Use sharded_state_dict_default in MLP.sharded_state_dict#4693

gdengk wants to merge 2 commits into

NVIDIA:mainNVIDIA/Megatron-LM:mainfrom

gdengk:gdeng/main-pr-4325-activation-mlp-state-dictgdengk/Megatron-LM:gdeng/main-pr-4325-activation-mlp-state-dictCopy head branch name to clipboard