CORTEXA
← Browse
arxivcs.CVcs.AI2026-07-31

TraceViT: Grounded Trace Supervision for Visual Abstract Reasoning

Binnan Liu, Yechi Ma, Tian Xie, Wei Hua

The Abstraction and Reasoning Corpus (ARC) tests whether a model can infer an unseen transformation from a few input-output examples and apply it to a new grid. Looped visual reasoners refine predictions over multiple iterations, but conventional training constrains only the final output, leaving intermediate refinements unconstrained. We propose that these refinements should instead follow the transformation step by step. We introduce TraceViT, a looped visual reasoner trained with semantically monotonic transformation chains. We obtain these chains by rewriting and verifying programmatic task implementations, decomposing each solution into intermediate grid states. Each iteration is grounded by a task reference derived from the few-shot demonstrations and an object workspace representing the current grid state. Because these chains may differ in length from the loop, soft trace alignment enforces only their ordering, letting the model allocate iterations freely. TraceViT achieves 67.8% pass@2 on ARC-AGI-1 and 24.3% on ARC-AGI-2. Controlled ablations on ARC-AGI-1 show that trace supervision becomes beneficial only when paired with grounding. Code and data will be available at https://github.com/LiuBinnan/TraceViT.

View free PDFSource page

Related papers

arxivcs.CVcs.AI2026-07-07

Segmentation before Answering: Pixel Grounding for MLLM Visual Reasoning

Yake Wei, Yuan Wang, Fengyun Rao, Jing Lyu, Di Hu

Recent advancements in Multimodal Large Language Models (MLLMs) have evolved from static perception to interleaved visual-language reasoning, often referred to as ``thinking with images''. A basic operation in this reasoning process is to zoom in on regions of interest (often rep…

View free PDFSource page
arxivcs.AIcs.CV2026-07-07

Bridging Physical Reasoning and Task Generalization via Visual Action Outcome Reasoning Alignment

Han-Jun Ko, Jr-Jen Chen, Haobo Yuan, Hsin-Ying Lee, Tiancheng Shen, Ming-Hsuan Yang, et al.

Vision-language models (VLMs) struggle to generalize in interactive physical reasoning, particularly under unseen tasks and environments. Two key failure modes are prominent: hallucinated chain-of-thought (CoT) reasoning that contradicts physical reality, and misalignment between…

View free PDFSource page
arxivcs.CVcs.AI2026-07-10

OmniMapBench: Benchmarking Visual-Centric Reasoning on Diverse Map Documents

Yang Chen, Yunwen Li, Yufan Shen, Minghao Liu, Tianyu Zheng, Bin Fu, et al.

Recent advancements in LVLMs necessitate robust benchmarks for complex, visually grounded reasoning. A critical limitation is identified in many document understanding benchmarks: visual content is often reducible to text, enabling high performance without genuine visual groundin…

View free PDFSource page
arxivcs.CVcs.AI2026-07-15

GHR-VLM: Making Zero-Shot Transit Video Analytics Realizable with Grounded Hybrid Reasoning

Kaicong Huang, Weiheng Oh, Jack M. Reilly, Thomas Guggisberg, Ruimin Ke

Transit video understanding can provide valuable fine-grained data that conventional passenger counters and fare systems cannot capture. However, supervised video models require task-specific annotations, while applying vision-language models (VLMs) directly to long onboard video…

View free PDFSource page
arxivcs.CVcs.AIcs.CL2026-06-30

PruneGround: Plug-and-play Spatial Pruning for 3D Visual Grounding

Duc Cao Dinh, Khai Le-Duc, Florent Draye, Chris Ngo, Terry Jingchen Zhang, Bernhard Schölkopf, et al.

3D Visual Grounding (3DVG) aims to localize target objects in 3D scenes given natural language descriptions. Existing approaches typically perform reasoning over the entire scene, leading to ambiguous predictions and high computational cost, especially in cluttered environments.…

View free PDFSource page
arxivcs.AIcs.CV2026-07-18

Constraint-Anchored Reasoning Traces

Zehua Cheng, Wei Dai, Jiahao Sun

Autoregressive multimodal large language models (MLLMs) suffer from error snowballing: a single incorrect inference early in a chainof-thought (CoT) trace corrupts all downstream reasoning. We find that in state-of-the-art open-source MLLMs, once the first error occurs, the reaso…

View free PDFSource page