{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.0,
  "global_step": 295000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.991559322033899e-05,
      "loss": 0.0122,
      "step": 500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.983084745762712e-05,
      "loss": 0.0065,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.974610169491526e-05,
      "loss": 0.0059,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.966135593220339e-05,
      "loss": 0.0056,
      "step": 2000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9576610169491526e-05,
      "loss": 0.0053,
      "step": 2500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9491864406779666e-05,
      "loss": 0.0058,
      "step": 3000
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.94071186440678e-05,
      "loss": 0.0054,
      "step": 3500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.932237288135593e-05,
      "loss": 0.0049,
      "step": 4000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.923762711864407e-05,
      "loss": 0.0053,
      "step": 4500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9152881355932206e-05,
      "loss": 0.005,
      "step": 5000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9068135593220346e-05,
      "loss": 0.0051,
      "step": 5500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.898338983050847e-05,
      "loss": 0.0049,
      "step": 6000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.889864406779661e-05,
      "loss": 0.0046,
      "step": 6500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.881389830508475e-05,
      "loss": 0.005,
      "step": 7000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.8729152542372885e-05,
      "loss": 0.0046,
      "step": 7500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.864440677966102e-05,
      "loss": 0.0046,
      "step": 8000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.855966101694915e-05,
      "loss": 0.0045,
      "step": 8500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.847491525423729e-05,
      "loss": 0.0046,
      "step": 9000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.8390169491525425e-05,
      "loss": 0.0046,
      "step": 9500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.830542372881356e-05,
      "loss": 0.0044,
      "step": 10000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.82206779661017e-05,
      "loss": 0.0046,
      "step": 10500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.813593220338983e-05,
      "loss": 0.0043,
      "step": 11000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.805118644067797e-05,
      "loss": 0.0042,
      "step": 11500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.7966440677966104e-05,
      "loss": 0.0045,
      "step": 12000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.7881864406779666e-05,
      "loss": 0.0044,
      "step": 12500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.77971186440678e-05,
      "loss": 0.0046,
      "step": 13000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.771237288135593e-05,
      "loss": 0.0043,
      "step": 13500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.7627627118644066e-05,
      "loss": 0.0043,
      "step": 14000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.7542881355932206e-05,
      "loss": 0.0045,
      "step": 14500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.745830508474576e-05,
      "loss": 0.0041,
      "step": 15000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.73735593220339e-05,
      "loss": 0.0041,
      "step": 15500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.7288813559322034e-05,
      "loss": 0.0044,
      "step": 16000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.7204067796610174e-05,
      "loss": 0.0044,
      "step": 16500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.711949152542373e-05,
      "loss": 0.0044,
      "step": 17000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.703474576271187e-05,
      "loss": 0.004,
      "step": 17500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.695e-05,
      "loss": 0.0041,
      "step": 18000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.6865423728813565e-05,
      "loss": 0.004,
      "step": 18500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.67806779661017e-05,
      "loss": 0.0039,
      "step": 19000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.669593220338983e-05,
      "loss": 0.004,
      "step": 19500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.6611186440677964e-05,
      "loss": 0.0041,
      "step": 20000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.6526440677966104e-05,
      "loss": 0.004,
      "step": 20500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.644186440677966e-05,
      "loss": 0.0038,
      "step": 21000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.635728813559322e-05,
      "loss": 0.0041,
      "step": 21500
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.6272542372881355e-05,
      "loss": 0.0037,
      "step": 22000
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.6187796610169495e-05,
      "loss": 0.0039,
      "step": 22500
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.610305084745763e-05,
      "loss": 0.004,
      "step": 23000
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.601830508474577e-05,
      "loss": 0.0041,
      "step": 23500
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.5933559322033895e-05,
      "loss": 0.0037,
      "step": 24000
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.5848983050847464e-05,
      "loss": 0.0042,
      "step": 24500
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.57642372881356e-05,
      "loss": 0.0041,
      "step": 25000
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.567949152542373e-05,
      "loss": 0.004,
      "step": 25500
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.559474576271186e-05,
      "loss": 0.0039,
      "step": 26000
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.551e-05,
      "loss": 0.0039,
      "step": 26500
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.5425254237288136e-05,
      "loss": 0.0038,
      "step": 27000
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.534050847457627e-05,
      "loss": 0.0041,
      "step": 27500
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.525576271186441e-05,
      "loss": 0.0037,
      "step": 28000
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.517118644067797e-05,
      "loss": 0.004,
      "step": 28500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.5086440677966105e-05,
      "loss": 0.004,
      "step": 29000
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.500169491525424e-05,
      "loss": 0.0042,
      "step": 29500
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.491694915254238e-05,
      "loss": 0.004,
      "step": 30000
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.483220338983051e-05,
      "loss": 0.0034,
      "step": 30500
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.474762711864407e-05,
      "loss": 0.0039,
      "step": 31000
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.4662881355932206e-05,
      "loss": 0.0037,
      "step": 31500
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.4578135593220346e-05,
      "loss": 0.0039,
      "step": 32000
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.449338983050847e-05,
      "loss": 0.0035,
      "step": 32500
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.440864406779661e-05,
      "loss": 0.0041,
      "step": 33000
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.4323898305084746e-05,
      "loss": 0.0041,
      "step": 33500
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.4239152542372886e-05,
      "loss": 0.004,
      "step": 34000
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.415440677966102e-05,
      "loss": 0.0038,
      "step": 34500
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.406983050847458e-05,
      "loss": 0.0038,
      "step": 35000
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.3985084745762714e-05,
      "loss": 0.0037,
      "step": 35500
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.390033898305085e-05,
      "loss": 0.004,
      "step": 36000
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.381559322033899e-05,
      "loss": 0.0038,
      "step": 36500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9991040475613329,
      "eval_f1": 0.0,
      "eval_loss": 0.0030166972428560257,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 368.0366,
      "eval_samples_per_second": 107.622,
      "eval_steps_per_second": 6.728,
      "step": 36875
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.373084745762712e-05,
      "loss": 0.0037,
      "step": 37000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.364610169491526e-05,
      "loss": 0.0026,
      "step": 37500
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.356135593220339e-05,
      "loss": 0.0026,
      "step": 38000
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.347661016949153e-05,
      "loss": 0.0025,
      "step": 38500
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.339186440677967e-05,
      "loss": 0.0024,
      "step": 39000
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.33071186440678e-05,
      "loss": 0.002,
      "step": 39500
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.322237288135593e-05,
      "loss": 0.0024,
      "step": 40000
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.3137796610169495e-05,
      "loss": 0.0031,
      "step": 40500
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.305338983050848e-05,
      "loss": 0.0034,
      "step": 41000
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.296864406779661e-05,
      "loss": 0.0031,
      "step": 41500
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.2883898305084746e-05,
      "loss": 0.0029,
      "step": 42000
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.279932203389831e-05,
      "loss": 0.0025,
      "step": 42500
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.271457627118644e-05,
      "loss": 0.0025,
      "step": 43000
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.2629830508474575e-05,
      "loss": 0.0026,
      "step": 43500
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.2545084745762715e-05,
      "loss": 0.0028,
      "step": 44000
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.2460338983050855e-05,
      "loss": 0.0026,
      "step": 44500
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.237559322033898e-05,
      "loss": 0.0028,
      "step": 45000
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.229084745762712e-05,
      "loss": 0.0023,
      "step": 45500
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.2206101694915254e-05,
      "loss": 0.0025,
      "step": 46000
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.2121355932203394e-05,
      "loss": 0.0027,
      "step": 46500
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.203661016949153e-05,
      "loss": 0.0024,
      "step": 47000
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.195186440677966e-05,
      "loss": 0.0025,
      "step": 47500
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.18671186440678e-05,
      "loss": 0.0026,
      "step": 48000
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.1782372881355934e-05,
      "loss": 0.0028,
      "step": 48500
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.1697627118644074e-05,
      "loss": 0.0029,
      "step": 49000
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.161288135593221e-05,
      "loss": 0.0025,
      "step": 49500
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.152813559322034e-05,
      "loss": 0.0024,
      "step": 50000
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.144338983050848e-05,
      "loss": 0.0025,
      "step": 50500
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.135864406779661e-05,
      "loss": 0.0025,
      "step": 51000
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.1274067796610175e-05,
      "loss": 0.0032,
      "step": 51500
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.118949152542373e-05,
      "loss": 0.0026,
      "step": 52000
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.1104745762711864e-05,
      "loss": 0.0026,
      "step": 52500
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.1020000000000004e-05,
      "loss": 0.0025,
      "step": 53000
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.093525423728814e-05,
      "loss": 0.0028,
      "step": 53500
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.085050847457627e-05,
      "loss": 0.0027,
      "step": 54000
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.076593220338983e-05,
      "loss": 0.0027,
      "step": 54500
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.068118644067797e-05,
      "loss": 0.0027,
      "step": 55000
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.0596440677966105e-05,
      "loss": 0.0024,
      "step": 55500
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.051169491525424e-05,
      "loss": 0.0026,
      "step": 56000
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.042694915254238e-05,
      "loss": 0.0025,
      "step": 56500
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.034220338983051e-05,
      "loss": 0.0026,
      "step": 57000
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.0257457627118645e-05,
      "loss": 0.0023,
      "step": 57500
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.017271186440678e-05,
      "loss": 0.0027,
      "step": 58000
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.008796610169492e-05,
      "loss": 0.0025,
      "step": 58500
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.000322033898306e-05,
      "loss": 0.0026,
      "step": 59000
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.9918474576271184e-05,
      "loss": 0.0028,
      "step": 59500
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.9833728813559324e-05,
      "loss": 0.0027,
      "step": 60000
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.974898305084746e-05,
      "loss": 0.0026,
      "step": 60500
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.966440677966102e-05,
      "loss": 0.0026,
      "step": 61000
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.957966101694915e-05,
      "loss": 0.0026,
      "step": 61500
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.9495084745762715e-05,
      "loss": 0.0026,
      "step": 62000
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.941033898305085e-05,
      "loss": 0.0026,
      "step": 62500
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.932559322033899e-05,
      "loss": 0.0028,
      "step": 63000
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.924084745762712e-05,
      "loss": 0.0028,
      "step": 63500
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.9156101694915254e-05,
      "loss": 0.0028,
      "step": 64000
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.907135593220339e-05,
      "loss": 0.0028,
      "step": 64500
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.898661016949153e-05,
      "loss": 0.0027,
      "step": 65000
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.890186440677967e-05,
      "loss": 0.0027,
      "step": 65500
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.8817118644067794e-05,
      "loss": 0.0027,
      "step": 66000
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.8732542372881356e-05,
      "loss": 0.0024,
      "step": 66500
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.8647796610169496e-05,
      "loss": 0.0028,
      "step": 67000
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.856305084745763e-05,
      "loss": 0.0026,
      "step": 67500
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.847830508474576e-05,
      "loss": 0.0028,
      "step": 68000
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.83935593220339e-05,
      "loss": 0.0025,
      "step": 68500
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.830898305084746e-05,
      "loss": 0.0026,
      "step": 69000
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.822457627118644e-05,
      "loss": 0.0025,
      "step": 69500
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.8139830508474575e-05,
      "loss": 0.0026,
      "step": 70000
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.8055084745762715e-05,
      "loss": 0.0027,
      "step": 70500
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.797033898305085e-05,
      "loss": 0.0027,
      "step": 71000
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.788559322033898e-05,
      "loss": 0.0027,
      "step": 71500
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.780084745762712e-05,
      "loss": 0.0026,
      "step": 72000
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.7716101694915255e-05,
      "loss": 0.0026,
      "step": 72500
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.7631355932203395e-05,
      "loss": 0.0022,
      "step": 73000
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.754677966101695e-05,
      "loss": 0.0026,
      "step": 73500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9992288213058589,
      "eval_f1": 0.0,
      "eval_loss": 0.002819355810061097,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 364.7923,
      "eval_samples_per_second": 108.58,
      "eval_steps_per_second": 6.787,
      "step": 73750
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.746203389830509e-05,
      "loss": 0.0021,
      "step": 74000
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.737728813559322e-05,
      "loss": 0.0017,
      "step": 74500
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.729271186440678e-05,
      "loss": 0.0016,
      "step": 75000
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.720796610169492e-05,
      "loss": 0.0018,
      "step": 75500
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.712322033898305e-05,
      "loss": 0.0016,
      "step": 76000
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.7038644067796614e-05,
      "loss": 0.0017,
      "step": 76500
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.695389830508475e-05,
      "loss": 0.0018,
      "step": 77000
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.686932203389831e-05,
      "loss": 0.0022,
      "step": 77500
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.678457627118644e-05,
      "loss": 0.0017,
      "step": 78000
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.6699830508474575e-05,
      "loss": 0.0016,
      "step": 78500
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.6615084745762715e-05,
      "loss": 0.0017,
      "step": 79000
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.653033898305085e-05,
      "loss": 0.0018,
      "step": 79500
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.644559322033899e-05,
      "loss": 0.0018,
      "step": 80000
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.636084745762712e-05,
      "loss": 0.0017,
      "step": 80500
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.627627118644068e-05,
      "loss": 0.0018,
      "step": 81000
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.619152542372882e-05,
      "loss": 0.0016,
      "step": 81500
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.610677966101695e-05,
      "loss": 0.0016,
      "step": 82000
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.6022033898305083e-05,
      "loss": 0.0015,
      "step": 82500
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.5937288135593223e-05,
      "loss": 0.0017,
      "step": 83000
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.5852542372881357e-05,
      "loss": 0.0018,
      "step": 83500
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.576779661016949e-05,
      "loss": 0.0015,
      "step": 84000
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.568305084745763e-05,
      "loss": 0.0019,
      "step": 84500
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.559830508474576e-05,
      "loss": 0.0017,
      "step": 85000
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.55135593220339e-05,
      "loss": 0.0018,
      "step": 85500
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.5428813559322036e-05,
      "loss": 0.0018,
      "step": 86000
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.534406779661017e-05,
      "loss": 0.0018,
      "step": 86500
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.525932203389831e-05,
      "loss": 0.0018,
      "step": 87000
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.517457627118644e-05,
      "loss": 0.0017,
      "step": 87500
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.508983050847458e-05,
      "loss": 0.0019,
      "step": 88000
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.500508474576271e-05,
      "loss": 0.002,
      "step": 88500
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.492033898305085e-05,
      "loss": 0.0016,
      "step": 89000
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.483559322033899e-05,
      "loss": 0.0014,
      "step": 89500
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.4751016949152544e-05,
      "loss": 0.0018,
      "step": 90000
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.466627118644068e-05,
      "loss": 0.0019,
      "step": 90500
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.458152542372882e-05,
      "loss": 0.0019,
      "step": 91000
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.449677966101695e-05,
      "loss": 0.0018,
      "step": 91500
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.4412033898305084e-05,
      "loss": 0.0021,
      "step": 92000
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.4327288135593224e-05,
      "loss": 0.0019,
      "step": 92500
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.424254237288136e-05,
      "loss": 0.0018,
      "step": 93000
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.41577966101695e-05,
      "loss": 0.0019,
      "step": 93500
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.407305084745763e-05,
      "loss": 0.0017,
      "step": 94000
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.398864406779661e-05,
      "loss": 0.0016,
      "step": 94500
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.390389830508475e-05,
      "loss": 0.0017,
      "step": 95000
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.381915254237288e-05,
      "loss": 0.0015,
      "step": 95500
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.373440677966102e-05,
      "loss": 0.0019,
      "step": 96000
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.3649830508474576e-05,
      "loss": 0.0019,
      "step": 96500
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.3565084745762716e-05,
      "loss": 0.0019,
      "step": 97000
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.348033898305085e-05,
      "loss": 0.0019,
      "step": 97500
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.339559322033898e-05,
      "loss": 0.002,
      "step": 98000
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.3311016949152544e-05,
      "loss": 0.0019,
      "step": 98500
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.322627118644068e-05,
      "loss": 0.0018,
      "step": 99000
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.314152542372882e-05,
      "loss": 0.0016,
      "step": 99500
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.305677966101695e-05,
      "loss": 0.0019,
      "step": 100000
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.2972203389830506e-05,
      "loss": 0.0016,
      "step": 100500
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.2887457627118646e-05,
      "loss": 0.0021,
      "step": 101000
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.280271186440678e-05,
      "loss": 0.0017,
      "step": 101500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.271796610169492e-05,
      "loss": 0.0017,
      "step": 102000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.2633389830508474e-05,
      "loss": 0.0021,
      "step": 102500
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.2548813559322037e-05,
      "loss": 0.0018,
      "step": 103000
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.246406779661017e-05,
      "loss": 0.0017,
      "step": 103500
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.237932203389831e-05,
      "loss": 0.0018,
      "step": 104000
    },
    {
      "epoch": 2.83,
      "learning_rate": 3.229457627118644e-05,
      "loss": 0.0017,
      "step": 104500
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.2209830508474576e-05,
      "loss": 0.0018,
      "step": 105000
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.212508474576271e-05,
      "loss": 0.0021,
      "step": 105500
    },
    {
      "epoch": 2.87,
      "learning_rate": 3.204033898305085e-05,
      "loss": 0.0017,
      "step": 106000
    },
    {
      "epoch": 2.89,
      "learning_rate": 3.195559322033899e-05,
      "loss": 0.0019,
      "step": 106500
    },
    {
      "epoch": 2.9,
      "learning_rate": 3.1870847457627116e-05,
      "loss": 0.0018,
      "step": 107000
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.1786101694915256e-05,
      "loss": 0.0017,
      "step": 107500
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.170135593220339e-05,
      "loss": 0.0019,
      "step": 108000
    },
    {
      "epoch": 2.94,
      "learning_rate": 3.161661016949153e-05,
      "loss": 0.0018,
      "step": 108500
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.1532033898305084e-05,
      "loss": 0.0019,
      "step": 109000
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.1447288135593224e-05,
      "loss": 0.0034,
      "step": 109500
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.136254237288136e-05,
      "loss": 0.0019,
      "step": 110000
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.127779661016949e-05,
      "loss": 0.0021,
      "step": 110500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.999242562907679,
      "eval_f1": 0.0,
      "eval_loss": 0.003269583685323596,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 363.311,
      "eval_samples_per_second": 109.022,
      "eval_steps_per_second": 6.815,
      "step": 110625
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.119322033898305e-05,
      "loss": 0.0013,
      "step": 111000
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.110864406779661e-05,
      "loss": 0.0011,
      "step": 111500
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.102389830508475e-05,
      "loss": 0.001,
      "step": 112000
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.093915254237289e-05,
      "loss": 0.0011,
      "step": 112500
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.0854406779661014e-05,
      "loss": 0.0013,
      "step": 113000
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.0769661016949154e-05,
      "loss": 0.0012,
      "step": 113500
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.068491525423729e-05,
      "loss": 0.0013,
      "step": 114000
    },
    {
      "epoch": 3.11,
      "learning_rate": 3.060016949152543e-05,
      "loss": 0.0013,
      "step": 114500
    },
    {
      "epoch": 3.12,
      "learning_rate": 3.0515423728813557e-05,
      "loss": 0.0012,
      "step": 115000
    },
    {
      "epoch": 3.13,
      "learning_rate": 3.043084745762712e-05,
      "loss": 0.0015,
      "step": 115500
    },
    {
      "epoch": 3.15,
      "learning_rate": 3.034610169491526e-05,
      "loss": 0.0013,
      "step": 116000
    },
    {
      "epoch": 3.16,
      "learning_rate": 3.026135593220339e-05,
      "loss": 0.0013,
      "step": 116500
    },
    {
      "epoch": 3.17,
      "learning_rate": 3.0176610169491526e-05,
      "loss": 0.0015,
      "step": 117000
    },
    {
      "epoch": 3.19,
      "learning_rate": 3.0091864406779662e-05,
      "loss": 0.0014,
      "step": 117500
    },
    {
      "epoch": 3.2,
      "learning_rate": 3.000728813559322e-05,
      "loss": 0.0015,
      "step": 118000
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.9922711864406783e-05,
      "loss": 0.0013,
      "step": 118500
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.9837966101694913e-05,
      "loss": 0.0012,
      "step": 119000
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.975322033898305e-05,
      "loss": 0.0012,
      "step": 119500
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.966847457627119e-05,
      "loss": 0.0015,
      "step": 120000
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.9583728813559326e-05,
      "loss": 0.0014,
      "step": 120500
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.9498983050847456e-05,
      "loss": 0.0012,
      "step": 121000
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.9414237288135592e-05,
      "loss": 0.0011,
      "step": 121500
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.932949152542373e-05,
      "loss": 0.0012,
      "step": 122000
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.9244745762711865e-05,
      "loss": 0.0013,
      "step": 122500
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.9160000000000005e-05,
      "loss": 0.0014,
      "step": 123000
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.9075254237288135e-05,
      "loss": 0.0012,
      "step": 123500
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.8990508474576272e-05,
      "loss": 0.0014,
      "step": 124000
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.890593220338983e-05,
      "loss": 0.0013,
      "step": 124500
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.8821355932203393e-05,
      "loss": 0.0016,
      "step": 125000
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.873661016949153e-05,
      "loss": 0.0012,
      "step": 125500
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.865186440677966e-05,
      "loss": 0.0012,
      "step": 126000
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.85671186440678e-05,
      "loss": 0.0013,
      "step": 126500
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.8482372881355936e-05,
      "loss": 0.0012,
      "step": 127000
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.8397627118644072e-05,
      "loss": 0.0012,
      "step": 127500
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.8312881355932202e-05,
      "loss": 0.0013,
      "step": 128000
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.822813559322034e-05,
      "loss": 0.0011,
      "step": 128500
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.81435593220339e-05,
      "loss": 0.0014,
      "step": 129000
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.8058813559322034e-05,
      "loss": 0.0015,
      "step": 129500
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.797406779661017e-05,
      "loss": 0.0012,
      "step": 130000
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.7889322033898307e-05,
      "loss": 0.0012,
      "step": 130500
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.7804576271186444e-05,
      "loss": 0.0013,
      "step": 131000
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.7719830508474577e-05,
      "loss": 0.0011,
      "step": 131500
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.7635084745762713e-05,
      "loss": 0.0013,
      "step": 132000
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.755033898305085e-05,
      "loss": 0.0012,
      "step": 132500
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.7465593220338986e-05,
      "loss": 0.0013,
      "step": 133000
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.7380847457627116e-05,
      "loss": 0.0012,
      "step": 133500
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.7296271186440682e-05,
      "loss": 0.0013,
      "step": 134000
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.7211694915254237e-05,
      "loss": 0.0015,
      "step": 134500
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.7126949152542374e-05,
      "loss": 0.0012,
      "step": 135000
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.704220338983051e-05,
      "loss": 0.0013,
      "step": 135500
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.6957457627118643e-05,
      "loss": 0.0013,
      "step": 136000
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.687271186440678e-05,
      "loss": 0.0013,
      "step": 136500
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.6787966101694917e-05,
      "loss": 0.0012,
      "step": 137000
    },
    {
      "epoch": 3.73,
      "learning_rate": 2.6703220338983053e-05,
      "loss": 0.0014,
      "step": 137500
    },
    {
      "epoch": 3.74,
      "learning_rate": 2.6618644067796612e-05,
      "loss": 0.0012,
      "step": 138000
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.6534067796610167e-05,
      "loss": 0.0012,
      "step": 138500
    },
    {
      "epoch": 3.77,
      "learning_rate": 2.6449322033898304e-05,
      "loss": 0.0013,
      "step": 139000
    },
    {
      "epoch": 3.78,
      "learning_rate": 2.636457627118644e-05,
      "loss": 0.0012,
      "step": 139500
    },
    {
      "epoch": 3.8,
      "learning_rate": 2.627983050847458e-05,
      "loss": 0.0012,
      "step": 140000
    },
    {
      "epoch": 3.81,
      "learning_rate": 2.6195084745762717e-05,
      "loss": 0.0013,
      "step": 140500
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.6110338983050847e-05,
      "loss": 0.0013,
      "step": 141000
    },
    {
      "epoch": 3.84,
      "learning_rate": 2.6025593220338983e-05,
      "loss": 0.0011,
      "step": 141500
    },
    {
      "epoch": 3.85,
      "learning_rate": 2.594084745762712e-05,
      "loss": 0.0012,
      "step": 142000
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.5856101694915256e-05,
      "loss": 0.0012,
      "step": 142500
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.577135593220339e-05,
      "loss": 0.0015,
      "step": 143000
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.5686610169491526e-05,
      "loss": 0.0012,
      "step": 143500
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.5601864406779663e-05,
      "loss": 0.0012,
      "step": 144000
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.551728813559322e-05,
      "loss": 0.0011,
      "step": 144500
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.5432542372881358e-05,
      "loss": 0.0011,
      "step": 145000
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.5347796610169495e-05,
      "loss": 0.0013,
      "step": 145500
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.526322033898305e-05,
      "loss": 0.0013,
      "step": 146000
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.5178474576271187e-05,
      "loss": 0.0012,
      "step": 146500
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.5093728813559327e-05,
      "loss": 0.0011,
      "step": 147000
    },
    {
      "epoch": 4.0,
      "learning_rate": 2.5008983050847456e-05,
      "loss": 0.0014,
      "step": 147500
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9992986286431048,
      "eval_f1": 0.0,
      "eval_loss": 0.003328560618683696,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 365.3105,
      "eval_samples_per_second": 108.426,
      "eval_steps_per_second": 6.778,
      "step": 147500
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.4924237288135593e-05,
      "loss": 0.0008,
      "step": 148000
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.483949152542373e-05,
      "loss": 0.0009,
      "step": 148500
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.4754745762711866e-05,
      "loss": 0.0009,
      "step": 149000
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.4670000000000003e-05,
      "loss": 0.0007,
      "step": 149500
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.458542372881356e-05,
      "loss": 0.0009,
      "step": 150000
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.4500677966101695e-05,
      "loss": 0.0008,
      "step": 150500
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.441593220338983e-05,
      "loss": 0.0006,
      "step": 151000
    },
    {
      "epoch": 4.11,
      "learning_rate": 2.4331186440677968e-05,
      "loss": 0.0009,
      "step": 151500
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.4246440677966104e-05,
      "loss": 0.0007,
      "step": 152000
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.4161694915254237e-05,
      "loss": 0.0007,
      "step": 152500
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.4077288135593222e-05,
      "loss": 0.0009,
      "step": 153000
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.399254237288136e-05,
      "loss": 0.0008,
      "step": 153500
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.3907966101694917e-05,
      "loss": 0.0007,
      "step": 154000
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.382322033898305e-05,
      "loss": 0.0009,
      "step": 154500
    },
    {
      "epoch": 4.2,
      "learning_rate": 2.373847457627119e-05,
      "loss": 0.0008,
      "step": 155000
    },
    {
      "epoch": 4.22,
      "learning_rate": 2.3653728813559323e-05,
      "loss": 0.001,
      "step": 155500
    },
    {
      "epoch": 4.23,
      "learning_rate": 2.356898305084746e-05,
      "loss": 0.001,
      "step": 156000
    },
    {
      "epoch": 4.24,
      "learning_rate": 2.3484237288135593e-05,
      "loss": 0.0009,
      "step": 156500
    },
    {
      "epoch": 4.26,
      "learning_rate": 2.339949152542373e-05,
      "loss": 0.0007,
      "step": 157000
    },
    {
      "epoch": 4.27,
      "learning_rate": 2.3314745762711866e-05,
      "loss": 0.001,
      "step": 157500
    },
    {
      "epoch": 4.28,
      "learning_rate": 2.3230000000000003e-05,
      "loss": 0.0008,
      "step": 158000
    },
    {
      "epoch": 4.3,
      "learning_rate": 2.3145254237288136e-05,
      "loss": 0.0006,
      "step": 158500
    },
    {
      "epoch": 4.31,
      "learning_rate": 2.3060677966101695e-05,
      "loss": 0.0009,
      "step": 159000
    },
    {
      "epoch": 4.33,
      "learning_rate": 2.297593220338983e-05,
      "loss": 0.001,
      "step": 159500
    },
    {
      "epoch": 4.34,
      "learning_rate": 2.2891186440677968e-05,
      "loss": 0.0007,
      "step": 160000
    },
    {
      "epoch": 4.35,
      "learning_rate": 2.2806440677966105e-05,
      "loss": 0.001,
      "step": 160500
    },
    {
      "epoch": 4.37,
      "learning_rate": 2.2721694915254238e-05,
      "loss": 0.001,
      "step": 161000
    },
    {
      "epoch": 4.38,
      "learning_rate": 2.2637118644067796e-05,
      "loss": 0.0008,
      "step": 161500
    },
    {
      "epoch": 4.39,
      "learning_rate": 2.2552372881355933e-05,
      "loss": 0.0008,
      "step": 162000
    },
    {
      "epoch": 4.41,
      "learning_rate": 2.246762711864407e-05,
      "loss": 0.0009,
      "step": 162500
    },
    {
      "epoch": 4.42,
      "learning_rate": 2.2382881355932203e-05,
      "loss": 0.001,
      "step": 163000
    },
    {
      "epoch": 4.43,
      "learning_rate": 2.229813559322034e-05,
      "loss": 0.0008,
      "step": 163500
    },
    {
      "epoch": 4.45,
      "learning_rate": 2.2213389830508476e-05,
      "loss": 0.0009,
      "step": 164000
    },
    {
      "epoch": 4.46,
      "learning_rate": 2.2128644067796613e-05,
      "loss": 0.001,
      "step": 164500
    },
    {
      "epoch": 4.47,
      "learning_rate": 2.204389830508475e-05,
      "loss": 0.0008,
      "step": 165000
    },
    {
      "epoch": 4.49,
      "learning_rate": 2.1959152542372882e-05,
      "loss": 0.0008,
      "step": 165500
    },
    {
      "epoch": 4.5,
      "learning_rate": 2.187457627118644e-05,
      "loss": 0.0009,
      "step": 166000
    },
    {
      "epoch": 4.52,
      "learning_rate": 2.1789830508474578e-05,
      "loss": 0.0008,
      "step": 166500
    },
    {
      "epoch": 4.53,
      "learning_rate": 2.1705084745762714e-05,
      "loss": 0.0009,
      "step": 167000
    },
    {
      "epoch": 4.54,
      "learning_rate": 2.1620508474576273e-05,
      "loss": 0.0009,
      "step": 167500
    },
    {
      "epoch": 4.56,
      "learning_rate": 2.1535762711864406e-05,
      "loss": 0.0007,
      "step": 168000
    },
    {
      "epoch": 4.57,
      "learning_rate": 2.1451016949152543e-05,
      "loss": 0.0009,
      "step": 168500
    },
    {
      "epoch": 4.58,
      "learning_rate": 2.136627118644068e-05,
      "loss": 0.0008,
      "step": 169000
    },
    {
      "epoch": 4.6,
      "learning_rate": 2.1281525423728816e-05,
      "loss": 0.0008,
      "step": 169500
    },
    {
      "epoch": 4.61,
      "learning_rate": 2.119677966101695e-05,
      "loss": 0.0008,
      "step": 170000
    },
    {
      "epoch": 4.62,
      "learning_rate": 2.1112203389830508e-05,
      "loss": 0.0008,
      "step": 170500
    },
    {
      "epoch": 4.64,
      "learning_rate": 2.1027457627118648e-05,
      "loss": 0.001,
      "step": 171000
    },
    {
      "epoch": 4.65,
      "learning_rate": 2.094271186440678e-05,
      "loss": 0.001,
      "step": 171500
    },
    {
      "epoch": 4.66,
      "learning_rate": 2.0857966101694917e-05,
      "loss": 0.0008,
      "step": 172000
    },
    {
      "epoch": 4.68,
      "learning_rate": 2.077322033898305e-05,
      "loss": 0.0009,
      "step": 172500
    },
    {
      "epoch": 4.69,
      "learning_rate": 2.0688474576271187e-05,
      "loss": 0.0008,
      "step": 173000
    },
    {
      "epoch": 4.71,
      "learning_rate": 2.0603728813559324e-05,
      "loss": 0.0009,
      "step": 173500
    },
    {
      "epoch": 4.72,
      "learning_rate": 2.051898305084746e-05,
      "loss": 0.001,
      "step": 174000
    },
    {
      "epoch": 4.73,
      "learning_rate": 2.0434406779661016e-05,
      "loss": 0.0008,
      "step": 174500
    },
    {
      "epoch": 4.75,
      "learning_rate": 2.0349661016949152e-05,
      "loss": 0.0009,
      "step": 175000
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.026491525423729e-05,
      "loss": 0.0008,
      "step": 175500
    },
    {
      "epoch": 4.77,
      "learning_rate": 2.0180169491525425e-05,
      "loss": 0.0009,
      "step": 176000
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.0095593220338984e-05,
      "loss": 0.0009,
      "step": 176500
    },
    {
      "epoch": 4.8,
      "learning_rate": 2.0010847457627117e-05,
      "loss": 0.0008,
      "step": 177000
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.9926101694915257e-05,
      "loss": 0.0008,
      "step": 177500
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.984135593220339e-05,
      "loss": 0.0007,
      "step": 178000
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.9756610169491527e-05,
      "loss": 0.0007,
      "step": 178500
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.967186440677966e-05,
      "loss": 0.0009,
      "step": 179000
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.9587118644067797e-05,
      "loss": 0.0009,
      "step": 179500
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.950254237288136e-05,
      "loss": 0.0008,
      "step": 180000
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.9417966101694914e-05,
      "loss": 0.001,
      "step": 180500
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.933322033898305e-05,
      "loss": 0.0008,
      "step": 181000
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.9248474576271187e-05,
      "loss": 0.0008,
      "step": 181500
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.9163728813559324e-05,
      "loss": 0.0008,
      "step": 182000
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.907898305084746e-05,
      "loss": 0.0009,
      "step": 182500
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.8994237288135594e-05,
      "loss": 0.0008,
      "step": 183000
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.890949152542373e-05,
      "loss": 0.0007,
      "step": 183500
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.8824745762711864e-05,
      "loss": 0.0009,
      "step": 184000
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9993343568078369,
      "eval_f1": 0.0,
      "eval_loss": 0.003261522389948368,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 364.1478,
      "eval_samples_per_second": 108.772,
      "eval_steps_per_second": 6.799,
      "step": 184375
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.8740000000000004e-05,
      "loss": 0.0008,
      "step": 184500
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.865542372881356e-05,
      "loss": 0.0006,
      "step": 185000
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.8570677966101695e-05,
      "loss": 0.0005,
      "step": 185500
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.848593220338983e-05,
      "loss": 0.0005,
      "step": 186000
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.840118644067797e-05,
      "loss": 0.0005,
      "step": 186500
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.8316440677966102e-05,
      "loss": 0.0005,
      "step": 187000
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.823169491525424e-05,
      "loss": 0.0006,
      "step": 187500
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.814694915254237e-05,
      "loss": 0.0006,
      "step": 188000
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.8062203389830508e-05,
      "loss": 0.0006,
      "step": 188500
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.7977457627118645e-05,
      "loss": 0.0006,
      "step": 189000
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.789271186440678e-05,
      "loss": 0.0005,
      "step": 189500
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.780813559322034e-05,
      "loss": 0.0006,
      "step": 190000
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.7723389830508473e-05,
      "loss": 0.0006,
      "step": 190500
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.763864406779661e-05,
      "loss": 0.0006,
      "step": 191000
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.7553898305084746e-05,
      "loss": 0.0006,
      "step": 191500
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.7469152542372883e-05,
      "loss": 0.0006,
      "step": 192000
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.7384406779661016e-05,
      "loss": 0.0006,
      "step": 192500
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.7299661016949153e-05,
      "loss": 0.0006,
      "step": 193000
    },
    {
      "epoch": 5.25,
      "learning_rate": 1.721491525423729e-05,
      "loss": 0.0004,
      "step": 193500
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.7130338983050848e-05,
      "loss": 0.0006,
      "step": 194000
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.7045593220338985e-05,
      "loss": 0.0006,
      "step": 194500
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.6960847457627118e-05,
      "loss": 0.0006,
      "step": 195000
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.687627118644068e-05,
      "loss": 0.0005,
      "step": 195500
    },
    {
      "epoch": 5.32,
      "learning_rate": 1.6791525423728816e-05,
      "loss": 0.0006,
      "step": 196000
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.670677966101695e-05,
      "loss": 0.0006,
      "step": 196500
    },
    {
      "epoch": 5.34,
      "learning_rate": 1.6622033898305086e-05,
      "loss": 0.0006,
      "step": 197000
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.653728813559322e-05,
      "loss": 0.0006,
      "step": 197500
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.645254237288136e-05,
      "loss": 0.0005,
      "step": 198000
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.6367796610169493e-05,
      "loss": 0.0007,
      "step": 198500
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.6283389830508473e-05,
      "loss": 0.0007,
      "step": 199000
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.619864406779661e-05,
      "loss": 0.0006,
      "step": 199500
    },
    {
      "epoch": 5.42,
      "learning_rate": 1.6113898305084747e-05,
      "loss": 0.0006,
      "step": 200000
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.6029152542372883e-05,
      "loss": 0.0005,
      "step": 200500
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.5944406779661016e-05,
      "loss": 0.0005,
      "step": 201000
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.5859661016949153e-05,
      "loss": 0.0006,
      "step": 201500
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.5775084745762715e-05,
      "loss": 0.0006,
      "step": 202000
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.5690338983050848e-05,
      "loss": 0.0004,
      "step": 202500
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.5605593220338985e-05,
      "loss": 0.0005,
      "step": 203000
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.5520847457627118e-05,
      "loss": 0.0006,
      "step": 203500
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.5436101694915255e-05,
      "loss": 0.0005,
      "step": 204000
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.535135593220339e-05,
      "loss": 0.0006,
      "step": 204500
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.5266610169491528e-05,
      "loss": 0.0005,
      "step": 205000
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.5181864406779661e-05,
      "loss": 0.0005,
      "step": 205500
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.5097118644067797e-05,
      "loss": 0.0004,
      "step": 206000
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.5012372881355932e-05,
      "loss": 0.0005,
      "step": 206500
    },
    {
      "epoch": 5.61,
      "learning_rate": 1.4927627118644069e-05,
      "loss": 0.0008,
      "step": 207000
    },
    {
      "epoch": 5.63,
      "learning_rate": 1.4842881355932204e-05,
      "loss": 0.0006,
      "step": 207500
    },
    {
      "epoch": 5.64,
      "learning_rate": 1.475813559322034e-05,
      "loss": 0.0004,
      "step": 208000
    },
    {
      "epoch": 5.65,
      "learning_rate": 1.46735593220339e-05,
      "loss": 0.0005,
      "step": 208500
    },
    {
      "epoch": 5.67,
      "learning_rate": 1.4588813559322034e-05,
      "loss": 0.0006,
      "step": 209000
    },
    {
      "epoch": 5.68,
      "learning_rate": 1.450406779661017e-05,
      "loss": 0.0006,
      "step": 209500
    },
    {
      "epoch": 5.69,
      "learning_rate": 1.4419322033898305e-05,
      "loss": 0.0006,
      "step": 210000
    },
    {
      "epoch": 5.71,
      "learning_rate": 1.4334576271186442e-05,
      "loss": 0.0005,
      "step": 210500
    },
    {
      "epoch": 5.72,
      "learning_rate": 1.4249999999999999e-05,
      "loss": 0.0004,
      "step": 211000
    },
    {
      "epoch": 5.74,
      "learning_rate": 1.4165254237288136e-05,
      "loss": 0.0004,
      "step": 211500
    },
    {
      "epoch": 5.75,
      "learning_rate": 1.4080677966101696e-05,
      "loss": 0.0007,
      "step": 212000
    },
    {
      "epoch": 5.76,
      "learning_rate": 1.3995932203389831e-05,
      "loss": 0.0004,
      "step": 212500
    },
    {
      "epoch": 5.78,
      "learning_rate": 1.3911186440677968e-05,
      "loss": 0.0005,
      "step": 213000
    },
    {
      "epoch": 5.79,
      "learning_rate": 1.38264406779661e-05,
      "loss": 0.0005,
      "step": 213500
    },
    {
      "epoch": 5.8,
      "learning_rate": 1.3741694915254239e-05,
      "loss": 0.0005,
      "step": 214000
    },
    {
      "epoch": 5.82,
      "learning_rate": 1.3656949152542372e-05,
      "loss": 0.0005,
      "step": 214500
    },
    {
      "epoch": 5.83,
      "learning_rate": 1.3572372881355933e-05,
      "loss": 0.0006,
      "step": 215000
    },
    {
      "epoch": 5.84,
      "learning_rate": 1.348762711864407e-05,
      "loss": 0.0006,
      "step": 215500
    },
    {
      "epoch": 5.86,
      "learning_rate": 1.3402881355932204e-05,
      "loss": 0.0005,
      "step": 216000
    },
    {
      "epoch": 5.87,
      "learning_rate": 1.331813559322034e-05,
      "loss": 0.0006,
      "step": 216500
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.3233389830508476e-05,
      "loss": 0.0004,
      "step": 217000
    },
    {
      "epoch": 5.9,
      "learning_rate": 1.3148644067796612e-05,
      "loss": 0.0006,
      "step": 217500
    },
    {
      "epoch": 5.91,
      "learning_rate": 1.3063898305084745e-05,
      "loss": 0.0005,
      "step": 218000
    },
    {
      "epoch": 5.93,
      "learning_rate": 1.2979152542372882e-05,
      "loss": 0.0005,
      "step": 218500
    },
    {
      "epoch": 5.94,
      "learning_rate": 1.2894406779661017e-05,
      "loss": 0.0006,
      "step": 219000
    },
    {
      "epoch": 5.95,
      "learning_rate": 1.2809661016949153e-05,
      "loss": 0.0005,
      "step": 219500
    },
    {
      "epoch": 5.97,
      "learning_rate": 1.2724915254237288e-05,
      "loss": 0.0006,
      "step": 220000
    },
    {
      "epoch": 5.98,
      "learning_rate": 1.2640338983050849e-05,
      "loss": 0.0004,
      "step": 220500
    },
    {
      "epoch": 5.99,
      "learning_rate": 1.2555762711864407e-05,
      "loss": 0.0006,
      "step": 221000
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9993843762384619,
      "eval_f1": 0.0,
      "eval_loss": 0.0035067975986748934,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 363.9269,
      "eval_samples_per_second": 108.838,
      "eval_steps_per_second": 6.804,
      "step": 221250
    },
    {
      "epoch": 6.01,
      "learning_rate": 1.2471016949152544e-05,
      "loss": 0.0003,
      "step": 221500
    },
    {
      "epoch": 6.02,
      "learning_rate": 1.2386271186440679e-05,
      "loss": 0.0003,
      "step": 222000
    },
    {
      "epoch": 6.03,
      "learning_rate": 1.2301525423728814e-05,
      "loss": 0.0004,
      "step": 222500
    },
    {
      "epoch": 6.05,
      "learning_rate": 1.221677966101695e-05,
      "loss": 0.0003,
      "step": 223000
    },
    {
      "epoch": 6.06,
      "learning_rate": 1.2132033898305085e-05,
      "loss": 0.0003,
      "step": 223500
    },
    {
      "epoch": 6.07,
      "learning_rate": 1.2047457627118644e-05,
      "loss": 0.0004,
      "step": 224000
    },
    {
      "epoch": 6.09,
      "learning_rate": 1.1962711864406779e-05,
      "loss": 0.0004,
      "step": 224500
    },
    {
      "epoch": 6.1,
      "learning_rate": 1.1877966101694917e-05,
      "loss": 0.0004,
      "step": 225000
    },
    {
      "epoch": 6.12,
      "learning_rate": 1.1793220338983052e-05,
      "loss": 0.0004,
      "step": 225500
    },
    {
      "epoch": 6.13,
      "learning_rate": 1.170864406779661e-05,
      "loss": 0.0004,
      "step": 226000
    },
    {
      "epoch": 6.14,
      "learning_rate": 1.1623898305084746e-05,
      "loss": 0.0004,
      "step": 226500
    },
    {
      "epoch": 6.16,
      "learning_rate": 1.1539152542372882e-05,
      "loss": 0.0004,
      "step": 227000
    },
    {
      "epoch": 6.17,
      "learning_rate": 1.1454406779661017e-05,
      "loss": 0.0004,
      "step": 227500
    },
    {
      "epoch": 6.18,
      "learning_rate": 1.1369661016949152e-05,
      "loss": 0.0004,
      "step": 228000
    },
    {
      "epoch": 6.2,
      "learning_rate": 1.1284915254237288e-05,
      "loss": 0.0004,
      "step": 228500
    },
    {
      "epoch": 6.21,
      "learning_rate": 1.1200338983050849e-05,
      "loss": 0.0003,
      "step": 229000
    },
    {
      "epoch": 6.22,
      "learning_rate": 1.1115593220338984e-05,
      "loss": 0.0004,
      "step": 229500
    },
    {
      "epoch": 6.24,
      "learning_rate": 1.1030847457627119e-05,
      "loss": 0.0003,
      "step": 230000
    },
    {
      "epoch": 6.25,
      "learning_rate": 1.0946101694915255e-05,
      "loss": 0.0004,
      "step": 230500
    },
    {
      "epoch": 6.26,
      "learning_rate": 1.086135593220339e-05,
      "loss": 0.0004,
      "step": 231000
    },
    {
      "epoch": 6.28,
      "learning_rate": 1.0776610169491527e-05,
      "loss": 0.0005,
      "step": 231500
    },
    {
      "epoch": 6.29,
      "learning_rate": 1.0691864406779662e-05,
      "loss": 0.0002,
      "step": 232000
    },
    {
      "epoch": 6.31,
      "learning_rate": 1.0607118644067796e-05,
      "loss": 0.0004,
      "step": 232500
    },
    {
      "epoch": 6.32,
      "learning_rate": 1.0522372881355933e-05,
      "loss": 0.0004,
      "step": 233000
    },
    {
      "epoch": 6.33,
      "learning_rate": 1.0437627118644068e-05,
      "loss": 0.0003,
      "step": 233500
    },
    {
      "epoch": 6.35,
      "learning_rate": 1.0352881355932203e-05,
      "loss": 0.0004,
      "step": 234000
    },
    {
      "epoch": 6.36,
      "learning_rate": 1.0268305084745763e-05,
      "loss": 0.0004,
      "step": 234500
    },
    {
      "epoch": 6.37,
      "learning_rate": 1.01835593220339e-05,
      "loss": 0.0003,
      "step": 235000
    },
    {
      "epoch": 6.39,
      "learning_rate": 1.0098813559322035e-05,
      "loss": 0.0006,
      "step": 235500
    },
    {
      "epoch": 6.4,
      "learning_rate": 1.001406779661017e-05,
      "loss": 0.0003,
      "step": 236000
    },
    {
      "epoch": 6.41,
      "learning_rate": 9.929322033898306e-06,
      "loss": 0.0004,
      "step": 236500
    },
    {
      "epoch": 6.43,
      "learning_rate": 9.844576271186441e-06,
      "loss": 0.0004,
      "step": 237000
    },
    {
      "epoch": 6.44,
      "learning_rate": 9.760000000000001e-06,
      "loss": 0.0003,
      "step": 237500
    },
    {
      "epoch": 6.45,
      "learning_rate": 9.675254237288136e-06,
      "loss": 0.0004,
      "step": 238000
    },
    {
      "epoch": 6.47,
      "learning_rate": 9.590508474576273e-06,
      "loss": 0.0003,
      "step": 238500
    },
    {
      "epoch": 6.48,
      "learning_rate": 9.505762711864408e-06,
      "loss": 0.0003,
      "step": 239000
    },
    {
      "epoch": 6.49,
      "learning_rate": 9.421016949152543e-06,
      "loss": 0.0003,
      "step": 239500
    },
    {
      "epoch": 6.51,
      "learning_rate": 9.33627118644068e-06,
      "loss": 0.0003,
      "step": 240000
    },
    {
      "epoch": 6.52,
      "learning_rate": 9.251525423728814e-06,
      "loss": 0.0003,
      "step": 240500
    },
    {
      "epoch": 6.54,
      "learning_rate": 9.16677966101695e-06,
      "loss": 0.0004,
      "step": 241000
    },
    {
      "epoch": 6.55,
      "learning_rate": 9.082033898305086e-06,
      "loss": 0.0004,
      "step": 241500
    },
    {
      "epoch": 6.56,
      "learning_rate": 8.997457627118644e-06,
      "loss": 0.0005,
      "step": 242000
    },
    {
      "epoch": 6.58,
      "learning_rate": 8.912881355932203e-06,
      "loss": 0.0004,
      "step": 242500
    },
    {
      "epoch": 6.59,
      "learning_rate": 8.82813559322034e-06,
      "loss": 0.0003,
      "step": 243000
    },
    {
      "epoch": 6.6,
      "learning_rate": 8.743389830508474e-06,
      "loss": 0.0006,
      "step": 243500
    },
    {
      "epoch": 6.62,
      "learning_rate": 8.658644067796611e-06,
      "loss": 0.0004,
      "step": 244000
    },
    {
      "epoch": 6.63,
      "learning_rate": 8.57406779661017e-06,
      "loss": 0.0003,
      "step": 244500
    },
    {
      "epoch": 6.64,
      "learning_rate": 8.489322033898306e-06,
      "loss": 0.0004,
      "step": 245000
    },
    {
      "epoch": 6.66,
      "learning_rate": 8.404576271186441e-06,
      "loss": 0.0002,
      "step": 245500
    },
    {
      "epoch": 6.67,
      "learning_rate": 8.319830508474578e-06,
      "loss": 0.0004,
      "step": 246000
    },
    {
      "epoch": 6.68,
      "learning_rate": 8.235084745762713e-06,
      "loss": 0.0003,
      "step": 246500
    },
    {
      "epoch": 6.7,
      "learning_rate": 8.150508474576271e-06,
      "loss": 0.0003,
      "step": 247000
    },
    {
      "epoch": 6.71,
      "learning_rate": 8.065762711864406e-06,
      "loss": 0.0004,
      "step": 247500
    },
    {
      "epoch": 6.73,
      "learning_rate": 7.981016949152543e-06,
      "loss": 0.0002,
      "step": 248000
    },
    {
      "epoch": 6.74,
      "learning_rate": 7.896271186440678e-06,
      "loss": 0.0003,
      "step": 248500
    },
    {
      "epoch": 6.75,
      "learning_rate": 7.811694915254238e-06,
      "loss": 0.0004,
      "step": 249000
    },
    {
      "epoch": 6.77,
      "learning_rate": 7.726949152542373e-06,
      "loss": 0.0003,
      "step": 249500
    },
    {
      "epoch": 6.78,
      "learning_rate": 7.642203389830508e-06,
      "loss": 0.0003,
      "step": 250000
    },
    {
      "epoch": 6.79,
      "learning_rate": 7.5574576271186445e-06,
      "loss": 0.0003,
      "step": 250500
    },
    {
      "epoch": 6.81,
      "learning_rate": 7.472711864406779e-06,
      "loss": 0.0004,
      "step": 251000
    },
    {
      "epoch": 6.82,
      "learning_rate": 7.387966101694915e-06,
      "loss": 0.0003,
      "step": 251500
    },
    {
      "epoch": 6.83,
      "learning_rate": 7.303220338983051e-06,
      "loss": 0.0003,
      "step": 252000
    },
    {
      "epoch": 6.85,
      "learning_rate": 7.218474576271187e-06,
      "loss": 0.0003,
      "step": 252500
    },
    {
      "epoch": 6.86,
      "learning_rate": 7.133728813559322e-06,
      "loss": 0.0004,
      "step": 253000
    },
    {
      "epoch": 6.87,
      "learning_rate": 7.048983050847457e-06,
      "loss": 0.0002,
      "step": 253500
    },
    {
      "epoch": 6.89,
      "learning_rate": 6.964237288135593e-06,
      "loss": 0.0002,
      "step": 254000
    },
    {
      "epoch": 6.9,
      "learning_rate": 6.879491525423729e-06,
      "loss": 0.0003,
      "step": 254500
    },
    {
      "epoch": 6.92,
      "learning_rate": 6.794745762711864e-06,
      "loss": 0.0003,
      "step": 255000
    },
    {
      "epoch": 6.93,
      "learning_rate": 6.710338983050849e-06,
      "loss": 0.0003,
      "step": 255500
    },
    {
      "epoch": 6.94,
      "learning_rate": 6.6255932203389835e-06,
      "loss": 0.0002,
      "step": 256000
    },
    {
      "epoch": 6.96,
      "learning_rate": 6.540847457627119e-06,
      "loss": 0.0003,
      "step": 256500
    },
    {
      "epoch": 6.97,
      "learning_rate": 6.456101694915255e-06,
      "loss": 0.0004,
      "step": 257000
    },
    {
      "epoch": 6.98,
      "learning_rate": 6.371355932203391e-06,
      "loss": 0.0003,
      "step": 257500
    },
    {
      "epoch": 7.0,
      "learning_rate": 6.2866101694915265e-06,
      "loss": 0.0004,
      "step": 258000
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9993948198558451,
      "eval_f1": 0.0,
      "eval_loss": 0.003731178119778633,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 371.4536,
      "eval_samples_per_second": 106.632,
      "eval_steps_per_second": 6.666,
      "step": 258125
    },
    {
      "epoch": 7.01,
      "learning_rate": 6.201864406779661e-06,
      "loss": 0.0003,
      "step": 258500
    },
    {
      "epoch": 7.02,
      "learning_rate": 6.117118644067797e-06,
      "loss": 0.0002,
      "step": 259000
    },
    {
      "epoch": 7.04,
      "learning_rate": 6.032372881355933e-06,
      "loss": 0.0002,
      "step": 259500
    },
    {
      "epoch": 7.05,
      "learning_rate": 5.9477966101694915e-06,
      "loss": 0.0002,
      "step": 260000
    },
    {
      "epoch": 7.06,
      "learning_rate": 5.863050847457627e-06,
      "loss": 0.0003,
      "step": 260500
    },
    {
      "epoch": 7.08,
      "learning_rate": 5.778305084745763e-06,
      "loss": 0.0002,
      "step": 261000
    },
    {
      "epoch": 7.09,
      "learning_rate": 5.6937288135593226e-06,
      "loss": 0.0003,
      "step": 261500
    },
    {
      "epoch": 7.11,
      "learning_rate": 5.6089830508474574e-06,
      "loss": 0.0002,
      "step": 262000
    },
    {
      "epoch": 7.12,
      "learning_rate": 5.524237288135593e-06,
      "loss": 0.0002,
      "step": 262500
    },
    {
      "epoch": 7.13,
      "learning_rate": 5.439661016949153e-06,
      "loss": 0.0002,
      "step": 263000
    },
    {
      "epoch": 7.15,
      "learning_rate": 5.3549152542372885e-06,
      "loss": 0.0002,
      "step": 263500
    },
    {
      "epoch": 7.16,
      "learning_rate": 5.270169491525423e-06,
      "loss": 0.0003,
      "step": 264000
    },
    {
      "epoch": 7.17,
      "learning_rate": 5.18542372881356e-06,
      "loss": 0.0001,
      "step": 264500
    },
    {
      "epoch": 7.19,
      "learning_rate": 5.100677966101696e-06,
      "loss": 0.0003,
      "step": 265000
    },
    {
      "epoch": 7.2,
      "learning_rate": 5.0159322033898305e-06,
      "loss": 0.0003,
      "step": 265500
    },
    {
      "epoch": 7.21,
      "learning_rate": 4.931186440677966e-06,
      "loss": 0.0002,
      "step": 266000
    },
    {
      "epoch": 7.23,
      "learning_rate": 4.846440677966102e-06,
      "loss": 0.0002,
      "step": 266500
    },
    {
      "epoch": 7.24,
      "learning_rate": 4.761694915254238e-06,
      "loss": 0.0002,
      "step": 267000
    },
    {
      "epoch": 7.25,
      "learning_rate": 4.6769491525423735e-06,
      "loss": 0.0002,
      "step": 267500
    },
    {
      "epoch": 7.27,
      "learning_rate": 4.592372881355932e-06,
      "loss": 0.0004,
      "step": 268000
    },
    {
      "epoch": 7.28,
      "learning_rate": 4.507627118644068e-06,
      "loss": 0.0002,
      "step": 268500
    },
    {
      "epoch": 7.29,
      "learning_rate": 4.422881355932204e-06,
      "loss": 0.0002,
      "step": 269000
    },
    {
      "epoch": 7.31,
      "learning_rate": 4.3381355932203385e-06,
      "loss": 0.0002,
      "step": 269500
    },
    {
      "epoch": 7.32,
      "learning_rate": 4.253389830508474e-06,
      "loss": 0.0002,
      "step": 270000
    },
    {
      "epoch": 7.34,
      "learning_rate": 4.168644067796611e-06,
      "loss": 0.0002,
      "step": 270500
    },
    {
      "epoch": 7.35,
      "learning_rate": 4.0840677966101696e-06,
      "loss": 0.0003,
      "step": 271000
    },
    {
      "epoch": 7.36,
      "learning_rate": 3.999322033898305e-06,
      "loss": 0.0003,
      "step": 271500
    },
    {
      "epoch": 7.38,
      "learning_rate": 3.914576271186441e-06,
      "loss": 0.0002,
      "step": 272000
    },
    {
      "epoch": 7.39,
      "learning_rate": 3.829830508474577e-06,
      "loss": 0.0002,
      "step": 272500
    },
    {
      "epoch": 7.4,
      "learning_rate": 3.745084745762712e-06,
      "loss": 0.0002,
      "step": 273000
    },
    {
      "epoch": 7.42,
      "learning_rate": 3.660338983050848e-06,
      "loss": 0.0002,
      "step": 273500
    },
    {
      "epoch": 7.43,
      "learning_rate": 3.575593220338983e-06,
      "loss": 0.0003,
      "step": 274000
    },
    {
      "epoch": 7.44,
      "learning_rate": 3.490847457627119e-06,
      "loss": 0.0003,
      "step": 274500
    },
    {
      "epoch": 7.46,
      "learning_rate": 3.406101694915254e-06,
      "loss": 0.0003,
      "step": 275000
    },
    {
      "epoch": 7.47,
      "learning_rate": 3.3215254237288133e-06,
      "loss": 0.0002,
      "step": 275500
    },
    {
      "epoch": 7.48,
      "learning_rate": 3.236779661016949e-06,
      "loss": 0.0002,
      "step": 276000
    },
    {
      "epoch": 7.5,
      "learning_rate": 3.1520338983050847e-06,
      "loss": 0.0002,
      "step": 276500
    },
    {
      "epoch": 7.51,
      "learning_rate": 3.0672881355932205e-06,
      "loss": 0.0002,
      "step": 277000
    },
    {
      "epoch": 7.53,
      "learning_rate": 2.9827118644067796e-06,
      "loss": 0.0002,
      "step": 277500
    },
    {
      "epoch": 7.54,
      "learning_rate": 2.8979661016949154e-06,
      "loss": 0.0001,
      "step": 278000
    },
    {
      "epoch": 7.55,
      "learning_rate": 2.813220338983051e-06,
      "loss": 0.0003,
      "step": 278500
    },
    {
      "epoch": 7.57,
      "learning_rate": 2.728474576271187e-06,
      "loss": 0.0002,
      "step": 279000
    },
    {
      "epoch": 7.58,
      "learning_rate": 2.643898305084746e-06,
      "loss": 0.0002,
      "step": 279500
    },
    {
      "epoch": 7.59,
      "learning_rate": 2.559322033898305e-06,
      "loss": 0.0002,
      "step": 280000
    },
    {
      "epoch": 7.61,
      "learning_rate": 2.474576271186441e-06,
      "loss": 0.0002,
      "step": 280500
    },
    {
      "epoch": 7.62,
      "learning_rate": 2.3898305084745766e-06,
      "loss": 0.0002,
      "step": 281000
    },
    {
      "epoch": 7.63,
      "learning_rate": 2.305084745762712e-06,
      "loss": 0.0002,
      "step": 281500
    },
    {
      "epoch": 7.65,
      "learning_rate": 2.2203389830508476e-06,
      "loss": 0.0002,
      "step": 282000
    },
    {
      "epoch": 7.66,
      "learning_rate": 2.1355932203389833e-06,
      "loss": 0.0002,
      "step": 282500
    },
    {
      "epoch": 7.67,
      "learning_rate": 2.0508474576271186e-06,
      "loss": 0.0001,
      "step": 283000
    },
    {
      "epoch": 7.69,
      "learning_rate": 1.9661016949152544e-06,
      "loss": 0.0002,
      "step": 283500
    },
    {
      "epoch": 7.7,
      "learning_rate": 1.8813559322033897e-06,
      "loss": 0.0002,
      "step": 284000
    },
    {
      "epoch": 7.72,
      "learning_rate": 1.7967796610169492e-06,
      "loss": 0.0002,
      "step": 284500
    },
    {
      "epoch": 7.73,
      "learning_rate": 1.7120338983050848e-06,
      "loss": 0.0001,
      "step": 285000
    },
    {
      "epoch": 7.74,
      "learning_rate": 1.6272881355932205e-06,
      "loss": 0.0002,
      "step": 285500
    },
    {
      "epoch": 7.76,
      "learning_rate": 1.542542372881356e-06,
      "loss": 0.0002,
      "step": 286000
    },
    {
      "epoch": 7.77,
      "learning_rate": 1.4579661016949154e-06,
      "loss": 0.0001,
      "step": 286500
    },
    {
      "epoch": 7.78,
      "learning_rate": 1.3733898305084745e-06,
      "loss": 0.0002,
      "step": 287000
    },
    {
      "epoch": 7.8,
      "learning_rate": 1.2886440677966103e-06,
      "loss": 0.0002,
      "step": 287500
    },
    {
      "epoch": 7.81,
      "learning_rate": 1.2038983050847458e-06,
      "loss": 0.0003,
      "step": 288000
    },
    {
      "epoch": 7.82,
      "learning_rate": 1.1191525423728813e-06,
      "loss": 0.0001,
      "step": 288500
    },
    {
      "epoch": 7.84,
      "learning_rate": 1.034406779661017e-06,
      "loss": 0.0002,
      "step": 289000
    },
    {
      "epoch": 7.85,
      "learning_rate": 9.496610169491526e-07,
      "loss": 0.0001,
      "step": 289500
    },
    {
      "epoch": 7.86,
      "learning_rate": 8.649152542372882e-07,
      "loss": 0.0002,
      "step": 290000
    },
    {
      "epoch": 7.88,
      "learning_rate": 7.801694915254238e-07,
      "loss": 0.0002,
      "step": 290500
    },
    {
      "epoch": 7.89,
      "learning_rate": 6.954237288135593e-07,
      "loss": 0.0003,
      "step": 291000
    },
    {
      "epoch": 7.91,
      "learning_rate": 6.106779661016949e-07,
      "loss": 0.0001,
      "step": 291500
    },
    {
      "epoch": 7.92,
      "learning_rate": 5.259322033898306e-07,
      "loss": 0.0002,
      "step": 292000
    },
    {
      "epoch": 7.93,
      "learning_rate": 4.411864406779661e-07,
      "loss": 0.0002,
      "step": 292500
    },
    {
      "epoch": 7.95,
      "learning_rate": 3.5661016949152543e-07,
      "loss": 0.0002,
      "step": 293000
    },
    {
      "epoch": 7.96,
      "learning_rate": 2.71864406779661e-07,
      "loss": 0.0002,
      "step": 293500
    },
    {
      "epoch": 7.97,
      "learning_rate": 1.8711864406779662e-07,
      "loss": 0.0002,
      "step": 294000
    },
    {
      "epoch": 7.99,
      "learning_rate": 1.0237288135593222e-07,
      "loss": 0.0001,
      "step": 294500
    },
    {
      "epoch": 8.0,
      "learning_rate": 1.7627118644067797e-08,
      "loss": 0.0002,
      "step": 295000
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9994025151528644,
      "eval_f1": 0.0,
      "eval_loss": 0.003991865552961826,
      "eval_precision": 0.0,
      "eval_recall": 0.0,
      "eval_runtime": 383.2603,
      "eval_samples_per_second": 103.348,
      "eval_steps_per_second": 6.46,
      "step": 295000
    },
    {
      "epoch": 8.0,
      "step": 295000,
      "total_flos": 1.21405005598464e+18,
      "train_loss": 0.0015094334459279553,
      "train_runtime": 87513.7882,
      "train_samples_per_second": 53.934,
      "train_steps_per_second": 3.371
    }
  ],
  "max_steps": 295000,
  "num_train_epochs": 8,
  "total_flos": 1.21405005598464e+18,
  "trial_name": null,
  "trial_params": null
}