Is LRfinder and onecycle policy in keras work effectively? Any leads on how to choose appropriate learning rate to improve accuracy in CNN?

@srinijasiro wrote:

I tried to use the same concept of LRfinder and Cyclic learning rate which is demonstarted in fast ai using keras but never found good improvement . Any leads who can suggest why this would happen and is keras good at using one cycle policy?

Posts: 1

Participants: 1

Read full topic