<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.> CycleResearcher is an open-source ecosystem for AI-driven academic research and review. It targets researchers and engineers seeking to automate research generation, rigorous multi-perspective evaluation, and accelerate scientific discovery through integrated feedback loops.

How It Works

The system comprises three core components: CycleResearcher for paper generation, CycleReviewer for detailed academic reviews, and DeepReviewer for multi-perspective review simulations with self-verification. This creates a complete feedback loop between research output and evaluation, aiming to automate academic processes and accelerate discovery.

Quick Start & Requirements

Installation is via pip install ai_researcher. Requires Python 3.8+. Significant computational resources are implied by model sizes ranging from 7B to 123B parameters. DeepReviewer's Best Mode necessitates a Semantic Scholar API key and running local services (start_models.sh, openscholar_api.py). Relevant links include the project homepage (http://ai-researcher.cn), the arXiv paper (https://arxiv.org/abs/2411.00816), and the ai-researcher.net platform for direct arXiv paper interaction.

Highlighted Details

CycleResearcher-12B achieves a paper generation score of 5.36, nearing conference-accepted averages. CycleReviewer demonstrates superior review performance with a 48.77% reduction in Proxy MSE compared to human reviewers and 74.24% decision accuracy. DeepReviewer offers flexible review modes (Fast, Standard, Best) including background knowledge search and self-verification. An AI detection feature is also included.

Maintenance & Community

Direct contact is available via email: zhuminjun@westlake.edu.cn. No specific community channels (e.g., Discord, Slack) or roadmap details are provided in the README.

Licensing & Compatibility

The project is distributed under a custom CycleResearcher-License. Specific terms and restrictions for commercial use or closed-source linking are detailed in the LICENSE.md file.

Limitations & Caveats

Reliance on large language models may introduce inherent biases or hallucinations. The setup for DeepReviewer's Best Mode involves external API keys and local service deployment. The custom license requires careful review for compatibility with downstream applications.