Struct MtmdInputChunks 

pub struct MtmdInputChunks {
    pub chunks: NonNull<mtmd_input_chunks>,
}

Safe wrapper around mtmd_input_chunks.

This is a collection of input chunks created from tokenizing text and media. The chunks represent the tokenized input that can be processed by the model, with text chunks containing tokens and media chunks containing embeddings.

Fields

chunks: NonNull<mtmd_input_chunks>

Raw pointer to the underlying mtmd_input_chunks.

Implementations

impl MtmdInputChunks

pub fn new() -> Result<Self, MtmdInputChunksError>

Create a new empty input chunks collection.

Errors

Returns MtmdInputChunksError::ChunksCreationFailed if the underlying llama.cpp function returns null.

Examples

use llama_cpp_bindings::mtmd::MtmdInputChunks;

let chunks = MtmdInputChunks::new().unwrap();
assert_eq!(chunks.len(), 0);
assert!(chunks.is_empty());

pub fn len(&self) -> usize

Get the number of chunks

pub fn is_empty(&self) -> bool

Check if chunks collection is empty

pub fn get(&self, index: usize) -> Option<MtmdInputChunk>

Get a chunk by index

pub fn total_tokens(&self) -> usize

Get total number of tokens across all chunks.

pub fn total_positions(&self) -> i32

Get total position count across all chunks.

pub fn eval_chunks( &self, mtmd_ctx: &MtmdContext, llama_ctx: &LlamaContext<'_>, start_position: llama_pos, seq_id: llama_seq_id, n_batch: i32, logits_last: bool, ) -> Result<llama_pos, MtmdEvalError>

Evaluate chunks using the multimodal context and LLAMA context.

Errors

Returns MtmdEvalError::EvalFailure if any encoding or decoding operation fails.

Trait Implementations

impl Debug for MtmdInputChunks

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

Formats the value using the given formatter.

impl Drop for MtmdInputChunks

fn drop(&mut self)

Executes the destructor for this type.

fn pin_drop(self: Pin<&mut Self>)

🔬This is a nightly-only experimental API. (pin_ergonomics)

Execute the destructor for this type, but different to Drop::drop, it requires self to be pinned.