More Like this

1. Training Recommender Systems at Scale: Communication-Efficient Model and Data Parallelism

   https://doi.org/10.1145/3447548.3467080  

   Gupta, Vipul; Choudhary, Dhruv; Tang, Peter; Wei, Xiaohan; Wang, Xing; Huang, Yuzhen; Kejariwal, Arun; Ramchandran, Kannan; Mahoney, Michael W. (August 2021, KDD '21: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining)

   In this paper, we consider hybrid parallelism—a paradigm that em- ploys both Data Parallelism (DP) and Model Parallelism (MP)—to scale distributed training of large recommendation models. We propose a compression framework called Dynamic Communication Thresholding (DCT) for communication-efficient hybrid training. DCT filters the entities to be communicated across the network through a simple hard-thresholding function, allowing only the most relevant information to pass through. For communication efficient DP, DCT compresses the parameter gradients sent to the parameter server during model synchronization. The threshold is updated only once every few thousand iterations to reduce the computational overhead of compression. For communication efficient MP, DCT incorporates a novel technique to compress the activations and gradients sent across the network during the forward and backward propagation, respectively. This is done by identifying and updating only the most relevant neurons of the neural network for each training sample in the data. We evaluate DCT on publicly available natural language processing and recommender models and datasets, as well as recommendation systems used in production at Facebook. DCT reduces communication by at least 100× and 20× during DP and MP, respectively. The algorithm has been deployed in production, and it improves end-to-end training time for a state-of-the-art industrial recommender model by 37%, without any loss in performance. 

   more » « less
2. Universal Checkpointing: A Flexible and Efficient Distributed Checkpointing System for Large-Scale DNN Training with Reconfigurable Parallelism

   Lian, Xinyu; Jacobs, Sam_Ade; Kurilenko, Lev; Tanaka, Masahiro; Bekman, Stas; Ruwase, Olatunji; Zhang, Minjia (July 2025, USENIX Association)

   Accepted and published in the Proceedings of the 2025 USENIX Annual Technical Conference (USENIX ATC ’25). Deep neural network (DNN) training continues to scale rapidly in terms of model size, data volume, and sequence length, to the point where multiple machines are required to fit large models for training. Different distributed and parallel training strategies have been developed to support large-scale DNN training by partitioning the training state across GPUs. However, existing DNN training systems provide very limited support for reconfiguring parallelism strategies in the middle of the training via checkpointing. This limitation arises because distributed checkpoints are tightly coupled to specific model parallelism and hardware configurations, preventing large-scale training jobs from efficiently adapting to hardware failures or resource elasticity. This paper presents Universal Checkpointing (UCP), a novel checkpointing system that enables flexible and efficient DNN training with reconfigurable parallelism. UCP overcomes challenges in existing systems by decoupling checkpoint structure from parallel training strategies and hardware configurations. In addition, we present a pattern-based reconfiguration pipeline that enables automatic, flexible, and efficient mapping of checkpoint state to various parallelism strategies. Evaluation on a range of DNN models, including state-of-the-art dense and sparse LLMs, shows that UCP enables reconfiguration for a broader set of widely used parallelism strategies than existing solutions while adding negligible reconfiguration cost. UCP has been successfully employed in real LLM training workloads, greatly enhancing their flexibility and resilience to dynamic hardware environments. 

   more » « less
3. Accordion: Adaptive Gradient Communication via Critical Learning Regime Identification

   Agarwal, S.; Wang, H.; Lee, K.; Venkataraman, S.; Papailiopoulos, D. (April 2021, Fourth Conference on Machine Learning and Systems)

   null (Ed.)

   Distributed model training suffers from communication bottlenecks due to frequent model updates transmitted across compute nodes. To alleviate these bottlenecks, practitioners use gradient compression techniques like sparsification, quantization, or low-rank updates. The techniques usually require choosing a static compression ratio, often requiring users to balance the trade-off between model accuracy and per-iteration speedup. In this work, we show that such performance degradation due to choosing a high compression ratio is not fundamental. An adaptive compression strategy can reduce communication while maintaining final test accuracy. Inspired by recent findings on critical learning regimes, in which small gradient errors can have irrecoverable impact on model performance, we propose Accordion a simple yet effective adaptive compression algorithm. While Accordion maintains a high enough compression rate on average, it avoids over-compressing gradients whenever in critical learning regimes, detected by a simple gradient-norm based criterion. Our extensive experimental study over a number of machine learning tasks in distributed environments indicates that Accordion, maintains similar model accuracy to uncompressed training, yet achieves up to 5.5x better compression and up to 4.1x end-to-end speedup over static approaches. We show that Accordion also works for adjusting the batch size, another popular strategy for alleviating communication bottlenecks. 

   more » « less
4. Adaptive Gradient Communication via Critical Learning Regime Identification

   Agarwal, Saurabh; Wang, Hongyi; Lee, Kangwook; Venkataraman, Shivaram; Papailiopoulos, Dimitris (January 2021, Proceedings of Machine Learning and Systems)

   Smola, A.; Dimakis, A.; Stoica, I. (Ed.)

   Distributed model training suffers from communication bottlenecks due to frequent model updates transmitted across compute nodes. To alleviate these bottlenecks, practitioners use gradient compression techniques like sparsification, quantization, low rank updates etc. The techniques usually require choosing a static compression ratio, often requiring users to balance the trade-off between model accuracy and per-iteration speedup. In this work, we show that such performance degradation due to choosing a high compression ratio is not fundamental and that an adaptive compression strategy can reduce communication while maintaining final test accuracy.Inspired by recent findings on critical learning regimes, in which small gradient errors can have irrecoverable impact on model performance, we propose ACCORDION a simple yet effective adaptive compression algorithm. While ACCORDION maintains a high enough compression rate on average, it avoids detrimental impact by not compressing gradients too much whenever in critical learning regimes, detected by a simple gradient-norm based criterion. Our extensive experimental study over a number of machine learning tasks in distributed environments indicates that ACCORDION, maintains similar model accuracy to uncompressed training, yet achieves up to 5.5×better compression and up to 4.1×end-to-end speedup over static approaches. We show that ACCORDION also works for adjusting the batch size, another popular strategy for alleviating communication bottlenecks. Our code is available at https://github.com/uw-mad-dash/Accordion 

   more » « less
5. Step-Ahead Error Feedback for Distributed Training with Compressed Gradient

   An Xu, Zhouyuan Huo (February 2021, 35th AAAI Conference on Artificial Intelligence (AAAI 2021))

   null (Ed.)

   Although the distributed machine learning methods can speed up the training of large deep neural networks, the communication cost has become the non-negligible bottleneck to constrain the performance. To address this challenge, the gradient compression based communication-efficient distributed learning methods were designed to reduce the communication cost, and more recently the local error feedback was incorporated to compensate for the corresponding performance loss. However, in this paper, we will show that a new "gradient mismatch" problem is raised by the local error feedback in centralized distributed training and can lead to degraded performance compared with full-precision training. To solve this critical problem, we propose two novel techniques, 1) step ahead and 2) error averaging, with rigorous theoretical analysis. Both our theoretical and empirical results show that our new methods can handle the "gradient mismatch" problem. The experimental results show that we can even train faster with common gradient compression schemes than both the full-precision training and local error feedback regarding the training epochs and without performance loss. 

   more » « less