Replies: 1 comment
-
|
Hi @frankxyy, There is TP for when we use kernel-injection, it is just that the tensor sharding is done in a slightly different way and in different part of the code. If you are interested, you can follow the logic from here: https://github.com/microsoft/DeepSpeed/blob/master/deepspeed/module_inject/replace_module.py#L354 |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
In engine.py script, I found the code above. It seems that for input with kernel injection switch on, the process of autotp will not be executed. Is there a possibility that the two: autotp and kernel injection can both be utilized?
Beta Was this translation helpful? Give feedback.
All reactions