English
Related papers

Related papers: Architecture Determines Observability of Transform…

200 papers

Transformers resist surgical control. Ablating an attention head identified as critical for capitalization produces minimal behavioral change because distributed redundancy compensates for damage. This Hydra effect renders interpretability…

Machine Learning · Computer Science 2026-03-20 J. Clayton Kerce

As large language models are deployed as autonomous agents with tool execution privileges, a critical assumption underpins their security architecture: that model errors are detectable at runtime. We present empirical evidence that this…

Artificial Intelligence · Computer Science 2026-03-24 Gregory M. Ruddell

We present the first comprehensive analysis of massive activation development throughout transformer training, using the Pythia model family as our testbed, and release our full dataset publicly to support further research. Through…

Artificial Intelligence · Computer Science 2026-02-25 Jorge Gallego-Feliciano , S. Aaron McClendon , Juan Morinelli , Stavros Zervoudakis , Antonios Saravanos

Transformer architectures have revolutionized machine learning across a wide range of domains, from natural language processing to scientific computing. However, their growing deployment in high-stakes applications, such as computer vision,…

Machine Learning · Computer Science 2026-02-17 Trishit Mondal , Ameya D. Jagtap

The attention mechanism in its standard implementation contains extraneous rotational degrees of freedom that are carried through computation but do not affect model activations or outputs. We introduce a simple symmetry-breaking protocol…

Machine Learning · Computer Science 2026-02-13 Eva Silverstein , Daniel Kunin , Vasudev Shyam

Neural networks gain capabilities during training, but the internal changes that precede capability acquisition are not well understood. In particular, the relationship between geometric change and behavioral change, and the effect of task…

Machine Learning · Computer Science 2026-04-03 Jayadev Billa

When do transformers commit to a decision, and what prevents them from correcting it? We introduce \textbf{prolepsis}: a transformer commits early, task-specific attention heads sustain the commitment, and no layer corrects it. Replicating…

Machine Learning · Computer Science 2026-04-17 Éric Jacopin

We document empirical capability ceilings in decoder-only autoregressive language models across knowledge-intensive tasks. Systematic evaluation of OPT and Pythia model families (70M-30B parameters, spanning 240 times scaling) reveals that…

Artificial Intelligence · Computer Science 2025-10-28 Javier Marín

Image Quality Assessment (IQA) models are increasingly relied upon to evaluate image quality in real-world systems -- from compression and enhancement to generation and streaming. Yet their adoption brings a fundamental risk: these models…

Computer Vision and Pattern Recognition · Computer Science 2025-06-06 Igor Meleshin , Anna Chistyakova , Anastasia Antsiferova , Dmitriy Vatolin

While transformers achieve strong performance, their internal decision-making processes remain opaque. We investigate whether architectural constraints can enforce interpretability by design through architectural stream independence:…

Machine Learning · Computer Science 2026-03-10 Clayton Kerce , Alexis Fox

Maintaining stability in feedback systems, from aircraft and autonomous robots to biological and physiological systems, relies on monitoring their behavior and continuously adjusting their inputs. Incremental damage can make such control…

Current AI safety relies on behavioral monitoring and post-training alignment, yet empirical measurement shows these approaches produce no detectable pre-commitment signal in a majority of instruction-tuned models tested. We present an…

Artificial Intelligence · Computer Science 2026-04-07 Gregory M. Ruddell

Symbolic regression aims to recover closed-form expressions from numerical data, but in differentiable symbolic regression the recovered expression depends not only on the grammar but also on the fixed architecture through which variables…

Neural and Evolutionary Computing · Computer Science 2026-05-29 Chakshu Gupta , Theodore J. LaGrow

Transformers trained on huge text corpora exhibit a remarkable set of capabilities, e.g., performing basic arithmetic. Given the inherent compositional nature of language, one can expect the model to learn to compose these capabilities,…

Machine Learning · Computer Science 2024-02-07 Rahul Ramesh , Ekdeep Singh Lubana , Mikail Khona , Robert P. Dick , Hidenori Tanaka

Prompt injection poses a critical threat to the safe deployment of large language models, yet existing detection approaches are typically evaluated under limited settings that do not reflect real-world operating constraints. In this work,…

Computation and Language · Computer Science 2026-05-27 Akindoyin Akinrele , Shreyank N Gowda

Transformers often appear to perform Bayesian reasoning in context, but verifying this rigorously has been impossible: natural data lack analytic posteriors, and large models conflate reasoning with memorization. We address this by…

Machine Learning · Computer Science 2026-05-19 Naman Agarwal , Siddhartha R. Dalal , Vishal Misra

Interpretability methods aim to understand the algorithm implemented by a trained model (e.g., a Transofmer) by examining various aspects of the model, such as the weight matrices or the attention patterns. In this work, through a…

Machine Learning · Computer Science 2023-12-05 Kaiyue Wen , Yuchen Li , Bingbin Liu , Andrej Risteski

Recent work has shown that Transformers' compositional generalization is governed by \emph{complexity control}, initialization scale and weight decay, which steers training toward low-complexity reasoning solutions rather than…

Machine Learning · Computer Science 2026-05-07 Sarwan Ali

We find that models report highest confidence precisely when they are fabricating. Across four model families (OLMo-3, Llama-3.1, Qwen3, Mistral), self-reported confidence inversely correlates with accuracy, with AUC ranging from 0.28 to…

Distributed, Parallel, and Cluster Computing · Computer Science 2026-05-08 Tony Mason , Vaastav Anand

In this paper we prove the existence of a fundamental trade-off between accuracy and robustness in perception-based control, where control decisions rely solely on data-driven, and often incompletely trained, perception maps. In particular,…

Optimization and Control · Mathematics 2020-03-18 Abed AlRahman Al Makdah , Vaibhav Katewa , Fabio Pasqualetti
‹ Prev 1 2 3 10 Next ›