paper-with-me

Papers

Taming the Memory Beast: Strategies for Reliable ML Training on Kubernetes

2024-12-19 · Jaideep Ray

Kubernetes offers a powerful orchestration platform for machine learning training, but memory management can be challenging due to specialized needs and resource constraints. This paper outlines how Kubernetes handles memory requests, limits, Quality of Service classes, and eviction policies for ML workloads, with special focus on GPU memory and ephemeral storage. Common pitfalls such as overcommitment, memory leaks, and ephemeral volume exhaustion are examined. We then provide best practices for stable, scalable memory utilization to help ML practitioners prevent out-of-memory events and ensure high-performance ML training pipelines.

📄 PDF Abstract BibTeX arXiv:2412.14701

Code (0)

등록된 구현이 없습니다.

Tasks

GPUManagement

Methods 이 논문이 사용한 방법론

Golden Queue Managers 설명 없음
Focus 설명 없음

Similar Papers 제목 키워드 기반

Harnessing intuitive local evolution rules for physical learning

2025-07-25 · Roie Ezraty, Menachem Stern, Shmuel M. Rubinstein arxiv

Machine Learning, however popular and accessible, is computationally intensive and highly power-consuming, prompting interest in alternative physical implementations of learning tasks. We introduce a training scheme for …

TorchBeast: A PyTorch Platform for Distributed RL

2019-10-08 · Heinrich Küttler, Nantas Nardelli, Thibaut Lavril, Marco Selvatici 외

TorchBeast is a platform for reinforcement learning (RL) research in PyTorch. It implements a version of the popular IMPALA algorithm for fast, asynchronous, parallel training of RL agents. Additionally, TorchBeast has s…

OpenAI GymReinforcement LearningReinforcement Learning (RL)

BEAST: Efficient Tokenization of B-Splines Encoded Action Sequences for Imitation Learning

2025-06-06 · Hongyi Zhou, Weiran Liao, Xi Huang, Yucheng Tang 외

We present the B-spline Encoded Action Sequence Tokenizer (BEAST), a novel action tokenizer that encodes action sequences into compact discrete or continuous tokens using B-splines. In contrast to existing action tokeniz…

continuous-controlContinuous ControlDecoderImitation Learning+1

Predicting human decisions with behavioral theories and machine learning

2019-04-15 · Ori Plonsky, Reut Apel, Eyal Ert, Moshe Tennenholtz 외

Predicting human decisions under risk and uncertainty remains a fundamental challenge across disciplines. Existing models often struggle even in highly stylized tasks like choice between lotteries. We introduce BEAST Gra…

BIG-bench Machine LearningDecision MakingDescriptiveDomain Generalization

Fast Adversarial Attacks on Language Models In One GPU Minute

2024-02-23 · Vinu Sankar Sadasivan, Shoumik Saha, Gaurang Sriramanan, Priyatham Kattakinda 외

In this paper, we introduce a novel class of fast, beam search-based adversarial attack (BEAST) for Language Models (LMs). BEAST employs interpretable parameters, enabling attackers to balance between attack speed, succe…

Adversarial AttackComputational EfficiencyGPU