English
Related papers

Related papers: Toy Models of Superposition

200 papers

Interpretability research often adopts a neuron-centric lens, treating individual neurons as the fundamental units of explanation. However, neuron-level explanations can be undermined by superposition, where single units respond to mixtures…

Computer Vision and Pattern Recognition · Computer Science 2026-03-17 Julien Colin , Lore Goetschalckx , Thomas Fel , Victor Boutin , Thomas Serre , Nuria Oliver

The success of neural networks builds to a large extent on their ability to create internal knowledge representations from real-world high-dimensional data, such as images, sound, or text. Approaches to extract and present these…

Artificial Intelligence · Computer Science 2023-01-03 Lars Holmberg , Paul Davidsson , Per Linde

Despite the success of language models using neural networks, it remains unclear to what extent neural models have the generalization ability to perform inferences. In this paper, we introduce a method for evaluating whether neural models…

Computation and Language · Computer Science 2020-05-05 Hitomi Yanaka , Koji Mineshima , Daisuke Bekki , Kentaro Inui

Neural networks represent more features than they have dimensions via superposition, forcing features to share representational space. Current methods decompose activations into sparse linear features but discard geometric structure. We…

Machine Learning · Computer Science 2026-02-03 Georgi Ivanov , Narmeen Oozeer , Shivam Raval , Tasana Pejovic , Shriyash Upadhyay , Amir Abdullah

Research in Artificial Intelligence (AI) has focused mostly on two extremes: either on small improvements in narrow AI domains, or on universal theoretical frameworks which are usually uncomputable, incompatible with theories of biological…

We computationally implement and experimentally test the behavioral predictions of a dynamic neural model of lexical meaning in the framework of Dynamic Field Theory. We demonstrate the architecture and behavior of the model using as a test…

Computation and Language · Computer Science 2025-09-18 Michael C. Stern , Maria M. Piñango

Models in which causation arises from higher level structures as well as from microdynamics may be relevant to unifying quantum theory with classical physics or general relativity. They also give a way of defining a form of panprotopsychist…

Quantum Physics · Physics 2020-11-03 Adrian Kent

The effectiveness of a language model is influenced by its token representations, which must encode contextual information and handle the same word form having a plurality of meanings (polysemy). Currently, none of the common language…

Computation and Language · Computer Science 2022-06-02 Andrea Lekkas , Peter Schneider-Kamp , Isabelle Augenstein

In physics we often use very simple models to describe systems with many degrees of freedom, but it is not clear why or how this success can be transferred to the more complex biological context. We consider models for the joint…

Neurons and Cognition · Quantitative Biology 2024-12-06 Luisa Ramirez , William Bialek , Stephanie E. Palmer , David J. Schwab

Comparing the internal representations of neural networks is a central goal in both neuroscience and machine learning. Standard alignment metrics operate on raw neural activations, implicitly assuming that similar representations produce…

Machine Learning · Computer Science 2026-04-02 Sunny Liu , Habon Issa , André Longon , Liv Gorton , Meenakshi Khosla , David Klindt

A central idea in mechanistic interpretability is that neural networks represent more features than they have dimensions, arranging them in superposition to form an over-complete basis. This framing has been influential, motivating…

Machine Learning · Computer Science 2026-03-11 Lucas Prieto , Edward Stevinson , Melih Barsbey , Tolga Birdal , Pedro A. M. Mediano

Understanding how information is represented in neural networks is a fundamental challenge in both neuroscience and artificial intelligence. Despite their nonlinear architectures, recent evidence suggests that neural networks encode…

Machine Learning · Computer Science 2025-03-04 David Klindt , Charles O'Neill , Patrik Reizinger , Harald Maurer , Nina Miolane

We present a method for storing multiple models within a single set of parameters. Models can coexist in superposition and still be retrieved individually. In experiments with neural networks, we show that a surprisingly large number of…

Machine Learning · Computer Science 2019-06-18 Brian Cheung , Alex Terekhov , Yubei Chen , Pulkit Agrawal , Bruno Olshausen

Recent probing studies reveal that large language models exhibit linear subspaces that separate true from false statements, yet the mechanism behind their emergence is unclear. We introduce a transparent, one-layer transformer toy model…

Computation and Language · Computer Science 2025-10-20 Shauli Ravfogel , Gilad Yehudai , Tal Linzen , Joan Bruna , Alberto Bietti

The generalization properties of an attractive network of non monotonic neurons which infers concepts from samples are studied. The macroscopic dynamics for the overlap between the state of the neurons with the concepts, well as the…

Statistical Mechanics · Physics 2009-10-31 D. R. C. Dominguez

Mechanistic interpretability aims to reverse engineer neural networks by uncovering which high-level algorithms they implement. Causal abstraction provides a precise notion of when a network implements an algorithm, i.e., a causal model of…

Machine Learning · Computer Science 2025-03-17 Theodora-Mara Pîslar , Sara Magliacane , Atticus Geiger

Distributional data tells us that a man can swallow candy, but not that a man can swallow a paintball, since this is never attested. However both are physically plausible events. This paper introduces the task of semantic plausibility:…

Computation and Language · Computer Science 2018-04-11 Su Wang , Greg Durrett , Katrin Erk

Artificial Intelligence models are becoming increasingly more powerful and accurate, supporting or even replacing humans' decision making. But with increased power and accuracy also comes higher complexity, making it hard for users to…

Artificial Intelligence · Computer Science 2019-07-10 Vivian S. Silva , André Freitas , Siegfried Handschuh

We demonstrate that final-state uncertainty is ubiquitous in multistable systems of coupled neuronal maps, meaning that predicting whether one such system will eventually be chaotic or nonchaotic is often nearly impossible. We propose a…

Neurons and Cognition · Quantitative Biology 2025-07-22 Brandon B. Le , Bennett Lamb , Luke Benfer , Sriharsha Sambangi , Nisal Geemal Vismith , Akshaj Jagarapu

Soft set theory provides a direct framework for parameterized decision modeling by assigning to each attribute (parameter) a subset of a given universe, thereby representing uncertainty in a structured way [1, 2]. Over the past decades, the…

Artificial Intelligence · Computer Science 2026-03-17 Takaaki Fujita , Florentin Smarandache