Closed bosaiiasliu closed 2 years ago
This is intended. The code is designed to work in cpu not cuda (so that it runs for everyone, even those who do not have a dedicated gpu). You should convert the parts that work on cpu to work on gpu to accelerate the speed.
I found that using some methods like CAM to generate feature importance interpretations is slow, taking about 2 seconds for one image. If I want to generate all the image explanations in the Imagenet dataset, the time it takes seems very long. Or, is there a suitable acceleration method?