Struct MtmdImageTokens 

pub struct MtmdImageTokens<'chunks> { /* private fields */ }

Image/audio token metadata attached to a non-text MtmdInputChunk.

Implementations

impl MtmdImageTokens<'_>

pub fn n_tokens(&self) -> usize

Total number of embedding tokens.

pub fn nx(&self) -> usize

Width of the token grid.

pub fn ny(&self) -> usize

Height of the token grid.

pub fn n_pos(&self) -> i32

Number of temporal positions (M-RoPE variant; equals n_tokens otherwise).

pub fn id(&self) -> Option<&str>

Optional ID for KV cache tracking.

pub fn decoder_positions(&self, pos_0: i32) -> Vec<MtmdDecoderPos>

Compute the per-token decoder positions used by M-RoPE models.

Returns a vector of length n_tokens. Each entry is relative to pos_0; for non-M-RoPE models this typically reduces to (0, i, 0, 0) for the i-th token.

Wraps mtmd_helper_image_get_decoder_pos.

Trait Implementations

impl<'chunks> Debug for MtmdImageTokens<'chunks>

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

Formats the value using the given formatter.