Files

T

iomgaa 271d1682c9 feat(question_gen): add lightweight 4-gate quality check

Implement 4 concurrent LLM-based quality gates for generated questions:
- key_verify: validates answer evidence in source material
- blind_answer: rejects questions answerable without video context
- multi_true: detects ambiguous multi-correct options
- leak_test: per-family shortcut detection (5 probe templates)

Includes run_gates orchestrator with verbatim_ratio short-circuit,
JSON response parsing with fallback, and 9 unit tests (all passing).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

2026-07-11 23:30:13 -04:00

938 B

Raw Permalink Blame History

Leak Test: Frequency Shortcut (REASONING Family)

You are detecting whether a question exploits word frequency patterns as a shortcut.

Question

{question}

Options

{options}

Designated Correct Answer

{answer}

Instructions

Check if the question can be answered by picking the option that shares the most words or phrases with the question stem, without genuine causal/logical reasoning.

Does the correct answer have significantly more word overlap with the question than distractors?
Are distractor options phrased in notably different vocabulary from the question?
Could a student "game" this question by matching keywords between question and options?

If frequency-based word matching is an exploitable shortcut → verdict "fail". If the question requires genuine reasoning → verdict "pass".

Response Format (strict JSON)

{{"verdict": "pass" or "fail", "reason": "brief explanation"}}

938 B Raw Permalink Blame History