flashlight_tensor 0.4.5

gpu/cpu tensor library focused around matrix and neural network operations
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

#[cfg(test)]
mod runner{
    use flashlight_tensor::prelude::*;
    
    #[tokio::test]
    async fn relu(){
        if std::env::var("CI").is_ok() {
            eprintln!("Skipping GPU test in CI");
            return;
        }
        let mut runner: GpuRunner = GpuRunner::init(1, MemoryMetric::GB);
        
        let tensor: Tensor<f32> = Tensor::rand(100.0, &[100]);
        
        let sample = Sample::from_data(vec!{tensor.clone()}, vec!{}, &[]);

        runner.append(sample);

        let full_gpu_output: Vec<Tensor<f32>> = runner.relu().await;
        let gpu_output = &full_gpu_output[0];
        let cpu_output = tensor.relu();

        let epsilon = 1e-4;
        for (a, b) in gpu_output.get_data().iter().zip(cpu_output.get_data()) {
            assert!((a - b).abs() < epsilon, "Values differ: GPU={} CPU={}", a, b);
        }
        assert_eq!(gpu_output.get_shape(), cpu_output.get_shape());
    }
    #[tokio::test]
    async fn relu_der(){
        if std::env::var("CI").is_ok() {
            eprintln!("Skipping GPU test in CI");
            return;
        }
        let mut runner: GpuRunner = GpuRunner::init(1, MemoryMetric::GB);
        
        let tensor: Tensor<f32> = Tensor::rand(100.0, &[100]);
        
        let sample = Sample::from_data(vec!{tensor.clone()}, vec!{}, &[]);

        runner.append(sample);

        let full_gpu_output: Vec<Tensor<f32>> = runner.relu_der().await;
        let gpu_output = &full_gpu_output[0];
        let cpu_output = tensor.relu_der();

        let epsilon = 1e-4;
        for (a, b) in gpu_output.get_data().iter().zip(cpu_output.get_data()) {
            assert!((a - b).abs() < epsilon, "Values differ: GPU={} CPU={}", a, b);
        }
        assert_eq!(gpu_output.get_shape(), cpu_output.get_shape());
    }

    #[tokio::test]
    async fn sigmoid(){
        if std::env::var("CI").is_ok() {
            eprintln!("Skipping GPU test in CI");
            return;
        }
        let mut runner: GpuRunner = GpuRunner::init(1, MemoryMetric::GB);
        
        let tensor: Tensor<f32> = Tensor::rand(100.0, &[100]);
        
        let sample = Sample::from_data(vec!{tensor.clone()}, vec!{}, &[3]);

        runner.append(sample);

        let full_gpu_output: Vec<Tensor<f32>> = runner.sigmoid().await;
        let gpu_output = &full_gpu_output[0];
        let cpu_output = tensor.sigmoid();

        let epsilon = 1e-4;
        for (a, b) in gpu_output.get_data().iter().zip(cpu_output.get_data()) {
            assert!((a - b).abs() < epsilon, "Values differ: GPU={} CPU={}", a, b);
        }
        assert_eq!(gpu_output.get_shape(), cpu_output.get_shape());
    }

    #[tokio::test]
    async fn sigmoid_der(){
        if std::env::var("CI").is_ok() {
            eprintln!("Skipping GPU test in CI");
            return;
        }
        let mut runner: GpuRunner = GpuRunner::init(1, MemoryMetric::GB);

        let tensor: Tensor<f32> = Tensor::rand(100.0, &[100]);
        
        let sample = Sample::from_data(vec!{tensor.clone()}, vec!{}, &[3]);

        runner.append(sample);

        let full_gpu_output: Vec<Tensor<f32>> = runner.sigmoid_der().await;
        let gpu_output = &full_gpu_output[0];
        let cpu_output = tensor.sigmoid_der();

        let epsilon = 1e-4;
        for (a, b) in gpu_output.get_data().iter().zip(cpu_output.get_data()) {
            assert!((a - b).abs() < epsilon, "Values differ: GPU={} CPU={}", a, b);
        }
        assert_eq!(gpu_output.get_shape(), cpu_output.get_shape());
    }
}