Multi-hop Question Answering on MorehopQA (AUROC, AUARC)
0.6457AUROCMATU
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MATULLM=Qwen2.5-7B2026.04 | 0.6457 | 0.7029 | |
| MATULLM=Llama3.1-8B2026.04 | 0.6262 | 0.6493 | |
| SAUP-MultipleLLM=Qwen2.5-7B2026.04 | 0.6242 | 0.6914 | |
| P(true)LLM=Qwen2.5-7B2026.04 | 0.6205 | 0.6853 | |
| P(true)LLM=Llama3.1-8B2026.04 | 0.6158 | 0.6416 | |
| Eigv(Agre)-finalLLM=Llama3.1-8B2026.04 | 0.611 | 0.6369 | |
| SAUP-SingleLLM=Qwen2.5-7B2026.04 | 0.6088 | 0.677 | |
| Eigv(Agre)-finalLLM=Qwen2.5-7B2026.04 | 0.6079 | 0.6741 | |
| SAUP-MultipleLLM=Llama3.1-8B2026.04 | 0.6055 | 0.6322 | |
| Eigv(Agre)-WholeLLM=Llama3.1-8B2026.04 | 0.6034 | 0.63 | |
| Eigv(Agre)-WholeLLM=Qwen2.5-7B2026.04 | 0.6021 | 0.6681 | |
| SAUP-SingleLLM=Llama3.1-8B2026.04 | 0.5918 | 0.6277 | |
| MATULLM=GPT-4o2026.04 | 0.5671 | 0.4336 | |
| P(true)LLM=GPT-4o2026.04 | 0.5455 | 0.4121 | |
| Eigv(Agre)-finalLLM=GPT-4o2026.04 | 0.5257 | 0.3992 | |
| Eigv(Agre)-WholeLLM=GPT-4o2026.04 | 0.5203 | 0.401 |