8 papers · 1 filter
From Mechanistic to Compositional Interpretability
Ward Gauderis, Thomas Dooms, Steven T. Homer +2
Mechanistic interpretability aims to explain neural model behaviour by reverse-engineering learned computational structure into human-understandable components. Without a formal fr…
When Are Two Networks the Same? Tensor Similarity for Mechanistic Interpretability
ML Nissen Gonzalez, Melwina Albuquerque, Laurence Wroe +3
Mechanistic interpretability aims to break models into meaningful parts; verifying that two such parts implement the same computation is a prerequisite. Existing similarity measure…
Bilinear autoencoders find interpretable manifolds
Thomas Dooms, Ward Gauderis, Geraint Wiggins +1
Sparse autoencoders have become a standard tool for uncovering interpretable latent representations in neural networks. Yet salient concepts often span manifolds that current linea…
Finding Manifolds With Bilinear Autoencoders
Thomas Dooms, Ward Gauderis
Sparse autoencoders are a standard tool for uncovering interpretable latent representations in neural networks. Yet, their interpretation depends on the inputs, making their isolat…
Compositionality Unlocks Deep Interpretable Models
Thomas Dooms, Ward Gauderis, Geraint A. Wiggins +1
We propose -net, an intrinsically interpretable architecture combining the compositional multilinear structure of tensor networks with the expressivity and efficiency of deep ne…
Tokenized SAEs: Disentangling SAE Reconstructions
Thomas Dooms, Daniel Wilhelm
Sparse auto-encoders (SAEs) have become a prevalent tool for interpreting language models' inner workings. However, it is unknown how tightly SAE features correspond to computation…