Struct VitPatchConfig 

pub struct VitPatchConfig {
    pub image_size: usize,
    pub patch_size: usize,
    pub n_channels: usize,
    pub d_model: usize,
}

Configuration for VitPatchEmbed.

The image is [n_channels × image_size × image_size]. It is tiled into (image_size / patch_size)² non-overlapping patches, each flattened to patch_size² · n_channels values and projected to d_model.

Fields

image_size: usize

Square image spatial size (height = width).

patch_size: usize

Square patch spatial size (stride = patch_size).

n_channels: usize

Number of input channels (e.g. 3 for RGB).

d_model: usize

Output token embedding dimension.

Implementations

impl VitPatchConfig

pub fn validate(&self) -> VisionResult<()>

Validate the configuration.

Errors

• VisionError::InvalidPatchSize if patch_size == 0 or image_size is not an exact multiple of patch_size.
• VisionError::InvalidEmbedDim if d_model == 0.
• VisionError::InvalidImageSize if image_size == 0 or n_channels == 0.

pub fn grid_size(&self) -> usize

Patches along one spatial axis: image_size / patch_size.

pub fn patch_dim(&self) -> usize

Flattened patch length: patch_size² · n_channels.

Trait Implementations

impl Clone for VitPatchConfig

fn clone(&self) -> VitPatchConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for VitPatchConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Eq for VitPatchConfig

impl PartialEq for VitPatchConfig

fn eq(&self, other: &VitPatchConfig) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for VitPatchConfig