xnn 0.2.0

A lightweight ML framework with GPU-first architecture
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
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99

//! Tests for `Tensor::copy` operation.

use std::f32::consts::PI;

use approx::assert_relative_eq;
use xnn::{Context, Tensor};

#[test]
fn test_copy_f32() {
    let ctx = Context::try_default().unwrap();
    let data = vec![1.0, 2.0, 3.0, 4.0];
    let t = Tensor::<f32>::from_slice(&ctx, &data).unwrap();
    let copy = t.copy().unwrap();
    assert_eq!(copy.dimensions(), t.dimensions());
    for (a, b) in copy.to_vec().unwrap().iter().zip(data.iter()) {
        assert_relative_eq!(a, b, epsilon = 1e-4);
    }
}

#[test]
fn test_copy_i32() {
    let ctx = Context::try_default().unwrap();
    let data = vec![1, 2, 3, 4];
    let t = Tensor::<i32>::from_slice(&ctx, &data).unwrap();
    let copy = t.copy().unwrap();
    assert_eq!(copy.dimensions(), t.dimensions());
    assert_eq!(copy.to_vec().unwrap(), data);
}

#[test]
fn test_copy_u32() {
    let ctx = Context::try_default().unwrap();
    let data = vec![1u32, 2, 3, 4];
    let t = Tensor::<u32>::from_slice(&ctx, &data).unwrap();
    let copy = t.copy().unwrap();
    assert_eq!(copy.dimensions(), t.dimensions());
    assert_eq!(copy.to_vec().unwrap(), data);
}

#[test]
fn test_copy_2d() {
    let ctx = Context::try_default().unwrap();
    let data = vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0];
    let t = Tensor::<f32>::from_shape_slice(&ctx, &[2, 3], &data).unwrap();
    let copy = t.copy().unwrap();
    assert_eq!(copy.dimensions(), &[2, 3]);
    for (a, b) in copy.to_vec().unwrap().iter().zip(data.iter()) {
        assert_relative_eq!(a, b, epsilon = 1e-4);
    }
}

#[test]
fn test_copy_non_aligned() {
    let ctx = Context::try_default().unwrap();
    let data: Vec<f32> = (0_u8..42).map(f32::from).collect();
    let t = Tensor::<f32>::from_slice(&ctx, &data).unwrap();
    let copy = t.copy().unwrap();
    assert_eq!(copy.dimensions(), &[42]);
    for (a, b) in copy.to_vec().unwrap().iter().zip(data.iter()) {
        assert_relative_eq!(a, b, epsilon = 1e-4);
    }
}

#[test]
fn test_copy_large() {
    let ctx = Context::try_default().unwrap();
    let len = 4096 * 4096;
    let t = Tensor::<f32>::constant(&ctx, &[len], &[PI]).unwrap();
    let copy = t.copy().unwrap();
    assert_eq!(copy.dimensions(), &[len]);
    for val in &copy.to_vec().unwrap() {
        assert_relative_eq!(*val, PI, epsilon = 1e-4);
    }
}

#[test]
fn test_copy_scalar() {
    let ctx = Context::try_default().unwrap();
    let t = Tensor::<f32>::constant(&ctx, &[], &[42.0]).unwrap();
    let copy = t.copy().unwrap();
    assert_eq!(copy.dimensions(), &[] as &[usize]);
    assert_relative_eq!(copy.to_vec().unwrap()[0], 42.0, epsilon = 1e-4);
}

#[test]
fn test_copy_independence() {
    let ctx = Context::try_default().unwrap();
    let data = vec![1.0, 2.0, 3.0, 4.0];
    let original = Tensor::<f32>::from_slice(&ctx, &data).unwrap();
    let copy = original.copy().unwrap();
    let out_orig = original.to_vec().unwrap();
    let out_copy = copy.to_vec().unwrap();
    for (a, b) in out_orig.iter().zip(data.iter()) {
        assert_relative_eq!(a, b, epsilon = 1e-4);
    }
    for (a, b) in out_copy.iter().zip(data.iter()) {
        assert_relative_eq!(a, b, epsilon = 1e-4);
    }
}