| { |
| "meta_info": { |
| "case_id": "BCIC2020-3_04.edf", |
| "bench_subset": "NeuroBench-Core", |
| "difficult": 1.5, |
| "original_dataset": "BCIC2020-3" |
| }, |
| "agent_input": { |
| "data_path": "data/core/BCIC2020-3_04.edf", |
| "instruction": "Please select EEG channels from the raw signal, apply an 8-13Hz alpha-band FIR (linear-phase) bandpass filter, and then use the first 30 seconds of the filtered signal to compute inter-channel correlations using the Pearson correlation coefficient. Then report: (1) the global synchronization value defined as the mean absolute correlation across all channel pairs, and (2) the single channel pair with the highest correlation coefficient." |
| }, |
| "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 two final outputs from the agent report:\n1) global synchronization value in alpha band\n2) highest-correlation channel pair\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks.\n3. DO NOT include any conversational text or explanations.\n4. Keys in JSON must be EXACTLY \"global_sync\" and \"top_pair\".\n5. \"global_sync\" must be a FLOAT.\n6. To extract \"top_pair\", first identify the two complete source channel labels. A referenced label such as F4-A1 is one complete channel label, and its hyphen must not be treated as the pair separator.\n7. For each source channel separately, uppercase it and remove one terminal reference suffix if present: -A1, -A2, -LE, -REF, -M1, -M2, or -AVG. Never output a reference electrode (A1/A2/LE/REF/M1/M2/AVG) as a pair member.\n8. Convert legacy aliases after removing the reference suffix: T3->T7, T4->T8, T5->P7, T6->P8.\n9. Sort the two normalized channel labels in ascending lexicographic order, then join them with exactly one hyphen using format \"CH1-CH2\" (uppercase, no spaces). If the report already uses a canonical pair such as F4-C4, normalize it to C4-F4. Each final label may contain only uppercase A-Z letters and digits; mixed-case forms such as Fp1 or Fp2 are forbidden and must become FP1 or FP2.\n10. Required examples: \"F4-A1 and C4-A1\" -> \"C4-F4\"; \"F4-C4\" -> \"C4-F4\"; \"T4-A1 and T3-A2\" -> \"T7-T8\"; \"Fp2 and Fp1\" -> \"FP1-FP2\".\n11. If the pair cannot be resolved into exactly two non-reference channels, set \"top_pair\" to null; do not guess. If global_sync is missing, set only global_sync to null.\n\n### OUTPUT TEMPLATE\n{\"global_sync\": <float|null>, \"top_pair\": <string|null>}", |
| "metrics": [ |
| { |
| "metric_id": "alpha_global_sync_accuracy", |
| "type": "numeric_check", |
| "target_key": "global_sync", |
| "weight": 50, |
| "params": { |
| "gt_value": 0.514571, |
| "tolerance": 0.025728550000000003 |
| } |
| }, |
| { |
| "metric_id": "highest_corr_pair_accuracy", |
| "type": "categorical_check", |
| "target_key": "top_pair", |
| "weight": 50, |
| "params": { |
| "gt_value": "F7-FT7" |
| } |
| } |
| ] |
| } |
| } |
|
|
