Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).

Construct the CLS token, mask token, position and patch embeddings.

Fuse X in to Z with cross attention

This class turns `pixel_values` of shape `(batch_size, num_channels, height, width)` into the initial `hidden_states` (patch embeddings) of s

This corresponds to the Block class in the original implementation.

This class turns `pixel_values` of shape `(batch_size, num_channels, height, width)` into the initial `hidden_states` (patch embeddings) of s

The residual connection is defined in Dinov2Layer instead of here (as is the case with other models), due to the layernorm applied before eac

r""" A variant of the gated linear unit activation function from https://arxiv.org/abs/2002.05202. Parameters: dim_in (`int`): The nu

BVH structure to accelerate point-triangle intersection

minimalist AABB struct with grow functionality

32-byte BVH node struct

Construct the CLS token, mask token, position and patch embeddings.

An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained models.

Define a struct for queue entries

minimalist triangle struct