You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
How should we understand data parallelism and fully sharded data parallelism?They should be a definition in terms of the previous understanding?
Another question is that SP and TP constitute MP groups. Does it mean that as many SP groups as many complete model copies are needed?
The text was updated successfully, but these errors were encountered:
How should we understand data parallelism and fully sharded data parallelism?They should be a definition in terms of the previous understanding?
Another question is that SP and TP constitute MP groups. Does it mean that as many SP groups as many complete model copies are needed?
The text was updated successfully, but these errors were encountered: