https://github.com/mafda/lightweight_fine_tuning_project
This repository provides a Jupyter notebook demonstrating parameter-efficient fine-tuning (PEFT) with LoRA on Hugging Face models.
https://github.com/mafda/lightweight_fine_tuning_project
huggingface huggingface-datasets huggingface-transformers lora peft peft-fine-tuning-llm pytorch
Last synced: 3 months ago
JSON representation
This repository provides a Jupyter notebook demonstrating parameter-efficient fine-tuning (PEFT) with LoRA on Hugging Face models.
- Host: GitHub
- URL: https://github.com/mafda/lightweight_fine_tuning_project
- Owner: mafda
- Created: 2024-10-26T23:35:17.000Z (8 months ago)
- Default Branch: main
- Last Pushed: 2024-11-11T13:34:20.000Z (7 months ago)
- Last Synced: 2025-01-17T05:09:14.722Z (5 months ago)
- Topics: huggingface, huggingface-datasets, huggingface-transformers, lora, peft, peft-fine-tuning-llm, pytorch
- Language: Jupyter Notebook
- Homepage:
- Size: 2.6 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Lightweight Fine-Tuning Project
Lightweight fine-tuning is one of the most important techniques for adapting
foundation models, because it allows you to modify foundation models for your
needs without needing substantial computational resources.
In this project, I will apply parameter-efficient fine-tuning using the
Hugging Face `peft` library.
> [See notebook](src/lightweight_finetuning.ipynb)
## Methodology
In this project, I will bring together all of the essential components of a
PyTorch + Hugging Face training and inference process. Specifically, I will:
1. Load a pre-trained model and evaluate its performance Perform
2. parameter-efficient fine tuning using the pre-trained model Perform inference
3. using the fine-tuned model and compare its performance to the original model
## Description
* PEFT technique: I used **LoRA** as my PEFT technique. LoRA is the only PEFT
technique that is compatible with all models at this time.
* Model: I used **distilbert-base-uncased** as my model. This is a relatively
small model that is compatible with sequence classification and **LoRA**.
* Evaluation approach: The evaluation approach covered in this project was the
`evaluate` method with a **Hugging Face** `Trainer`.
* Fine-tuning dataset: I use a dataset from Hugging Face's datasets library
[stanfordnlp/imdb](https://huggingface.co/datasets/stanfordnlp/imdb).
## Project Setup
### Clone this repository
```shell
(base)$: [email protected]:mafda/lightweight_fine_tuning_project.git
(base)$: cd lightweight_fine_tuning_project
```
### Configure environment
- Create the conda environment
```shell
(base)$: conda env create -f environment.yml
```
- Activate the environment
```shell
(base)$: conda activate peft
```
- Download the [base
model](https://drive.google.com/drive/folders/10oHiSKabEgOMj1p88Dw0hSIN1gSHoyvG?usp=share_link).
Unzip the folder and place it in the repo, at location
`path/to/lightweight_fine_tuning_project/model/`.
- Download the [peft
model](https://drive.google.com/drive/folders/10oHiSKabEgOMj1p88Dw0hSIN1gSHoyvG?usp=share_link).
Unzip the folder and place it in the repo, at location
`path/to/lightweight_fine_tuning_project/model/`.
- Donwload the [lora
model](https://drive.google.com/drive/folders/10oHiSKabEgOMj1p88Dw0hSIN1gSHoyvG?usp=share_link)
for the dog dataset. Place it in the repo, at location
`path/to/lightweight_fine_tuning_project/model/`.
## Project Structure
```shell
.
├── README.md
├── environment.yml
├── model
│ ├── base
│ ├── lora
│ └── peft
└── src
└── lightweight_finetuning.ipynb
```
## References
- [Generative AI Nanodegree
Program](https://www.udacity.com/course/generative-ai--nd608)
---
made with 💙 by [mafda](https://mafda.github.io/)