Benchmark
PromptEmbeddingMap

Starts an asynchronous computation of the benchmark's prompt embedding map.

Changed on

Restricted to platform staff. Computing a map is what decides whether a benchmark exposes one on the anonymous read route, so it is a curation call rather than a benchmark-owner capability — a benchmark's owner cannot trigger or re-trigger it.

Returns a tracking id immediately; poll GET /benchmark/{benchmarkId}/prompt-embedding-map/status for progress. Responds with 409 Conflict when no embedding provider is configured, or when a computation for this benchmark is already in flight. Any map already being served stays in place until the new one succeeds, so triggering a recomputation is safe.

post/benchmark/{benchmarkId}/prompt-embedding-map

Request

  • Base URL
    https://api.rapidata.ai
  • URL
    https://api.rapidata.ai/benchmark/{benchmarkId}/prompt-embedding-map
  • Auth
    OpenID Connect (https://auth.rapidata.ai/.well-known/openid-configuration, scopes: openid, profile, email, offline_access, api)

Path parameters

benchmarkIdstring required

The benchmark whose prompts should be embedded.

Response

OK

idstring required

The id of the queued prompt embedding map.

Changes