提出新型推理模型，显著提升复杂任务表现。
Universal Reasoning Model
- 基于循环归纳偏置与Transformer非线性结构改进
- 在ARC-AGI1上达53.8%准确率，刷新记录
- 适合需要强逻辑推理的AI研究者使用
通用变换器（UTs）被广泛应用于复杂推理任务，如ARC-AGI和数独，但其性能提升的具体来源仍不明确。本文系统分析了多种UT变体，发现其在ARC-AGI上的表现提升主要源于循环归纳偏置和Transformer强大的非线性组件，而非复杂的架构设计。受此启发，我们提出通用推理模型（URM），通过引入短卷积与截断反向传播增强原模型。该方法显著提升推理性能，在ARC-AGI1上达到53.8% pass@1，在ARC-AGI2上达到16.0% pass@1，均达当前最优水平。代码已开源：https://github.com/UbiquantAI/URM。
原文摘要 · Abstract (English)
Universal transformers (UTs) have been widely used for complex reasoning tasks such as ARC-AGI and Sudoku, yet the specific sources of their performance gains remain underexplored. In this work, we systematically analyze UTs variants and show that improvements on ARC-AGI primarily arise from the recurrent inductive bias and strong nonlinear components of Transformer, rather than from elaborate architectural designs. Motivated by this finding, we propose the Universal Reasoning Model (URM), which enhances the UT with short convolution and truncated backpropagation. Our approach substantially improves reasoning performance, achieving state-of-the-art 53.8% pass@1 on ARC-AGI 1 and 16.0% pass@1 on ARC-AGI 2. Our code is avaliable at https://github.com/UbiquantAI/URM.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。