Search papers, labs, and topics across Lattice.
Affiliation:
2
0
5
Pre-trained models unlock surprisingly aggressive quantization in federated learning, slashing communication costs by 40% without sacrificing accuracy on MNIST and CIFAR-100.
