pytorch

af9c6bc8 - [FSDP] Add `keep_low_precision_grads` support when CPU offloading (#86495)

[FSDP] Add `keep_low_precision_grads` support when CPU offloading (#86495) When CPU offloading, FSDP uses `_cpu_grad`, not `_saved_grad_shard`. This adds support for `keep_low_precision_grads` for that case. Pull Request resolved: https://github.com/pytorch/pytorch/pull/86495 Approved by: https://github.com/rohan-varma