Episode
LatentSkill: From In-Context Textual Skills to In-Weight Latent Skills for LLM Agents
- Podcast
- Daily Paper Cast
- Published
- Jun 10, 2026
- Duration seconds
- 1322
- Processing state
not_requested- Canonical source
- https://share.transistor.fm/s/7f04106c
Actions
POST https://stenobird.com/v1/public/podcasts/daily-paper-cast-7079649/episodes/latentskill-from-in-context-textual-skills-to-in-weight-latent-skills-for-llm-agents/transcription-requests
Idempotently request low-priority transcript generation for this episode.GET https://stenobird.com/podcast/daily-paper-cast-7079649/latentskill-from-in-context-textual-skills-to-in-weight-latent-skills-for-llm-agents.md
Read the agent-friendly Markdown representation of this episode resource.
Summary
🤗 Upvotes: 52 | cs.CL, cs.AI Authors: Aofan Yu, Chenyu Zhou, Tianyi Xu, Zihan Guo, Rong Shan, Zhihui Fu, Jun Wang, Weiwen Liu, Yong Yu, Weinan Zhang, Jianghao Lin Title: LatentSkill: From In-Context Textual Skills to In-Weight Latent Skills for LLM Agents Arxiv: http://arxiv.org/abs/2606.06087v1 Abstract: Agent systems increasingly use textual skills to encode reusable task procedures, but injecting these skills into the prompt at every step incurs substantial context overhead and exposes skill content as plaintext. We present LatentSkill, a framework that converts textual skills into plug-and-play LoRA adapters through a pretrained hypernetwork. LatentSkill stores skill knowledge in weight space rather than context space, removing per-step skill tokens while preserving modular loading, scaling, and composition. On ALFWorld and Search-QA, LatentSkill outperforms the corresponding in-context skill baseline while using substantially fewer prefill tokens: it improves ALFWorld success by 21.4 and 13.4 points on the seen and unseen splits with 64.1% fewer prefill tokens, and improves Search-QA exact match by 3.0 points with 72.2% lower skill-token overhead. Further analysis shows that generated skill LoRAs form a structured semantic geometry, can be precisely controlled via the LoRA scaling coefficient, and can be composed through parameter-space arithmetic when skill components are aligned. These findings suggest that weight-space skills provide an efficient, modular, and less exposed substrate for extending LLM agents.