Laurence Aitchison

10

Papers

24

Total Citations

Papers (10)

Position: Don't Use the CLT in LLM Evals With Fewer Than a Few Hundred Datapoints

Jacobian Sparse Autoencoders: Sparsify Computations, Not Just Activations

Function-Space Learning Rates

Scale-invariant attention

Model-based Bayesian inference of neural activity and connectivity from all-optical interrogation of a neural circuit

Position: Bayesian Deep Learning is Needed in the Age of Large-Scale AI

Tensor Monte Carlo: Particle Methods for the GPU era

Bayesian filtering unifies adaptive and non-adaptive neural network optimization methods

A variational approximate posterior for the deep Wishart process

Taylor TD-learning