I understand how to load/save sharded FSDP state dicts from this tutorial: https://www.youtube.com/watch?v=uBLgprhJn_8.
But, how can I load a sharded state dict onto a single rank?
I understand how to load/save sharded FSDP state dicts from this tutorial: https://www.youtube.com/watch?v=uBLgprhJn_8.
But, how can I load a sharded state dict onto a single rank?