Standard Workflow

Workflow is a pipeline tool that developers use to deploy models or applications based on real-world business scenarios. Its core function is to break down the complete machine learning task into multiple steps of a workflow, where each step is a manageable component that can be developed, optimized, configured, and automated individually. Workflow helps standardize the process of generating machine learning models, enabling teams to execute AI tasks at a large scale and improve the efficiency of model generation.

ModelArts Workflow provides a standardized MLOps solution that reduces model training costs.

• Supports steps such as data labeling, data processing, model development/training, model evaluation, application development, and application evaluation.
• Automatically coordinates all dependencies between workflow steps, providing features like run records, monitoring, and continuous execution.
• For workflow development, Workflow offers the necessary functionality and parameter descriptions for the pipeline, allowing you to define the steps and their relationships using the SDK.
• For workflow reuse, users can save the pipeline after development, making it available for future use or for other team members, without needing to worry about the algorithms or implementation details included in the pipeline.

Figure 1 Workflow process