List your company

Multi-Hop Question Answering on 2Wiki (Exact Match (EM))

46.9Exact Match (EM)Thinker-Instruct
21.21227.88134.5541.219May 12, 2026May 13, 2026May 14, 2026May 15, 2026May 16, 2026May 17, 2026May 18, 2026

Evaluation results

MethodLinks
Thinker-InstructTraining Pipeline=Synthesize → SFT, External Teacher=Qwen2.5-72B, Backbone=Qwen2.5-7B2026.0546.9
SD-Search-InstructTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0544.1
SD-Search-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0543.8
MR-Search-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0543.6
SKILLGRAPHEvaluation Protocol=Zero-shot2026.0543.4
EvolveREvaluation Protocol=Zero-shot2026.0542
Search-R1-InstructTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0541.4
AutoRefine-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0540.5
SkillRLEvaluation Protocol=Zero-shot2026.0540.3
Search-R1Evaluation Protocol=Zero-shot2026.0540.1
StepSearch-BaseTraining Pipeline=Pure RL, External Teacher=GPT-4o, Backbone=Qwen2.5-7B2026.0538.5
StepSearch-InstructTraining Pipeline=Pure RL, External Teacher=GPT-4o, Backbone=Qwen2.5-7B2026.0536.6
ZeroSearchEvaluation Protocol=Zero-shot2026.0535.2
Search-R1-BaseTraining Pipeline=Pure RL, External Teacher=None, Backbone=Qwen2.5-7B2026.0527.2
Search-o1Evaluation Protocol=Zero-shot2026.0527
RAGEvaluation Protocol=Zero-shot2026.0523.2
CoTEvaluation Protocol=Zero-shot2026.0522.6
Qwen2.5-7B-InstructEvaluation Protocol=Zero-shot2026.0522.2