List your company

Multi-Hop Question Answering on HotpotQA (EM, ACC)

44.6Exact Match (EM)IGRPO
19.84826.27432.739.126Oct 18, 2025Nov 30, 2025Jan 13, 2026Feb 26, 2026Apr 10, 2026May 24, 2026Jul 7, 2026

Evaluation results

MethodLinks
IGRPOBackbone=Qwen2.5-7B-Instruct, Type=Tree-based2026.0744.6—
IGPOBackbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.0742.9—
GiGPOBackbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.0741.6—
Tree-GRPOBackbone=Qwen2.5-7B-Instruct, Type=Tree-based2026.0739.9—
IGRPOBackbone=Qwen2.5-3B-Instruct, Type=Tree-based2026.0738.9—
StepSearchBackbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.0738.6—
DTKG2025.1038.285.8
TOG2025.1037.183.5
Search-R1Backbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.0737—
GiGPOBackbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.0736.9—
IGPOBackbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.0735.7—
COT2025.1035.681
KGR2025.1035.583.5
Original2025.103568.2
ZeroSearchBackbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.0734.6—
StepSearchBackbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.0734.5—
CRITIC2025.103478
Search-R1Backbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.0732.4—
Tree-GRPOBackbone=Qwen2.5-3B-Instruct, Type=Tree-based2026.0731.1—
ZeroSearchBackbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.0727.4—
R1-InstructBackbone=Qwen2.5-3B-Instruct, Type=RL Training2026.0723.7—
R1-InstructBackbone=Qwen2.5-7B-Instruct, Type=RL Training2026.0720.8—