Large neural network models are commonly trained through a combination of advanced parallelism strategies in a single program, multiple data (SPMD) paradigm. For example, training large transformer models requires combining data, model, and pipeline partitioning; and optimizer sharding techniques. However, identifying efficient combinations for many model architectures and accelerator systems requires significant manual analysis. In this work, we present an automatic partitioner that identifies these combinations through a goal-oriented search. Our key findings are that a Monte Carlo Tree Search-based partitioner leveraging partition-specific compiler analysis directly into the search and guided goals matches expert-level strategies for various models.
@article{arxiv.2210.06352,
title = {Automatic Discovery of Composite SPMD Partitioning Strategies in PartIR},
author = {Sami Alabed and Dominik Grewe and Juliana Franco and Bart Chrzaszcz and Tom Natan and Tamara Norman and Norman A. Rink and Dimitrios Vytiniotis and Michael Schaarschmidt},
journal= {arXiv preprint arXiv:2210.06352},
year = {2022}
}