Recent large language models (LLMs) have tended to leverage sparsity to reduce computations, employing the sparsely activated mixture-of-experts (MoE) technique. MoE introduces four modules, including token routing, token communication, expert computation, and expert parallelism, that impact model quality and training efficiency. To enable versatile usage of MoE models, we introduce FSMoE, a flexible training system optimizing task scheduling with three novel techniques: 1) Unified abstraction and online profiling of MoE modules for task scheduling across various MoE implementations. 2) Co-scheduling intra-node and inter-node communications with computations to minimize communication overheads. 3) To support near-optimal task scheduling, we design an adaptive gradient partitioning method for gradient aggregation and a schedule to adaptively pipeline communications and computations. We conduct extensive experiments with configured MoE layers and real-world MoE models on two GPU clusters. Experimental results show that 1) our FSMoE supports four popular types of MoE routing functions and is more efficient than existing implementations (with up to a 1.42$\times$ speedup), and 2) FSMoE outperforms the state-of-the-art MoE training systems (DeepSpeed-MoE and Tutel) by 1.18$\times$-1.22$\times$ on 1458 MoE layers and 1.19$\times$-3.01$\times$ on real-world MoE models based on GPT-2 and Mixtral using a popular routing function.
FSMoE: A Flexible and Scalable Training System for Sparse Mixture-of-Experts Models
FSMoE is a flexible training system that optimizes task scheduling for sparsely activated mixture-of-experts models, improving efficiency and outperforming existing implementations.
- Year
- 2025
- Venue
- arXiv 2025
- Authors
- 8
- Hosting
- Abstract onlyARXIV-DEFAULT
Cite
Notes
Only stored in your browser.
Attribution
- Abstract & full text
- arxiv.org/abs/2501.10714ARXIV-DEFAULT
- TL;DR
- Semantic Scholar