Skip to content

Latest commit

 

History

History
100 lines (75 loc) · 3.75 KB

README.md

File metadata and controls

100 lines (75 loc) · 3.75 KB

CritiqueFineTuning

This repo contains the code for Critique Fine-Tuning: Learning to Critique is More Effective than Learning to Imitate. In this paper, we introduce Critique Fine-Tuning (CFT) - a paradigm shift in LLM training where models learn to critique rather than imitate!


Highlights

Our fine-tuning method can achieve on par results with RL training!

abs

News

  • [2025/01/30] ⚡️ The paper, code, data, and model for CritiqueFineTuning are all available online.

Getting Started

Installation

  1. First install LLaMA-Factory:
git clone --depth 1 https://github.com/hiyouga/LLaMA-Factory.git
cd LLaMA-Factory
pip install -e ".[torch,metrics]"
  1. Install additional requirements: pip install -r requirements.txt

Training Steps

  1. First, clone the repository and download the dataset:
git clone https://github.com/TIGER-AI-Lab/CritiqueFineTuning.git
cd tools/scripts
bash download_data.sh

  1. Configure model paths in train/scripts/train_qwen2_5-math-7b-cft/qwen2.5-math-7b-cft-webinstruct-50k.yaml

  2. Start training:

cd ../../train/scripts/train_qwen2_5-math-7b-cft
bash train.sh

For training the 32B model, follow a similar process but refer to the configuration in train/scripts/train_qwen2_5-32b-instruct-cft/qwen2.5-32b-cft-webinstruct-4k.yaml.

Note: In our paper experiments, we used MATH-500 as the validation set to select the final checkpoint. After training is complete, run the following commands to generate validation scores:

cd train/Validation
bash start_validate.sh

This will create a validation_summary.txt file containing MATH-500 scores for each checkpoint. Select the checkpoint with the highest score as your final model.

Evaluation

Fill in the model path and evaluation result save path in tools/scripts/evaluate.sh, then run:

cd tools/scripts
bash evaluate.sh

Note: Our evaluation code is modified from Qwen2.5-Math and MAmmoTH.

Construct Critique Data

To create your own critique data, you can use our data generation script:

cd tools/self_construct_critique_data
bash run.sh

Simply modify the model_name parameter in run.sh to specify which model you want to use as the critique teacher. The script will generate critique data following our paper's approach.

Citation

Cite our paper as

@misc{wang2025critiquefinetuninglearningcritique,
      title={Critique Fine-Tuning: Learning to Critique is More Effective than Learning to Imitate},
      author={Yubo Wang and Xiang Yue and Wenhu Chen},
      year={2025},
      eprint={2501.17703},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2501.17703},
}