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 0x7f2db60e9e50> ├── 'a' --> tensor([1, 2, 3]) └── 'b' --> <Tensor 0x7f2db60e9e80> └── 'x' --> tensor([[4, 5], [6, 7]]) new random native tensor: tensor([[ 0.1875, -0.2799, -0.5441], [-0.1897, 0.4704, -0.5737]]) new random tree tensor: <Tensor 0x7f2db58df070> ├── 'a' --> tensor([[ 0.1016, -1.6815, -1.2394], │ [ 2.4800, -0.3574, -0.2664]]) └── 'b' --> <Tensor 0x7f2db65bf190> └── 'x' --> tensor([[ 0.6640, -1.4243, 0.6565, 0.4917], [-1.0477, -0.4803, 0.9895, 1.3029], [ 1.4071, -1.3841, -0.3736, 0.3501]]) |