-
Notifications
You must be signed in to change notification settings - Fork 5.5k
/
input.py
344 lines (278 loc) · 11.9 KB
/
input.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
# 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 paddle
from paddle.fluid import Variable, core
from paddle.fluid.data_feeder import check_type
from paddle.fluid.framework import convert_np_dtype_to_dtype_, static_only
from paddle.fluid.layer_helper import LayerHelper
__all__ = []
@static_only
def data(name, shape, dtype=None, lod_level=0):
"""
This function creates a variable on the global block. The global variable
can be accessed by all the following operators in the graph. The variable
is a placeholder that could be fed with input, such as Executor can feed
input into the variable. When `dtype` is None, the dtype
will get from the global dtype by `paddle.get_default_dtype()`.
Args:
name (str): The name/alias of the variable, see :ref:`api_guide_Name`
for more details.
shape (list|tuple): List|Tuple of integers declaring the shape. You can
set None or -1 at a dimension to indicate the dimension can be of any
size. For example, it is useful to set changeable batch size as None or -1.
dtype (np.dtype|str, optional): The type of the data. Supported
dtype: bool, float16, float32, float64, int8, int16, int32, int64,
uint8. Default: None. When `dtype` is not set, the dtype will get
from the global dtype by `paddle.get_default_dtype()`.
lod_level (int, optional): The LoD level of the LoDTensor. Usually users
don't have to set this value. Default: 0.
Returns:
Variable: The global variable that gives access to the data.
Examples:
.. code-block:: python
import numpy as np
import paddle
paddle.enable_static()
# Creates a variable with fixed size [3, 2, 1]
# User can only feed data of the same shape to x
# the dtype is not set, so it will set "float32" by
# paddle.get_default_dtype(). You can use paddle.get_default_dtype() to
# change the global dtype
x = paddle.static.data(name='x', shape=[3, 2, 1])
# Creates a variable with changeable batch size -1.
# Users can feed data of any batch size into y,
# but size of each data sample has to be [2, 1]
y = paddle.static.data(name='y', shape=[-1, 2, 1], dtype='float32')
z = x + y
# In this example, we will feed x and y with np-ndarray "1"
# and fetch z, like implementing "1 + 1 = 2" in PaddlePaddle
feed_data = np.ones(shape=[3, 2, 1], dtype=np.float32)
exe = paddle.static.Executor(paddle.framework.CPUPlace())
out = exe.run(paddle.static.default_main_program(),
feed={
'x': feed_data,
'y': feed_data
},
fetch_list=[z.name])
# np-ndarray of shape=[3, 2, 1], dtype=float32, whose elements are 2
print(out)
"""
helper = LayerHelper('data', **locals())
check_type(name, 'name', (bytes, str), 'data')
check_type(shape, 'shape', (list, tuple), 'data')
shape = list(shape)
for i in range(len(shape)):
if shape[i] is None:
shape[i] = -1
if dtype:
return helper.create_global_variable(
name=name,
shape=shape,
dtype=dtype,
type=core.VarDesc.VarType.LOD_TENSOR,
stop_gradient=True,
lod_level=lod_level,
is_data=True,
need_check_feed=True,
)
else:
return helper.create_global_variable(
name=name,
shape=shape,
dtype=paddle.get_default_dtype(),
type=core.VarDesc.VarType.LOD_TENSOR,
stop_gradient=True,
lod_level=lod_level,
is_data=True,
need_check_feed=True,
)
class InputSpec:
"""
InputSpec describes the signature information of the model input, such as ``shape`` , ``dtype`` , ``name`` .
This interface is often used to specify input tensor information of models in high-level API.
It's also used to specify the tensor information for each input parameter of the forward function
decorated by `@paddle.jit.to_static`.
Args:
shape (tuple(integers)|list[integers]): List|Tuple of integers
declaring the shape. You can set "None" or -1 at a dimension
to indicate the dimension can be of any size. For example,
it is useful to set changeable batch size as "None" or -1.
dtype (np.dtype|str, optional): The type of the data. Supported
dtype: bool, float16, float32, float64, int8, int16, int32, int64,
uint8. Default: float32.
name (str): The name/alias of the variable, see :ref:`api_guide_Name`
for more details.
Examples:
.. code-block:: python
from paddle.static import InputSpec
input = InputSpec([None, 784], 'float32', 'x')
label = InputSpec([None, 1], 'int64', 'label')
print(input) # InputSpec(shape=(-1, 784), dtype=paddle.float32, name=x)
print(label) # InputSpec(shape=(-1, 1), dtype=paddle.int64, name=label)
"""
def __init__(self, shape, dtype='float32', name=None, stop_gradient=False):
# replace `None` in shape with -1
self.shape = self._verify(shape)
# convert dtype into united represention
if dtype is not None:
if not isinstance(dtype, core.VarDesc.VarType):
dtype = convert_np_dtype_to_dtype_(dtype)
self.dtype = dtype
self.name = name
self.stop_gradient = stop_gradient
def _create_feed_layer(self):
return data(self.name, shape=self.shape, dtype=self.dtype)
def __repr__(self):
return '{}(shape={}, dtype={}, name={}, stop_gradient={})'.format(
type(self).__name__,
self.shape,
self.dtype,
self.name,
self.stop_gradient,
)
@classmethod
def from_tensor(cls, tensor, name=None):
"""
Generates a InputSpec based on the description of input tensor.
Args:
tensor(Tensor): the source tensor to generate a InputSpec instance
Returns:
A InputSpec instance generated from Tensor.
Examples:
.. code-block:: python
import paddle
from paddle.static import InputSpec
paddle.disable_static()
x = paddle.ones([2, 2], dtype="float32")
x_spec = InputSpec.from_tensor(x, name='x')
print(x_spec) # InputSpec(shape=(2, 2), dtype=paddle.float32, name=x)
"""
if isinstance(tensor, (Variable, core.eager.Tensor)):
return cls(tensor.shape, tensor.dtype, name or tensor.name)
else:
raise ValueError(
"Input `tensor` should be a Tensor, but received {}.".format(
type(tensor).__name__
)
)
@classmethod
def from_numpy(cls, ndarray, name=None):
"""
Generates a InputSpec based on the description of input np.ndarray.
Args:
tensor(Tensor): the source numpy ndarray to generate a InputSpec instance
Returns:
A InputSpec instance generated from Tensor.
Examples:
.. code-block:: python
import numpy as np
from paddle.static import InputSpec
x = np.ones([2, 2], np.float32)
x_spec = InputSpec.from_numpy(x, name='x')
print(x_spec) # InputSpec(shape=(2, 2), dtype=paddle.float32, name=x)
"""
return cls(ndarray.shape, ndarray.dtype, name)
def batch(self, batch_size):
"""
Inserts `batch_size` in front of the `shape`.
Args:
batch_size(int): the inserted integer value of batch size.
Returns:
The original InputSpec instance by inserting `batch_size` in front of `shape`.
Examples:
.. code-block:: python
from paddle.static import InputSpec
x_spec = InputSpec(shape=[64], dtype='float32', name='x')
x_spec.batch(4)
print(x_spec) # InputSpec(shape=(4, 64), dtype=paddle.float32, name=x)
"""
if isinstance(batch_size, (list, tuple)):
if len(batch_size) != 1:
raise ValueError(
"Length of batch_size: {} shall be 1, but received {}.".format(
batch_size, len(batch_size)
)
)
batch_size = batch_size[1]
elif not isinstance(batch_size, int):
raise TypeError(
"type(batch_size) shall be `int`, but received {}.".format(
type(batch_size).__name__
)
)
new_shape = [batch_size] + list(self.shape)
self.shape = tuple(new_shape)
return self
def unbatch(self):
"""
Removes the first element of `shape`.
Returns:
The original InputSpec instance by removing the first element of `shape` .
Examples:
.. code-block:: python
from paddle.static import InputSpec
x_spec = InputSpec(shape=[4, 64], dtype='float32', name='x')
x_spec.unbatch()
print(x_spec) # InputSpec(shape=(64,), dtype=paddle.float32, name=x)
"""
if len(self.shape) == 0:
raise ValueError(
"Not support to unbatch a InputSpec when len(shape) == 0."
)
self.shape = self._verify(self.shape[1:])
return self
def _verify(self, shape):
"""
Verifies the input shape and modifies `None` into `-1`.
"""
if not isinstance(shape, (list, tuple)):
raise TypeError(
"Type of `shape` in InputSpec should be one of (tuple, list), but received {}.".format(
type(shape).__name__
)
)
for i, ele in enumerate(shape):
if ele is not None:
if not isinstance(ele, int):
raise ValueError(
"shape[{}] should be an `int`, but received `{}`:{}.".format(
i, type(ele).__name__, ele
)
)
if ele is None or ele < -1:
shape[i] = -1
return tuple(shape)
def __hash__(self):
# Note(Aurelius84): `name` is not considered as a field to compute hashkey.
# Because it's no need to generate a new program in following cases while using
# @paddle.jit.to_static.
#
# Case 1:
# foo(x_var)
# foo(y_var)
# x_var and y_var hold same shape and dtype, they should share a same program.
#
#
# Case 2:
# foo(x_var)
# foo(x_np) # x_np is a numpy.ndarray.
# x_var and x_np hold same shape and dtype, they should also share a same program.
return hash((tuple(self.shape), self.dtype, self.stop_gradient))
def __eq__(self, other):
slots = ['shape', 'dtype', 'name', 'stop_gradient']
return type(self) is type(other) and all(
getattr(self, attr) == getattr(other, attr) for attr in slots
)
def __ne__(self, other):
return not self == other