pyexp-guide

Structure reproducible machine learning experiments by defining functions with the @experiment decorator and managing parameters through Config objects. This approach ensures organized, repeatable ML workflows and simplifies complex project tracking.

Manage ML configuration parameters by using Config objects, and perform sweeping by utilizing the built-in sweep and merge functions. This enables flexible parameterization and comprehensive hyperparameter tuning across multiple experiment runs.

Handle dependencies between machine learning experiments by specifying DAG-based relationships using the depends_on attribute. This ensures correct execution order and manages complex workflow pipelines efficiently across your project.

Load and analyze results from previous experiment runs by accessing the Runs collection and extracting data through the Result dataclass. This provides structured access to output metrics for organized analysis and indexing.

Yes, you can merge multiple configurations for a single machine learning workflow using the built-in merge functionality. This allows you to combine base settings with specific overrides, ensuring flexible parameterization across different experiment runs.

No, pyexp does not require external dependencies to manage experiment reproducibility. It operates independently to provide experiment definition, configuration management, and dependency tracking for organized and repeatable ML workflows.