alibi_detect.cd.pytorch.preprocess

HiddenOutput

Inherits from: Module

Constructor

HiddenOutput(self, model: Union[torch.nn.modules.module.Module, torch.nn.modules.container.Sequential], layer: int = -1, flatten: bool = False) -> None

Name	Type	Default	Description
model	Union[torch.nn.modules.module.Module, torch.nn.modules.container.Sequential]
layer	int	-1
flatten	bool	False

Methods

forward

forward(x: torch.Tensor) -> torch.Tensor

Name	Type	Default	Description
x	torch.Tensor

Returns

• Type: torch.Tensor

UAE

Inherits from: Module

Constructor

UAE(self, encoder_net: Optional[torch.nn.modules.module.Module] = None, input_layer: Optional[torch.nn.modules.module.Module] = None, shape: Optional[tuple] = None, enc_dim: Optional[int] = None) -> None

Name	Type	Default	Description
encoder_net	Optional[torch.nn.modules.module.Module]	None
input_layer	Optional[torch.nn.modules.module.Module]	None
shape	Optional[tuple]	None
enc_dim	Optional[int]	None

Methods

forward

forward(x: Union[numpy.ndarray, torch.Tensor, Dict[str, torch.Tensor]]) -> torch.Tensor

Name	Type	Default	Description
x	Union[numpy.ndarray, torch.Tensor, Dict[str, torch.Tensor]]

Returns

• Type: torch.Tensor

Functions

preprocess_drift

preprocess_drift(x: Union[numpy.ndarray, list], model: Union[torch.nn.modules.module.Module, torch.nn.modules.container.Sequential], device: Union[Literal[cuda, gpu, cpu], torch.device, None] = None, preprocess_batch_fn: Optional[Callable] = None, tokenizer: Optional[Callable] = None, max_len: Optional[int] = None, batch_size: int = 10000000000, dtype: Union[type[numpy.generic], torch.dtype] = <class 'numpy.float32'>) -> Union[numpy.ndarray, torch.Tensor, tuple]

Prediction function used for preprocessing step of drift detector.

Name	Type	Default	Description
x	Union[numpy.ndarray, list]		Batch of instances.
model	Union[torch.nn.modules.module.Module, torch.nn.modules.container.Sequential]		Model used for preprocessing.
device	Union[Literal[cuda, gpu, cpu], torch.device, None]	None	Device type used. The default tries to use the GPU and falls back on CPU if needed. Can be specified by passing either 'cuda', 'gpu', 'cpu' or an instance of torch.device.
preprocess_batch_fn	Optional[Callable]	None	Optional batch preprocessing function. For example to convert a list of objects to a batch which can be processed by the PyTorch model.
tokenizer	Optional[Callable]	None	Optional tokenizer for text drift.
max_len	Optional[int]	None	Optional max token length for text drift.
batch_size	int	10000000000	Batch size used during prediction.
dtype	Union[type[numpy.generic], torch.dtype]	<class 'numpy.float32'>	Model output type, e.g. np.float32 or torch.float32.

Returns

• Type: Union[numpy.ndarray, torch.Tensor, tuple]