Mixtures of Subspaces for Bandwidth Efficient Context Parallel Training

• URL: http://arxiv.org/abs/2606.16384v1
• Date: Mon, 15 Jun 2026 08:17:13 GMT
• Title: Mixtures of Subspaces for Bandwidth Efficient Context Parallel Training
• Abstract summary: We propose a compression method for communication-efficient context parallelism in decentralized settings.<n>Our key insight is to exploit the intrinsic low-rank structure of activation outputs by dynamically constraining them to learned mixtures of subspaces.
• Score: 54.43685601795611
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Pretraining language models with extended context windows enhances their ability to leverage rich information during generation. Existing methods split input sequences into chunks, broadcast them across multiple devices, and compute attention block by block which incurs significant communication overhead. While feasible in high-speed clusters, these methods are impractical for decentralized training over low-bandwidth connections. We propose a compression method for communication-efficient context parallelism in decentralized settings, achieving a remarkable compression rate of over 95\% with negligible overhead and no loss in convergence. Our key insight is to exploit the intrinsic low-rank structure of activation outputs by dynamically constraining them to learned mixtures of subspaces via efficient reparameterizations. We demonstrate scaling billion-parameter decentralized models to context lengths exceeding 100K tokens on networks as slow as 300Mbps, matching the wall-clock convergence speed of centralized models on 100Gbps interconnects.

Related papers

• Communication-Efficient LLM Adaptation over Decentralized GPU Meshes [48.95736135870457]
  We propose an asynchronous two-circuit system for post-pretraining adaptation.<n>A fast compressed training circuit drives throughput using masking activation for pipeline-parallel transfer and compressed data-parallel synchronization.<n>We show that masking supports post-pretraining adaptation at high compression rates when anchored this way.
  arXiv  Detail & Related papers  (2026-09-13T06:56:50Z)
• AsyncMesh: Fully Asynchronous Optimization for Data and Pipeline Parallelism [54.8494905524997]
  We introduce asynchronous updates across both parallelism axes, relaxing the co-location requirement.<n>We provide convergence guarantees for both sparse averaging and asynchronous updates.<n>Experiments on large-scale language models demonstrate that our approach matches the performance of the fully synchronous baseline.
  arXiv  Detail & Related papers  (2026-01-30T01:24:47Z)
• Distributed Low-Communication Training with Decoupled Momentum Optimization [38.33322656231618]
  Training large models requires substantial computational resources, typically available only in data centers with high-bandwidth interconnects.<n>We propose an approach that further reduces communication by combining infrequent synchronizations across distributed model replicas with momentum gradient compression.<n>In particular, we treat the momentum as a signal and decompose the Nesterov momentum into high- and low-frequency components via the discrete cosine transform.
  arXiv  Detail & Related papers  (2025-10-03T08:25:21Z)
• DiLoCoX: A Low-Communication Large-Scale Training Framework for Decentralized Cluster [7.597885871452736]
  We propose DiLoCoX, a low-communication large-scale decentralized cluster training framework.<n>It combines Pipeline Parallelism with Dual-Step-Delay Overlap of Communication and Local Training, and an Adaptive Gradient Compression Scheme.<n>We show that DiLoCoX can achieve a 357x speedup in distributed training while maintaining negligible degradation in model convergence.
  arXiv  Detail & Related papers  (2025-06-26T13:45:04Z)
• Protocol Models: Scaling Decentralized Training with Communication-Efficient Model Parallelism [59.79227116582264]
  Scaling models has led to significant advancements in deep learning, but training these models in decentralized settings remains challenging.<n>We propose a novel compression algorithm that compresses both forward and backward passes, enabling up to 99% compression with no convergence degradation.
  arXiv  Detail & Related papers  (2025-06-02T02:19:22Z)
• Flash Communication: Reducing Tensor Parallelization Bottleneck for Fast Large Language Model Inference [14.805702987440512]
  We introduce Flash Communication, a novel low-bit compression technique designed to alleviate the tensor-parallelism communication bottleneck during inference.<n>Our method substantially boosts intra-node communication speed by more than 3x and reduces the time-to-first-token by 2x, with nearly no sacrifice in model accuracy.
  arXiv  Detail & Related papers  (2024-12-06T11:29:32Z)
• Accelerating Distributed Deep Learning using Lossless Homomorphic Compression [17.654138014999326]
  We introduce a novel compression algorithm that effectively merges worker-level compression with in-network aggregation. We show up to a 6.33$times$ improvement in aggregation throughput and a 3.74$times$ increase in per-iteration training speed.
  arXiv  Detail & Related papers  (2024-02-12T09:57:47Z)
• Asynchronous Parallel Incremental Block-Coordinate Descent for Decentralized Machine Learning [55.198301429316125]
  Machine learning (ML) is a key technique for big-data-driven modelling and analysis of massive Internet of Things (IoT) based intelligent and ubiquitous computing. For fast-increasing applications and data amounts, distributed learning is a promising emerging paradigm since it is often impractical or inefficient to share/aggregate data. This paper studies the problem of training an ML model over decentralized systems, where data are distributed over many user devices.
  arXiv  Detail & Related papers  (2022-02-07T15:04:15Z)
• PowerGossip: Practical Low-Rank Communication Compression in Decentralized Deep Learning [62.440827696638664]
  We introduce a simple algorithm that directly compresses the model differences between neighboring workers. Inspired by the PowerSGD for centralized deep learning, this algorithm uses power steps to maximize the information transferred per bit.
  arXiv  Detail & Related papers  (2020-08-04T09:14:52Z)
• Joint Parameter-and-Bandwidth Allocation for Improving the Efficiency of Partitioned Edge Learning [73.82875010696849]
  Machine learning algorithms are deployed at the network edge for training artificial intelligence (AI) models. This paper focuses on the novel joint design of parameter (computation load) allocation and bandwidth allocation.
  arXiv  Detail & Related papers  (2020-03-10T05:52:15Z)

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.