Struct EmbeddingDistanceEvaluator 

pub struct EmbeddingDistanceEvaluator { /* private fields */ }

Evaluator that computes cosine similarity between embeddings of prediction and reference.

Implementations

impl EmbeddingDistanceEvaluator

pub fn new(embeddings: Arc<dyn Embeddings>, threshold: f64) -> Self

Create a new embedding distance evaluator.

• embeddings: The embeddings model to use.
• threshold: Minimum cosine similarity to pass (default suggestion: 0.8).

Trait Implementations

impl Evaluator for EmbeddingDistanceEvaluator

fn evaluate<'life0, 'life1, 'life2, 'life3, 'async_trait>( &'life0 self, prediction: &'life1 str, reference: &'life2 str, _input: &'life3 str, ) -> Pin<Box<dyn Future<Output = Result<EvalResult, SynapticError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait, 'life3: 'async_trait,

Evaluate a prediction against a reference, given the original input.