Creating Pipelines
Overview​
Abacus.AI allows you to create pipelines to orchestrate and automate tasks within your projects.
Use Cases​
Pipelines can be used to automate various workflows, such as:
- Refreshing data
- Retraining models
- Deploying models
Pipelines are extremely flexible and free-form, allowing you to orchestrate virtually any workflow.
Pipeline Components​
Pipelines consist of three main components:
1. Python Functions​
Contains the core logic and functionality of your pipeline operations.
2. Pipeline Steps​
Defines how Python functions are transformed and executed within the pipeline.
3. Pipeline​
The complete workflow created by connecting all pipeline steps together.
Creating a Pipeline​
Step 1: Navigate to Pipelines​
Click on Pipelines after navigating inside a project. This ensures the pipeline is connected to your project.
Step 2: Select a Template​
Click on Create Pipeline and choose from multiple example templates to get started quickly.
Abacus.AI will open a notebook that you can use to create and configure your pipeline.
Managing Your Pipeline​
Once your pipeline is set up, you can manage it directly from the UI:
- Setup Refresh Schedule: Automate pipeline execution on a schedule
- Change Input Mappings: Modify input parameters and configurations
- Edit Pipeline: Opens a notebook for making changes to the pipeline logic
Monitoring Pipeline Runs​
At the bottom of the pipeline screen, you can view historical pipeline runs. If you used a logger or print statements in your code, you can view the output in the Logs section.
Handling Pipeline Failures​
The Detail page allows you to reset the pipeline in case of failure. You can selectively reset only the steps that failed, rather than rerunning the entire pipeline.
