paper-with-me

홈 › Papers

STAR: Mitigating Cascading Errors in Spatial Reasoning via Turn-point Alignment and Segment-level DPO

2026-04-01 · Pukun Zhao, Longxiang Wang, Chen Chen, Peicheng Wang, Fanqing Zhou, Runze Li, Haojian Huang arxiv

Structured spatial navigation is a core benchmark for Large Language Models (LLMs) spatial reasoning. Existing paradigms like Visualization-of-Thought (VoT) are prone to cascading errors in complex topologies. To solve this, we propose STAR, a two-stage framework grounded on topological anchors, and introduce the RedMaze-23K dataset with human-inspired turnpoint annotations. The first stage uses supervised fine-tuning to help models internalize spatial semantics and prune redundant paths. The second adopts Spatial-aware Segment-level Direct Preference Optimization (SDPO) to refine self-correction in long-horizon navigation. Experiments show STAR achieves state-of-the-art performance among open-source models: its 32B variant outperforms DeepSeek-V3 (29.27% vs. 25.00%) and reaches 82.4% of GPT-4's performance.

📄 PDF Abstract BibTeX arXiv:2604.00558

Code (0)

등록된 구현이 없습니다.

Tasks

Spatial Reasoning

Similar Papers 제목 키워드 기반

LARAD: Layout-Aware Road Anomaly Detection via Spatial-Logic Reasoning

2026-07-14 · Shiyi Mu, Xujie Chen, Shugong Xu arxiv

Accurate open-world obstacle detection is critical for autonomous driving. Current anomaly segmentation methods suffer from a fundamental blind spot: they over-rely on texture novelty to identify out-of-distribution (OoD…

Autonomous DrivingAnomaly Detection

Vesta: A Generalist Embodied Reasoning Model

2026-06-18 · Johan Bjorck, Zhiqi Li, Yunze Man, Jing Wang 외 arxiv

Robots operating in open-world environments must seamlessly integrate localization, spatial reasoning, navigation, and long-horizon planning. While specialist models excel at individual tasks, deploying a multi-model sta…

Spatial Reasoning

C2F-Thinker: Coarse-to-Fine Reasoning with Hint-Guided Reinforcement Learning for Multimodal Sentiment Analysis

2026-03-10 · Miaosen Luo, Zhenhao Yang, Jieshen Long, Jinghu Sun 외 arxiv

Multimodal sentiment analysis aims to integrate textual, acoustic, and visual information for deep emotional understanding. Despite the progress of multimodal large language models (MLLMs) via supervised fine-tuning, the…

Multimodal Sentiment AnalysisReinforcement LearningDomain Generalization

Process In-Context Learning: Enhancing Mathematical Reasoning via Dynamic Demonstration Insertion

2026-01-17 · Ang Gao, Changshuo Zhang, Xiao Zhang, Deyang Li 외 arxiv

In-context learning (ICL) has proven highly effective across diverse large language model (LLM) tasks. However, its potential for enhancing tasks that demand step-by-step logical deduction, such as mathematical reasoning…

Mathematical Reasoning

LiLo-VLA: Compositional Long-Horizon Manipulation via Linked Object-Centric Policies

2026-02-25 · Yue Yang, Shuo Cheng, Yu Fang, Homanga Bharadhwaj 외 arxiv

General-purpose robots must master long-horizon manipulation, defined as tasks involving multiple kinematic structure changes (e.g., attaching or detaching objects) in unstructured environments. While Vision-Language-Act…

Zero-shot Generalization