fdb89a6eb2fb
Queries a benchmark's prompt ratings — a page of prompts, each with its ranked entries and the pairwise win matrix over them.
Filters apply to the underlying votes and use the same field set as the standings query. Ordering (extremeness by default) selects which prompts a page fits; a rating is an entry's standing against the other entries on its prompt, so a participant filter narrows the entries but does not refit. A prompt with no matching votes yields no entries. groupBy decides what competes. Per image, every image is its own entry and a participant with several images on a prompt appears several times. Per participant, its images become one entry rated from their pooled wins over other participants; comparisons between two of its own images are excluded from the rating, from both matrices, and from responseCount. Grouping changes only how prompts are scored, never which prompts the page holds. matrix.wins and matrix.votes are square and aligned to entries by position: row i, column j holds entry i's count against entry j. Read the axes off entries — the matrix carries no labels of its own.
Path parameters
The id of the benchmark.
Query parameters
Whether to weight ratings by annotator reliability.
What each rated entry of a prompt represents.
What each rated entry represents.
The 1-based page index.
The number of items per page.
Sort fields. Prefix with - for descending order (e.g. -created_at).
Filter by country.
Filter by language.
Filter by gender.
Filter by age_bucket.
Filter by occupation.
Filter by tags.
Filter by participant_id.
Filter by leaderboard_id.
Filter by run_id.
Filter by prompt_identifier.
Filter by voted_at.
How to combine the field filters: "and" (default) requires every filter to match, "or" requires any of them to match.
Response
OK