{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.977333242302178,
  "eval_steps": 1000000,
  "global_step": 291280,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 5.000000000000001e-07,
      "loss": 10.131,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 9.5844,
      "step": 1000
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.5e-06,
      "loss": 9.2458,
      "step": 1500
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 8.8439,
      "step": 2000
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.5e-06,
      "loss": 8.3653,
      "step": 2500
    },
    {
      "epoch": 0.04,
      "learning_rate": 3e-06,
      "loss": 7.896,
      "step": 3000
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.5e-06,
      "loss": 7.5097,
      "step": 3500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.000000000000001e-06,
      "loss": 7.243,
      "step": 4000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.5e-06,
      "loss": 7.0838,
      "step": 4500
    },
    {
      "epoch": 0.07,
      "learning_rate": 5e-06,
      "loss": 6.9762,
      "step": 5000
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.500000000000001e-06,
      "loss": 6.9075,
      "step": 5500
    },
    {
      "epoch": 0.08,
      "learning_rate": 6e-06,
      "loss": 6.8523,
      "step": 6000
    },
    {
      "epoch": 0.09,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 6.8056,
      "step": 6500
    },
    {
      "epoch": 0.1,
      "learning_rate": 7e-06,
      "loss": 6.7608,
      "step": 7000
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.500000000000001e-06,
      "loss": 6.7278,
      "step": 7500
    },
    {
      "epoch": 0.11,
      "learning_rate": 8.000000000000001e-06,
      "loss": 6.697,
      "step": 8000
    },
    {
      "epoch": 0.12,
      "learning_rate": 8.5e-06,
      "loss": 6.6687,
      "step": 8500
    },
    {
      "epoch": 0.12,
      "learning_rate": 9e-06,
      "loss": 6.6394,
      "step": 9000
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.5e-06,
      "loss": 6.6134,
      "step": 9500
    },
    {
      "epoch": 0.14,
      "learning_rate": 1e-05,
      "loss": 6.5997,
      "step": 10000
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.999316332809189e-06,
      "loss": 6.571,
      "step": 10500
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.998632665618377e-06,
      "loss": 6.5458,
      "step": 11000
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.997948998427567e-06,
      "loss": 6.5301,
      "step": 11500
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.997265331236755e-06,
      "loss": 6.5043,
      "step": 12000
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.996581664045943e-06,
      "loss": 6.4904,
      "step": 12500
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.995897996855133e-06,
      "loss": 6.4727,
      "step": 13000
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.99521432966432e-06,
      "loss": 6.4556,
      "step": 13500
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.994530662473509e-06,
      "loss": 6.4457,
      "step": 14000
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.993846995282698e-06,
      "loss": 6.4316,
      "step": 14500
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.993163328091886e-06,
      "loss": 6.4167,
      "step": 15000
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.992479660901074e-06,
      "loss": 6.4068,
      "step": 15500
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.991795993710262e-06,
      "loss": 6.4019,
      "step": 16000
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.99111232651945e-06,
      "loss": 6.385,
      "step": 16500
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.990428659328639e-06,
      "loss": 6.3783,
      "step": 17000
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.989744992137828e-06,
      "loss": 6.3659,
      "step": 17500
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.989061324947016e-06,
      "loss": 6.3611,
      "step": 18000
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.988377657756204e-06,
      "loss": 6.3483,
      "step": 18500
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.987693990565394e-06,
      "loss": 6.3364,
      "step": 19000
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.987010323374582e-06,
      "loss": 6.3351,
      "step": 19500
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.98632665618377e-06,
      "loss": 6.3213,
      "step": 20000
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.98564298899296e-06,
      "loss": 6.3141,
      "step": 20500
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.984959321802148e-06,
      "loss": 6.3102,
      "step": 21000
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.984275654611336e-06,
      "loss": 6.3055,
      "step": 21500
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.983591987420524e-06,
      "loss": 6.2931,
      "step": 22000
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.982908320229712e-06,
      "loss": 6.2941,
      "step": 22500
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.9822246530389e-06,
      "loss": 6.2862,
      "step": 23000
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.98154098584809e-06,
      "loss": 6.2811,
      "step": 23500
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.980857318657278e-06,
      "loss": 6.2707,
      "step": 24000
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.980173651466468e-06,
      "loss": 6.2685,
      "step": 24500
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.979489984275656e-06,
      "loss": 6.2648,
      "step": 25000
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.978806317084844e-06,
      "loss": 6.2582,
      "step": 25500
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.978122649894032e-06,
      "loss": 6.2545,
      "step": 26000
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.977438982703222e-06,
      "loss": 6.252,
      "step": 26500
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.97675531551241e-06,
      "loss": 6.2425,
      "step": 27000
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.976071648321598e-06,
      "loss": 6.2362,
      "step": 27500
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.975387981130786e-06,
      "loss": 6.2306,
      "step": 28000
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.974704313939974e-06,
      "loss": 6.2367,
      "step": 28500
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.974020646749164e-06,
      "loss": 6.2269,
      "step": 29000
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.973336979558352e-06,
      "loss": 6.2226,
      "step": 29500
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.972653312367541e-06,
      "loss": 6.2205,
      "step": 30000
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.97196964517673e-06,
      "loss": 6.2161,
      "step": 30500
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.971285977985918e-06,
      "loss": 6.2141,
      "step": 31000
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.970602310795106e-06,
      "loss": 6.2032,
      "step": 31500
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.969918643604294e-06,
      "loss": 6.1959,
      "step": 32000
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.969234976413483e-06,
      "loss": 6.202,
      "step": 32500
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.968551309222671e-06,
      "loss": 6.1938,
      "step": 33000
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.96786764203186e-06,
      "loss": 6.1909,
      "step": 33500
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.967183974841048e-06,
      "loss": 6.1947,
      "step": 34000
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.966500307650237e-06,
      "loss": 6.1876,
      "step": 34500
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.965816640459425e-06,
      "loss": 6.1803,
      "step": 35000
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.965132973268613e-06,
      "loss": 6.1811,
      "step": 35500
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.964449306077803e-06,
      "loss": 6.1823,
      "step": 36000
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.963765638886991e-06,
      "loss": 6.1775,
      "step": 36500
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.96308197169618e-06,
      "loss": 6.1735,
      "step": 37000
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.962398304505367e-06,
      "loss": 6.1693,
      "step": 37500
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.961714637314555e-06,
      "loss": 6.167,
      "step": 38000
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.961030970123745e-06,
      "loss": 6.1609,
      "step": 38500
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.960347302932933e-06,
      "loss": 6.1611,
      "step": 39000
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.959663635742121e-06,
      "loss": 6.1662,
      "step": 39500
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.95897996855131e-06,
      "loss": 6.1569,
      "step": 40000
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.958296301360499e-06,
      "loss": 6.158,
      "step": 40500
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.957612634169687e-06,
      "loss": 6.1419,
      "step": 41000
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.956928966978877e-06,
      "loss": 6.145,
      "step": 41500
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.956245299788065e-06,
      "loss": 6.1445,
      "step": 42000
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.955561632597253e-06,
      "loss": 6.1469,
      "step": 42500
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.954877965406441e-06,
      "loss": 6.1414,
      "step": 43000
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.954194298215629e-06,
      "loss": 6.1408,
      "step": 43500
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.953510631024817e-06,
      "loss": 6.1394,
      "step": 44000
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.952826963834007e-06,
      "loss": 6.1403,
      "step": 44500
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.952143296643195e-06,
      "loss": 6.1324,
      "step": 45000
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.951459629452383e-06,
      "loss": 6.1277,
      "step": 45500
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.950775962261573e-06,
      "loss": 6.13,
      "step": 46000
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.95009229507076e-06,
      "loss": 6.1265,
      "step": 46500
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.949408627879949e-06,
      "loss": 6.1261,
      "step": 47000
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.948724960689138e-06,
      "loss": 6.1207,
      "step": 47500
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.948041293498326e-06,
      "loss": 6.1266,
      "step": 48000
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.947357626307515e-06,
      "loss": 6.1177,
      "step": 48500
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.946673959116703e-06,
      "loss": 6.1173,
      "step": 49000
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.94599029192589e-06,
      "loss": 6.115,
      "step": 49500
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.945306624735079e-06,
      "loss": 6.1069,
      "step": 50000
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.944622957544268e-06,
      "loss": 6.1106,
      "step": 50500
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.943939290353456e-06,
      "loss": 6.1105,
      "step": 51000
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.943255623162645e-06,
      "loss": 6.1098,
      "step": 51500
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.942571955971834e-06,
      "loss": 6.1038,
      "step": 52000
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.941888288781022e-06,
      "loss": 6.1072,
      "step": 52500
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.94120462159021e-06,
      "loss": 6.0991,
      "step": 53000
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.9405209543994e-06,
      "loss": 6.1033,
      "step": 53500
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.939837287208588e-06,
      "loss": 6.0987,
      "step": 54000
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.939153620017776e-06,
      "loss": 6.0999,
      "step": 54500
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.938469952826964e-06,
      "loss": 6.0953,
      "step": 55000
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.937786285636152e-06,
      "loss": 6.0919,
      "step": 55500
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.93710261844534e-06,
      "loss": 6.0937,
      "step": 56000
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.93641895125453e-06,
      "loss": 6.0957,
      "step": 56500
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.935735284063718e-06,
      "loss": 6.0826,
      "step": 57000
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.935051616872908e-06,
      "loss": 6.0903,
      "step": 57500
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.934367949682096e-06,
      "loss": 6.0857,
      "step": 58000
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.933684282491284e-06,
      "loss": 6.0848,
      "step": 58500
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.933000615300472e-06,
      "loss": 6.0859,
      "step": 59000
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.932316948109662e-06,
      "loss": 6.0842,
      "step": 59500
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.93163328091885e-06,
      "loss": 6.0816,
      "step": 60000
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.930949613728038e-06,
      "loss": 6.0753,
      "step": 60500
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.930265946537226e-06,
      "loss": 6.0789,
      "step": 61000
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.929582279346414e-06,
      "loss": 6.0788,
      "step": 61500
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.928898612155604e-06,
      "loss": 6.0758,
      "step": 62000
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.928214944964792e-06,
      "loss": 6.0771,
      "step": 62500
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.927531277773982e-06,
      "loss": 6.068,
      "step": 63000
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.92684761058317e-06,
      "loss": 6.0746,
      "step": 63500
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.926163943392358e-06,
      "loss": 6.0752,
      "step": 64000
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.925480276201546e-06,
      "loss": 6.0771,
      "step": 64500
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.924796609010734e-06,
      "loss": 6.0671,
      "step": 65000
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.924112941819923e-06,
      "loss": 6.0693,
      "step": 65500
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.923429274629112e-06,
      "loss": 6.0655,
      "step": 66000
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.9227456074383e-06,
      "loss": 6.0644,
      "step": 66500
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.922061940247488e-06,
      "loss": 6.0648,
      "step": 67000
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.921378273056677e-06,
      "loss": 6.0651,
      "step": 67500
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.920694605865865e-06,
      "loss": 6.0659,
      "step": 68000
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.920010938675053e-06,
      "loss": 6.0643,
      "step": 68500
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.919327271484243e-06,
      "loss": 6.0582,
      "step": 69000
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.918643604293431e-06,
      "loss": 6.0616,
      "step": 69500
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.91795993710262e-06,
      "loss": 6.0586,
      "step": 70000
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.917276269911807e-06,
      "loss": 6.0601,
      "step": 70500
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.916592602720995e-06,
      "loss": 6.0566,
      "step": 71000
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.915908935530185e-06,
      "loss": 6.0577,
      "step": 71500
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.915225268339373e-06,
      "loss": 6.054,
      "step": 72000
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.914541601148561e-06,
      "loss": 6.0532,
      "step": 72500
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.91385793395775e-06,
      "loss": 6.0513,
      "step": 73000
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.913174266766939e-06,
      "loss": 6.0501,
      "step": 73500
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.912490599576127e-06,
      "loss": 6.053,
      "step": 74000
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.911806932385317e-06,
      "loss": 6.05,
      "step": 74500
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.911123265194505e-06,
      "loss": 6.0497,
      "step": 75000
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.910439598003693e-06,
      "loss": 6.0512,
      "step": 75500
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.909755930812881e-06,
      "loss": 6.0432,
      "step": 76000
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.909072263622069e-06,
      "loss": 6.0403,
      "step": 76500
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.908388596431257e-06,
      "loss": 6.0399,
      "step": 77000
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.907704929240447e-06,
      "loss": 6.045,
      "step": 77500
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.907021262049635e-06,
      "loss": 6.0427,
      "step": 78000
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.906337594858823e-06,
      "loss": 6.0438,
      "step": 78500
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.905653927668013e-06,
      "loss": 6.0402,
      "step": 79000
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.9049702604772e-06,
      "loss": 6.0419,
      "step": 79500
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.904286593286389e-06,
      "loss": 6.0365,
      "step": 80000
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.903602926095579e-06,
      "loss": 6.04,
      "step": 80500
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.902919258904767e-06,
      "loss": 6.031,
      "step": 81000
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.902235591713955e-06,
      "loss": 6.0366,
      "step": 81500
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.901551924523143e-06,
      "loss": 6.0359,
      "step": 82000
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.90086825733233e-06,
      "loss": 6.0353,
      "step": 82500
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.900184590141519e-06,
      "loss": 6.0311,
      "step": 83000
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.899500922950709e-06,
      "loss": 6.0341,
      "step": 83500
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.898817255759897e-06,
      "loss": 6.0333,
      "step": 84000
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.898133588569086e-06,
      "loss": 6.0319,
      "step": 84500
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.897449921378274e-06,
      "loss": 6.0284,
      "step": 85000
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.896766254187462e-06,
      "loss": 6.0292,
      "step": 85500
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.89608258699665e-06,
      "loss": 6.0307,
      "step": 86000
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.89539891980584e-06,
      "loss": 6.0264,
      "step": 86500
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.894715252615028e-06,
      "loss": 6.0239,
      "step": 87000
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.894031585424216e-06,
      "loss": 6.0246,
      "step": 87500
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.893347918233404e-06,
      "loss": 6.0247,
      "step": 88000
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.892664251042592e-06,
      "loss": 6.0192,
      "step": 88500
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.89198058385178e-06,
      "loss": 6.0235,
      "step": 89000
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.89129691666097e-06,
      "loss": 6.0225,
      "step": 89500
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.890613249470158e-06,
      "loss": 6.0189,
      "step": 90000
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.889929582279348e-06,
      "loss": 6.0178,
      "step": 90500
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.889245915088536e-06,
      "loss": 6.0215,
      "step": 91000
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.888562247897724e-06,
      "loss": 6.0145,
      "step": 91500
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.887878580706912e-06,
      "loss": 6.0162,
      "step": 92000
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.887194913516102e-06,
      "loss": 6.019,
      "step": 92500
    },
    {
      "epoch": 1.27,
      "learning_rate": 9.88651124632529e-06,
      "loss": 6.0157,
      "step": 93000
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.885827579134478e-06,
      "loss": 6.0173,
      "step": 93500
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.885143911943666e-06,
      "loss": 6.0208,
      "step": 94000
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.884460244752854e-06,
      "loss": 6.0189,
      "step": 94500
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.883776577562044e-06,
      "loss": 6.0121,
      "step": 95000
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.883092910371232e-06,
      "loss": 6.0109,
      "step": 95500
    },
    {
      "epoch": 1.31,
      "learning_rate": 9.882409243180422e-06,
      "loss": 6.0191,
      "step": 96000
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.88172557598961e-06,
      "loss": 6.0133,
      "step": 96500
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.881041908798798e-06,
      "loss": 6.0076,
      "step": 97000
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.880358241607986e-06,
      "loss": 6.0142,
      "step": 97500
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.879674574417174e-06,
      "loss": 6.0056,
      "step": 98000
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.878990907226364e-06,
      "loss": 6.0098,
      "step": 98500
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.878307240035552e-06,
      "loss": 6.0128,
      "step": 99000
    },
    {
      "epoch": 1.36,
      "learning_rate": 9.87762357284474e-06,
      "loss": 6.0097,
      "step": 99500
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.876939905653928e-06,
      "loss": 6.0072,
      "step": 100000
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.876256238463117e-06,
      "loss": 5.9995,
      "step": 100500
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.875572571272305e-06,
      "loss": 6.0029,
      "step": 101000
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.874888904081494e-06,
      "loss": 6.0034,
      "step": 101500
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.874205236890683e-06,
      "loss": 6.0061,
      "step": 102000
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.873521569699871e-06,
      "loss": 6.0025,
      "step": 102500
    },
    {
      "epoch": 1.41,
      "learning_rate": 9.87283790250906e-06,
      "loss": 5.9943,
      "step": 103000
    },
    {
      "epoch": 1.41,
      "learning_rate": 9.872154235318247e-06,
      "loss": 5.9901,
      "step": 103500
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.871470568127435e-06,
      "loss": 5.9888,
      "step": 104000
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.870786900936625e-06,
      "loss": 5.9835,
      "step": 104500
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.870103233745813e-06,
      "loss": 5.9757,
      "step": 105000
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.869419566555001e-06,
      "loss": 5.9705,
      "step": 105500
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.86873589936419e-06,
      "loss": 5.958,
      "step": 106000
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.868052232173379e-06,
      "loss": 5.9471,
      "step": 106500
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.867368564982567e-06,
      "loss": 5.9333,
      "step": 107000
    },
    {
      "epoch": 1.47,
      "learning_rate": 9.866684897791757e-06,
      "loss": 5.9207,
      "step": 107500
    },
    {
      "epoch": 1.47,
      "learning_rate": 9.866001230600945e-06,
      "loss": 5.9089,
      "step": 108000
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.865317563410133e-06,
      "loss": 5.8966,
      "step": 108500
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.864633896219321e-06,
      "loss": 5.8896,
      "step": 109000
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.863950229028509e-06,
      "loss": 5.8734,
      "step": 109500
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.863266561837697e-06,
      "loss": 5.8704,
      "step": 110000
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.862582894646887e-06,
      "loss": 5.8559,
      "step": 110500
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.861899227456075e-06,
      "loss": 5.8555,
      "step": 111000
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.861215560265263e-06,
      "loss": 5.8431,
      "step": 111500
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.860531893074453e-06,
      "loss": 5.8343,
      "step": 112000
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.85984822588364e-06,
      "loss": 5.8331,
      "step": 112500
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.859164558692829e-06,
      "loss": 5.8299,
      "step": 113000
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.858480891502019e-06,
      "loss": 5.8192,
      "step": 113500
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.857797224311207e-06,
      "loss": 5.8182,
      "step": 114000
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.857113557120395e-06,
      "loss": 5.8136,
      "step": 114500
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.856429889929583e-06,
      "loss": 5.8016,
      "step": 115000
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.85574622273877e-06,
      "loss": 5.7994,
      "step": 115500
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.855062555547959e-06,
      "loss": 5.793,
      "step": 116000
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.854378888357149e-06,
      "loss": 5.7919,
      "step": 116500
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.853695221166337e-06,
      "loss": 5.789,
      "step": 117000
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.853011553975526e-06,
      "loss": 5.7812,
      "step": 117500
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.852327886784714e-06,
      "loss": 5.7737,
      "step": 118000
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.851644219593902e-06,
      "loss": 5.7713,
      "step": 118500
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.85096055240309e-06,
      "loss": 5.7668,
      "step": 119000
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.85027688521228e-06,
      "loss": 5.7586,
      "step": 119500
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.849593218021468e-06,
      "loss": 5.763,
      "step": 120000
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.848909550830656e-06,
      "loss": 5.7516,
      "step": 120500
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.848225883639844e-06,
      "loss": 5.754,
      "step": 121000
    },
    {
      "epoch": 1.66,
      "learning_rate": 9.847542216449032e-06,
      "loss": 5.7465,
      "step": 121500
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.846858549258222e-06,
      "loss": 5.739,
      "step": 122000
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.84617488206741e-06,
      "loss": 5.7388,
      "step": 122500
    },
    {
      "epoch": 1.68,
      "learning_rate": 9.845491214876598e-06,
      "loss": 5.7324,
      "step": 123000
    },
    {
      "epoch": 1.69,
      "learning_rate": 9.844807547685788e-06,
      "loss": 5.7321,
      "step": 123500
    },
    {
      "epoch": 1.69,
      "learning_rate": 9.844123880494976e-06,
      "loss": 5.7235,
      "step": 124000
    },
    {
      "epoch": 1.7,
      "learning_rate": 9.843440213304164e-06,
      "loss": 5.7229,
      "step": 124500
    },
    {
      "epoch": 1.71,
      "learning_rate": 9.842756546113352e-06,
      "loss": 5.7163,
      "step": 125000
    },
    {
      "epoch": 1.71,
      "learning_rate": 9.842072878922542e-06,
      "loss": 5.7159,
      "step": 125500
    },
    {
      "epoch": 1.72,
      "learning_rate": 9.84138921173173e-06,
      "loss": 5.7104,
      "step": 126000
    },
    {
      "epoch": 1.73,
      "learning_rate": 9.840705544540918e-06,
      "loss": 5.7094,
      "step": 126500
    },
    {
      "epoch": 1.73,
      "learning_rate": 9.840021877350106e-06,
      "loss": 5.7052,
      "step": 127000
    },
    {
      "epoch": 1.74,
      "learning_rate": 9.839338210159294e-06,
      "loss": 5.704,
      "step": 127500
    },
    {
      "epoch": 1.75,
      "learning_rate": 9.838654542968484e-06,
      "loss": 5.6962,
      "step": 128000
    },
    {
      "epoch": 1.75,
      "learning_rate": 9.837970875777672e-06,
      "loss": 5.6941,
      "step": 128500
    },
    {
      "epoch": 1.76,
      "learning_rate": 9.837287208586862e-06,
      "loss": 5.6885,
      "step": 129000
    },
    {
      "epoch": 1.77,
      "learning_rate": 9.83660354139605e-06,
      "loss": 5.686,
      "step": 129500
    },
    {
      "epoch": 1.78,
      "learning_rate": 9.835919874205238e-06,
      "loss": 5.6832,
      "step": 130000
    },
    {
      "epoch": 1.78,
      "learning_rate": 9.835236207014426e-06,
      "loss": 5.6758,
      "step": 130500
    },
    {
      "epoch": 1.79,
      "learning_rate": 9.834552539823614e-06,
      "loss": 5.6751,
      "step": 131000
    },
    {
      "epoch": 1.8,
      "learning_rate": 9.833868872632804e-06,
      "loss": 5.6679,
      "step": 131500
    },
    {
      "epoch": 1.8,
      "learning_rate": 9.833185205441992e-06,
      "loss": 5.6644,
      "step": 132000
    },
    {
      "epoch": 1.81,
      "learning_rate": 9.83250153825118e-06,
      "loss": 5.6645,
      "step": 132500
    },
    {
      "epoch": 1.82,
      "learning_rate": 9.831817871060368e-06,
      "loss": 5.6606,
      "step": 133000
    },
    {
      "epoch": 1.82,
      "learning_rate": 9.831134203869558e-06,
      "loss": 5.6558,
      "step": 133500
    },
    {
      "epoch": 1.83,
      "learning_rate": 9.830450536678746e-06,
      "loss": 5.6458,
      "step": 134000
    },
    {
      "epoch": 1.84,
      "learning_rate": 9.829766869487934e-06,
      "loss": 5.6493,
      "step": 134500
    },
    {
      "epoch": 1.84,
      "learning_rate": 9.829083202297123e-06,
      "loss": 5.642,
      "step": 135000
    },
    {
      "epoch": 1.85,
      "learning_rate": 9.828399535106311e-06,
      "loss": 5.6355,
      "step": 135500
    },
    {
      "epoch": 1.86,
      "learning_rate": 9.8277158679155e-06,
      "loss": 5.6328,
      "step": 136000
    },
    {
      "epoch": 1.86,
      "learning_rate": 9.827032200724688e-06,
      "loss": 5.6272,
      "step": 136500
    },
    {
      "epoch": 1.87,
      "learning_rate": 9.826348533533876e-06,
      "loss": 5.6234,
      "step": 137000
    },
    {
      "epoch": 1.88,
      "learning_rate": 9.825664866343065e-06,
      "loss": 5.6167,
      "step": 137500
    },
    {
      "epoch": 1.88,
      "learning_rate": 9.824981199152253e-06,
      "loss": 5.6098,
      "step": 138000
    },
    {
      "epoch": 1.89,
      "learning_rate": 9.824297531961441e-06,
      "loss": 5.6053,
      "step": 138500
    },
    {
      "epoch": 1.9,
      "learning_rate": 9.82361386477063e-06,
      "loss": 5.6072,
      "step": 139000
    },
    {
      "epoch": 1.9,
      "learning_rate": 9.82293019757982e-06,
      "loss": 5.5956,
      "step": 139500
    },
    {
      "epoch": 1.91,
      "learning_rate": 9.822246530389007e-06,
      "loss": 5.5886,
      "step": 140000
    },
    {
      "epoch": 1.92,
      "learning_rate": 9.821562863198197e-06,
      "loss": 5.5851,
      "step": 140500
    },
    {
      "epoch": 1.93,
      "learning_rate": 9.820879196007385e-06,
      "loss": 5.5827,
      "step": 141000
    },
    {
      "epoch": 1.93,
      "learning_rate": 9.820195528816573e-06,
      "loss": 5.577,
      "step": 141500
    },
    {
      "epoch": 1.94,
      "learning_rate": 9.819511861625761e-06,
      "loss": 5.5718,
      "step": 142000
    },
    {
      "epoch": 1.95,
      "learning_rate": 9.81882819443495e-06,
      "loss": 5.57,
      "step": 142500
    },
    {
      "epoch": 1.95,
      "learning_rate": 9.818144527244139e-06,
      "loss": 5.563,
      "step": 143000
    },
    {
      "epoch": 1.96,
      "learning_rate": 9.817460860053327e-06,
      "loss": 5.5568,
      "step": 143500
    },
    {
      "epoch": 1.97,
      "learning_rate": 9.816777192862515e-06,
      "loss": 5.5546,
      "step": 144000
    },
    {
      "epoch": 1.97,
      "learning_rate": 9.816093525671703e-06,
      "loss": 5.545,
      "step": 144500
    },
    {
      "epoch": 1.98,
      "learning_rate": 9.815409858480893e-06,
      "loss": 5.5436,
      "step": 145000
    },
    {
      "epoch": 1.99,
      "learning_rate": 9.814726191290081e-06,
      "loss": 5.5376,
      "step": 145500
    },
    {
      "epoch": 1.99,
      "learning_rate": 9.81404252409927e-06,
      "loss": 5.5359,
      "step": 146000
    },
    {
      "epoch": 2.0,
      "learning_rate": 9.813358856908459e-06,
      "loss": 5.5284,
      "step": 146500
    },
    {
      "epoch": 2.01,
      "learning_rate": 9.812675189717647e-06,
      "loss": 5.5249,
      "step": 147000
    },
    {
      "epoch": 2.01,
      "learning_rate": 9.811991522526835e-06,
      "loss": 5.5171,
      "step": 147500
    },
    {
      "epoch": 2.02,
      "learning_rate": 9.811307855336023e-06,
      "loss": 5.517,
      "step": 148000
    },
    {
      "epoch": 2.03,
      "learning_rate": 9.810624188145211e-06,
      "loss": 5.5045,
      "step": 148500
    },
    {
      "epoch": 2.03,
      "learning_rate": 9.8099405209544e-06,
      "loss": 5.5045,
      "step": 149000
    },
    {
      "epoch": 2.04,
      "learning_rate": 9.809256853763589e-06,
      "loss": 5.4991,
      "step": 149500
    },
    {
      "epoch": 2.05,
      "learning_rate": 9.808573186572777e-06,
      "loss": 5.4926,
      "step": 150000
    },
    {
      "epoch": 2.06,
      "learning_rate": 9.807889519381966e-06,
      "loss": 5.4873,
      "step": 150500
    },
    {
      "epoch": 2.06,
      "learning_rate": 9.807205852191155e-06,
      "loss": 5.4837,
      "step": 151000
    },
    {
      "epoch": 2.07,
      "learning_rate": 9.806522185000343e-06,
      "loss": 5.4751,
      "step": 151500
    },
    {
      "epoch": 2.08,
      "learning_rate": 9.805838517809532e-06,
      "loss": 5.4719,
      "step": 152000
    },
    {
      "epoch": 2.08,
      "learning_rate": 9.80515485061872e-06,
      "loss": 5.4636,
      "step": 152500
    },
    {
      "epoch": 2.09,
      "learning_rate": 9.804471183427908e-06,
      "loss": 5.4613,
      "step": 153000
    },
    {
      "epoch": 2.1,
      "learning_rate": 9.803787516237096e-06,
      "loss": 5.4544,
      "step": 153500
    },
    {
      "epoch": 2.1,
      "learning_rate": 9.803103849046285e-06,
      "loss": 5.4498,
      "step": 154000
    },
    {
      "epoch": 2.11,
      "learning_rate": 9.802420181855473e-06,
      "loss": 5.4499,
      "step": 154500
    },
    {
      "epoch": 2.12,
      "learning_rate": 9.801736514664662e-06,
      "loss": 5.4336,
      "step": 155000
    },
    {
      "epoch": 2.12,
      "learning_rate": 9.80105284747385e-06,
      "loss": 5.4305,
      "step": 155500
    },
    {
      "epoch": 2.13,
      "learning_rate": 9.800369180283038e-06,
      "loss": 5.4256,
      "step": 156000
    },
    {
      "epoch": 2.14,
      "learning_rate": 9.799685513092228e-06,
      "loss": 5.4226,
      "step": 156500
    },
    {
      "epoch": 2.14,
      "learning_rate": 9.799001845901416e-06,
      "loss": 5.4176,
      "step": 157000
    },
    {
      "epoch": 2.15,
      "learning_rate": 9.798318178710604e-06,
      "loss": 5.4114,
      "step": 157500
    },
    {
      "epoch": 2.16,
      "learning_rate": 9.797634511519794e-06,
      "loss": 5.4036,
      "step": 158000
    },
    {
      "epoch": 2.16,
      "learning_rate": 9.796950844328982e-06,
      "loss": 5.4011,
      "step": 158500
    },
    {
      "epoch": 2.17,
      "learning_rate": 9.79626717713817e-06,
      "loss": 5.3903,
      "step": 159000
    },
    {
      "epoch": 2.18,
      "learning_rate": 9.795583509947358e-06,
      "loss": 5.3841,
      "step": 159500
    },
    {
      "epoch": 2.18,
      "learning_rate": 9.794899842756546e-06,
      "loss": 5.3809,
      "step": 160000
    },
    {
      "epoch": 2.19,
      "learning_rate": 9.794216175565734e-06,
      "loss": 5.3761,
      "step": 160500
    },
    {
      "epoch": 2.2,
      "learning_rate": 9.793532508374924e-06,
      "loss": 5.3604,
      "step": 161000
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.792848841184112e-06,
      "loss": 5.3578,
      "step": 161500
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.792165173993302e-06,
      "loss": 5.3522,
      "step": 162000
    },
    {
      "epoch": 2.22,
      "learning_rate": 9.79148150680249e-06,
      "loss": 5.3483,
      "step": 162500
    },
    {
      "epoch": 2.23,
      "learning_rate": 9.790797839611678e-06,
      "loss": 5.3422,
      "step": 163000
    },
    {
      "epoch": 2.23,
      "learning_rate": 9.790114172420866e-06,
      "loss": 5.3314,
      "step": 163500
    },
    {
      "epoch": 2.24,
      "learning_rate": 9.789430505230056e-06,
      "loss": 5.3233,
      "step": 164000
    },
    {
      "epoch": 2.25,
      "learning_rate": 9.788746838039244e-06,
      "loss": 5.3163,
      "step": 164500
    },
    {
      "epoch": 2.25,
      "learning_rate": 9.788063170848432e-06,
      "loss": 5.3056,
      "step": 165000
    },
    {
      "epoch": 2.26,
      "learning_rate": 9.78737950365762e-06,
      "loss": 5.3054,
      "step": 165500
    },
    {
      "epoch": 2.27,
      "learning_rate": 9.786695836466808e-06,
      "loss": 5.2922,
      "step": 166000
    },
    {
      "epoch": 2.27,
      "learning_rate": 9.786012169275998e-06,
      "loss": 5.2893,
      "step": 166500
    },
    {
      "epoch": 2.28,
      "learning_rate": 9.785328502085186e-06,
      "loss": 5.2822,
      "step": 167000
    },
    {
      "epoch": 2.29,
      "learning_rate": 9.784644834894374e-06,
      "loss": 5.2734,
      "step": 167500
    },
    {
      "epoch": 2.29,
      "learning_rate": 9.783961167703563e-06,
      "loss": 5.2629,
      "step": 168000
    },
    {
      "epoch": 2.3,
      "learning_rate": 9.783277500512752e-06,
      "loss": 5.2583,
      "step": 168500
    },
    {
      "epoch": 2.31,
      "learning_rate": 9.78259383332194e-06,
      "loss": 5.2534,
      "step": 169000
    },
    {
      "epoch": 2.31,
      "learning_rate": 9.781910166131128e-06,
      "loss": 5.2404,
      "step": 169500
    },
    {
      "epoch": 2.32,
      "learning_rate": 9.781226498940317e-06,
      "loss": 5.2284,
      "step": 170000
    },
    {
      "epoch": 2.33,
      "learning_rate": 9.780542831749505e-06,
      "loss": 5.2255,
      "step": 170500
    },
    {
      "epoch": 2.33,
      "learning_rate": 9.779859164558693e-06,
      "loss": 5.2181,
      "step": 171000
    },
    {
      "epoch": 2.34,
      "learning_rate": 9.779175497367882e-06,
      "loss": 5.2164,
      "step": 171500
    },
    {
      "epoch": 2.35,
      "learning_rate": 9.77849183017707e-06,
      "loss": 5.2012,
      "step": 172000
    },
    {
      "epoch": 2.36,
      "learning_rate": 9.77780816298626e-06,
      "loss": 5.1913,
      "step": 172500
    },
    {
      "epoch": 2.36,
      "learning_rate": 9.777124495795447e-06,
      "loss": 5.1779,
      "step": 173000
    },
    {
      "epoch": 2.37,
      "learning_rate": 9.776440828604637e-06,
      "loss": 5.1783,
      "step": 173500
    },
    {
      "epoch": 2.38,
      "learning_rate": 9.775757161413825e-06,
      "loss": 5.1676,
      "step": 174000
    },
    {
      "epoch": 2.38,
      "learning_rate": 9.775073494223013e-06,
      "loss": 5.1562,
      "step": 174500
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.774389827032201e-06,
      "loss": 5.1464,
      "step": 175000
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.77370615984139e-06,
      "loss": 5.1406,
      "step": 175500
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.773022492650579e-06,
      "loss": 5.1324,
      "step": 176000
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.772338825459767e-06,
      "loss": 5.1245,
      "step": 176500
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.771655158268955e-06,
      "loss": 5.111,
      "step": 177000
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.770971491078143e-06,
      "loss": 5.1032,
      "step": 177500
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.770287823887333e-06,
      "loss": 5.0954,
      "step": 178000
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.769604156696521e-06,
      "loss": 5.0877,
      "step": 178500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.76892048950571e-06,
      "loss": 5.0766,
      "step": 179000
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.768236822314899e-06,
      "loss": 5.0635,
      "step": 179500
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.767553155124087e-06,
      "loss": 5.0623,
      "step": 180000
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.766869487933275e-06,
      "loss": 5.0469,
      "step": 180500
    },
    {
      "epoch": 2.47,
      "learning_rate": 9.766185820742463e-06,
      "loss": 5.0407,
      "step": 181000
    },
    {
      "epoch": 2.48,
      "learning_rate": 9.765502153551651e-06,
      "loss": 5.035,
      "step": 181500
    },
    {
      "epoch": 2.49,
      "learning_rate": 9.76481848636084e-06,
      "loss": 5.0213,
      "step": 182000
    },
    {
      "epoch": 2.49,
      "learning_rate": 9.764134819170029e-06,
      "loss": 5.0149,
      "step": 182500
    },
    {
      "epoch": 2.5,
      "learning_rate": 9.763451151979217e-06,
      "loss": 5.0034,
      "step": 183000
    },
    {
      "epoch": 2.51,
      "learning_rate": 9.762767484788407e-06,
      "loss": 4.9937,
      "step": 183500
    },
    {
      "epoch": 2.51,
      "learning_rate": 9.762083817597595e-06,
      "loss": 4.987,
      "step": 184000
    },
    {
      "epoch": 2.52,
      "learning_rate": 9.761400150406783e-06,
      "loss": 4.9696,
      "step": 184500
    },
    {
      "epoch": 2.53,
      "learning_rate": 9.760716483215972e-06,
      "loss": 4.9591,
      "step": 185000
    },
    {
      "epoch": 2.53,
      "learning_rate": 9.76003281602516e-06,
      "loss": 4.9649,
      "step": 185500
    },
    {
      "epoch": 2.54,
      "learning_rate": 9.759349148834348e-06,
      "loss": 4.9479,
      "step": 186000
    },
    {
      "epoch": 2.55,
      "learning_rate": 9.758665481643537e-06,
      "loss": 4.9324,
      "step": 186500
    },
    {
      "epoch": 2.55,
      "learning_rate": 9.757981814452725e-06,
      "loss": 4.9248,
      "step": 187000
    },
    {
      "epoch": 2.56,
      "learning_rate": 9.757298147261913e-06,
      "loss": 4.9161,
      "step": 187500
    },
    {
      "epoch": 2.57,
      "learning_rate": 9.756614480071102e-06,
      "loss": 4.9119,
      "step": 188000
    },
    {
      "epoch": 2.57,
      "learning_rate": 9.75593081288029e-06,
      "loss": 4.8988,
      "step": 188500
    },
    {
      "epoch": 2.58,
      "learning_rate": 9.755247145689478e-06,
      "loss": 4.8902,
      "step": 189000
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.754563478498668e-06,
      "loss": 4.8801,
      "step": 189500
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.753879811307856e-06,
      "loss": 4.8741,
      "step": 190000
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.753196144117044e-06,
      "loss": 4.8587,
      "step": 190500
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.752512476926234e-06,
      "loss": 4.8536,
      "step": 191000
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.751828809735422e-06,
      "loss": 4.8431,
      "step": 191500
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.75114514254461e-06,
      "loss": 4.8352,
      "step": 192000
    },
    {
      "epoch": 2.63,
      "learning_rate": 9.750461475353798e-06,
      "loss": 4.8329,
      "step": 192500
    },
    {
      "epoch": 2.64,
      "learning_rate": 9.749777808162986e-06,
      "loss": 4.814,
      "step": 193000
    },
    {
      "epoch": 2.64,
      "learning_rate": 9.749094140972174e-06,
      "loss": 4.7995,
      "step": 193500
    },
    {
      "epoch": 2.65,
      "learning_rate": 9.748410473781364e-06,
      "loss": 4.7996,
      "step": 194000
    },
    {
      "epoch": 2.66,
      "learning_rate": 9.747726806590552e-06,
      "loss": 4.7841,
      "step": 194500
    },
    {
      "epoch": 2.66,
      "learning_rate": 9.747043139399742e-06,
      "loss": 4.7709,
      "step": 195000
    },
    {
      "epoch": 2.67,
      "learning_rate": 9.74635947220893e-06,
      "loss": 4.7714,
      "step": 195500
    },
    {
      "epoch": 2.68,
      "learning_rate": 9.745675805018118e-06,
      "loss": 4.7585,
      "step": 196000
    },
    {
      "epoch": 2.68,
      "learning_rate": 9.744992137827306e-06,
      "loss": 4.7531,
      "step": 196500
    },
    {
      "epoch": 2.69,
      "learning_rate": 9.744308470636496e-06,
      "loss": 4.7437,
      "step": 197000
    },
    {
      "epoch": 2.7,
      "learning_rate": 9.743624803445684e-06,
      "loss": 4.7318,
      "step": 197500
    },
    {
      "epoch": 2.7,
      "learning_rate": 9.742941136254872e-06,
      "loss": 4.7264,
      "step": 198000
    },
    {
      "epoch": 2.71,
      "learning_rate": 9.74225746906406e-06,
      "loss": 4.7121,
      "step": 198500
    },
    {
      "epoch": 2.72,
      "learning_rate": 9.741573801873248e-06,
      "loss": 4.7024,
      "step": 199000
    },
    {
      "epoch": 2.72,
      "learning_rate": 9.740890134682438e-06,
      "loss": 4.7017,
      "step": 199500
    },
    {
      "epoch": 2.73,
      "learning_rate": 9.740206467491626e-06,
      "loss": 4.6881,
      "step": 200000
    },
    {
      "epoch": 2.74,
      "learning_rate": 9.739522800300815e-06,
      "loss": 4.6773,
      "step": 200500
    },
    {
      "epoch": 2.74,
      "learning_rate": 9.738839133110004e-06,
      "loss": 4.6667,
      "step": 201000
    },
    {
      "epoch": 2.75,
      "learning_rate": 9.738155465919192e-06,
      "loss": 4.6611,
      "step": 201500
    },
    {
      "epoch": 2.76,
      "learning_rate": 9.73747179872838e-06,
      "loss": 4.6543,
      "step": 202000
    },
    {
      "epoch": 2.77,
      "learning_rate": 9.736788131537568e-06,
      "loss": 4.6361,
      "step": 202500
    },
    {
      "epoch": 2.77,
      "learning_rate": 9.736104464346757e-06,
      "loss": 4.6358,
      "step": 203000
    },
    {
      "epoch": 2.78,
      "learning_rate": 9.735420797155945e-06,
      "loss": 4.6277,
      "step": 203500
    },
    {
      "epoch": 2.79,
      "learning_rate": 9.734737129965134e-06,
      "loss": 4.6078,
      "step": 204000
    },
    {
      "epoch": 2.79,
      "learning_rate": 9.734053462774322e-06,
      "loss": 4.6072,
      "step": 204500
    },
    {
      "epoch": 2.8,
      "learning_rate": 9.73336979558351e-06,
      "loss": 4.5963,
      "step": 205000
    },
    {
      "epoch": 2.81,
      "learning_rate": 9.7326861283927e-06,
      "loss": 4.5914,
      "step": 205500
    },
    {
      "epoch": 2.81,
      "learning_rate": 9.732002461201887e-06,
      "loss": 4.5756,
      "step": 206000
    },
    {
      "epoch": 2.82,
      "learning_rate": 9.731318794011077e-06,
      "loss": 4.5767,
      "step": 206500
    },
    {
      "epoch": 2.83,
      "learning_rate": 9.730635126820265e-06,
      "loss": 4.5702,
      "step": 207000
    },
    {
      "epoch": 2.83,
      "learning_rate": 9.729951459629453e-06,
      "loss": 4.5584,
      "step": 207500
    },
    {
      "epoch": 2.84,
      "learning_rate": 9.729267792438641e-06,
      "loss": 4.5453,
      "step": 208000
    },
    {
      "epoch": 2.85,
      "learning_rate": 9.72858412524783e-06,
      "loss": 4.542,
      "step": 208500
    },
    {
      "epoch": 2.85,
      "learning_rate": 9.727900458057019e-06,
      "loss": 4.528,
      "step": 209000
    },
    {
      "epoch": 2.86,
      "learning_rate": 9.727216790866207e-06,
      "loss": 4.5251,
      "step": 209500
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.726533123675395e-06,
      "loss": 4.5135,
      "step": 210000
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.725849456484583e-06,
      "loss": 4.5113,
      "step": 210500
    },
    {
      "epoch": 2.88,
      "learning_rate": 9.725165789293773e-06,
      "loss": 4.4909,
      "step": 211000
    },
    {
      "epoch": 2.89,
      "learning_rate": 9.724482122102961e-06,
      "loss": 4.4943,
      "step": 211500
    },
    {
      "epoch": 2.89,
      "learning_rate": 9.72379845491215e-06,
      "loss": 4.4798,
      "step": 212000
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.723114787721339e-06,
      "loss": 4.4761,
      "step": 212500
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.722431120530527e-06,
      "loss": 4.4651,
      "step": 213000
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.721747453339715e-06,
      "loss": 4.4564,
      "step": 213500
    },
    {
      "epoch": 2.92,
      "learning_rate": 9.721063786148903e-06,
      "loss": 4.4532,
      "step": 214000
    },
    {
      "epoch": 2.93,
      "learning_rate": 9.720380118958091e-06,
      "loss": 4.4456,
      "step": 214500
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.71969645176728e-06,
      "loss": 4.4338,
      "step": 215000
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.719012784576469e-06,
      "loss": 4.4347,
      "step": 215500
    },
    {
      "epoch": 2.95,
      "learning_rate": 9.718329117385657e-06,
      "loss": 4.4247,
      "step": 216000
    },
    {
      "epoch": 2.96,
      "learning_rate": 9.717645450194847e-06,
      "loss": 4.4179,
      "step": 216500
    },
    {
      "epoch": 2.96,
      "learning_rate": 9.716961783004035e-06,
      "loss": 4.4076,
      "step": 217000
    },
    {
      "epoch": 2.97,
      "learning_rate": 9.716278115813223e-06,
      "loss": 4.4,
      "step": 217500
    },
    {
      "epoch": 2.98,
      "learning_rate": 9.715594448622412e-06,
      "loss": 4.3962,
      "step": 218000
    },
    {
      "epoch": 2.98,
      "learning_rate": 9.7149107814316e-06,
      "loss": 4.3872,
      "step": 218500
    },
    {
      "epoch": 2.99,
      "learning_rate": 9.714227114240789e-06,
      "loss": 4.3838,
      "step": 219000
    },
    {
      "epoch": 3.0,
      "learning_rate": 9.713543447049977e-06,
      "loss": 4.3698,
      "step": 219500
    },
    {
      "epoch": 3.0,
      "learning_rate": 9.712859779859165e-06,
      "loss": 4.364,
      "step": 220000
    },
    {
      "epoch": 3.01,
      "learning_rate": 9.712176112668353e-06,
      "loss": 4.3682,
      "step": 220500
    },
    {
      "epoch": 3.02,
      "learning_rate": 9.711492445477542e-06,
      "loss": 4.3537,
      "step": 221000
    },
    {
      "epoch": 3.02,
      "learning_rate": 9.71080877828673e-06,
      "loss": 4.3459,
      "step": 221500
    },
    {
      "epoch": 3.03,
      "learning_rate": 9.710125111095919e-06,
      "loss": 4.3442,
      "step": 222000
    },
    {
      "epoch": 3.04,
      "learning_rate": 9.709441443905108e-06,
      "loss": 4.3352,
      "step": 222500
    },
    {
      "epoch": 3.04,
      "learning_rate": 9.708757776714296e-06,
      "loss": 4.3311,
      "step": 223000
    },
    {
      "epoch": 3.05,
      "learning_rate": 9.708074109523484e-06,
      "loss": 4.3182,
      "step": 223500
    },
    {
      "epoch": 3.06,
      "learning_rate": 9.707390442332674e-06,
      "loss": 4.3108,
      "step": 224000
    },
    {
      "epoch": 3.07,
      "learning_rate": 9.706706775141862e-06,
      "loss": 4.3074,
      "step": 224500
    },
    {
      "epoch": 3.07,
      "learning_rate": 9.70602310795105e-06,
      "loss": 4.3053,
      "step": 225000
    },
    {
      "epoch": 3.08,
      "learning_rate": 9.705339440760238e-06,
      "loss": 4.294,
      "step": 225500
    },
    {
      "epoch": 3.09,
      "learning_rate": 9.704655773569426e-06,
      "loss": 4.2862,
      "step": 226000
    },
    {
      "epoch": 3.09,
      "learning_rate": 9.703972106378614e-06,
      "loss": 4.2796,
      "step": 226500
    },
    {
      "epoch": 3.1,
      "learning_rate": 9.703288439187804e-06,
      "loss": 4.2754,
      "step": 227000
    },
    {
      "epoch": 3.11,
      "learning_rate": 9.702604771996992e-06,
      "loss": 4.2725,
      "step": 227500
    },
    {
      "epoch": 3.11,
      "learning_rate": 9.701921104806182e-06,
      "loss": 4.2657,
      "step": 228000
    },
    {
      "epoch": 3.12,
      "learning_rate": 9.70123743761537e-06,
      "loss": 4.2615,
      "step": 228500
    },
    {
      "epoch": 3.13,
      "learning_rate": 9.700553770424558e-06,
      "loss": 4.2557,
      "step": 229000
    },
    {
      "epoch": 3.13,
      "learning_rate": 9.699870103233746e-06,
      "loss": 4.2414,
      "step": 229500
    },
    {
      "epoch": 3.14,
      "learning_rate": 9.699186436042936e-06,
      "loss": 4.2314,
      "step": 230000
    },
    {
      "epoch": 3.15,
      "learning_rate": 9.698502768852124e-06,
      "loss": 4.2325,
      "step": 230500
    },
    {
      "epoch": 3.15,
      "learning_rate": 9.697819101661312e-06,
      "loss": 4.2295,
      "step": 231000
    },
    {
      "epoch": 3.16,
      "learning_rate": 9.6971354344705e-06,
      "loss": 4.2268,
      "step": 231500
    },
    {
      "epoch": 3.17,
      "learning_rate": 9.696451767279688e-06,
      "loss": 4.2214,
      "step": 232000
    },
    {
      "epoch": 3.17,
      "learning_rate": 9.695768100088878e-06,
      "loss": 4.2133,
      "step": 232500
    },
    {
      "epoch": 3.18,
      "learning_rate": 9.695084432898066e-06,
      "loss": 4.2071,
      "step": 233000
    },
    {
      "epoch": 3.19,
      "learning_rate": 9.694400765707256e-06,
      "loss": 4.2031,
      "step": 233500
    },
    {
      "epoch": 3.2,
      "learning_rate": 9.693717098516444e-06,
      "loss": 4.1969,
      "step": 234000
    },
    {
      "epoch": 3.2,
      "learning_rate": 9.693033431325632e-06,
      "loss": 4.1932,
      "step": 234500
    },
    {
      "epoch": 3.21,
      "learning_rate": 9.69234976413482e-06,
      "loss": 4.1839,
      "step": 235000
    },
    {
      "epoch": 3.22,
      "learning_rate": 9.691666096944008e-06,
      "loss": 4.1797,
      "step": 235500
    },
    {
      "epoch": 3.22,
      "learning_rate": 9.690982429753198e-06,
      "loss": 4.1727,
      "step": 236000
    },
    {
      "epoch": 3.23,
      "learning_rate": 9.690298762562386e-06,
      "loss": 4.1645,
      "step": 236500
    },
    {
      "epoch": 3.24,
      "learning_rate": 9.689615095371574e-06,
      "loss": 4.1612,
      "step": 237000
    },
    {
      "epoch": 3.24,
      "learning_rate": 9.688931428180762e-06,
      "loss": 4.1605,
      "step": 237500
    },
    {
      "epoch": 3.25,
      "learning_rate": 9.68824776098995e-06,
      "loss": 4.1531,
      "step": 238000
    },
    {
      "epoch": 3.26,
      "learning_rate": 9.68756409379914e-06,
      "loss": 4.1487,
      "step": 238500
    },
    {
      "epoch": 3.26,
      "learning_rate": 9.686880426608328e-06,
      "loss": 4.1448,
      "step": 239000
    },
    {
      "epoch": 3.27,
      "learning_rate": 9.686196759417517e-06,
      "loss": 4.1333,
      "step": 239500
    },
    {
      "epoch": 3.28,
      "learning_rate": 9.685513092226705e-06,
      "loss": 4.1319,
      "step": 240000
    },
    {
      "epoch": 3.28,
      "learning_rate": 9.684829425035893e-06,
      "loss": 4.1274,
      "step": 240500
    },
    {
      "epoch": 3.29,
      "learning_rate": 9.684145757845081e-06,
      "loss": 4.1209,
      "step": 241000
    },
    {
      "epoch": 3.3,
      "learning_rate": 9.68346209065427e-06,
      "loss": 4.1169,
      "step": 241500
    },
    {
      "epoch": 3.3,
      "learning_rate": 9.68277842346346e-06,
      "loss": 4.1097,
      "step": 242000
    },
    {
      "epoch": 3.31,
      "learning_rate": 9.682094756272647e-06,
      "loss": 4.1047,
      "step": 242500
    },
    {
      "epoch": 3.32,
      "learning_rate": 9.681411089081835e-06,
      "loss": 4.1056,
      "step": 243000
    },
    {
      "epoch": 3.32,
      "learning_rate": 9.680727421891023e-06,
      "loss": 4.0931,
      "step": 243500
    },
    {
      "epoch": 3.33,
      "learning_rate": 9.680043754700213e-06,
      "loss": 4.0962,
      "step": 244000
    },
    {
      "epoch": 3.34,
      "learning_rate": 9.679360087509401e-06,
      "loss": 4.0919,
      "step": 244500
    },
    {
      "epoch": 3.35,
      "learning_rate": 9.678676420318591e-06,
      "loss": 4.0818,
      "step": 245000
    },
    {
      "epoch": 3.35,
      "learning_rate": 9.677992753127779e-06,
      "loss": 4.0743,
      "step": 245500
    },
    {
      "epoch": 3.36,
      "learning_rate": 9.677309085936967e-06,
      "loss": 4.0739,
      "step": 246000
    },
    {
      "epoch": 3.37,
      "learning_rate": 9.676625418746155e-06,
      "loss": 4.0648,
      "step": 246500
    },
    {
      "epoch": 3.37,
      "learning_rate": 9.675941751555343e-06,
      "loss": 4.0663,
      "step": 247000
    },
    {
      "epoch": 3.38,
      "learning_rate": 9.675258084364531e-06,
      "loss": 4.0574,
      "step": 247500
    },
    {
      "epoch": 3.39,
      "learning_rate": 9.674574417173721e-06,
      "loss": 4.0498,
      "step": 248000
    },
    {
      "epoch": 3.39,
      "learning_rate": 9.673890749982909e-06,
      "loss": 4.0531,
      "step": 248500
    },
    {
      "epoch": 3.4,
      "learning_rate": 9.673207082792097e-06,
      "loss": 4.0533,
      "step": 249000
    },
    {
      "epoch": 3.41,
      "learning_rate": 9.672523415601287e-06,
      "loss": 4.0436,
      "step": 249500
    },
    {
      "epoch": 3.41,
      "learning_rate": 9.671839748410475e-06,
      "loss": 4.0352,
      "step": 250000
    },
    {
      "epoch": 3.42,
      "learning_rate": 9.671156081219663e-06,
      "loss": 4.0404,
      "step": 250500
    },
    {
      "epoch": 3.43,
      "learning_rate": 9.670472414028853e-06,
      "loss": 4.0255,
      "step": 251000
    },
    {
      "epoch": 3.43,
      "learning_rate": 9.66978874683804e-06,
      "loss": 4.0258,
      "step": 251500
    },
    {
      "epoch": 3.44,
      "learning_rate": 9.669105079647229e-06,
      "loss": 4.0224,
      "step": 252000
    },
    {
      "epoch": 3.45,
      "learning_rate": 9.668421412456417e-06,
      "loss": 4.0209,
      "step": 252500
    },
    {
      "epoch": 3.45,
      "learning_rate": 9.667737745265605e-06,
      "loss": 4.0128,
      "step": 253000
    },
    {
      "epoch": 3.46,
      "learning_rate": 9.667054078074793e-06,
      "loss": 4.0086,
      "step": 253500
    },
    {
      "epoch": 3.47,
      "learning_rate": 9.666370410883983e-06,
      "loss": 4.0056,
      "step": 254000
    },
    {
      "epoch": 3.48,
      "learning_rate": 9.66568674369317e-06,
      "loss": 4.0003,
      "step": 254500
    },
    {
      "epoch": 3.48,
      "learning_rate": 9.665003076502359e-06,
      "loss": 3.9991,
      "step": 255000
    },
    {
      "epoch": 3.49,
      "learning_rate": 9.664319409311548e-06,
      "loss": 3.9966,
      "step": 255500
    },
    {
      "epoch": 3.5,
      "learning_rate": 9.663635742120736e-06,
      "loss": 3.9876,
      "step": 256000
    },
    {
      "epoch": 3.5,
      "learning_rate": 9.662952074929925e-06,
      "loss": 3.9887,
      "step": 256500
    },
    {
      "epoch": 3.51,
      "learning_rate": 9.662268407739114e-06,
      "loss": 3.9789,
      "step": 257000
    },
    {
      "epoch": 3.52,
      "learning_rate": 9.661584740548302e-06,
      "loss": 3.9782,
      "step": 257500
    },
    {
      "epoch": 3.52,
      "learning_rate": 9.66090107335749e-06,
      "loss": 3.9779,
      "step": 258000
    },
    {
      "epoch": 3.53,
      "learning_rate": 9.660217406166678e-06,
      "loss": 3.9723,
      "step": 258500
    },
    {
      "epoch": 3.54,
      "learning_rate": 9.659533738975866e-06,
      "loss": 3.9663,
      "step": 259000
    },
    {
      "epoch": 3.54,
      "learning_rate": 9.658850071785054e-06,
      "loss": 3.9648,
      "step": 259500
    },
    {
      "epoch": 3.55,
      "learning_rate": 9.658166404594244e-06,
      "loss": 3.9545,
      "step": 260000
    },
    {
      "epoch": 3.56,
      "learning_rate": 9.657482737403432e-06,
      "loss": 3.9504,
      "step": 260500
    },
    {
      "epoch": 3.56,
      "learning_rate": 9.656799070212622e-06,
      "loss": 3.9559,
      "step": 261000
    },
    {
      "epoch": 3.57,
      "learning_rate": 9.65611540302181e-06,
      "loss": 3.9409,
      "step": 261500
    },
    {
      "epoch": 3.58,
      "learning_rate": 9.655431735830998e-06,
      "loss": 3.9424,
      "step": 262000
    },
    {
      "epoch": 3.58,
      "learning_rate": 9.654748068640186e-06,
      "loss": 3.9376,
      "step": 262500
    },
    {
      "epoch": 3.59,
      "learning_rate": 9.654064401449376e-06,
      "loss": 3.9339,
      "step": 263000
    },
    {
      "epoch": 3.6,
      "learning_rate": 9.653380734258564e-06,
      "loss": 3.9297,
      "step": 263500
    },
    {
      "epoch": 3.6,
      "learning_rate": 9.652697067067752e-06,
      "loss": 3.9336,
      "step": 264000
    },
    {
      "epoch": 3.61,
      "learning_rate": 9.65201339987694e-06,
      "loss": 3.9259,
      "step": 264500
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.651329732686128e-06,
      "loss": 3.9252,
      "step": 265000
    },
    {
      "epoch": 3.63,
      "learning_rate": 9.650646065495318e-06,
      "loss": 3.9139,
      "step": 265500
    },
    {
      "epoch": 3.63,
      "learning_rate": 9.649962398304506e-06,
      "loss": 3.9229,
      "step": 266000
    },
    {
      "epoch": 3.64,
      "learning_rate": 9.649278731113696e-06,
      "loss": 3.9092,
      "step": 266500
    },
    {
      "epoch": 3.65,
      "learning_rate": 9.648595063922884e-06,
      "loss": 3.9111,
      "step": 267000
    },
    {
      "epoch": 3.65,
      "learning_rate": 9.647911396732072e-06,
      "loss": 3.9084,
      "step": 267500
    },
    {
      "epoch": 3.66,
      "learning_rate": 9.64722772954126e-06,
      "loss": 3.8996,
      "step": 268000
    },
    {
      "epoch": 3.67,
      "learning_rate": 9.646544062350448e-06,
      "loss": 3.9002,
      "step": 268500
    },
    {
      "epoch": 3.67,
      "learning_rate": 9.645860395159638e-06,
      "loss": 3.8964,
      "step": 269000
    },
    {
      "epoch": 3.68,
      "learning_rate": 9.645176727968826e-06,
      "loss": 3.8946,
      "step": 269500
    },
    {
      "epoch": 3.69,
      "learning_rate": 9.644493060778014e-06,
      "loss": 3.888,
      "step": 270000
    },
    {
      "epoch": 3.69,
      "learning_rate": 9.643809393587202e-06,
      "loss": 3.8861,
      "step": 270500
    },
    {
      "epoch": 3.7,
      "learning_rate": 9.643125726396392e-06,
      "loss": 3.8864,
      "step": 271000
    },
    {
      "epoch": 3.71,
      "learning_rate": 9.64244205920558e-06,
      "loss": 3.8757,
      "step": 271500
    },
    {
      "epoch": 3.71,
      "learning_rate": 9.641758392014768e-06,
      "loss": 3.8774,
      "step": 272000
    },
    {
      "epoch": 3.72,
      "learning_rate": 9.641074724823957e-06,
      "loss": 3.8705,
      "step": 272500
    },
    {
      "epoch": 3.73,
      "learning_rate": 9.640391057633145e-06,
      "loss": 3.8759,
      "step": 273000
    },
    {
      "epoch": 3.73,
      "learning_rate": 9.639707390442333e-06,
      "loss": 3.8685,
      "step": 273500
    },
    {
      "epoch": 3.74,
      "learning_rate": 9.639023723251521e-06,
      "loss": 3.8645,
      "step": 274000
    },
    {
      "epoch": 3.75,
      "learning_rate": 9.63834005606071e-06,
      "loss": 3.8649,
      "step": 274500
    },
    {
      "epoch": 3.76,
      "learning_rate": 9.6376563888699e-06,
      "loss": 3.8557,
      "step": 275000
    },
    {
      "epoch": 3.76,
      "learning_rate": 9.636972721679087e-06,
      "loss": 3.8595,
      "step": 275500
    },
    {
      "epoch": 3.77,
      "learning_rate": 9.636289054488275e-06,
      "loss": 3.8495,
      "step": 276000
    },
    {
      "epoch": 3.78,
      "learning_rate": 9.635605387297463e-06,
      "loss": 3.8462,
      "step": 276500
    },
    {
      "epoch": 3.78,
      "learning_rate": 9.634921720106653e-06,
      "loss": 3.8496,
      "step": 277000
    },
    {
      "epoch": 3.79,
      "learning_rate": 9.634238052915841e-06,
      "loss": 3.8508,
      "step": 277500
    },
    {
      "epoch": 3.8,
      "learning_rate": 9.633554385725031e-06,
      "loss": 3.8317,
      "step": 278000
    },
    {
      "epoch": 3.8,
      "learning_rate": 9.632870718534219e-06,
      "loss": 3.847,
      "step": 278500
    },
    {
      "epoch": 3.81,
      "learning_rate": 9.632187051343407e-06,
      "loss": 3.8339,
      "step": 279000
    },
    {
      "epoch": 3.82,
      "learning_rate": 9.631503384152595e-06,
      "loss": 3.8371,
      "step": 279500
    },
    {
      "epoch": 3.82,
      "learning_rate": 9.630819716961783e-06,
      "loss": 3.8284,
      "step": 280000
    },
    {
      "epoch": 3.83,
      "learning_rate": 9.630136049770973e-06,
      "loss": 3.8198,
      "step": 280500
    },
    {
      "epoch": 3.84,
      "learning_rate": 9.629452382580161e-06,
      "loss": 3.8166,
      "step": 281000
    },
    {
      "epoch": 3.84,
      "learning_rate": 9.628768715389349e-06,
      "loss": 3.8156,
      "step": 281500
    },
    {
      "epoch": 3.85,
      "learning_rate": 9.628085048198537e-06,
      "loss": 3.8178,
      "step": 282000
    },
    {
      "epoch": 3.86,
      "learning_rate": 9.627401381007727e-06,
      "loss": 3.819,
      "step": 282500
    },
    {
      "epoch": 3.86,
      "learning_rate": 9.626717713816915e-06,
      "loss": 3.8155,
      "step": 283000
    },
    {
      "epoch": 3.87,
      "learning_rate": 9.626034046626103e-06,
      "loss": 3.8127,
      "step": 283500
    },
    {
      "epoch": 3.88,
      "learning_rate": 9.625350379435293e-06,
      "loss": 3.8059,
      "step": 284000
    },
    {
      "epoch": 3.88,
      "learning_rate": 9.62466671224448e-06,
      "loss": 3.7996,
      "step": 284500
    },
    {
      "epoch": 3.89,
      "learning_rate": 9.623983045053669e-06,
      "loss": 3.8023,
      "step": 285000
    },
    {
      "epoch": 3.9,
      "learning_rate": 9.623299377862857e-06,
      "loss": 3.7941,
      "step": 285500
    },
    {
      "epoch": 3.91,
      "learning_rate": 9.622615710672045e-06,
      "loss": 3.7924,
      "step": 286000
    },
    {
      "epoch": 3.91,
      "learning_rate": 9.621932043481235e-06,
      "loss": 3.7875,
      "step": 286500
    },
    {
      "epoch": 3.92,
      "learning_rate": 9.621248376290423e-06,
      "loss": 3.788,
      "step": 287000
    },
    {
      "epoch": 3.93,
      "learning_rate": 9.62056470909961e-06,
      "loss": 3.7897,
      "step": 287500
    },
    {
      "epoch": 3.93,
      "learning_rate": 9.619881041908799e-06,
      "loss": 3.7887,
      "step": 288000
    },
    {
      "epoch": 3.94,
      "learning_rate": 9.619197374717988e-06,
      "loss": 3.7784,
      "step": 288500
    },
    {
      "epoch": 3.95,
      "learning_rate": 9.618513707527177e-06,
      "loss": 3.7826,
      "step": 289000
    },
    {
      "epoch": 3.95,
      "learning_rate": 9.617830040336366e-06,
      "loss": 3.778,
      "step": 289500
    },
    {
      "epoch": 3.96,
      "learning_rate": 9.617146373145554e-06,
      "loss": 3.7705,
      "step": 290000
    },
    {
      "epoch": 3.97,
      "learning_rate": 9.616462705954742e-06,
      "loss": 3.7723,
      "step": 290500
    },
    {
      "epoch": 3.97,
      "learning_rate": 9.61577903876393e-06,
      "loss": 3.7705,
      "step": 291000
    },
    {
      "epoch": 3.98,
      "step": 291280,
      "total_flos": 2.0552858723952886e+18,
      "train_loss": 5.310745644091381,
      "train_runtime": 197999.4599,
      "train_samples_per_second": 2958.977,
      "train_steps_per_second": 36.987
    }
  ],
  "logging_steps": 500,
  "max_steps": 7323500,
  "num_train_epochs": 100,
  "save_steps": 1000000,
  "total_flos": 2.0552858723952886e+18,
  "trial_name": null,
  "trial_params": null
}