General Question Answering on TriviaQA (EM, LLM)
66.4Exact Match (EM)DAC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DACBackbone=Qwen3-8B, FT=LoRA2026.06 | 66.4 | 77.8 | |
| Search-R1 (GRPO/OF)Backbone=Qwen3-8B, FT=Full2026.06 | 66.1 | 77.1 | |
| Search-R1 (GRPO/O)Backbone=Qwen3-8B, FT=Full2026.06 | 66 | 76.6 | |
| Search-R1 (GRPO/O)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 62 | 72 | |
| Search-R1 (PPO/OF)Backbone=Qwen3-8B, FT=Full2026.06 | 61.8 | 73.1 | |
| Search-R1 (PPO/O)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 61.7 | 73.1 | |
| DACBackbone=Qwen2.5-7B-Instruct, FT=LoRA2026.06 | 61.2 | 73.5 | |
| Search-R1 (GRPO/OF)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 60.9 | 72.9 | |
| Search-R1 (PPO/OF)Backbone=Qwen2.5-7B-Instruct, FT=Full2026.06 | 60.8 | 71.8 | |
| Search-R1 (PPO/O)Backbone=Qwen3-8B, FT=Full2026.06 | 60.5 | 70.6 | |
| Naive RAGBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 58.9 | 71.4 | |
| Naive RAGBackbone=Qwen3-8B, FT=-2026.06 | 58.5 | 70.3 | |
| IRCoTBackbone=Qwen3-8B, FT=-2026.06 | 57.9 | 74.5 | |
| Search-o1Backbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 52.3 | 65.2 | |
| Search-o1Backbone=Qwen3-8B, FT=-2026.06 | 49.9 | 72.1 | |
| CoTBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 47 | 56.2 | |
| CoTBackbone=Qwen3-8B, FT=-2026.06 | 45.3 | 53.8 | |
| DirectBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 44.8 | 52.3 | |
| DirectBackbone=Qwen3-8B, FT=-2026.06 | 42.4 | 52.5 | |
| IRCoTBackbone=Qwen2.5-7B-Instruct, FT=-2026.06 | 29.8 | 59.4 |