Research map / Large language models / Agents, code and math
DecoVer: A Decompose-and-Verify Neuro-Symbolic Framework for Embodied Task Planning with BC+
YiXiang Jiang, Binqian Xu, Xiangbo Shu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add DecoVer neuro-symbolic planning to the LLM | Claude-Opus-4.5 · VirtualHome (298 tasks) | Task SR (%): 57.4 -> 91.6 | Improved |
| add DecoVer to the LLM | Gemini-3-Pro-Preview · VirtualHome (298 tasks) | Task SR (%): 76.7 -> 89.6 | Improved |
| add DecoVer to the LLM | GPT-5.1 · VirtualHome logically extended (23 cases) | Task SR (%): 13.0 -> 73.9 | Improved |
| add DecoVer to the LLM | GPT-5.1 · VirtualHome temporally extended (20 cases) | Task SR (%): 0.0 -> 30.0 | Improved |
| add DecoVer to the LLM | Claude-Opus-4.5-thinking · VirtualHome temporally extended (20 cases) | Task SR (%): 45.0 -> 90.0 | Improved |
| add DecoVer to the LLM | Gemini-3-Pro-Preview-thinking · VirtualHome goal satisfaction (logically extended) | Total goal satisfaction (%): 84.8 -> 81.0 | Hurt |
| add DecoVer to the LLM | GPT-4o · VirtualHome goal satisfaction (temporally extended) | Total goal satisfaction (%): 42.9 -> 41.1 | Hurt |
| add action verification (ablation) | DecoVer on Claude-Opus-4.5 · VirtualHome ablation (11 hard instances) | Task SR (%): 54.5 -> 81.8 | Improved |
| add action verification (ablation) | DecoVer on Gemini-3-Pro-Preview · VirtualHome ablation (11 hard instances) | Task SR (%): 90.9 -> 90.9 | No clear effect |
| add ground rules verification (ablation) | DecoVer on Claude-Opus-4.5 · VirtualHome ablation (11 hard instances) | Task SR (%): 72.7 -> 81.8 | Improved |
| add flag grouping in S4 (ablation) | DecoVer on Claude-Opus-4.5 · VirtualHome ablation (11 hard instances) | Task SR (%): 36.4 -> 81.8 | Improved |
| restrict S3 to selected objects and actions instead of all (ablation) | DecoVer on Claude-Opus-4.5 · VirtualHome ablation (11 hard instances) | Task SR (%): 36.4 -> 81.8 | Improved |
Closest papers by meaning
- Towards Reliable Code-as-Policies: A Neuro-Symbolic Framework for Embodied Task Planning (NeurIPS 2025 · similarity 0.93)
- NeSyPr: Neurosymbolic Proceduralization For Efficient Embodied Reasoning (NeurIPS 2025 · similarity 0.90)
- Embodied Task Planning via Graph-Informed Action Generation with Large Language Models (ICML 2026 · similarity 0.90)
- Embodied Agent Interface: Benchmarking LLMs for Embodied Decision Making (NeurIPS 2024 · similarity 0.89)
- NeSyC: A Neuro-symbolic Continual Learner For Complex Embodied Tasks In Open Domains (ICLR 2025 · similarity 0.89)
- R2C: Mapping Room to Chessboard to Unlock LLM As Low-Level Action Planner (CVPR 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
