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

use crate::grad_clipping::GradientClippingConfig;
use crate::module::ADModule;
use crate::{self as burn, LearningRate};

use super::decay::{WeightDecay, WeightDecayConfig};
use super::momentum::{Momentum, MomentumConfig, MomentumState};
use super::SimpleOptimizer;
use crate::config::Config;
use crate::optim::adaptor::OptimizerAdaptor;
use crate::record::Record;
use crate::tensor::Tensor;
use burn_tensor::backend::{ADBackend, Backend};

/// Configuration to create the [Sgd](Sgd) optimizer.
#[derive(Config)]
pub struct SgdConfig {
    /// [Weight decay](WeightDecayConfig) config.
    weight_decay: Option<WeightDecayConfig>,
    /// [Momentum](MomentumConfig) config.
    momentum: Option<MomentumConfig>,
    /// [Gradient Clipping](GradientClippingConfig) config.
    gradient_clipping: Option<GradientClippingConfig>,
}

/// Optimizer that implements stochastic gradient descent with momentum.
///
/// The optimizer can be configured with [SgdConfig](SgdConfig).
pub struct Sgd<B: Backend> {
    momentum: Option<Momentum<B>>,
    weight_decay: Option<WeightDecay<B>>,
}

/// State of [Sgd](Sgd).
#[derive(Record, Clone, new)]
pub struct SgdState<B: Backend, const D: usize> {
    momentum: Option<MomentumState<B, D>>,
}

impl SgdConfig {
    /// Creates a new [SgdConfig](SgdConfig) with default values.
    pub fn init<B: ADBackend, M: ADModule<B>>(
        &self,
    ) -> OptimizerAdaptor<Sgd<B::InnerBackend>, M, B> {
        let momentum = self.momentum.as_ref().map(Momentum::new);
        let weight_decay = self.weight_decay.as_ref().map(WeightDecay::new);

        let mut optim = OptimizerAdaptor::from(Sgd {
            momentum,
            weight_decay,
        });
        if let Some(config) = &self.gradient_clipping {
            optim = optim.with_grad_clipping(config.init());
        }
        optim
    }
}

impl<B: Backend> SimpleOptimizer<B> for Sgd<B> {
    type State<const D: usize> = SgdState<B, D>;

    fn step<const D: usize>(
        &self,
        lr: LearningRate,
        tensor: Tensor<B, D>,
        mut grad: Tensor<B, D>,
        state: Option<Self::State<D>>,
    ) -> (Tensor<B, D>, Option<Self::State<D>>) {
        let mut state_momemtum = None;

        if let Some(state) = state {
            state_momemtum = state.momentum;
        }

        if let Some(weight_decay) = &self.weight_decay {
            grad = weight_decay.transform(grad, tensor.clone());
        }

        if let Some(momentum) = &self.momentum {
            let (grad_out, state) = momentum.transform(grad, state_momemtum);
            state_momemtum = Some(state);
            grad = grad_out;
        }

        let state = SgdState::new(state_momemtum);
        let delta = grad.mul_scalar(lr);

        (tensor - delta, Some(state))
    }

    fn to_device<const D: usize>(mut state: Self::State<D>, device: &B::Device) -> Self::State<D> {
        state.momentum = state.momentum.map(|state| state.to_device(device));
        state
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::{
        grad_clipping::GradientClipping,
        nn::{Linear, LinearConfig},
        optim::{GradientsParams, Optimizer},
        tensor::{Distribution, Shape},
        TestADBackend, TestBackend,
    };

    const LEARNING_RATE: LearningRate = 0.02;

    #[test]
    fn with_updated_params_should_have_state() {
        let layer = layer();
        let mut optim = sgd_with_all();
        let loss = layer.forward(random_tensor());
        let grads = loss.backward();
        let grads = GradientsParams::from_grads(grads, &layer);
        let _layer = optim.step(LEARNING_RATE, layer, grads);

        let record = optim.to_record();

        assert!(!record.is_empty());
    }

    #[test]
    fn without_updated_params_should_not_have_state() {
        let optim = sgd_with_all();
        let record = optim.to_record();
        assert!(record.is_empty());
    }

    #[test]
    fn can_attach_gradient_clipping() {
        let optim = sgd_with_all().with_grad_clipping(GradientClipping::Value(0.5));
        assert!(optim.has_gradient_clipping());
    }

    #[test]
    fn should_load_state() {
        let layer = layer();
        let mut optim = sgd_with_all();
        let loss = layer.forward(random_tensor());
        let grads = loss.backward();
        let grads = GradientsParams::from_grads(grads, &layer);
        let _layer = optim.step(LEARNING_RATE, layer, grads);

        let record = optim.to_record();
        let optim_new = sgd_with_all();
        let record_new = optim_new.to_record();
        let optim_new = optim_new.load_record(record.clone());
        let state_restored = optim_new.to_record();

        assert_ne!(record.len(), record_new.len());
        assert_eq!(record.len(), state_restored.len());
    }

    fn random_tensor() -> Tensor<TestADBackend, 2> {
        Tensor::<TestADBackend, 2>::random(Shape::new([2, 20]), Distribution::Default)
    }

    fn layer() -> Linear<TestADBackend> {
        LinearConfig::new(20, 20).with_bias(true).init()
    }

    fn sgd_with_all() -> OptimizerAdaptor<Sgd<TestBackend>, Linear<TestADBackend>, TestADBackend> {
        SgdConfig {
            weight_decay: Some(WeightDecayConfig { penalty: 0.05 }),
            momentum: Some(MomentumConfig {
                momentum: 0.9,
                dampening: 0.1,
                nesterov: true,
            }),
            gradient_clipping: None,
        }
        .init()
    }
}