Question Answering on SciFact
100Answerability RateVanilla RAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Vanilla RAG2026.05 | 100 | — | |
| RobustRAG2026.05 | 100 | — | |
| Paraphrase2026.05 | 100 | — | |
| Debate2026.05 | 100 | — | |
| Cordon-MAS2026.05 | 74 | — | |
| TrustRAG2026.05 | 54 | — | |
| Cordon-MASLLM Backbone=DeepSeek2026.05 | — | 86 | |
| Cordon-MASLLM Backbone=GPT-4o2026.05 | — | 88 | |
| Vanilla RAGLLM Backbone=DeepSeek2026.05 | — | 46 |