Skip to content
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

Jump to bottom

[AMP] Add bfloat16 and float16 tests for compare ops #51978

Merged
merged 3 commits into from
Mar 23, 2023
Merged

[AMP] Add bfloat16 and float16 tests for compare ops #51978

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
d4f0be9
add bf16 and fp16 tests
yeliang2258 Mar 22, 2023
a1f6663
Merge remote-tracking branch 'upstream/develop' into amp_compare_dev
yeliang2258 Mar 22, 2023
709309e
fix dtype check
yeliang2258 Mar 22, 2023
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
6 changes: 4 additions & 2 deletions paddle/phi/kernels/cpu/compare_kernel.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -94,7 +94,8 @@ PD_REGISTER_KERNEL(equal_all,
int64_t, \
float, \
double, \
phi::dtype::float16) { \
phi::dtype::float16, \
phi::dtype::bfloat16) { \
kernel->OutputAt(0).SetDataType(phi::DataType::BOOL); \
} \
PD_REGISTER_KERNEL(name##_raw, \
Expand All @@ -107,7 +108,8 @@ PD_REGISTER_KERNEL(equal_all,
int64_t, \
float, \
double, \
phi::dtype::float16) { \
phi::dtype::float16, \
phi::dtype::bfloat16) { \
kernel->OutputAt(0).SetDataType(phi::DataType::BOOL); \
}
PD_REGISTER_COMPARE_KERNEL(less_than, LessThan)
Expand Down
42 changes: 42 additions & 0 deletions python/paddle/fluid/tests/unittests/test_compare_op.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -136,6 +136,15 @@ def test_dynamic_api_float(self):
self.assertEqual((out.numpy() == self.real_result).all(), True)
paddle.enable_static()

def test_dynamic_api_float16(self):
paddle.disable_static()
x = paddle.to_tensor(self.input_x, dtype="float16")
y = paddle.to_tensor(self.input_y, dtype="float16")
op = eval("paddle.%s" % (self.op_type))
out = op(x, y)
self.assertEqual((out.numpy() == self.real_result).all(), True)
paddle.enable_static()

def test_dynamic_api_inf_1(self):
if self.op_type == "equal":
paddle.disable_static()
Expand Down Expand Up @@ -434,6 +443,39 @@ def test_attr_name(self):
create_paddle_case('not_equal', lambda _a, _b: _a != _b)


# add bf16 tests
def create_bf16_case(op_type, callback):
class TestCompareOpBF16Op(op_test.OpTest):
def setUp(self):
self.op_type = op_type
self.dtype = np.uint16
self.python_api = eval("paddle." + op_type)

x = np.random.uniform(0, 1, [5, 5]).astype(np.float32)
y = np.random.uniform(0, 1, [5, 5]).astype(np.float32)
real_result = callback(x, y)
self.inputs = {
'X': op_test.convert_float_to_uint16(x),
'Y': op_test.convert_float_to_uint16(y),
}
self.outputs = {'Out': real_result}

def test_check_output(self):
self.check_output()

cls_name = "BF16TestCase_{}".format(op_type)
TestCompareOpBF16Op.__name__ = cls_name
globals()[cls_name] = TestCompareOpBF16Op


create_bf16_case('less_than', lambda _a, _b: _a < _b)
create_bf16_case('less_equal', lambda _a, _b: _a <= _b)
create_bf16_case('greater_than', lambda _a, _b: _a > _b)
create_bf16_case('greater_equal', lambda _a, _b: _a >= _b)
create_bf16_case('equal', lambda _a, _b: _a == _b)
create_bf16_case('not_equal', lambda _a, _b: _a != _b)


class TestCompareOpError(unittest.TestCase):
def test_errors(self):
paddle.enable_static()
Expand Down
4 changes: 2 additions & 2 deletions python/paddle/tensor/logic.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -736,13 +736,13 @@ def not_equal(x, y, name=None):
check_variable_and_dtype(
x,
"x",
["bool", "float32", "float64", "int32", "int64"],
["bool", "float16", "float32", "float64", "int32", "int64"],
Copy link
Contributor

Choose a reason for hiding this comment

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

加一下uint16

Copy link
Contributor Author

Choose a reason for hiding this comment

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

#52016

"not_equal",
)
check_variable_and_dtype(
y,
"y",
["bool", "float32", "float64", "int32", "int64"],
["bool", "float16", "float32", "float64", "int32", "int64"],
"not_equal",
)
helper = LayerHelper("not_equal", **locals())
Expand Down