List your company

Multi-hop Question Answering on 2-wiki out-of-domain transfer

34.2Exact MatchWAPO
12.46418.10723.7529.393Jun 15, 2026

Evaluation results

MethodLinks
WAPOModel=Qwen3-4B2026.0634.2
WAPOModel=SmolLM3-3B2026.0631.4
DAPOModel=SmolLM3-3B2026.0630.3
GSPOModel=SmolLM3-3B2026.0629.8
GSPOModel=Qwen3-4B2026.0629.7
GRPOModel=SmolLM3-3B2026.0628.7
GRPOModel=Qwen3-4B2026.0626.6
WAPOModel=Gemma3-4B2026.0625.6
GSPOModel=Gemma3-4B2026.0624.4
GRPOModel=Gemma3-4B2026.0623.3
DAPOModel=Gemma3-4B2026.0622
DAPOModel=Qwen3-4B2026.0613.3