[ghstack-poisoned]

This PR removes the property `params_with_grad` from `FullyShardedDataParallel`. It was introduced when implementing `clip_grad_norm_()` but was not consistently used. Personally, I do not think it makes sense for `FullyShardedDataParallel` to expose this helper because it is not a common paradigm.

This PR is technically BC-breaking. However, I checked that no one internally is using this API.


[ghstack-poisoned]

This PR removes the property `params_with_grad` from `FullyShardedDataParallel`. It was introduced when implementing `clip_grad_norm_()` but was not consistently used. Personally, I do not think it makes sense for `FullyShardedDataParallel` to expose this helper because it is not a common paradigm.

This PR is technically BC-breaking. However, I checked that no one internally is using this API.


[ghstack-poisoned]

This PR removes the property `params_with_grad` from `FullyShardedDataParallel`. It was introduced when implementing `clip_grad_norm_()` but was not consistently used. Personally, I do not think it makes sense for `FullyShardedDataParallel` to expose this helper because it is not a common paradigm.

This PR is technically BC-breaking. However, I checked that no one internally is using this API.


cc @ezyang @gchanan

[ghstack-poisoned]