LLM-Finetuner

this repo is dedicated for a framework that facilitates the process of fine tuning LLMs from data preprocessing to evaluation with an emphasize on efficiency and cost-effectiveness

Table of Contents

• Overview
• Features
• Installation
• Usage
• Contributing
• License

Overview

This framework consists of the following modules:

• Preprocessing: where you provide a link to a huggingface hosted dataset and define the preprocessing criterias
• Fine-tuning: where you specify the base LLM and the processed dataset as well as the training hyperparameters
• Generation: where you can test your fine-tuned model and generate answers and make conversations
• Evaluation: many evaluation metrics are available where you can evaluate your finetuned model on many aspects

Features

• Easy setup and configuration
• Support for huggingface models and datasets
• Comprehensive experiment management
• Detailed performance evaluation and reporting through weights and biases integration
• Customizable and extensible

Installation

To install LLM-Finetuner, clone the repository and install the required dependencies:

git clone https://github.com/imadken/LLM-Finetuner.git
cd LLM-Finetuner

Usage

Follow these steps to use LLM-Finetuner for your fine-tuning tasks:

1. Clone the Project

2. setup conda env

create a conda env from environment.yml

conda env create -f environment.yml

conda activate myenv

conda env list

conda activate myenv

if you need to evaluate using LLM-based evaluation then you should switch to environment-test.yml

3. Configure the necessary Yaml files

• For each module, there is a yaml file to configure
• Before the evaluation, you need to generate answers and feed them to the evaluation module
• in the end, configure the main.yaml and run main.py

Contributing

We welcome contributions to LLM-Finetuner! To contribute, please follow these steps:

1. Fork the repository.
2. Create a new branch (git checkout -b feature-branch).
3. Commit your changes (git commit -am 'Add new feature').
4. Push to the branch (git push origin feature-branch).
5. Create a new Pull Request.

Please ensure your code adheres to the project's coding standards and includes appropriate tests.

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

For more detailed documentation and examples, please refer to the project's for any questions or support.