mipcandy.inference

Module Contents

Classes

Predictor

Functions

parse_predictant

API

mipcandy.inference.parse_predictant(x: mipcandy.types.SupportedPredictant, loader: type[mipcandy.data.Loader], *, as_label: bool = False) → tuple[list[torch.Tensor], list[str] | None]

class mipcandy.inference.Predictor(experiment_folder: str | os.PathLike[str], example_shape: mipcandy.types.AmbiguousShape, *, checkpoint: str = 'checkpoint_best.pth', device: mipcandy.types.Device = 'cpu')

Bases: mipcandy.layer.WithPaddingModule, mipcandy.layer.WithNetwork

Initialization

lazy_load_model() → None

predict_image(image: torch.Tensor, *, batch: bool = False) → torch.Tensor

_predict(x: mipcandy.types.SupportedPredictant | mipcandy.data.UnsupervisedDataset) → tuple[list[torch.Tensor], list[str] | None]

predict(x: mipcandy.types.SupportedPredictant | mipcandy.data.UnsupervisedDataset) → list[torch.Tensor]

static save_prediction(output: torch.Tensor, path: str | os.PathLike[str]) → None

save_predictions(outputs: Sequence[torch.Tensor], folder: str | os.PathLike[str], *, filenames: Sequence[str | os.PathLike[str]] | None = None) → None

predict_to_files(x: mipcandy.types.SupportedPredictant | mipcandy.data.UnsupervisedDataset, folder: str | os.PathLike[str]) → list[str] | None

__call__(x: mipcandy.types.SupportedPredictant | mipcandy.data.UnsupervisedDataset) → list[torch.Tensor]