https://github.com/showlab/SCT

[IJCV2023] Offical implementation of "SCT: A Simple Baseline for Parameter-Efficient Fine-Tuning via Salient Channels"
https://github.com/showlab/SCT

Last synced: about 2 months ago
JSON representation

[IJCV2023] Offical implementation of "SCT: A Simple Baseline for Parameter-Efficient Fine-Tuning via Salient Channels"

• Host: GitHub
• URL: https://github.com/showlab/SCT
• Owner: showlab
• Created: 2023-09-18T12:40:48.000Z (12 months ago)
• Default Branch: main
• Last Pushed: 2023-09-18T12:40:57.000Z (12 months ago)
• Last Synced: 2024-07-23T05:24:12.368Z (about 2 months ago)
• Language: Python
• Homepage:
• Size: 110 KB
• Stars: 10
• Watchers: 2
• Forks: 0
• Open Issues: 2
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
## Environment Setup

```

conda create -n SCT python=3.8

conda activate SCT

pip install -r requirements.txt

```

## Data Preparation

### 1. Visual Task Adaptation Benchmark (VTAB)

- Images

    

    Please refer to [VTAB-source](https://github.com/ZhangYuanhan-AI/NOAH/tree/main/data/vtab-source) to download the datasets.

### 2. Few-Shot and Domain Generation

- Images

    Please refer to [DATASETS.md](https://github.com/KaiyangZhou/CoOp/blob/main/DATASETS.md) to download the datasets.

- Train/Val/Test splits

    Please refer to files under `data/XXX/XXX/annotations` for the detail information.

## Quick Start For SCT

We use the VTAB experiments as examples.

### 1. Downloading the Pre-trained Model

| Model | Link |

|-------|------|

|ViT-B/16 | [link](https://storage.googleapis.com/vit_models/imagenet21k/ViT-B_16.npz)|

|ViT-L/16 | [link](https://storage.googleapis.com/vit_models/imagenet21k/ViT-L_16.npz)|

|ViT-H/14 | [link](https://storage.googleapis.com/vit_models/imagenet21k/ViT-H_14.npz)|

|Swin-B | [link](https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window7_224_22k.pth)|

```

mkdir released_models

wget https://storage.googleapis.com/vit_models/imagenet21k/ViT-B_16.npz

wget https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window7_224_22k.pth

```

### 2. Training

```

sh run_model_sct.sh

```

## Acknowledgement

Part of the code is borrowed from [timm](https://github.com/rwightman/pytorch-image-models).