qiita https://qiita.com/Shoelife2022/items/f69def09a6a272b42d16

claudeで生成したREADME↓

Gemini Auto Summarize

This repository contains a Python script that automatically summarizes academic papers using the Gemini large language model. The script reads a PDF file of a research paper, feeds it to the Gemini model, and generates responses to a set of predefined prompts. The generated summaries provide an in-depth explanation of the paper's content, experiments, results, and conclusions.

Features

• Automatically reads a PDF file of a research paper
• Utilizes the Gemini-1.5-Flash or Gemini-1.5-Pro model for text generation
• Generates detailed summaries based on a set of predefined prompts
• Supports single file processing or batch processing of multiple files
• Handles request limits and server errors by switching between Flash and Pro models
• Generates a markdown file with the summarized content

Requirements

• Python 3.x
• tqdm
• google-generativeai
• argparse
• pypdf

Usage

1. Clone the repository:

git clone [email protected]:shogo-d-nakamura/gemini_auto_summarize.git

2. Install the required dependencies:

conda env create -f env.yml

3. Set up your Gemini API key:

• Create a file named gemini_api in the project directory.
• Paste your Gemini API key into the file and save it.

4. Run the script:

• To process a single PDF file:

python main.py --input <filename>

• To process multiple PDF files in a batch:

python main.py --batch <dirname>

• To use the Gemini-1.5-Pro model (default is Gemini-1.5-Flash):

python main.py --input <filename> --pro

5. The script will generate a markdown file with the same name as the input PDF file, containing the summarized content.

Prompts

The script uses a set of predefined prompts to generate summaries for different sections of the research paper. The prompts cover the following aspects:

• Overall content of the paper
• Introduction
• Important cited papers
• Datasets used
• Experiments and their results
• Proposed methods
• Conclusion
• Related work
• Novelty of the paper
• Advantages over existing research
• Challenges and limitations

Limitations

• The script relies on the availability and performance of the Gemini language model. Request limits or server issues may affect the summarization process.
• The quality of the generated summaries depends on the quality and clarity of the input PDF file.
• The script assumes the PDF file contains a single research paper. Processing multiple papers in a single PDF file may lead to unexpected results.

Contributing

Contributions to this project are welcome. If you find any issues or have suggestions for improvements, please open an issue or submit a pull request.

License

This project is licensed under the MIT License.

Acknowledgements

• This project utilizes the Gemini language model developed by Google.
• The script is built using various open-source libraries and tools.