2 citations · 2 across the 7 of their papers we have counts for
7 papers
Innocuous-Seeming Data, Latent Ideology: Ideological Generalisation in Finetuned LLMs
Robert Graham, Edward Stevinson, Yariv Barsheshat
Finetuning language models on small, curated datasets is standard practice for adapting them to specific policies or domains. We show that finetuning on narrow, factually-defensibl…
From Data Statistics to Feature Geometry: How Correlations Shape Superposition
Lucas Prieto, Edward Stevinson, Melih Barsbey +2
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 b…
Adversarial Attacks Leverage Interference Between Features in Superposition
Edward Stevinson, Lucas Prieto, Melih Barsbey +1
Why do adversarial examples exist, and why do they transfer between models? Existing explanations appeal to high-dimensional geometry, non-robust patterns in the input, and decisio…
ContextBench: Modifying Contexts for Targeted Latent Activation
Robert Graham, Edward Stevinson, Leo Richter +3
Identifying inputs that trigger specific behaviours or latent features in language models could have a wide range of safety use cases. We investigate a class of methods capable of…
Prisma: An Open Source Toolkit for Mechanistic Interpretability in Vision and Video
Sonia Joseph, Praneet Suresh, Lorenz Hufe +7
Robust tooling and publicly available pre-trained models have helped drive recent advances in mechanistic interpretability for language models. However, similar progress in vision…
A Scalable Approach to Probabilistic Neuro-Symbolic Robustness Verification
Vasileios Manginas, Nikolaos Manginas, Edward Stevinson +4
Neuro-Symbolic Artificial Intelligence (NeSy AI) has emerged as a promising direction for integrating neural learning with symbolic reasoning. Typically, in the probabilistic varia…
Leveraging knowledge graphs to update scientific word embeddings using latent semantic imputation
Jason Hoelscher-Obermaier, Edward Stevinson, Valentin Stauber +4
The most interesting words in scientific texts will often be novel or rare. This presents a challenge for scientific word embedding models to determine quality embedding vectors fo…