Short-form Question Answering on Natural Questions (NQ) (test)
36EMVSPO + PRS_short
Evaluation Results
| Method | Links | |
|---|---|---|
| VSPO + PRS_shortBackbone=Qwen2.5-3B-Instruct2025.12 | 36 | |
| GRPO + PRS_shortBackbone=Qwen2.5-3B-Instruct2025.12 | 35.18 | |
| PPO + PRS_shortBackbone=Qwen2.5-3B-Instruct2025.12 | 32.1 | |
| SFTBackbone=Qwen2.5-3B-Instruct2025.12 | 24.9 | |
| UntrainedBackbone=Qwen2.5-3B-Instruct2025.12 | 4.15 |