TaskVine Hands-on Exercises
The exercises progress from a Linux command to Python functions, declared data, and persistent serverless execution. Complete them in order during the tutorial, or use each page independently afterward.
1. TaskVine Quickstart
Run the TaskVine Quickstart to create a manager, submit five Linux command tasks, declare a shared remote input, connect a worker directly, and collect standard output.
This is the environment and access validation exercise.
2. Matrix Multiplication
Run Matrix Multiplication with TaskVine to execute Python functions
with PythonTask and start workers through vine_factory. Choose either the
in-memory program or its file-based extension.
This exercise introduces Python return values, resource requirements, declared inputs and outputs, and worker sandbox filenames.
3. MobileNet Batch Inference
Run MobileNet Batch Inference to compare two versions of the same image-classification application: ordinary PythonTasks that load the model inside each task, and Function Calls that reuse a model held by a persistent Function Library.
This is the advanced exercise. It may be demonstrated during the live session and completed independently afterward.