Alibaba Cloud Container Service for Kubernetes (ACK) provides a deep learning solution that is based on the open source Kubeflow project and the powerful computing capabilities of Alibaba Cloud. This solution allows you to build easy-to-use, open source, and end-to-end deep learning platforms. Deep learning models trained on these platforms can be quickly deployed as API services and integrated into applications. The deep learning solution of ACK allows data and algorithm engineers to use Alibaba Cloud resources for data preparation, model development, model training, and evaluation and prediction.

Features

The deep learning solution of ACK provides the following features:
  • Ease of use: does not require technology expertise to build and manage deep learning platforms.
  • High performance: improves the performance and utilization of heterogeneous computing resources, such as CPU and GPU resources.
  • Integration: supports multiple mainstream deep learning frameworks, such as TensorFlow and PyTorch. Supports custom runtimes. All tools used in this solution are open source.
  • End to end: provides best practices for you to manage the lifecycle of deep learning tasks based on services of Alibaba Cloud.
  • Servitization: allows you to deploy deep learning models as services and integrate them with applications on the cloud.
  • Coordination: supports coordination with data scientists.