8000 [testing] Port `torch.{repeat, tile}` tests to use OpInfo machinery by kshitij12345 · Pull Request #50199 · pytorch/pytorch · GitHub
[go: up one dir, main page]
Skip to content
8000

[testing] Port torch.{repeat, tile} tests to use OpInfo machinery #50199

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 6 commits into from
Closed

[testing] Port torch.{repeat, tile} tests to use OpInfo machinery #50199

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
c20dd9f
port repeat tile tests to use OpInfo machinery
kshitij12345 Jan 7, 2021
126058c
use list comprehension instead of filter
kshitij12345 Jan 7, 2021
af87706
address comments
kshitij12345 Jan 8, 2021
ea92bca
Merge branch 'master' into develop/opinfo/repeat-tile
kshitij12345 Jan 14, 2021
1973767
Merge branch 'master' into develop/opinfo/repeat-tile
kshitij12345 Jan 15, 2021
f7171b7
Merge branch 'master' into develop/opinfo/repeat-tile
kshitij12345 Jan 18, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 16 additions & 1 deletion test/test_shape_ops.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,9 +8,10 @@
from torch._six import nan
from torch.testing._internal.common_utils import (
TestCase, run_tests, make_tensor, torch_to_numpy_dtype_dict)
from torch.testing._internal.common_methods_invocations import shape_funcs
from torch.testing._internal.common_device_type import (
instantiate_device_type_tests, onlyCPU, dtypes, onlyOnCPUAndCUDA,
dtypesIfCPU, dtypesIfCUDA)
dtypesIfCPU, dtypesIfCUDA, ops)

# TODO: replace with make_tensor
def _generate_input(shape, dtype, device, with_extremal):
Expand Down Expand Up @@ -599,7 +600,21 @@ def test_nonzero_non_diff(self, device):
nz = x.nonzero()
self.assertFalse(nz.requires_grad)

class TestShapeFuncs(TestCase):
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Add a comment here describing what this class is for.

Do we really want to add another test class instead of just putting this function into the existing TestShapeOps?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I just thought, going forward all ShapeFuncsInfo Op tests might as well go under TestShapeFuncs. But we can just move it to the existing class as well.

Do let me know if that is preferred.

Thanks!

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Either way is fine, just add a comment explaining why people should put a test here, for example, vs. the other test suite.

"""Test suite for Shape manipulating operators using the ShapeFuncInfo."""

@dtypes(*(torch.uint8, torch.int64, torch.double, torch.complex128))
@ops([op for op in shape_funcs if op.name in ['tile', 'repeat']])
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is a good way to filter the OpInfos. In the future we may want to consider allowing filtering directly by name or function. That is,

@ops('tile', 'repeat')
@ops(torch.tile, torch.repeat)

Which might be simpler to remember and more readable.

def test_repeat_tile_vs_numpy(self, device, dtype, op):
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Since torch.repeat is more restrictive than tile, does this test not test where torch.tile and np.tile are compatible but torch.repeat isn't?

samples = op.sample_inputs(device, dtype, requires_grad=False)
for sample in samples:
(t, dims) = sample.input
expected = op.ref(t.cpu().numpy(), dims, **sample.kwargs)
result = op(t, d 10000 ims, **sample.kwargs).cpu().numpy()
self.assertEqual(expected, result)

instantiate_device_type_tests(TestShapeOps, globals())
instantiate_device_type_tests(TestShapeFuncs, globals())

if __name__ == '__main__':
run_tests()
60 changes: 0 additions & 60 deletions test/test_tensor_creation_ops.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -702,27 +702,6 @@ def test_cat_different_devices(self, devices):
"input tensors must be on the same device"):
torch.cat((cpu, cuda0))

@onlyOnCPUAndCUDA
def test_tile(self, device):
shapes = ((6, 4, 3),
(1,),
())
reps = ((1, 10, 10, 99),
(25, 1, 1),
(3, 3, 3),
(1, 2, 0),
(2, 2),
(2,),
(1,),
())
for shape in shapes:
tensor = torch.randn(shape, device=device)
for t in (tensor, tensor.T):
for dims in reps:
expected = np.tile(t.cpu().numpy(), dims)
result = torch.tile(t, dims).cpu().numpy()
self.assertEqual(expected, result)

# TODO: reconcile with other cat tests
# TODO: Compare with a NumPy reference instead of CPU
@onlyCUDA
Expand Down Expand Up @@ -1127,25 +1106,6 @@ def test_stack_out(self, device):
self.assertEqual(res_out.select(dim, 1), y, atol=0, rtol=0)
self.assertEqual(res_out.select(dim, 2), z, atol=0, rtol=0)

def test_repeat(self, device):
initial_shape = (8, 4)
tensor = torch.rand(*initial_shape, device=device)

size = (3, 1, 1)
torchSize = torch.Size(size)
target = [3, 8, 4]
self.assertEqual(tensor.repeat(*size).size(), target, msg='Error in repeat')
self.assertEqual(tensor.repeat(torchSize).size(), target,
msg='Error in repeat using LongStorage')
result = tensor.repeat(*size)
self.assertEqual(result.size(), target, msg='Error in repeat using result')
result = tensor.repeat(torchSize)
self.assertEqual(result.size(), target, msg='Error in repeat using result and LongStorage')
self.assertEqual(result.mean(0).view(8, 4), tensor, msg='Error in repeat (not equal)')

zeroDimTarget = torch.Size([24, 0])
self.assertEqual(tensor.repeat((3, 0)).size(), zeroDimTarget, msg="Error when calling with 0 repeats")

def test_repeat_interleave(self, device):
x = torch.tensor([0, 1, 2, 3], device=device)
expected = torch.tensor([1, 2, 2, 3, 3, 3], device=device)
Expand Down Expand Up @@ -1196,26 +1156,6 @@ def test_repeat_interleave(self, device):
y = torch.repeat_interleave(x, x)
self.assertEqual(y, x)

def test_repeat_tile(self, device):
initial_shape = (8, 4)

repeats = ((3, 1, 1),
(3, 3, 3),
(1, 2, 1),
(2, 2, 2, 2))

def _generate_noncontiguous_input():
out = np.broadcast_to(np.random.random((1, 4)), initial_shape)
# Note: non-writeable NumPy arrays will warn if converted to tensors
out.setflags(write=True)
assert not (out.flags.c_contiguous or out.flags.f_contiguous)
return out

for repeat in repeats:
for tensor in (torch.from_numpy(np.random.random(initial_shape)).to(device),
torch.from_numpy(_generate_noncontiguous_input()).to(device)):
self.assertEqual(tensor.repeat(*repeat).cpu().numpy(), np.tile(tensor.cpu().numpy(), repeat))

# TODO: udpate to work on CUDA, too
@onlyCPU
def test_new_methods_requires_grad(self, device):
Expand Down
1 change: 0 additions & 1 deletion test/test_torch.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6834,7 +6834,6 @@ def inner(self, device, dtype):
('renorm', '2_norm', _small_3d, lambda t, d: [2, 1, 1], 1e-3, 1e-5, 1e-5, _float_types),
('renorm', '2_norm_neg_dim', _small_3d, lambda t, d: [2, -1, 1], 1e-3, 1e-5, 1e-5, _float_types),
('renorm', '1_5_norm', _small_3d, lambda t, d: [1.5, 1, 1], 1e-3, 1e-5, 1e-5, _float_types),
('repeat', '', _small_2d, lambda t, d: [2, 2, 2], 1e-5, 1e-5, 1e-5, _types, _cpu_types, False),
('size', '', _new_t((1, 2, 3, 4)), lambda t, d: [], 1e-5, 1e-5, 1e-5, _types, _cpu_types, False),
('size', 'dim', _new_t((1, 2, 3, 4)), lambda t, d: [1], 1e-5, 1e-5, 1e-5, _types, _cpu_types, False),
('size', 'neg_dim', _new_t((1, 2, 3, 4)), lambda t, d: [-2], 1e-5, 1e-5, 1e-5, _types, _cpu_types, False),
Expand Down
83 changes: 69 additions & 14 deletions torch/testing/_internal/common_methods_invocations.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -443,6 +443,34 @@ def sample_movedim_moveaxis(op_info, device, dtype, requires_grad):
requires_grad=requires_grad),
(0, -1, -2, -3), (-3, -2, -1, -0))))


def sample_repeat_tile(op_info, device, dtype, requires_grad):
rep_dims = ((), (0, ), (1, ), (0, 2), (1, 1), (2, 3), (2, 3, 2), (0, 2, 3), (2, 1, 1, 1),)
shapes = ((), (0,), (2,), (3, 0), (3, 2), (3, 0, 1))

if requires_grad:
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is a good way to filter the samples for gradcheck and gradgradcheck.

# Tests for variant_consistency_jit, grad, gradgrad
# are slower. Use smaller bags of `rep_dims` and `shapes`
# in this case.
rep_dims = ((), (0, ), (0, 2), (1, 1), (2, 3), (1, 3, 2), (3, 1, 1)) # type: ignore
shapes = ((), (0,), (2,), (3, 2)) # type: ignore

tensors = [make_tensor(shape, device, dtype,
low=None, high=None,
requires_grad=requires_grad) for shape in shapes]

samples = []
for rep_dim, tensor in product(rep_dims, tensors):
for t in (tensor, tensor.T):
if op_info.name == 'repeat' and len(rep_dim) >= t.dim():
# `torch.repeat` errors for `len(rep_dims) < t.dim()`,
# so we filter such combinations.
samples.append(SampleInput((t, rep_dim),))
elif op_info.name == 'tile':
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Add a comment here explaining the filtering for tile and repeat. This is a clever way to filter, and this answers my previous question about test coverage.

samples.append(SampleInput((t, rep_dim),))

return samples

def np_unary_ufunc_integer_promotion_wrapper(fn):
# Wrapper that passes PyTorch's default scalar
# type as an argument to the wrapped NumPy
Expand Down Expand Up @@ -529,6 +557,28 @@ def sample_inputs(self, device, dtype, requires_grad=False):
]


class ShapeFuncInfo(OpInfo):
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Add a brief comment here explaining what this derived class is intended for (maybe something like, "Early version of a specialized OpInfo for "shape" operations like tile and roll"?)

"""Early version of a specialized OpInfo for Shape manipulating operations like tile and roll"""
def __init__(self,
name, # the string name of the function
*,
ref, # a reference function
dtypes=floating_types(),
dtypesIfCPU=None,
dtypesIfCUDA=None,
dtypesIfROCM=None,
sample_inputs_func=None,
**kwargs):
super(ShapeFuncInfo, self).__init__(name,
dtypes=dtypes,
dtypesIfCPU=dtypesIfCPU,
dtypesIfCUDA=dtypesIfCUDA,
dtypesIfROCM=dtypesIfROCM,
sample_inputs_func=sample_inputs_func,
**kwargs)
self.ref = ref


class HermitianOpInfo(OpInfo):
"""Operator information for Hermitian functions
These are functions that take Hermitian matrices as input.
Expand Down Expand Up @@ -578,7 +628,6 @@ def sample_inputs_linalg_pinv_hermitian(op_info, device, dtype, requires_grad=Fa
o.kwargs = {"hermitian": True}
return out


def sample_inputs_linalg_solve(op_info, device, dtype, requires_grad=False):
"""
This function generates always solvable input for torch.linalg.solve
Expand Down Expand Up @@ -1405,6 +1454,24 @@ def sample_inputs_fliplr_flipud(op_info, device, dtype, requires_grad):
test_inplace_grad=False,
supports_tensor_out=False,
sample_inputs_func=sample_movedim_moveaxis),
ShapeFuncInfo('repeat',
op=lambda x, dims: x.repeat(dims),
ref=np.tile,
dtypes=all_types_and_complex_and(torch.bool, torch.float16, torch.bfloat16),
supports_tensor_out=False,
test_inplace_grad=False,
skips=(
# torch.repeat does not exist so we get a RuntimeError.
SkipInfo('TestCommon', 'test_variant_consistency_jit',
dtypes=all_types_and_complex_and(torch.bool, torch.float16, torch.bfloat16)),
),
sample_inputs_func=sample_repeat_tile),
ShapeFuncInfo('tile',
ref=np.tile,
dtypes=all_types_and_complex_and(torch.bool, torch.float16, torch.bfloat16),
supports_tensor_out=False,
test_inplace_grad=False,
sample_inputs_func=sample_repeat_tile),
]

if TEST_SCIPY:
Expand Down Expand Up @@ -1506,6 +1573,7 @@ def reference_sigmoid(x):
unary_ufuncs = [op for op in op_db if isinstance(op, UnaryUfuncInfo)]
spectral_funcs = [op for op in op_db if isinstance(op, SpectralFuncInfo)]
sparse_unary_ufuncs = [op for op in op_db if isinstance(op, UnaryUfuncInfo) and op.supports_sparse is True]
shape_funcs = [op for op in op_db if isinstance(op, ShapeFuncInfo)]

def index_variable(shape, max_indices, device=torch.device('cpu')):
if not isinstance(shape, tuple):
Expand Down Expand Up @@ -1982,14 +2050,6 @@ def method_tests():
('renorm', (S, S, S), (2, 1, 0.5), 'dim', (), [1]),
('renorm', (S, S, S), (1, 2, 3), 'norm_1'),
('renorm', (S, S, S), (inf, 2, 0.5), 'norm_inf'),
('repeat', (S,), (2,), 'single_number'),
('repeat', (), (2, 3), 'scalar'),
('repeat', (2, 2), (3, 2)),
('repeat', (2, 2), (1, 3, 1, 2), 'unsqueeze'),
('repeat', (S, S), (1, 1), 'keepdim0'),
('repeat', (S, S), (3, 1, 1), 'keepdim1'),
('repeat', (S,), (0, ), 'zero_dim'),
('repeat', (S,), (0, 2), 'zero_dim_multi'),
('logcumsumexp', (S, S, S), (0,), 'dim0', (), [0]),
('logcumsumexp', (S, S, S), (1,), 'dim1', (), [0]),
('logcumsumexp', (), (0,), 'dim0_scalar', (), [0]),
Expand Down Expand Up @@ -2206,11 +2266,6 @@ def method_tests():
('diagonal', (M, M, M), (1, 1, 2), '3d_1'),
('diagonal', (M, M, M), (2, 0, 1), '3d_2'),
('diagonal', (M, M, M), (-2, 0, 1), '3d_3'),
('tile', (2, 2), ([2, 2, 2],), 'more_reps_dims', (False,)),
('tile', (2, 2), ([2, 2],), 'same_reps_dims', (False,)),
('tile', (2, 2), ([2, 3],), 'less_reps_dims', (False,)),
('tile', (2, 2, 2), ([2, 2, 0],), 'zero_rep_dim', (False,)),
('tile', (), ([S, S, S],), 'empty_tensor', (False,)),
('tril', (M, M), NO_ARGS),
('tril', (M, M), (2,), 'idx'),
('tril', (S, M, M), NO_ARGS, 'batched'),
Expand Down
0