readme.md

Data Preparation

We provide two ways for preparing VTAB-1k:

• Download the source datasets, please refer to NOAH.
• We provide the prepared datasets, which can be download from google drive. Download FGVC, Please refer to VPT.

After that, the file structure should look like:

$ROOT/data
|-- cifar
|-- caltech101
......
|-- FGVC/
    |-- CUB_200_2011
    ......
|-- diabetic_retinopathy

• Download the pretrained ViT-B/16 to ./ViT-B_16.npz

Environment settings

conda create -n cvpt python=3.7

conda activate cvpt

conda install pytorch==1.10.0 torchvision==0.11.0 torchaudio==0.10.0 cudatoolkit=10.2 -c pytorch

pip install timm==0.5.4

pip install avalanche-lib==0.2.1

Testing

bash scripts/test_all.sh

Noting

Influenced by the dataset, the best result of CIFAR we reported is run under the VPT project.

The comparative experiments with VPT in the paper are carried out under our project.