Open
Conversation
6a57908 to
b6bd9e0
Compare
Collaborator
|
Sorry, I'm still mulling over how to best handle this. Ideally we would support some kind of "raw" column which gets merged with the other columns in the formatter. The problem here is that each jsonl row is a raw request so making each row a single key under a new dict level would necessitate changing the dataset format. |
Signed-off-by: rebel-jinhwan <jinhwan.suk@rebellions.ai>
Signed-off-by: rebel-jinhwan <jinhwan.suk@rebellions.ai>
Signed-off-by: rebel-jinhwan <jinhwan.suk@rebellions.ai>
Signed-off-by: rebel-jinhwan <jinhwan.suk@rebellions.ai>
070558d to
1c2016e
Compare
This file contains hidden or 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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
This PR adds support for multi-turn chat messages in GuideLLM. When the dataset provides a
messages(orconversations/conversation) column, the OpenAI-style request handler uses that column as the full conversation for each request instead of building a single-turn message fromprefix_column+text_column+ media columnsDetails
"messages_column"to the literal type for dataset column types.src/guidellm/data/preprocessors/mappers.py):messages_columnmaps from names["messages", "conversations", "conversation"]so datasets with those column names are recognized.src/guidellm/backends/openai/request_handlers.py): Informat(), when building the request body:raw_messages = data.columns.get("messages_column", []).raw_messagesis a non-empty list and the first element is a list of dicts withroleandcontent, setarguments.body["messages"] = conversationand return.messagesfromprefix_columnandtext_column(and image/video/audio columns) as a single system + user turn.No changes to scheduler or backend interfaces are required for this; the backend still receives a single request per row, with that request’s
body["messages"]already containing the full conversation whenmessages_columnis used.Test Plan
Related Issues
Use of AI
## WRITTEN BY AI ##)