LoReFT

Hyper fast fine tuning for LLMs using PyReft.

See it live and in action 📺

<a href="https://youtu.be/iy9Z4DyHxvE"><img src="https://i.imgur.com/DLy9byB.jpeg"/></a>

Startup 🚀

1. Install pip install torch==2.2.0 transformers pyreft
2. Replace your HF token with access tokens from here https://huggingface.co/settings/tokens
3. Run the training pipeline python train.py
4. Score using the scoring pipeline python score.py

Other References 🔗

<p>-<a href="https://github.com/stanfordnlp/pyreft">PyReft</a>:Pyreft library from the awesome Stanford NLP team.</p> <p>-<a href="https://arxiv.org/pdf/2404.03592">ReFt Paper</a>:Research paper talking about Representation Fine Tuning.</p>

Who, When, Why?

👨🏾‍💻 Author: Nick Renotte <br /> 📅 Version: 1.x<br /> 📜 License: This project is licensed under the MIT License </br>