Training language models across low-bandwidth networks requires reducing synchronization overhead. Local optimization methods such as DiLoCo communicate only after several local steps, but outer gradient averaging and the outer update remain on the critical path. Existing computation-communication overlap methods avoid this stall through delayed or replica-local updates, which can degrade convergence or cause replicas to drift. We propose LACCO (Local Accumulation with Computation-Communication Overlap), a two-stage algorithm that overlaps global communication with local computation while applying globally aggregated updates shared by every replica, without introducing staleness in the outer update. We establish convergence guarantees for smooth non-convex objectives. Experiments on language-model pre-training with FineWeb show that LACCO consistently outperforms communication-overlapped baselines such as Eager and Delayed DiLoCo. These gains remain robust as model size scales up to 0.5B parameters and across a range of communication intervals and numbers of replicas up to 16 workers, while incurring only negligible-or fully overlappable-computational or memory overhead, which demonstrates the strength of LACCO.
Communication-efficient distributed training reduces the data exchanged among devices so models can be trained faster and over limited-bandwidth networks.
Department of Electrical Engineering, California Institute of Technology · Department of Computer Science and Technology, Tsinghua University · Department of Computer Sciences, University of Wisconsin - Madison