Hugging Face's logo

Holarissun
/
RM-HH-GPT2Large_helpful_gpt3_loraR64_40000_gpt2-large_shuffleTrue_extractchosenTrue

PEFT
Safetensors
trl
reward-trainer
Generated from Trainer
Model card Files Community
RM-HH-GPT2Large_helpful_gpt3_loraR64_40000_gpt2-large_shuffleTrue_extractchosenTrue / README.md
Holarissun's picture
Holarissun
End of training
496dca8 verified
preview code
|
raw
history blame contribute delete
No virus
7.55 kB
metadata
license: mit
library_name: peft
tags:
  - trl
  - reward-trainer
  - generated_from_trainer
metrics:
  - accuracy
base_model: openai-community/gpt2-large
model-index:
  - name: >-
      RM-HH-GPT2Large_helpful_gpt3_loraR64_40000_gpt2-large_shuffleTrue_extractchosenTrue
    results: []

RM-HH-GPT2Large_helpful_gpt3_loraR64_40000_gpt2-large_shuffleTrue_extractchosenTrue

This model is a fine-tuned version of openai-community/gpt2-large on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4911
  • Accuracy: 0.7362

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 1.41e-05
  • train_batch_size: 1
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 4
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 2.0

Training results

Training Loss Epoch Step Validation Loss Accuracy
0.8065 0.02 250 0.7702 0.4763
0.7485 0.04 500 0.6903 0.5578
0.6625 0.06 750 0.6116 0.6516
0.5815 0.08 1000 0.5742 0.6817
0.5657 0.1 1250 0.5565 0.6940
0.5608 0.13 1500 0.5479 0.7015
0.5611 0.15 1750 0.5418 0.7083
0.5342 0.17 2000 0.5386 0.7105
0.5842 0.19 2250 0.5319 0.7124
0.5096 0.21 2500 0.5293 0.7171
0.5234 0.23 2750 0.5258 0.7173
0.5321 0.25 3000 0.5243 0.7202
0.5295 0.27 3250 0.5212 0.7202
0.5211 0.29 3500 0.5220 0.7200
0.5119 0.31 3750 0.5215 0.7205
0.509 0.33 4000 0.5200 0.7226
0.5393 0.36 4250 0.5155 0.7230
0.5343 0.38 4500 0.5143 0.7267
0.4944 0.4 4750 0.5195 0.7277
0.5198 0.42 5000 0.5188 0.7258
0.523 0.44 5250 0.5206 0.7282
0.53 0.46 5500 0.5082 0.7264
0.5107 0.48 5750 0.5103 0.7307
0.502 0.5 6000 0.5163 0.7284
0.5198 0.52 6250 0.5132 0.7305
0.5879 0.54 6500 0.5067 0.7313
0.5174 0.57 6750 0.5061 0.7311
0.5062 0.59 7000 0.5053 0.7298
0.5265 0.61 7250 0.5064 0.7303
0.5043 0.63 7500 0.5096 0.7309
0.5291 0.65 7750 0.5073 0.7299
0.4966 0.67 8000 0.5141 0.7305
0.5361 0.69 8250 0.5086 0.7288
0.534 0.71 8500 0.5051 0.7288
0.5073 0.73 8750 0.5104 0.7286
0.5155 0.75 9000 0.5138 0.7290
0.5041 0.77 9250 0.5149 0.7294
0.5552 0.8 9500 0.5030 0.7288
0.5177 0.82 9750 0.4995 0.7294
0.4882 0.84 10000 0.5007 0.7337
0.5409 0.86 10250 0.4992 0.7320
0.5044 0.88 10500 0.4994 0.7311
0.4897 0.9 10750 0.5013 0.7322
0.509 0.92 11000 0.4999 0.7331
0.5256 0.94 11250 0.4950 0.7360
0.4976 0.96 11500 0.4937 0.7356
0.5033 0.98 11750 0.4952 0.7358
0.4917 1.0 12000 0.4939 0.7333
0.4615 1.03 12250 0.5005 0.7328
0.4797 1.05 12500 0.4981 0.7347
0.4872 1.07 12750 0.4997 0.7362
0.5106 1.09 13000 0.5012 0.7343
0.482 1.11 13250 0.5021 0.7365
0.4916 1.13 13500 0.4946 0.7367
0.4957 1.15 13750 0.4972 0.7379
0.4822 1.17 14000 0.5072 0.7379
0.4911 1.19 14250 0.5080 0.7343
0.5042 1.21 14500 0.5148 0.7343
0.4966 1.23 14750 0.5055 0.7350
0.527 1.26 15000 0.4945 0.7345
0.4544 1.28 15250 0.5070 0.7354
0.5198 1.3 15500 0.4993 0.7335
0.5138 1.32 15750 0.4958 0.7358
0.5324 1.34 16000 0.4917 0.7348
0.4695 1.36 16250 0.4951 0.7347
0.5016 1.38 16500 0.4938 0.7360
0.478 1.4 16750 0.4969 0.7345
0.4955 1.42 17000 0.4958 0.7345
0.5072 1.44 17250 0.4908 0.7341
0.4764 1.46 17500 0.4957 0.7345
0.5096 1.49 17750 0.4928 0.7347
0.4944 1.51 18000 0.4923 0.7331
0.4766 1.53 18250 0.4931 0.7333
0.515 1.55 18500 0.4897 0.7339
0.4672 1.57 18750 0.4920 0.7348
0.5122 1.59 19000 0.4921 0.7337
0.5395 1.61 19250 0.4899 0.7333
0.5088 1.63 19500 0.4892 0.7326
0.4864 1.65 19750 0.4895 0.7358
0.4605 1.67 20000 0.4968 0.7358
0.5165 1.7 20250 0.4940 0.7354
0.4955 1.72 20500 0.4919 0.7348
0.4923 1.74 20750 0.4906 0.7348
0.5121 1.76 21000 0.4905 0.7337
0.5068 1.78 21250 0.4892 0.7356
0.4767 1.8 21500 0.4900 0.7350
0.4976 1.82 21750 0.4904 0.7354
0.4934 1.84 22000 0.4893 0.7356
0.479 1.86 22250 0.4905 0.7352
0.4698 1.88 22500 0.4909 0.7347
0.4894 1.9 22750 0.4907 0.7352
0.509 1.93 23000 0.4907 0.7354
0.4805 1.95 23250 0.4914 0.7350
0.5152 1.97 23500 0.4911 0.7358
0.4935 1.99 23750 0.4911 0.7362

Framework versions

  • PEFT 0.9.0
  • Transformers 4.38.2
  • Pytorch 2.1.2
  • Datasets 2.18.0
  • Tokenizers 0.15.2