Add chat template for gemma models - #2183
Open
C00kieFact0ry wants to merge 1 commit into
Open
Conversation
Copy link
Copy Markdown
Sorry, something went wrong.
For multimodal models, we need to change the user/assistant roles and add the proper start and end tokens
Copy link
Copy Markdown
Author
|
@abetlen Please check at your convenience🙏. |
Sorry, something went wrong.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
Using the standard Llava15ChatHandler with multimodal Gemma models fail.
Gemma models expect <start_of_turn>user / <start_of_turn>model / <end_of_turn> tokens instead. There is no vision-capable handler subclass for Gemma models, even though the library already follows the pattern of subclassing Llava15ChatHandler for other model families (e.g., ObsidianChatHandler, MoondreamChatHandler, Llava16ChatHandler) in llama_chat_format.py.
Although there is a format_gemma for text based Gemma models, this is bypassed by the chat handler
Solution
For multimodal models, we need to change the user/assistant roles and add the proper start and end tokens using a GemmaChatHandler