Support FullyShardedDataParallel
#24
Labels
enhancement
New features or improvements to existing functionality
Milestone
Describe the Request
Support
FullyShardedDataParallel
for training larger models or on memory constrained hardware. We'll need to consider how the user passes configuration options to FDSP as there are a lot.Sample Code
No response
Additional Code or Information
No response
The text was updated successfully, but these errors were encountered: