Automatic partitioning of machine learning models for training across multiple devices

US12189717B1 · kind B1 · utility

Assignee

• AMAZON TECHNOLOGIES, INC. · US

Inventors

• Can Karakus · Redwood City, US
• Rahul Raghavendra Huilgol · San Jose, US
• Anirudh Subramanian · Redwood City, US
• Fei Wu · Jiaocheng District, CN
• Christopher Cade Daniel · San Jose, US
• Akhil Mehra · Saratoga, US
• Ajay Paidi · Newark, US
• Yutong Zhang · Belmont, US
• Indu Thangakrishnan · Santa Clara, US
• Luis Alves Pereira Quintela · Mountain View, US

Key dates

• Filing date: Nov 27, 2020
• Grant date: Jan 7, 2025
• Expiry date: Nov 9, 2043

Classification

• Technology area (CPC G): Physics
• CPC primary: G06F2209/5017
• WIPO field: Computer technology
• WIPO sector: Electrical engineering

Abstract

Automatic partitioning of a machine learning model may be performed for training across multiple processing devices. A training job for a machine learning model may specify a number of partitions for a machine learning model. An optimization parameter may be determined for the machine learning model. Different partitions of the machine learning model to be trained across multiple processing devices may be determined based on the specified number of partitions and the optimization parameter. A schedule for executing the training job may be generated according to the respective partitions of the machine learning model. The training job may be executed according to the schedule.