The LLM Paper QA is a tool designed to read, parse, and generate summaries from scientific papers hosted on the arXiv repository. This tool uses OpenAI's GPT-4 (gpt-4-32k) to generate the summaries and can be useful for quickly understanding the key points of a paper without having to read the entire document.

Before running the script, you need to set up your environment variables. An example .env.example file has been provided in the repository. Copy the .env.example file to a new file called .env, then replace the placeholders with your actual information.

cp .env.example .env
# edit .env

After setting up your environment variables, install the required packages using pip:

pip install -r requirements.txt

Once the required packages are installed, you can run the script with the following command:

python src/read_paper.py [arxiv_id]

Replace [arxiv_id] with the ID of the paper from arXiv that you wish to summarize.

The script supports several command-line options:

--dry-run: If True, don't actually generate a summary. The paper will still be downloaded and cached, but no calls will be made to the LLM.

--keep-ref: If True, keep the references in the paper before summarizing.

--keep-app: If True, keep the appendices in the paper before summarizing.

--keep-latex: If True, keep the LaTeX in the paper before summarizing, otherwise it will be converted to plain text.

--use-ar5iv: If True, use ar5iv instead of arxiv-vanity.

--force-refresh: If True, force the paper to be refreshed, ignoring the cache.

The script outputs a text file containing the generated summary of the paper. This summary includes the paper's metadata, number of tokens in the input prompt, number of tokens in the generated content, and the content itself.

If the CodiMD client is enabled and not in dry-run mode, the summary is also published to a CodiMD document and its URL is printed to the console.

If a paper is not found, is still being rendered, or has failed to render, the script will print an error message and exit.

Please report any issues you encounter here.

Contributions to LLM Paper QA are welcomed.

This project is licensed under the MIT License - see the LICENSE.md file for details.

(README.md generated using GPT-4.)