Platform For AI:Development workflows

The modules of PAI are displayed in the left-side navigation pane of the workspace details page. You can use different modules for AI development based on your business scenarios. This section describes how to use the modules in common business scenarios.

• Cloud-native development

  

  Module	Description	Reference
  ①	High-quality datasets are essential to high-precision models. To manage datasets in a centralized manner and prepare for data labeling and model training, you can use the dataset management module to register public datasets, datasets on on-premises machines, and datasets in Alibaba Cloud storage services. You can also create index datasets by scanning Object Storage Service (OSS) objects.	Create and manage datasets
  ②	Data Science Workshop (DSW) is a cloud-based, interactive integrated development environment (IDE) for machine learning. You can use DSW notebooks to read data, develop algorithms, and train and deploy models.	DSW overview
  ③	The image management module allows you to use the official PAI images or custom images and manage the images in a centralized manner.	Custom images
  ④	Deep Learning Containers (DLC) provides a flexible, stable, easy-to-use, and high-performance training environment for machine learning. DLC supports mainstream and custom algorithm frameworks and allows you to run ultra-large-scale distributed deep learning tasks.	Before you begin
  ⑤	You can use the datasets that are stored in Apsara File Storage NAS (NAS) or OSS and the code that is stored in Git repositories when you submit training jobs.	Before you begin
  ⑥	The model management module allows you to manage trained models in a centralized manner and deploy a model to Elastic Algorithm Service (EAS) as an online service.	Register and manage models
  ⑦	EAS allows you to load models to CPUs and GPUs at the same time, provides high throughput with low latency, and supports real-time auto scaling. You can deploy a large-scale and complex model with a few clicks. Note EAS does not support DSW images or CPFS datasets.	EAS overview

• AI and big data

  

  Module	Description	Reference
  ①	The source data is stored in MaxCompute tables, preprocessed in DataWorks, and then used for model training in PAI.	Create tables Import data to tables Data development process
  ②	Machine Learning Designer can combine streaming training and batch training and supports large-scale distributed training for traditional machine learning, deep learning, and reinforcement learning. Machine Learning Designer provides hundreds of algorithm components that support automatic parameter tuning. You can drag the components to create models with minimal coding.	Overview of Machine Learning Designer
  ③	DataWorks schedules tasks based on the scheduling parameters and time properties that you configured.	Supported formats of scheduling parameters Configure time properties
  ④	The job management module allows you to store the pipeline data of Machine Learning Designer and the execution records of custom jobs. This helps you compare the results of different pipelines or jobs.	Create and manage container training jobs
  ⑤	The model management module allows you to manage trained models in a centralized manner and deploy a model to EAS as an online service.	Register and manage models
  ⑥	EAS allows you to load models to CPUs and GPUs at the same time, provides high throughput with low latency, and supports real-time auto scaling. You can deploy a large-scale and complex model with a few clicks. Note EAS does not support DSW images or CPFS datasets.	EAS overview