Search papers, labs, and topics across Lattice.
1
0
2
Training a neural network with $L^4$ loss enables it to compute more functions than neurons, revealing a surprising efficiency in representation.
