arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2602.02156cs.CV
收藏

LoopViT: 通过循环变换器扩展视觉ARC

LoopViT: Scaling Visual ARC with Looped Transformers

HKUST(香港科技大学);CASIA(中国科学院自动化研究所);UC Santa Cruz(加州大学圣克ruz分校)

Wen-Jie Shu, Xuerui Qiu, Rui-Jie Zhu, Harold Haodong Chen, Yexin Liu, Harry Yang

AI总结:

LoopViT通过循环变换器实现视觉推理的高效扩展,采用权重绑定递归结构和动态退出机制,提升ARC-AGI基准测试性能。

摘要:

最近的视觉推理进展利用视觉变换器来解决ARC-AGI基准测试。然而,我们认为馈送前馈架构,其中计算深度严格受限于参数大小,无法捕捉人类归纳的迭代、算法性质。在本工作中,我们提出了一种递归架构,称为Loop-ViT,通过权重绑定的递归解耦推理深度与模型容量。Loop-ViT迭代一个权重绑定的混合块,结合局部卷积和全局注意力,形成潜在的思维链。关键的是,我们引入了一个无参数的动态退出机制,基于预测熵:当模型内部状态“结晶”成低不确定性吸引子时,模型停止推理。在ARC-AGI-1基准测试上的实验证实了这一观点:我们的1800万参数模型在准确性上达到65.8%,优于7300万参数的大型集合。这些发现表明,适应性迭代计算比单纯增加网络宽度提供了更高效的视觉推理扩展轴。代码可在https://github.com/WenjieShu/LoopViT上获取。

英文摘要:

Recent advances in visual reasoning have leveraged vision transformers to tackle the ARC-AGI benchmark. However, we argue that the feed-forward architecture, where computational depth is strictly bound to parameter size, falls short of capturing the iterative, algorithmic nature of human induction. In this work, we propose a recursive architecture called Loop-ViT, which decouples reasoning depth from model capacity through weight-tied recurrence. Loop-ViT iterates a weight-tied Hybrid Block, combining local convolutions and global attention, to form a latent chain of thought. Crucially, we introduce a parameter-free Dynamic Exit mechanism based on predictive entropy: the model halts inference when its internal state ``crystallizes" into a low-uncertainty attractor. Empirical results on the ARC-AGI-1 benchmark validate this perspective: our 18M model achieves 65.8% accuracy, outperforming massive 73M-parameter ensembles. These findings demonstrate that adaptive iterative computation offers a far more efficient scaling axis for visual reasoning than simply increasing network width. The code is available at https://github.com/WenjieShu/LoopViT.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑