Struct EmbeddingModelConfig 

pub struct EmbeddingModelConfig {
    pub weights_path: String,
    pub tokenizer_path: Option<String>,
    pub dimension: u32,
    pub max_seq_len: u32,
    pub backend: InferenceBackend,
    pub model_name: String,
    pub model_version: String,
}

Configuration for loading a real embedding model from disk.

This is the configuration type callers populate to construct a real EmbeddingModel implementation via a future BackendLoader. The MockEmbeddingModel ignores this; it is used only when testing the pipeline without model files.

Once a candle or onnx-runtime integration crate is added (M12 full implementation), it will consume this config and return a Box<dyn EmbeddingModel>.

Fields

weights_path: String

Path to the model weights file (ONNX .onnx or safetensors).

tokenizer_path: Option<String>

Tokenizer config path (tokenizer.json for HuggingFace tokenizers).

dimension: u32

Expected embedding dimension.

max_seq_len: u32

Maximum input token length (truncation limit).

backend: InferenceBackend

Compute backend selection.

model_name: String

Model name for registry (e.g. “nomic-embed-text-v1.5”).

model_version: String

Model version string.

Implementations

impl EmbeddingModelConfig

pub fn weights_exist(&self) -> bool

Check that the model weights file exists on disk.

Trait Implementations

impl Clone for EmbeddingModelConfig

fn clone(&self) -> EmbeddingModelConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for EmbeddingModelConfig

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

Formats the value using the given formatter.