Hello, I would like to use the encoder in the EfficientSam method to extract features from images for downstream tasks. How should I write the code?