Multi-Hop Question Answering on 2WikiMultihopQA (EM, LLM)
46EMDAC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DACBackbone=Qwen3-8B, FT=LoRA2026.06 | 46 | 52.8 | |
| Search-R1 (GRPO/OF)Backbone=Qwen3-8B, FT=Full2026.06 | 43.9 | 50.1 | |
| Search-R1 (PPO/OF)Backbone=Qwen3-8B, FT=Full2026.06 | 42.2 | 47.9 | |
| Search-R1 (GRPO/O)Backbone=Qwen3-8B, FT=Full2026.06 | 40.5 | 46.3 | |
| Search-R1 (PPO/O)Backbone=Qwen3-8B, FT=Full2026.06 | 38.2 | 43.6 | |
| DACBackbone=Qwen2.5-7B-Instruct, FT=LoRA2026.06 | 35.4 | 41.4 | |
| Search-R1 (GRPO/O)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 34.8 | 39.5 | |
| Search-R1 (GRPO/OF)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 34.7 | 40.5 | |
| Search-R1 (PPO/O)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 34.4 | 40 | |
| Search-R1 (PPO/OF)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 32.9 | 38.8 | |
| Search-o1Backbone=Qwen3-8B, FT=-2026.06 | 31.7 | 40.9 | |
| IRCoTBackbone=Qwen3-8B, FT=-2026.06 | 26.3 | 34.4 | |
| Search-o1Backbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 26 | 32.9 | |
| CoTBackbone=Qwen3-8B, FT=-2026.06 | 25.5 | 27.6 | |
| DirectBackbone=Qwen3-8B, FT=-2026.06 | 25.2 | 27.1 | |
| DirectBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 24.6 | 25.8 | |
| CoTBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 21.2 | 22.8 | |
| IRCoTBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 19.5 | 31.1 | |
| Naive RAGBackbone=Qwen3-8B, FT=-2026.06 | 19.3 | 22.7 | |
| Naive RAGBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 17.9 | 21.2 |