training-stack-change

To keep ML training pipelines in sync, coordinate changes across training workflows, selectors, and embedding pipelines to prevent runtime drift and ensure artifact alignment. This process verifies that surface edits match updated contracts.

Training stack change coordination detects edits under src/training and aligns them with evaluation artifacts and runtime-facing outputs. It satisfies documentation, validation, and debt-tracking requirements for evolving artifact contracts.

Validate alignment between training outputs and downstream consumers by applying artifact governance to reflect updated contracts in code and docs. Run make agent-ci-gate with changed files to verify surface alignment.

Yes, this workflow governance process provides a structured review to record debt for misalignments between training artifacts and runtime outputs. It detects surface changes to prevent drift in ML pipelines.

The best way to verify changes across an ML training stack is running make agent-report with environment and changed files to generate a report, followed by agent-ci-gate to enforce validation and debt tracking.