diff options
Diffstat (limited to 'src/embedders/vecmetric.rs')
-rw-r--r-- | src/embedders/vecmetric.rs | 22 |
1 files changed, 12 insertions, 10 deletions
diff --git a/src/embedders/vecmetric.rs b/src/embedders/vecmetric.rs index 1bda3a8..9f2f143 100644 --- a/src/embedders/vecmetric.rs +++ b/src/embedders/vecmetric.rs @@ -4,29 +4,31 @@ use serde::{Deserialize, Serialize}; pub trait VecMetric: MetricElem + From<Vec<f32>> {} #[derive(Deserialize, Serialize)] -pub struct CosineSimilarity(pub Vec<f32>); +pub struct AngularDistance(pub Vec<f32>); #[derive(Deserialize, Serialize)] pub struct EuclidianDistance(pub Vec<f32>); #[derive(Deserialize, Serialize)] pub struct ManhattenDistance(pub Vec<f32>); -impl VecMetric for CosineSimilarity {} +impl VecMetric for AngularDistance {} impl VecMetric for EuclidianDistance {} impl VecMetric for ManhattenDistance {} -#[rustfmt::skip] impl From<Vec<f32>> for CosineSimilarity { fn from(value: Vec<f32>) -> Self { Self(value) } } +#[rustfmt::skip] impl From<Vec<f32>> for AngularDistance { fn from(value: Vec<f32>) -> Self { Self(value) } } #[rustfmt::skip] impl From<Vec<f32>> for EuclidianDistance { fn from(value: Vec<f32>) -> Self { Self(value) } } #[rustfmt::skip] impl From<Vec<f32>> for ManhattenDistance { fn from(value: Vec<f32>) -> Self { Self(value) } } -impl MetricElem for CosineSimilarity { +impl MetricElem for AngularDistance { fn dist(&self, other: &Self) -> f64 { - let len_a = self.0.iter().map(|x| x.powi(2)).sum::<f32>().sqrt(); - let len_b = other.0.iter().map(|x| x.powi(2)).sum::<f32>().sqrt(); - self.0 + let x = self + .0 .iter() .zip(other.0.iter()) - .map(|(a, b)| (*a / len_a - *b / len_b).powi(2)) - .sum::<f32>() - .sqrt() as f64 + .map(|(a, b)| *a * *b) + .sum::<f32>(); + let mag_a = self.0.iter().map(|x| x.powi(2)).sum::<f32>(); + let mag_b = other.0.iter().map(|x| x.powi(2)).sum::<f32>(); + let cossim = x / (mag_a * mag_b).sqrt(); + cossim.acos() as f64 } } impl MetricElem for EuclidianDistance { |