387 lines
10 KiB
Python
387 lines
10 KiB
Python
# Copyright (c) 2020 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
|
|
from dygraph_to_static_utils import (
|
|
Dy2StTestBase,
|
|
test_ast_only,
|
|
)
|
|
|
|
import paddle
|
|
from paddle import base
|
|
|
|
SEED = 2020
|
|
np.random.seed(SEED)
|
|
|
|
|
|
# Situation 1: Test list append
|
|
def test_list_append_without_control_flow(x):
|
|
# Python list will not be transformed.
|
|
x = paddle.assign(x)
|
|
a = []
|
|
# It's a plain python control flow which won't be transformed
|
|
if 2 > 1:
|
|
a.append(x)
|
|
return a
|
|
|
|
|
|
def test_list_append_in_if(x):
|
|
x = paddle.assign(x)
|
|
a = []
|
|
if x.numpy()[0] > 0:
|
|
a.append(x)
|
|
else:
|
|
a.append(paddle.full(shape=[1, 2], fill_value=9, dtype="float32"))
|
|
# TODO(Aurelius84): Currently, run_program_op doesn't support output DenseTensorArray.
|
|
return a[0]
|
|
|
|
|
|
def test_list_append_in_for_loop(x, iter_num):
|
|
x = paddle.assign(x)
|
|
# Use `full` so that static analysis can analyze the type of iter_num is Tensor
|
|
iter_num = paddle.full(
|
|
shape=[1], fill_value=iter_num, dtype="int32"
|
|
) # TODO(liym27): Delete it if the type of parameter iter_num can be resolved
|
|
a = []
|
|
for i in range(iter_num):
|
|
a.append(x)
|
|
return a[0]
|
|
|
|
|
|
def test_list_append_in_for_subscript(x):
|
|
x = paddle.assign(x)
|
|
iter_num = paddle.shape(x)[0]
|
|
a = []
|
|
for i in range(iter_num):
|
|
x = x + 1
|
|
a.append(x)
|
|
out = paddle.concat(a)
|
|
return out[0]
|
|
|
|
|
|
def test_list_append_in_while_loop_subscript(x):
|
|
x = paddle.assign(x)
|
|
iter_num = paddle.shape(x)[0]
|
|
a = []
|
|
i = 0
|
|
while i < iter_num:
|
|
x = x + 1
|
|
a.append(x)
|
|
i += 1
|
|
out = paddle.concat(a)
|
|
return out[0]
|
|
|
|
|
|
def test_list_append_in_for_loop_with_concat(x, iter_num):
|
|
x = paddle.assign(x)
|
|
a = []
|
|
# Use `full` so that static analysis can analyze the type of iter_num is Tensor
|
|
iter_num = paddle.full(
|
|
shape=[1], fill_value=iter_num, dtype="int32"
|
|
) # TODO(liym27): Delete it if the type of parameter iter_num can be resolved
|
|
for i in range(iter_num):
|
|
a.append(x)
|
|
a = paddle.concat(a, axis=0)
|
|
return a
|
|
|
|
|
|
def test_list_append_in_while_loop(x, iter_num):
|
|
x = paddle.assign(x)
|
|
iter_num = paddle.full(shape=[1], fill_value=iter_num, dtype="int32")
|
|
a = []
|
|
i = 0
|
|
while i < iter_num:
|
|
a.append(x)
|
|
i += 1
|
|
return a[0]
|
|
|
|
|
|
def test_list_append_in_while_loop_with_stack(x, iter_num):
|
|
x = paddle.assign(x)
|
|
iter_num = paddle.full(shape=[1], fill_value=iter_num, dtype="int32")
|
|
a = []
|
|
i = 0
|
|
while i < iter_num.numpy()[0]:
|
|
a.append(x)
|
|
i += 1
|
|
out = paddle.stack(a, axis=1)
|
|
return out
|
|
|
|
|
|
def test_tensor_array_slice(x, iter_num):
|
|
a = []
|
|
for i in range(paddle.to_tensor(3)):
|
|
a.append(paddle.to_tensor(float(i)))
|
|
t = a[1:3]
|
|
return a[2]
|
|
|
|
|
|
# Situation 2: Test list pop
|
|
def test_list_pop_without_control_flow_1(x):
|
|
x = paddle.assign(x)
|
|
a = []
|
|
if 2 > 1:
|
|
a.append(x)
|
|
a.pop()
|
|
return a
|
|
|
|
|
|
def test_list_pop_without_control_flow_2(x):
|
|
x = paddle.assign(x)
|
|
a = []
|
|
if 2 > 1:
|
|
a.append(x)
|
|
a.append(x + 1)
|
|
last_item = a.pop(1)
|
|
return last_item
|
|
|
|
|
|
def test_list_pop_in_if(x):
|
|
x = paddle.assign(x)
|
|
a = []
|
|
b = [x * 2 + (x + 1)]
|
|
if x.numpy()[0] > 0:
|
|
a.append(x)
|
|
b.append(x + 1)
|
|
a.append(paddle.full(shape=[1], fill_value=1, dtype="int64"))
|
|
else:
|
|
a.append(x + 1)
|
|
b.append(x - 1)
|
|
a.append(paddle.full(shape=[2], fill_value=2, dtype="int64"))
|
|
item1 = a.pop(1)
|
|
return item1, b[-1]
|
|
|
|
|
|
def test_list_pop_in_for_loop(x, iter_num):
|
|
x = paddle.assign(x)
|
|
# Use `full` so that static analysis can analyze the type of iter_num is Tensor
|
|
iter_num = paddle.full(
|
|
shape=[1], fill_value=iter_num, dtype="int32"
|
|
) # TODO(liym27): Delete it if the type of parameter iter_num can be resolved
|
|
|
|
a = []
|
|
b = [x - 1, x + 1]
|
|
for i in range(iter_num):
|
|
a.append(x + i)
|
|
b.append(x * 2)
|
|
|
|
one = paddle.ones(shape=[1], dtype="int32")
|
|
for i in range(one.numpy()[0]):
|
|
item = a.pop()
|
|
return a[0], item, b[1]
|
|
|
|
|
|
def test_list_pop_in_while_loop(x, iter_num):
|
|
x = paddle.assign(x)
|
|
iter_num = paddle.full(shape=[1], fill_value=iter_num, dtype="int32")
|
|
a = []
|
|
b = [x]
|
|
b.append(x)
|
|
b.pop()
|
|
i = 0
|
|
|
|
while i < iter_num:
|
|
a.append(x + i)
|
|
b.append(x - i)
|
|
i += 1
|
|
if i % 2 == 1:
|
|
a.pop()
|
|
return a[0], b[2]
|
|
|
|
|
|
class TestListWithoutControlFlowConfig(Dy2StTestBase):
|
|
def setUp(self):
|
|
self.place = base.CPUPlace()
|
|
if base.is_compiled_with_cuda():
|
|
self.place = base.CUDAPlace(0)
|
|
if base.is_compiled_with_xpu():
|
|
self.place = base.XPUPlace(0)
|
|
|
|
self.init_data()
|
|
self.init_dygraph_func()
|
|
|
|
def init_data(self):
|
|
self.input = np.random.random(3).astype('float32')
|
|
|
|
def init_dygraph_func(self):
|
|
self.all_dygraph_funcs = [
|
|
test_list_append_without_control_flow,
|
|
test_list_pop_without_control_flow_1,
|
|
test_list_pop_without_control_flow_2,
|
|
]
|
|
|
|
def result_to_numpy(self, res):
|
|
if isinstance(res, (list, tuple)):
|
|
res = paddle.utils.map_structure(lambda x: x.numpy(), res)
|
|
else:
|
|
res = [res.numpy()]
|
|
return res
|
|
|
|
def run_static_mode(self):
|
|
return self.train(to_static=True)
|
|
|
|
def run_dygraph_mode(self):
|
|
return self.train(to_static=False)
|
|
|
|
def train(self, to_static=False):
|
|
with base.dygraph.guard():
|
|
if to_static:
|
|
res = paddle.jit.to_static(self.dygraph_func)(self.input)
|
|
else:
|
|
res = self.dygraph_func(self.input)
|
|
return self.result_to_numpy(res)
|
|
|
|
def compare_transformed_static_result(self):
|
|
for dyfunc in self.all_dygraph_funcs:
|
|
self.dygraph_func = dyfunc
|
|
static_res_list = self.run_static_mode()
|
|
dygraph_res_list = self.run_dygraph_mode()
|
|
|
|
self.assertEqual(len(static_res_list), len(dygraph_res_list))
|
|
for stat_res, dy_res in zip(static_res_list, dygraph_res_list):
|
|
np.testing.assert_allclose(
|
|
stat_res,
|
|
dy_res,
|
|
rtol=1e-05,
|
|
err_msg=f'dygraph_res is {dy_res}\nstatic_res is {stat_res}',
|
|
)
|
|
|
|
|
|
class TestListWithoutControlFlow(TestListWithoutControlFlowConfig):
|
|
def test_transformed_static_result(self):
|
|
self.compare_transformed_static_result()
|
|
|
|
|
|
class TestListInIf(TestListWithoutControlFlow):
|
|
def init_dygraph_func(self):
|
|
self.all_dygraph_funcs = [test_list_append_in_if]
|
|
|
|
|
|
class TestListInWhileLoop(TestListWithoutControlFlowConfig):
|
|
def init_data(self):
|
|
self.input = np.random.random(3).astype('float32')
|
|
self.iter_num = 3
|
|
|
|
def init_dygraph_func(self):
|
|
self.all_dygraph_funcs = [
|
|
test_list_append_in_while_loop,
|
|
test_list_pop_in_while_loop,
|
|
]
|
|
|
|
# TODO(zhangbo): Refine BuildOpFrom for op with sub_block
|
|
def train(self, to_static=False):
|
|
with base.dygraph.guard():
|
|
if to_static:
|
|
res = paddle.jit.to_static(self.dygraph_func)(
|
|
self.input, self.iter_num
|
|
)
|
|
else:
|
|
res = self.dygraph_func(self.input, self.iter_num)
|
|
return self.result_to_numpy(res)
|
|
|
|
def test_transformed_static_result(self):
|
|
self.compare_transformed_static_result()
|
|
|
|
|
|
class TestListInWhileLoopWithStack(TestListInWhileLoop):
|
|
def init_dygraph_func(self):
|
|
self.all_dygraph_funcs = [test_list_append_in_while_loop_with_stack]
|
|
|
|
|
|
class TestTensorArraySlice(TestListInWhileLoop):
|
|
def init_dygraph_func(self):
|
|
self.all_dygraph_funcs = [test_tensor_array_slice]
|
|
|
|
|
|
class TestListInForLoop(TestListInWhileLoop):
|
|
def init_dygraph_func(self):
|
|
self.all_dygraph_funcs = [
|
|
test_list_append_in_for_loop,
|
|
test_list_pop_in_for_loop,
|
|
]
|
|
|
|
|
|
class TestListInForLoopWithConcat(TestListInWhileLoop):
|
|
def init_dygraph_func(self):
|
|
self.all_dygraph_funcs = [
|
|
test_list_append_in_for_loop_with_concat,
|
|
]
|
|
|
|
|
|
class TestListInForLoopWithSubscript(TestListWithoutControlFlow):
|
|
def init_dygraph_func(self):
|
|
self.all_dygraph_funcs = [
|
|
test_list_append_in_for_subscript,
|
|
test_list_append_in_while_loop_subscript,
|
|
]
|
|
|
|
def init_data(self):
|
|
self.input = np.random.random((3, 4)).astype('float32')
|
|
|
|
|
|
class ListWithCondNet(paddle.nn.Layer):
|
|
def __init__(self):
|
|
super().__init__()
|
|
|
|
# Add *args to test function.__self__ in FunctionSpec.
|
|
# DO NOT remove *args.
|
|
def forward(self, x, index, *args):
|
|
y = paddle.nn.functional.relu(x)
|
|
a = []
|
|
|
|
for i in y:
|
|
a.append(i)
|
|
|
|
if index > 0:
|
|
res = a[0] * a[0]
|
|
y = y + 1
|
|
else:
|
|
res = a[-1] * a[-1]
|
|
y = y - 1
|
|
|
|
z = a[-1] * res * y[0]
|
|
return z
|
|
|
|
|
|
class TestListWithCondGradInferVarType(Dy2StTestBase):
|
|
def test_to_static(self):
|
|
net = ListWithCondNet()
|
|
x = paddle.to_tensor([2, 3, 4], dtype='float32')
|
|
index = paddle.to_tensor([1])
|
|
res = paddle.jit.to_static(net)(x, index)
|
|
self.assertEqual(res, 48.0)
|
|
|
|
|
|
def tensor_array_dtype():
|
|
l = []
|
|
for i in range(paddle.to_tensor(3)):
|
|
l.append(i)
|
|
return l[0]
|
|
|
|
|
|
class TestTensorArrayDtype(Dy2StTestBase):
|
|
@test_ast_only
|
|
def test_tensor_array_dtype(self):
|
|
fn = tensor_array_dtype
|
|
static_fn = paddle.jit.to_static(fn)
|
|
st_out = static_fn()
|
|
self.assertEqual(st_out.dtype, paddle.int64)
|
|
|
|
|
|
if __name__ == '__main__':
|
|
unittest.main()
|