MLPerf Power
收藏资源简介:
MLPerf Power数据集由MLCommons开发,旨在评估从微瓦到兆瓦级别机器学习系统的能效。该数据集包含1841条来自60个不同系统的测量数据,覆盖了从物联网设备到大型数据中心的广泛应用场景。数据集的创建过程涉及多个组织的协作,确保了测量方法的一致性和可比性。MLPerf Power数据集主要用于研究机器学习系统的能效优化,旨在为未来的可持续AI解决方案提供基准。
The MLPerf Power dataset, developed by MLCommons, aims to evaluate the energy efficiency of machine learning (ML) systems across scales from microwatts to megawatts. This dataset contains 1,841 measurement records from 60 unique systems, spanning a broad spectrum of application scenarios ranging from Internet of Things (IoT) devices to large-scale data centers. The dataset's development involved cross-organizational collaboration to guarantee the consistency and comparability of its measurement methodologies. The MLPerf Power dataset is primarily utilized for research on energy efficiency optimization of ML systems, with the goal of providing standardized benchmarks for future sustainable AI solutions.

- 1MLPerf Power: Benchmarking the Energy Efficiency of Machine Learning Systems from μWatts to MWatts for Sustainable AIMLCommons · 2024年



