Module patch_embed 

Patch embedding for Vision Transformers.

Converts a CHW image to a sequence of patch tokens by applying a strided Conv2D with kernel_size == stride == patch_size. Also provides 2-D sinusoidal and learnable positional encodings.

Re-exports

pub use conv2d_patch::PatchEmbed;

pub use conv2d_patch::PatchEmbedConfig;

pub use conv2d_patch::PatchEmbedWeights;

pub use conv2d_patch::prepend_cls;

pub use pos_embed::LearnablePosEmbed;

pub use pos_embed::add_pos_embed;

pub use pos_embed::pos_2d_sincos;

Modules

conv2d_patch

Patch embedder: strided Conv2D producing [N_patches, embed_dim].

pos_embed

2-D positional encodings for Vision Transformers.