English
Related papers

Related papers: Toy Models of Superposition

200 papers

Universality is a key hypothesis in mechanistic interpretability -- that different models learn similar features and circuits when trained on similar tasks. In this work, we study the universality hypothesis by examining how small neural…

Machine Learning · Computer Science 2023-05-26 Bilal Chughtai , Lawrence Chan , Neel Nanda

The human ability to flexibly reason using analogies with domain-general content depends on mechanisms for identifying relations between concepts, and for mapping concepts and their relations across analogs. Building on a recent model of…

Artificial Intelligence · Computer Science 2021-10-06 Hongjing Lu , Nicholas Ichien , Keith J. Holyoak

Mechanistic Interpretability (MI) promises a path toward fully understanding how neural networks make their predictions. Prior work demonstrates that even when trained to perform simple arithmetic, models can implement a variety of…

Machine Learning · Computer Science 2024-05-28 Ouail Kitouni , Niklas Nolte , Víctor Samuel Pérez-Díaz , Sokratis Trifinopoulos , Mike Williams

People regularly make inferences about objects in the world that they cannot see by flexibly integrating information from multiple sources: auditory and visual cues, language, and our prior beliefs and knowledge about the scene. How are we…

Artificial Intelligence · Computer Science 2025-06-18 Lance Ying , Daniel Xu , Alicia Zhang , Katherine M. Collins , Max H. Siegel , Joshua B. Tenenbaum

Revealing the implicit semantic relation between the constituents of a noun-compound is important for many NLP applications. It has been addressed in the literature either as a classification task to a set of pre-defined relations or by…

Computation and Language · Computer Science 2018-05-08 Vered Shwartz , Ido Dagan

Foundation models are powerful yet often opaque in their decision-making. A topic of continued interest in both neuroscience and artificial intelligence is whether some neurons behave like grandmother cells, i.e., neurons that are…

Machine Learning · Computer Science 2026-01-08 Ricardo Knauer , Erik Rodner

Interpretability benefits the theoretical understanding of representations. Existing word embeddings are generally dense representations. Hence, the meaning of latent dimensions is difficult to interpret. This makes word embeddings like a…

Computation and Language · Computer Science 2023-06-27 Minxue Xia , Hao Zhu

Toy models are highly idealized and deliberately simplified models that retain only the essential features of a system in order to explore specific theoretical questions. Long used in physics and other sciences, they have recently begun to…

Neurons and Cognition · Quantitative Biology 2025-08-04 Larissa Albantakis

Neural network models often generalize poorly to mismatched domains or distributions. In NLP, this issue arises in particular when models are expected to generalize compositionally, that is, to novel combinations of familiar words and…

Computation and Language · Computer Science 2021-11-10 Wang Zhu , Peter Shaw , Tal Linzen , Fei Sha

Associative learning--forming links between co-occurring items--is fundamental to human cognition, reshaping internal representations in complex ways. Testing hypotheses on how representational changes occur in biological systems is…

Machine Learning · Computer Science 2025-10-27 Camila Kolling , Vy Ai Vo , Mariya Toneva

Multilingual language models (MLLMs) have demonstrated remarkable abilities to transfer knowledge across languages, despite being trained without explicit cross-lingual supervision. We analyze the parameter spaces of three MLLMs to study…

Computation and Language · Computer Science 2025-06-03 Frederick Riemenschneider , Anette Frank

Humans can systematically generalize to novel compositions of existing concepts. Recent studies argue that neural networks appear inherently ineffective in such cognitive capacity, leading to a pessimistic view and a lack of attention to…

Computation and Language · Computer Science 2022-10-19 Ning Shi , Boxin Wang , Wei Wang , Xiangyu Liu , Zhouhan Lin

Lexical semantic typology has identified important cross-linguistic generalizations about the variation and commonalities in polysemy patterns---how languages package up meanings into words. Recent computational research has enabled…

Computation and Language · Computer Science 2020-06-04 Ella Rabinovich , Yang Xu , Suzanne Stevenson

We train multi-task autoencoders on linguistic tasks and analyze the learned hidden sentence representations. The representations change significantly when translation and part-of-speech decoders are added. The more decoders a model…

Computation and Language · Computer Science 2018-01-19 Gino Brunner , Yuyi Wang , Roger Wattenhofer , Michael Weigelt

Neural language models learn word representations, or embeddings, that capture rich linguistic and conceptual information. Here we investigate the embeddings learned by neural machine translation models, a recently-developed class of neural…

Computation and Language · Computer Science 2015-04-06 Felix Hill , Kyunghyun Cho , Sebastien Jean , Coline Devin , Yoshua Bengio

Spoken communication occurs in a "noisy channel" characterized by high levels of environmental noise, variability within and between speakers, and lexical and syntactic ambiguity. Given these properties of the received linguistic input,…

Computation and Language · Computer Science 2021-01-26 Stephan C. Meylan , Sathvik Nair , Thomas L. Griffiths

Recent research (arXiv:2310.11453, arXiv:2402.17764) has proposed binary and ternary transformer networks as a way to significantly reduce memory and improve inference speed in Large Language Models (LLMs) while maintaining accuracy. In…

Machine Learning · Computer Science 2024-05-29 Jason Li

Modern machine learning systems based on neural networks have shown great success in learning complex data patterns while being able to make good predictions on unseen data points. However, the limited interpretability of these systems…

Machine Learning · Computer Science 2020-07-22 Sarath Shekkizhar , Antonio Ortega

Neural networks are powerful tools for cognitive modeling due to their flexibility and emergent properties. However, interpreting their learned representations remains challenging due to their sub-symbolic semantics. In this work, we…

Machine Learning · Computer Science 2026-04-07 Andrew Nam , Declan Campbell , Thomas Griffiths , Jonathan Cohen , Sarah-Jane Leslie

When an action potential is transmitted to a postsynaptic neuron, a small change in the postsynaptic neuron's membrane potential occurs. These small changes, known as a postsynaptic potentials (PSPs), are highly variable, and current models…

Neurons and Cognition · Quantitative Biology 2015-07-14 Laurence Aitchison , Peter E. Latham