raitap.models

Models module — loads pretrained or custom models and selects a backend (native PyTorch / ONNX).

Public Python surface

ModelConfig

Dataclass section of AppConfig describing how to locate / construct the model under test.

Model

The loaded model object passed to assessors and explainers.

class raitap.models.Model(config, *, resolved_preprocessing=None, allow_unsafe_pickle=False)

Bases: Trackable

log(tracker, **kwargs)

Log the object's artifacts or metadata to the provided tracker.

class raitap.models.ModelConfig(source: 'str | None' = None, arch: 'str | None' = None, num_classes: 'int | None' = None, pretrained: 'bool' = False, class_names: 'list[str] | None' = None, task_kind: 'TaskKind | None' = None)

Bases: object