# Create a conda environment
conda create -n dassl python=3.8

# Activate the environment
conda activate dassl

# Install torch (version >= 1.8.1) and torchvision
conda install pytorch==1.9.0 torchvision==0.10.0 torchaudio==0.9.0 cudatoolkit=11.1 -c pytorch -c nvidia

# Install dependencies
pip install -r requirements.txt

Please follow the CoOp instructions: DATASETS.md.

First, you need set the dataset path in ./scripts/***.sh. Then, navigate to the scripts folder:

cd ./scripts

Run the following command, where gpuid specifies the GPU ID you want to use.

bash base2new.sh gpuid

For domain generalization and cross-dataset evaluation, we first train on ImageNet with multiple GPUs, as shown below:

cd ./scripts
bash xd_train.sh gpuid1,gpuid2

Next, we perform Domain Generalization and Cross-dataset evaluation on the new dataset.

bash xd_test_cross.sh gpuid1,gpuid2
bash xd_test_dg.sh gpuid1,gpuid2

For few-shot evaluation, run the following command.
The first two arguments specify the GPU IDs, and the third argument sets the number of shots.

bash few-shot.sh gpuid1,gpuid2 shots