Search papers, labs, and topics across Lattice.
School of Mathematics, Sun Yat-sen University
1
0
2
Transformers, like ReLU networks, are universal approximators, and their expressive power for piecewise linear functions grows exponentially with depth.
