What is your score using this algorithm

[2g-XzenG]

Could you tell me roughly what score you reach using this method? Thanks

[akshataph]

around 99.2 ...

[2g-XzenG]

Hello, Thanks for answering, I ran both methods (rnn+cnn and cnn), and rnn+cnn is much better than cnn itself (I got around 85 even I increase the epoch to 20). I wonder do you come up with this rnn+cnn yourself? Or do you implement some paper's method (if yes, can you sent me a link). I am really interested in this method.

[davidnguyen2015]

Hi jiegzhan, I start learning about neural network and researching for CNN-RNN for text classification. I found solution about RNN in this post on github, but in this source of your, i can't find memo about CNN. I think this solution for https://www.kaggle.com/c/sf-crime/data is combined of CNN and RNN in tensorflow. Have you help me about CNN model in solution for text classification. You may confirm in this github or send mail to email nguyenthaian.vn@gmail.com. Thank you so much

[gxdalu-yaya]

@1230pitchanqw 我看了一下，基本和a c-lstm neural network for text classification这篇论文的差不多，有几个区别： 1.卷积核用了多个，用padding的形式，让不通卷积核大小的输出是一样的 2.卷积完之后做了个k-max-pooling，论文的没用 3.LSTM的输出，论文里说最后一个节点的输出，这个代码是real_len的输出。

[davidnguyen2015]

Thank you so much.

2017-09-29 9:56 GMT+07:00 gxdalu-yaya notifications@github.com:

@1230pitchanqw https://github.com/1230pitchanqw 我看了一下，基本和a c-lstm neural network for text classification这篇论文的差不多，有几个区别： 1.卷积核用了多个，用padding的形式，让不通卷积核大小的输出是一样的 2.卷积完之后做了个k-max-pooling，论文的没用 3.LSTM的输出，论文里说最后一个节点的输出，这个代码是real_len的输出。

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/jiegzhan/multi-class-text-classification-cnn-rnn/issues/6#issuecomment-333018263, or mute the thread https://github.com/notifications/unsubscribe-auth/AZS1RD0J-3hP85SHOq_2k3jx3blYjRS_ks5snFx5gaJpZM4MHM2U .

[EagleYing]

the environment is python2 or python3?

[553270311]

@1230pitchanqw 我看了一下，基本和a c-lstm neural network for text classification这篇论文的差不多，有几个区别： 1.卷积核用了多个，用padding的形式，让不通卷积核大小的输出是一样的 2.卷积完之后做了个k-max-pooling，论文的没用 3.LSTM的输出，论文里说最后一个节点的输出，这个代码是real_len的输出。

老哥，请教一下