tch 0.6.1

Rust wrappers for the PyTorch C++ api (libtorch).
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

import torch

class Foo(torch.jit.ScriptModule):
    def __init__(self, v):
        super(Foo, self).__init__()
        self.register_buffer('value', v)

    @torch.jit.script_method
    def forward(self, x, y):
        return 2 * x + y + self.value

foo = Foo(torch.Tensor([42.0]))
foo.save('foo.pt')

class Foo1(torch.jit.ScriptModule):
    def __init__(self):
        super(Foo1, self).__init__()

    def forward(self, x, y):
        return 2 * x + y

foo = Foo1()
traced_foo = torch.jit.trace(foo, (torch.rand(3), torch.rand(3)))
traced_foo.save('foo1.pt')

class Foo2(torch.jit.ScriptModule):
    def __init__(self):
        super(Foo2, self).__init__()

    def forward(self, x, y):
        return (2 * x + y, x - y)

foo = Foo2()
traced_foo = torch.jit.trace(foo, (torch.rand(3), torch.rand(3)))
traced_foo.save('foo2.pt')

class Foo3(torch.jit.ScriptModule):
    def __init__(self):
        super(Foo3, self).__init__()

    @torch.jit.script_method
    def forward(self, x):
        result = x[0]
        for i in range(x.size(0)):
            if i: result = result * x[i]
        return result

foo = Foo3()
foo.save('foo3.pt')

from typing import Tuple, List

class Foo4(torch.jit.ScriptModule):
    def __init__(self):
        super(Foo4, self).__init__()

    @torch.jit.script_method
    def forward(self, x: Tuple[float, float, int]):
        return x[0] + x[1] * x[2]

foo = Foo4()
foo.save('foo4.pt')

class Foo5(torch.jit.ScriptModule):
    def __init__(self):
        super(Foo5, self).__init__()

    @torch.jit.script_method
    def forward(self, xs: List[str]):
      return [x[:-1] for x in xs]

foo = Foo5()
foo.save('foo5.pt')