-
A Geometric Account of Activation Steering through Angle–Norm Decomposition
Decomposing linear activation steering into angle and norm components, and what each contributes to controlling LLM behaviour.
-
Inductive bias in neural networks
What inductive bias is, where it is in neural networks ans why we need it.
-
Important moments regarding doing PhD in London
What you need to know if you consider doing a PhD in the UK (especially if you're international student)