提交 0bd75a57 编写于 作者: Q qiaolongfei

change layers to layer

上级 281250f5
...@@ -10,7 +10,7 @@ import random ...@@ -10,7 +10,7 @@ import random
import numpy as np import numpy as np
import paddle.trainer.PyDataProvider2 as dp import paddle.trainer.PyDataProvider2 as dp
import paddle.v2 import paddle.v2 as paddle_v2
import py_paddle.swig_paddle as api import py_paddle.swig_paddle as api
from paddle.trainer_config_helpers import * from paddle.trainer_config_helpers import *
from py_paddle import DataProviderConverter from py_paddle import DataProviderConverter
...@@ -58,7 +58,7 @@ def input_order_converter(generator): ...@@ -58,7 +58,7 @@ def input_order_converter(generator):
def main(): def main():
api.initPaddle("-use_gpu=false", "-trainer_count=4") # use 4 cpu cores api.initPaddle("-use_gpu=false", "-trainer_count=4") # use 4 cpu cores
optimizer = paddle.v2.optimizer.Adam( optimizer = paddle_v2.optimizer.Adam(
learning_rate=1e-4, learning_rate=1e-4,
batch_size=1000, batch_size=1000,
model_average=ModelAverage(average_window=0.5), model_average=ModelAverage(average_window=0.5),
...@@ -71,16 +71,17 @@ def main(): ...@@ -71,16 +71,17 @@ def main():
assert isinstance(updater, api.ParameterUpdater) assert isinstance(updater, api.ParameterUpdater)
# define network # define network
images = paddle.v2.layers.data_layer(name='pixel', size=784) images = paddle_v2.layer.data(name='pixel', size=784)
label = paddle.v2.layers.data_layer(name='label', size=10) label = paddle_v2.layer.data(name='label', size=10)
hidden1 = paddle.v2.layers.fc_layer(input=images, size=200) hidden1 = paddle_v2.layer.fc(input=images, size=200)
hidden2 = paddle.v2.layers.fc_layer(input=hidden1, size=200) hidden2 = paddle_v2.layer.fc(input=hidden1, size=200)
inference = paddle.v2.layers.fc_layer( inference = paddle_v2.layer.fc(input=hidden2,
input=hidden2, size=10, act=SoftmaxActivation()) size=10,
cost = paddle.v2.layers.classification_cost(input=inference, label=label) act=SoftmaxActivation())
cost = paddle_v2.layer.classification_cost(input=inference, label=label)
# Create Simple Gradient Machine. # Create Simple Gradient Machine.
model_config = paddle.v2.layers.parse_network(cost) model_config = paddle_v2.layer.parse_network(cost)
m = api.GradientMachine.createFromConfigProto(model_config, m = api.GradientMachine.createFromConfigProto(model_config,
api.CREATE_MODE_NORMAL, api.CREATE_MODE_NORMAL,
optimizer.enable_types()) optimizer.enable_types())
......
...@@ -13,6 +13,6 @@ ...@@ -13,6 +13,6 @@
# limitations under the License. # limitations under the License.
import optimizer import optimizer
import layers import layer
__all__ = ['optimizer', 'layers'] __all__ = ['optimizer', 'layer']
...@@ -19,6 +19,21 @@ from paddle.trainer_config_helpers.default_decorators import wrap_name_default ...@@ -19,6 +19,21 @@ from paddle.trainer_config_helpers.default_decorators import wrap_name_default
import collections import collections
def parse_network(*outputs):
"""
parse all output layers and then generate a model config proto.
:param outputs:
:return:
"""
def __real_func__():
context = dict()
real_output = [each.to_proto(context=context) for each in outputs]
conf_helps.outputs(real_output)
return __parse__(__real_func__)
class Layer(object): class Layer(object):
def __init__(self, name, parent_layer): def __init__(self, name, parent_layer):
assert isinstance(parent_layer, dict) assert isinstance(parent_layer, dict)
...@@ -49,22 +64,13 @@ class Layer(object): ...@@ -49,22 +64,13 @@ class Layer(object):
raise NotImplementedError() raise NotImplementedError()
def parse_network(*outputs): def __convert_to_v2__(method_name, name_prefix, parent_names):
def __real_func__():
context = dict()
real_output = [each.to_proto(context=context) for each in outputs]
conf_helps.outputs(real_output)
return __parse__(__real_func__)
def __convert__(method_name, name_prefix, parent_names):
if name_prefix is not None: if name_prefix is not None:
wrapper = wrap_name_default(name_prefix=name_prefix) wrapper = wrap_name_default(name_prefix=name_prefix)
else: else:
wrapper = None wrapper = None
class __Impl__(Layer): class V2LayerImpl(Layer):
def __init__(self, name=None, **kwargs): def __init__(self, name=None, **kwargs):
parent_layers = dict() parent_layers = dict()
other_kwargs = dict() other_kwargs = dict()
...@@ -75,7 +81,7 @@ def __convert__(method_name, name_prefix, parent_names): ...@@ -75,7 +81,7 @@ def __convert__(method_name, name_prefix, parent_names):
if key not in parent_names: if key not in parent_names:
other_kwargs[key] = kwargs[key] other_kwargs[key] = kwargs[key]
super(__Impl__, self).__init__(name, parent_layers) super(V2LayerImpl, self).__init__(name, parent_layers)
self.__other_kwargs__ = other_kwargs self.__other_kwargs__ = other_kwargs
if wrapper is not None: if wrapper is not None:
...@@ -89,24 +95,38 @@ def __convert__(method_name, name_prefix, parent_names): ...@@ -89,24 +95,38 @@ def __convert__(method_name, name_prefix, parent_names):
args[each] = self.__other_kwargs__[each] args[each] = self.__other_kwargs__[each]
return getattr(conf_helps, method_name)(name=self.name, **args) return getattr(conf_helps, method_name)(name=self.name, **args)
return __Impl__ return V2LayerImpl
data_layer = __convert__('data_layer', None, []) data = __convert_to_v2__('data_layer', None, [])
fc_layer = __convert__('fc_layer', name_prefix='fc', parent_names=['input']) fc = __convert_to_v2__('fc_layer', name_prefix='fc', parent_names=['input'])
classification_cost = __convert__( max_id = __convert_to_v2__(
'maxid_layer', name_prefix='maxid_layer', parent_names=['input'])
classification_cost = __convert_to_v2__(
'classification_cost', 'classification_cost',
name_prefix='classification_cost', name_prefix='classification_cost',
parent_names=['input', 'label']) parent_names=['input', 'label'])
cross_entropy_cost = __convert_to_v2__(
'cross_entropy',
name_prefix='cross_entropy',
parent_names=['input', 'label'])
__all__ = ['data_layer', 'fc_layer', 'classification_cost', 'parse_network'] __all__ = [
'parse_network', 'data', 'fc', 'max_id', 'classification_cost',
'cross_entropy_cost'
]
if __name__ == '__main__': if __name__ == '__main__':
data = data_layer(name='pixel', size=784) pixel = data(name='pixel', size=784)
hidden = fc_layer(input=data, size=100, act=conf_helps.SigmoidActivation()) label = data(name='label', size=10)
predict = fc_layer( hidden = fc(input=pixel, size=100, act=conf_helps.SigmoidActivation())
input=[hidden, data], size=10, act=conf_helps.SoftmaxActivation()) inference = fc(input=hidden, size=10, act=conf_helps.SoftmaxActivation())
cost = classification_cost( maxid = max_id(input=inference)
input=predict, label=data_layer( cost1 = classification_cost(input=inference, label=label)
name='label', size=10)) cost2 = cross_entropy_cost(input=inference, label=label)
print parse_network(cost)
print parse_network(cost1)
print parse_network(cost2)
print parse_network(cost1, cost2)
print parse_network(cost2)
print parse_network(inference, maxid)
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册