arXiv cs.AI· Jisen Lei, Huizhi Liang·· 12 小时前AI 评分24
Generate-then-Rank：为具身智能体学习符号化目标解释
Generate-then-Rank: Learning Symbolic Goal Interpretation for Embodied Agents
AI 导读
研究提出两阶段“生成-排序”方法 Generate-then-Rank，用于具身智能体接口（EAI）中的目标解释：先通过监督微调让 LLM 适配 EAI 输出规范，再用 Candidate-Output Ranking from Errors（CORE）从模型自身生成的候选中学习偏好。
正文 · 原文
Abstract:Goal interpretation is to translate a natural language instruction into a formal symbolic goal that a planner can execute. It is critical for embodied agents because any error here propagates to every downstream planning and action step. Doing it well requires both generating suitable candidates and reliably selecting among them. We present a two-stage generate-then-rank approach to Goal Interpretation in the Embodied Agent Interface (EAI). First, supervised fine-tuning adapts a Large Language Model (LLM) to EAI's output convention. Second, Candidate-Output Ranking from Errors (CORE) learns preferences over complete outputs from the model's own generated alternatives. We conducted experiments on the household scenes of the Embodied Agent Interface Benchmark. Multiple open source LLMs including DeepSeek-V3 chat, DeepSeek-Reasoner, and Qwen2.5-7B and Qwen2.5-72B were evaluated. The proposed method outperformed the baseline models including LLMs with simple prompting, chain-of-thought, and knowledge graph based approaches.
| Comments: | 14 pages, 2 figures. Accepted at the 10th Euro-China Conference on Intelligent Data Analysis and Applications (ECC 2026) |
| Subjects: | Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Robotics (cs.RO) |
| Cite as: | arXiv:2611.07825 [cs.AI] |
| (or arXiv:2611.07825v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.07825 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Jisen Lei [view email]
[v1]
Thu, 5 Nov 2026 11:05:53 UTC (1,010 KB)
来源：arXiv cs.AI · arxiv.org