Struct Qwen3EmbeddingProvider 

pub struct Qwen3EmbeddingProvider { /* private fields */ }

Qwen3-Embedding-0.6B provider — decoder-ONNX, last-token pooling, MRL-512.

Implementations

impl Qwen3EmbeddingProvider

pub fn new( model_path: impl Into<PathBuf>, tokenizer_path: impl Into<PathBuf>, ) -> Self

Create a provider pointing at model_fp16.onnx + tokenizer.json.

Degrades to Mock when the feature is off, files are missing, or loading fails — never panics, never requires the model to construct.

pub fn is_runtime_available(&self) -> bool

Whether real decoder-ONNX inference is active (vs. Mock fallback).

pub fn model_path(&self) -> &Path

Path to the configured ONNX model file.

pub fn tokenizer_path(&self) -> &Path

Path to the configured tokenizer file.

pub async fn embed_query(&self, query: &str) -> Result<Vec<f32>, EmbeddingError>

Embed a query (asymmetric lane): prepends the Qwen3 instruction prefix.

Trait Implementations

impl EmbeddingProvider for Qwen3EmbeddingProvider

fn embed_query<'life0, 'life1, 'async_trait>( &'life0 self, query: &'life1 str, ) -> Pin<Box<dyn Future<Output = Result<Vec<f32>, EmbeddingError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Asymmetric query lane — delegates to the inherent Self::embed_query so trait-object call sites (SessionView::graft_text) get the instruction prefix.

fn embed<'life0, 'life1, 'async_trait>( &'life0 self, text: &'life1 str, ) -> Pin<Box<dyn Future<Output = Result<Vec<f32>, EmbeddingError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Embed a single text chunk into a vector.

fn embed_batch<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, texts: &'life1 [&'life2 str], ) -> Pin<Box<dyn Future<Output = Result<Vec<Vec<f32>>, EmbeddingError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Embed a batch of text chunks.

fn dimensions(&self) -> usize

Dimensionality of the output vectors.

fn model_name(&self) -> &str

Name of the embedding model (for metadata tracking).

fn warm<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = ()> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Pre-warm the backend: run one throwaway embed so the model graph, runtime thread pool, and allocator are hot before the first real call. Best-effort — errors are swallowed (a backend with nothing to warm, like the Mock, just pays one cheap call). Intended to be called once at daemon startup so the first conversation turn doesn’t pay the model’s first-inference cost (ADR-058 Phase 5.3; mirrors the 6.2 STT pre-warm).