Oobleck enables resilient distributed training of large DNN models with guaranteed fault
tolerance. It takes a planning-execution co-design approach, where it first generates a set of …

This paper proposes ElasticFlow, an elastic serverless training platform for distributed deep
learning. ElasticFlow provides a serverless interface with two distinct features:(i) users …

Training large-scale language models is increasingly critical in various domains, but it is
hindered by frequent failures, leading to significant time and economic costs. Current failure …

Deep learning tasks (DLT) include training and inference tasks, where training DLTs have
requirements on minimizing average job completion time (JCT) and inference tasks need …

Large language models (LLMs) represented by chartGPT have achieved profound
applications and breakthroughs in various fields. This demonstrates that LLMs with …

This paper presents the latest advances in machine learning techniques and highlights
deep learning (DL) methods in recent studies. This technology has recently received great …

A robust solution that incorporates fault tolerance and elastic scaling capabilities for
distributed deep learning. Taking advantage of MPI resilient capabilities, aka. User-Level …

In the realm of distributed computing, collective operations involve coordinated
communication and synchronization among multiple processing units, enabling efficient …

After the spread of autonomous driving systems in cars, the next step is the development of
autonomous drone systems. This will have an application in many military and civil fields …

GPU 클러스터의 자원을 효율적으로 관리하기 위해 동적으로 스케일링하는 것이 중요하다.
체크포인트 기반의 중지 후 재개 스케일링이 널리 사용되어 왔지만, 최근 프레임워크는 학습된 …