In PyTorch, DTensor typically requires creating multiple processes that run the same code concurrently. However, in some cases, you might want to avoid starting multiple processes in your main code and instead use this feature directly, similar to Jax Sharding.
-
Notifications
You must be signed in to change notification settings - Fork 0
abdussamettrkr/SPDTensor
About
Single Process Jax like sharding for PyTorch Dtensors
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published