autograd 1.1.1

Tensors and differentiable operations in Rust
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
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176

extern crate autograd as ag;
extern crate ndarray;
use self::ag::NdArray;
use ag::{tensor::Constant, tensor::Variable, with};
use ndarray::array;

#[test]
fn reduce_prod() {
    with(|g| {
        let rng = ag::ndarray_ext::ArrayRng::<f64>::default();
        let v = g.constant(rng.standard_normal(&[3, 2]));
        let z = g.reduce_prod(v, &[0, 1], false); // keep_dims=false
        let empty_shape: &[usize] = &[];
        assert_eq!(z.eval(&[]).unwrap().shape(), empty_shape);
    });
}

#[test]
fn argmax() {
    with(|g| {
        let x = g.constant(array![[3., 4.], [5., 6.]]);
        let y = g.argmax(x, -1, false);
        assert_eq!(y.eval(&[]), Ok(ndarray::arr1(&[1., 1.]).into_dyn()));
    });
}

#[test]
fn argmax2() {
    with(|g| {
        let test = g.constant(array![84.0, 16.0, 0.04, 85.0, 16.0, 85.0]);
        let max_dis_index = g.argmax(test, 0, false);
        assert_eq!(max_dis_index.eval(&[]), Ok(ndarray::arr0(3.).into_dyn()));
    });
}

#[test]
fn argmax_with_multi_max_args() {
    with(|g| {
        let x = g.constant(array![1., 2., 3., 3.]);
        let y = g.argmax(x, 0, false);
        assert_eq!(2., y.eval(&[]).unwrap()[ndarray::IxDyn(&[])]);
    });
}

#[test]
fn reduce_mean() {
    with(|g| {
        let v = g.variable(array![2., 3., 4.]);
        let z = g.reduce_mean(v, &[0], false); // keep_dims=false
        assert_eq!(3., z.eval(&[]).unwrap()[ndarray::IxDyn(&[])]);
    });
}

#[test]
fn reduce_grad() {
    with(|g| {
        let v = g.variable(array![2., 3., 4.]);
        let z = g.reduce_mean(v, &[0], false); // keep_dims=false
        let g = g.grad(&[z], &[v])[0];
        assert_eq!(g.eval(&[]).unwrap().shape(), &[3]);
    });
}

#[test]
fn transpose_matmul_square() {
    with(|g| {
        let x = g.constant(array![[0., 1.], [2., 3.]]);
        let w = g.constant(array![[0., 1.], [2., 3.]]);
        let w2 = g.transpose(w, &[1, 0]);
        let mm = g.matmul(x, w2);
        assert_eq!(
            mm.eval(&[]).unwrap().as_slice().unwrap(),
            &[1., 3., 3., 13.]
        );
    });
}

#[test]
fn transpose_matmul() {
    with(|g| {
        let x = g.constant(array![[0., 1., 2.], [3., 4., 5.]]);
        let w = g.constant(array![[0., 1.], [2., 3.]]);
        let x2 = g.transpose(x, &[1, 0]);
        let mm = g.matmul(x2, w);
        assert_eq!(
            mm.eval(&[]).unwrap().as_slice().unwrap(),
            &[6., 9., 8., 13., 10., 17.]
        );
    });
}

#[test]
fn test_mm() {
    with(|g: &mut ag::Graph<f32>| {
        let a = g.ones(&[2, 5]);
        let b = g.ones(&[5, 1]);
        let c = g.matmul(&a, &b);
        let d = c.eval(&[]).unwrap();
        assert_eq!(d.as_slice().unwrap(), &[5., 5.]);
    });
}

#[test]
fn test_batch_matmul_normal() {
    // blas is used
    with(|g: &mut ag::Graph<f32>| {
        let a: ag::Tensor<f32> = g.ones(&[2, 3, 4, 2]);
        let b: ag::Tensor<f32> = g.ones(&[2, 3, 2, 3]);
        let c = g.batch_matmul(a, b);
        let shape = &[2, 3, 4, 3];
        let size = shape.iter().product();
        let ans = NdArray::<_>::from_shape_vec(ndarray::IxDyn(shape), vec![2f32; size]).unwrap();
        let ret = c.eval(&[]).unwrap();
        ret.all_close(&ans, 1e-4);
    });
}

#[test]
fn test_batch_matmul_trans_not_square() {
    // blas is not used
    with(|g: &mut ag::Graph<f32>| {
        let a: ag::Tensor<f32> = g.constant(array![[[1., 2.], [3., 4.]], [[1., 2.], [3., 4.]]]);
        let b: ag::Tensor<f32> = g.constant(array![[[1., 2.], [3., 4.]], [[1., 2.], [3., 4.]]]);
        let c = g.batch_matmul(a, b);
        let ans = array![[[7., 10.], [15., 22.]], [[7., 10.], [15., 22.]]].into_dyn();
        let ret = c.eval(&[]).unwrap();
        assert!(ret.all_close(&ans, 1e-4));
        assert_eq!(ret.shape(), &[2, 2, 2]);
    });
}

#[test]
fn test_batch_matmul_trans_square_both() {
    // blas is not used
    with(|g: &mut ag::Graph<f32>| {
        let a_: ag::Tensor<f32> = g.constant(array![[[1., 2.], [3., 4.]], [[1., 2.], [3., 4.]]]);
        let b_: ag::Tensor<f32> = g.constant(array![[[1., 2.], [3., 4.]], [[1., 2.], [3., 4.]]]);
        let a: ag::Tensor<f32> = g.transpose(a_, &[0, 2, 1]);
        let b: ag::Tensor<f32> = g.transpose(b_, &[0, 2, 1]);
        let c = g.batch_matmul(a, b);
        let ans = array![[[7., 15.], [10., 22.]], [[7., 15.], [10., 22.]]].into_dyn();
        let ret = c.eval(&[]).unwrap();
        assert!(ret.all_close(&ans, 1e-4));
        assert_eq!(ret.shape(), &[2, 2, 2]);
    });
}

#[test]
fn test_batch_matmul_trans_square_lhs() {
    // blas is used
    with(|g: &mut ag::Graph<f32>| {
        let a_: ag::Tensor<f32> = g.constant(array![[[1., 2.], [3., 4.]], [[1., 2.], [3., 4.]]]);
        let a: ag::Tensor<f32> = g.transpose(a_, &[0, 2, 1]);
        let b: ag::Tensor<f32> = g.constant(array![[[1., 2.], [3., 4.]], [[1., 2.], [3., 4.]]]);
        let c = g.batch_matmul(a, b);
        let ans = array![[[10., 14.], [14., 20.]], [[10., 14.], [14., 20.]]].into_dyn();
        let ret = c.eval(&[]).unwrap();
        assert!(ret.all_close(&ans, 1e-4));
        assert_eq!(ret.shape(), &[2, 2, 2]);
    });
}

#[test]
fn test_batch_matmul_with_copy() {
    // blas is used
    with(|g: &mut ag::Graph<f32>| {
        let a_: ag::Tensor<f32> = g.constant(array![[[1., 2.], [3., 4.]], [[1., 2.], [3., 4.]]]);
        let a: ag::Tensor<f32> = g.transpose(a_, &[0, 2, 1]);
        let b: ag::Tensor<f32> = g.constant(array![[[1., 2.], [3., 4.]], [[1., 2.], [3., 4.]]]);
        let c = g.batch_matmul(a, b);
        let ans = array![[[10., 14.], [14., 20.]], [[10., 14.], [14., 20.]]].into_dyn();
        let ret = c.eval(&[]).unwrap();
        assert!(ret.all_close(&ans, 1e-4));
        assert_eq!(ret.shape(), &[2, 2, 2]);
    });
}