xref: /aosp_15_r20/external/pytorch/test/jit/test_tensor_creation_ops.py (revision da0073e96a02ea20f0ac840b70461e3646d07c45) 
1# Owner(s): ["oncall: jit"]
2
3import os
4import sys
5
6import torch
7
8
9# Make the helper files in test/ importable
10pytorch_test_dir = os.path.dirname(os.path.dirname(os.path.realpath(__file__)))
11sys.path.append(pytorch_test_dir)
12from torch.testing._internal.jit_utils import JitTestCase
13
14
15if __name__ == "__main__":
16    raise RuntimeError(
17        "This test file is not meant to be run directly, use:\n\n"
18        "\tpython test/test_jit.py TESTNAME\n\n"
19        "instead."
20    )
21
22
23class TestTensorCreationOps(JitTestCase):
24    """
25    A suite of tests for ops that create tensors.
26    """
27
28    def test_randperm_default_dtype(self):
29        def randperm(x: int):
30            perm = torch.randperm(x)
31            # Have to perform assertion here because TorchScript returns dtypes
32            # as integers, which are not comparable against eager torch.dtype.
33            assert perm.dtype == torch.int64
34
35        self.checkScript(randperm, (3,))
36
37    def test_randperm_specifed_dtype(self):
38        def randperm(x: int):
39            perm = torch.randperm(x, dtype=torch.float)
40            # Have to perform assertion here because TorchScript returns dtypes
41            # as integers, which are not comparable against eager torch.dtype.
42            assert perm.dtype == torch.float
43
44        self.checkScript(randperm, (3,))
45
46    def test_triu_indices_default_dtype(self):
47        def triu_indices(rows: int, cols: int):
48            indices = torch.triu_indices(rows, cols)
49            # Have to perform assertion here because TorchScript returns dtypes
50            # as integers, which are not comparable against eager torch.dtype.
51            assert indices.dtype == torch.int64
52
53        self.checkScript(triu_indices, (3, 3))
54
55    def test_triu_indices_specified_dtype(self):
56        def triu_indices(rows: int, cols: int):
57            indices = torch.triu_indices(rows, cols, dtype=torch.int32)
58            # Have to perform assertion here because TorchScript returns dtypes
59            # as integers, which are not comparable against eager torch.dtype.
60            assert indices.dtype == torch.int32
61
62        self.checkScript(triu_indices, (3, 3))
63
64    def test_tril_indices_default_dtype(self):
65        def tril_indices(rows: int, cols: int):
66            indices = torch.tril_indices(rows, cols)
67            # Have to perform assertion here because TorchScript returns dtypes
68            # as integers, which are not comparable against eager torch.dtype.
69            assert indices.dtype == torch.int64
70
71        self.checkScript(tril_indices, (3, 3))
72
73    def test_tril_indices_specified_dtype(self):
74        def tril_indices(rows: int, cols: int):
75            indices = torch.tril_indices(rows, cols, dtype=torch.int32)
76            # Have to perform assertion here because TorchScript returns dtypes
77            # as integers, which are not comparable against eager torch.dtype.
78            assert indices.dtype == torch.int32
79
80        self.checkScript(tril_indices, (3, 3))
81