burn-train 0.20.1

Training crate for the Burn 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

use crate::metric::TopKAccuracyInput;
use crate::metric::{
    AccuracyInput, Adaptor, ConfusionStatsInput, HammingScoreInput, LossInput, PerplexityInput,
    processor::ItemLazy,
};
use burn_core::tensor::backend::Backend;
use burn_core::tensor::{Int, Tensor, Transaction};
use burn_ndarray::NdArray;

/// Simple classification output adapted for multiple metrics.
#[derive(new)]
pub struct ClassificationOutput<B: Backend> {
    /// The loss.
    pub loss: Tensor<B, 1>,

    /// The output.
    pub output: Tensor<B, 2>,

    /// The targets.
    pub targets: Tensor<B, 1, Int>,
}

impl<B: Backend> ItemLazy for ClassificationOutput<B> {
    type ItemSync = ClassificationOutput<NdArray>;

    fn sync(self) -> Self::ItemSync {
        let [output, loss, targets] = Transaction::default()
            .register(self.output)
            .register(self.loss)
            .register(self.targets)
            .execute()
            .try_into()
            .expect("Correct amount of tensor data");

        let device = &Default::default();

        ClassificationOutput {
            output: Tensor::from_data(output, device),
            loss: Tensor::from_data(loss, device),
            targets: Tensor::from_data(targets, device),
        }
    }
}

impl<B: Backend> Adaptor<AccuracyInput<B>> for ClassificationOutput<B> {
    fn adapt(&self) -> AccuracyInput<B> {
        AccuracyInput::new(self.output.clone(), self.targets.clone())
    }
}

impl<B: Backend> Adaptor<LossInput<B>> for ClassificationOutput<B> {
    fn adapt(&self) -> LossInput<B> {
        LossInput::new(self.loss.clone())
    }
}

impl<B: Backend> Adaptor<TopKAccuracyInput<B>> for ClassificationOutput<B> {
    fn adapt(&self) -> TopKAccuracyInput<B> {
        TopKAccuracyInput::new(self.output.clone(), self.targets.clone())
    }
}

impl<B: Backend> Adaptor<PerplexityInput<B>> for ClassificationOutput<B> {
    fn adapt(&self) -> PerplexityInput<B> {
        PerplexityInput::new(self.output.clone(), self.targets.clone())
    }
}

impl<B: Backend> Adaptor<ConfusionStatsInput<B>> for ClassificationOutput<B> {
    fn adapt(&self) -> ConfusionStatsInput<B> {
        let [_, num_classes] = self.output.dims();
        if num_classes > 1 {
            ConfusionStatsInput::new(
                self.output.clone(),
                self.targets.clone().one_hot(num_classes).bool(),
            )
        } else {
            ConfusionStatsInput::new(
                self.output.clone(),
                self.targets.clone().unsqueeze_dim(1).bool(),
            )
        }
    }
}

/// Multi-label classification output adapted for multiple metrics.
#[derive(new)]
pub struct MultiLabelClassificationOutput<B: Backend> {
    /// The loss.
    pub loss: Tensor<B, 1>,

    /// The output.
    pub output: Tensor<B, 2>,

    /// The targets.
    pub targets: Tensor<B, 2, Int>,
}

impl<B: Backend> ItemLazy for MultiLabelClassificationOutput<B> {
    type ItemSync = MultiLabelClassificationOutput<NdArray>;

    fn sync(self) -> Self::ItemSync {
        let [output, loss, targets] = Transaction::default()
            .register(self.output)
            .register(self.loss)
            .register(self.targets)
            .execute()
            .try_into()
            .expect("Correct amount of tensor data");

        let device = &Default::default();

        MultiLabelClassificationOutput {
            output: Tensor::from_data(output, device),
            loss: Tensor::from_data(loss, device),
            targets: Tensor::from_data(targets, device),
        }
    }
}

impl<B: Backend> Adaptor<HammingScoreInput<B>> for MultiLabelClassificationOutput<B> {
    fn adapt(&self) -> HammingScoreInput<B> {
        HammingScoreInput::new(self.output.clone(), self.targets.clone())
    }
}

impl<B: Backend> Adaptor<LossInput<B>> for MultiLabelClassificationOutput<B> {
    fn adapt(&self) -> LossInput<B> {
        LossInput::new(self.loss.clone())
    }
}

impl<B: Backend> Adaptor<ConfusionStatsInput<B>> for MultiLabelClassificationOutput<B> {
    fn adapt(&self) -> ConfusionStatsInput<B> {
        ConfusionStatsInput::new(self.output.clone(), self.targets.clone().bool())
    }
}