-
Notifications
You must be signed in to change notification settings - Fork 103
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Tests]: Context and Sequence Parallel #438
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for the MR! I suggest to change names of some files.
@@ -0,0 +1,18 @@ | |||
import os | |||
import sys |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
run_test_diffusers_adapter.py
@@ -0,0 +1,6 @@ | |||
HOST=127.0.0.1 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
change name to run_test_clinet.sh?
@@ -0,0 +1,44 @@ | |||
import os |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
launch_flux_usp_service_example.py?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sure, these were temporal scripts meant for debugging, thus we can directly remove it. Or we can keep it as a blueprint for debugging.
Before merging I wanted to have feedback about adding tests covering all the models supported by xDiT. What do you think? |
Thank you for your thoughtful consideration. I certainly hope you can cover as many model services as possible. |
Guarantee distributed inference integrity with newer
xFuser
versions (e.g. world_size -> dit_parallel_size) and gather benchmarking results. ParaAttention testing is our main reference.