Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2511.14670 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912716729352192 |
|---|---|
| author | Ding, Ruomeng Cheng, Wei Shao, Minglai Zhao, Chen |
| author_facet | Ding, Ruomeng Cheng, Wei Shao, Minglai Zhao, Chen |
| contents | Large language models (LLMs) are increasingly applied to sequential decision-making through in-context learning (ICL), yet their effectiveness is highly sensitive to prompt quality. Effective prompts should meet three principles: focus on decision-critical information, provide step-level granularity, and minimize reliance on expert annotations through label efficiency. However, existing ICL methods often fail to satisfy all three criteria simultaneously. Motivated by these challenges, we introduce SkillGen, a skill-based ICL framework for structured sequential reasoning. It constructs an action-centric, domain-level graph from sampled trajectories, identifies high-utility actions via temporal-difference credit assignment, and retrieves step-wise skills to generate fine-grained, context-aware prompts. We further present a theoretical analysis showing that focusing on high-utility segments supports task identifiability and informs more effective ICL prompt design. Experiments on ALFWorld, BabyAI, and ScienceWorld, using both open-source and proprietary LLMs, show that SkillGen achieves consistent gains, improving progress rate by 5.9%-16.5% on average across models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_14670 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SkillGen: Learning Domain Skills for In-Context Sequential Decision Making Ding, Ruomeng Cheng, Wei Shao, Minglai Zhao, Chen Artificial Intelligence Large language models (LLMs) are increasingly applied to sequential decision-making through in-context learning (ICL), yet their effectiveness is highly sensitive to prompt quality. Effective prompts should meet three principles: focus on decision-critical information, provide step-level granularity, and minimize reliance on expert annotations through label efficiency. However, existing ICL methods often fail to satisfy all three criteria simultaneously. Motivated by these challenges, we introduce SkillGen, a skill-based ICL framework for structured sequential reasoning. It constructs an action-centric, domain-level graph from sampled trajectories, identifies high-utility actions via temporal-difference credit assignment, and retrieves step-wise skills to generate fine-grained, context-aware prompts. We further present a theoretical analysis showing that focusing on high-utility segments supports task identifiability and informs more effective ICL prompt design. Experiments on ALFWorld, BabyAI, and ScienceWorld, using both open-source and proprietary LLMs, show that SkillGen achieves consistent gains, improving progress rate by 5.9%-16.5% on average across models. |
| title | SkillGen: Learning Domain Skills for In-Context Sequential Decision Making |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2511.14670 |