List your company

Deep search QA on WebWalkerQA

23.01AccuracyProCeedRL
18.444419.629720.81522.0003Apr 2, 2026

Evaluation results

MethodLinks
ProCeedRLBackbone=Qwen3-8B2026.0423.01
Rewinding MoreBackbone=Qwen3-8B, Type=ablation2026.0420.16
Qwen3-8B-v3-SFTBackbone=Qwen3-8B2026.0419.85
RFTBackbone=Qwen3-8B2026.0419.56
DAPO/Search-R1Backbone=Qwen3-8B2026.0419.51
ReAct PromptingBackbone=Qwen3-8B2026.0418.62