今日值得读

聚焦大语言模型与智能体研究,不按热度排名。Agent 每天分别选出真正值得读的新工作,先生成可检查的研究计划,再持续补齐复现结果。

1大语言模型arXiv 2609.015320/76

Knowledge Distillation During Mid-Training Favors Reasoning over Factual Recall

Jacqueline He, Howard Yen, Shuyue Stella Li, Margaret Li, Hanqing Zeng, Yinglong Xia, Benyu Zhang, Zhuokai Zhao, Qiang Zhang, Pang Wei Koh, Luke Zettlemoyer, Wen-tau Yih

为什么值得读

This paper uncovers a critical trade-off in language model training: applying standard knowledge distillation during mid-training accelerates reasoning capabilities but actively impedes the acquisition of factual recall. The authors diagnose the mechanism behind this phenomenon—namely, that teachers exhibit high predictive uncertainty on factual tokens, thereby diluting the ground-truth training signal for facts that the student has not yet acquired. To resolve this, they introduce Switch Distillation, a simple and computationally lightweight objective that routes confident tokens to reverse-KL distillation and uncertain tokens to standard cross-entropy. For practitioners developing foundation models, this provides a practical, drop-in training objective that substantially improves reasoning performance without sacrificing factual knowledge, with benefits that persist through subsequent alignment and post-training pipelines.

核心研究结论

  • Mid-training ablation results for Always CE relative difference across Reasoning, Factual Recall, and Knowledge task groups.
  • Downstream task evaluation after post-training stage RLVR1 for 13b_rkd.
  • The stage-dependent distillation tradeoff generalizes to the SmolLM family (SmolLM2 360M student, 1.7B Instruct teacher, SmolLM3 Stage-3 data), where mid-training KD exhibits the reasoning-recall deficit and Switch Distillation mitigates it.
计算与语言人工智能机器学习 (cs.LG)
2智能体arXiv 2609.014870/14

Defense-as-Skill: Evolving Runtime Guard Skill for Skill-Augmented Agents

Xiaofang Yang, Ziqi Miao, Dianbo Sui, Jing Shao, Lijun Li

为什么值得读

This paper establishes that runtime safety defenses for autonomous coding agents can be natively packaged and iteratively evolved as agent skills themselves, rather than requiring intrusive host-runtime modifications, platform-level kernel sandboxing, or external moderation endpoints. By introducing the SCOPE-R taxonomy and closed-loop guard-skill evolution via MCTS, the work provides developers with modular, transparent, and portable defense primitives that achieve a favorable balance between risk mitigation and benign operational utility across diverse frontier foundation models.

核心研究结论

  • SkillSonar generalizes zero-shot to external safety benchmarks, achieving the lowest ASR of 52.90% on SkillSafetyBench and highest safety score of 54.00% on WildClawBench compared to prompt, permission, and allowlist baselines.
  • On benign tasks containing no malicious skill behavior, SkillSonar preserves high utility on Claude Code (0.817 on GLM-5, 0.836 on Claude Haiku 4.5, and 0.724 on GPT-5.4), demonstrating minimal false-positive disruption of safe execution.
  • Over five matched evaluation rounds, SkillSonar reduces ASR across all individual risk families, achieving reductions of 50.0 pp on Resource & Reliability, 42.9 pp on Execution Safety, 35.3 pp on Specification Integrity, 18.3 pp on Operational Side Effects, 61.5 pp on Privacy & Data Flow (OOD), and 44.8 pp on Capability Control (OOD).
密码学与安全人工智能软件工程