DistLearn: Zero-Order Decentralized Layer-wise Training via Game-Theoretic Block Coordinate Descent Deja un comentario / Por Julian / julio 10, 2026