Short-form Question Answering on Bamboogle (test)
0.3489EMVSPO + PRS_short
Evaluation Results
| Method | Links | |
|---|---|---|
| VSPO + PRS_shortBackbone=Qwen2.5-3B-Instruct2025.12 | 0.3489 | |
| GRPO + PRS_shortBackbone=Qwen2.5-3B-Instruct2025.12 | 0.3333 | |
| PPO + PRS_shortBackbone=Qwen2.5-3B-Instruct2025.12 | 0.3325 | |
| UntrainedBackbone=Qwen2.5-3B-Instruct2025.12 | 0.1667 | |
| SFTBackbone=Qwen2.5-3B-Instruct2025.12 | 0.112 |