Simple Workflow

A basic example demonstrating the structure of a Mixtrain workflow with inputs.

To follow along, import the Hello Workflow example into Mixtrain.

Code

from mixtrain import MixFlow


class ExampleWorkflow(MixFlow):
    def setup(self):
        print("Optional: Setting up workflow...")

    def run(
        self,
        model_name: str,
        learning_rate: float = 0.001,
        batch_size: int = 32,
        epochs: int = 10,
        use_gpu: bool = True,
    ):
        print(f"Training {model_name} for {epochs} epochs")
        print(f"Batch size: {batch_size}")
        print(f"Learning rate: {learning_rate}")
        print(f"GPU: {'enabled' if use_gpu else 'disabled'}")

    def cleanup(self):
        print("Optional: Cleaning up...")

Running the Workflow

Create the workflow

mixtrain workflow create example_workflow.py \
  --name my-workflow \
  --description "Example workflow"

Run with default inputs

mixtrain workflow run my-workflow

Run with custom inputs

mixtrain workflow run my-workflow \
  --input '{"model_name": "resnet50", "epochs": 20, "batch_size": 64}'

Key Concepts

Inputs in `run()` signature

Define inputs as parameters in the run() method:

Pattern	Description
`param: str`	Required input (no default)
`param: int = 10`	Optional input with default

Lifecycle Methods

Method	Description
`setup()`	Initialize resources before execution. Optional.
`run()`	Main workflow logic (required)
`cleanup()`	Release resources after execution. Optional.

Next Steps

Continuous VLM Post-Training - A complete training and evaluation loop
Workflows Guide - Full documentation

On this page