Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/JamesQFreeman/LoRA-ViT
Low rank adaptation for Vision Transformer
https://github.com/JamesQFreeman/LoRA-ViT
Last synced: 9 days ago
JSON representation
Low rank adaptation for Vision Transformer
- Host: GitHub
- URL: https://github.com/JamesQFreeman/LoRA-ViT
- Owner: JamesQFreeman
- License: gpl-3.0
- Created: 2023-02-23T10:56:13.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2024-03-18T08:37:22.000Z (8 months ago)
- Last Synced: 2024-08-02T15:30:51.337Z (3 months ago)
- Language: Python
- Size: 2.98 MB
- Stars: 329
- Watchers: 4
- Forks: 14
- Open Issues: 2
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# MeLo: Low-rank Adaptation is Better than Finetuning for Medical Image
## Useful links
## Feature
- [x] Supported DeepLab segmentation for ```lukemelas/PyTorch-Pretrained-ViT```. 2023-03-15
- [x] Supported ```timm```. 2023-03-16
- [x] Supported ```multi-lora```. 2023-11-15
- [ ] Repo clean up.
## Installation
Gii clone. My ```torch.__version__==1.13.0```, other version newer than ```torch.__version__==1.10.0``` should also work, I guess.
You may also need a ```safetensors``` from huggingface to load and save weight.
## Examples
You may find examples in ```examples.ipynb```
## Usage
You may use Vision Transformer from ```timm```:
```python
import timm
import torch
from lora import LoRA_ViT_timm
img = torch.randn(2, 3, 224, 224)
model = timm.create_model('vit_base_patch16_224', pretrained=True)
lora_vit = LoRA_ViT_timm(vit_model=model, r=4, alpha=4, num_classes=10)
pred = lora_vit(img)
print(pred.shape)
```
If timm is too complicated, you can use a simpler implementation of ViT from ```lukemelas/PyTorch-Pretrained-ViT```.
Wrap you ViT using LoRA-ViT, this a simple example of classifer
```python
from base_vit import ViT
import torch
from lora import LoRA_ViT
model = ViT('B_16_imagenet1k')
model.load_state_dict(torch.load('B_16_imagenet1k.pth'))
preds = model(img) # preds.shape = torch.Size([1, 1000])
num_params = sum(p.numel() for p in model.parameters() if p.requires_grad)
print(f"trainable parameters: {num_params}") #trainable parameters: 86859496
lora_model = LoRA_ViT(model, r=4, alpha=4, num_classes=10)
num_params = sum(p.numel() for p in lora_model.parameters() if p.requires_grad)
print(f"trainable parameters: {num_params}") # trainable parameters: 147456
```
this an example for segmentation tasks, using deeplabv3
```python
model = ViT('B_16_imagenet1k')
model.load_state_dict(torch.load('B_16_imagenet1k.pth'))
lora_model = LoRA_ViT(model, r=4, alpha=4)
seg_lora_model = SegWrapForViT(vit_model=lora_model, image_size=384,
patches=16, dim=768, n_classes=10)
num_params = sum(p.numel() for p in seg_lora_model.parameters() if p.requires_grad)
print(f"Number of trainable parameters: {num_params/2**20:.3f}") # trainable parameters: 6.459
```
Save and load LoRA:
```python
lora_model.save_lora_parameters('mytask.lora.safetensors') # save
lora_model.load_lora_parameters('mytask.lora.safetensors') # load
```
## Performance
In M1 Pro, LoRA is about 1.8x~1.9x faster.
```python performance_profile.py``` should do the time profiler now. More test will come soon.
## Citation
Use this bibtex to cite this repository:
```
@misc{zhu2023melo,
title={MeLo: Low-rank Adaptation is Better than Fine-tuning for Medical Image Diagnosis},
author={Yitao Zhu and Zhenrong Shen and Zihao Zhao and Sheng Wang and Xin Wang and Xiangyu Zhao and Dinggang Shen and Qian Wang},
year={2023},
eprint={2311.08236},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```
## Credit
ViT code and imagenet pretrained weight come from ```lukemelas/PyTorch-Pretrained-ViT```