UniAP: Unifying Inter- and Intra-Layer Automatic Parallelism by Mixed Integer Quadratic Programming
This addresses the inefficiency and human effort in manual parallelism for training large models, offering a novel automated solution for distributed learning practitioners.
The paper tackles the problem of sub-optimal solutions in automatic parallelism for distributed deep learning by proposing UniAP, a method that unifies inter- and intra-layer parallelism using mixed integer quadratic programming, resulting in up to 3.80x higher throughput and up to 107x faster strategy optimization compared to state-of-the-art methods.
Distributed learning is commonly used for training deep learning models, especially large models. In distributed learning, manual parallelism (MP) methods demand considerable human effort and have limited flexibility. Hence, automatic parallelism (AP) methods have recently been proposed for automating the parallel strategy optimization process. Existing AP methods suffer from sub-optimal solutions because they do not jointly optimize the two categories of parallel strategies (i.e., inter-layer parallelism and intra-layer parallelism). In this paper, we propose a novel AP method called UniAP, which unifies inter- and intra-layer automatic parallelism by mixed integer quadratic programming. To the best of our knowledge, UniAP is the first parallel method that can jointly optimize the two categories of parallel strategies to find an optimal solution. Experimental results show that UniAP outperforms state-of-the-art methods by up to 3.80$\times$ in throughput and reduces strategy optimization time by up to 107$\times$ across five Transformer-based models.