6 papers
Beyond Linear Steering: Unified Multi-Attribute Control for Language Models
Narmeen Oozeer, Luke Marks, Shreyans Jain +2
Controlling multiple behavioral attributes in large language models (LLMs) at inference time is a challenging problem due to interference between attributes and the limitations of…
Automated Interpretability Metrics Do Not Distinguish Trained and Random Transformers
Thomas Heap, Tim Lawson, Lucy Farnik +1
Sparse autoencoders (SAEs) are widely used to extract sparse, interpretable latents from transformer activations. We test whether commonly used SAE quality metrics and automatic ex…
Jacobian Sparse Autoencoders: Sparsify Computations, Not Just Activations
Lucy Farnik, Tim Lawson, Conor Houghton +1
Sparse autoencoders (SAEs) have been successfully used to discover sparse and human-interpretable representations of the latent activations of LLMs. However, we would ultimately li…
Residual Stream Analysis with Multi-Layer SAEs
Tim Lawson, Lucy Farnik, Conor Houghton +1
Sparse autoencoders (SAEs) are a promising approach to interpreting the internal representations of transformer language models. However, SAEs are usually trained separately on eac…
STARC: A General Framework For Quantifying Differences Between Reward Functions
Joar Skalse, Lucy Farnik, Sumeet Ramesh Motwani +3
In order to solve a task using reinforcement learning, it is necessary to first formalise the goal of that task as a reward function. However, for many real-world tasks, it is very…
Inducing Human-like Biases in Moral Reasoning Language Models
Artem Karpov, Seong Hah Cho, Austin Meek +3
In this work, we study the alignment (BrainScore) of large language models (LLMs) fine-tuned for moral reasoning on behavioral data and/or brain data of humans performing the same…