About the Code

[AmingWu]

Dear Authors,

Have you released all the code?

[whcpumpkin]

Hi,
I am still refactoring the code. The current code does not include training and testing code. I will release all the code as soon as possible.

[AmingWu]

Dear Authors,

How long does your method need to train?

[whcpumpkin]

Dear Authors,

How long does your method need to train?

Hi,
On an A100 40G, each epoch takes about 4 hours, and approximately 20 epochs are needed to obtain the optimal model.
I think that increasing the number of workers in the dataloader can effectively reduce the train time, but this requires a sufficient number of CPU cores.

[AmingWu]

Does this code support multiple GPUs?

[whcpumpkin]

Does this code support multiple GPUs?

Currently, it only supports single GPU.

[AmingWu]

Dear Authors,

How to perform test?

[whcpumpkin]

Dear Authors,

How to perform test?

Hi,
I released the testing code just now.
If it has any problem, please let me know.