neuronika 0.2.0

Tensors and dynamic neural networks.
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
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122

use super::{Data, Forward, Gradient, Input, InputBackward, Overwrite, Tensor};
use std::cell::{Cell, RefCell};

mod forward {

    use super::{Cell, Data, Forward, Input, RefCell, Tensor};

    #[test]
    fn creation() {
        let input = Input {
            data: RefCell::new(Tensor::zeros((3, 3))),
            computed: Cell::new(false),
        };
        assert!(!input.was_computed());
        assert_eq!(*input.data(), Tensor::from_elem((3, 3), 0.));
        assert_eq!(*input.data_mut(), Tensor::from_elem((3, 3), 0.));
    }

    #[test]
    fn computation_was_computed_transition() {
        let input = Input {
            data: RefCell::new(Tensor::zeros((3, 3))),
            computed: Cell::new(false),
        };

        input.forward();
        assert!(input.was_computed());

        input.forward();
        assert!(input.was_computed());

        input.reset_computation();
        assert!(!input.was_computed());

        input.reset_computation();
        assert!(!input.was_computed());
    }

    #[test]
    fn debug() {
        let node = Input {
            data: RefCell::new(Tensor::zeros(1)),
            computed: Cell::new(false),
        };
        let output =
            "Input { data: [0.0], shape=[1], strides=[1], layout=CFcf (0xf), const ndim=1, computed: false }";

        assert_eq!(output, format!("{:?}", node));
    }

    #[test]
    fn display() {
        let node = Input {
            data: RefCell::new(Tensor::zeros(1)),
            computed: Cell::new(false),
        };

        assert_eq!(format!("{}", node.data()), format!("{}", node));
    }
}

mod backward {
    use super::{Cell, Gradient, InputBackward, Overwrite, RefCell, Tensor};

    #[test]
    fn creation() {
        let input = InputBackward {
            gradient: RefCell::new(Some(Tensor::zeros((3, 3)))),
            overwrite: Cell::new(true),
        };
        assert!(input.can_overwrite());
        assert_eq!(*input.gradient(), Tensor::from_elem((3, 3), 0.));
        assert_eq!(*input.gradient_mut(), Tensor::from_elem((3, 3), 0.));
    }

    #[test]
    fn computation_state_transition() {
        let input = InputBackward {
            gradient: RefCell::new(Some(Tensor::zeros((3, 3)))),
            overwrite: Cell::new(true),
        };

        assert!(input.can_overwrite());
        input.set_overwrite(false);
        assert!(!input.can_overwrite());
        input.set_overwrite(true);
        assert!(input.can_overwrite());
    }

    #[test]
    fn zero_grad() {
        let input = InputBackward {
            gradient: RefCell::new(Some(Tensor::ones((3, 3)))),
            overwrite: Cell::new(true),
        };

        input.zero_grad();
        assert_eq!(*input.gradient(), Tensor::zeros((3, 3)));
    }

    #[test]
    fn debug() {
        let node = InputBackward {
            gradient: RefCell::new(Some(Tensor::zeros(1))),
            overwrite: Cell::new(false),
        };
        let output =
            "InputBackward { gradient: Some([0.0], shape=[1], strides=[1], layout=CFcf (0xf), const ndim=1), overwrite: false }";

        assert_eq!(output, format!("{:?}", node));
    }

    #[test]
    fn display() {
        let node = InputBackward {
            gradient: RefCell::new(Some(Tensor::zeros(1))),
            overwrite: Cell::new(false),
        };

        assert_eq!(format!("{}", node.gradient()), format!("{}", node));
    }
}