lstm rnn＆＃x27;序列＆＃x27;对象没有属性。
我正在对RNN模型进行预测，并且错误地说：
AttributeError Traceback (most recent call last)
<ipython-input-18-cd3fe876f68d> in <module>
1 # Make sentiment predictions
----> 2 predictions = model.predict_classes(X_test)
AttributeError: 'Sequential' object has no attribute 'predict_classes'
我正在做一个复杂的情感分析模型，该模型将编码/代币注释用于使用LSTM层的深度NN工作，目标是可以将所有这些注释和所有评论提供给RNN在此.CSV文件中预测每个评论中每个评论的积极分数。该模型将输出0表示负面评论或1以供积极评论。
所以我处于最后阶段。我对所有数据进行了预处理，编码/令牌化，并构建了NN模型，现在我正在运行训练。我决定训练几个时期，因为它正在吃笔记本电脑的记忆。
# Training the model
batch_size = 1000
epochs = 10
model.fit(
X_train_rnn,
y_train_rnn,
validation_data = (X_val_rnn, y_val_rnn),
epochs = epochs,
batch_size = batch_size,
verbose = 1
)
在我的计算机训练该型号几分钟后，我尝试做出预测，在此遇到错误的地方：
y_rnn_pred = model.predict_classes(X_test_rnn, batch_size=1000)
I'm running predictions on an RNN model and it errors saying:
AttributeError Traceback (most recent call last)
<ipython-input-18-cd3fe876f68d> in <module>
1 # Make sentiment predictions
----> 2 predictions = model.predict_classes(X_test)
AttributeError: 'Sequential' object has no attribute 'predict_classes'
I'm doing a complicated sentiment analysis model that feeds encoded/tokens comments into a Deep NN work using LSTM Layers, the goal is can a RNN be fed all these comments and predict the positivity score of the each comment in this .csv file of 25,000 comments. The model will output either 0 for negative comment or 1 for positive comment.
So I'm at the final stages. I pre-processed and encoded/tokenized all my data, built the NN model, and now I'm running it to train. I decided to train on a few epochs because it's eating the memory of laptop.
# Training the model
batch_size = 1000
epochs = 10
model.fit(
X_train_rnn,
y_train_rnn,
validation_data = (X_val_rnn, y_val_rnn),
epochs = epochs,
batch_size = batch_size,
verbose = 1
)
After my computer trains this model for minutes on end, then I try to make predictions and this where I get the error:
y_rnn_pred = model.predict_classes(X_test_rnn, batch_size=1000)
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
如果您使用的是TF.KERAS版本2.5+
替换
为
If you are using tf.keras version 2.5+
replace
with