Tom Angsten

Message

Tom Angsten

Message

Do VPD's Explanations Aggregate? An Audit of the Released Decomposition

TL;DR: adVersarial Parameter Decomposition (VPD) decomposes model weights into simple components and then labels each component as "needed here" or "safe to remove here" for each token of the input. These labels make up the "explanation" of that input. A core aspiration of VPD is that inputs' explanations can aggregate...

Sep 30•29

Limitations on the Interpretability of Learned Features from Sparse Dictionary Learning

Overview When I began my attempt to replicate Anthropic's Towards Monosemanticity paper, I had high expectations for how interpretable the extracted features should be. If I judged a feature to be uninterpretable, I attributed this to either sub-optimal sparse autoencoder hyperparameters, a bug in my implementation, or possible shortcomings of...

Jul 30, 2024•6

Ground-Truth Label Imbalance Impairs the Performance of Contrast-Consistent Search (and Other Contrast-Pair-Based Unsupervised Methods)

Summary Contrast-Consistent Search (CCS) is a method for finding truthful directions within the activation spaces of large language models (LLMs) in an unsupervised way, introduced in Burns et al., 2022. However, all experiments in that study involve training datasets that are balanced with respect to the ground-truth labels of the...

Aug 5, 2023•6