What (philosophically valid) method could be used to tell whether or not a neuron (or set of neurons/feature/head) is related to lexical disambiguation of some words? #16

Open Zmavli opened 1 month ago