{ "benchmark_manifest_sha256": "0ea5ab07f46c509a7a24943b31d5e9bfd6368920609227bc47613e18e52c4642", "case_count": 9, "evaluation_results_canonical_json_sha256": "55cdba9cce4dcdc88cee0829cd510d7e624a3472feb9d33f7e681621a792eba3", "evidence_path": "artifacts/evidence/accuracy/P4/runs/p4-2.0.1-9677d0ef37db82bcf39b", "evidence_run_id": "p4-2.0.1-9677d0ef37db82bcf39b", "failure_example_count": 14, "implemented_capability_count": 15, "local_harness_status": "pass", "phase4_done": false, "phase5_ready": false, "phase_decision": "blocked", "product_benchmark_status": "fail", "product_gate_evidence_sha256": "c9f5e3de3ab6d64911f8807b26caa661ccf911a5fd6b995107edda2004836e0c", "promotion_allowed": false, "reference_baseline_sha256": "aeb0f6510352f4e205e4fc4e8c423110ca72c2142016c7af846d758865f258c6", "schema_version": 2, "split_counts": { "background-test": 2, "calibration": 2, "challenge": 4, "test": 7, "train": 3, "val": 3 }, "status": "fail", "task_family_count": 7 }