Repository navigation
Keep role bytes when the template role cache is full - #1483
Merged
martindevans merged 1 commit intoNov 6, 2026
Merged
Conversation
martindevans
approved these changes
Nov 6, 2026
martindevans
left a comment
Member
There was a problem hiding this comment.
Oops, thanks for fixing this!
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
LLamaTemplate.TextMessagecaches the UTF-8 bytes of each role, with a hard limit of 128 entries. The assignmentroleBytes = rArr;was inside theif (roleCache.Count < 128)block, so once the cache was full any new role ended up with emptyRoleBytes. When the template is applied, that empty memory is pinned to a null pointer and passed tollama_chat_apply_templateas the role.This moves the assignment out of the size check, so the converted bytes are always used and only the caching is limited.
Added a test that adds 200 distinct roles to a template and checks every message keeps its role bytes. It doesn't need a model, so it runs without the downloaded test models.