GPU平台上大语言模型推理工作负载的请求与词元能耗成本特征分析.docx

GPU平台上大语言模型推理工作负载的请求与词元能耗成本特征分析.docx

CharacterizationofRequestandTokenEnergyCostsforLLMInferenceWorkloadsonGPUPlatforms

PrabhuVellaisamy,VanessaLam,ShawnBlanton,andJohnPaulShen

ElectricalandComputerEngineeringDepartment

CarnegieMellonUniversity

Pittsburgh,PA,USA

arXiv:260828044v][cs.PF]28Aug20

您可能关注的文档

文档评论(0)

1亿VIP精品文档

相关文档