[fix] restrict per request streaming to rolling batch use-cases #1670
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
xyang16
reviewed
Mar 26, 2024
| @@ -189,6 +188,8 @@ def parse_input(self, inputs): | |||
|
|
|||
| if not "output_formatter" in param: | |||
| param["output_formatter"] = self.config.output_formatter | |||
| if self.rolling_batch: | |||
| param["stream"] = input_map.pop("stream", False) | |||
There was a problem hiding this comment.
trtllm is only rolling batch right now, but i know @sindhuvahinis is adding support for dynamic batch there.
Sindhu, should i add the non rolling batch condition in trtllm as well? Not sure how the changes you will be raising will look, so maybe you can add it in your PR as well?
There was a problem hiding this comment.
Yeah will add it in my PR.
xyang16
approved these changes
Mar 26, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
Fix an issue with "stream" being populated for dynamic batch use-cases.
We support per request streaming only for rolling batch (I think it may be possible with dynamic batching, but only in the case where we serve 1 request at a time, so it's probably not worth it).