dfdx 0.13.0

Ergonomic auto differentiation in Rust, with pytorch like apis.
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
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210

use num_traits::Float;
use rand_distr::uniform::SampleUniform;

use crate::{
    nn::modules::*,
    shapes::Dtype,
    tensor::{PutTape, SplitTape, Storage},
    tensor_ops::Device,
};

use super::mha::MultiHeadAttention;

/// A transformer encoder.
///
/// Generics
/// - `MODEL_DIM`: The size of query/key/value tensors. Given to [MultiHeadAttention].
/// - `NUM_HEADS`: The number of heads in [MultiHeadAttention].
/// - `FF_DIM`: The size of the hidden layer in
///   the feedforward network in [TransformerEncoderBlock].
/// - `NUM_LAYERS`: The number of [TransformerEncoderBlock] to use.
pub type TransformerEncoder<
    const MODEL_DIM: usize,
    const NUM_HEADS: usize,
    const FF_DIM: usize,
    const NUM_LAYERS: usize,
    E,
    D,
> = Repeated<TransformerEncoderBlock<MODEL_DIM, NUM_HEADS, FF_DIM, E, D>, NUM_LAYERS>;

pub mod builder {
    #[derive(Debug)]
    pub struct TransformerEncoder<
        const MODEL_DIM: usize,
        const NUM_HEADS: usize,
        const FF_DIM: usize,
        const NUM_LAYERS: usize,
    >;

    #[derive(Debug)]
    pub struct TransformerEncoderBlock<
        const MODEL_DIM: usize,
        const NUM_HEADS: usize,
        const FF_DIM: usize,
    >;
}

impl<const M: usize, const H: usize, const F: usize, const L: usize, E: Dtype, D: Device<E>>
    BuildOnDevice<D, E> for builder::TransformerEncoder<M, H, F, L>
where
    TransformerEncoder<M, H, F, L, E, D>: BuildModule<D, E>,
{
    type Built = TransformerEncoder<M, H, F, L, E, D>;
    fn try_build_on_device(device: &D) -> Result<Self::Built, <D>::Err> {
        Self::Built::try_build(device)
    }
}

impl<const M: usize, const H: usize, const F: usize, E: Dtype, D: Device<E>> BuildOnDevice<D, E>
    for builder::TransformerEncoderBlock<M, H, F>
where
    TransformerEncoderBlock<M, H, F, E, D>: BuildModule<D, E>,
{
    type Built = TransformerEncoderBlock<M, H, F, E, D>;
    fn try_build_on_device(device: &D) -> Result<Self::Built, <D>::Err> {
        Self::Built::try_build(device)
    }
}

/// A single transformer encoder block
///
/// Generics
/// - `MODEL_DIM`: The size of query/key/value tensors. Given to [MultiHeadAttention].
/// - `NUM_HEADS`: The number of heads in [MultiHeadAttention].
/// - `FF_DIM`: The size of the hidden layer in the feedforward network.
///
/// **Pytorch equivalent**:
/// ```python
/// encoder = torch.nn.TransformerEncoderLayer(
///    EMBED_DIM, NUM_HEADS, dim_feedforward=FF_DIM, batch_first=True, dropout=0.0
/// )
/// ```
#[derive(Clone, Debug)]
pub struct TransformerEncoderBlock<
    const MODEL_DIM: usize,
    const NUM_HEADS: usize,
    const FF_DIM: usize,
    E: Dtype,
    D: Storage<E>,
> {
    pub self_attn: MultiHeadAttention<MODEL_DIM, NUM_HEADS, MODEL_DIM, MODEL_DIM, E, D>,
    pub norm1: LayerNorm1D<MODEL_DIM, E, D>,
    pub ff: FF<MODEL_DIM, FF_DIM, E, D>,
    pub norm2: LayerNorm1D<MODEL_DIM, E, D>,
}

type FF<const M: usize, const F: usize, E, D> =
    Residual<(Linear<M, F, E, D>, ReLU, Linear<F, M, E, D>)>;

impl<const M: usize, const H: usize, const F: usize, E, D: Device<E>> TensorCollection<E, D>
    for TransformerEncoderBlock<M, H, F, E, D>
where
    E: Dtype + Float + SampleUniform,
{
    type To<E2: Dtype, D2: Device<E2>> = TransformerEncoderBlock<M, H, F, E2, D2>;

    fn iter_tensors<V: ModuleVisitor<Self, E, D>>(
        visitor: &mut V,
    ) -> Result<Option<Self::To<V::E2, V::D2>>, V::Err> {
        visitor.visit_fields(
            (
                Self::module("self_attn", |s| &s.self_attn, |s| &mut s.self_attn),
                Self::module("norm1", |s| &s.norm1, |s| &mut s.norm1),
                Self::module("ff", |s| &s.ff, |s| &mut s.ff),
                Self::module("norm2", |s| &s.norm2, |s| &mut s.norm2),
            ),
            |(self_attn, norm1, ff, norm2)| TransformerEncoderBlock {
                self_attn,
                norm1,
                ff,
                norm2,
            },
        )
    }
}

impl<const M: usize, const H: usize, const F: usize, E: Dtype, D: Device<E>, Src> Module<Src>
    for TransformerEncoderBlock<M, H, F, E, D>
where
    Src: SplitTape + std::ops::Add<Src::NoTape, Output = Src>,
    MultiHeadAttention<M, H, M, M, E, D>: Module<Src, Output = Src, Error = D::Err>,
    LayerNorm1D<M, E, D>: Module<Src, Output = Src, Error = D::Err>,
    FF<M, F, E, D>: Module<Src, Output = Src, Error = D::Err>,
{
    type Output = Src;
    type Error = D::Err;

    fn try_forward(&self, src: Src) -> Result<Self::Output, D::Err> {
        let (src, tape) = src.split_tape();
        let x = self.self_attn.try_forward(src.clone().put_tape(tape))?;
        let x = x + src;
        let x = self.norm1.try_forward(x)?;
        let x = self.ff.try_forward(x)?;
        self.norm2.try_forward(x)
    }
}

impl<const M: usize, const H: usize, const F: usize, E: Dtype, D: Device<E>> NonMutableModule
    for TransformerEncoderBlock<M, H, F, E, D>
{
}

#[cfg(test)]
#[allow(clippy::excessive_precision)]
mod tests {
    use super::*;
    use crate::{shapes::Rank3, tensor::*, tests::*};

    #[test]
    fn test_encoder_block_forward() {
        let dev = TestDevice::seed_from_u64(2);

        const BATCH: usize = 3;
        const SEQ_LEN: usize = 5;
        const EMBED_DIM: usize = 9;
        const NUM_HEADS: usize = 3;
        const FF_DIM: usize = 16;

        type Dtype = f32;

        let encoder = dev
            .build_module::<builder::TransformerEncoderBlock<EMBED_DIM, NUM_HEADS, FF_DIM>, Dtype>(
            );

        let x: Tensor<Rank3<BATCH, SEQ_LEN, EMBED_DIM>, Dtype, _> = dev.sample_normal();
        let y = encoder.forward(x);

        // This expected y was generated by:
        // 1. saving `encoder` parameters, `x` and `y` to a npz files
        // 2. Running pytorch with the same values
        // 3. printing out the output
        // See https://github.com/coreylowman/dfdx/wiki/Exporting-MultiHeadAttention-to-pytorch-for-unit-tests
        #[rustfmt::skip]
        assert_close_to_literal!(
            y,
            [
                [
                    [0.83316803, 0.85057360, 0.37431455, 1.48506296,-0.38405111,-1.89352179,-1.07049453,-0.50913972, 0.31408834],
                    [-0.57205188, 0.64078861,-0.56589824, 0.67155081, 0.65419787, 0.28409126,-1.75282931, 1.68111539,-1.04096484],
                    [-0.01414229, 1.34985816, 0.09684382, 0.13165890,-1.39875984,-1.61741352, 1.28747427, 0.75574619,-0.59126562],
                    [0.12542287, 2.60457349, 0.21064451,-0.81285846,-0.15861531,-0.87273139,-0.81707120,-0.17004849,-0.10931605],
                    [-1.54970682,-0.77183282, 1.37495196,-0.69562960,-0.66684282, 0.24720824, 1.38581741,-0.35962212, 1.03565681],
                ],
                [
                    [-0.15229249,-0.90768278,-0.85165489, 0.12768827, 1.61459768, 1.25826979,-0.46860829, 0.87496787,-1.49528503],
                    [-1.35595357, 1.13305736,-0.08542954, 1.01601434,-0.04678532,-1.69470263, 0.76144469,-0.68443829, 0.95679283],
                    [-1.49877191, 0.64559501, 0.33383703, 1.73698330,-0.14289393, 1.17869902,-1.01659226,-0.61038357,-0.62647283],
                    [0.78263682, 0.78481543,-0.16064386, 1.03396618, 1.49144781,-1.55002558,-1.11833119,-0.62120575,-0.64265978],
                    [-1.58957553, 1.75000548, 0.01272983, 0.11212827,-0.34744453,-1.45086825, 0.95842224, 0.50071126, 0.05389150],
                ],
                [
                    [-1.13160479,-0.21202824, 0.25907388,-0.64313424,-0.76302397,-0.16797650,-0.75345570, 2.01765633, 1.39449334],
                    [-0.16463053,-0.73241645,-0.69120175, 0.13771832, 0.72443259,-2.06525135, 1.02475107, 1.40244913, 0.36414924],
                    [0.38766465,-0.19543301,-1.80767059, 1.11545098, 0.21692322,-1.22834778, 0.13580292, 1.63094711,-0.25533777],
                    [1.22877085, 0.05472810, 0.65142977, 0.73869365,-0.74706972,-1.29277837, 1.07350135, 0.06228387,-1.76955938],
                    [-0.01733636,-1.57447529, 0.79691470, 1.00687420, 1.65637493,-0.75668150,-0.54616517, 0.45799020,-1.02349579],
                ],
            ]
        );
    }
}