DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Rhu, Minsoo | - |
dc.contributor.advisor | 유민수 | - |
dc.contributor.author | Bang, Jehyeon | - |
dc.date.accessioned | 2023-06-26T19:34:20Z | - |
dc.date.available | 2023-06-26T19:34:20Z | - |
dc.date.issued | 2023 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=1033105&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/309962 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 전기및전자공학부, 2023.2,[iv, 24 p. :] | - |
dc.description.abstract | As the size of deep neural network (DNN) models is rapidly increasing to improve performance, the demand for compute resources required for DNN training is exponentially increasing. Such large-scale training is performed on distributed systems with various parallelism techniques, and the training performance in distributed systems varies drastically depending on DNN model architecture, the network topology, and the combination of parallelism techniques. However, finding the optimal training configuration incurs immense expenses, leading to the inability to effectively use compute resources in large-scale training. To address this, this thesis proposes a simulation framework to predict the training iteration time of distributed training. The proposed framework accurately predicts the training iteration time for various configurations with a mean absolute error of 12.80%, facilitating efficient exploration for the optimal training configuration. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | Distributed training▼aDeep neural networks▼aSimulation▼aParallelization▼aGPU | - |
dc.subject | 분산 학습▼a심층신경망▼a시뮬레이션▼a병렬화▼a그래픽처리장치 | - |
dc.title | (A) software framework for estimating training time of trillion-parameter scale distributed machine learning | - |
dc.title.alternative | 대규모 분산형 기계학습의 학습 시간 예측을 위한 소프트웨어 프레임워크 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전기및전자공학부, | - |
dc.contributor.alternativeauthor | 방제현 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.