Core Configuration¶

spectrans.config.core ¶

Base configuration models for spectrans components.

This module provides the foundational Pydantic models that define common configuration patterns used across spectrans components. These base classes are extended by specific component configuration models in submodules.

Classes:

Name	Description
`BaseLayerConfig`	Base configuration for all neural network layers.
`UnitaryLayerConfig`	Configuration for layers that preserve energy/unitarity.
`FilterLayerConfig`	Configuration for layers using learnable spectral filters.
`AttentionLayerConfig`	Configuration for attention-based layers.

Notes

All configuration models use Pydantic v2 BaseModel for validation and type safety. The base classes here provide common parameter patterns that are inherited by specific component configurations.

Examples:

>>> from spectrans.config.core import BaseLayerConfig
>>> class MyLayerConfig(BaseLayerConfig):
...     special_param: int = 42
>>> config = MyLayerConfig(hidden_dim=768)
>>> print(config.hidden_dim)
768

Classes¶

BaseLayerConfig ¶

Bases: BaseModel

Base configuration for all neural network layers.

Attributes:

Name	Type	Description
`hidden_dim`	`int`	Hidden dimension size, must be positive.
`dropout`	`float`	Dropout probability, must be between 0.0 and 1.0, defaults to 0.0.

UnitaryLayerConfig ¶

Bases: BaseLayerConfig

Configuration for layers that preserve energy/unitarity.

Attributes:

Name	Type	Description
`norm_eps`	`float`	Epsilon for numerical stability in normalization, defaults to 1e-5.
`energy_tolerance`	`float`	Tolerance for energy preservation checks, defaults to 1e-4.
`fft_norm`	`FFTNorm`	FFT normalization mode, defaults to "ortho".

FilterLayerConfig ¶

Bases: BaseLayerConfig

Configuration for layers using learnable spectral filters.

Attributes:

Name	Type	Description
`sequence_length`	`int`	Input sequence length, must be positive.
`learnable_filters`	`bool`	Whether filters are learnable, defaults to True.
`fft_norm`	`FFTNorm`	FFT normalization mode, defaults to "ortho".
`filter_init_std`	`float`	Standard deviation for filter initialization, defaults to 0.02.

AttentionLayerConfig ¶

Bases: BaseLayerConfig

Configuration for attention-based layers.

Attributes:

Name	Type	Description
`num_heads`	`int`	Number of attention heads, must be positive, defaults to 8.
`head_dim`	`int \| None`	Dimension per head, defaults to None (auto-computed).