Args: in_channels (int): Number of input channels. out_channels (int): Number of output channels. bottlen

Args: dim (int): Number of input channels. num_heads (int): Number of attention heads in each ViT block.

Args: net (Backbone): module representing the subnetwork backbone. Must be a subclass of :class:`Backbone`.

The `__init__` method of any subclass can specify its own set of arguments.

Multi-Scale Deformable Attention Module :param d_model hidden dimension :param n_levels number of feature levels

Inputs: embed_dim - Dimensionality of input and attention feature vectors hidden_dim - Dimensionality of hidden layer

Args: input_shape: shapes (channels and stride) of the input features num_classes: number of classes to predict

Return an activation function given a string

Calculate decomposed Relative Positional Embeddings from :paper:`mvitv2`. https://github.com/facebookresearch/mvit/blob/19786631e330df9f3622e

Forward function of contrastive learning.

Input: - tgt: nq, bs, d_model - memory: hw, bs, d_model - pos: hw, bs, d_model - refpoin

Input: - tgt/tgt_query_pos: nq, bs, d_model -

Args: x: Tensor of shape (N,C,H,W). H, W must be a multiple of ``self.size_divisibility``. Returns: dict[str

Subclasses must override this method, but adhere to the same return type. Returns: dict[str->Tensor]: mapping from featu

:param query (N, Length_{query}, C) :param reference_points (N, Length_{query}, n_levels, 2), range

_summary_ Args: v (_type_): bs, n_img, dim l (_type_): bs, n_text, dim attention_mask_v (_type_, optional

Input: - src: [bs, sum(hi*wi), 256] - pos: pos embed for src. [bs, sum(hi*wi), 256] - spatial_shapes: h,w

Input: - tgt: nq, bs, d_model - memory: hw, bs, d_model - pos: hw, bs, d_model - refpoints_un

Input: - tgt/tgt_query_pos: nq, bs, d_model -

Args: text_attention_mask: bs, num_token memory_text: bs, num_token, d_model Raises: RuntimeErr

:param features: multi-scale features from the backbone :param masks: image mask :return: enhanced multi-scale features and m

query: (n_context, queryL, d) context: (n_context, sourceL, d)