StarTrail: Concentric Ring Sequence Parallelism for Efficient Near-Infinite-Context Transformer Model Training

TL;DR

StarTrail introduces a multi-dimensional parallelism approach for efficient training of long-sequence Transformer models, significantly reducing communication overhead and improving performance across NLP and CV tasks.

Contribution

It proposes a novel concentric ring parallelism method with an extra dimension and sub-ring communication to enhance scalability and efficiency in long-sequence Transformer training.

Findings

Achieves up to 77.12% performance improvement on GPT models

Achieves up to 114.33% performance improvement on DiT models

Reduces communication volume and bandwidth bottlenecks significantly

Abstract

Training Transformer models on long sequences in a distributed setting poses significant challenges in terms of efficiency and scalability. Current methods are either constrained by the number of attention heads or excessive communication overheads. To address this problem, we propose StarTrail, a multi-dimensional concentric distributed training system for long sequences, fostering an efficient communication paradigm and providing additional tuning flexibility for communication arrangements. Specifically, StarTrail introduces an extra parallel dimension and divides the peer-to-peer communication into sub-rings to substantially reduce communication volume and avoid bandwidth bottlenecks. Through comprehensive experiments across diverse hardware environments and on both Natural Language Processing (NLP) and Computer Vision (CV) tasks, we demonstrate that our approach significantly surpasses state-of-the-art methods that support Long sequence lengths, achieving performance improvements of up to 77.12% on GPT-style models and up to 114.33% on DiT (Diffusion Transformer) models without affecting the computations results.