vantadb 0.1.4

VantaDB: An embedded persistent memory and vector retrieval engine for local-first AI applications.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48

use std::path::Path;

#[path = "../common/mod.rs"]
mod common;

use common::sift_loader::{read_fvecs, read_ivecs};

#[test]
fn validate_sift_dataset_integrity() {
    let base_path = Path::new("datasets/sift/sift_base.fvecs");
    let query_path = Path::new("datasets/sift/sift_query.fvecs");
    let groundtruth_path = Path::new("datasets/sift/sift_groundtruth.ivecs");

    // Skip if not downloaded
    if !base_path.exists() || !query_path.exists() || !groundtruth_path.exists() {
        println!("SIFT dataset not found. Skipping integrity check.");
        return;
    }

    println!("Loading SIFT1M base vectors...");
    let base = read_fvecs(base_path).expect("Failed to read base.fvecs");

    println!("Loading SIFT1M query vectors...");
    let query = read_fvecs(query_path).expect("Failed to read query.fvecs");

    println!("Loading SIFT1M ground truth...");
    let groundtruth = read_ivecs(groundtruth_path).expect("Failed to read groundtruth.ivecs");

    // Phase 2.1 Validation Logic from Roadmap
    assert_eq!(base.len(), 1_000_000, "Base must have 1M vectors");
    assert_eq!(base[0].len(), 128, "Base vectors must be 128D");

    assert_eq!(query.len(), 10_000, "Query must have 10K vectors");
    assert_eq!(query[0].len(), 128, "Query vectors must be 128D");

    assert_eq!(
        groundtruth.len(),
        10_000,
        "Groundtruth must have 10K entries"
    );
    assert_eq!(
        groundtruth[0].len(),
        100,
        "Groundtruth usually provides top 100 nearest neighbors"
    );

    println!("SIFT1M Dataset Integrity: PASSED ✅");
}