Struct VlmPrompt 

pub struct VlmPrompt<'a> {
    pub parts: Vec<VlmPart<'a>>,
}

An ordered, interleaved multimodal prompt — text <img> text <img> text.

Order is the payload. “Compare the first image to the second” is not expressible as a set of images plus a question, and a model that receives the images in the wrong order answers the wrong question fluently. So the prompt is a sequence and the engine splices its image-token blocks at the positions the sequence gives it.

Fields

parts: Vec<VlmPart<'a>>

Implementations

impl<'a> VlmPrompt<'a>

pub fn single(image: &'a ImageBuffer, text: Option<&'a str>) -> Self

The single-image case: the image, then the instruction if there is one.

pub fn image_count(&self) -> usize

Number of images in the prompt — what an engine checks against the image-token placeholders it is about to splice.

Trait Implementations

impl<'a> Clone for VlmPrompt<'a>

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

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<'a> Debug for VlmPrompt<'a>

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

Formats the value using the given formatter.

impl<'a> Default for VlmPrompt<'a>

fn default() -> VlmPrompt<'a>

Returns the “default value” for a type.

impl<'a> Eq for VlmPrompt<'a>

impl<'a> PartialEq for VlmPrompt<'a>

fn eq(&self, other: &VlmPrompt<'a>) -> bool

Equality operator ==.

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

Inequality operator !=.

impl<'a> StructuralPartialEq for VlmPrompt<'a>