Question Answering on TATQA
10.2F1Oracle
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Oracle2025.09 | 10.2 | 26.4 | |
| OursBackbone=GPT-4o2025.09 | 9.91 | 22 | |
| OursBackbone=Qwen2.52025.09 | 9.78 | 22 | |
| Ours-cBackbone=GPT-4o2025.09 | 9.52 | 21.2 | |
| Ours-cBackbone=Qwen2.52025.09 | 9.44 | 21.8 | |
| OursBackbone=LLaMA-32025.09 | 9.36 | 21.2 | |
| Ours-cBackbone=LLaMA-32025.09 | 9 | 19.4 | |
| Hybrid2025.09 | 8.72 | 19 | |
| Agent-BasedBackbone=Qwen2.52025.09 | 8 | 18.8 | |
| Rule AgentBackbone=GPT-4o2025.09 | 8 | 18.2 | |
| Score AgentBackbone=GPT-4o2025.09 | 7.96 | 19.6 | |
| Rule AgentBackbone=Qwen2.52025.09 | 7.9 | 18.2 | |
| OursBackbone=GPT-4.12025.09 | 7.79 | 18.4 | |
| Ours-cBackbone=GPT-4.12025.09 | 7.54 | 17.8 | |
| Score AgentBackbone=Qwen2.52025.09 | 7.51 | 18 | |
| Rule AgentBackbone=LLaMA-32025.09 | 7.47 | 17.6 | |
| Score AgentBackbone=LLaMA-32025.09 | 7.37 | 18.8 | |
| Score AgentBackbone=GPT-4.12025.09 | 7.26 | 17.8 | |
| Rule AgentBackbone=GPT-4.12025.09 | 7.09 | 18 | |
| Agent-BasedBackbone=GPT-4o2025.09 | 6.93 | 13.8 | |
| Rule-Based2025.09 | 6.56 | 17 | |
| Doc2025.09 | 6.08 | 10 | |
| Agent-BasedBackbone=GPT-4.12025.09 | 6.02 | 15.8 | |
| Agent-BasedBackbone=LLaMA-32025.09 | 5.76 | 13 | |
| DB2025.09 | 5.3 | 15 | |
| Basic QA2025.09 | 3.5 | 5 | |
| Adaptive-RAG2025.09 | 3.45 | 18 |