A new preprint on a decentralized method GT-NSGDm that achieves optimal convergence under heavy-tailed gradient noise.