Decoupled DiLoCo enables asynchronous distributed pre-training with zero global downtime under simulated failures while preserving competitive performance on text and vision tasks.
DiLoCo and related methods.An alternative approach to pre-training involves using periodic synchronization across accelerators, reducing net- workbottlenecksintraining
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Decoupled DiLoCo for Resilient Distributed Pre-training
Decoupled DiLoCo enables asynchronous distributed pre-training with zero global downtime under simulated failures while preserving competitive performance on text and vision tasks.