arXiv Paper Recommender

A hackable tool to swim through daily arXiv papers, powered by LLMs.

Usage

Customize the following content, then run python main.py.

Put your OpenAI API key (started with sk-) in .openai_api_key.
Customize CATEGORY_LIST in main.py to be the list of categories you want to keep tracks of (e.g. ['cs.LG'])
arXiv-Recommender will filter out papers with authors in MY_AUTHOR_LIST of filter_util.py. Please customize so you won't miss any paper by them.
arXiv-Recommender will use ChatGPT to filter out papers that are relevant to your research topics defined in research_topic.py. Please customize so you won't miss relevant papers.

Advanced

In llm.py, you can customize prompts, the LLM model to use (default is GPT-3.5)
You can setup your email account in mail_config.py and arXiv-Recommender can send a summary email to your mailbox.

Acknowledgements

Thank you to arXiv for use of its open access interoperability.

Contribute

Feel free to contribute to the project via Pull Requests, Issues, or contact me via kaixuanh@princeton.edu