On-policy self-distillation (OPSD) provides dense token-level supervision without a second model: one network acts as teacher with the reference solution and as student with only the problem. We identify a specific failure mode of this recipe. During training, student token entropy rises past the teacher's and remains...
Yi-Fei Liu, Ming-Hao Fang, Xin-Yu Gu et al.· 0 citations
GDPevo is presented, an evolution-native benchmark grounded in GDP-related enterprise workflows, together with the fully automated data pipeline that generates it, and the best evolved agents remain far below the fully informed oracle ceiling, indicating that the self-evolution ability of current agents remains far fro...
Leijun Zhou, Zhihao Liu, Xiang Qu et al.· 1 citation
SciForge is a multimodal research-native AI workbench that reserves the graphical interface for human judgment while search, parsing, model routing, workflow execution, plotting, writing, and presentation generation run as modular agent-accessible services.
SciForge Team Zhangyang Gao, Minghao Fang, Yifei Liu et al.· arXiv.org· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.