牛哥精选 · 半年

📋 全部 🤖 AI·大模型 ⚡ 效率工具 📝 深度技术 🚀 产品观察 💰 商业科技 🔓 开源项目 🎨 设计创意 📖 阅读推荐 🏷 资源合集 🌱 成长效率

📝 深度技术 arXiv AI 2026-05-28

Skill-Conditioned Gated Self-Distillation for LLM Reasoning

用特权信息生成密集token级监督，提升LLM推理能力的新蒸馏方法。

arXiv:2605.28791v1 Announce Type: cross Abstract: On-policy self-distillation (SD) improves LLM reasoning by using teacher-side privileged information…

自我蒸馏 llm推理特权信息 token级监督门控机制

📝 深度技术 arXiv AI 2026-05-20

Toward Privileged Foundation Models:LUPI for Accelerated and Improved Learning

新框架PIQL利用特权信息同时加速表格基础模型训练并提升泛化能力。

arXiv:2605.07799v2 Announce Type: replace-cross Abstract: Training foundation models is computationally intensive and often slow to converge. We intro…

特权信息基础模型表格数据加速学习泛化

📅 日期

2026-05-20 2026-05-19

🐂 牛哥精选

Skill-Conditioned Gated Self-Distillation for LLM Reasoning

Toward Privileged Foundation Models:LUPI for Accelerated and Improved Learning

📅 日期