Custom Training Phase Tutorial#

Illustrate how to create bespoke phases (warm-up, fine-tuning, evaluation) and plug them into the trainer. Includes callback orchestration and scheduler considerations.

Scenario#

Describe the business or research question being addressed and why this workflow matters.

Setup#

Outline dataset prep, FeatureSet definitions, and experiment context initialization.

Training#

Demonstrate how to assemble nodes, losses, optimizers, and training phases for this scenario.

Evaluation#

Summarize metrics, qualitative checks, and how to inspect :class:LossRecord outputs.

What’s Next#

Link to relevant how-to guides, explanations, or follow-up tutorials.