Attiri, an extension of the LLaMa, aims to build and share an instruction-following LLaMA model for the Tamil language. Our project includes the similar 52K data used for fine-tuning the original model but in Tamil language, as well as the code for generating the data and fine-tuning the model. As of now, the project is under development and will be available soon.

The repository contains

• Dataset
• Code to generate the data
• Code to fine tune LLaMA 7B model

Table of Contents

1. Preparation
   1. Setup
   2. Dataset
2. Usage
   1. Translate
   2. Finetuning
3. Citation
4. To Contribute
5. To-Do
6. Acknowledgments
7. License
8. Fun-Fact

Preparation

Setup

To use the program, you must have Python 3.9+ (recommended = 3.9) and the necessary packages installed. You can install the necessary packages using pip:

Create a new Conda environment with Python 3.9:

conda create --name attiri python=3.9

Activate the new environment: