Configuration API Reference

This module defines the Pydantic schemas used for validating the Hydra configurations. These schemas ensure that all experiment parameters are typed and consistent across different runs.

config_schema

Pydantic configuration schemas for validation.

Classes

ExperimentConfig

Bases: BaseModel

Top-level experiment configuration.

DataConfig

Bases: BaseModel

Configuration for data loading.

PipelineConfig

Bases: BaseModel

Configuration for preprocessing pipelines.

ModelConfig

Bases: BaseModel

Configuration for models.

TrainingConfig

Bases: BaseModel

Configuration for training.

LossConfig

Bases: BaseModel

Configuration for loss functions.

Supports all registered loss types: mse, physics_informed, huber, mape, mae. Pass loss-specific parameters directly in this config.

TrackingConfig

Bases: BaseModel

Configuration for experiment tracking.