持续创作，加速成长！这是我参与「掘金日新计划 · 10 月更文挑战」的第9天，点击查看活动详情
前言
之前的文章中，描述了TextCNN的模型结构以及模型代码，今天将记录模型训练代码阅读以及模型训练。
train
import datetime
import os
import time
import numpy as np
import tensorflow as tf
from data_helpers import batch_iter
from text_cnn import TextCNN
import pickle
import codecs
writer = codecs.open('dev_res.txt','w','UTF-8')
def pickle_reader(inputs):
f = open(inputs, 'rb')
lines = pickle.load(f)
f.close()
print("Finish load {}".format(inputs))
return lines
def pickle_writer(inputs, name):
output = open(name, 'wb')
pickle.dump(inputs, output, protocol=2)
output.close()
print("Finish save {}".format(name))
tf.flags.DEFINE_string("filter_sizes", "3,4,5", "Comma-separated filter sizes (default: '3,4,5')")
tf.flags.DEFINE_integer("num_filters", 128, "Number of filters per filter size (default: 128)")
tf.flags.DEFINE_float("dropout_keep_prob", 0.5, "Dropout keep probability (default: 0.5)")
tf.flags.DEFINE_float("l2_reg_lambda", 0.0, "L2 regularization lambda (default: 0.0)")
# Training parameters
tf.flags.DEFINE_integer("batch_size", 32, "Batch Size (default: 64)")
tf.flags.DEFINE_integer("num_epochs", 40, "Number of training epochs (default: 200)")
tf.flags.DEFINE_integer("evaluate_every", 2000, "Evaluate model on dev set after this many steps (default: 100)")
tf.flags.DEFINE_integer("checkpoint_every", 1000, "Save model after this many steps (default: 100)")
tf.flags.DEFINE_integer("num_checkpoints", 2, "Number of checkpoints to store (default: 5)")
# Misc Parameters
tf.flags.DEFINE_boolean("allow_soft_placement", True, "Allow device soft device placement")
tf.flags.DEFINE_boolean("log_device_placement", False, "Log placement of ops on devices")
FLAGS = tf.flags.FLAGS
# FLAGS._parse_flags()
print("\nParameters:")
for attr, value in sorted(FLAGS.__flags.items()):
print("{}={}".format(attr.upper(), value))
print("")
# Data Preparation
# ==================================================
# train_data = codecs.open(r'F:\TextCNN_4_URL\data\train_data.txt','r','UTF-8').readlines()
train_data = codecs.open(r'E:\TextCNN\data\VlunData\train.txt','r','UTF-8').readlines()
np.random.shuffle(train_data)
urls = []
labels = []
index = int(0.8 * float(len(train_data)))
max_len = 0
for line in train_data:
url = line.strip().split('<split>')[0].split()
if len(url) > max_len:
max_len = len(url)
label = line.strip().split('<split>')[1]
urls.append(url)
labels.append(label)
print('文本最大长度为 {}'.format(max_len))
vocabulary = {'UNK':0,'PAD':1}
for url in urls:
for word in url:
if word in vocabulary:
pass
else:
vocabulary[word] = len(vocabulary)
pickle_writer(vocabulary,'vocabulary.pkl')
def convert2id(inputs):
res = []
for line in inputs:
ids = [1 for _ in range(max_len)]
for index, word in enumerate(line):
if word in vocabulary:
ids[index] = vocabulary[word]
else:
ids[index] = 0
res.append(ids)
return res
def convert2label(inputs):
res = []
for index in inputs:
if index == '0':
res.append([0,1])
else:
res.append([1,0])
return res
train_url, train_label = convert2id(urls[:index]), convert2label(labels[:index])
dev_url, dev_label = convert2id(urls[index:]), convert2label(labels[index:])
# Training
# ==================================================
with tf.Graph().as_default():
session_conf = tf.ConfigProto(
allow_soft_placement=FLAGS.allow_soft_placement,
log_device_placement=FLAGS.log_device_placement)
sess = tf.Session(config=session_conf)
with sess.as_default():
cnn = TextCNN(sequence_length=max_len,
num_classes=2,
vocab_size=len(vocabulary),
embedding_size=300,
filter_sizes=list(map(int, FLAGS.filter_sizes.split(","))),
num_filters=FLAGS.num_filters,
l2_reg_lambda=FLAGS.l2_reg_lambda)
# Define Training procedure
global_step = tf.Variable(0, name="global_step", trainable=False)
optimizer = tf.train.AdamOptimizer(1e-3)
grads_and_vars = optimizer.compute_gradients(cnn.loss)
train_op = optimizer.apply_gradients(grads_and_vars, global_step=global_step)
# Keep track of gradient values and sparsity (optional)
grad_summaries = []
for g, v in grads_and_vars:
if g is not None:
grad_hist_summary = tf.summary.histogram("{}/grad/hist".format(v.name), g)
sparsity_summary = tf.summary.scalar("{}/grad/sparsity".format(v.name), tf.nn.zero_fraction(g))
grad_summaries.append(grad_hist_summary)
grad_summaries.append(sparsity_summary)
grad_summaries_merged = tf.summary.merge(grad_summaries)
# Output directory for models and summaries
timestamp = str(int(time.time()))
out_dir = os.path.abspath(os.path.join(os.path.curdir, "runs", timestamp))
print("Writing to {}\n".format(out_dir))
# Summaries for loss and accuracy
loss_summary = tf.summary.scalar("loss", cnn.loss)
acc_summary = tf.summary.scalar("accuracy", cnn.accuracy)
# Train Summaries
train_summary_op = tf.summary.merge([loss_summary, acc_summary, grad_summaries_merged])
train_summary_dir = os.path.join(out_dir, "summaries", "train")
train_summary_writer = tf.summary.FileWriter(train_summary_dir, sess.graph)
# Dev summaries
dev_summary_op = tf.summary.merge([loss_summary, acc_summary])
dev_summary_dir = os.path.join(out_dir, "summaries", "dev")
dev_summary_writer = tf.summary.FileWriter(dev_summary_dir, sess.graph)
# Checkpoint directory. Tensorflow assumes this directory already exists so we need to create it
checkpoint_dir = os.path.abspath(os.path.join(out_dir, "checkpoints"))
checkpoint_prefix = os.path.join(checkpoint_dir, "model")
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
saver = tf.train.Saver(tf.global_variables(), max_to_keep=FLAGS.num_checkpoints)
# Write vocabulary
# Initialize all variables
sess.run(tf.global_variables_initializer())
def train_step(x_batch, y_batch):
"""
A single training step
"""
feed_dict = {
cnn.input_x: x_batch,
cnn.input_y: y_batch,
cnn.dropout_keep_prob: FLAGS.dropout_keep_prob
}
_, step, summaries, loss, accuracy = sess.run(
[train_op, global_step, train_summary_op, cnn.loss, cnn.accuracy],
feed_dict)
time_str = datetime.datetime.now().isoformat()
print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
train_summary_writer.add_summary(summaries, step)
def dev_step(x_batch, y_batch, writer=None):
"""
Evaluates model on a dev set
"""
feed_dict = {
cnn.input_x: x_batch,
cnn.input_y: y_batch,
cnn.dropout_keep_prob: 1.0
}
step, summaries, loss, accuracy = sess.run(
[global_step, dev_summary_op, cnn.loss, cnn.accuracy],
feed_dict)
time_str = datetime.datetime.now().isoformat()
if writer:
writer.add_summary(summaries, step)
return loss,accuracy
# Generate batches
batches = batch_iter(
list(zip(train_url, train_label)), FLAGS.batch_size, FLAGS.num_epochs)
# Training loop. For each batch...
for batch in batches:
x_batch, y_batch = zip(*batch)
train_step(x_batch, y_batch)
current_step = tf.train.global_step(sess, global_step)
if current_step % FLAGS.evaluate_every == 0:
print("\nEvaluation:")
t_cost = 0.0
t_acc = 0.0
count = 0
dev_batches = batch_iter(list(zip(dev_url, dev_label)), FLAGS.batch_size, 1)
for dev_batch in dev_batches:
x_dev_batch, y_dev_batch = zip(*dev_batch)
loss, accuracy = dev_step(x_dev_batch, y_dev_batch, writer=dev_summary_writer)
t_cost += loss
t_acc += accuracy
count += 1
print("loss {:g}, acc {:g}".format(t_cost/count, t_acc/count)+'\n')
writer.write("loss {:g}, acc {:g}".format(t_cost/count, t_acc/count)+'\n')
writer.flush()
if current_step % FLAGS.checkpoint_every == 0:
path = saver.save(sess, checkpoint_prefix, global_step=current_step)
print("Saved model checkpoint to {}\n".format(path))
参数解读：
- dev_sample_percentage 测试集比例
- positive_data_file 正样本文件
- negative_data_file 负样本文件
- embedding_dim 词向量维度
- filter_sizes filter维度
- num_filters filter个数
- dropout_keep_prob 随机失活比例
- l2_reg_lambda l2正则化系数
- batch_size batch 批次大小
- num_epochs 训练轮数
- evaluate_every 每隔多少step进行测试
- checkpoint_every 每隔多少step保存一个checkpoint
- num_checkpoints 一共保存多少个checkpoint
- allow_soft_placement 允许动态分配资源
- log_device_placement=True，让我们可以看到我们的tensor、op是在哪台设备、哪颗CPU上运行
train_data = codecs.open(r'E:\TextCNN\data\VlunData\train.txt','r','UTF-8').readlines()
np.random.shuffle(train_data)
urls = []
labels = []
index = int(0.8 * float(len(train_data)))
max_len = 0
for line in train_data:
url = line.strip().split('<split>')[0].split()
if len(url) > max_len:
max_len = len(url)
label = line.strip().split('<split>')[1]
urls.append(url)
labels.append(label)
print('文本最大长度为 {}'.format(max_len))
vocabulary = {'UNK':0,'PAD':1}
for url in urls:
for word in url:
if word in vocabulary:
pass
else:
vocabulary[word] = len(vocabulary)
pickle_writer(vocabulary,'vocabulary.pkl')
train_data 加载训练数据，即dataPreprocessing数据处理之后的训练数据，格式如下：
apple os x 是 一款 苹果 分发 的 基于 bsd 的 操作系统 。<split>0
cpanel 是 美国 cpanel 公司 的 一套 基于 web 的 自动化 主机 托管 平台 。 该 平台 主要 用于 自动化 管理 网站 和 服务器 。<split>0
mozilla firefox 是 美国 mozilla 基金会 的 一款 开源 web 浏览器 。<split>1
iscripts autohoster 是 一款 基于 php 的 web 应用程序 。<split>0
dataease v1.11 . 1 was discovered to contain a sql injection vulnerability via the parameter datasourceid .<split>0
d - bus 是 一款 进程 间通信 ( ipc ) 实现 。 用于 在 应用程序 间 发送 消息 。<split>1
chromium - browser 是 一个 开放源码 的 web 浏览器 项目 ， 由谷歌 启动 ， 为 专有 的 谷歌 浏览器 浏览器 提供 源代码 。<split>1
在 固件 版本 为 1012 的 insteon hub 2245 - 222 设备 上 ， 从 pubnub 服务 收到 的 特别 精心制作 的 回复 可能 会 在 覆盖 任意 数据 的 全局 区域 上 导致 缓冲区 溢出 。 攻击者 应该 模拟 pubnub 并 响应 https get 请求 来 触发 此 漏洞 。 strcpy 溢出 了 insteon _ pubnub 缓冲区 。 channel _ ad _ r ， 它 的 大小 为 16 字节 。 攻击者 可以 发送 任意 长 的 “ ad _ r ” 参数 来 利用 这个 漏洞 。<split>1
mattermost server 是 美国 mattermost 公司 的 一套 开源 的 消息传递 平台 。<split>0
malwarebytes premium 是 美国 malwarebytes 公司 的 一套 反 恶意 间谍 软件 。 该软件 支持 删除 蠕虫 、 拨号 程序 、 木马 、 rootkit 、 间谍 软件 、 漏洞 、 僵尸 和 其他 恶意软件 等 。<split>0
np.random.shuffle(train_data)，对训练数据进展随机打乱，
index = int(0.8 * float(len(train_data)))： 用于切分训练测试集
max_len = 0
for line in train_data:
url = line.strip().split('<split>')[0].split()
if len(url) > max_len:
max_len = len(url)
label = line.strip().split('<split>')[1]
urls.append(url)
labels.append(label)
获取文本最大长度，以及将文本与标签分别存于不同的list，用于模型训练。
print('文本最大长度为 {}'.format(max_len))
vocabulary = {'UNK':0,'PAD':1}
for url in urls:
for word in url:
if word in vocabulary:
pass
else:
vocabulary[word] = len(vocabulary)
pickle_writer(vocabulary,'vocabulary.pkl')
def convert2id(inputs):
res = []
for line in inputs:
ids = [1 for _ in range(max_len)]
for index, word in enumerate(line):
if word in vocabulary:
ids[index] = vocabulary[word]
else:
ids[index] = 0
res.append(ids)
return res
def convert2label(inputs):
res = []
for index in inputs:
if index == '0':
res.append([0,1])
else:
res.append([1,0])
return res
生成词典，用于测试的时候使用，并且将token_list和label_list转为对应的index，作为模型input。 词典方面，增加pad和unk字符，分别为补长字符和未登录词
train_url, train_label = convert2id(urls[:index]), convert2label(labels[:index])
dev_url, dev_label = convert2id(urls[index:]), convert2label(labels[index:])
切分训练集和测试机，index为上面生成的int类型。
with tf.Graph().as_default():
session_conf = tf.ConfigProto(
allow_soft_placement=FLAGS.allow_soft_placement,
log_device_placement=FLAGS.log_device_placement)
sess = tf.Session(config=session_conf)
with sess.as_default():
cnn = TextCNN(sequence_length=max_len,
num_classes=2,
vocab_size=len(vocabulary),
embedding_size=300,
filter_sizes=list(map(int, FLAGS.filter_sizes.split(","))),
num_filters=FLAGS.num_filters,
l2_reg_lambda=FLAGS.l2_reg_lambda)
# Define Training procedure
global_step = tf.Variable(0, name="global_step", trainable=False)
optimizer = tf.train.AdamOptimizer(1e-3)
grads_and_vars = optimizer.compute_gradients(cnn.loss)
train_op = optimizer.apply_gradients(grads_and_vars, global_step=global_step)
# Keep track of gradient values and sparsity (optional)
grad_summaries = []
for g, v in grads_and_vars:
if g is not None:
grad_hist_summary = tf.summary.histogram("{}/grad/hist".format(v.name), g)
sparsity_summary = tf.summary.scalar("{}/grad/sparsity".format(v.name), tf.nn.zero_fraction(g))
grad_summaries.append(grad_hist_summary)
grad_summaries.append(sparsity_summary)
grad_summaries_merged = tf.summary.merge(grad_summaries)
# Output directory for models and summaries
timestamp = str(int(time.time()))
out_dir = os.path.abspath(os.path.join(os.path.curdir, "runs", timestamp))
print("Writing to {}\n".format(out_dir))
# Summaries for loss and accuracy
loss_summary = tf.summary.scalar("loss", cnn.loss)
acc_summary = tf.summary.scalar("accuracy", cnn.accuracy)
# Train Summaries
train_summary_op = tf.summary.merge([loss_summary, acc_summary, grad_summaries_merged])
train_summary_dir = os.path.join(out_dir, "summaries", "train")
train_summary_writer = tf.summary.FileWriter(train_summary_dir, sess.graph)
# Dev summaries
dev_summary_op = tf.summary.merge([loss_summary, acc_summary])
dev_summary_dir = os.path.join(out_dir, "summaries", "dev")
dev_summary_writer = tf.summary.FileWriter(dev_summary_dir, sess.graph)
# Checkpoint directory. Tensorflow assumes this directory already exists so we need to create it
checkpoint_dir = os.path.abspath(os.path.join(out_dir, "checkpoints"))
checkpoint_prefix = os.path.join(checkpoint_dir, "model")
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
saver = tf.train.Saver(tf.global_variables(), max_to_keep=FLAGS.num_checkpoints)
# Write vocabulary
# Initialize all variables
sess.run(tf.global_variables_initializer())
初始化模型，增加summary，用于记录模型训练过程。
def train_step(x_batch, y_batch):
"""
A single training step
"""
feed_dict = {
cnn.input_x: x_batch,
cnn.input_y: y_batch,
cnn.dropout_keep_prob: FLAGS.dropout_keep_prob
}
_, step, summaries, loss, accuracy = sess.run(
[train_op, global_step, train_summary_op, cnn.loss, cnn.accuracy],
feed_dict)
time_str = datetime.datetime.now().isoformat()
print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
train_summary_writer.add_summary(summaries, step)
def dev_step(x_batch, y_batch, writer=None):
"""
Evaluates model on a dev set
"""
feed_dict = {
cnn.input_x: x_batch,
cnn.input_y: y_batch,
cnn.dropout_keep_prob: 1.0
}
step, summaries, loss, accuracy = sess.run(
[global_step, dev_summary_op, cnn.loss, cnn.accuracy],
feed_dict)
time_str = datetime.datetime.now().isoformat()
if writer:
writer.add_summary(summaries, step)
return loss,accuracy
# Generate batches
batches = batch_iter(
list(zip(train_url, train_label)), FLAGS.batch_size, FLAGS.num_epochs)
# Training loop. For each batch...
for batch in batches:
x_batch, y_batch = zip(*batch)
train_step(x_batch, y_batch)
current_step = tf.train.global_step(sess, global_step)
if current_step % FLAGS.evaluate_every == 0:
print("\nEvaluation:")
t_cost = 0.0
t_acc = 0.0
count = 0
dev_batches = batch_iter(list(zip(dev_url, dev_label)), FLAGS.batch_size, 1)
for dev_batch in dev_batches:
x_dev_batch, y_dev_batch = zip(*dev_batch)
loss, accuracy = dev_step(x_dev_batch, y_dev_batch, writer=dev_summary_writer)
t_cost += loss
t_acc += accuracy
count += 1
print("loss {:g}, acc {:g}".format(t_cost/count, t_acc/count)+'\n')
writer.write("loss {:g}, acc {:g}".format(t_cost/count, t_acc/count)+'\n')
writer.flush()
if current_step % FLAGS.checkpoint_every == 0:
path = saver.save(sess, checkpoint_prefix, global_step=current_step)
print("Saved model checkpoint to {}\n".format(path))
模型的训练过程feed_dict部分，值得一提的是，程序之前的代码是将测试集全部送入模型，会导致OOM，因此也是用batch的方法，计算累计loss和acc，在计算平均值。
训练过程
由于时间问题，训练到近拟合就停止了，模型基本收敛
最终训练过程准确率0.91，测试集准确率0.88，作为任务的baseline。
模型的计算graph如下：
模型全部代码如下：
dataPreprocessing.py
import codecs
import jieba
from tqdm import tqdm
writer = codecs.open("train.txt",'w','UTF-8')
lines = codecs.open("aliyunSpider.txt","r",'UTF-8').readlines()
lines = list(set(lines))
print(len(lines))
count0 = 0
count1 = 0
for line in tqdm(lines):
try:
content = eval(line.strip())["content"].strip()
if len(content) > 20 and count0 <= 50000:
if eval(line.strip())["type"] == "系统":
count0 += 1
print(count0)
writer.write("{}<split>1".format(' '.join(list(jieba.cut(content)))))
writer.write("\n")
elif eval(line.strip())["type"] == "应用" and count1 <= 50000:
count1 += 1
print(count1)
writer.write("{}<split>0".format(' '.join(list(jieba.cut(content)))))
writer.write("\n")
except Exception as e:
pass
text_cnn.py
import codecs
import jieba
from tqdm import tqdm
writer = codecs.open("train.txt",'w','UTF-8')
lines = codecs.open("aliyunSpider.txt","r",'UTF-8').readlines()
lines = list(set(lines))
print(len(lines))
count0 = 0
count1 = 0
for line in tqdm(lines):
try:
content = eval(line.strip())["content"].strip()
if len(content) > 20 and count0 <= 50000:
if eval(line.strip())["type"] == "系统":
count0 += 1
print(count0)
writer.write("{}<split>1".format(' '.join(list(jieba.cut(content)))))
writer.write("\n")
elif eval(line.strip())["type"] == "应用" and count1 <= 50000:
count1 += 1
print(count1)
writer.write("{}<split>0".format(' '.join(list(jieba.cut(content)))))
writer.write("\n")
except Exception as e:
pass
data_helpers.py
import numpy as np
import re
import itertools
from collections import Counter
def clean_str(string):
"""
Tokenization/string cleaning for all datasets except for SST.
Original taken from https://github.com/yoonkim/CNN_sentence/blob/master/process_data.py
"""
string = re.sub(r"[^A-Za-z0-9(),!?'`]", " ", string)
string = re.sub(r"'s", " 's", string)
string = re.sub(r"'ve", " 've", string)
string = re.sub(r"n't", " n't", string)
string = re.sub(r"'re", " 're", string)
string = re.sub(r"'d", " 'd", string)
string = re.sub(r"'ll", " 'll", string)
string = re.sub(r",", " , ", string)
string = re.sub(r"!", " ! ", string)
string = re.sub(r"(", " ( ", string)
string = re.sub(r")", " ) ", string)
string = re.sub(r"?", " ? ", string)
string = re.sub(r"\s{2,}", " ", string)
return string.strip().lower()
def load_data_and_labels(positive_data_file, negative_data_file):
"""
Loads MR polarity data from files, splits the data into words and generates labels.
Returns split sentences and labels.
"""
# Load data from files
positive_examples = list(open(positive_data_file, "r",encoding='UTF-8').readlines()[:60000])
positive_examples = [s.strip() for s in positive_examples]
negative_examples = list(open(negative_data_file, "r",encoding='utf-8').readlines()[:60000])
negative_examples = [s.strip() for s in negative_examples]
# Split by words
x_text = positive_examples + negative_examples
x_text = [clean_str(sent) for sent in x_text]
# Generate labels
positive_labels = [[0, 1] for _ in positive_examples]
negative_labels = [[1, 0] for _ in negative_examples]
y = np.concatenate([positive_labels, negative_labels], 0)
return [x_text, y]
def batch_iter(data, batch_size, num_epochs, shuffle=True):
"""
Generates a batch iterator for a dataset.
"""
data = np.array(data)
data_size = len(data)
num_batches_per_epoch = int((len(data)-1)/batch_size) + 1
for epoch in range(num_epochs):
# Shuffle the data at each epoch
if shuffle:
shuffle_indices = np.random.permutation(np.arange(data_size))
shuffled_data = data[shuffle_indices]
else:
shuffled_data = data
for batch_num in range(num_batches_per_epoch):
start_index = batch_num * batch_size
end_index = min((batch_num + 1) * batch_size, data_size)
yield shuffled_data[start_index:end_index]
train.py
import datetime
import os
import time
import numpy as np
import tensorflow as tf
from data_helpers import batch_iter
from text_cnn import TextCNN
import pickle
import codecs
writer = codecs.open('dev_res.txt','w','UTF-8')
def pickle_reader(inputs):
f = open(inputs, 'rb')
lines = pickle.load(f)
f.close()
print("Finish load {}".format(inputs))
return lines
def pickle_writer(inputs, name):
output = open(name, 'wb')
pickle.dump(inputs, output, protocol=2)
output.close()
print("Finish save {}".format(name))
tf.flags.DEFINE_string("filter_sizes", "3,4,5", "Comma-separated filter sizes (default: '3,4,5')")
tf.flags.DEFINE_integer("num_filters", 128, "Number of filters per filter size (default: 128)")
tf.flags.DEFINE_float("dropout_keep_prob", 0.5, "Dropout keep probability (default: 0.5)")
tf.flags.DEFINE_float("l2_reg_lambda", 0.0, "L2 regularization lambda (default: 0.0)")
# Training parameters
tf.flags.DEFINE_integer("batch_size", 32, "Batch Size (default: 64)")
tf.flags.DEFINE_integer("num_epochs", 40, "Number of training epochs (default: 200)")
tf.flags.DEFINE_integer("evaluate_every", 2000, "Evaluate model on dev set after this many steps (default: 100)")
tf.flags.DEFINE_integer("checkpoint_every", 1000, "Save model after this many steps (default: 100)")
tf.flags.DEFINE_integer("num_checkpoints", 2, "Number of checkpoints to store (default: 5)")
# Misc Parameters
tf.flags.DEFINE_boolean("allow_soft_placement", True, "Allow device soft device placement")
tf.flags.DEFINE_boolean("log_device_placement", False, "Log placement of ops on devices")
FLAGS = tf.flags.FLAGS
# FLAGS._parse_flags()
print("\nParameters:")
for attr, value in sorted(FLAGS.__flags.items()):
print("{}={}".format(attr.upper(), value))
print("")
# Data Preparation
# ==================================================
train_data = codecs.open(r'E:\TextCNN\data\VlunData\train.txt','r','UTF-8').readlines()
np.random.shuffle(train_data)
urls = []
labels = []
index = int(0.8 * float(len(train_data)))
max_len = 0
for line in train_data:
url = line.strip().split('<split>')[0].split()
if len(url) > max_len:
max_len = len(url)
label = line.strip().split('<split>')[1]
urls.append(url)
labels.append(label)
print('文本最大长度为 {}'.format(max_len))
vocabulary = {'UNK':0,'PAD':1}
for url in urls:
for word in url:
if word in vocabulary:
pass
else:
vocabulary[word] = len(vocabulary)
pickle_writer(vocabulary,'vocabulary.pkl')
def convert2id(inputs):
res = []
for line in inputs:
ids = [1 for _ in range(max_len)]
for index, word in enumerate(line):
if word in vocabulary:
ids[index] = vocabulary[word]
else:
ids[index] = 0
res.append(ids)
return res
def convert2label(inputs):
res = []
for index in inputs:
if index == '0':
res.append([0,1])
else:
res.append([1,0])
return res
train_url, train_label = convert2id(urls[:index]), convert2label(labels[:index])
dev_url, dev_label = convert2id(urls[index:]), convert2label(labels[index:])
# Training
# ==================================================
with tf.Graph().as_default():
session_conf = tf.ConfigProto(
allow_soft_placement=FLAGS.allow_soft_placement,
log_device_placement=FLAGS.log_device_placement)
sess = tf.Session(config=session_conf)
with sess.as_default():
cnn = TextCNN(sequence_length=max_len,
num_classes=2,
vocab_size=len(vocabulary),
embedding_size=300,
filter_sizes=list(map(int, FLAGS.filter_sizes.split(","))),
num_filters=FLAGS.num_filters,
l2_reg_lambda=FLAGS.l2_reg_lambda)
# Define Training procedure
global_step = tf.Variable(0, name="global_step", trainable=False)
optimizer = tf.train.AdamOptimizer(1e-3)
grads_and_vars = optimizer.compute_gradients(cnn.loss)
train_op = optimizer.apply_gradients(grads_and_vars, global_step=global_step)
# Keep track of gradient values and sparsity (optional)
grad_summaries = []
for g, v in grads_and_vars:
if g is not None:
grad_hist_summary = tf.summary.histogram("{}/grad/hist".format(v.name), g)
sparsity_summary = tf.summary.scalar("{}/grad/sparsity".format(v.name), tf.nn.zero_fraction(g))
grad_summaries.append(grad_hist_summary)
grad_summaries.append(sparsity_summary)
grad_summaries_merged = tf.summary.merge(grad_summaries)
# Output directory for models and summaries
timestamp = str(int(time.time()))
out_dir = os.path.abspath(os.path.join(os.path.curdir, "runs", timestamp))
print("Writing to {}\n".format(out_dir))
# Summaries for loss and accuracy
loss_summary = tf.summary.scalar("loss", cnn.loss)
acc_summary = tf.summary.scalar("accuracy", cnn.accuracy)
# Train Summaries
train_summary_op = tf.summary.merge([loss_summary, acc_summary, grad_summaries_merged])
train_summary_dir = os.path.join(out_dir, "summaries", "train")
train_summary_writer = tf.summary.FileWriter(train_summary_dir, sess.graph)
# Dev summaries
dev_summary_op = tf.summary.merge([loss_summary, acc_summary])
dev_summary_dir = os.path.join(out_dir, "summaries", "dev")
dev_summary_writer = tf.summary.FileWriter(dev_summary_dir, sess.graph)
# Checkpoint directory. Tensorflow assumes this directory already exists so we need to create it
checkpoint_dir = os.path.abspath(os.path.join(out_dir, "checkpoints"))
checkpoint_prefix = os.path.join(checkpoint_dir, "model")
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
saver = tf.train.Saver(tf.global_variables(), max_to_keep=FLAGS.num_checkpoints)
# Write vocabulary
# Initialize all variables
sess.run(tf.global_variables_initializer())
def train_step(x_batch, y_batch):
"""
A single training step
"""
feed_dict = {
cnn.input_x: x_batch,
cnn.input_y: y_batch,
cnn.dropout_keep_prob: FLAGS.dropout_keep_prob
}
_, step, summaries, loss, accuracy = sess.run(
[train_op, global_step, train_summary_op, cnn.loss, cnn.accuracy],
feed_dict)
time_str = datetime.datetime.now().isoformat()
print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
train_summary_writer.add_summary(summaries, step)
def dev_step(x_batch, y_batch, writer=None):
"""
Evaluates model on a dev set
"""
feed_dict = {
cnn.input_x: x_batch,
cnn.input_y: y_batch,
cnn.dropout_keep_prob: 1.0
}
step, summaries, loss, accuracy = sess.run(
[global_step, dev_summary_op, cnn.loss, cnn.accuracy],
feed_dict)
time_str = datetime.datetime.now().isoformat()
if writer:
writer.add_summary(summaries, step)
return loss,accuracy
# Generate batches
batches = batch_iter(
list(zip(train_url, train_label)), FLAGS.batch_size, FLAGS.num_epochs)
# Training loop. For each batch...
for batch in batches:
x_batch, y_batch = zip(*batch)
train_step(x_batch, y_batch)
current_step = tf.train.global_step(sess, global_step)
if current_step % FLAGS.evaluate_every == 0:
print("\nEvaluation:")
t_cost = 0.0
t_acc = 0.0
count = 0
dev_batches = batch_iter(list(zip(dev_url, dev_label)), FLAGS.batch_size, 1)
for dev_batch in dev_batches:
x_dev_batch, y_dev_batch = zip(*dev_batch)
loss, accuracy = dev_step(x_dev_batch, y_dev_batch, writer=dev_summary_writer)
t_cost += loss
t_acc += accuracy
count += 1
print("loss {:g}, acc {:g}".format(t_cost/count, t_acc/count)+'\n')
writer.write("loss {:g}, acc {:g}".format(t_cost/count, t_acc/count)+'\n')
writer.flush()
if current_step % FLAGS.checkpoint_every == 0:
path = saver.save(sess, checkpoint_prefix, global_step=current_step)
print("Saved model checkpoint to {}\n".format(path))
`;-. ___,
`.`\_...._/`.-"`
\ / ,
/() () \ .' `-._
|) . ()\ / _.'
\ -'- ,; '. <
;.__ ,;| > \
/ , / , |.-'.-'
(_/ (_/ ,;|.<`
\ , ;-`
> \ /
(_,-'`> .'
(_,'
蟹蟹~
