Haodong Duan

2 articles on SOTA Papers

Professional GUI Agents Stall on Long-Horizon Workflows

Workflow-GYM evaluates 338 tasks across 56 virtual-machine software environments, where the best model averages 30.67% success.

Jul 28, 20265 min2606.11042

Automated Research Produces Reviewable Papers, With Weak Evidence

FARS runs ideation, planning, experiments, and writing end to end, producing 166 papers with 282 human reviews exposing quality and integrity gaps.

Jul 28, 20265 min2606.31651 Code available