End of training

all_results.json

@@ -1,8 +1,8 @@
 {
  "epoch": 3.25,
- "eval_accuracy": 1.0,
- "eval_loss": 0.08780594170093536,
- "eval_runtime": 76.2242,
- "eval_samples_per_second": 4.552,
- "eval_steps_per_second": 0.761
+ "eval_accuracy": 0.9654178674351584,
+ "eval_loss": 0.10321669280529022,
+ "eval_runtime": 81.3764,
+ "eval_samples_per_second": 4.264,
+ "eval_steps_per_second": 0.713
 }

test_results.json

@@ -1,8 +1,8 @@
 {
  "epoch": 3.25,
- "eval_accuracy": 1.0,
- "eval_loss": 0.08780594170093536,
- "eval_runtime": 76.2242,
- "eval_samples_per_second": 4.552,
- "eval_steps_per_second": 0.761
+ "eval_accuracy": 0.9654178674351584,
+ "eval_loss": 0.10321669280529022,
+ "eval_runtime": 81.3764,
+ "eval_samples_per_second": 4.264,
+ "eval_steps_per_second": 0.713
 }

trainer_state.json

@@ -10,205 +10,205 @@
  "log_history": [
  {
  "epoch": 0.05,
- "grad_norm": 14.361767768859863,
+ "grad_norm": 13.759540557861328,
  "learning_rate": 2.5e-05,
- "loss": 2.3816,
+ "loss": 2.1835,
  "step": 10
  },
  {
  "epoch": 0.1,
- "grad_norm": 8.448980331420898,
+ "grad_norm": 6.8759236335754395,
  "learning_rate": 5e-05,
- "loss": 1.5686,
+ "loss": 1.4605,
  "step": 20
  },
  {
  "epoch": 0.15,
- "grad_norm": 7.0012526512146,
+ "grad_norm": 4.969621658325195,
  "learning_rate": 4.722222222222222e-05,
- "loss": 0.6191,
+ "loss": 0.5362,
  "step": 30
  },
  {
  "epoch": 0.2,
- "grad_norm": 4.961479187011719,
+ "grad_norm": 7.053426265716553,
  "learning_rate": 4.4444444444444447e-05,
- "loss": 0.3468,
+ "loss": 0.3213,
  "step": 40
  },
  {
  "epoch": 0.25,
- "grad_norm": 2.13403058052063,
+ "grad_norm": 1.6918387413024902,
  "learning_rate": 4.166666666666667e-05,
- "loss": 0.1523,
+ "loss": 0.1339,
  "step": 50
  },
  {
  "epoch": 0.25,
  "eval_accuracy": 1.0,
- "eval_loss": 0.07354684174060822,
- "eval_runtime": 34.9344,
- "eval_samples_per_second": 4.265,
- "eval_steps_per_second": 0.716,
+ "eval_loss": 0.06609966605901718,
+ "eval_runtime": 33.5797,
+ "eval_samples_per_second": 4.437,
+ "eval_steps_per_second": 0.744,
  "step": 50
  },
  {
  "epoch": 1.05,
- "grad_norm": 0.933645486831665,
+ "grad_norm": 0.6457386612892151,
  "learning_rate": 3.888888888888889e-05,
- "loss": 0.0521,
+ "loss": 0.0278,
  "step": 60
  },
  {
  "epoch": 1.1,
- "grad_norm": 0.865109920501709,
+ "grad_norm": 0.6831403374671936,
  "learning_rate": 3.611111111111111e-05,
- "loss": 0.0226,
+ "loss": 0.0171,
  "step": 70
  },
  {
  "epoch": 1.15,
- "grad_norm": 0.17869924008846283,
+ "grad_norm": 0.2744472622871399,
  "learning_rate": 3.3333333333333335e-05,
- "loss": 0.108,
+ "loss": 0.0444,
  "step": 80
  },
  {
  "epoch": 1.2,
- "grad_norm": 0.23531177639961243,
+ "grad_norm": 0.17638139426708221,
  "learning_rate": 3.055555555555556e-05,
- "loss": 0.0107,
+ "loss": 0.0083,
  "step": 90
  },
  {
  "epoch": 1.25,
- "grad_norm": 0.1640271246433258,
+ "grad_norm": 0.08078416436910629,
  "learning_rate": 2.777777777777778e-05,
- "loss": 0.024,
+ "loss": 0.0302,
  "step": 100
  },
  {
  "epoch": 1.25,
- "eval_accuracy": 0.9865771812080537,
- "eval_loss": 0.046999599784612656,
- "eval_runtime": 32.3832,
- "eval_samples_per_second": 4.601,
- "eval_steps_per_second": 0.772,
+ "eval_accuracy": 1.0,
+ "eval_loss": 0.016043640673160553,
+ "eval_runtime": 32.5864,
+ "eval_samples_per_second": 4.572,
+ "eval_steps_per_second": 0.767,
  "step": 100
  },
  {
  "epoch": 2.05,
- "grad_norm": 0.47139155864715576,
+ "grad_norm": 6.167383193969727,
  "learning_rate": 2.5e-05,
- "loss": 0.0056,
+ "loss": 0.0247,
  "step": 110
  },
  {
  "epoch": 2.1,
- "grad_norm": 0.11248086392879486,
+ "grad_norm": 0.060583747923374176,
  "learning_rate": 2.2222222222222223e-05,
- "loss": 0.0045,
+ "loss": 0.0037,
  "step": 120
  },
  {
  "epoch": 2.15,
- "grad_norm": 0.07324650138616562,
+ "grad_norm": 0.06478825956583023,
  "learning_rate": 1.9444444444444445e-05,
- "loss": 0.0042,
+ "loss": 0.0044,
  "step": 130
  },
  {
  "epoch": 2.2,
- "grad_norm": 0.46895766258239746,
+ "grad_norm": 0.18354220688343048,
  "learning_rate": 1.6666666666666667e-05,
- "loss": 0.0202,
+ "loss": 0.0069,
  "step": 140
  },
  {
  "epoch": 2.25,
- "grad_norm": 0.048765964806079865,
+ "grad_norm": 0.03793413192033768,
  "learning_rate": 1.388888888888889e-05,
- "loss": 0.0583,
+ "loss": 0.0119,
  "step": 150
  },
  {
  "epoch": 2.25,
- "eval_accuracy": 0.9865771812080537,
- "eval_loss": 0.030154313892126083,
- "eval_runtime": 32.7398,
- "eval_samples_per_second": 4.551,
- "eval_steps_per_second": 0.764,
+ "eval_accuracy": 1.0,
+ "eval_loss": 0.021267052739858627,
+ "eval_runtime": 32.5518,
+ "eval_samples_per_second": 4.577,
+ "eval_steps_per_second": 0.768,
  "step": 150
  },
  {
  "epoch": 3.05,
- "grad_norm": 0.11046731472015381,
+ "grad_norm": 0.049735669046640396,
  "learning_rate": 1.1111111111111112e-05,
- "loss": 0.0032,
+ "loss": 0.0033,
  "step": 160
  },
  {
  "epoch": 3.1,
- "grad_norm": 0.07638247311115265,
+ "grad_norm": 0.042223431169986725,
  "learning_rate": 8.333333333333334e-06,
- "loss": 0.0027,
+ "loss": 0.0023,
  "step": 170
  },
  {
  "epoch": 3.15,
- "grad_norm": 0.0314478874206543,
+ "grad_norm": 0.02806258387863636,
  "learning_rate": 5.555555555555556e-06,
- "loss": 0.0053,
+ "loss": 0.0033,
  "step": 180
  },
  {
  "epoch": 3.2,
- "grad_norm": 0.23283490538597107,
+ "grad_norm": 0.08276186883449554,
  "learning_rate": 2.777777777777778e-06,
- "loss": 0.0032,
+ "loss": 0.003,
  "step": 190
  },
  {
  "epoch": 3.25,
- "grad_norm": 0.3815344274044037,
+ "grad_norm": 0.4868297278881073,
  "learning_rate": 0.0,
- "loss": 0.0036,
+ "loss": 0.0028,
  "step": 200
  },
  {
  "epoch": 3.25,
- "eval_accuracy": 0.9865771812080537,
- "eval_loss": 0.02967873215675354,
- "eval_runtime": 32.6048,
- "eval_samples_per_second": 4.57,
- "eval_steps_per_second": 0.767,
+ "eval_accuracy": 0.9932885906040269,
+ "eval_loss": 0.025193922221660614,
+ "eval_runtime": 32.4856,
+ "eval_samples_per_second": 4.587,
+ "eval_steps_per_second": 0.77,
  "step": 200
  },
  {
  "epoch": 3.25,
  "step": 200,
  "total_flos": 1.0514301109272576e+18,
- "train_loss": 0.2698267618194222,
- "train_runtime": 1064.2955,
- "train_samples_per_second": 1.128,
- "train_steps_per_second": 0.188
+ "train_loss": 0.24147296573035418,
+ "train_runtime": 1050.525,
+ "train_samples_per_second": 1.142,
+ "train_steps_per_second": 0.19
  },
  {
  "epoch": 3.25,
- "eval_accuracy": 1.0,
- "eval_loss": 0.08780594170093536,
- "eval_runtime": 86.4118,
- "eval_samples_per_second": 4.016,
- "eval_steps_per_second": 0.671,
+ "eval_accuracy": 0.9654178674351584,
+ "eval_loss": 0.10321671515703201,
+ "eval_runtime": 78.8722,
+ "eval_samples_per_second": 4.4,
+ "eval_steps_per_second": 0.735,
  "step": 200
  },
  {
  "epoch": 3.25,
- "eval_accuracy": 1.0,
- "eval_loss": 0.08780594170093536,
- "eval_runtime": 76.2242,
- "eval_samples_per_second": 4.552,
- "eval_steps_per_second": 0.761,
+ "eval_accuracy": 0.9654178674351584,
+ "eval_loss": 0.10321669280529022,
+ "eval_runtime": 81.3764,
+ "eval_samples_per_second": 4.264,
+ "eval_steps_per_second": 0.713,
  "step": 200
  }
  ],