An attention layer that allows for downscaling the size of the embedding after projection to queries, keys, and values.

Computes and stores the average and current value

data_dict['output_key'] will be a list of segmentations scaled according to ds_scales

HuggingFace tokenizer wrapper

A structure for storing masks and their related data in batched format. Implements basic filtering and concatenation.

input must be logits, not probabilities!

Transformer encoder consisting of `config.num_hidden_layers` self attention layers. Each layer is a [`BiomedCLIPEncoderLayer`]. Args:

Target function

HuggingFace model adapter

Image feature wrapper

Resize and Keep Ratio Copy & paste from `timm`

this is just a compatibility layer because my target tensor is float and has an extra dimension input must be logits, not probabilities!

Text feature wrapper

Multi-head Attention block with relative position embeddings.

Transformer blocks with support of window attention and residual propagation blocks

Crops the given image at the center. If the image is torch Tensor, it is expected to have [..., H, W] shape, where ... means an arbitrary numb

Useful to calculate the empirical mean and variance of intermediate feature maps.

Hard Negative Noise Contrastive Estimation proposed in https://arxiv.org/abs/2301.02280 beta1: hardness parameter for image features beta

A ResNet class that is similar to torchvision's but contains the following changes: - There are now 3 "stem" convolutions as opposed to 1, wi

ONLY SUPPORTS 2D IMAGES!!!

Nibabel loads the images in a different order than sitk. We convert the axes to the sitk order to be consistent. This is of course considered

Reorients images to RAS Nibabel loads the images in a different order than sitk. We convert the axes to the sitk order to be consistent.

https://arxiv.org/abs/2212.00794

Image to Patch Embedding.

Positional encoding using random spatial frequencies.

An iterable dataset yielding a list of urls.

Resizes images to the longest side 'target_length', as well as provides methods for resizing coordinates and boxes. Provides methods for

This model should not be called directly, but is used in ONNX export. It combines the prompt encoder, mask decoder, and mask postprocessing