feat(accuracy): build hardened phase 4 evaluation harness

2026-08-02 02:43:41 +02:00
parent 7b92e29e49
commit bd2fd9780f
13 changed files with 5468 additions and 7 deletions
@@ -28,7 +28,7 @@ from training_release_manifest import (  # noqa: E402
 )


-PROTECTED_SPLITS = {"calibration", "test", "background-test"}
+PROTECTED_SPLITS = {"calibration", "test", "background-test", "challenge"}
 PROPOSAL_DATASET_PROVENANCE_NAME = "proposal-dataset-provenance.json"
 PROPOSAL_DATASET_EVIDENCE_NAME = "proposal-dataset-evidence.json"
 PROPOSAL_DATASET_SCHEMA_VERSION = 1