lazyslide_models.vision.CHIEFSlideEncoder

Contents

lazyslide_models.vision.CHIEFSlideEncoder#

class CHIEFSlideEncoder(model_path=None, token=None)#

Bases: SlideEncoderModel

chief-slide-encoder GitHub Paper Params: 1.2M AGPL-3.0 [Wang et al., 2024] Clinical Histopathology Imaging Evaluation Foundation (CHIEF)

encode_slide(embeddings, coords=None, **kwargs)#: Encode the slide using the CHIEF slide encoder. The embeddings should be a tensor of shape [B, T, N]. T is the number of tiles, and N is the feature dimension.