FeedForward Training Module
Feedforward TE training entry point and terminal-reporting helpers.
- scripts.training.train_feedforward_network.load_training_config(config_path=DEFAULT_CONFIG_PATH)[source]
Load one training configuration file through the shared infrastructure.
- Parameters:
config_path (str | Path) – Training YAML path to load.
- Returns:
Parsed training configuration dictionary.
- Return type:
dict
- scripts.training.train_feedforward_network.resolve_runtime_config(training_config)[source]
Resolve runtime execution options for the current training configuration.
- Parameters:
training_config (dict) – Parsed training configuration dictionary.
- Returns:
Runtime configuration dictionary after repository-specific checks and warning-driven adjustments.
- Return type:
dict[str, object]
- scripts.training.train_feedforward_network.train_feedforward_network(config_path=DEFAULT_CONFIG_PATH)[source]
Run the full feedforward TE training workflow for one configuration.
This entry point loads the configuration, prepares artifact directories, initializes the datamodule and model stack, runs training, validation, and held-out testing, then serializes the common artifact contract used by the repository registries and campaign tooling.
- Parameters:
config_path (str | Path) – Path to the YAML training configuration to execute.
- Return type:
None