LLama 3.2 Summarizor

This is just the simple summarization tool for evaluating the LLama3 model's performance using cli commands.

How to use

Clone the repository from the github repository:

git clone https://github.com/pereira90-ai/llama3-cli-summarizor

cd llaba3-cli-summarizor

Environment Setup

• Install conda environment from yaml file:

conda env create -f environment.yaml -n summarizor
conda activate summarizor

• Install cuda enabled torch on the conda environment:

pip uninstall torch -y
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

Download LLama 3.2 model from huggingface

First of all make sure you are in the project directory: llama3-cli-summarizor And then:

python download_model.py

Run the summarizor with command

You can use text or PDF file as an input

python app.py <your_pdf.txt> <output_file.txt> <max_token_len>

Sample usage:

python app.py meeting_summary.pdf out.txt 256

Here your_pdf.txt is the path of your meeting log file to summarize while output_file.txt is the path of the result. And max_token_len is the optional parameter to change the length of the summary length.

Enjoy!