Search papers, labs, and topics across Lattice.
This paper introduces Wuying-Browser-Agent, a comprehensive framework designed to enhance browser agents' performance in real-world scenarios by addressing execution, supervision, optimization, and evaluation challenges. The framework incorporates structured execution primitives, a specialized training curriculum for complex UI interactions, and an innovative reward shaping method to improve long-horizon decision-making. The results show that Wuying-Browser-Agent-27B sets new benchmarks in browser-use tasks, achieving state-of-the-art performance while also demonstrating robust generalization capabilities across various agentic tasks.
Wuying-Browser-Agent achieves a groundbreaking 80.6% on WebVoyager, revealing that real-world browser agents can excel beyond short, clean demonstrations.
Browser agents perform well on short, clean demonstrations, but real deployment is fundamentally different: agents must sustain dozens of decisions on live websites while recovering from mistakes and navigating complex UIs. We argue that closing this gap requires alignment at every level of the pipeline, including execution, supervision, optimization, and evaluation, rather than scale alone. We present Wuying-Browser-Agent, a unified framework that addresses each of these levels. A structured browser harness provides stable execution primitives and decision-oriented context management. Reflection and UI-specialized Curriculum SFT (RUIC-SFT) explicitly trains on recovery trajectories and complex-UI interactions. Divergence-Aware Online GRPO (DAO-GRPO) improves long-horizon credit assignment through potential-based reward shaping and divergence-aware step weighting. Finally, we introduce BrowserBench, a bilingual real-web benchmark of 350 tasks averaging 37.9 steps, because most existing benchmarks are too short to expose long-horizon failure modes. Wuying-Browser-Agent-27B achieves 80.6\% on WebVoyager, 66.7\% on Online-Mind2Web, and 65.1\% on BrowserBench, establishing a new open-source state of the art on browser-use benchmarks. The same pipeline also transfers beyond browser use, demonstrating strong general agentic ability and reaching an average score of 73.8 on Tau2-Bench, Claw-Eval, and BFCL-v4.
