doplhin-dpo-1-epoch / README.md
kyryl-opens-ml's picture
Model save
2361139 verified
metadata
license: apache-2.0
library_name: peft
tags:
  - trl
  - dpo
  - generated_from_trainer
base_model: cognitivecomputations/dolphin-2.1-mistral-7b
model-index:
  - name: doplhin-dpo-1-epoch
    results: []

doplhin-dpo-1-epoch

This model is a fine-tuned version of cognitivecomputations/dolphin-2.1-mistral-7b on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6046
  • Rewards/chosen: -6.6173
  • Rewards/rejected: -10.5237
  • Rewards/accuracies: 0.7880
  • Rewards/margins: 3.9064
  • Logps/rejected: -431.3049
  • Logps/chosen: -422.0936
  • Logits/rejected: -2.5993
  • Logits/chosen: -2.6739

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: 5e-05
  • train_batch_size: 2
  • eval_batch_size: 2
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 8
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 0.9

Training results

Training Loss Epoch Step Validation Loss Rewards/chosen Rewards/rejected Rewards/accuracies Rewards/margins Logps/rejected Logps/chosen Logits/rejected Logits/chosen
0.5544 0.62 700 0.6046 -6.6173 -10.5237 0.7880 3.9064 -431.3049 -422.0936 -2.5993 -2.6739

Framework versions

  • PEFT 0.8.2
  • Transformers 4.37.2
  • Pytorch 2.1.0+cu118
  • Datasets 2.16.1
  • Tokenizers 0.15.2