AI brief
A LessWrong post provides a visual explainer of empirical Neural Tangent Kernels as a mechanistic interpretability method. The author writes that they found the method difficult to grasp at first and produced the explainer to build their own intuitions, noting they are not an expert and that mistakes are possible.
Written by AI from LessWrong's published text.
Source
“There is a new mech interp method in town. I found it difficult to get my head around at first, but once I understood it, I found it beautiful.”
Interpretation or community commentary rather than straight reporting.
Read original story ↗