Overview

Fine-tuned Llama-2 chat 7B with new dataset including posts from Telegram channel about cyber security. Used QLoRA for fine-tuning. Trained for one epoch using Google Collab.

More about training this model - https://github.com/ArzhkvJl/PDFchat/blob/TestDockerSupport/fune-tuning.py.
How to use - https://github.com/ArzhkvJl/PDFchat.