Model

DataLoader.one_batch[source]

DataLoader.one_batch()

Pathed functionality that grabs one batch of data from the DataLoader and deletes the iter

Returns:

• <class 'torch.Tensor'>

  A batch of data

GatherPredsCallback.after_validate[source]

GatherPredsCallback.after_validate()

Patched functionality that does nothing

class CudaCallback[source]

CudaCallback(device:str=None) :: Callback

Move data to CUDA device

Parameters:

• device : <class 'str'>, optional

  A device to move the data to, such as 'cuda:0' or 'cpu'

class AdaptiveModel[source]

AdaptiveModel() :: ABC

Helper class that provides a standard way to create an ABC using inheritance.

AdaptiveModel.set_model[source]

AdaptiveModel.set_model(model)

Sets model in _learn

Parameters:

• model : <class 'inspect._empty'>

  A PyTorch model

AdaptiveModel.set_as_dict[source]

AdaptiveModel.set_as_dict(as_dict:bool=False)

Sets as_dict in _learn

Parameters:

• as_dict : <class 'bool'>, optional

  Whether to return the inputs as a dictionary when predicting or training

AdaptiveModel.set_device[source]

AdaptiveModel.set_device(device:str='cpu')

Sets the device for CudaCallback in __learn

Parameters:

• device : <class 'str'>, optional

  A device for the `CudaCallback`, such as 'cuda:0' or 'cpu'

AdaptiveModel.get_preds[source]

AdaptiveModel.get_preds(dl=None, cbs=[])

Get raw predictions based on dl with cbs.

For basic inference, cbs should include any Callbacks needed to do general inference

Parameters:

• dl : <class 'NoneType'>, optional

  An iterable DataLoader or DataLoader-like object

• cbs : <class 'list'>, optional

  Optional fastai `Callbacks`

AdaptiveModel.load[source]

AdaptiveModel.load(model_name_or_path:Union[str, Path])

Load model into the AdaptiveModel object as alternative constructor

Parameters:

• model_name_or_path : typing.Union[str, pathlib.Path]

  A model file location to use

AdaptiveModel.predict[source]

AdaptiveModel.predict(text:Union[List[Sentence], Sentence, List[str], str], mini_batch_size:int=32, **kwargs)

Run inference on the model

Parameters:

• text : typing.Union[typing.List[flair.data.Sentence], flair.data.Sentence, typing.List[str], str]

  Some text to predict on

• mini_batch_size : <class 'int'>, optional

  A batch size for if a list of texts are passed in

• kwargs : <class 'inspect._empty'>

Returns:

• typing.List[flair.data.Sentence]

  A list of predicted sentences