clinicadl.networks.config.ViTConfig¶

clinicadl.networks.config.ViTConfig[source]¶

Config class for clinicadl.networks.nn.ViT.

parameter in_shape: Sequence[PositiveInt] [Required]¶

parameter patch_size: Union[Sequence[PositiveInt], PositiveInt] [Required]¶

parameter num_outputs: Optional[PositiveInt] [Required]¶

parameter embedding_dim: PositiveInt = 768¶

parameter num_layers: PositiveInt = 12¶

parameter num_heads: PositiveInt = 12¶

parameter mlp_dim: PositiveInt = 3072¶

parameter pos_embed_type: Optional[PosEmbedType] = PosEmbedType.LEARN¶

parameter output_act: Union[ActFunction, tuple[ActFunction, dict[str, Any]], None] = ActFunction.TANH¶

parameter dropout: Optional[NonNegativeFloat] = None¶