Required prerequisites
Motivation
Hi, TorchOpt team!
I am writing to inquire if there is a function that converts torch.optim state checkpoint 100% to a torchopt object, that can replay completely the same resulting behavior as how the former behaves.
I have searched for function declarations like torchopt.from_torch etc, but seemed not to discover any.
Could you please point me to the function, inform me that it does not exist anywhere, or perhaps suggest a viable near-identical ways to implement this as close and convenient as possible? I'd be greatly indebted!
Solution
No response
Alternatives
No response
Additional context
No response