Add inspect-ai backend support for MT-Bench (English + Finnish)#9
Open
ezosa wants to merge 1 commit into
Open
Conversation
Adds native inspect-ai task variants so MT-Bench can run via the inspect backend, which only uses an inspect scorer (it ignores lighteval JudgeLLM metrics). Each task does true 2-turn generation + per-turn LLM-judge scoring. - mt_bench/main_inspect.py, mt_bench_fi/main_inspect.py: custom multi-turn solver (generate turn 1, append turn-2 question, generate turn 2) plus two per-turn scorers that grade via a separate judge model reached over the OpenAI-compatible API (SCORER_MODEL_BASE_URL), reusing the existing 1-5 MT-Bench rubric and <score> parsing. - main_inspect.py: fix GenerateConfig kwargs for newer inspect-ai (frequence_penalty -> frequency_penalty, log_probs -> logprobs, drop response_format).
Collaborator
|
@ezosa added some comments I noticed. I think the linting should also be fixed so the CI tests pass and then it's good to go |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds native inspect-ai task variants so MT-Bench can run via the inspect backend, which only uses an inspect scorer (it ignores lighteval JudgeLLM metrics). Each task does true 2-turn generation + per-turn LLM-judge scoring.