FlexSP: Accelerating Large Language Model Training via Flexible Sequence Parallelism

• URL: http://arxiv.org/abs/2412.01523v3
• Date: Tue, 11 Feb 2025 07:31:03 GMT
• Title: FlexSP: Accelerating Large Language Model Training via Flexible Sequence Parallelism
• Authors: Yujie Wang, Shiju Wang, Shenhan Zhu, Fangcheng Fu, Xinyi Liu, Xuefeng Xiao, Huixia Li, Jiashi Li, Faming Wu, Bin Cui,
• Abstract summary: Existing sequence parallelism methods assume homogeneous sequence lengths (i.e., all input sequences are equal in length) and therefore leverages a single, static scattering strategy for all input sequences. We show that the sequence lengths in LLM training corpora exhibit substantial variability, often following a long-tail distribution. We propose a Heterogeneous-adaptive sequence parallelism method to address this problem.
• Score: 33.23902060961886
• License:
• Abstract: Extending the context length (i.e., the maximum supported sequence length) of LLMs is of paramount significance. To facilitate long context training of LLMs, sequence parallelism has emerged as an essential technique, which scatters each input sequence across multiple devices and necessitates communication to process the sequence. In essence, existing sequence parallelism methods assume homogeneous sequence lengths (i.e., all input sequences are equal in length) and therefore leverages a single, static scattering strategy for all input sequences. However, in reality, the sequence lengths in LLM training corpora exhibit substantial variability, often following a long-tail distribution, which leads to workload heterogeneity. In this paper, we show that employing a single, static strategy results in inefficiency and resource under-utilization, highlighting the need for adaptive approaches to handle the heterogeneous workloads across sequences. To address this, we propose a heterogeneity-adaptive sequence parallelism method. For each training step, our approach captures the variability in sequence lengths and assigns the optimal combination of scattering strategies based on workload characteristics. We model this problem as a linear programming optimization and design an efficient and effective solver to find the optimal solution. Furthermore, we implement our method in a high-performance system that supports adaptive parallelization in distributed LLM training. Experimental results demonstrate that our system outperforms state-of-the-art training frameworks by up to 1.98x.

Related papers

• Transformer-Squared: Self-adaptive LLMs [29.1326358746118]
  We introduce Transformer-Squared, a novel self-adaptation framework that adapts large language models for unseen tasks in real-time. Our method consistently outperforms ubiquitous approaches such as LoRA, with fewer parameters and greater efficiency. Transformer-Squared represents a significant leap forward, offering a scalable, efficient solution for enhancing the adaptability and task-specific performance of LLMs.
  arXiv  Detail & Related papers  (2025-01-09T01:19:21Z)
• Demystifying Workload Imbalances in Large Transformer Model Training over Variable-length Sequences [31.232756326457277]
  We develop Hydraulis, which jointly optimize the parallel strategies and data assignment. Empirical results demonstrate that Hydraulis outperforms existing systems by 1.32-2.66 times.
  arXiv  Detail & Related papers  (2024-12-10T20:01:53Z)
• Short-Long Convolutions Help Hardware-Efficient Linear Attention to Focus on Long Sequences [60.489682735061415]
  We propose CHELA, which replaces state space models with short-long convolutions and implements linear attention in a divide-and-conquer manner. Our experiments on the Long Range Arena benchmark and language modeling tasks demonstrate the effectiveness of the proposed method.
  arXiv  Detail & Related papers  (2024-06-12T12:12:38Z)
• Linear Attention Sequence Parallelism [33.06590170649837]
  We introduce Linear Attention Sequence Parallelism (LASP) for linear attention-based transformer models. LASP takes advantage of the right-product kernel trick of linear attention, which sharply decreases the communication overhead. LASP scales sequence length up to 4096K on 128 GPUs, which is 8$times$ longer than existing SP methods.
  arXiv  Detail & Related papers  (2024-04-03T17:33:21Z)
• How Many Pretraining Tasks Are Needed for In-Context Learning of Linear Regression? [92.90857135952231]
  Transformers pretrained on diverse tasks exhibit remarkable in-context learning (ICL) capabilities. We study ICL in one of its simplest setups: pretraining a linearly parameterized single-layer linear attention model for linear regression.
  arXiv  Detail & Related papers  (2023-10-12T15:01:43Z)
• Amortizing intractable inference in large language models [56.92471123778389]
  We use amortized Bayesian inference to sample from intractable posterior distributions. We empirically demonstrate that this distribution-matching paradigm of LLM fine-tuning can serve as an effective alternative to maximum-likelihood training. As an important application, we interpret chain-of-thought reasoning as a latent variable modeling problem.
  arXiv  Detail & Related papers  (2023-10-06T16:36:08Z)
• Low-Rank Multitask Learning based on Tensorized SVMs and LSSVMs [65.42104819071444]
  Multitask learning (MTL) leverages task-relatedness to enhance performance. We employ high-order tensors, with each mode corresponding to a task index, to naturally represent tasks referenced by multiple indices. We propose a general framework of low-rank MTL methods with tensorized support vector machines (SVMs) and least square support vector machines (LSSVMs)
  arXiv  Detail & Related papers  (2023-08-30T14:28:26Z)
• Improving Automatic Parallel Training via Balanced Memory Workload Optimization [36.87527680184956]
  Transformer models have emerged as the leading approach for achieving state-of-the-art performance across various application domains. We present Galvatron-BMW, a novel system framework that integrates multiple parallelism prevalent dimensions and automatically identifies the most efficient hybrid parallelism strategy. Our evaluations on different Transformer models demonstrate the capabilities of Galvatron-BMW in automating distributed training under varying GPU memory constraints.
  arXiv  Detail & Related papers  (2023-07-05T05:28:38Z)
• EOS: a Parallel, Self-Adaptive, Multi-Population Evolutionary Algorithm for Constrained Global Optimization [68.8204255655161]
  EOS is a global optimization algorithm for constrained and unconstrained problems of real-valued variables. It implements a number of improvements to the well-known Differential Evolution (DE) algorithm. Results prove that EOSis capable of achieving increased performance compared to state-of-the-art single-population self-adaptive DE algorithms.
  arXiv  Detail & Related papers  (2020-07-09T10:19:22Z)
• Accelerating Feedforward Computation via Parallel Nonlinear Equation Solving [106.63673243937492]
  Feedforward computation, such as evaluating a neural network or sampling from an autoregressive model, is ubiquitous in machine learning. We frame the task of feedforward computation as solving a system of nonlinear equations. We then propose to find the solution using a Jacobi or Gauss-Seidel fixed-point method, as well as hybrid methods of both. Our method is guaranteed to give exactly the same values as the original feedforward computation with a reduced (or equal) number of parallelizable iterations, and hence reduced time given sufficient parallel computing power.
  arXiv  Detail & Related papers  (2020-02-10T10:11:31Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.