Enum ModalityInput 

pub enum ModalityInput<'a> {
    Text(Cow<'a, str>),
    Image {
        data: Cow<'a, [u8]>,
        format: ImageFormat,
    },
    Hybrid {
        text: Cow<'a, str>,
        visual_positions: Vec<VisualPosition>,
    },
}

Input modality for the encoder.

Supports text, images, and hybrid (OCR + visual) inputs. This enables ColPali-style visual document understanding.

Variants

Text(Cow<'a, str>)

Plain text input

Image

Image bytes (PNG/JPEG)

Fields

data: Cow<'a, [u8]>

Raw image bytes

format: ImageFormat

Image format hint

Hybrid

Hybrid: text with visual location (e.g., OCR result)

Fields

text: Cow<'a, str>

Extracted text

visual_positions: Vec<VisualPosition>

Visual bounding boxes for each token/word

Trait Implementations

impl<'a> Clone for ModalityInput<'a>

fn clone(&self) -> ModalityInput<'a>

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<'a> Debug for ModalityInput<'a>

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

Formats the value using the given formatter.