Skip to content

Conversation

wweic
Copy link
Owner

@wweic wweic commented Aug 1, 2025

No description provided.

return numpy.attr("empty")(0, py::dtype("object"));
}

// First pass: count the number of strings and calculate total size

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is this change to mainly optimize string inputs?

@Swipe4057
Copy link

Will there be a merger of this great feature?

@wweic
Copy link
Owner Author

wweic commented Aug 26, 2025

@Swipe4057 We are working with Nvidia team on triton-inference-server/server#8348. We have engaged with their engineering team.

@yinggeh
Copy link

yinggeh commented Sep 17, 2025

Hi @wweic. You should merge to main instead of r25.05.

@wweic
Copy link
Owner Author

wweic commented Sep 17, 2025

@yinggeh Yeah, this PR is for personal review with my team. I was waiting for triton team's greenlight to upstream so I can prepare the proper PR to main. I just got response here: triton-inference-server/server#8348 . So I guess it's OK to send a PR now to the pythonbackend repo?

@wweic
Copy link
Owner Author

wweic commented Sep 17, 2025

proper PR to triton repo: triton-inference-server#416

@wweic wweic closed this Sep 17, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants