downlad the data from https://drive.google.com/file/d/1YZQFSonulXuagMIfbJkZeTFJ6qEUuUxL/view?usp=sharing and extract files to a ./data folder
Do something like this
.
├── data
│ ├── dataset.json
│ ├── imagesTr
│ ├── imagesTs
│ └── labelsTr
├── V_NAS.py
├── main.py
...
...
...
Model articture for VNAS, this model has ability to update specific parameters
Model articture for searched NAS, you can try differnet combination of model(s) by passing the desired block index list to each Encoder/Decoder
Train the best or worst model
Differentialable articture search