{
  "schema_version": 1,
  "builder_plan": {
    "scenario": "An agent looks good in demos, but you cannot tell which product tasks it fails.",
    "change": "Use the study's evaluation idea to inspect a small set of tasks with human-checked answers.",
    "baseline": "Your current evaluator and a manually reviewed reference set.",
    "steps": [
      "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.",
      "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.",
      "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.",
      "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test."
    ],
    "measure": "Agreement with human review, missed failures, false alarms, and evaluation cost per task.",
    "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation."
  },
  "builder_review": null,
  "arxiv_id": "2609.28991",
  "title": "Beneath the Scores: Rethinking Hallucination Evaluation for Video Understanding Models",
  "summary": "The method replaces reliance on aggregate benchmark scores for diagnosing hallucinations in multi-stage video LLM agents with a causal intervention protocol that corrupts or overwrites individual stages (temporal grounding, visual observation, reasoning) while holding the downstream task fixed. It costs evaluation engineering effort: instrumenting the agent to allow stage-level intervention and running many controlled trials (the paper used 60,008 runs across three architectures), which is feasible on a single GPU or via APIs but adds nontrivial evaluation budget. It can fail if the intervention setup does not match the production architecture's stage boundaries, and the finding that benchmark scores do not predict causal cascade sensitivity means teams must build their own intervention harness rather than reuse existing benchmarks.",
  "rationale": "For teams deploying multi-stage video agents, the result that grounding errors dominate downstream hallucinations (about 4x the impact of visual corruption) and that incorrect evidence is worse than missing evidence is directly actionable for prioritizing debugging and guardrail placement, but the intervention protocol itself must be validated against the team's specific agent pipeline before adoption.",
  "technique": "Causal stage-intervention evaluation for video agents",
  "family": "avaliacao_de_agente",
  "recommendation": "testar",
  "published": "2026-09-24",
  "as_of": "2026-10-18",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2609.28991",
  "independent_implementations": 0,
  "total_implementations": 0,
  "stars": 0,
  "citations": 0,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "The method replaces reliance on aggregate benchmark scores for diagnosing hallucinations in multi-stage video LLM agents with a causal intervention protocol that corrupts or overwrites individual stages (temporal grounding, visual observation, reasoning) while holding the downstream task fixed. It costs evaluation engineering effort: instrumenting the agent to allow stage-level intervention and running many controlled trials (the paper used 60,008 runs across three architectures), which is feasible on a single GPU or via APIs but adds nontrivial evaluation budget. It can fail if the intervention setup does not match the production architecture's stage boundaries, and the finding that benchmark scores do not predict causal cascade sensitivity means teams must build their own intervention harness rather than reuse existing benchmarks.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2609.28991",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "exposure_map": [
    {
      "dimension": "quality",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compute",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "latency",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "operations",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compatibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "security",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "data_and_training",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "reproducibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "risks": [],
  "minimum_test": [],
  "open_questions": [],
  "independent_tests": [],
  "lab_reproductions": [],
  "report_available": false,
  "validation_tier": "unknown",
  "evidence_tier": "unknown",
  "infrastructure_basis": "unknown",
  "equations": [],
  "equations_status": "not_fetched",
  "core_kind": "",
  "equations_fetched_at": "",
  "equations_source_url": "",
  "equations_source_sha256": "",
  "exposures_reviewed_at": "",
  "exposures_source_sha256": ""
}