General Question Answering on PopQA (EM, LLM)
47.9EMDAC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DACBackbone=Qwen3-8B, FT=LoRA2026.06 | 47.9 | 52.9 | |
| Search-R1 (GRPO/O)Backbone=Qwen3-8B, FT=Full2026.06 | 46.9 | 51.6 | |
| Search-R1 (PPO/OF)Backbone=Qwen3-8B, FT=Full2026.06 | 45.8 | 50.4 | |
| Search-R1 (GRPO/OF)Backbone=Qwen3-8B, FT=Full2026.06 | 45.3 | 50.4 | |
| Search-R1 (PPO/O)Backbone=Qwen3-8B, FT=Full2026.06 | 43.4 | 48 | |
| Search-R1 (GRPO/O)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 42.6 | 47.5 | |
| DACBackbone=Qwen2.5-7B-Instruct, FT=LoRA2026.06 | 41.1 | 47.6 | |
| Search-R1 (GRPO/OF)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 39.9 | 46.5 | |
| Search-R1 (PPO/O)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 39.5 | 45.9 | |
| Naive RAGBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 38.9 | 46.8 | |
| Naive RAGBackbone=Qwen3-8B, FT=-2026.06 | 38.7 | 45.5 | |
| Search-R1 (PPO/OF)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 37.9 | 44.6 | |
| IRCoTBackbone=Qwen3-8B, FT=-2026.06 | 36.3 | 50 | |
| Search-o1Backbone=Qwen3-8B, FT=-2026.06 | 35.7 | 45.3 | |
| Search-o1Backbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 33.6 | 41.1 | |
| IRCoTBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 20.7 | 45.6 | |
| DirectBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 16.5 | 19.2 | |
| CoTBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 16 | 19 | |
| CoTBackbone=Qwen3-8B, FT=-2026.06 | 15.9 | 19.9 | |
| DirectBackbone=Qwen3-8B, FT=-2026.06 | 15.4 | 20.5 |