TreeReview

This repository contains the implementation code and benchmark dataset for the paper "TreeReview: A Dynamic Tree of Questions Framework for Deep and Efficient LLM-based Scientific Peer Review". It provides a framework for generating comprehensive and in-depth peer reviews using Large Language Models (LLMs) through a hierarchical question-answering process.

Setup

1. Create a Conda Environment:

conda create -n treereview python=3.11
conda activate treereview

2. Install Dependencies:

pip install -r requirements.txt

Usage

To generate a review for a single paper, run the following command:

python main.py --paper-id <paper_id> --mmd-path <path_to_paper_mmd> --output-path <output_file>