Book summarizer

Reads an epub format book. Use an LLM to summarize it chapter by chapter, and write it to output.

Dependencies

1. Install the software packages.

pip install EbookLib beautifulsoup4
pip install transformers torch

2. To allow Slack integration, set an incoming webhook URL and export it to the environment variable SLACK_WEBHOOK_URL. Otherwise, uncomment the line post_via_webhook in the main script summarize_book.py.

Usage

1. Place a book in the eBooks folder. Let's say the file name is book_name.epub.
2. Enter the scripts folder.
3. Call sbatch script_run.sbatch to launch the computation job to slurm.
4. When the job completes, the resulting book summary (in markdown) will appear in eBooks folder as book_name_summary-modelname.md

Configurations

• To change the GPUs to run on, modify the script_run.sbatch script.
• To change the model, modify the --model and --model_shortname when calling summarize_book.py.
• To change the prompt, modify the command in the summarize_book.py