returns what functions to use for resampling data and seg, respectively. Also returns kwargs resampling function must be callable(dat

function must be callable(data, new_shape, current_spacing, new_spacing, **kwargs). The new_shape should be used as target. current_s

Calculates the Dice coefficient, a measure of similarity, between two binary label volumes. Note: the dice output is per image and per checkp

Feature: different configurations can now have different splits

Returns the fingerprint as a dictionary (additionally to saving it)

Finds a free port on localhost. It is useful in single-node training when we don't want to connect to a real main node but have to set the `M

Generates masks for the given image. Arguments: image (np.ndarray): The image to generate masks for, in HWC uint8 form

Generates a list of crop boxes of different sizes. Each layer has (2**i)**2 boxes for the ith layer.

grid_size: int of the grid height and width return: pos_embed: [grid_size*grid_size, embed_dim] or [1+grid_size*grid_size, embed_dim] (w/

we may have to change this in the future to accommodate other plans -> network mappings num_input_channels can differ depending on whether w

If we find the channel_name in channel_name_to_normalization_mapping return the corresponding normalization. If it is not found, use the defa

Get parameters

Choose strategy for dropping (masking) tokens to achieve target context length

Return function over iterator that groups key, value pairs into samples. :param keys: function that splits the key into key and extension (base_p

Filter masks at the edge of a crop, but not at the edge of the original image.

enumerate available model architectures based on config files

Returns the names of available CLIP models

return all models having the specified pretrain tag

return all pretrain tags for the specified model architecture

Load weights from .npz checkpoints for official Google big_vision image-text models Currently the SigLIP source models are supported and a Custo

Transfers all weights between matching keys in state_dicts. matching is done by name and we only transfer if the shape is also the same. Segm

This is used by the nnUNetTrainer to initialize nnUNetPredictor for the final validation