{
  "best_metric": 0.5688490867614746,
  "best_model_checkpoint": "german_xlm_xnli/checkpoint-24544",
  "epoch": 3.0,
  "global_step": 36816,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 4.932094741416776e-05,
      "loss": 1.0183,
      "step": 500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.864189482833551e-05,
      "loss": 0.8804,
      "step": 1000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.7962842242503265e-05,
      "loss": 0.8122,
      "step": 1500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.728378965667101e-05,
      "loss": 0.7686,
      "step": 2000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.660473707083877e-05,
      "loss": 0.7689,
      "step": 2500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.592568448500652e-05,
      "loss": 0.7629,
      "step": 3000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.5246631899174276e-05,
      "loss": 0.7423,
      "step": 3500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.456757931334203e-05,
      "loss": 0.7304,
      "step": 4000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.3888526727509784e-05,
      "loss": 0.7108,
      "step": 4500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.320947414167754e-05,
      "loss": 0.707,
      "step": 5000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.2530421555845286e-05,
      "loss": 0.6951,
      "step": 5500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.185136897001304e-05,
      "loss": 0.6974,
      "step": 6000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.1172316384180795e-05,
      "loss": 0.6861,
      "step": 6500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.049326379834855e-05,
      "loss": 0.6804,
      "step": 7000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.98142112125163e-05,
      "loss": 0.6778,
      "step": 7500
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.913515862668405e-05,
      "loss": 0.6702,
      "step": 8000
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.8456106040851805e-05,
      "loss": 0.6665,
      "step": 8500
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.777705345501956e-05,
      "loss": 0.6647,
      "step": 9000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.7098000869187314e-05,
      "loss": 0.6572,
      "step": 9500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.641894828335506e-05,
      "loss": 0.6462,
      "step": 10000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.5739895697522816e-05,
      "loss": 0.6498,
      "step": 10500
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.506084311169057e-05,
      "loss": 0.6489,
      "step": 11000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.4381790525858324e-05,
      "loss": 0.639,
      "step": 11500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.370273794002607e-05,
      "loss": 0.6453,
      "step": 12000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7558232931726908,
      "eval_loss": 0.6077868342399597,
      "eval_runtime": 4.7892,
      "eval_samples_per_second": 519.922,
      "eval_steps_per_second": 65.147,
      "step": 12272
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.3023685354193826e-05,
      "loss": 0.6051,
      "step": 12500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.234463276836158e-05,
      "loss": 0.5721,
      "step": 13000
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.1665580182529335e-05,
      "loss": 0.5778,
      "step": 13500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.098652759669709e-05,
      "loss": 0.5651,
      "step": 14000
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0307475010864843e-05,
      "loss": 0.5542,
      "step": 14500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9628422425032598e-05,
      "loss": 0.5584,
      "step": 15000
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.8949369839200345e-05,
      "loss": 0.5563,
      "step": 15500
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.82703172533681e-05,
      "loss": 0.5423,
      "step": 16000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7591264667535854e-05,
      "loss": 0.5594,
      "step": 16500
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.6912212081703608e-05,
      "loss": 0.5648,
      "step": 17000
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6233159495871362e-05,
      "loss": 0.5483,
      "step": 17500
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5554106910039117e-05,
      "loss": 0.5565,
      "step": 18000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4875054324206867e-05,
      "loss": 0.5519,
      "step": 18500
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4196001738374622e-05,
      "loss": 0.5571,
      "step": 19000
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3516949152542376e-05,
      "loss": 0.5519,
      "step": 19500
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2837896566710127e-05,
      "loss": 0.5462,
      "step": 20000
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.215884398087788e-05,
      "loss": 0.5483,
      "step": 20500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1479791395045636e-05,
      "loss": 0.5503,
      "step": 21000
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.0800738809213386e-05,
      "loss": 0.5506,
      "step": 21500
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.012168622338114e-05,
      "loss": 0.5355,
      "step": 22000
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.944263363754889e-05,
      "loss": 0.5424,
      "step": 22500
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8763581051716646e-05,
      "loss": 0.545,
      "step": 23000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8084528465884397e-05,
      "loss": 0.5304,
      "step": 23500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.740547588005215e-05,
      "loss": 0.5455,
      "step": 24000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6726423294219905e-05,
      "loss": 0.5289,
      "step": 24500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7819277108433735,
      "eval_loss": 0.5688490867614746,
      "eval_runtime": 4.7774,
      "eval_samples_per_second": 521.2,
      "eval_steps_per_second": 65.307,
      "step": 24544
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.6047370708387656e-05,
      "loss": 0.4449,
      "step": 25000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.536831812255541e-05,
      "loss": 0.4432,
      "step": 25500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4689265536723165e-05,
      "loss": 0.4347,
      "step": 26000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.4010212950890916e-05,
      "loss": 0.4341,
      "step": 26500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.333116036505867e-05,
      "loss": 0.4321,
      "step": 27000
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2652107779226425e-05,
      "loss": 0.4294,
      "step": 27500
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1973055193394177e-05,
      "loss": 0.449,
      "step": 28000
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1294002607561931e-05,
      "loss": 0.4334,
      "step": 28500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0614950021729684e-05,
      "loss": 0.4316,
      "step": 29000
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.935897435897435e-06,
      "loss": 0.435,
      "step": 29500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.25684485006519e-06,
      "loss": 0.4287,
      "step": 30000
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.577792264232942e-06,
      "loss": 0.445,
      "step": 30500
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.898739678400696e-06,
      "loss": 0.4271,
      "step": 31000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.219687092568449e-06,
      "loss": 0.4276,
      "step": 31500
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.540634506736203e-06,
      "loss": 0.4202,
      "step": 32000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.861581920903955e-06,
      "loss": 0.4137,
      "step": 32500
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.182529335071708e-06,
      "loss": 0.4312,
      "step": 33000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.503476749239461e-06,
      "loss": 0.4147,
      "step": 33500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.824424163407214e-06,
      "loss": 0.426,
      "step": 34000
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.1453715775749674e-06,
      "loss": 0.4262,
      "step": 34500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.466318991742721e-06,
      "loss": 0.4247,
      "step": 35000
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.7872664059104738e-06,
      "loss": 0.4191,
      "step": 35500
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.108213820078227e-06,
      "loss": 0.4202,
      "step": 36000
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.2916123424598005e-07,
      "loss": 0.4199,
      "step": 36500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7867469879518072,
      "eval_loss": 0.588763415813446,
      "eval_runtime": 4.7887,
      "eval_samples_per_second": 519.969,
      "eval_steps_per_second": 65.153,
      "step": 36816
    },
    {
      "epoch": 3.0,
      "step": 36816,
      "total_flos": 7.74938740264658e+16,
      "train_loss": 0.5654819974273456,
      "train_runtime": 7177.4758,
      "train_samples_per_second": 164.139,
      "train_steps_per_second": 5.129
    }
  ],
  "max_steps": 36816,
  "num_train_epochs": 3,
  "total_flos": 7.74938740264658e+16,
  "trial_name": null,
  "trial_params": null
}