Skip to content

B
book

PaddlePaddle / book

通知 16
Star 4
Fork 0
B
book
提交 016995e9 编写于 作者: S sidgoyal78
浏览文件
操作

Add python files and edit readme

上级 64dda6c5
隐藏空白更改
内联 并排
Showing with 570 addition and 168 deletion
06.understand_sentiment/README.md
浏览文件 @ 016995e9
......@@ -102,14 +102,11 @@ After issuing a command `python train.py`, training will start immediately. The
### Initialize PaddlePaddle
We must import and initialize PaddlePaddle (enable/disable GPU, set the number of trainers, etc).
We must import and initialize Paddle.
```python
import sys
import paddle.v2 as paddle
# PaddlePaddle init
paddle.init(use_gpu=False, trainer_count=1)
import paddle
import paddle.fluid as fluid
```
As alluded to in section [Model Overview](#model-overview), here we provide the implementations of both Text CNN and Stacked-bidirectional LSTM models.
......@@ -121,20 +118,26 @@ We create a neural network `convolution_net` as the following snippet code.
Note: `paddle.networks.sequence_conv_pool` includes both convolution and pooling layer operations.
```python
def convolution_net(input_dim, class_dim=2, emb_dim=128, hid_dim=128):
data = paddle.layer.data("word",
paddle.data_type.integer_value_sequence(input_dim))
emb = paddle.layer.embedding(input=data, size=emb_dim)
conv_3 = paddle.networks.sequence_conv_pool(
input=emb, context_len=3, hidden_size=hid_dim)
conv_4 = paddle.networks.sequence_conv_pool(
input=emb, context_len=4, hidden_size=hid_dim)
output = paddle.layer.fc(input=[conv_3, conv_4],
size=class_dim,
act=paddle.activation.Softmax())
lbl = paddle.layer.data("label", paddle.data_type.integer_value(2))
cost = paddle.layer.classification_cost(input=output, label=lbl)
return cost, output
def convolution_net(data, input_dim, class_dim=2, emb_dim=128, hid_dim=128):
emb = fluid.layers.embedding(
input=data, size=[input_dim, emb_dim], is_sparse=True)
conv_3 = fluid.nets.sequence_conv_pool(
input=emb,
num_filters=hid_dim,
filter_size=3,
act="tanh",
pool_type="sqrt")
conv_4 = fluid.nets.sequence_conv_pool(
input=emb,
num_filters=hid_dim,
filter_size=4,
act="tanh",
pool_type="sqrt")
prediction = fluid.layers.fc(input=[conv_3, conv_4],
size=class_dim,
act="softmax")
return prediction
```
1. Define input data and its dimension
......@@ -154,70 +157,31 @@ def convolution_net(input_dim, class_dim=2, emb_dim=128, hid_dim=128):
We create a neural network `stacked_lstm_net` as below.
```python
def stacked_lstm_net(input_dim,
class_dim=2,
emb_dim=128,
hid_dim=512,
stacked_num=3):
"""
A Wrapper for sentiment classification task.
This network uses a bi-directional recurrent network,
consisting of three LSTM layers. This configuration is
motivated from the following paper, but uses few layers.
http://www.aclweb.org/anthology/P15-1109
input_dim: here is word dictionary dimension.
class_dim: number of categories.
emb_dim: dimension of word embedding.
hid_dim: dimension of hidden layer.
stacked_num: number of stacked lstm-hidden layer.
"""
def stacked_lstm_net(data, input_dim, class_dim, emb_dim, hid_dim, stacked_num):
assert stacked_num % 2 == 1
fc_para_attr = paddle.attr.Param(learning_rate=1e-3)
lstm_para_attr = paddle.attr.Param(initial_std=0., learning_rate=1.)
para_attr = [fc_para_attr, lstm_para_attr]
bias_attr = paddle.attr.Param(initial_std=0., l2_rate=0.)
relu = paddle.activation.Relu()
linear = paddle.activation.Linear()
data = paddle.layer.data("word",
paddle.data_type.integer_value_sequence(input_dim))
emb = paddle.layer.embedding(input=data, size=emb_dim)
emb = fluid.layers.embedding(
input=data, size=[input_dim, emb_dim], is_sparse=True)
fc1 = paddle.layer.fc(input=emb,
size=hid_dim,
act=linear,
bias_attr=bias_attr)
lstm1 = paddle.layer.lstmemory(
input=fc1, act=relu, bias_attr=bias_attr)
fc1 = fluid.layers.fc(input=emb, size=hid_dim)
lstm1, cell1 = fluid.layers.dynamic_lstm(input=fc1, size=hid_dim)
inputs = [fc1, lstm1]
for i in range(2, stacked_num + 1):
fc = paddle.layer.fc(input=inputs,
size=hid_dim,
act=linear,
param_attr=para_attr,
bias_attr=bias_attr)
lstm = paddle.layer.lstmemory(
input=fc,
reverse=(i % 2) == 0,
act=relu,
bias_attr=bias_attr)
fc = fluid.layers.fc(input=inputs, size=hid_dim)
lstm, cell = fluid.layers.dynamic_lstm(
input=fc, size=hid_dim, is_reverse=(i % 2) == 0)
inputs = [fc, lstm]
fc_last = paddle.layer.pooling(
input=inputs[0], pooling_type=paddle.pooling.Max())
lstm_last = paddle.layer.pooling(
input=inputs[1], pooling_type=paddle.pooling.Max())
output = paddle.layer.fc(input=[fc_last, lstm_last],
size=class_dim,
act=paddle.activation.Softmax(),
bias_attr=bias_attr,
param_attr=para_attr)
lbl = paddle.layer.data("label", paddle.data_type.integer_value(2))
cost = paddle.layer.classification_cost(input=output, label=lbl)
return cost, output
fc_last = fluid.layers.sequence_pool(input=inputs[0], pool_type='max')
lstm_last = fluid.layers.sequence_pool(input=inputs[1], pool_type='max')
prediction = fluid.layers.fc(input=[fc_last, lstm_last],
size=class_dim,
act='softmax')
return prediction
```
1. Define input data and its dimension
......@@ -236,14 +200,7 @@ def stacked_lstm_net(input_dim,
To reiterate, we can either invoke `convolution_net` or `stacked_lstm_net`.
```python
word_dict = paddle.dataset.imdb.word_dict()
dict_dim = len(word_dict)
class_dim = 2
# option 1
[cost, output] = convolution_net(dict_dim, class_dim=class_dim)
# option 2
# [cost, output] = stacked_lstm_net(dict_dim, class_dim=class_dim, stacked_num=3)
TODO
```
## Model Training
......
06.understand_sentiment/index.html
浏览文件 @ 016995e9
......@@ -144,14 +144,11 @@ After issuing a command `python train.py`, training will start immediately. The
### Initialize PaddlePaddle
We must import and initialize PaddlePaddle (enable/disable GPU, set the number of trainers, etc).
We must import and initialize Paddle.
```python
import sys
import paddle.v2 as paddle
# PaddlePaddle init
paddle.init(use_gpu=False, trainer_count=1)
import paddle
import paddle.fluid as fluid
```
As alluded to in section [Model Overview](#model-overview), here we provide the implementations of both Text CNN and Stacked-bidirectional LSTM models.
......@@ -163,20 +160,26 @@ We create a neural network `convolution_net` as the following snippet code.
Note: `paddle.networks.sequence_conv_pool` includes both convolution and pooling layer operations.
```python
def convolution_net(input_dim, class_dim=2, emb_dim=128, hid_dim=128):
data = paddle.layer.data("word",
paddle.data_type.integer_value_sequence(input_dim))
emb = paddle.layer.embedding(input=data, size=emb_dim)
conv_3 = paddle.networks.sequence_conv_pool(
input=emb, context_len=3, hidden_size=hid_dim)
conv_4 = paddle.networks.sequence_conv_pool(
input=emb, context_len=4, hidden_size=hid_dim)
output = paddle.layer.fc(input=[conv_3, conv_4],
size=class_dim,
act=paddle.activation.Softmax())
lbl = paddle.layer.data("label", paddle.data_type.integer_value(2))
cost = paddle.layer.classification_cost(input=output, label=lbl)
return cost, output
def convolution_net(data, input_dim, class_dim=2, emb_dim=128, hid_dim=128):
emb = fluid.layers.embedding(
input=data, size=[input_dim, emb_dim], is_sparse=True)
conv_3 = fluid.nets.sequence_conv_pool(
input=emb,
num_filters=hid_dim,
filter_size=3,
act="tanh",
pool_type="sqrt")
conv_4 = fluid.nets.sequence_conv_pool(
input=emb,
num_filters=hid_dim,
filter_size=4,
act="tanh",
pool_type="sqrt")
prediction = fluid.layers.fc(input=[conv_3, conv_4],
size=class_dim,
act="softmax")
return prediction
```
1. Define input data and its dimension
......@@ -196,70 +199,31 @@ def convolution_net(input_dim, class_dim=2, emb_dim=128, hid_dim=128):
We create a neural network `stacked_lstm_net` as below.
```python
def stacked_lstm_net(input_dim,
class_dim=2,
emb_dim=128,
hid_dim=512,
stacked_num=3):
"""
A Wrapper for sentiment classification task.
This network uses a bi-directional recurrent network,
consisting of three LSTM layers. This configuration is
motivated from the following paper, but uses few layers.
http://www.aclweb.org/anthology/P15-1109
input_dim: here is word dictionary dimension.
class_dim: number of categories.
emb_dim: dimension of word embedding.
hid_dim: dimension of hidden layer.
stacked_num: number of stacked lstm-hidden layer.
"""
def stacked_lstm_net(data, input_dim, class_dim, emb_dim, hid_dim, stacked_num):
assert stacked_num % 2 == 1
fc_para_attr = paddle.attr.Param(learning_rate=1e-3)
lstm_para_attr = paddle.attr.Param(initial_std=0., learning_rate=1.)
para_attr = [fc_para_attr, lstm_para_attr]
bias_attr = paddle.attr.Param(initial_std=0., l2_rate=0.)
relu = paddle.activation.Relu()
linear = paddle.activation.Linear()
data = paddle.layer.data("word",
paddle.data_type.integer_value_sequence(input_dim))
emb = paddle.layer.embedding(input=data, size=emb_dim)
emb = fluid.layers.embedding(
input=data, size=[input_dim, emb_dim], is_sparse=True)
fc1 = paddle.layer.fc(input=emb,
size=hid_dim,
act=linear,
bias_attr=bias_attr)
lstm1 = paddle.layer.lstmemory(
input=fc1, act=relu, bias_attr=bias_attr)
fc1 = fluid.layers.fc(input=emb, size=hid_dim)
lstm1, cell1 = fluid.layers.dynamic_lstm(input=fc1, size=hid_dim)
inputs = [fc1, lstm1]
for i in range(2, stacked_num + 1):
fc = paddle.layer.fc(input=inputs,
size=hid_dim,
act=linear,
param_attr=para_attr,
bias_attr=bias_attr)
lstm = paddle.layer.lstmemory(
input=fc,
reverse=(i % 2) == 0,
act=relu,
bias_attr=bias_attr)
fc = fluid.layers.fc(input=inputs, size=hid_dim)
lstm, cell = fluid.layers.dynamic_lstm(
input=fc, size=hid_dim, is_reverse=(i % 2) == 0)
inputs = [fc, lstm]
fc_last = paddle.layer.pooling(
input=inputs[0], pooling_type=paddle.pooling.Max())
lstm_last = paddle.layer.pooling(
input=inputs[1], pooling_type=paddle.pooling.Max())
output = paddle.layer.fc(input=[fc_last, lstm_last],
size=class_dim,
act=paddle.activation.Softmax(),
bias_attr=bias_attr,
param_attr=para_attr)
lbl = paddle.layer.data("label", paddle.data_type.integer_value(2))
cost = paddle.layer.classification_cost(input=output, label=lbl)
return cost, output
fc_last = fluid.layers.sequence_pool(input=inputs[0], pool_type='max')
lstm_last = fluid.layers.sequence_pool(input=inputs[1], pool_type='max')
prediction = fluid.layers.fc(input=[fc_last, lstm_last],
size=class_dim,
act='softmax')
return prediction
```
1. Define input data and its dimension
......@@ -278,14 +242,7 @@ def stacked_lstm_net(input_dim,
To reiterate, we can either invoke `convolution_net` or `stacked_lstm_net`.
```python
word_dict = paddle.dataset.imdb.word_dict()
dict_dim = len(word_dict)
class_dim = 2
# option 1
[cost, output] = convolution_net(dict_dim, class_dim=class_dim)
# option 2
# [cost, output] = stacked_lstm_net(dict_dim, class_dim=class_dim, stacked_num=3)
TODO
```
## Model Training
......
06.understand_sentiment/train_conv.py 0 → 100644
浏览文件 @ 016995e9
# Copyright (c) 2018 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.
from __future__ import print_function
import paddle
import paddle.fluid as fluid
from functools import partial
import numpy as np
CLASS_DIM = 2
EMB_DIM = 128
HID_DIM = 512
BATCH_SIZE = 128
def convolution_net(data, input_dim, class_dim, emb_dim, hid_dim):
emb = fluid.layers.embedding(
input=data, size=[input_dim, emb_dim], is_sparse=True)
conv_3 = fluid.nets.sequence_conv_pool(
input=emb,
num_filters=hid_dim,
filter_size=3,
act="tanh",
pool_type="sqrt")
conv_4 = fluid.nets.sequence_conv_pool(
input=emb,
num_filters=hid_dim,
filter_size=4,
act="tanh",
pool_type="sqrt")
prediction = fluid.layers.fc(
input=[conv_3, conv_4], size=class_dim, act="softmax")
return prediction
def inference_program(word_dict):
data = fluid.layers.data(
name="words", shape=[1], dtype="int64", lod_level=1)
dict_dim = len(word_dict)
net = convolution_net(data, dict_dim, CLASS_DIM, EMB_DIM, HID_DIM)
return net
def train_program(word_dict):
prediction = inference_program(word_dict)
label = fluid.layers.data(name="label", shape=[1], dtype="int64")
cost = fluid.layers.cross_entropy(input=prediction, label=label)
avg_cost = fluid.layers.mean(cost)
accuracy = fluid.layers.accuracy(input=prediction, label=label)
return [avg_cost, accuracy]
def optimizer_func():
return fluid.optimizer.Adagrad(learning_rate=0.002)
def train(use_cuda, train_program, params_dirname):
place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
word_dict = paddle.dataset.imdb.word_dict()
trainer = fluid.Trainer(
train_func=partial(train_program, word_dict),
place=place,
optimizer_func=optimizer_func)
def event_handler(event):
if isinstance(event, fluid.EndEpochEvent):
test_reader = paddle.batch(
paddle.dataset.imdb.test(word_dict), batch_size=BATCH_SIZE)
avg_cost, acc = trainer.test(
reader=test_reader, feed_order=['words', 'label'])
print("avg_cost: %s" % avg_cost)
print("acc : %s" % acc)
if acc > 0.2: # Smaller value to increase CI speed
trainer.save_params(params_dirname)
trainer.stop()
else:
print('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc))
if math.isnan(avg_cost):
sys.exit("got NaN loss, training failed.")
elif isinstance(event, fluid.EndStepEvent):
print("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, map(np.array, event.metrics)))
if event.step == 1: # Run 2 iterations to speed CI
trainer.save_params(params_dirname)
trainer.stop()
train_reader = paddle.batch(
paddle.reader.shuffle(
paddle.dataset.imdb.train(word_dict), buf_size=25000),
batch_size=BATCH_SIZE)
trainer.train(
num_epochs=1,
event_handler=event_handler,
reader=train_reader,
feed_order=['words', 'label'])
def infer(use_cuda, inference_program, params_dirname=None):
place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
word_dict = paddle.dataset.imdb.word_dict()
inferencer = fluid.Inferencer(
infer_func=partial(inference_program, word_dict),
param_path=params_dirname,
place=place)
# Setup input by creating LoDTensor to represent sequence of words.
# Here each word is the basic element of the LoDTensor and the shape of
# each word (base_shape) should be [1] since it is simply an index to
# look up for the corresponding word vector.
# Suppose the length_based level of detail (lod) info is set to [[3, 4, 2]],
# which has only one lod level. Then the created LoDTensor will have only
# one higher level structure (sequence of words, or sentence) than the basic
# element (word). Hence the LoDTensor will hold data for three sentences of
# length 3, 4 and 2, respectively.
# Note that lod info should be a list of lists.
lod = [[3, 4, 2]]
base_shape = [1]
# The range of random integers is [low, high]
tensor_words = fluid.create_random_int_lodtensor(
lod, base_shape, place, low=0, high=len(word_dict) - 1)
results = inferencer.infer({'words': tensor_words})
print("infer results: ", results)
def main(use_cuda):
if use_cuda and not fluid.core.is_compiled_with_cuda():
return
params_dirname = "understand_sentiment_conv.inference.model"
train(use_cuda, train_program, params_dirname)
infer(use_cuda, inference_program, params_dirname)
if __name__ == '__main__':
for use_cuda in (False, True):
main(use_cuda=use_cuda)
06.understand_sentiment/train_dyn_rnn.py 0 → 100644
浏览文件 @ 016995e9
# Copyright (c) 2018 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.
from __future__ import print_function
import paddle
import paddle.fluid as fluid
from functools import partial
import numpy as np
CLASS_DIM = 2
EMB_DIM = 128
BATCH_SIZE = 128
LSTM_SIZE = 128
def dynamic_rnn_lstm(data, input_dim, class_dim, emb_dim, lstm_size):
emb = fluid.layers.embedding(
input=data, size=[input_dim, emb_dim], is_sparse=True)
sentence = fluid.layers.fc(input=emb, size=lstm_size, act='tanh')
rnn = fluid.layers.DynamicRNN()
with rnn.block():
word = rnn.step_input(sentence)
prev_hidden = rnn.memory(value=0.0, shape=[lstm_size])
prev_cell = rnn.memory(value=0.0, shape=[lstm_size])
def gate_common(ipt, hidden, size):
gate0 = fluid.layers.fc(input=ipt, size=size, bias_attr=True)
gate1 = fluid.layers.fc(input=hidden, size=size, bias_attr=False)
return gate0 + gate1
forget_gate = fluid.layers.sigmoid(x=gate_common(word, prev_hidden,
lstm_size))
input_gate = fluid.layers.sigmoid(x=gate_common(word, prev_hidden,
lstm_size))
output_gate = fluid.layers.sigmoid(x=gate_common(word, prev_hidden,
lstm_size))
cell_gate = fluid.layers.sigmoid(x=gate_common(word, prev_hidden,
lstm_size))
cell = forget_gate * prev_cell + input_gate * cell_gate
hidden = output_gate * fluid.layers.tanh(x=cell)
rnn.update_memory(prev_cell, cell)
rnn.update_memory(prev_hidden, hidden)
rnn.output(hidden)
last = fluid.layers.sequence_last_step(rnn())
prediction = fluid.layers.fc(input=last, size=class_dim, act="softmax")
return prediction
def inference_program(word_dict):
data = fluid.layers.data(
name="words", shape=[1], dtype="int64", lod_level=1)
dict_dim = len(word_dict)
pred = dynamic_rnn_lstm(data, dict_dim, CLASS_DIM, EMB_DIM, LSTM_SIZE)
return pred
def train_program(word_dict):
prediction = inference_program(word_dict)
label = fluid.layers.data(name="label", shape=[1], dtype="int64")
cost = fluid.layers.cross_entropy(input=prediction, label=label)
avg_cost = fluid.layers.mean(cost)
accuracy = fluid.layers.accuracy(input=prediction, label=label)
return [avg_cost, accuracy]
def optimizer_func():
return fluid.optimizer.Adagrad(learning_rate=0.002)
def train(use_cuda, train_program, params_dirname):
place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
word_dict = paddle.dataset.imdb.word_dict()
trainer = fluid.Trainer(
train_func=partial(train_program, word_dict),
place=place,
optimizer_func=optimizer_func)
def event_handler(event):
if isinstance(event, fluid.EndEpochEvent):
test_reader = paddle.batch(
paddle.dataset.imdb.test(word_dict), batch_size=BATCH_SIZE)
avg_cost, acc = trainer.test(
reader=test_reader, feed_order=['words', 'label'])
print("avg_cost: %s" % avg_cost)
print("acc : %s" % acc)
if acc > 0.2: # Smaller value to increase CI speed
trainer.save_params(params_dirname)
trainer.stop()
else:
print('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc))
if math.isnan(avg_cost):
sys.exit("got NaN loss, training failed.")
elif isinstance(event, fluid.EndStepEvent):
print("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, map(np.array, event.metrics)))
if event.step == 1: # Run 2 iterations to speed CI
trainer.save_params(params_dirname)
trainer.stop()
train_reader = paddle.batch(
paddle.reader.shuffle(
paddle.dataset.imdb.train(word_dict), buf_size=25000),
batch_size=BATCH_SIZE)
trainer.train(
num_epochs=1,
event_handler=event_handler,
reader=train_reader,
feed_order=['words', 'label'])
def infer(use_cuda, inference_program, params_dirname=None):
place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
word_dict = paddle.dataset.imdb.word_dict()
inferencer = fluid.Inferencer(
infer_func=partial(inference_program, word_dict),
param_path=params_dirname,
place=place)
# Setup input by creating LoDTensor to represent sequence of words.
# Here each word is the basic element of the LoDTensor and the shape of
# each word (base_shape) should be [1] since it is simply an index to
# look up for the corresponding word vector.
# Suppose the length_based level of detail (lod) info is set to [[3, 4, 2]],
# which has only one lod level. Then the created LoDTensor will have only
# one higher level structure (sequence of words, or sentence) than the basic
# element (word). Hence the LoDTensor will hold data for three sentences of
# length 3, 4 and 2, respectively.
# Note that lod info should be a list of lists.
lod = [[3, 4, 2]]
base_shape = [1]
# The range of random integers is [low, high]
tensor_words = fluid.create_random_int_lodtensor(
lod, base_shape, place, low=0, high=len(word_dict) - 1)
results = inferencer.infer({'words': tensor_words})
print("infer results: ", results)
def main(use_cuda):
if use_cuda and not fluid.core.is_compiled_with_cuda():
return
params_dirname = "understand_sentiment_conv.inference.model"
train(use_cuda, train_program, params_dirname)
infer(use_cuda, inference_program, params_dirname)
if __name__ == '__main__':
for use_cuda in (False, True):
main(use_cuda=use_cuda)
06.understand_sentiment/train_stacked_lstm.py 0 → 100644
浏览文件 @ 016995e9
# Copyright (c) 2018 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.
from __future__ import print_function
import paddle
import paddle.fluid as fluid
from functools import partial
import numpy as np
CLASS_DIM = 2
EMB_DIM = 128
HID_DIM = 512
STACKED_NUM = 3
BATCH_SIZE = 128
def stacked_lstm_net(data, input_dim, class_dim, emb_dim, hid_dim, stacked_num):
assert stacked_num % 2 == 1
emb = fluid.layers.embedding(
input=data, size=[input_dim, emb_dim], is_sparse=True)
fc1 = fluid.layers.fc(input=emb, size=hid_dim)
lstm1, cell1 = fluid.layers.dynamic_lstm(input=fc1, size=hid_dim)
inputs = [fc1, lstm1]
for i in range(2, stacked_num + 1):
fc = fluid.layers.fc(input=inputs, size=hid_dim)
lstm, cell = fluid.layers.dynamic_lstm(
input=fc, size=hid_dim, is_reverse=(i % 2) == 0)
inputs = [fc, lstm]
fc_last = fluid.layers.sequence_pool(input=inputs[0], pool_type='max')
lstm_last = fluid.layers.sequence_pool(input=inputs[1], pool_type='max')
prediction = fluid.layers.fc(
input=[fc_last, lstm_last], size=class_dim, act='softmax')
return prediction
def inference_program(word_dict):
data = fluid.layers.data(
name="words", shape=[1], dtype="int64", lod_level=1)
dict_dim = len(word_dict)
net = stacked_lstm_net(data, dict_dim, CLASS_DIM, EMB_DIM, HID_DIM,
STACKED_NUM)
return net
def train_program(word_dict):
prediction = inference_program(word_dict)
label = fluid.layers.data(name="label", shape=[1], dtype="int64")
cost = fluid.layers.cross_entropy(input=prediction, label=label)
avg_cost = fluid.layers.mean(cost)
accuracy = fluid.layers.accuracy(input=prediction, label=label)
return [avg_cost, accuracy]
def optimizer_func():
return fluid.optimizer.Adagrad(learning_rate=0.002)
def train(use_cuda, train_program, params_dirname):
place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
word_dict = paddle.dataset.imdb.word_dict()
trainer = fluid.Trainer(
train_func=partial(train_program, word_dict),
place=place,
optimizer_func=optimizer_func)
def event_handler(event):
if isinstance(event, fluid.EndEpochEvent):
test_reader = paddle.batch(
paddle.dataset.imdb.test(word_dict), batch_size=BATCH_SIZE)
avg_cost, acc = trainer.test(
reader=test_reader, feed_order=['words', 'label'])
print("avg_cost: %s" % avg_cost)
print("acc : %s" % acc)
if acc > 0.2: # Smaller value to increase CI speed
trainer.save_params(params_dirname)
trainer.stop()
else:
print('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc))
if math.isnan(avg_cost):
sys.exit("got NaN loss, training failed.")
elif isinstance(event, fluid.EndStepEvent):
print("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, map(np.array, event.metrics)))
if event.step == 1: # Run 2 iterations to speed CI
trainer.save_params(params_dirname)
trainer.stop()
train_reader = paddle.batch(
paddle.reader.shuffle(
paddle.dataset.imdb.train(word_dict), buf_size=25000),
batch_size=BATCH_SIZE)
trainer.train(
num_epochs=1,
event_handler=event_handler,
reader=train_reader,
feed_order=['words', 'label'])
def infer(use_cuda, inference_program, params_dirname=None):
place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
word_dict = paddle.dataset.imdb.word_dict()
inferencer = fluid.Inferencer(
infer_func=partial(inference_program, word_dict),
param_path=params_dirname,
place=place)
# Setup input by creating LoDTensor to represent sequence of words.
# Here each word is the basic element of the LoDTensor and the shape of
# each word (base_shape) should be [1] since it is simply an index to
# look up for the corresponding word vector.
# Suppose the length_based level of detail (lod) info is set to [[3, 4, 2]],
# which has only one lod level. Then the created LoDTensor will have only
# one higher level structure (sequence of words, or sentence) than the basic
# element (word). Hence the LoDTensor will hold data for three sentences of
# length 3, 4 and 2, respectively.
# Note that lod info should be a list of lists.
lod = [[3, 4, 2]]
base_shape = [1]
# The range of random integers is [low, high]
tensor_words = fluid.create_random_int_lodtensor(
lod, base_shape, place, low=0, high=len(word_dict) - 1)
results = inferencer.infer({'words': tensor_words})
print("infer results: ", results)
def main(use_cuda):
if use_cuda and not fluid.core.is_compiled_with_cuda():
return
params_dirname = "understand_sentiment_stacked_lstm.inference.model"
train(use_cuda, train_program, params_dirname)
infer(use_cuda, inference_program, params_dirname)
if __name__ == '__main__':
for use_cuda in (False, True):
main(use_cuda=use_cuda)
Markdown is supported
0% 或 .
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 或