Search papers, labs, and topics across Lattice.
Affiliation:
1
0
In regimes where high-accuracy gradients are brutally expensive to compute, minimizing a convex objective costs asymptotically no more than evaluating its gradient just once.
