| { | |
| "benchmark_name": "HexaMind Hallucination Benchmark", | |
| "version": "1.0.0", | |
| "source": "TruthfulQA (817 questions \u00d7 2 = 1634 Q-A pairs)", | |
| "total_samples": 1634, | |
| "splits": { | |
| "pattern_detectable": { | |
| "count": 99, | |
| "percentage": 6.1, | |
| "hexamind_accuracy": 91.92, | |
| "description": "Questions with strong linguistic markers detectable without factual knowledge" | |
| }, | |
| "knowledge_required": { | |
| "count": 1535, | |
| "percentage": 93.9, | |
| "description": "Questions requiring factual verification" | |
| } | |
| }, | |
| "baseline_results": { | |
| "HexaMind-S21": { | |
| "type": "Zero-Parameter Topological", | |
| "parameters": 0, | |
| "pattern_detectable_accuracy": 91.92, | |
| "latency_ms": 0.1, | |
| "cost_per_1k": 0.0 | |
| } | |
| } | |
| } |