Quick Start

Create a Tree-based Tensor

You can create a tree-based tensor or a native tensor like the following example code.

 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
import builtins
import os
from functools import partial

import treetensor.torch as torch

print = partial(builtins.print, sep=os.linesep)

if __name__ == '__main__':
    t1 = torch.tensor([[1, 2, 3],
                       [4, 5, 6]])
    print('new native tensor:', t1)

    t2 = torch.tensor({
        'a': [1, 2, 3],
        'b': {'x': [[4, 5], [6, 7]]},
    })
    print('new tree tensor:', t2)

    t3 = torch.randn(2, 3)
    print('new random native tensor:', t3)

    t4 = torch.randn({
        'a': (2, 3),
        'b': {'x': (3, 4)},
    })
    print('new random tree tensor:', t4)

The output should be like below.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
new native tensor:
tensor([[1, 2, 3],
        [4, 5, 6]])
new tree tensor:
<Tensor 0x7f4bf7cd6250>
├── 'a' --> tensor([1, 2, 3])
└── 'b' --> <Tensor 0x7f4bf79fba30>
    └── 'x' --> tensor([[4, 5],
                        [6, 7]])

new random native tensor:
tensor([[0.4173, 0.4777, 0.7328],
        [0.6321, 0.2784, 1.0704]])
new random tree tensor:
<Tensor 0x7f4bf75e9ee0>
├── 'a' --> tensor([[-1.2351,  1.1970, -0.2185],
│                   [-0.4415,  0.7200, -0.2820]])
└── 'b' --> <Tensor 0x7f4bf7cc4460>
    └── 'x' --> tensor([[-1.1630, -1.7738, -0.0266,  0.9209],
                        [-2.4688,  0.0977, -0.4680,  1.1742],
                        [-0.0277, -1.0709,  0.6655,  1.9821]])