~/arbitrary-packingSuperposition as a learning principle: unrelated things share a neuron because they rarely co-occur, so they do not interfere. The essay reports no surviving canonical circuit of this shape.
~/bad-learningThe architecture learned poorly. InceptionV1's last layer had almost no legible neurons; a parameter-matched ConvNeXt with residuals and LayerNorm on the same data is about 99% legible.
no votes on this pair in this scope yet
you’ll log in, then return to this pair to cast your vote.