Health Misinformation Detection on HealthFC
84Macro F1 ScoreiPOE-h
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| iPOE-hModel=Qwen3-4B, Protocol=Zero-shot2026.05 | 84 | — | — | |
| iPOE-lb-llmModel=Qwen3-4B, Protocol=Zero-shot2026.05 | 84 | — | — | |
| iPOE-hModel=LLama3-8B, Protocol=Zero-shot2026.05 | 84 | — | — | |
| iPOE-lb-hModel=Qwen3-4B, Protocol=Zero-shot2026.05 | 83 | — | — | |
| Multi-Agent DebateEvaluation stages=1st stage + 2nd stage2025.11 | 82.4 | 82.1 | 82.7 | |
| StepByStep2025.11 | 81 | 72.6 | 91.6 | |
| iPOE-llmModel=Qwen3-4B, Protocol=Zero-shot2026.05 | 80 | — | — | |
| iPOE-hModel=Qwen3-30B, Protocol=Zero-shot2026.05 | 80 | — | — | |
| iPOE-llmModel=LLama3-8B, Protocol=Zero-shot2026.05 | 79 | — | — | |
| iPOE-llmModel=Qwen3-30B, Protocol=Zero-shot2026.05 | 79 | — | — | |
| iPOE-lb-llmModel=Qwen3-30B, Protocol=Zero-shot2026.05 | 79 | — | — | |
| WebAgent2025.11 | 78.1 | 78 | 78.3 | |
| iPOE-lb-llmModel=LLama3-8B, Protocol=Zero-shot2026.05 | 78 | — | — | |
| Rand-hModel=Qwen3-30B, Protocol=Zero-shot2026.05 | 77 | — | — | |
| iPOE-lb-hModel=Qwen3-30B, Protocol=Zero-shot2026.05 | 77 | — | — | |
| iPOE-lb-hModel=LLama3-8B, Protocol=Zero-shot2026.05 | 76 | — | — | |
| Rand-llmModel=Qwen3-30B, Protocol=Zero-shot2026.05 | 76 | — | — | |
| Agreement Score PredictionEvaluation stages=1st stage only2025.11 | 74.3 | 76.9 | 73.4 | |
| VanillaModel=Qwen3-30B, Protocol=Zero-shot2026.05 | 71 | — | — | |
| Rand-llmModel=Qwen3-4B, Protocol=Zero-shot2026.05 | 65 | — | — | |
| Rand-llmModel=LLama3-8B, Protocol=Zero-shot2026.05 | 63 | — | — | |
| VanillaModel=Qwen3-4B, Protocol=Zero-shot2026.05 | 60 | — | — | |
| VanillaModel=LLama3-8B, Protocol=Zero-shot2026.05 | 48 | — | — | |
| Rand-hModel=Qwen3-4B, Protocol=Zero-shot2026.05 | 46 | — | — | |
| Rand-hModel=LLama3-8B, Protocol=Zero-shot2026.05 | 38 | — | — |