neuron explanation

deep neural network (3) concept discovery (2) concept synthesis (1) textual description (1) robustness analysis (1) concept labels (1) sanity checks (1) evaluation metric (1) mechanistic interpretability (1) adversarial perturbation (1)