### 🚀 The feature Should add some tests to ensure the right sharded grad scaler, no_sync ctx manager, etc is picked out when using composable FSDP ### Motivation, pitch . ### Alternatives _No response_ ### Additional context _No response_
🚀 The feature
Should add some tests to ensure the right sharded grad scaler, no_sync ctx manager, etc is picked out when using composable FSDP
Motivation, pitch
.
Alternatives
No response
Additional context
No response