lumen-core 0.6.1

A tiny ML framework
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

use std::sync::Arc;
use crate::{AutogradMetaT, NumDType, Result, Shape, WithDType};
use crate::{Tensor, TensorId, TensorImpl};

impl<T: WithDType> Tensor<T> {
    /// Broadcast the input tensor to the target shape. This returns an error if the input shape is
    /// not compatible with the target shape.
    ///
    /// If the input shape is `i_1, i_2, ... i_k`, the target shape has to have `k` dimensions or
    /// more and shape `j_1, ..., j_l, t_1, t_2, ..., t_k`. The dimensions `j_1` to `j_l` can have
    /// any value, the dimension `t_a` must be equal to `i_a` if `i_a` is different from 1. If
    /// `i_a` is equal to 1, any value can be used.
    pub fn broadcast_as<S: Into<Shape>>(&self, shape: S) -> Result<Self> {
        let ndarry_ = TensorImpl {
            id: TensorId::new(),
            storage: self.0.storage.clone(),
            layout: self.layout().broadcast_as(shape)?,
            meta: T::AutogradMeta::on_broadcast_op(self)
        };
        Ok(Tensor(Arc::new(ndarry_)))
    }
}

macro_rules! broadcast_binary_op {
    ($fn_name:ident, $inner_fn_name:ident) => {
        pub fn $fn_name(&self, rhs: &Self) -> Result<Self> {
            if self.shape() == rhs.shape() {
                return self.$inner_fn_name(rhs);
            }
            let lhs = self;
            let shape = lhs
                .shape()
                .broadcast_shape_binary_op(rhs.shape(), stringify!($fn_name))?;
            let l_broadcast = shape != *lhs.shape();
            let r_broadcast = shape != *rhs.shape();
            match (l_broadcast, r_broadcast) {
                (true, true) => lhs
                    .broadcast_as(&shape)?
                    .$inner_fn_name(&rhs.broadcast_as(&shape)?),
                (false, true) => lhs.$inner_fn_name(&rhs.broadcast_as(&shape)?),
                (true, false) => lhs.broadcast_as(&shape)?.$inner_fn_name(rhs),
                (false, false) => lhs.$inner_fn_name(rhs),
            }
        }
    };
}

macro_rules! broadcast_cmp_op {
    ($fn_name:ident, $inner_fn_name:ident) => {
        pub fn $fn_name(&self, rhs: &Self) -> Result<Tensor<bool>> {
            if self.shape() == rhs.shape() {
                return self.$inner_fn_name(rhs);
            }
            let lhs = self;
            let shape = lhs
                .shape()
                .broadcast_shape_binary_op(rhs.shape(), stringify!($fn_name))?;
            let l_broadcast = shape != *lhs.shape();
            let r_broadcast = shape != *rhs.shape();
            match (l_broadcast, r_broadcast) {
                (true, true) => lhs
                    .broadcast_as(&shape)?
                    .$inner_fn_name(&rhs.broadcast_as(&shape)?),
                (false, true) => lhs.$inner_fn_name(&rhs.broadcast_as(&shape)?),
                (true, false) => lhs.broadcast_as(&shape)?.$inner_fn_name(rhs),
                (false, false) => lhs.$inner_fn_name(rhs),
            }
        }
    };
}

impl<T: NumDType> Tensor<T> {
    broadcast_binary_op!(broadcast_add, add);
    broadcast_binary_op!(broadcast_mul, mul);
    broadcast_binary_op!(broadcast_sub, sub);
    broadcast_binary_op!(broadcast_div, div);
    broadcast_binary_op!(broadcast_maximum, maximum);
    broadcast_binary_op!(broadcast_minimum, minimum);
    broadcast_cmp_op!(broadcast_eq, eq);
    broadcast_cmp_op!(broadcast_ne, ne);
    broadcast_cmp_op!(broadcast_lt, lt);
    broadcast_cmp_op!(broadcast_le, le);
    broadcast_cmp_op!(broadcast_gt, gt);
    broadcast_cmp_op!(broadcast_ge, ge);
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_broadcast_add_scalar() {
        let a = Tensor::new(&[1., 2., 3.]).unwrap();
        let b = Tensor::new(&[10.]).unwrap(); // 标量 (1,)
        let res = a.broadcast_add(&b).unwrap();

        let expected = Tensor::new(&[11., 12., 13.]).unwrap();
        assert!(res.allclose(&expected, 1e-6, 1e-6).unwrap());
    }

    #[test]
    fn test_broadcast_add_vector_to_matrix() {
        let a = Tensor::new(&[[1., 2., 3.], [4., 5., 6.]]).unwrap(); // (2,3)
        let b = Tensor::new(&[10., 20., 30.]).unwrap(); // (3,)
        let res = a.broadcast_add(&b).unwrap();

        let expected = Tensor::new(&[[11., 22., 33.], [14., 25., 36.]]).unwrap();
        assert!(res.allclose(&expected, 1e-6, 1e-6).unwrap());
    }

    #[test]
    fn test_broadcast_mul() {
        let a = Tensor::new(&[[1., 2.], [3., 4.]]).unwrap(); // (2,2)
        let b = Tensor::new(&[10.]).unwrap(); // (1,)
        let res = a.broadcast_mul(&b).unwrap();

        let expected = Tensor::new(&[[10., 20.], [30., 40.]]).unwrap();
        assert!(res.allclose(&expected, 1e-6, 1e-6).unwrap());
    }

    #[test]
    fn test_broadcast_maximum_minimum() {
        let a = Tensor::new(&[1., 5., 3.]).unwrap();
        let b = Tensor::new(&[2., 2., 2.]).unwrap();

        let max_res = a.broadcast_maximum(&b).unwrap();
        let min_res = a.broadcast_minimum(&b).unwrap();

        let expected_max = Tensor::new(&[2., 5., 3.]).unwrap();
        let expected_min = Tensor::new(&[1., 2., 2.]).unwrap();

        assert!(max_res.allclose(&expected_max, 1e-6, 1e-6).unwrap());
        assert!(min_res.allclose(&expected_min, 1e-6, 1e-6).unwrap());
    }

    #[test]
    fn test_broadcast_comparisons() {
        let a = Tensor::new(&[1., 2., 3.]).unwrap();
        let b = Tensor::new(&[2.]).unwrap();

        let eq = a.broadcast_eq(&b).unwrap();
        let lt = a.broadcast_lt(&b).unwrap();
        let le = a.broadcast_le(&b).unwrap();
        let gt = a.broadcast_gt(&b).unwrap();
        let ge = a.broadcast_ge(&b).unwrap();
        let ne = a.broadcast_ne(&b).unwrap();

        assert_eq!(eq.to_vec().unwrap(), [false, true, false]);
        assert_eq!(lt.to_vec().unwrap(), [true, false, false]);
        assert_eq!(le.to_vec().unwrap(), [true, true, false]);
        assert_eq!(gt.to_vec().unwrap(), [false, false, true]);
        assert_eq!(ge.to_vec().unwrap(), [false, true, true]);
        assert_eq!(ne.to_vec().unwrap(), [true, false, true]);
    }

    #[test]
    fn test_broadcast_div() {
        let a = Tensor::new(&[10., 20., 30.]).unwrap();
        let b = Tensor::new(&[10.]).unwrap();
        let res = a.broadcast_div(&b).unwrap();

        let expected = Tensor::new(&[1., 2., 3.]).unwrap();
        assert!(res.allclose(&expected, 1e-6, 1e-6).unwrap());
    }
}