README.md

# 🦙🎛️ LLaMA-LoRA

<a href="https://colab.research.google.com/github/zetavg/LLaMA-LoRA/blob/main/LLaMA_LoRA.ipynb" target="_parent"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"/></a>

Making evaluating and fine-tuning LLaMA models with low-rank adaptation (LoRA) easy.


## Features

* [1-click up and running in Google Colab](https://colab.research.google.com/github/zetavg/LLaMA-LoRA/blob/main/LLaMA_LoRA.ipynb).
  * Loads and stores data in Google Drive.
* Evaluate various LLaMA LoRA models stored in your folder or from Hugging Face.<br /><a href="https://user-images.githubusercontent.com/3784687/230273443-3b1738ca-fbc4-427b-a1e6-9dabca160288.mp4"><img width="640px" src="https://user-images.githubusercontent.com/3784687/230272844-09f7a35b-46bf-4101-b15d-4ddf243b8bef.gif" /></a>
* Fine-tune LLaMA models with different prompt templates and training dataset format.<br /><a href="https://user-images.githubusercontent.com/3784687/230276439-d729f900-c8c3-47b8-b6f6-93a5cfcecbe8.mp4"><img width="640px" src="https://user-images.githubusercontent.com/3784687/230277315-9a91d983-1690-4594-9d54-912eda8963ee.gif" /></a>
  * Load JSON and JSONL datasets from your folder, or even paste plain text directly into the UI.
  * Supports Stanford Alpaca [seed_tasks](https://github.com/tatsu-lab/stanford_alpaca/blob/main/seed_tasks.jsonl), [alpaca_data](https://github.com/tatsu-lab/stanford_alpaca/blob/main/alpaca_data.json) and [OpenAI "prompt"-"completion"](https://platform.openai.com/docs/guides/fine-tuning/data-formatting) format.


## Acknowledgements

* https://github.com/tloen/alpaca-lora
* https://github.com/lxe/simple-llama-finetuner
* ...

TBC