Dr. Jia,
hello, I recently read your article FedLPS and the accompanying code, and I have some questions to ask you.
The article claims to solve the problem of training costs caused by multiple tasks on a single client. I would like to ask how you designed the "multi-task" scenario in the code? Is one dataset representing one classification task?
I see that the code is an example experiment of a single dataset. If a dataset represents a task, how does the code implement local parameter sharing across multiple tasks? I did not find the corresponding part in the code.
Dr. Jia, hello, I recently read your article FedLPS and the accompanying code, and I have some questions to ask you.
I'm looking forward to your warm reply.