AnalysisAI ModelsJuly 15, 2026
Paper disentangles a convolutional neuron in InceptionV1
A first paper on mechanistic interpretability studies a single 1x1 convolution neuron in InceptionV1. The method is applied to other neurons in the same layer.
A first paper on mechanistic interpretability studies a single 1x1 convolution neuron in InceptionV1. The method is applied to other neurons in the same layer.