List your company

Deep Search QA on MuSiQue

29.52AccuracyProCeedRL
18.204821.142424.0827.0176Apr 2, 2026

Evaluation results

MethodLinks
ProCeedRLBackbone=Qwen3-8B2026.0429.52
DeepResearcher2026.0429.3
DAPO/Search-R1Backbone=Qwen3-8B2026.0423.6
RFTBackbone=Qwen3-8B2026.0422.49
Qwen3-8B-v3-SFTBackbone=Qwen3-8B2026.0420.07
Rewinding MoreBackbone=Qwen3-8B, Type=ablation2026.0419.11
GiGPO2026.0418.9
ReAct PromptingBackbone=Qwen3-8B2026.0418.64