Remove legacy Foundational Analysis case paths after numeric ordering

foundational_analysis/cases/case1/case1_01.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "ISRUC_01.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "ISRUC"
-    },
-    "agent_input": {
-        "data_path": "data/core/ISRUC_01.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 11.81567963841674,
-                    "tolerance": 1.181567963841674
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_02.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "ISRUC_02.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "ISRUC"
-    },
-    "agent_input": {
-        "data_path": "data/core/ISRUC_02.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 9.26960186255792,
-                    "tolerance": 0.926960186255792
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_03.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "ISRUC_03.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "ISRUC"
-    },
-    "agent_input": {
-        "data_path": "data/core/ISRUC_03.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 11.932442037349787,
-                    "tolerance": 1.1932442037349789
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_04.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "ISRUC_04.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "ISRUC"
-    },
-    "agent_input": {
-        "data_path": "data/core/ISRUC_04.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 5.878455525394873,
-                    "tolerance": 0.5878455525394873
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_05.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "ISRUC_05.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "ISRUC"
-    },
-    "agent_input": {
-        "data_path": "data/core/ISRUC_05.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 7.919897825685701,
-                    "tolerance": 0.7919897825685701
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_06.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "BCIC2020-3_01.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "BCIC2020-3"
-    },
-    "agent_input": {
-        "data_path": "data/core/BCIC2020-3_01.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 13.560618171116563,
-                    "tolerance": 1.3560624385726978
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_07.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "BCIC2020-3_02.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "BCIC2020-3"
-    },
-    "agent_input": {
-        "data_path": "data/core/BCIC2020-3_02.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 12.075784079421299,
-                    "tolerance": 1.2075783110068763
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_08.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "BCIC2020-3_03.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "BCIC2020-3"
-    },
-    "agent_input": {
-        "data_path": "data/core/BCIC2020-3_03.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 12.149602175318517,
-                    "tolerance": 1.2149599789598067
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_09.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "BCIC2020-3_04.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "BCIC2020-3"
-    },
-    "agent_input": {
-        "data_path": "data/core/BCIC2020-3_04.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 14.652083592026083,
-                    "tolerance": 1.4652082512405658
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_10.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "BCIC2020-3_05.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "BCIC2020-3"
-    },
-    "agent_input": {
-        "data_path": "data/core/BCIC2020-3_05.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 13.166214068045193,
-                    "tolerance": 1.3166212097326537
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_11.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "SEED-V-01.cnt",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "SEED-V"
-    },
-    "agent_input": {
-        "data_path": "data/core/SEED-V-01.cnt",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 4.836610733624954,
-                    "tolerance": 0.4836610733624954
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_12.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "SEED-V-02.cnt",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "SEED-V"
-    },
-    "agent_input": {
-        "data_path": "data/core/SEED-V-02.cnt",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 6.833748557175907,
-                    "tolerance": 0.6833748557175907
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_13.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "SEED-V-03.cnt",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "SEED-V"
-    },
-    "agent_input": {
-        "data_path": "data/core/SEED-V-03.cnt",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 7.544371519964062,
-                    "tolerance": 0.7544371519964063
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_14.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "SEED-V-04.cnt",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "SEED-V"
-    },
-    "agent_input": {
-        "data_path": "data/core/SEED-V-04.cnt",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 13.23650829225669,
-                    "tolerance": 1.3236508292256692
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_15.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "SEED-V-05.cnt",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "SEED-V"
-    },
-    "agent_input": {
-        "data_path": "data/core/SEED-V-05.cnt",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 9.461418170390266,
-                    "tolerance": 0.9461418170390266
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_16.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "Mumtaz2016_01.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "Mumtaz2016"
-    },
-    "agent_input": {
-        "data_path": "data/core/Mumtaz2016_01.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 25.60839634754504,
-                    "tolerance": 2.560839634754504
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_17.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "Mumtaz2016_02.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "Mumtaz2016"
-    },
-    "agent_input": {
-        "data_path": "data/core/Mumtaz2016_02.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 26.963042816030004,
-                    "tolerance": 2.6963042816030005
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_18.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "Mumtaz2016_03.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "Mumtaz2016"
-    },
-    "agent_input": {
-        "data_path": "data/core/Mumtaz2016_03.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 58.63472717305611,
-                    "tolerance": 5.863472717305612
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_19.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "Mumtaz2016_04.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "Mumtaz2016"
-    },
-    "agent_input": {
-        "data_path": "data/core/Mumtaz2016_04.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 21.885057050117236,
-                    "tolerance": 2.188505705011724
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_20.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "Mumtaz2016_05.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "Mumtaz2016"
-    },
-    "agent_input": {
-        "data_path": "data/core/Mumtaz2016_05.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 36.11783061563106,
-                    "tolerance": 3.6117830615631057
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_21.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "MentalArithmetic_01.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "MentalArithmetic"
-    },
-    "agent_input": {
-        "data_path": "data/core/MentalArithmetic_01.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 41.33018074506845,
-                    "tolerance": 4.133018074506846
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_22.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "MentalArithmetic_02.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "MentalArithmetic"
-    },
-    "agent_input": {
-        "data_path": "data/core/MentalArithmetic_02.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 18.30015063387382,
-                    "tolerance": 1.830015063387382
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_23.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "MentalArithmetic_03.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "MentalArithmetic"
-    },
-    "agent_input": {
-        "data_path": "data/core/MentalArithmetic_03.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 23.478518914155703,
-                    "tolerance": 2.3478518914155706
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_24.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "MentalArithmetic_04.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "MentalArithmetic"
-    },
-    "agent_input": {
-        "data_path": "data/core/MentalArithmetic_04.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 40.95759770354007,
-                    "tolerance": 4.0957597703540065
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case1/case1_25.json

@@ -1,27 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "MentalArithmetic_05.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.0,
-        "original_dataset": "MentalArithmetic"
-    },
-    "agent_input": {
-        "data_path": "data/core/MentalArithmetic_05.edf",
-        "instruction": "Please first extract only the EEG channels from the raw signal, and then apply a 0.5-40Hz FIR bandpass filter to these channels. After filtering, obtain each channel's Alpha-band power and total filtered-signal power by integrating the PSD over frequency. Calculate Alpha relative power separately for each channel, then average the channel-wise ratios and report the final percentage value clearly in your response."
-    },
-    "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 'Alpha Relative Power Percentage' from the agent's report.\n\n### STRICT CONSTRAINTS (MANDATORY)\n1. Output ONLY a valid JSON object.\n2. DO NOT include Markdown code blocks (e.g., no backticks or json tags).\n3. DO NOT include any conversational text, explanations, or units (%).\n4. The Key in the JSON must be EXACTLY \"alpha_ratio\".\n5. The Value must be a FLOAT (e.g., if the report says 25.5%, output 25.5).\n6. If the value is missing or cannot be found, return {\"alpha_ratio\": null}.\n\n### OUTPUT TEMPLATE\n{\"alpha_ratio\": <float>}",
-        "metrics": [
-            {
-                "metric_id": "alpha_ratio_accuracy",
-                "type": "numeric_check",
-                "target_key": "alpha_ratio",
-                "weight": 100,
-                "params": {
-                    "gt_value": 18.87056907818833,
-                    "tolerance": 1.8870569078188328
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_01.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "ISRUC_01.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "ISRUC"
-    },
-    "agent_input": {
-        "data_path": "data/core/ISRUC_01.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": 0.306056,
-                    "tolerance": 0.0153028
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "C4-F4"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_02.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "ISRUC_02.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "ISRUC"
-    },
-    "agent_input": {
-        "data_path": "data/core/ISRUC_02.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": 0.431466,
-                    "tolerance": 0.021573300000000004
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "C4-F4"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_03.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "ISRUC_03.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "ISRUC"
-    },
-    "agent_input": {
-        "data_path": "data/core/ISRUC_03.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": 0.304628,
-                    "tolerance": 0.0152314
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "C4-F4"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_04.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "ISRUC_04.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "ISRUC"
-    },
-    "agent_input": {
-        "data_path": "data/core/ISRUC_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": 0.291297,
-                    "tolerance": 0.014564849999999999
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "C3-F3"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_05.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "ISRUC_05.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "ISRUC"
-    },
-    "agent_input": {
-        "data_path": "data/core/ISRUC_05.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": 0.354238,
-                    "tolerance": 0.0177119
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "C3-F3"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_06.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "BCIC2020-3_01.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "BCIC2020-3"
-    },
-    "agent_input": {
-        "data_path": "data/core/BCIC2020-3_01.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": 0.466999,
-                    "tolerance": 0.02334995
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "P3-P5"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_07.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "BCIC2020-3_02.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "BCIC2020-3"
-    },
-    "agent_input": {
-        "data_path": "data/core/BCIC2020-3_02.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": 0.334351,
-                    "tolerance": 0.01671755
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "FC4-FC6"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_08.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "BCIC2020-3_03.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "BCIC2020-3"
-    },
-    "agent_input": {
-        "data_path": "data/core/BCIC2020-3_03.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": 0.446754,
-                    "tolerance": 0.022337700000000002
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "AF7-F5"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_09.json

@@ -1,36 +0,0 @@
-{
-    "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"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_10.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "BCIC2020-3_05.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "BCIC2020-3"
-    },
-    "agent_input": {
-        "data_path": "data/core/BCIC2020-3_05.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": 0.490313,
-                    "tolerance": 0.02451565
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "P1-P3"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_11.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "SEED-V-01.cnt",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "SEED-V"
-    },
-    "agent_input": {
-        "data_path": "data/core/SEED-V-01.cnt",
-        "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": 0.352766,
-                    "tolerance": 0.017638300000000003
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "PO6-PO8"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_12.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "SEED-V-02.cnt",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "SEED-V"
-    },
-    "agent_input": {
-        "data_path": "data/core/SEED-V-02.cnt",
-        "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": 0.347057,
-                    "tolerance": 0.01735285
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "PO3-PO5"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_13.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "SEED-V-03.cnt",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "SEED-V"
-    },
-    "agent_input": {
-        "data_path": "data/core/SEED-V-03.cnt",
-        "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": 0.366734,
-                    "tolerance": 0.0183367
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "P5-PO5"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_14.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "SEED-V-04.cnt",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "SEED-V"
-    },
-    "agent_input": {
-        "data_path": "data/core/SEED-V-04.cnt",
-        "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": 0.474302,
-                    "tolerance": 0.023715100000000003
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "O2-PO4"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_15.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "SEED-V-05.cnt",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "SEED-V"
-    },
-    "agent_input": {
-        "data_path": "data/core/SEED-V-05.cnt",
-        "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": 0.42731,
-                    "tolerance": 0.021365500000000003
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "P4-P6"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_16.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "Mumtaz2016_01.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "Mumtaz2016"
-    },
-    "agent_input": {
-        "data_path": "data/core/Mumtaz2016_01.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": 0.368373,
-                    "tolerance": 0.01841865
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "FP1-FP2"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_17.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "Mumtaz2016_02.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "Mumtaz2016"
-    },
-    "agent_input": {
-        "data_path": "data/core/Mumtaz2016_02.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": 0.495031,
-                    "tolerance": 0.02475155
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "F3-FP1"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_18.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "Mumtaz2016_03.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "Mumtaz2016"
-    },
-    "agent_input": {
-        "data_path": "data/core/Mumtaz2016_03.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": 0.608099,
-                    "tolerance": 0.03040495
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "F4-FZ"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_19.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "Mumtaz2016_04.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "Mumtaz2016"
-    },
-    "agent_input": {
-        "data_path": "data/core/Mumtaz2016_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": 0.404867,
-                    "tolerance": 0.02024335
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "FP1-FZ"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_20.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "Mumtaz2016_05.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "Mumtaz2016"
-    },
-    "agent_input": {
-        "data_path": "data/core/Mumtaz2016_05.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": 0.479867,
-                    "tolerance": 0.02399335
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "F3-FP1"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_21.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "MentalArithmetic_01.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "MentalArithmetic"
-    },
-    "agent_input": {
-        "data_path": "data/core/MentalArithmetic_01.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": 0.463025,
-                    "tolerance": 0.02315125
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "F4-FZ"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_22.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "MentalArithmetic_02.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "MentalArithmetic"
-    },
-    "agent_input": {
-        "data_path": "data/core/MentalArithmetic_02.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": 0.577661,
-                    "tolerance": 0.02888305
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "F7-FP1"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_23.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "MentalArithmetic_03.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "MentalArithmetic"
-    },
-    "agent_input": {
-        "data_path": "data/core/MentalArithmetic_03.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": 0.478912,
-                    "tolerance": 0.0239456
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "C4-CZ"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_24.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "MentalArithmetic_04.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "MentalArithmetic"
-    },
-    "agent_input": {
-        "data_path": "data/core/MentalArithmetic_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": 0.551636,
-                    "tolerance": 0.027581800000000004
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "F4-FZ"
-                }
-            }
-        ]
-    }
-}

foundational_analysis/cases/case10/case10_25.json

@@ -1,36 +0,0 @@
-{
-    "meta_info": {
-        "case_id": "MentalArithmetic_05.edf",
-        "bench_subset": "NeuroBench-Core",
-        "difficult": 1.5,
-        "original_dataset": "MentalArithmetic"
-    },
-    "agent_input": {
-        "data_path": "data/core/MentalArithmetic_05.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": 0.479741,
-                    "tolerance": 0.02398705
-                }
-            },
-            {
-                "metric_id": "highest_corr_pair_accuracy",
-                "type": "categorical_check",
-                "target_key": "top_pair",
-                "weight": 50,
-                "params": {
-                    "gt_value": "F8-FP2"
-                }
-            }
-        ]
-    }
-}