List your company

Short-form Question Answering on TriviaQA (test)

50Exact MatchVSPO + PRS_short
16.210424.982733.75542.5273Dec 8, 2025

Evaluation results

MethodLinks
VSPO + PRS_shortBackbone=Qwen2.5-3B-Instruct2025.1250
GRPO + PRS_shortBackbone=Qwen2.5-3B-Instruct2025.1249.87
PPO + PRS_shortBackbone=Qwen2.5-3B-Instruct2025.1248.87
SFTBackbone=Qwen2.5-3B-Instruct2025.1229.2
UntrainedBackbone=Qwen2.5-3B-Instruct2025.1217.51