-
GraphiT, Teaching Language Models to Read Graphs
how to use language models to do node classification on graphs.
-
Efficient Parameter Estimation with Hamiltonian Monte Carlo And Its Applications
How to use Hamiltonian dynamics to do Bayesian inference more efficiently.
-
Physics as Computation
How light, heat, chemistry, and noise are replacing transistors as the fundamental substrate of intelligence — a critical survey of six paradigms reshaping the field.