English
Related papers

Related papers: Critical Windows of Complexity Control: When Trans…

200 papers

Transformers have demonstrated impressive capabilities across various tasks, yet their performance on compositional problems remains a subject of debate. In this study, we investigate the internal mechanisms underlying Transformers'…

Computation and Language · Computer Science 2025-01-16 Zhongwang Zhang , Pengxiao Lin , Zhiwei Wang , Yaoyu Zhang , Zhi-Qin John Xu

Transformers trained on modular arithmetic exhibit sharp transitions between memorization, generalization, and collapse. We show that weight decay acts as a scalar empirical control parameter for these regimes, and introduce two cheap…

Machine Learning · Computer Science 2026-05-21 Lucky Verma

Transformers have shown impressive capabilities across various tasks, but their performance on compositional problems remains a topic of debate. In this work, we investigate the mechanisms of how transformers behave on unseen compositional…

Machine Learning · Computer Science 2025-01-14 Zhongwang Zhang , Pengxiao Lin , Zhiwei Wang , Yaoyu Zhang , Zhi-Qin John Xu

Mechanistic interpretability has transformed the analysis of transformer circuits by decomposing model behavior into competing algorithms, identifying phase transitions during training, and deriving closed-form predictions for when and why…

Machine Learning · Computer Science 2026-03-19 Alma Lago

When writing programs, people have the ability to tackle a new complex task by decomposing it into smaller and more familiar subtasks. While it is difficult to measure whether neural program synthesis methods have similar capabilities, what…

Machine Learning · Computer Science 2023-10-31 Kensen Shi , Joey Hong , Manzil Zaheer , Pengcheng Yin , Charles Sutton

Computers are deterministic dynamical systems (CHAOS 19:033124, 2009). Among other things, that implies that one should be able to use deterministic forecast rules to predict their behavior. That statement is sometimes-but not always-true.…

Chaotic Dynamics · Physics 2013-05-24 Joshua Garland , Ryan James , Elizabeth Bradley

Humans exhibit remarkable compositional reasoning by integrating knowledge from various sources. For example, if someone learns ( B = f(A) ) from one source and ( C = g(B) ) from another, they can deduce ( C=g(B)=g(f(A)) ) even without…

Artificial Intelligence · Computer Science 2025-10-14 Yutong Yin , Zhaoran Wang

We investigate the integration of human-like working memory constraints into the Transformer architecture and implement several cognitively inspired attention variants, including fixed-width windows based and temporal decay based attention…

Computation and Language · Computer Science 2026-04-24 Pranava Madhyastha , Dagmar Adamcova

We study implicit reasoning, i.e. the ability to combine knowledge or rules within a single forward pass. While transformer-based large language models store substantial factual knowledge and rules, they often fail to compose this knowledge…

Computation and Language · Computer Science 2026-04-10 Harsh Kohli , Srinivasan Parthasarathy , Huan Sun , Yuekun Yao

While there are many studies on weight regularization, the study on structure regularization is rare. Many existing systems on structured prediction focus on increasing the level of structural dependencies within the model. However, this…

Machine Learning · Computer Science 2015-02-02 Xu Sun

Compression has been a critical lens to understand the success of Transformers. In the past, we have typically taken the target distribution as a criterion to evaluate a model's compression performance. Nevertheless,it often remains…

Machine Learning · Computer Science 2025-04-29 Ruifeng Ren , Yong Liu

We study the approximation capabilities, convergence speeds and on-convergence behaviors of transformers trained on in-context recall tasks -- which requires to recognize the \emph{positional} association between a pair of tokens from…

Machine Learning · Computer Science 2025-10-22 Quan Nguyen , Thanh Nguyen-Tang

Memorization in language models is widely studied but remains difficult to isolate and control. Understanding when and what models memorize is essential for explaining their predictions, yet existing approaches are post-hoc: they can detect…

Computation and Language · Computer Science 2026-04-08 Xiangbo Zhang , Ali Emami

The ability to reason lies at the core of artificial intelligence (AI), and challenging problems usually call for deeper and longer reasoning to tackle. A crucial question about AI reasoning is whether models can extrapolate learned…

Machine Learning · Computer Science 2025-11-11 Yu Huang , Zixin Wen , Aarti Singh , Yuejie Chi , Yuxin Chen

Can Transformers predict new syllogisms by composing established ones? More generally, what type of targets can be learned by such models from scratch? Recent works show that Transformers can be Turing-complete in terms of expressivity, but…

Machine Learning · Computer Science 2024-11-05 Emmanuel Abbe , Samy Bengio , Aryo Lotfi , Colin Sandon , Omid Saremi

When trained on language data, do transformers learn some arbitrary computation that utilizes the full capacity of the architecture or do they learn a simpler, tree-like computation, hypothesized to underlie compositional meaning systems…

Computation and Language · Computer Science 2022-11-07 Shikhar Murty , Pratyusha Sharma , Jacob Andreas , Christopher D. Manning

The transformer architecture is central to the success of modern Large Language Models (LLMs), in part due to its surprising ability to perform a wide range of tasks - including mathematical reasoning, memorization, and retrieval - using…

Machine Learning · Computer Science 2025-09-05 Yihe Dong , Lorenzo Noci , Mikhail Khodak , Mufan Li

Systematic compositionality is an essential mechanism in human language, allowing the recombination of known parts to create novel expressions. However, existing neural models have been shown to lack this basic ability in learning symbolic…

Computation and Language · Computer Science 2021-10-01 Yichen Jiang , Mohit Bansal

Transformers can under some circumstances generalize to novel problem instances whose constituent parts might have been encountered during training, but whose compositions have not. What mechanisms underlie this ability for compositional…

Machine Learning · Computer Science 2025-02-18 Simon Schug , Seijin Kobayashi , Yassir Akram , João Sacramento , Razvan Pascanu

Standard Transformers excel at semantic modeling but struggle with rigid sequential logic and state tracking. Theoretical work establishes that self-attention is limited to $\AC^0$ (under hard attention) or $\TC^0$ (under soft attention),…

Machine Learning · Computer Science 2026-02-10 Mehryar Mohri
‹ Prev 1 2 3 10 Next ›