{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 76518,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 1.9869311795917304e-05,
      "loss": 1.2439,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9738623591834603e-05,
      "loss": 0.6806,
      "step": 1000
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9607935387751905e-05,
      "loss": 0.6678,
      "step": 1500
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9477247183669204e-05,
      "loss": 0.6594,
      "step": 2000
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9346558979586506e-05,
      "loss": 0.6537,
      "step": 2500
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9215870775503805e-05,
      "loss": 0.6493,
      "step": 3000
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9085182571421107e-05,
      "loss": 0.646,
      "step": 3500
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8954494367338406e-05,
      "loss": 0.6475,
      "step": 4000
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8823806163255708e-05,
      "loss": 0.6358,
      "step": 4500
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8693117959173007e-05,
      "loss": 0.6404,
      "step": 5000
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.856242975509031e-05,
      "loss": 0.636,
      "step": 5500
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.8431741551007608e-05,
      "loss": 0.6307,
      "step": 6000
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.830105334692491e-05,
      "loss": 0.6292,
      "step": 6500
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.817036514284221e-05,
      "loss": 0.6254,
      "step": 7000
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.803967693875951e-05,
      "loss": 0.632,
      "step": 7500
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.790898873467681e-05,
      "loss": 0.6276,
      "step": 8000
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.7778300530594112e-05,
      "loss": 0.6295,
      "step": 8500
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.764761232651141e-05,
      "loss": 0.625,
      "step": 9000
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.751692412242871e-05,
      "loss": 0.6272,
      "step": 9500
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.7386235918346012e-05,
      "loss": 0.6241,
      "step": 10000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.725554771426331e-05,
      "loss": 0.6208,
      "step": 10500
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.7124859510180613e-05,
      "loss": 0.6201,
      "step": 11000
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.699417130609791e-05,
      "loss": 0.6271,
      "step": 11500
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.6863483102015214e-05,
      "loss": 0.6173,
      "step": 12000
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.6732794897932513e-05,
      "loss": 0.6172,
      "step": 12500
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.6602106693849815e-05,
      "loss": 0.6135,
      "step": 13000
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.6471418489767114e-05,
      "loss": 0.6115,
      "step": 13500
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.6340730285684416e-05,
      "loss": 0.6191,
      "step": 14000
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.6210042081601715e-05,
      "loss": 0.6177,
      "step": 14500
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.6079353877519017e-05,
      "loss": 0.6171,
      "step": 15000
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.5948665673436316e-05,
      "loss": 0.6225,
      "step": 15500
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.5817977469353618e-05,
      "loss": 0.6078,
      "step": 16000
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.5687289265270917e-05,
      "loss": 0.6115,
      "step": 16500
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.555660106118822e-05,
      "loss": 0.6085,
      "step": 17000
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.5425912857105518e-05,
      "loss": 0.6011,
      "step": 17500
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.529522465302282e-05,
      "loss": 0.6096,
      "step": 18000
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.5164536448940119e-05,
      "loss": 0.6144,
      "step": 18500
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.5033848244857421e-05,
      "loss": 0.6042,
      "step": 19000
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.490316004077472e-05,
      "loss": 0.6149,
      "step": 19500
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.4772471836692022e-05,
      "loss": 0.6024,
      "step": 20000
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.464178363260932e-05,
      "loss": 0.603,
      "step": 20500
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.4511095428526623e-05,
      "loss": 0.6062,
      "step": 21000
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4380407224443922e-05,
      "loss": 0.6073,
      "step": 21500
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.4249719020361224e-05,
      "loss": 0.6058,
      "step": 22000
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.4119030816278523e-05,
      "loss": 0.6066,
      "step": 22500
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.3988342612195825e-05,
      "loss": 0.6008,
      "step": 23000
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.3857654408113126e-05,
      "loss": 0.622,
      "step": 23500
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.3726966204030424e-05,
      "loss": 0.6013,
      "step": 24000
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.3596277999947727e-05,
      "loss": 0.6073,
      "step": 24500
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.3465589795865025e-05,
      "loss": 0.5973,
      "step": 25000
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.3334901591782328e-05,
      "loss": 0.6065,
      "step": 25500
    },
    {
      "epoch": 1.0,
      "eval_gen_len": 18.7995,
      "eval_loss": 0.547476589679718,
      "eval_rouge1": 34.0929,
      "eval_rouge2": 12.225,
      "eval_rougeL": 27.465,
      "eval_rougeLsum": 27.4736,
      "eval_runtime": 1500.9794,
      "eval_samples_per_second": 7.55,
      "eval_steps_per_second": 1.887,
      "step": 25506
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.3204213387699626e-05,
      "loss": 0.5931,
      "step": 26000
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.3073525183616929e-05,
      "loss": 0.5981,
      "step": 26500
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.2942836979534227e-05,
      "loss": 0.5865,
      "step": 27000
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.281214877545153e-05,
      "loss": 0.5894,
      "step": 27500
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.2681460571368828e-05,
      "loss": 0.5844,
      "step": 28000
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.255077236728613e-05,
      "loss": 0.5928,
      "step": 28500
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.242008416320343e-05,
      "loss": 0.5891,
      "step": 29000
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.2289395959120732e-05,
      "loss": 0.5868,
      "step": 29500
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.215870775503803e-05,
      "loss": 0.5831,
      "step": 30000
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.2028019550955333e-05,
      "loss": 0.5894,
      "step": 30500
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.1897331346872632e-05,
      "loss": 0.5926,
      "step": 31000
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.1766643142789934e-05,
      "loss": 0.5891,
      "step": 31500
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.1635954938707233e-05,
      "loss": 0.5925,
      "step": 32000
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.1505266734624533e-05,
      "loss": 0.582,
      "step": 32500
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.1374578530541834e-05,
      "loss": 0.5904,
      "step": 33000
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.1243890326459134e-05,
      "loss": 0.596,
      "step": 33500
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.1113202122376435e-05,
      "loss": 0.5851,
      "step": 34000
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.0982513918293735e-05,
      "loss": 0.5852,
      "step": 34500
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.0851825714211037e-05,
      "loss": 0.5952,
      "step": 35000
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.0721137510128336e-05,
      "loss": 0.5857,
      "step": 35500
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.0590449306045638e-05,
      "loss": 0.5875,
      "step": 36000
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.0459761101962937e-05,
      "loss": 0.5901,
      "step": 36500
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.032907289788024e-05,
      "loss": 0.5883,
      "step": 37000
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0198384693797538e-05,
      "loss": 0.5893,
      "step": 37500
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.006769648971484e-05,
      "loss": 0.5853,
      "step": 38000
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.93700828563214e-06,
      "loss": 0.5839,
      "step": 38500
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.80632008154944e-06,
      "loss": 0.5877,
      "step": 39000
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.67563187746674e-06,
      "loss": 0.5904,
      "step": 39500
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.54494367338404e-06,
      "loss": 0.5753,
      "step": 40000
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.414255469301343e-06,
      "loss": 0.5855,
      "step": 40500
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.283567265218643e-06,
      "loss": 0.5917,
      "step": 41000
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.152879061135942e-06,
      "loss": 0.5818,
      "step": 41500
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.022190857053243e-06,
      "loss": 0.578,
      "step": 42000
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.891502652970543e-06,
      "loss": 0.5831,
      "step": 42500
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.760814448887844e-06,
      "loss": 0.583,
      "step": 43000
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.630126244805144e-06,
      "loss": 0.5869,
      "step": 43500
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.499438040722445e-06,
      "loss": 0.5802,
      "step": 44000
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.368749836639745e-06,
      "loss": 0.5776,
      "step": 44500
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.238061632557046e-06,
      "loss": 0.5833,
      "step": 45000
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.107373428474346e-06,
      "loss": 0.5844,
      "step": 45500
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.976685224391647e-06,
      "loss": 0.5834,
      "step": 46000
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.845997020308947e-06,
      "loss": 0.5829,
      "step": 46500
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.715308816226248e-06,
      "loss": 0.5849,
      "step": 47000
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.584620612143548e-06,
      "loss": 0.5874,
      "step": 47500
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.453932408060849e-06,
      "loss": 0.5879,
      "step": 48000
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.323244203978149e-06,
      "loss": 0.577,
      "step": 48500
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.19255599989545e-06,
      "loss": 0.5807,
      "step": 49000
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.06186779581275e-06,
      "loss": 0.5865,
      "step": 49500
    },
    {
      "epoch": 1.96,
      "learning_rate": 6.931179591730051e-06,
      "loss": 0.5765,
      "step": 50000
    },
    {
      "epoch": 1.98,
      "learning_rate": 6.8004913876473514e-06,
      "loss": 0.5883,
      "step": 50500
    },
    {
      "epoch": 2.0,
      "learning_rate": 6.669803183564651e-06,
      "loss": 0.5787,
      "step": 51000
    },
    {
      "epoch": 2.0,
      "eval_gen_len": 18.8155,
      "eval_loss": 0.5370269417762756,
      "eval_rouge1": 34.4751,
      "eval_rouge2": 12.6984,
      "eval_rougeL": 27.9605,
      "eval_rougeLsum": 27.9586,
      "eval_runtime": 1516.6364,
      "eval_samples_per_second": 7.472,
      "eval_steps_per_second": 1.868,
      "step": 51012
    },
    {
      "epoch": 2.02,
      "learning_rate": 6.539114979481952e-06,
      "loss": 0.5674,
      "step": 51500
    },
    {
      "epoch": 2.04,
      "learning_rate": 6.408426775399254e-06,
      "loss": 0.5708,
      "step": 52000
    },
    {
      "epoch": 2.06,
      "learning_rate": 6.277738571316554e-06,
      "loss": 0.5746,
      "step": 52500
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.147050367233854e-06,
      "loss": 0.5726,
      "step": 53000
    },
    {
      "epoch": 2.1,
      "learning_rate": 6.0163621631511545e-06,
      "loss": 0.5789,
      "step": 53500
    },
    {
      "epoch": 2.12,
      "learning_rate": 5.885673959068455e-06,
      "loss": 0.5693,
      "step": 54000
    },
    {
      "epoch": 2.14,
      "learning_rate": 5.7549857549857555e-06,
      "loss": 0.5677,
      "step": 54500
    },
    {
      "epoch": 2.16,
      "learning_rate": 5.624297550903056e-06,
      "loss": 0.5724,
      "step": 55000
    },
    {
      "epoch": 2.18,
      "learning_rate": 5.4936093468203565e-06,
      "loss": 0.5737,
      "step": 55500
    },
    {
      "epoch": 2.2,
      "learning_rate": 5.362921142737657e-06,
      "loss": 0.5758,
      "step": 56000
    },
    {
      "epoch": 2.22,
      "learning_rate": 5.2322329386549575e-06,
      "loss": 0.5693,
      "step": 56500
    },
    {
      "epoch": 2.23,
      "learning_rate": 5.101544734572258e-06,
      "loss": 0.5772,
      "step": 57000
    },
    {
      "epoch": 2.25,
      "learning_rate": 4.9708565304895586e-06,
      "loss": 0.5793,
      "step": 57500
    },
    {
      "epoch": 2.27,
      "learning_rate": 4.840168326406859e-06,
      "loss": 0.576,
      "step": 58000
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.70948012232416e-06,
      "loss": 0.5768,
      "step": 58500
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.57879191824146e-06,
      "loss": 0.5685,
      "step": 59000
    },
    {
      "epoch": 2.33,
      "learning_rate": 4.448103714158761e-06,
      "loss": 0.5706,
      "step": 59500
    },
    {
      "epoch": 2.35,
      "learning_rate": 4.31741551007606e-06,
      "loss": 0.5796,
      "step": 60000
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.186727305993362e-06,
      "loss": 0.5763,
      "step": 60500
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.056039101910662e-06,
      "loss": 0.5707,
      "step": 61000
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.925350897827963e-06,
      "loss": 0.5748,
      "step": 61500
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.794662693745263e-06,
      "loss": 0.5762,
      "step": 62000
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.6639744896625632e-06,
      "loss": 0.5693,
      "step": 62500
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.5332862855798637e-06,
      "loss": 0.5754,
      "step": 63000
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.4025980814971643e-06,
      "loss": 0.5832,
      "step": 63500
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.2719098774144648e-06,
      "loss": 0.5653,
      "step": 64000
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.1412216733317653e-06,
      "loss": 0.5635,
      "step": 64500
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.0105334692490658e-06,
      "loss": 0.5743,
      "step": 65000
    },
    {
      "epoch": 2.57,
      "learning_rate": 2.8798452651663663e-06,
      "loss": 0.5633,
      "step": 65500
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.7491570610836672e-06,
      "loss": 0.5741,
      "step": 66000
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.6184688570009677e-06,
      "loss": 0.5691,
      "step": 66500
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.487780652918268e-06,
      "loss": 0.5758,
      "step": 67000
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.3570924488355683e-06,
      "loss": 0.5637,
      "step": 67500
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.226404244752869e-06,
      "loss": 0.5753,
      "step": 68000
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.0957160406701694e-06,
      "loss": 0.5763,
      "step": 68500
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.96502783658747e-06,
      "loss": 0.5738,
      "step": 69000
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.8343396325047702e-06,
      "loss": 0.5698,
      "step": 69500
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.7036514284220707e-06,
      "loss": 0.5725,
      "step": 70000
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.5729632243393714e-06,
      "loss": 0.5733,
      "step": 70500
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.442275020256672e-06,
      "loss": 0.578,
      "step": 71000
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.3115868161739722e-06,
      "loss": 0.5755,
      "step": 71500
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.1808986120912727e-06,
      "loss": 0.5806,
      "step": 72000
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.0502104080085732e-06,
      "loss": 0.5736,
      "step": 72500
    },
    {
      "epoch": 2.86,
      "learning_rate": 9.195222039258737e-07,
      "loss": 0.5665,
      "step": 73000
    },
    {
      "epoch": 2.88,
      "learning_rate": 7.888339998431741e-07,
      "loss": 0.5711,
      "step": 73500
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.581457957604747e-07,
      "loss": 0.5699,
      "step": 74000
    },
    {
      "epoch": 2.92,
      "learning_rate": 5.274575916777753e-07,
      "loss": 0.5727,
      "step": 74500
    },
    {
      "epoch": 2.94,
      "learning_rate": 3.967693875950757e-07,
      "loss": 0.5702,
      "step": 75000
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.660811835123762e-07,
      "loss": 0.5697,
      "step": 75500
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.3539297942967668e-07,
      "loss": 0.5658,
      "step": 76000
    },
    {
      "epoch": 3.0,
      "learning_rate": 4.704775346977182e-09,
      "loss": 0.5692,
      "step": 76500
    },
    {
      "epoch": 3.0,
      "eval_gen_len": 18.8131,
      "eval_loss": 0.5340152382850647,
      "eval_rouge1": 34.6791,
      "eval_rouge2": 12.8236,
      "eval_rougeL": 28.1201,
      "eval_rougeLsum": 28.1241,
      "eval_runtime": 1504.5288,
      "eval_samples_per_second": 7.532,
      "eval_steps_per_second": 1.883,
      "step": 76518
    }
  ],
  "max_steps": 76518,
  "num_train_epochs": 3,
  "total_flos": 3.727644383379456e+17,
  "trial_name": null,
  "trial_params": null
}