Datasets:
| { | |
| "meta_info": { | |
| "case_id": "case39_07", | |
| "bench_subset": "NeuroBench-Multi", | |
| "difficult": 1, | |
| "original_dataset": "Multi_QA" | |
| }, | |
| "agent_input": { | |
| "data_path": [], | |
| "instruction": "Answer this standalone multimodal neurophysiology multiple-choice question: What is the purpose of pre-event baseline correction in an ERP analysis? Options: A: It deletes all signal samples after the event. B: It expresses post-event activity relative to a pre-event reference level. C: It guarantees removal of every physiological artifact. D: It changes the recording sampling rate. Select exactly one option, A, B, C, or D. Return only one JSON object with the single field answer and no explanation or extra fields." | |
| }, | |
| "eval_config": { | |
| "parser_prompt": "### ROLE\nYou are a precise JSON Extraction Engine for neurophysiological data. Your sole task is to convert natural language reports into structured JSON data.\n\n### TASK\nExtract the answer choice from the agent's report.\n- answer: exactly one of A, B, C, or D.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks.\n3. DO NOT include conversational text, explanations, or extra fields.\n4. The JSON key must be EXACTLY \"answer\".\n5. The answer value must be exactly one of \"A\", \"B\", \"C\", or \"D\".\n6. If the answer is missing or unclear, return {\"answer\": null}.\n\n### OUTPUT TEMPLATE\n{\"answer\": \"<A | B | C | D or null>\"}", | |
| "metrics": [ | |
| { | |
| "metric_id": "knowledge_answer_accuracy", | |
| "type": "categorical_check", | |
| "target_key": "answer", | |
| "weight": 100, | |
| "params": { | |
| "gt_value": "B" | |
| } | |
| } | |
| ] | |
| } | |
| } | |