9019,
      "step": 79000
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.457538658543684e-05,
      "loss": 0.928,
      "step": 79500
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.4478376438175435e-05,
      "loss": 0.9387,
      "step": 80000
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.438136629091403e-05,
      "loss": 0.9649,
      "step": 80500
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.428435614365263e-05,
      "loss": 0.903,
      "step": 81000
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.418734599639122e-05,
      "loss": 0.9183,
      "step": 81500
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.4090335849129825e-05,
      "loss": 0.9201,
      "step": 82000
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.3993325701868414e-05,
      "loss": 0.9025,
      "step": 82500
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.389631555460701e-05,
      "loss": 0.9034,
      "step": 83000
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.379930540734561e-05,
      "loss": 0.94,
      "step": 83500
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.37022952600842e-05,
      "loss": 0.9241,
      "step": 84000
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.3605285112822804e-05,
      "loss": 0.9674,
      "step": 84500
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.35082749655614e-05,
      "loss": 0.9141,
      "step": 85000
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.3411264818299995e-05,
      "loss": 0.9069,
      "step": 85500
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.331425467103859e-05,
      "loss": 0.9461,
      "step": 86000
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.3217244523777193e-05,
      "loss": 0.9103,
      "step": 86500
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.312023437651578e-05,
      "loss": 0.9241,
      "step": 87000
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.302322422925438e-05,
      "loss": 0.951,
      "step": 87500
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.292621408199298e-05,
      "loss": 0.9509,
      "step": 88000
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.282920393473157e-05,
      "loss": 0.9311,
      "step": 88500
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.273219378747017e-05,
      "loss": 0.9085,
      "step": 89000
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.263518364020877e-05,
      "loss": 0.9402,
      "step": 89500
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.2538173492947364e-05,
      "loss": 0.9181,
      "step": 90000
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.244116334568596e-05,
      "loss": 0.8801,
      "step": 90500
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.234415319842456e-05,
      "loss": 0.9246,
      "step": 91000
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.224714305116315e-05,
      "loss": 0.949,
      "step": 91500
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.2150132903901747e-05,
      "loss": 0.9081,
      "step": 92000
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.205312275664035e-05,
      "loss": 0.9665,
      "step": 92500
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.195611260937894e-05,
      "loss": 0.9274,
      "step": 93000
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.185910246211754e-05,
      "loss": 0.9407,
      "step": 93500
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.1762092314856136e-05,
      "loss": 0.9362,
      "step": 94000
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.166508216759473e-05,
      "loss": 0.9304,
      "step": 94500
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.156807202033333e-05,
      "loss": 0.8939,
      "step": 95000
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.147106187307193e-05,
      "loss": 0.9145,
      "step": 95500
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.137405172581052e-05,
      "loss": 0.941,
      "step": 96000
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.1277041578549115e-05,
      "loss": 0.9381,
      "step": 96500
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.118003143128772e-05,
      "loss": 0.8844,
      "step": 97000
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.1083021284026306e-05,
      "loss": 0.8916,
      "step": 97500
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.098601113676491e-05,
      "loss": 0.9436,
      "step": 98000
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.0889000989503505e-05,
      "loss": 0.9383,
      "step": 98500
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.07919908422421e-05,
      "loss": 0.9086,
      "step": 99000
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.0694980694980696e-05,
      "loss": 0.8912,
      "step": 99500
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.059797054771929e-05,
      "loss": 0.9412,
      "step": 100000
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.050096040045789e-05,
      "loss": 0.9268,
      "step": 100500
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.0403950253196483e-05,
      "loss": 0.9141,
      "step": 101000
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.0306940105935082e-05,
      "loss": 0.9069,
      "step": 101500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.0209929958673678e-05,
      "loss": 0.94,
      "step": 102000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.0112919811412277e-05,
      "loss": 0.9342,
      "step": 102500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.001590966415087e-05,
      "loss": 0.9068,
      "step": 103000
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.991889951688947e-05,
      "loss": 0.9693,
      "step": 103500
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9821889369628065e-05,
      "loss": 0.9291,
      "step": 104000
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.9724879222366657e-05,
      "loss": 0.9161,
      "step": 104500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.962786907510526e-05,
      "loss": 0.8943,
      "step": 105000
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.9530858927843852e-05,
      "loss": 0.9375,
      "step": 105500
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.943384878058245e-05,
      "loss": 0.9247,
      "step": 106000
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.9336838633321047e-05,
      "loss": 0.955,
      "step": 106500
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.9239828486059646e-05,
      "loss": 0.9331,
      "step": 107000
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.9142818338798238e-05,
      "loss": 0.9402,
      "step": 107500
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.9045808191536837e-05,
      "loss": 0.9499,
      "step": 108000
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.8948798044275433e-05,
      "loss": 0.956,
      "step": 108500
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.8851787897014025e-05,
      "loss": 0.9277,
      "step": 109000
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.8754777749752624e-05,
      "loss": 0.9601,
      "step": 109500
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.865776760249122e-05,
      "loss": 0.9473,
      "step": 110000
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.856075745522982e-05,
      "loss": 0.9423,
      "step": 110500
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.8463747307968415e-05,
      "loss": 0.9162,
      "step": 111000
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.8366737160707014e-05,
      "loss": 0.9544,
      "step": 111500
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.8269727013445606e-05,
      "loss": 0.8595,
      "step": 112000
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.8172716866184206e-05,
      "loss": 0.9092,
      "step": 112500
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.80757067189228e-05,
      "loss": 0.9022,
      "step": 113000
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.7978696571661394e-05,
      "loss": 0.8945,
      "step": 113500
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.7881686424399993e-05,
      "loss": 0.8888,
      "step": 114000
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.778467627713859e-05,
      "loss": 0.9063,
      "step": 114500
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.7687666129877188e-05,
      "loss": 0.8755,
      "step": 115000
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.7590655982615783e-05,
      "loss": 0.9199,
      "step": 115500
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.7493645835354382e-05,
      "loss": 0.9013,
      "step": 116000
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.7396635688092975e-05,
      "loss": 0.9043,
      "step": 116500
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.7299625540831574e-05,
      "loss": 0.8963,
      "step": 117000
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.720261539357017e-05,
      "loss": 0.9173,
      "step": 117500
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.7105605246308762e-05,
      "loss": 0.9345,
      "step": 118000
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.700859509904736e-05,
      "loss": 0.9415,
      "step": 118500
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.6911584951785957e-05,
      "loss": 0.9149,
      "step": 119000
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.6814574804524556e-05,
      "loss": 0.958,
      "step": 119500
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.671756465726315e-05,
      "loss": 0.9151,
      "step": 120000
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.662055451000175e-05,
      "loss": 0.918,
      "step": 120500
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.6523544362740343e-05,
      "loss": 0.8934,
      "step": 121000
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.6426534215478942e-05,
      "loss": 0.9143,
      "step": 121500
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.6329524068217538e-05,
      "loss": 0.9048,
      "step": 122000
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.623251392095613e-05,
      "loss": 0.9183,
      "step": 122500
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.613550377369473e-05,
      "loss": 0.9278,
      "step": 123000
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.6038493626433325e-05,
      "loss": 0.9219,
      "step": 123500
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.5941483479171924e-05,
      "loss": 0.899,
      "step": 124000
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.5844473331910517e-05,
      "loss": 0.8905,
      "step": 124500
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.574746318464912e-05,
      "loss": 0.9157,
      "step": 125000
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.565045303738771e-05,
      "loss": 0.9317,
      "step": 125500
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.555344289012631e-05,
      "loss": 0.931,
      "step": 126000
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.5456432742864906e-05,
      "loss": 0.9274,
      "step": 126500
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.53594225956035e-05,
      "loss": 0.84,
      "step": 127000
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.5262412448342098e-05,
      "loss": 0.8753,
      "step": 127500
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.5165402301080694e-05,
      "loss": 0.8975,
      "step": 128000
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.5068392153819293e-05,
      "loss": 0.9369,
      "step": 128500
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.4971382006557885e-05,
      "loss": 0.9171,
      "step": 129000
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.4874371859296484e-05,
      "loss": 0.8922,
      "step": 129500
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.477736171203508e-05,
      "loss": 0.8931,
      "step": 130000
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.4680351564773676e-05,
      "loss": 0.8667,
      "step": 130500
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.4583341417512275e-05,
      "loss": 0.9311,
      "step": 131000
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.448633127025087e-05,
      "loss": 0.9077,
      "step": 131500
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.4389321122989466e-05,
      "loss": 0.8988,
      "step": 132000
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.4292310975728062e-05,
      "loss": 0.8927,
      "step": 132500
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.4195300828466658e-05,
      "loss": 0.8995,
      "step": 133000
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.4098290681205254e-05,
      "loss": 0.9698,
      "step": 133500
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.4001280533943853e-05,
      "loss": 0.8552,
      "step": 134000
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.390427038668245e-05,
      "loss": 0.9115,
      "step": 134500
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.3807260239421044e-05,
      "loss": 0.8817,
      "step": 135000
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.371025009215964e-05,
      "loss": 0.9012,
      "step": 135500
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.361323994489824e-05,
      "loss": 0.8953,
      "step": 136000
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.3516229797636835e-05,
      "loss": 0.8849,
      "step": 136500
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.341921965037543e-05,
      "loss": 0.9138,
      "step": 137000
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.3322209503114026e-05,
      "loss": 0.9085,
      "step": 137500
    },
    {
      "epoch": 3.75,
      "learning_rate": 2.3225199355852622e-05,
      "loss": 0.8695,
      "step": 138000
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.3128189208591218e-05,
      "loss": 0.9493,
      "step": 138500
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.3031179061329817e-05,
      "loss": 0.8985,
      "step": 139000
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.2934168914068413e-05,
      "loss": 0.8838,
      "step": 139500
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.2837158766807008e-05,
      "loss": 0.8823,
      "step": 140000
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.2740148619545607e-05,
      "loss": 0.933,
      "step": 140500
    },
    {
      "epoch": 3.83,
      "learning_rate": 2.2643138472284203e-05,
      "loss": 0.9226,
      "step": 141000
    },
    {
      "epoch": 3.84,
      "learning_rate": 2.25461283250228e-05,
      "loss": 0.9095,
      "step": 141500
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.2449118177761395e-05,
      "loss": 0.9371,
      "step": 142000
    },
    {
      "epoch": 3.87,
      "learning_rate": 2.235210803049999e-05,
      "loss": 0.9062,
      "step": 142500
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.2255097883238586e-05,
      "loss": 0.9118,
      "step": 143000
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.2158087735977185e-05,
      "loss": 0.888,
      "step": 143500
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.206107758871578e-05,
      "loss": 0.926,
      "step": 144000
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.1964067441454377e-05,
      "loss": 0.9618,
      "step": 144500
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.1867057294192976e-05,
      "loss": 0.9088,
      "step": 145000
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.177004714693157e-05,
      "loss": 0.9351,
      "step": 145500
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.1673036999670164e-05,
      "loss": 0.8912,
      "step": 146000
    },
    {
      "epoch": 3.98,
      "learning_rate": 2.1576026852408763e-05,
      "loss": 0.9464,
      "step": 146500
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.147901670514736e-05,
      "loss": 0.9041,
      "step": 147000
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.1382006557885955e-05,
      "loss": 0.9354,
      "step": 147500
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.1284996410624554e-05,
      "loss": 0.8652,
      "step": 148000
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.118798626336315e-05,
      "loss": 0.8903,
      "step": 148500
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.1090976116101745e-05,
      "loss": 0.9394,
      "step": 149000
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.0993965968840344e-05,
      "loss": 0.9253,
      "step": 149500
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.0896955821578937e-05,
      "loss": 0.9139,
      "step": 150000
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.0799945674317532e-05,
      "loss": 0.9204,
      "step": 150500
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.070293552705613e-05,
      "loss": 0.883,
      "step": 151000
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.0605925379794727e-05,
      "loss": 0.9354,
      "step": 151500
    },
    {
      "epoch": 4.13,
      "learning_rate": 2.0508915232533323e-05,
      "loss": 0.8801,
      "step": 152000
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.0411905085271922e-05,
      "loss": 0.8879,
      "step": 152500
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.0314894938010518e-05,
      "loss": 0.8884,
      "step": 153000
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.0217884790749113e-05,
      "loss": 0.9264,
      "step": 153500
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.0120874643487713e-05,
      "loss": 0.9586,
      "step": 154000
    },
    {
      "epoch": 4.2,
      "learning_rate": 2.0023864496226305e-05,
      "loss": 0.8777,
      "step": 154500
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.99268543489649e-05,
      "loss": 0.9138,
      "step": 155000
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.98298442017035e-05,
      "loss": 0.9085,
      "step": 155500
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.9732834054442096e-05,
      "loss": 0.8894,
      "step": 156000
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.963582390718069e-05,
      "loss": 0.8704,
      "step": 156500
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.953881375991929e-05,
      "loss": 0.8793,
      "step": 157000
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.9441803612657886e-05,
      "loss": 0.8966,
      "step": 157500
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.9344793465396482e-05,
      "loss": 0.9199,
      "step": 158000
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.9247783318135078e-05,
      "loss": 0.8879,
      "step": 158500
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.9150773170873673e-05,
      "loss": 0.8834,
      "step": 159000
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.905376302361227e-05,
      "loss": 0.9081,
      "step": 159500
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.8956752876350868e-05,
      "loss": 0.8915,
      "step": 160000
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.8859742729089464e-05,
      "loss": 0.9092,
      "step": 160500
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.876273258182806e-05,
      "loss": 0.8787,
      "step": 161000
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.8665722434566655e-05,
      "loss": 0.927,
      "step": 161500
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.8568712287305255e-05,
      "loss": 0.8968,
      "step": 162000
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.847170214004385e-05,
      "loss": 0.887,
      "step": 162500
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.8374691992782446e-05,
      "loss": 0.8938,
      "step": 163000
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.8277681845521042e-05,
      "loss": 0.8835,
      "step": 163500
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.8180671698259638e-05,
      "loss": 0.8994,
      "step": 164000
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.8083661550998233e-05,
      "loss": 0.8839,
      "step": 164500
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.7986651403736832e-05,
      "loss": 0.8901,
      "step": 165000
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.7889641256475428e-05,
      "loss": 0.8754,
      "step": 165500
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.7792631109214024e-05,
      "loss": 0.8991,
      "step": 166000
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.7695620961952623e-05,
      "loss": 0.902,
      "step": 166500
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.759861081469122e-05,
      "loss": 0.8808,
      "step": 167000
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.7501600667429814e-05,
      "loss": 0.9244,
      "step": 167500
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.740459052016841e-05,
      "loss": 0.8822,
      "step": 168000
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.7307580372907006e-05,
      "loss": 0.8777,
      "step": 168500
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.72105702256456e-05,
      "loss": 0.9138,
      "step": 169000
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.71135600783842e-05,
      "loss": 0.9311,
      "step": 169500
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.7016549931122796e-05,
      "loss": 0.843,
      "step": 170000
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.6919539783861392e-05,
      "loss": 0.8975,
      "step": 170500
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.682252963659999e-05,
      "loss": 0.9105,
      "step": 171000
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.6725519489338587e-05,
      "loss": 0.8941,
      "step": 171500
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.662850934207718e-05,
      "loss": 0.8893,
      "step": 172000
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.653149919481578e-05,
      "loss": 0.9068,
      "step": 172500
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.6434489047554374e-05,
      "loss": 0.8709,
      "step": 173000
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.633747890029297e-05,
      "loss": 0.8782,
      "step": 173500
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.624046875303157e-05,
      "loss": 0.9319,
      "step": 174000
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.6143458605770165e-05,
      "loss": 0.9162,
      "step": 174500
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.604644845850876e-05,
      "loss": 0.9282,
      "step": 175000
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.594943831124736e-05,
      "loss": 0.9002,
      "step": 175500
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.5852428163985952e-05,
      "loss": 0.8695,
      "step": 176000
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.5755418016724548e-05,
      "loss": 0.8604,
      "step": 176500
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.5658407869463147e-05,
      "loss": 0.8867,
      "step": 177000
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.5561397722201743e-05,
      "loss": 0.8724,
      "step": 177500
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.546438757494034e-05,
      "loss": 0.8902,
      "step": 178000
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5367377427678938e-05,
      "loss": 0.8798,
      "step": 178500
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.5270367280417533e-05,
      "loss": 0.8536,
      "step": 179000
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.5173357133156129e-05,
      "loss": 0.9221,
      "step": 179500
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.5076346985894726e-05,
      "loss": 0.8698,
      "step": 180000
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.497933683863332e-05,
      "loss": 0.8493,
      "step": 180500
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.4882326691371918e-05,
      "loss": 0.8843,
      "step": 181000
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.4785316544110514e-05,
      "loss": 0.8792,
      "step": 181500
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.4688306396849111e-05,
      "loss": 0.8803,
      "step": 182000
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.4591296249587707e-05,
      "loss": 0.8858,
      "step": 182500
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.4494286102326304e-05,
      "loss": 0.9053,
      "step": 183000
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4397275955064902e-05,
      "loss": 0.9032,
      "step": 183500
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.4300265807803497e-05,
      "loss": 0.8996,
      "step": 184000
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.4203255660542095e-05,
      "loss": 0.8503,
      "step": 184500
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.4106245513280689e-05,
      "loss": 0.9006,
      "step": 185000
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.4009235366019285e-05,
      "loss": 0.888,
      "step": 185500
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.3912225218757882e-05,
      "loss": 0.9044,
      "step": 186000
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.381521507149648e-05,
      "loss": 0.8797,
      "step": 186500
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.3718204924235075e-05,
      "loss": 0.8769,
      "step": 187000
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.3621194776973673e-05,
      "loss": 0.9074,
      "step": 187500
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.352418462971227e-05,
      "loss": 0.8748,
      "step": 188000
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.3427174482450866e-05,
      "loss": 0.8623,
      "step": 188500
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.3330164335189463e-05,
      "loss": 0.8983,
      "step": 189000
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.3233154187928057e-05,
      "loss": 0.8834,
      "step": 189500
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.3136144040666653e-05,
      "loss": 0.9079,
      "step": 190000
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.303913389340525e-05,
      "loss": 0.8804,
      "step": 190500
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.2942123746143848e-05,
      "loss": 0.8942,
      "step": 191000
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.2845113598882444e-05,
      "loss": 0.8813,
      "step": 191500
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.2748103451621041e-05,
      "loss": 0.8861,
      "step": 192000
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.2651093304359637e-05,
      "loss": 0.8632,
      "step": 192500
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.2554083157098234e-05,
      "loss": 0.8615,
      "step": 193000
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.245707300983683e-05,
      "loss": 0.9083,
      "step": 193500
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.2360062862575426e-05,
      "loss": 0.8949,
      "step": 194000
    },
    {
      "epoch": 5.28,
      "learning_rate": 1.2263052715314021e-05,
      "loss": 0.8694,
      "step": 194500
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.2166042568052619e-05,
      "loss": 0.867,
      "step": 195000
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.2069032420791215e-05,
      "loss": 0.8697,
      "step": 195500
    },
    {
      "epoch": 5.32,
      "learning_rate": 1.1972022273529812e-05,
      "loss": 0.8807,
      "step": 196000
    },
    {
      "epoch": 5.34,
      "learning_rate": 1.187501212626841e-05,
      "loss": 0.8939,
      "step": 196500
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.1778001979007003e-05,
      "loss": 0.894,
      "step": 197000
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.1680991831745601e-05,
      "loss": 0.8861,
      "step": 197500
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.1583981684484198e-05,
      "loss": 0.8518,
      "step": 198000
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.1486971537222794e-05,
      "loss": 0.923,
      "step": 198500
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.138996138996139e-05,
      "loss": 0.8814,
      "step": 199000
    },
    {
      "epoch": 5.42,
      "learning_rate": 1.1292951242699987e-05,
      "loss": 0.9137,
      "step": 199500
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.1195941095438583e-05,
      "loss": 0.8618,
      "step": 200000
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.109893094817718e-05,
      "loss": 0.8896,
      "step": 200500
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.1001920800915776e-05,
      "loss": 0.8792,
      "step": 201000
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.0904910653654372e-05,
      "loss": 0.8633,
      "step": 201500
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.080790050639297e-05,
      "loss": 0.9201,
      "step": 202000
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.0710890359131567e-05,
      "loss": 0.9096,
      "step": 202500
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.0613880211870162e-05,
      "loss": 0.8737,
      "step": 203000
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.0516870064608758e-05,
      "loss": 0.8475,
      "step": 203500
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.0419859917347356e-05,
      "loss": 0.9165,
      "step": 204000
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.0322849770085951e-05,
      "loss": 0.8994,
      "step": 204500
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.0225839622824549e-05,
      "loss": 0.9323,
      "step": 205000
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.0128829475563145e-05,
      "loss": 0.9088,
      "step": 205500
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.003181932830174e-05,
      "loss": 0.928,
      "step": 206000
    },
    {
      "epoch": 5.61,
      "learning_rate": 9.934809181040338e-06,
      "loss": 0.8902,
      "step": 206500
    },
    {
      "epoch": 5.62,
      "learning_rate": 9.837799033778933e-06,
      "loss": 0.9114,
      "step": 207000
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.74078888651753e-06,
      "loss": 0.8386,
      "step": 207500
    },
    {
      "epoch": 5.65,
      "learning_rate": 9.643778739256127e-06,
      "loss": 0.868,
      "step": 208000
    },
    {
      "epoch": 5.66,
      "learning_rate": 9.546768591994722e-06,
      "loss": 0.9076,
      "step": 208500
    },
    {
      "epoch": 5.68,
      "learning_rate": 9.44975844473332e-06,
      "loss": 0.8464,
      "step": 209000
    },
    {
      "epoch": 5.69,
      "learning_rate": 9.352748297471917e-06,
      "loss": 0.8549,
      "step": 209500
    },
    {
      "epoch": 5.7,
      "learning_rate": 9.255738150210511e-06,
      "loss": 0.8513,
      "step": 210000
    },
    {
      "epoch": 5.72,
      "learning_rate": 9.158728002949109e-06,
      "loss": 0.8795,
      "step": 210500
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.061717855687706e-06,
      "loss": 0.8875,
      "step": 211000
    },
    {
      "epoch": 5.74,
      "learning_rate": 8.964707708426302e-06,
      "loss": 0.8778,
      "step": 211500
    },
    {
      "epoch": 5.76,
      "learning_rate": 8.867697561164898e-06,
      "loss": 0.8956,
      "step": 212000
    },
    {
      "epoch": 5.77,
      "learning_rate": 8.770687413903495e-06,
      "loss": 0.9028,
      "step": 212500
    },
    {
      "epoch": 5.79,
      "learning_rate": 8.67367726664209e-06,
      "loss": 0.8707,
      "step": 213000
    },
    {
      "epoch": 5.8,
      "learning_rate": 8.576667119380688e-06,
      "loss": 0.8967,
      "step": 213500
    },
    {
      "epoch": 5.81,
      "learning_rate": 8.479656972119284e-06,
      "loss": 0.8202,
      "step": 214000
    },
    {
      "epoch": 5.83,
      "learning_rate": 8.38264682485788e-06,
      "loss": 0.8801,
      "step": 214500
    },
    {
      "epoch": 5.84,
      "learning_rate": 8.285636677596477e-06,
      "loss": 0.8748,
      "step": 215000
    },
    {
      "epoch": 5.85,
      "learning_rate": 8.188626530335075e-06,
      "loss": 0.9012,
      "step": 215500
    },
    {
      "epoch": 5.87,
      "learning_rate": 8.09161638307367e-06,
      "loss": 0.8916,
      "step": 216000
    },
    {
      "epoch": 5.88,
      "learning_rate": 7.994606235812266e-06,
      "loss": 0.8725,
      "step": 216500
    },
    {
      "epoch": 5.89,
      "learning_rate": 7.897596088550863e-06,
      "loss": 0.9009,
      "step": 217000
    },
    {
      "epoch": 5.91,
      "learning_rate": 7.80058594128946e-06,
      "loss": 0.8911,
      "step": 217500
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.703575794028057e-06,
      "loss": 0.9234,
      "step": 218000
    },
    {
      "epoch": 5.94,
      "learning_rate": 7.6065656467666515e-06,
      "loss": 0.8613,
      "step": 218500
    },
    {
      "epoch": 5.95,
      "learning_rate": 7.509555499505248e-06,
      "loss": 0.8522,
      "step": 219000
    },
    {
      "epoch": 5.96,
      "learning_rate": 7.4125453522438455e-06,
      "loss": 0.9087,
      "step": 219500
    },
    {
      "epoch": 5.98,
      "learning_rate": 7.315535204982442e-06,
      "loss": 0.885,
      "step": 220000
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.218525057721039e-06,
      "loss": 0.8861,
      "step": 220500
    },
    {
      "epoch": 6.0,
      "learning_rate": 7.121514910459634e-06,
      "loss": 0.9002,
      "step": 221000
    },
    {
      "epoch": 6.02,
      "learning_rate": 7.024504763198231e-06,
      "loss": 0.8999,
      "step": 221500
    },
    {
      "epoch": 6.03,
      "learning_rate": 6.9274946159368276e-06,
      "loss": 0.9182,
      "step": 222000
    },
    {
      "epoch": 6.04,
      "learning_rate": 6.830484468675424e-06,
      "loss": 0.8479,
      "step": 222500
    },
    {
      "epoch": 6.06,
      "learning_rate": 6.73347432141402e-06,
      "loss": 0.9325,
      "step": 223000
    },
    {
      "epoch": 6.07,
      "learning_rate": 6.6364641741526165e-06,
      "loss": 0.8421,
      "step": 223500
    },
    {
      "epoch": 6.08,
      "learning_rate": 6.539454026891213e-06,
      "loss": 0.8992,
      "step": 224000
    },
    {
      "epoch": 6.1,
      "learning_rate": 6.4424438796298105e-06,
      "loss": 0.8851,
      "step": 224500
    },
    {
      "epoch": 6.11,
      "learning_rate": 6.345433732368405e-06,
      "loss": 0.8902,
      "step": 225000
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.248423585107002e-06,
      "loss": 0.8804,
      "step": 225500
    },
    {
      "epoch": 6.14,
      "learning_rate": 6.151413437845599e-06,
      "loss": 0.9066,
      "step": 226000
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.054403290584195e-06,
      "loss": 0.8402,
      "step": 226500
    },
    {
      "epoch": 6.17,
      "learning_rate": 5.957393143322792e-06,
      "loss": 0.817,
      "step": 227000
    },
    {
      "epoch": 6.18,
      "learning_rate": 5.860382996061388e-06,
      "loss": 0.8759,
      "step": 227500
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.763372848799985e-06,
      "loss": 0.9003,
      "step": 228000
    },
    {
      "epoch": 6.21,
      "learning_rate": 5.6663627015385814e-06,
      "loss": 0.8757,
      "step": 228500
    },
    {
      "epoch": 6.22,
      "learning_rate": 5.569352554277177e-06,
      "loss": 0.8991,
      "step": 229000
    },
    {
      "epoch": 6.23,
      "learning_rate": 5.472342407015775e-06,
      "loss": 0.8822,
      "step": 229500
    },
    {
      "epoch": 6.25,
      "learning_rate": 5.37533225975437e-06,
      "loss": 0.8906,
      "step": 230000
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.278322112492967e-06,
      "loss": 0.8844,
      "step": 230500
    },
    {
      "epoch": 6.27,
      "learning_rate": 5.1813119652315635e-06,
      "loss": 0.8714,
      "step": 231000
    },
    {
      "epoch": 6.29,
      "learning_rate": 5.08430181797016e-06,
      "loss": 0.8623,
      "step": 231500
    },
    {
      "epoch": 6.3,
      "learning_rate": 4.987291670708756e-06,
      "loss": 0.8191,
      "step": 232000
    },
    {
      "epoch": 6.32,
      "learning_rate": 4.890281523447353e-06,
      "loss": 0.8605,
      "step": 232500
    },
    {
      "epoch": 6.33,
      "learning_rate": 4.79327137618595e-06,
      "loss": 0.8809,
      "step": 233000
    },
    {
      "epoch": 6.34,
      "learning_rate": 4.6962612289245456e-06,
      "loss": 0.9204,
      "step": 233500
    },
    {
      "epoch": 6.36,
      "learning_rate": 4.599251081663142e-06,
      "loss": 0.8772,
      "step": 234000
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.502240934401739e-06,
      "loss": 0.8633,
      "step": 234500
    },
    {
      "epoch": 6.38,
      "learning_rate": 4.405230787140335e-06,
      "loss": 0.9188,
      "step": 235000
    },
    {
      "epoch": 6.4,
      "learning_rate": 4.308220639878931e-06,
      "loss": 0.8526,
      "step": 235500
    },
    {
      "epoch": 6.41,
      "learning_rate": 4.2112104926175285e-06,
      "loss": 0.8792,
      "step": 236000
    },
    {
      "epoch": 6.42,
      "learning_rate": 4.114200345356124e-06,
      "loss": 0.9183,
      "step": 236500
    },
    {
      "epoch": 6.44,
      "learning_rate": 4.017190198094721e-06,
      "loss": 0.8629,
      "step": 237000
    },
    {
      "epoch": 6.45,
      "learning_rate": 3.920180050833317e-06,
      "loss": 0.8728,
      "step": 237500
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.823169903571914e-06,
      "loss": 0.8578,
      "step": 238000
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.72615975631051e-06,
      "loss": 0.8989,
      "step": 238500
    },
    {
      "epoch": 6.49,
      "learning_rate": 3.6291496090491067e-06,
      "loss": 0.9086,
      "step": 239000
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.5321394617877033e-06,
      "loss": 0.889,
      "step": 239500
    },
    {
      "epoch": 6.52,
      "learning_rate": 3.4351293145262995e-06,
      "loss": 0.8783,
      "step": 240000
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.3381191672648965e-06,
      "loss": 0.8762,
      "step": 240500
    },
    {
      "epoch": 6.55,
      "learning_rate": 3.2411090200034926e-06,
      "loss": 0.9224,
      "step": 241000
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.144098872742089e-06,
      "loss": 0.8741,
      "step": 241500
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.0470887254806854e-06,
      "loss": 0.9123,
      "step": 242000
    },
    {
      "epoch": 6.59,
      "learning_rate": 2.950078578219282e-06,
      "loss": 0.849,
      "step": 242500
    },
    {
      "epoch": 6.6,
      "learning_rate": 2.8530684309578785e-06,
      "loss": 0.8696,
      "step": 243000
    },
    {
      "epoch": 6.61,
      "learning_rate": 2.756058283696475e-06,
      "loss": 0.8385,
      "step": 243500
    },
    {
      "epoch": 6.63,
      "learning_rate": 2.6590481364350713e-06,
      "loss": 0.8764,
      "step": 244000
    },
    {
      "epoch": 6.64,
      "learning_rate": 2.562037989173668e-06,
      "loss": 0.9135,
      "step": 244500
    },
    {
      "epoch": 6.65,
      "learning_rate": 2.465027841912264e-06,
      "loss": 0.8975,
      "step": 245000
    },
    {
      "epoch": 6.67,
      "learning_rate": 2.3680176946508606e-06,
      "loss": 0.894,
      "step": 245500
    },
    {
      "epoch": 6.68,
      "learning_rate": 2.271007547389457e-06,
      "loss": 0.8851,
      "step": 246000
    },
    {
      "epoch": 6.7,
      "learning_rate": 2.1739974001280533e-06,
      "loss": 0.872,
      "step": 246500
    },
    {
      "epoch": 6.71,
      "learning_rate": 2.07698725286665e-06,
      "loss": 0.8467,
      "step": 247000
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.9799771056052465e-06,
      "loss": 0.872,
      "step": 247500
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.8829669583438427e-06,
      "loss": 0.8703,
      "step": 248000
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.7859568110824392e-06,
      "loss": 0.8783,
      "step": 248500
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.6889466638210358e-06,
      "loss": 0.8553,
      "step": 249000
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.5919365165596324e-06,
      "loss": 0.865,
      "step": 249500
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.4949263692982286e-06,
      "loss": 0.8555,
      "step": 250000
    },
    {
      "epoch": 6.8,
      "learning_rate": 1.3979162220368251e-06,
      "loss": 0.8611,
      "step": 250500
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.3009060747754215e-06,
      "loss": 0.8721,
      "step": 251000
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.203895927514018e-06,
      "loss": 0.8965,
      "step": 251500
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.1068857802526145e-06,
      "loss": 0.8787,
      "step": 252000
    },
    {
      "epoch": 6.86,
      "learning_rate": 1.0098756329912108e-06,
      "loss": 0.8599,
      "step": 252500
    },
    {
      "epoch": 6.87,
      "learning_rate": 9.128654857298073e-07,
      "loss": 0.8675,
      "step": 253000
    },
    {
      "epoch": 6.89,
      "learning_rate": 8.158553384684039e-07,
      "loss": 0.9003,
      "step": 253500
    },
    {
      "epoch": 6.9,
      "learning_rate": 7.188451912070003e-07,
      "loss": 0.8969,
      "step": 254000
    },
    {
      "epoch": 6.91,
      "learning_rate": 6.218350439455968e-07,
      "loss": 0.8741,
      "step": 254500
    },
    {
      "epoch": 6.93,
      "learning_rate": 5.248248966841932e-07,
      "loss": 0.8867,
      "step": 255000
    },
    {
      "epoch": 6.94,
      "learning_rate": 4.278147494227896e-07,
      "loss": 0.885,
      "step": 255500
    },
    {
      "epoch": 6.95,
      "learning_rate": 3.308046021613861e-07,
      "loss": 0.8513,
      "step": 256000
    },
    {
      "epoch": 6.97,
      "learning_rate": 2.3379445489998255e-07,
      "loss": 0.8624,
      "step": 256500
    },
    {
      "epoch": 6.98,
      "learning_rate": 1.36784307638579e-07,
      "loss": 0.8955,
      "step": 257000
    },
    {
      "epoch": 6.99,
      "learning_rate": 3.9774160377175454e-08,
      "loss": 0.8978,
      "step": 257500
    },
    {
      "epoch": 7.0,
      "step": 257705,
      "total_flos": 4.149897903123333e+17,
      "train_loss": 0.9280026929055305,
      "train_runtime": 81540.4406,
      "train_samples_per_second": 18.962,
      "train_steps_per_second": 3.16
    }
  ],
  "max_steps": 257705,
  "num_train_epochs": 7,
  "total_flos": 4.149897903123333e+17,
  "trial_name": null,
  "trial_params": null
}