Skip to content

Example: Custom Training Pipeline

Create a fully customized training pipeline.

Overview

Build a pipeline that: - Uses custom data loading - Applies unique preprocessing - Implements custom metrics - Handles edge cases

Custom Configuration

agent:
  type: custom

custom:
  data_loader: my_loader.py
  preprocessor: my_preprocessor.py
  metrics: my_metrics.py

training:
  custom_callback: my_callback.py

Custom Data Loader

# my_loader.py
class CustomDataLoader:
    def __init__(self, path):
        self.path = path

    def load(self):
        # Custom loading logic
        return X, y

Custom Training Loop

# Train with full control
agent = Agent(config)
agent.fit_custom(
    train_data=X_train,
    valid_data=X_val,
    custom_callback=my_callback
)

Advanced Development

See Developer Guide →


Back to Examples →