ANTE: Anticipate, Then Execute -- Learning Internal World Models for Coding Agents from Verified Beliefs ▲ 0 💬 1 2026-10-17 arXiv HF 原文 摘要 （无摘要）