Skip to content

Comments

Feature/skip scored#228

Draft
anthonyduong9 wants to merge 1 commit intomeridianlabs-ai:mainfrom
anthonyduong9:feature/skip-scored
Draft

Feature/skip scored#228
anthonyduong9 wants to merge 1 commit intomeridianlabs-ai:mainfrom
anthonyduong9:feature/skip-scored

Conversation

@anthonyduong9
Copy link
Contributor

No description provided.

def _has_scanner_score(transcript_info: TranscriptInfo, scanner_name: str) -> bool:
"""Check if transcript already has a score from this scanner."""
if transcript_info.source_type != EVAL_LOG_SOURCE_TYPE:
return False

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why? Do only inspect logs support scores?
Note that the score could come from the original source (inspect eval log int his case) or from a previous scout scan

state.scores = {scorer_name: Score(value=1)}
return state

return solve # type: ignore[return-value]

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: Can we fix the argument name and remove type: ignore?

f"Available tables: {table_names}"
)

# Verify count first for clearer error messages

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why is this clearer?

If the expected value is [A, B, C] and real value is [A, B] I would much rather see the diff, which would let me see that C is missing. “Expected 3, got 2” seems much less helpful

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I like the exhaustiveness of the tests, but this feel a bit heavy. Can we simplify it? Maybe parametrize some of the tests?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants