CraFT

KO | EN

gitlite — search

#javascript #ai #go #rust #linux #macos #java #nodejs #awesome #awesome-list #pytorch #mcp #ios #open-source

(★ 19)

Official code for ICML 2024 paper, "Connecting the Dots: Collaborative Fine-tuning for Black-Box Vision-Language Models"

CraFT 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<h1 align='center' style="text-align:center; font-weight:bold; font-size:2.0em;letter-spacing:2.0px;"> Connecting the Dots: Collaborative Fine-tuning for Black-Box Vision-Language Models </h1>

<a href="https://zhengbo.wang/" target="_blank" style="text-decoration: none;">Zhengbo Wang1,2</a>&nbsp;,&nbsp;
 <a href="https://liangjian.xyz/" target="_blank" style="text-decoration: none;">Jian Liang2,3†</a>&nbsp;,&nbsp;
 <a href="https://scholar.google.com/citations?user=ayrg9AUAAAAJ&hl=en" target="_blank" style="text-decoration: none;">Ran He2,3</a>&nbsp;,&nbsp; 
 <a href="http://staff.ustc.edu.cn/~zlwang/index_en.html" target="_blank" style="text-decoration: none;">Zilei Wang1</a>&nbsp;,&nbsp; 
	<a href="https://scholar.google.com/citations?user=W-FGd_UAAAAJ&hl=en" target="_blank" style="text-decoration: none;">Tieniu Tan2,3,4</a>&nbsp;&nbsp;
	 
1University of Science and Technology of China&nbsp;&nbsp;&nbsp;
2NLPR & MAIS, Institute of Automation, Chinese Academy of Sciences&nbsp;&nbsp;&nbsp;
3School of Artificial Intelligence, University of Chinese Academy of Sciences&nbsp;&nbsp;&nbsp;
4Nanjing University

ICML, 2024 




 <a href="https://arxiv.org/abs/2402.04050" target="_blank" style="text-decoration: none;">[Paper]</a>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;
 <a href="https://github.com/mrflogs/CraFT" target="_blank" style="text-decoration: none;">[Code]</a>

## Requirements
### Installation
Create a conda environment and install dependencies:
```
conda create -n craft python=3.9
conda activate craft

pip install -r requirements.txt

# Install the according versions of torch and torchvision
conda install pytorch torchvision cudatoolkit
```

### Dataset
Follow [DATASET.md](DATASET.md) to install ImageNet and other 10 datasets referring to CoOp.

## Get Started
### Configs
The running configurations can be modified in `craft_configs/dataset.yaml`, including shot numbers, visual encoders, and hyperparamters.

### Running
For ImageNet dataset:
```bash
CUDA_VISIBLE_DEVICES=0 python main_craft_imagenet.py --config craft_configs/imagenet.yaml
```
For other 10 datasets:
```bash
CUDA_VISIBLE_DEVICES=0 python main_craft.py --config craft_configs/dataset.yaml
```

## Acknowledgement

This repo benefits from [CLIP](https://github.com/openai/CLIP), [CoOp](https://github.com/KaiyangZhou/CoOp), [SHIP](https://github.com/mrflogs/SHIP), and [H2B](https://github.com/mrflogs/ICLR24). Thanks for their wonderful work.

## Citation
```latex
@inproceedings{wang2024craft,
  title={Connecting the Dots: Collaborative Fine-tuning for Black-Box Vision-Language Models},
  author={Wang, Zhengbo and Liang, Jian and He, Ran and Wang, Zilei and Tan, Tieniu},
  booktitle={Proceedings of International Conference on Machine Learning (ICML)},
  year={2024}
}
```

## Contact

If you have any question, feel free to contact 📫zhengbowang@mail.ustc.edu.cn.

CraFT

CraFT 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

CraFT

CraFT 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README