Hyperparameter tuning selects values from a defined search space by optimizing an objective such as a loss function. Effective searches automate evaluation while keeping the model, configuration, data, and code reproducible. section 2.1.3
A production-oriented tuning workflow:
- Define the search space and loss function rather than changing values manually. section 2.1.3
- Represent the trained model as an artifact produced by applying configurations and data to modular code. section 2.1.3
- Record configurations, results, data versions, checkpoints, and model versions to preserve lineage and traceability. section 2.1.3
Figure 2.5 A view of a retraining pipeline using the modular codebase concept. This approach of keeping the model, code, configuration files, and data as distinct versioned components with lineage links ensures that the process remains flexible, fast, and adaptable while enabling experimentation, debugging, and iterative development.
Use the retraining-pipeline view to connect automated tuning with distinct versioned models, code, configuration files, and data. section 2.1.3
