admin

Invalidate an evaluation run

Invalidate an evaluation run (irreversible).

Marks the run as excluded from scoring and adjusts the work item's included_success_count. Reopens the work item to allow validators to claim replacement work.

post/v1/admin/evaluation-runs/{eval_run_id}/invalidate

Path parameters

eval_run_idstring uuid required

Evaluation run ID

Evaluation run ID

Request body

reasonstring required

Reason for invalidating the run

Response

Successful Response

eval_run_idstring uuid required

Evaluation run ID

agent_version_idstring uuid required

Agent version being evaluated

suite_idinteger required

Problem suite ID

validator_hotkeystring required

Validator performing the evaluation

status'CLAIMED' | 'RUNNING' | 'SUCCESS' | 'FAILED' | 'TIMED_OUT' | 'STALE' | 'CANCELLED' required

Status of an evaluation run through its lifecycle.

claimed_atstring date-time required

When the run was claimed

lease_expires_atstring date-time required

When the lease expires

is_includedboolean required

Whether this run is included in scoring

last_heartbeat_atstring date-time nullable

Last heartbeat from the validator

completed_atstring date-time nullable

When the run completed

invalidated_atstring date-time nullable

When the run was invalidated

invalidated_bystring nullable

Hotkey of the admin who invalidated the run

invalidation_reasonstring nullable

Reason the run was invalidated

validator_scorenumber nullable

Score assigned by the validator

score_components_summaryobject nullable

Summary of score components

Changes

No recorded changes to this endpoint across all 1 revision of this API.