liaogang
8 years ago
commit
23e47bb600
@ -0,0 +1,77 @@
|
||||
#edit-mode: -*- python -*-
|
||||
# Copyright (c) 2016 Baidu, Inc. 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.
|
||||
|
||||
from paddle.trainer_config_helpers import *
|
||||
|
||||
######################## data source ################################
|
||||
define_py_data_sources2(train_list='gserver/tests/Sequence/dummy.list',
|
||||
test_list=None,
|
||||
module='rnn_data_provider',
|
||||
obj='process_subseq')
|
||||
|
||||
|
||||
settings(batch_size=2, learning_rate=0.01)
|
||||
######################## network configure ################################
|
||||
dict_dim = 10
|
||||
word_dim = 8
|
||||
hidden_dim = 8
|
||||
label_dim = 3
|
||||
|
||||
data = data_layer(name="word", size=dict_dim)
|
||||
|
||||
emb = embedding_layer(input=data, size=word_dim)
|
||||
|
||||
# This hierachical RNN is designed to be equivalent to the simple RNN in
|
||||
# sequence_rnn.conf
|
||||
|
||||
def outer_step(wid, x):
|
||||
outer_mem = memory(name="outer_rnn_state", size=hidden_dim)
|
||||
def inner_step(y, wid):
|
||||
z = embedding_layer(input=wid, size=word_dim)
|
||||
inner_mem = memory(name="inner_rnn_state",
|
||||
size=hidden_dim,
|
||||
boot_layer=outer_mem)
|
||||
out = fc_layer(input=[y, z, inner_mem],
|
||||
size=hidden_dim,
|
||||
act=TanhActivation(),
|
||||
bias_attr=True,
|
||||
name="inner_rnn_state")
|
||||
return out
|
||||
|
||||
inner_rnn_output = recurrent_group(
|
||||
step=inner_step,
|
||||
name="inner",
|
||||
input=[x, wid])
|
||||
last = last_seq(input=inner_rnn_output, name="outer_rnn_state")
|
||||
|
||||
# "return last" should also work. But currently RecurrentGradientMachine
|
||||
# does not handle it correctly. Current implementation requires that
|
||||
# all the out links are from sequences. However, it does not report error
|
||||
# when the out links are not sequences.
|
||||
return inner_rnn_output
|
||||
|
||||
out = recurrent_group(
|
||||
name="outer",
|
||||
step=outer_step,
|
||||
input=[SubsequenceInput(data), SubsequenceInput(emb)])
|
||||
|
||||
rep = last_seq(input=out)
|
||||
prob = fc_layer(size=label_dim,
|
||||
input=rep,
|
||||
act=SoftmaxActivation(),
|
||||
bias_attr=True)
|
||||
|
||||
outputs(classification_cost(input=prob,
|
||||
label=data_layer(name="label", size=label_dim)))
|
||||
@ -0,0 +1,58 @@
|
||||
#edit-mode: -*- python -*-
|
||||
# Copyright (c) 2016 Baidu, Inc. 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.
|
||||
|
||||
from paddle.trainer_config_helpers import *
|
||||
|
||||
######################## data source ################################
|
||||
define_py_data_sources2(train_list='gserver/tests/Sequence/dummy.list',
|
||||
test_list=None,
|
||||
module='rnn_data_provider',
|
||||
obj='process_seq')
|
||||
|
||||
|
||||
settings(batch_size=2, learning_rate=0.01)
|
||||
######################## network configure ################################
|
||||
dict_dim = 10
|
||||
word_dim = 8
|
||||
hidden_dim = 8
|
||||
label_dim = 3
|
||||
|
||||
data = data_layer(name="word", size=dict_dim)
|
||||
|
||||
emb = embedding_layer(input=data, size=word_dim)
|
||||
|
||||
def step(y, wid):
|
||||
z = embedding_layer(input=wid, size=word_dim)
|
||||
mem = memory(name="rnn_state", size=hidden_dim)
|
||||
out = fc_layer(input=[y, z, mem],
|
||||
size=hidden_dim,
|
||||
act=TanhActivation(),
|
||||
bias_attr=True,
|
||||
name="rnn_state")
|
||||
return out
|
||||
|
||||
out = recurrent_group(
|
||||
name="rnn",
|
||||
step=step,
|
||||
input=[emb, data])
|
||||
|
||||
rep = last_seq(input=out)
|
||||
prob = fc_layer(size=label_dim,
|
||||
input=rep,
|
||||
act=SoftmaxActivation(),
|
||||
bias_attr=True)
|
||||
|
||||
outputs(classification_cost(input=prob,
|
||||
label=data_layer(name="label", size=label_dim)))
|
||||
File diff suppressed because it is too large
Load Diff
@ -0,0 +1 @@
|
||||
*protostr
|
||||
@ -0,0 +1,17 @@
|
||||
7e6919d17562516e9a1d9a88de1fb3b9 img_layers.protostr
|
||||
a5d9259ff1fd7ca23d0ef090052cb1f2 last_first_seq.protostr
|
||||
9c038249ec8ff719753a746cdb04c026 layer_activations.protostr
|
||||
5913f87b39cee3b2701fa158270aca26 projections.protostr
|
||||
6b39e34beea8dfb782bee9bd3dea9eb5 simple_rnn_layers.protostr
|
||||
0fc1409600f1a3301da994ab9d28b0bf test_cost_layers.protostr
|
||||
144bc6d3a509de74115fa623741797ed test_expand_layer.protostr
|
||||
2378518bdb71e8c6e888b1842923df58 test_fc.protostr
|
||||
8bb44e1e5072d0c261572307e7672bda test_grumemory_layer.protostr
|
||||
1f3510672dce7a9ed25317fc58579ac7 test_hsigmoid.protostr
|
||||
d350bd91a0dc13e854b1364c3d9339c6 test_lstmemory_layer.protostr
|
||||
251a948ba41c1071afcd3d9cf9c233f7 test_ntm_layers.protostr
|
||||
e6ff04e70aea27c7b06d808cc49c9497 test_print_layer.protostr
|
||||
2a75dd33b640c49a8821c2da6e574577 test_rnn_group.protostr
|
||||
67d6fde3afb54f389d0ce4ff14726fe1 test_sequence_pooling.protostr
|
||||
f586a548ef4350ba1ed47a81859a64cb unused_layers.protostr
|
||||
8122477f4f65244580cec09edc590041 util_layers.protostr
|
||||
@ -0,0 +1,18 @@
|
||||
#!/bin/bash
|
||||
|
||||
set -e
|
||||
cd `dirname $0`
|
||||
export PYTHONPATH=$PWD/../../../../
|
||||
|
||||
configs=(test_fc layer_activations projections test_print_layer
|
||||
test_sequence_pooling test_lstmemory_layer test_grumemory_layer
|
||||
last_first_seq test_expand_layer test_ntm_layers test_hsigmoid
|
||||
img_layers util_layers simple_rnn_layers unused_layers test_cost_layers
|
||||
test_rnn_group)
|
||||
|
||||
|
||||
for conf in ${configs[*]}
|
||||
do
|
||||
echo "Generating " $conf
|
||||
python -m paddle.utils.dump_config $conf.py > $conf.protostr
|
||||
done
|
||||
Some files were not shown because too many files have changed in this diff Show More
Loading…
Reference in new issue