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
177
178
179
180
181
182
183
184
185
186
187
188
189

//! A two dimension transposed convolution layer.
use super::Path;
use crate::Tensor;
use std::borrow::Borrow;

/// A generic transposed convolution configuration.
#[allow(clippy::upper_case_acronyms)]
#[derive(Debug, Clone, Copy)]
pub struct ConvTransposeConfigND<ND> {
    pub stride: ND,
    pub padding: ND,
    pub output_padding: ND,
    pub groups: i64,
    pub bias: bool,
    pub dilation: ND,
    pub ws_init: super::Init,
    pub bs_init: super::Init,
}

/// A transposed convolution configuration using the same values on each dimension.
pub type ConvTransposeConfig = ConvTransposeConfigND<i64>;

impl Default for ConvTransposeConfig {
    fn default() -> Self {
        ConvTransposeConfigND {
            stride: 1,
            padding: 0,
            output_padding: 0,
            dilation: 1,
            groups: 1,
            bias: true,
            ws_init: super::init::DEFAULT_KAIMING_UNIFORM,
            bs_init: super::Init::Const(0.),
        }
    }
}

/// A generic transposed convolution layer.
#[allow(clippy::upper_case_acronyms)]
#[derive(Debug)]
pub struct ConvTransposeND<ND> {
    pub ws: Tensor,
    pub bs: Option<Tensor>,
    config: ConvTransposeConfigND<ND>,
}

/// A one dimension transposed convolution layer.
pub type ConvTranspose1D = ConvTransposeND<[i64; 1]>;

/// A two dimension transposed convolution layer.
pub type ConvTranspose2D = ConvTransposeND<[i64; 2]>;

/// A three dimension transposed convolution layer.
pub type ConvTranspose3D = ConvTransposeND<[i64; 3]>;

fn conv_transpose<'a, ND: std::convert::AsRef<[i64]>, T: Borrow<super::Path<'a>>>(
    vs: T,
    in_dim: i64,
    out_dim: i64,
    ksizes: ND,
    config: ConvTransposeConfigND<ND>,
) -> ConvTransposeND<ND> {
    let vs = vs.borrow();
    let bs = if config.bias { Some(vs.var("bias", &[out_dim], config.bs_init)) } else { None };
    let mut weight_size = vec![in_dim, out_dim / config.groups];
    weight_size.extend(ksizes.as_ref().iter());
    let ws = vs.var("weight", weight_size.as_slice(), config.ws_init);
    ConvTransposeND { ws, bs, config }
}

trait Create: std::convert::AsRef<[i64]> + std::marker::Sized {
    fn make_array(i: i64) -> Self;

    fn conv_transpose<'a, T: Borrow<super::Path<'a>>>(
        vs: T,
        in_dim: i64,
        out_dim: i64,
        ksize: i64,
        config: ConvTransposeConfig,
    ) -> ConvTransposeND<Self> {
        let config = ConvTransposeConfigND::<Self> {
            stride: Self::make_array(config.stride),
            padding: Self::make_array(config.padding),
            output_padding: Self::make_array(config.output_padding),
            dilation: Self::make_array(config.dilation),
            groups: config.groups,
            bias: config.bias,
            ws_init: config.ws_init,
            bs_init: config.bs_init,
        };
        conv_transpose(vs, in_dim, out_dim, Self::make_array(ksize), config)
    }
}

impl Create for [i64; 1] {
    fn make_array(i: i64) -> Self {
        [i]
    }
}

impl Create for [i64; 2] {
    fn make_array(i: i64) -> Self {
        [i, i]
    }
}

impl Create for [i64; 3] {
    fn make_array(i: i64) -> Self {
        [i, i, i]
    }
}

/// Creates a one dimension transposed convolution layer.
pub fn conv_transpose1d<'a, T: Borrow<Path<'a>>>(
    vs: T,
    i: i64,
    o: i64,
    k: i64,
    c: ConvTransposeConfig,
) -> ConvTranspose1D {
    <[i64; 1]>::conv_transpose(vs, i, o, k, c)
}

/// Creates a two dimension transposed convolution layer.
pub fn conv_transpose2d<'a, T: Borrow<Path<'a>>>(
    vs: T,
    i: i64,
    o: i64,
    k: i64,
    c: ConvTransposeConfig,
) -> ConvTranspose2D {
    <[i64; 2]>::conv_transpose(vs, i, o, k, c)
}

/// Creates a three dimension transposed convolution layer.
pub fn conv_transpose3d<'a, T: Borrow<Path<'a>>>(
    vs: T,
    i: i64,
    o: i64,
    k: i64,
    c: ConvTransposeConfig,
) -> ConvTranspose3D {
    <[i64; 3]>::conv_transpose(vs, i, o, k, c)
}

impl super::module::Module for ConvTranspose1D {
    fn forward(&self, xs: &Tensor) -> Tensor {
        Tensor::conv_transpose1d(
            xs,
            &self.ws,
            self.bs.as_ref(),
            self.config.stride,
            self.config.padding,
            self.config.output_padding,
            self.config.groups,
            self.config.dilation,
        )
    }
}

impl super::module::Module for ConvTranspose2D {
    fn forward(&self, xs: &Tensor) -> Tensor {
        Tensor::conv_transpose2d(
            xs,
            &self.ws,
            self.bs.as_ref(),
            self.config.stride,
            self.config.padding,
            self.config.output_padding,
            self.config.groups,
            self.config.dilation,
        )
    }
}

impl super::module::Module for ConvTranspose3D {
    fn forward(&self, xs: &Tensor) -> Tensor {
        Tensor::conv_transpose3d(
            xs,
            &self.ws,
            self.bs.as_ref(),
            self.config.stride,
            self.config.padding,
            self.config.output_padding,
            self.config.groups,
            self.config.dilation,
        )
    }
}