English
Related papers

Related papers: All is Not Lost: LLM Recovery without Checkpoints

200 papers

LLMs have seen rapid adoption in all domains. They need to be trained on high-end high-performance computing (HPC) infrastructures and ingest massive amounts of input data. Unsurprisingly, at such a large scale, unexpected events (e.g.,…

Distributed, Parallel, and Cluster Computing · Computer Science 2024-06-18 Avinash Maurya , Robert Underwood , M. Mustafa Rafique , Franck Cappello , Bogdan Nicolae

As LLMs and foundation models scale, checkpoint/restore has become a critical pattern for training and inference. With 3D parallelism (tensor, pipeline, data), checkpointing involves many processes, each managing numerous tensors of varying…

Distributed, Parallel, and Cluster Computing · Computer Science 2026-01-01 Mikaila J. Gossman , Avinash Maurya , Bogdan Nicolae , Jon C. Calhoun

While checkpointing is typically combined with a restart of the whole application, localized recovery permits all but the affected processes to continue. In task-based cluster programming, for instance, the application can then be finished…

Distributed, Parallel, and Cluster Computing · Computer Science 2021-03-01 Claudia Fohry

Realistic simulations in engineering or in the materials sciences can consume enormous computing resources and thus require the use of massively parallel supercomputers. The probability of a failure increases both with the runtime and with…

Distributed, Parallel, and Cluster Computing · Computer Science 2018-01-30 Nils Kohl , Johannes Hötzer , Florian Schornbaum , Martin Bauer , Christian Godenschwager , Harald Köstler , Britta Nestler , Ulrich Rüde

Iterative methods are commonly used approaches to solve large, sparse linear systems, which are fundamental operations for many modern scientific simulations. When the large-scale iterative methods are running with a large number of ranks…

Distributed, Parallel, and Cluster Computing · Computer Science 2018-05-30 Dingwen Tao , Sheng Di , Xin Liang , Zizhong Chen , Franck Cappello

Large language models (LLMs) have made a profound impact across various fields due to their advanced capabilities. However, training these models at unprecedented scales requires extensive AI accelerator clusters and sophisticated…

Checkpointing is essential for fault tolerance in training large language models (LLMs). However, existing methods, regardless of their I/O strategies, periodically store the entire model and optimizer states, incurring substantial storage…

Distributed, Parallel, and Cluster Computing · Computer Science 2026-02-26 Minqiu Sun , Xin Huang , Luanzheng Guo , Nathan R. Tallent , Kento Sato , Dong Dai

Efficient construction of checkpoints/snapshots is a critical tool for training and diagnosing deep learning models. In this paper, we propose a lossy compression scheme for checkpoint constructions (called LC-Checkpoint). LC-Checkpoint…

Machine Learning · Computer Science 2020-09-29 Yu Chen , Zhenming Liu , Bin Ren , Xin Jin

Recent developments in large language models (LLMs) have introduced new requirements for efficient and robust training. As LLM clusters scale, node failures, lengthy recoveries, and bulky checkpoints erode efficiency. Infrequent…

Distributed, Parallel, and Cluster Computing · Computer Science 2025-12-04 Bohan Zhao , Yuanhong Wang , Chenglin Liu , Jiagi Pan , Guang Yang , Ruitao Liu , Tingrui Zhang , Kai Luo , Wei Xu

This paper presents Checkmate, a system that enables per-iteration checkpointing in DNN training without any training slowdown. The traditional approach to checkpointing requires a pause in training to copy model states to a separate…

Distributed, Parallel, and Cluster Computing · Computer Science 2025-07-21 Ankit Bhardwaj , Weiyang Wang , Jeremy Carin , Adam Belay , Manya Ghobadi

The success of Google's Pregel framework in distributed graph processing has inspired a surging interest in developing Pregel-like platforms featuring a user-friendly "think like a vertex" programming model. Existing Pregel-like systems…

Distributed, Parallel, and Cluster Computing · Computer Science 2016-01-26 Da Yan , James Cheng , Fan Yang

During the training of Large Language Models (LLMs), tensor data is periodically "checkpointed" to persistent storage to allow recovery of work done in the event of failure. The volume of data that must be copied during each checkpoint,…

Machine Learning · Computer Science 2025-05-16 Daniel Waddington , Cornel Constantinescu

Checkpointing to preserve training states is crucial during the development of Large Foundation Models (LFMs), for training resumption upon various failures or changes in GPU resources and parallelism configurations. In addition, saved…

Artificial Intelligence · Computer Science 2025-04-03 Borui Wan , Mingji Han , Yiyao Sheng , Yanghua Peng , Haibin Lin , Mofan Zhang , Zhichao Lai , Menghan Yu , Junda Zhang , Zuquan Song , Xin Liu , Chuan Wu

In this paper, we revisit traditional checkpointing and rollback recovery strategies, with a focus on silent data corruption errors. Contrarily to fail-stop failures, such latent errors cannot be detected immediately, and a mechanism to…

Distributed, Parallel, and Cluster Computing · Computer Science 2013-11-01 Guillaume Aupy , Anne Benoit , Thomas Hérault , Yves Robert , Frédéric Vivien , Dounia Zaidouni

Deep-learning-based recommendation models (DLRMs) are widely deployed to serve personalized content to users. DLRMs are large in size due to their use of large embedding tables, and are trained by distributing the model across the memory of…

Machine Learning · Computer Science 2021-04-06 Kaige Liu , Jack Kosaian , K. V. Rashmi

NVM-based systems are naturally fit candidates for incorporating periodic checkpointing (or snapshotting). This increases the reliability of the system, makes it more immune to power failures, and reduces wasted work in especially an HPC…

Hardware Architecture · Computer Science 2023-01-30 Akshin Singh , Smruti R. Sarangi

Fault localization (FL) is a critical step in debugging, which typically relies on repeated executions to pinpoint faulty code regions. However, repeated executions can be impractical in the presence of non-deterministic failures or high…

Software Engineering · Computer Science 2026-05-13 Ahmadreza Saboor Yaraghi , Golnaz Gharachorlu , Sakina Fatima , Lionel C. Briand , Ruiyuan Wan , Ruifeng Gao

Large Language Model (LLM) training is frequently interrupted by a heterogeneous spectrum of failures, from common GPU crashes to catastrophic cluster-wide outages. Existing checkpointing systems rely on monolithic, single-tier storage…

Distributed, Parallel, and Cluster Computing · Computer Science 2026-05-19 Shujie Han , Feng Jiang , Patrick P. C. Lee , Xiao Zhang , Zhijie Huang , Nannan Zhao , Xiaonan Zhao , Lichen Pan

The paper proposes and optimizes a partial recovery training system, CPR, for recommendation models. CPR relaxes the consistency requirement by enabling non-failed nodes to proceed without loading checkpoints when a node fails during…

Large language models frequently commit unrecoverable reasoning errors mid-generation: once a wrong step is taken, subsequent tokens compound the mistake rather than correct it. We introduce $\textbf{Latent Phase-Shift Rollback}$ (LPSR): at…

Machine Learning · Computer Science 2026-04-21 Manan Gupta , Dhruv Kumar
‹ Prev 1 2 3 10 Next ›