wehub-resource-sync
143 lines
4.6 KiB
Python
143 lines
4.6 KiB
Python
# Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
import unittest
|
|
|
|
import numpy as np
|
|
|
|
import paddle
|
|
|
|
|
|
class TestTakeAlongAxisOutAndParamDecorator(unittest.TestCase):
|
|
def setUp(self):
|
|
paddle.disable_static()
|
|
self.input_shape = [2, 3, 4]
|
|
self.axis = 1
|
|
self.indices = paddle.to_tensor([[[0]]], dtype='int64')
|
|
self.out_shape = [2, 2, 4]
|
|
self.x_np = np.random.rand(*self.input_shape).astype(np.float32)
|
|
|
|
self.apis = [
|
|
paddle.take_along_dim,
|
|
paddle.take_along_axis,
|
|
]
|
|
self.test_types = [
|
|
"decorator1",
|
|
"decorator2",
|
|
"out",
|
|
"out_decorator",
|
|
]
|
|
|
|
def do_test(self, api, test_type):
|
|
x = paddle.to_tensor(self.x_np, stop_gradient=False)
|
|
out = paddle.empty(self.out_shape, dtype='float32')
|
|
out.stop_gradient = False
|
|
|
|
if test_type == 'raw':
|
|
out = api(x, self.indices, self.axis)
|
|
out.mean().backward()
|
|
return out, x.grad
|
|
elif test_type == 'decorator1':
|
|
out = api(x, dim=self.axis, indices=self.indices)
|
|
out.mean().backward()
|
|
return out, x.grad
|
|
elif test_type == 'decorator2':
|
|
out = api(dim=self.axis, indices=self.indices, input=x)
|
|
out.mean().backward()
|
|
return out, x.grad
|
|
elif test_type == 'out':
|
|
api(x, self.indices, self.axis, out=out)
|
|
out.mean().backward()
|
|
return out, x.grad
|
|
elif test_type == 'out_decorator':
|
|
api(input=x, indices=self.indices, dim=self.axis, out=out)
|
|
out.mean().backward()
|
|
return out, x.grad
|
|
else:
|
|
raise ValueError(f"Unknown test type: {test_type}")
|
|
|
|
def test_take_along_dim(self):
|
|
out_std, grad_std = self.do_test(paddle.take_along_dim, 'raw')
|
|
for test_type in self.test_types:
|
|
out, grad = self.do_test(paddle.take_along_dim, test_type)
|
|
np.testing.assert_allclose(out.numpy(), out_std.numpy(), rtol=1e-20)
|
|
np.testing.assert_allclose(
|
|
grad.numpy(), grad_std.numpy(), rtol=1e-20
|
|
)
|
|
|
|
|
|
class TestTensorTakeAlongAxisParamDecorator(unittest.TestCase):
|
|
def setUp(self):
|
|
paddle.disable_static()
|
|
|
|
self.input_shape = [2, 3, 4]
|
|
self.axis = 1
|
|
self.out_shape = [2, 2, 4]
|
|
|
|
self.x_np = np.random.rand(*self.input_shape).astype(np.float32)
|
|
|
|
self.indices_np = np.random.randint(
|
|
0, self.input_shape[self.axis], size=self.out_shape
|
|
).astype('int64')
|
|
|
|
self.method_names = [
|
|
'take_along_dim',
|
|
'take_along_axis',
|
|
]
|
|
|
|
self.test_types = ["kwargs"]
|
|
|
|
def do_test(self, method_name, test_type):
|
|
x = paddle.to_tensor(self.x_np, stop_gradient=False)
|
|
indices = paddle.to_tensor(self.indices_np)
|
|
out_tensor = paddle.empty(self.out_shape, dtype='float32')
|
|
out_tensor.stop_gradient = False
|
|
|
|
api_to_call = getattr(x, method_name)
|
|
|
|
if test_type == 'raw':
|
|
result = api_to_call(indices, self.axis)
|
|
elif test_type == 'kwargs':
|
|
result = api_to_call(indices=indices, axis=self.axis)
|
|
else:
|
|
raise ValueError(f"Unknown test type: {test_type}")
|
|
|
|
result.mean().backward()
|
|
|
|
return result, x.grad
|
|
|
|
def test_tensor_methods(self):
|
|
for method in self.method_names:
|
|
out_std, grad_std = self.do_test(method, 'raw')
|
|
|
|
for test_type in self.test_types:
|
|
with self.subTest(method=method, type=test_type):
|
|
out, grad = self.do_test(method, test_type)
|
|
|
|
np.testing.assert_allclose(
|
|
out.numpy(),
|
|
out_std.numpy(),
|
|
rtol=1e-20,
|
|
)
|
|
|
|
np.testing.assert_allclose(
|
|
grad.numpy(),
|
|
grad_std.numpy(),
|
|
rtol=1e-20,
|
|
)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
unittest.main()
|