### 🚀 The feature, motivation and pitch DDP has mixed precision: https://github.com/pytorch/pytorch/pull/92882, code example can be found in torch/testing/_internal/distributed/distributed_test.py _test_ddp_native_mixed_precision FSDP2 has ignored_param, `fully_shard(ignore_params=)`. We need DDP mixed precision for ignore_params action: improve `test_fully_shard_ignore_params.py` to cover the above usecase ### Alternatives _No response_ ### Additional context _No response_ cc @H-Huang @awgu @wanchaol @fegin @fduwjj @wz337 @wconstab @d4l3k