Multi-hop Question Answering on HotpotQA (val) (ASR, ASR-G)
92ASRLogicPoison
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LogicPoisonLLM=GPT-4o-mini, RAG Framework=Naive RAG2026.04 | 92 | 91.6 | |
| LogicPoisonLLM=Llama-3.1-8B, RAG Framework=Naive RAG2026.04 | 88.2 | 86.6 | |
| LogicPoisonLLM=Qwen-3-32B, RAG Framework=Naive RAG2026.04 | 85 | 83.8 | |
| LogicPoisonLLM=Qwen-3-32B, RAG Framework=GraphRAG2026.04 | 84.2 | 92 | |
| LogicPoisonLLM=GPT-4o-mini, RAG Framework=GFM-RAG2026.04 | 81 | 78 | |
| LogicPoisonLLM=Llama-3.1-8B, RAG Framework=GFM-RAG2026.04 | 80.6 | 83.8 | |
| LogicPoisonLLM=Llama-3.1-8B, RAG Framework=GraphRAG2026.04 | 79.2 | 91.2 | |
| LogicPoisonLLM=GPT-4o-mini, RAG Framework=GraphRAG2026.04 | 78.4 | 92.2 | |
| LogicPoisonLLM=Qwen-3-32B, RAG Framework=GFM-RAG2026.04 | 76.6 | 76.8 | |
| LogicPoisonLLM=Qwen-3-32B, RAG Framework=HippoRAG22026.04 | 76.6 | 69.8 | |
| LogicPoisonLLM=GPT-4o-mini, RAG Framework=HippoRAG22026.04 | 73.6 | 66.2 | |
| LogicPoisonLLM=Llama-3.1-8B, RAG Framework=HippoRAG22026.04 | 72.6 | 73 | |
| PoisonedRAGLLM=Qwen-3-32B, RAG Framework=HippoRAG22026.04 | 71.8 | 67.6 | |
| PoisonedRAGLLM=GPT-4o-mini, RAG Framework=GFM-RAG2026.04 | 71.6 | 70.4 | |
| PoisonedRAGLLM=Qwen-3-32B, RAG Framework=GFM-RAG2026.04 | 71.6 | 35 | |
| PoisonedRAGLLM=Llama-3.1-8B, RAG Framework=HippoRAG22026.04 | 68.4 | 63.6 | |
| PoisonedRAGLLM=GPT-4o-mini, RAG Framework=HippoRAG22026.04 | 68 | 66 | |
| PoisonedRAGLLM=GPT-4o-mini, RAG Framework=GraphRAG2026.04 | 66.8 | 80 | |
| PoisonedRAGLLM=Qwen-3-32B, RAG Framework=GraphRAG2026.04 | 65.8 | 80.2 | |
| PoisonedRAGLLM=Llama-3.1-8B, RAG Framework=GraphRAG2026.04 | 64.4 | 78 | |
| PoisonedRAGLLM=GPT-4o-mini, RAG Framework=Naive RAG2026.04 | 61.8 | 72.8 | |
| PoisonedRAGLLM=Qwen-3-32B, RAG Framework=Naive RAG2026.04 | 56.8 | 68.2 | |
| PoisonedRAGLLM=Llama-3.1-8B, RAG Framework=GFM-RAG2026.04 | 55.8 | 49.4 | |
| PoisonedRAGLLM=Llama-3.1-8B, RAG Framework=Naive RAG2026.04 | 51 | 58.8 |