LeafSkill reduces large model use by reusing action grounding
LeafSkill: Efficient Leaf-Level Action Grounding for Recursive LLM Agents
Artificial Intelligence
Summary
Long tasks can be broken down by large language models that plan and act step-by-step. The authors found that while planning tends to be unique for each task, turning small actions into exact commands is often repeated across tasks. They created LeafSkill, a system that remembers and reuses these small action steps, making the process faster and more efficient. This approach keeps the main planning model unchanged but saves computing resources and improves success in complex tasks.
What this means in practice
- •For robotic software developers: Improve robot task completion by reusing leaf-level action groundings to reduce the need for repeated language model calls.
- •For autonomous system engineers: Lower computational costs in planning systems by offloading common action grounding to a lightweight selector, preserving high-level planning for the main model.
Authors
Shen Liu, Ze Zhou, Zhenguo Xu, Shaopu Wang, Yike Gao, Chunlei Wang
Abstract
Recursive large language model (LLM) agents solve long-horizon tasks through decomposition, execution, and environmental feedback. Yet after producing an atomic subtask, they often invoke the same large model again to translate a familiar instruction into an executable action. We observe that high-level planning is task-specific, whereas leaf-level action grounding recurs across tasks and can be checked against environment-admitted actions. We introduce LeafSkill, an external reuse layer that retrieves semantically related grounding experience, binds action templates to current objects and admissible actions, and uses a lightweight 1.7B selector to choose a bound action or defer the unchanged subtask to the original agent. Skills remain outside the model parameters, allowing a frozen selector to use post-training additions. Under lexical and syntactic paraphrases, exact matching retrieves none of the reference actions, while semantic retrieval preserves 96.0% and 97.0% candidate coverage; the frozen selector adopts 97.34% of unseen skill types. On an independent confirmation set of 200 Robotouille tasks, LeafSkill completes 137 tasks versus 121 for ReCAP (68.5% vs. 60.5%, exact McNemar p=.0259). On the 106 tasks completed by both, it reduces Qwen3.5-27B tokens by 14.42% and calls by 14.08%. A separate held-out repeat yields a consistent 9.0-point success gain and 18.98% token reduction. Reusable leaf grounding can thus improve task completion while reducing host-model inference without replacing high-level planning.