OpenDiLoCo: An Open-Source Framework for Globally Distributed Low-Communication Training

Title: OpenDiLoCo: An Open-Source Framework for Globally Distributed Low-Communication Training
Authors: Sami Jaghouar; Jack Min Ong; Johannes Hagemann
Venue: arXiv
Year: 2024
Canonical: arxiv:2407.07852v1
URL: http://arxiv.org/abs/2407.07852v1
Collector: arxiv
OA: true

ABSTRACT
OpenDiLoCo is an open-source implementation and replication of the Distributed Low-Communication (DiLoCo) training method for large language models. We provide a reproducible implementation of the DiLoCo experiments, offering it within a scalable, decentralized training framework using the Hivemind library. We demonstrate its effectiveness by training a model across two continents and three countries, while maintaining 90-95% compute utilization. Additionally, we conduct ablations studies focusing on the algorithm's compute efficiency, scalability in the number of workers and show that its gradients can be all-reduced using FP16 without any performance degradation. Furthermore, we scale OpenDiLoCo to 3x the size of the original work, demonstrating its effectiveness for billion parameter models.