https://github.com/pankti09/fine-tuning-llms-using-qlora
fine-tune LLaMA and BERT using the PEFT (Parameter-Efficient Fine-Tuning) method and LoRA (Low-Rank Adaptation) in Python.
https://github.com/pankti09/fine-tuning-llms-using-qlora
fine-tune llama llm
Last synced: 8 months ago
JSON representation
fine-tune LLaMA and BERT using the PEFT (Parameter-Efficient Fine-Tuning) method and LoRA (Low-Rank Adaptation) in Python.
- Host: GitHub
- URL: https://github.com/pankti09/fine-tuning-llms-using-qlora
- Owner: pankti09
- Created: 2024-09-06T02:21:39.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2024-09-06T03:25:54.000Z (about 1 year ago)
- Last Synced: 2024-12-31T04:13:11.628Z (9 months ago)
- Topics: fine-tune, llama, llm
- Language: Jupyter Notebook
- Homepage:
- Size: 5.86 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Fine-Tuning-LLMs-using-QLoRA
# Fine-Tuning LLMs with LoRA
This project demonstrates how to fine-tune large language models (LLMs) such as **LLaMA** and **BERT** using **LoRA (Low-Rank Adaptation)** with the **PEFT (Parameter-Efficient Fine-Tuning)** method. The models are fine-tuned on different tasks using Hugging Face's `transformers` library and PyTorch.
## Table of Contents
- [Project Overview](#project-overview)
- [Technologies Used](#technologies-used)
- [Installation](#installation)
- [Usage](#usage)
- [Training LLaMA Model](#training-llama-model)
- [Training BERT Model](#training-bert-model)
- [License](#license)
## Project Overview
In this project, we fine-tune two different models:
- **LLaMA** for causal language modeling using the **wikitext** dataset.
- **BERT** for sequence classification using the **IMDB movie reviews** dataset.
We leverage **LoRA**, a technique that allows efficient fine-tuning by reducing the number of trainable parameters, making the process more computationally efficient.
## Technologies Used
- Python
- Hugging Face `transformers`
- `datasets`
- `PEFT (Parameter-Efficient Fine-Tuning)`
- PyTorch
- `accelerate`
## Installation
To set up this project, you need to install the required dependencies listed in the `requirements.txt` file.
1. Clone the repository:
```bash
git clone https://github.com/your-username/llm-fine-tuning-lora.git
cd llm-fine-tuning-lora
```
2. Install the dependencies:
```bash
pip install -r requirements.txt
```
## Usage
1. Make sure to have access to a GPU for efficient fine-tuning.
2. Run the notebook to fine-tune either **LLaMA** or **BERT** models.
To train **LLaMA** and **BERT**, you can directly run the code in the notebook or use a Python script based on the notebook's content.
## Training LLaMA Model
To fine-tune the LLaMA model using the **Wikitext** dataset, load the LLaMA model, configure LoRA, and start training. The training process involves:
- Loading the tokenizer and model.
- Applying LoRA configuration.
- Fine-tuning the model with reduced parameters for efficiency.
## Training BERT Model
For **BERT**, the sequence classification task is conducted using the **IMDB** dataset. The process is similar:
- Loading the tokenizer and model.
- Applying LoRA configuration for parameter-efficient tuning.
- Fine-tuning the model for binary classification.
## License
This project is licensed under the MIT License. See the [LICENSE](LICENSE) file for more details.