peal.adaptors.interfaces.AdaptorConfig¶

class peal.adaptors.interfaces.AdaptorConfig[source]¶

Bases: BaseModel

The config template for an adaptor.

Every adaptor config extends this model. The adaptor_type and category fields tell the yaml loader which pydantic class to instantiate; the remaining fields control reproducibility and how much intermediate output is produced.

Parameters:

adaptor_type (str) – Name of the adaptor implementation, e.g. "CFKD".
category (str) – Always "adaptor"; used to route the yaml to this config family.
seed (int or None) – Seed for all sources of randomness.
tracking_level (int) – Verbosity/caching level from 0 (silent) to 5 (everything).
calculate_explainer_stats (bool) – Whether to compute explainer statistics such as sparsity/diversity.
in_memory (bool) – Whether datasets are fully loaded into RAM.

Methods

Attributes

`calculate_explainer_stats`	Whether to calculate explainer stats like sparsity, diversity, etc.
`category`	The category of the config.
`in_memory`	Whether to load all datasets into the RAM or not.
`seed`	The seed of all randomness to make results reproducible.
`tracking_level`	How many intermediate results are cached an visualized.
`adaptor_type`	The type of adaptor that shall be used.

adaptor_type: str¶: The type of adaptor that shall be used. This is necessary to know which pydantic class to use when loading from yaml.

category: str = 'adaptor'¶: The category of the config. Can not be changed for adaptor. This is also necessary to identify which pydantic class to use when loading from yaml.

seed: int | None = 0¶: The seed of all randomness to make results reproducible.

tracking_level: int = 0¶: How many intermediate results are cached an visualized. 0 -> None >=1 -> only progress bars >=2 -> prints >=3 -> caching >=4 -> visualizations >=5 -> everything, including expensive visualizations and tracking of values not mentioned by papers

calculate_explainer_stats: bool = False¶: Whether to calculate explainer stats like sparsity, diversity, etc.

in_memory: bool = False¶: Whether to load all datasets into the RAM or not. Careful with big datasets!