LitLLM is a powerful AI toolkit that transforms how researchers write literature reviews using advanced Retrieval-Augmented Generation (RAG) to create accurate, well-structured related work sections in minutes rather than hours or days.

🔗 LitLLM Web App

LitLLM is a tool that helps researchers write literature reviews with the assistance of Large Language Models (LLMs). Writing literature reviews is one of the most time-consuming aspects of academic research, particularly in rapidly evolving fields like machine learning. LitLLM addresses this challenge by decomposing the task into two key components:

1. Retrieval: Finding the most relevant papers for your research
2. Generation: Creating a coherent, well-structured literature review based on the retrieved papers

LitLLM follows principles of Retrieval-Augmented Generation (RAG):

1. Keyword Extraction: LLMs identify meaningful keywords from your research abstract
2. Multi-Strategy Search: Combines keyword-based and embedding-based search to query academic databases (Google Scholar, OpenAlex)
3. Re-ranking with Attribution: An LLM re-ranks search results to prioritize the most relevant papers
4. Structured Generation: Creates a literature review following a plan-based approach that organizes the content meaningfully

• Hybrid Retrieval: Combines keyword and embedding-based search for optimal coverage
• Attribution-based Re-ranking: Prioritizes papers by relevance and importance
• Plan-based Generation: Creates structured, coherent literature reviews with fewer hallucinations
• Interactive Interface: User-friendly web interface for generating literature reviews

Note: the current interface has some new features not shown in the screenshot below (exporting citation, adding papers directly, high-level plan generation)

To use LitLLM:

1. Provide your research idea or abstract in the textbox
2. Select papers from the search results to base your literature review on
3. (Optional) Provide a writing plan to guide the generation of literature review
4. Generate literature review!

LitLLM is based on the following two papers:

@article{agarwal2024llms,
  title={LitLLMs, LLMs for Literature Review: Are we there yet?},
  author={Agarwal*, Shubham and Sahu*, Gaurav and Puri*, Abhay and Laradji, Issam H and Dvijotham, Krishnamurthy DJ and Stanley, Jason and Charlin, Laurent and Pal, Christopher},
  journal={arXiv preprint arXiv:2412.15249},
  year={2024}
}

@article{agarwal2024litllm,
  title={Litllm: A toolkit for scientific literature review},
  author={Agarwal*, Shubham and Sahu*, Gaurav and Puri*, Abhay and Laradji, Issam H and Dvijotham, Krishnamurthy DJ and Stanley, Jason and Charlin, Laurent and Pal, Christopher},
  journal={arXiv preprint arXiv:2402.01788},
  year={2024}
}

For feedback:

• Open an issue in this repository
• For private feedback, please fill out our feedback form
• Alternatively, you can also email us at litllm [at] duck [dot] com

We welcome contributions from the community to make LitLLM even better! Please reach out to us at litllm [at] duck [dot] com with any queries.

Made with ❤️ by researchers at Mila and ServiceNow Research