| {
|
| "meta_info": {
|
| "case_id": "SEED-V-03.cnt",
|
| "bench_subset": "NeuroBench-Core",
|
| "difficult": 2.0,
|
| "original_dataset": "SEED-V"
|
| },
|
| "agent_input": {
|
| "data_path": "data/core/SEED-V-03.cnt",
|
| "instruction": "Please compare the strongest brain region in the first two 30-second windows using alpha-band relative energy across frontal, central, parietal, and occipital regions. Report the strongest region in each window and describe the region-state transition process. If the strongest region does not change, state that the status is stable. Use Welch PSD estimation. Define relative power as the integrated power in the target frequency band divided by the integrated total power from 0.5 to 45 Hz. Compute relative power separately for each channel, and define each region's relative-power score as the arithmetic mean of the channel-wise relative-power values within that region."
|
| },
|
| "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 three outputs from the agent report:\n1) strongest region in window 1\n2) strongest region in window 2\n3) transition process description\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 must be EXACTLY \"window_1_strongest_region\", \"window_2_strongest_region\", and \"transition_process\".\n5. The first two values must be one of: frontal, central, parietal, occipital, or null.\n6. \"transition_process\" must be a string or null.\n\n### OUTPUT TEMPLATE\n{\"window_1_strongest_region\": <string|null>, \"window_2_strongest_region\": <string|null>, \"transition_process\": <string|null>}",
|
| "metrics": [
|
| {
|
| "metric_id": "window_1_region_accuracy",
|
| "type": "categorical_check",
|
| "target_key": "window_1_strongest_region",
|
| "weight": 30,
|
| "params": {
|
| "gt_value": "occipital"
|
| }
|
| },
|
| {
|
| "metric_id": "window_2_region_accuracy",
|
| "type": "categorical_check",
|
| "target_key": "window_2_strongest_region",
|
| "weight": 30,
|
| "params": {
|
| "gt_value": "occipital"
|
| }
|
| },
|
| {
|
| "metric_id": "transition_process_consistency",
|
| "type": "semantic_consistency",
|
| "target_key": "full_text",
|
| "weight": 40,
|
| "params": {
|
| "judge_prompt": "You are judging semantic consistency for an EEG task that compares the strongest brain region across two time windows. The true strongest region in window 1 is occipital. The true strongest region in window 2 is occipital. Mark the response as correct only if its transition narrative matches the real relationship between these two windows. If the strongest region is the same in both windows, the response must clearly indicate stability or no transition. If the strongest region changes, the response must clearly describe the shift from the first region to the second region. The expected transition pattern is: stable. If the response describes the wrong direction, invents a transition when the state is stable, or claims stability when a shift occurred, mark it as incorrect. Output only a valid JSON object. The JSON must contain exactly one key named status. The value of status must be either True or False. Do not output Markdown, explanations, additional fields, or any natural language."
|
| }
|
| }
|
| ]
|
| }
|
| }
|
|
|
