Long-form Question Answering on Biography FullP (100% polluted)
61.87VeriScore F1@kMIRAGE
Evaluation Results
| Method | Links | |
|---|---|---|
| MIRAGEBackbone=GPT-4o-mini2026.07 | 61.87 | |
| No-RAGBackbone=GPT-4o-mini2026.07 | 58.99 | |
| MIRAGEBackbone=Qwen3-8B2026.07 | 55.49 | |
| No-RAGBackbone=Qwen3-8B2026.07 | 53.75 | |
| AstuteRAGBackbone=GPT-4o-mini2026.07 | 47.91 | |
| RetRobust-NLIBackbone=GPT-4o-mini2026.07 | 45.33 | |
| BRIDGEBackbone=GPT-4o-mini2026.07 | 43.5 | |
| SelfRAGBackbone=GPT-4o-mini2026.07 | 38.55 | |
| FLAREBackbone=GPT-4o-mini2026.07 | 37.38 | |
| DRAGINBackbone=GPT-4o-mini2026.07 | 32.09 | |
| MADAMRAGBackbone=GPT-4o-mini2026.07 | 25.48 | |
| RobustRAGBackbone=Qwen3-8B2026.07 | 25.22 | |
| TrustRAGBackbone=GPT-4o-mini2026.07 | 23.42 | |
| RARGBackbone=GPT-4o-mini2026.07 | 22.92 | |
| RAGBackbone=GPT-4o-mini2026.07 | 22.65 | |
| BRIDGEBackbone=Qwen3-8B2026.07 | 21.31 | |
| AstuteRAGBackbone=Qwen3-8B2026.07 | 20.7 | |
| RobustRAGBackbone=GPT-4o-mini2026.07 | 20.27 | |
| SelfRAGBackbone=Qwen3-8B2026.07 | 18.97 | |
| DRAGINBackbone=Qwen3-8B2026.07 | 17.5 | |
| RARGBackbone=Qwen3-8B2026.07 | 16.07 | |
| TrustRAGBackbone=Qwen3-8B2026.07 | 15.68 | |
| MADAMRAGBackbone=Qwen3-8B2026.07 | 15.15 | |
| RAGBackbone=Qwen3-8B2026.07 | 13.24 | |
| RetRobust-NLIBackbone=Qwen3-8B2026.07 | 12.37 | |
| FLAREBackbone=Qwen3-8B2026.07 | 12.26 |