Struct OnnxEntityEmbedder 

pub struct OnnxEntityEmbedder { /* private fields */ }

Runs entity features through an already-open model session to produce an Embedding.

The embedder never loads a model, chooses a backend, or moves data across a device boundary: the caller supplies an open session, names the input and output tensors, and selects the copy policy. This keeps backend identity and transfer semantics explicit and auditable.

Implementations

impl OnnxEntityEmbedder

pub fn try_new( input_name: impl Into<String>, output_name: impl Into<String>, input_descriptor: TensorDescriptor, output_descriptor: TensorDescriptor, copy_policy: CopyPolicy, ) -> SemanticResult<Self>

Creates an embedder for one model input/output pair.

pub fn embed_one( &self, session: &mut dyn ModelSession, features: &[f32], ) -> SemanticResult<Embedding>

Embeds one feature vector (flattened f32 values) into an embedding.

features must contain exactly the element count implied by input_descriptor. The session’s output_name tensor must match output_descriptor.

pub fn input_descriptor(&self) -> &TensorDescriptor

Returns the configured input descriptor.

pub fn output_descriptor(&self) -> &TensorDescriptor

Returns the configured output descriptor.

Trait Implementations

impl Clone for OnnxEntityEmbedder

fn clone(&self) -> OnnxEntityEmbedder

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for OnnxEntityEmbedder

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

Formats the value using the given formatter.