Struct BinaryBlocker 

pub struct BinaryBlocker {
    pub threshold: u32,
    /* private fields */
}

Blocker using binary embeddings for fast candidate filtering.

Usage Pattern

1. Pre-compute binary hashes for all entities in your KB
2. At query time, hash the query embedding
3. Find candidates within Hamming distance threshold
4. Run dense similarity only on candidates

Example

use anno::backends::inference::{BinaryBlocker, BinaryHash};

let mut blocker = BinaryBlocker::new(8); // 8-bit Hamming threshold

// Add entities to the index
let hash1 = BinaryHash::from_embedding(&vec![0.1; 768]);
let hash2 = BinaryHash::from_embedding(&vec![-0.1; 768]);
blocker.add(0, hash1);
blocker.add(1, hash2);

// Query
let query = BinaryHash::from_embedding(&vec![0.1; 768]);
let candidates = blocker.query(&query);
assert!(candidates.contains(&0)); // Similar to hash1

Fields

threshold: u32

Hamming distance threshold for candidates

Implementations

impl BinaryBlocker

pub fn new(threshold: u32) -> Self

Create a new blocker with the given threshold.

pub fn add(&mut self, id: usize, hash: BinaryHash)

Add an entity to the index.

pub fn add_batch( &mut self, entries: impl IntoIterator<Item = (usize, BinaryHash)>, )

Add multiple entities.

pub fn query(&self, query: &BinaryHash) -> Vec<usize>

Find candidate IDs within Hamming distance threshold.

pub fn query_with_distance(&self, query: &BinaryHash) -> Vec<(usize, u32)>

Find candidates with their distances.

pub fn len(&self) -> usize

Number of entries in the index.

pub fn is_empty(&self) -> bool

Check if index is empty.

pub fn clear(&mut self)

Clear the index.

Trait Implementations

impl Clone for BinaryBlocker

fn clone(&self) -> BinaryBlocker

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for BinaryBlocker

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

Formats the value using the given formatter.