Fix generating partially valid tokens #3
Conversation
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
Matching a regex partially can lead to generating a token which causes the whole generated sequence to be invalid.
When using partial=True, we ensure we don't generate invalid output, but also this makes it impossible to generate certain output sequences. Therefore it's necessary to allow generating tokens which match only partially, and then take the substring of that token which matches the regex.
mattiasarro
changed the title
|
I'm interested in this solution too, as I was having the same parserllm issues as in: r2d4/parserllm#4 Also, the outlines project may interest you. They precompile valid continuations, and then inference happens in The issue I have with |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
Matching a regex partially can lead to generating a token which causes the whole generated sequence to be invalid, even if a substring of the token would result in a valid output.
The other option would be to tweak
complete_rewe run theif stop_after_match:block after every character of the token (rather than the full token text) to the output text, but that's less clean. Or is that needed to be able to generate some output sequences which can only occur by generating a larger invalid token and then pruning the output?Edit: looks like we need the latter approach, see latest commit.