From Pareto to Preference: Personalized Test-Time Scaling via Amortized Agentic Policy Discovery ▲ 16 💬 1 2026-10-09 arXiv HF 原文 GitHub ★2 摘要 （无摘要）