List your company

Multi-Hop Question Answering on Bamboogle (Exact Match (EM))

73.8Exact Match (EM)SkillRL
8.59225.52142.4559.379May 12, 2026May 21, 2026May 30, 2026Jun 9, 2026Jun 18, 2026Jun 27, 2026Jul 7, 2026

Evaluation results

MethodLinks
SkillRLEvaluation Protocol=Zero-shot2026.0573.8
SKILLGRAPHEvaluation Protocol=Zero-shot2026.0572.6
IGRPOBackbone=Qwen2.5-7B-Instruct, Type=Tree-based2026.0771.8
GiGPOBackbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.0768.9
IGRPOBackbone=Qwen2.5-3B-Instruct, Type=Tree-based2026.0765.7
IGPOBackbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.0764.5
GiGPOBackbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.0764.1
EvolveREvaluation Protocol=Zero-shot2026.0554.4
Tree-GRPOBackbone=Qwen2.5-7B-Instruct, Type=Tree-based2026.0741.6
StepSearchBackbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.0740
IGPOBackbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.0739.9
Search-R1Evaluation Protocol=Zero-shot2026.0536.8
Search-R1Backbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.0736.8
StepSearchBackbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.0734.4
Search-o1Evaluation Protocol=Zero-shot2026.0530.4
R1-InstructBackbone=Qwen2.5-3B-Instruct, Type=RL Training2026.0729.3
Tree-GRPOBackbone=Qwen2.5-3B-Instruct, Type=Tree-based2026.0728
ZeroSearchEvaluation Protocol=Zero-shot2026.0527.8
ZeroSearchBackbone=Qwen2.5-7B-Instruct, Type=Chain-based2026.0727.8
Search-R1Backbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.0726.4
CoTEvaluation Protocol=Zero-shot2026.0524
R1-InstructBackbone=Qwen2.5-7B-Instruct, Type=RL Training2026.0719.2
RAGEvaluation Protocol=Zero-shot2026.0516.8
Qwen2.5-7B-InstructEvaluation Protocol=Zero-shot2026.0514.4
ZeroSearchBackbone=Qwen2.5-3B-Instruct, Type=Chain-based2026.0711.1