List your company

Web Navigation and E-commerce on WebShop In-Distribution

94.73Average ScoreSkillFlow
49.042860.903972.76584.6261May 13, 2026May 20, 2026May 27, 2026Jun 3, 2026Jun 10, 2026Jun 17, 2026Jun 25, 2026

Evaluation results

MethodLinks
SkillFlowMethod Category=Ours2026.0594.7393.75
SkillRLMethod Category=Agent+RL2026.0587.9682.81
FlowSteerMethod Category=Agent+RL2026.0585.4379.2
v4-flashMethod Category=Baseline2026.0585.1767.97
AgentFlowMethod Category=Agent+RL2026.0583.6174.22
Task-Perturbed NLL OptimizationBackbone=Qwen3-8B2026.0681—
Task-Perturbed NLL OptimizationBackbone=Qwen3-4B2026.0677—
GRPOBackbone=Qwen3-8B2026.0676—
GRPOBackbone=Qwen3-4B2026.0674—
Qwen3.5-AFlowMethod Category=AFlow, Backbone=Qwen3.52026.0571.0959.38
Qwen3.5-9BMethod Category=Baseline2026.0556.9332.03
Qwen3.5-GRPOMethod Category=GRPO, Backbone=Qwen3.52026.0554.136.71
Qwen3.5-SFTMethod Category=SFT, Backbone=Qwen3.52026.0550.835.16