在IMDB情感分类任务上训练双向LSTM

代码如下在这里插入代码片

from __future__ import print_function
import numpy as npfrom keras.preprocessing import sequence
from keras.models import Sequential
from keras.layers import Dense, Dropout, Embedding, LSTM, Bidirectional
from keras.datasets import imdbmax_features &＃61; 20000
# 在此数量的单词之后剪切文本&＃xff08;取最常见的 max_features 个单词&＃xff09;
maxlen &＃61; 100
batch_size &＃61; 32print(&＃39;Loading data...&＃39;)
(x_train, y_train), (x_test, y_test) &＃61; imdb.load_data(num_words&＃61;max_features)
print(len(x_train), &＃39;train sequences&＃39;)
print(len(x_test), &＃39;test sequences&＃39;)print(&＃39;Pad sequences (samples x time)&＃39;)
x_train &＃61; sequence.pad_sequences(x_train, maxlen&＃61;maxlen)
x_test &＃61; sequence.pad_sequences(x_test, maxlen&＃61;maxlen)
print(&＃39;x_train shape:&＃39;, x_train.shape)
print(&＃39;x_test shape:&＃39;, x_test.shape)
y_train &＃61; np.array(y_train)
y_test &＃61; np.array(y_test)model &＃61; Sequential()
model.add(Embedding(max_features, 128, input_length&＃61;maxlen))
model.add(Bidirectional(LSTM(64)))
model.add(Dropout(0.5))
model.add(Dense(1, activation&＃61;&＃39;sigmoid&＃39;))# 尝试使用不同的优化器和优化器配置
model.compile(&＃39;adam&＃39;, &＃39;binary_crossentropy&＃39;, metrics&＃61;[&＃39;accuracy&＃39;])print(&＃39;Train...&＃39;)
model.fit(x_train, y_train,batch_size&＃61;batch_size,epochs&＃61;4,validation_data&＃61;[x_test, y_test])

Output after 4 epochs on CPU: ~0.8146. Time per epoch on CPU (Core i7): ~150s.

在 CPU 上经过 4 个轮次后的输出&＃xff1a;〜0.8146。 CPU&＃xff08;Core i7&＃xff09;上每个轮次的时间&＃xff1a;〜150s。

参考来源:https://keras-zh.readthedocs.io/examples/imdb_bidirectional_lstm/