workaround for tool calling #253
Merged
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
Don't exclude the vision embeddings when regenerating for a tool call. Signed-off-by: kingbri <[email protected]>
|
For now, this should be okay. The draft model issue can be tested and fixed at a later date. |
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Outdated suggestions cannot be applied.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.
Is your pull request related to a problem? Please describe.
Related to #234.
Why should this feature be added?
Not a proper fix, but could be merged into a local branch if anyone needs a quick workaround.
Examples
Enables tool calling again. However, it still raises an error when combined with draft models. To work around that background_drop should be called early. I'm having a hard time understanding the code in the time I have, and SD+TC is probably only wanted by a few people, so this is all I have for now.
Additional context
I tried disabling the draft model just before generating the prompt for the tool call (mainly to not have to relaunch TabbyAPI to switch from using a draft for coding and back to using tool calling), but I got into a tool call loop. That's not the way to fix this anyway.