Related papers: Disentangling Dense Embeddings with Sparse Autoenc…

Sparse autoencoders (SAEs) have received considerable recent attention as tools for mechanistic interpretability, showing success at extracting interpretable features even from very large LLMs. However, this research has been largely…

Disentangling the encodings of neural models is a fundamental aspect for improving interpretability, semantic control and downstream task performance in Natural Language Processing. Currently, most disentanglement methods are unsupervised…

Sparse Autoencoders (SAEs) have shown to find interpretable features in neural networks from polysemantic neurons caused by superposition. Previous work has shown SAEs are an effective tool to extract interpretable features from the early…

While large language models provide strong compositional reasoning, existing reasoning segmentation pipelines fail to transparently connect this reasoning to visual perception. Current methods, such as latent query alignment, are end-to-end…

Steering vectors are a promising approach to control the behaviour of large language models. However, their underlying mechanisms remain poorly understood. While sparse autoencoders (SAEs) may offer a potential method to interpret steering…

This work presents a framework based on feature disentanglement to learn speaker embeddings that are robust to environmental variations. Our framework utilises an auto-encoder as a disentangler, dividing the input speaker embedding into…

One of the roadblocks to a better understanding of neural networks' internals is \textit{polysemanticity}, where neurons appear to activate in multiple, semantically distinct contexts. Polysemanticity prevents us from identifying concise,…

The linear representation hypothesis states that neural network activations encode high-level concepts as linear mixtures. However, under superposition, this encoding is a projection from a higher-dimensional concept space into a…

Sparse autoencoders (SAEs) offer a natural path toward comparable explanations across different representation spaces. However, current SAEs are trained per modality, producing dictionaries whose features are not directly understandable and…

Adapting foundation models for specific purposes has become a standard approach to build machine learning systems for downstream applications. Yet, it is an open question which mechanisms take place during adaptation. Here we develop a new…

Sparse autoencoders (SAEs) \citep{bricken2023monosemanticity,gao2024scalingevaluatingsparseautoencoders} rely on dictionary learning to extract interpretable features from neural networks at scale in an unsupervised manner, with…

What latent features are encoded in language model (LM) representations? Recent work on training sparse autoencoders (SAEs) to disentangle interpretable features in LM representations has shown significant promise. However, evaluating the…

Sparse autoencoders (SAEs) enable feature-level mechanistic interpretability and activation steering in large language models (LLMs), but SAE-based language control remains unreliable in multilingual settings: most SAEs are trained on…

Sparse autoencoders (SAEs) have emerged as a powerful technique for decomposing language model representations into interpretable features. Current interpretation methods infer feature semantics from activation patterns, but overlook that…

Scientific datasets present unique challenges for machine learning-driven compression methods, including more stringent requirements on accuracy and mitigation of potential invalidating artifacts. Drawing on results from compressed sensing…

Text-to-image diffusion models generate images through an iterative denoising process, so internal neural layers produce trajectories of activations rather than single static representations. Sparse autoencoders (SAEs) have recently been…

How do computers and intelligent agents view the world around them? Feature extraction and representation constitutes one the basic building blocks towards answering this question. Traditionally, this has been done with carefully engineered…

The fidelity with which neural networks can now generate content such as music presents a scientific opportunity: these systems appear to have learned implicit theories of such content's structure through statistical learning alone. This…

Sparse autoencoders (SAEs) and transcoders have become important tools for machine learning interpretability. However, measuring how interpretable they are remains challenging, with weak consensus about which benchmarks to use. Most…

Sparse Autoencoders (SAEs) have emerged as a predominant tool in mechanistic interpretability, aiming to identify interpretable monosemantic features. However, how does sparse encoding organize the representations of activation vector from…