---
title: README
emoji: 🏢
colorFrom: red
colorTo: green
sdk: static
pinned: false
tags: [llama, jupyter, colab, function calling, QLoRa, fine-tuning, scripts, fine-tuning]
---

# *QLoRa Fine-Tuning Scripts/Notebooks for Small Datasets*

## QLoRa Training Notebook for Llama 2 (FREE)
- Access the Google Colab script [here](https://colab.research.google.com/drive/1uMSS1o_8YOPyG1X_4k6ENEE3kJfBGGhH?usp=sharing).

- View the [Youtube Video Here](https://youtu.be/OQdp-OeG1as)

## QLoRa ADVANCED Training Notebook (PAID)
This advanced script provides improved performance when training with small datasets:
- Includes a prompt loss-mask for improved performance when structured responses are required.
- Includes a stop token after responses - allowing the model to provide a short reponse (e.g. a function call) and then stop.
- Request [access here](https://buy.stripe.com/5kA5l69K52Hxf3a006). €14.99 (or $16.49) per seat/user. Access will be provided within 24 hours of purchase.

# Function Calling Dataset
- Commercial dataset allowing language models to be fine-tuned for function calling ([Get access here](https://huggingface.co/datasets/Trelis/function_calling_extended)).
- Created using only human input or Apache 2 licensed datasets (no third party commercial licensing limitations)
- Models trained with this dataset:
- - [Llama-2-7B](https://huggingface.co/Trelis/Llama-2-7b-chat-hf-function-calling) - repo includes Google Colab notebook for inference
- - [Llama-2-13B](https://huggingface.co/Trelis/Llama-2-13b-chat-hf-function-calling) - repo includes Google Colab notebook for inference

# Jupyter Llama
A Chat Assistant in Jupyter Notebook - Built on Llama 2.
- Upload pdf or text files for analysis.
- No data goes to OpenAI.
- No data is used for training language models.

Try it:
- [Google Colab](https://colab.research.google.com/drive/1iRNIVACiRGjvCrCn4ClAAW9-KoHj9W2z?usp=sharing).
- [Github](https://github.com/TrelisResearch/jupyter-llama/).