Hugging Face's logo

chchen
/
Mistral-7B-Instruct-v0.2-ORPO

PEFT
Safetensors
llama-factory
lora
trl
dpo
Generated from Trainer
Model card Files Community
Mistral-7B-Instruct-v0.2-ORPO / README.md
chchen's picture
chchen
End of training
5238045 verified
preview code
|
raw
history blame
1.66 kB
metadata
license: apache-2.0
library_name: peft
tags:
  - llama-factory
  - lora
  - trl
  - dpo
  - generated_from_trainer
base_model: mistralai/Mistral-7B-Instruct-v0.2
model-index:
  - name: Mistral-7B-Instruct-v0.2-ORPO
    results: []

Mistral-7B-Instruct-v0.2-ORPO

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.2 on the dpo_mix_en dataset. It achieves the following results on the evaluation set:

  • Loss: 1.1249
  • Rewards/chosen: -0.1062
  • Rewards/rejected: -0.1197
  • Rewards/accuracies: 0.4400
  • Rewards/margins: 0.0135
  • Logps/rejected: -1.1975
  • Logps/chosen: -1.0620
  • Logits/rejected: -2.6819
  • Logits/chosen: -2.6777
  • Sft Loss: 1.0620
  • Odds Ratio Loss: 0.6295

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-06
  • train_batch_size: 2
  • eval_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 0.1
  • num_epochs: 3.0
  • mixed_precision_training: Native AMP

Training results

Framework versions

  • PEFT 0.10.0
  • Transformers 4.40.1
  • Pytorch 2.3.0
  • Datasets 2.19.0
  • Tokenizers 0.19.1