label size is to big, out of memory.

In the implementation, the final outputs of Bi-RNN are calculated as the reduce mean among all time stamps. Compared with output_rnn_last=output_rnn[:,-1,:], what is the difference between these two strategies on the impact of the final classification results?

when i run the p8_TextRNN_train.py, it does show the error:
No Such file or dictionary: 'zhihu-word2vect.bin-100'
where can i find the file?

it seems that the way you implement of attention mechanism is different from original paper, can you give more ideas?

不好意思，读了你的HAN_model.py代码感觉你的代码不太完整，缺少了textRNN.accuracy, textRNN.predictions, textRNN.W_projection这些部分。而且textRNN.input_y:没有定义。还有Attention求权重的方法好像和论文原著不太一样，论文中好像接入了个softmax在和隐藏层相乘累加。
请问能大概介绍一下你文章的思路吗？有点云里雾里的。对word级别的为什么要写成每篇文章的第一句，每篇文章的第二句这样循环输入呢？最后的Loss是什么意思？

where is hierAtten_word_voabulary.pik?

File "/home/wt/桌面/class/text_classification-master/a02_TextCNN/other_experiement/data_util_zhihu.py", line 28, in create_voabulary
model=word2vec.load(word2vec_model_path,kind='bin')
File "/usr/local/lib/python3.5/dist-packages/word2vec/io.py", line 18, in load
return word2vec.WordVectors.from_binary(fname, *args, **kwargs)
File "/usr/local/lib/python3.5/dist-packages/word2vec/wordvectors.py", line 185, in from_binary
with open(fname, 'rb') as fin:
FileNotFoundError: [Errno 2] No such file or directory: 'zhihu-word2vec.bin-100'

I try to run p7_TextCNN_train.py in a02_TextCNN, and I've downloaded the zhihu-word2vec-title-desc.bin-100 and put it in the same place as p7_TextCNN_train.py, then the error occurs: [Errno 2] No such file or directory: 'cache_vocabulary_label_pik/cnn2_word_voabulary.pik'
I've noticed that someone met almost the same problem as me, but I don't understand now that cnn2_word_voabulary.pik does not exist, how can the program utilize it? I can't figure out how to debug it. Should I update somewhere in the program?

flake8 testing of https://github.com/brightmart/text_classification on Python 2.7.13

$ flake8 . --count --select=E901,E999,F821,F822,F823 --show-source --statistics

./a07_Transformer/a2_predict.py:102:55: F821 undefined name 'predict'
            print("===============>",start,"predict:",predict)
                                                      ^

if not elsewhere reported, when you run the "python -u p7_TextCNN_train.py", a sample data is missing:

FileNotFoundError: [Errno 2] No such file or directory: '../data/train_label_single100_merge.txt' - and as noted in code

By the way, in which way the sample data is expected to be ? I am speaking of a normal text that needs to be classified. How long should each of the lines in the sample data be and what are they allowed to contain / not contain? And how they are pre-processed for the classification ?

hi,
I can not find the module named 'p4_zhihu_load_data', where is it?thanks.

I'm guessing that this code is intended to be open sourced otherwise it wouldn't be here, but is there a particular license that you'd like to choose for your work?

in a01_FastText/p5_fastTextB_train.py , p4_zhihu_load_data is imported, but this file is not appended.

FileNotFoundError: [Errno 2] No such file or directory: '../zhihu-word2vec.bin-100'

I can't find information about 'test-zhihu-forpredict-v4only-title.txt' from all closed issue, can you provide the url for it, or some example data for me? @brightmart

当我运行 p5_fastTextB_train.py时，报错ImportError: No module named p4_zhihu_load_data
该怎么办呢？

where to find zhihu-word2vec-title-desc.bin-100?

ModuleNotFoundError Traceback (most recent call last)
in ()
8 import numpy as np
9 from p5_fastTextB_model import fastTextB as fastText
---> 10 from p4_zhihu_load_data import load_data,create_voabulary,create_voabulary_label
11 from tflearn.data_utils import to_categorical, pad_sequences
12 import os

ModuleNotFoundError: No module named 'p4_zhihu_load_data'

Hello! I am new to tensorflow and when I run your model TextCNN, I get a issue, that is, sess.run() blocks.
I can only get the print before the code: "curr_loss,curr_acc,_=sess.run([textCNN.loss_val,textCNN.accuracy,textCNN.train_op],feed_dict=feed_dict)" and then , the program blocks! I already make sure the input data exists and I fail to figure it out.
Hope you can give me the answer, thanks for your patience.

when I run a01_FastText/p5_fastTextB_train.py, but I get a error "can not find p4_zhihu_load_data library".

Then, I modify the code as
"
sys.path.append('../aa1_data_util')
from data_util_zhihu import load_data,create_voabulary,create_voabulary_label
"

but I get:

cache_path: cache_vocabulary_label_pik/_word_voabulary.pik file_exists: False
create vocabulary. word2vec_model_path: zhihu-word2vec-title-desc.bin-100
Traceback (most recent call last):
File "p5_fastTextB_train.py", line 159, in
tf.app.run()
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/platform/app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "p5_fastTextB_train.py", line 42, in main
vocabulary_word2index, vocabulary_index2word = create_voabulary()
File "../aa1_data_util/data_util_zhihu.py", line 40, in create_voabulary
with open(cache_path, 'a') as data_f:
FileNotFoundError: [Errno 2] No such file or directory: 'cache_vocabulary_label_pik/_word_voabulary.pik'

So where is the _word_voabulary.pik file? how can I get it?

hdf5 is not supported on this machine (please install/reinstall h5py for optimal experience)
('cache_path:', 'cache_vocabulary_label_pik/_word_voabulary.pik', 'file_exists:', False)
('create vocabulary. word2vec_model_path:', 'zhihu-word2vec-title-desc.bin-100')
Traceback (most recent call last):
  File "p5_fastTextB_train.py", line 161, in <module>
    tf.app.run()
  File "/usr/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 48, in run
    _sys.exit(main(_sys.argv[:1] + flags_passthrough))
  File "p5_fastTextB_train.py", line 44, in main
    vocabulary_word2index, vocabulary_index2word = create_voabulary()
  File "../aa1_data_util/data_util_zhihu.py", line 26, in create_voabulary
    model=word2vec.load(word2vec_model_path,kind='bin')
  File "/usr/lib64/python2.7/site-packages/word2vec/io.py", line 18, in load
    return word2vec.WordVectors.from_binary(fname, *args, **kwargs)
  File "/usr/lib64/python2.7/site-packages/word2vec/wordvectors.py", line 154, in from_binary
    with open(fname, 'rb') as fin:
IOError: [Errno 2] No such file or directory: 'zhihu-word2vec-title-desc.bin-100'

please help.

rt，请问就这个场景，怎么构造的query

As discussed in #6, these four lines (or something similar) need to be added to the following files for them to work in Python 3.

./a8_predict.py:5:1: F821 undefined name 'reload'
./a8_train.py:5:1: F821 undefined name 'reload'
./a01_FastText/p5_fastTextB_predict.py:5:1: F821 undefined name 'reload'
./a01_FastText/p5_fastTextB_predict_multilabel.py:5:1: F821 undefined name 'reload'
./a01_FastText/p5_fastTextB_train.py:5:1: F821 undefined name 'reload'
./a02_TextCNN/p7_TextCNN_predict.py:5:1: F821 undefined name 'reload'
./a02_TextCNN/p7_TextCNN_train.py:5:1: F821 undefined name 'reload'
./a02_TextCNN/other_experiement/p7_TextCNN_predict_exp.py:5:1: F821 undefined name 'reload'
./a02_TextCNN/other_experiement/p7_TextCNN_predict_exp512.py:5:1: F821 undefined name 'reload'
./a02_TextCNN/other_experiement/p7_TextCNN_predict_exp512_0609.py:5:1: F821 undefined name 'reload'
./a02_TextCNN/other_experiement/p7_TextCNN_predict_exp512_simple.py:5:1: F821 undefined name 'reload'
./a02_TextCNN/other_experiement/p7_TextCNN_train_exp.py:5:1: F821 undefined name 'reload'
./a02_TextCNN/other_experiement/p7_TextCNN_train_exp512.py:5:1: F821 undefined name 'reload'
./a02_TextCNN/other_experiement/p7_TextCNN_train_exp_512_0609.py:5:1: F821 undefined name 'reload'
./a02_TextCNN/other_experiement/p8_TextCNN_predict_exp.py:5:1: F821 undefined name 'reload'
./a03_TextRNN/p8_TextRNN_predict.py:5:1: F821 undefined name 'reload'
./a03_TextRNN/p8_TextRNN_train.py:5:1: F821 undefined name 'reload'
./a04_TextRCNN/p71_TextRCNN_predict.py:5:1: F821 undefined name 'reload'
./a04_TextRCNN/p71_TextRCNN_train.py:5:1: F821 undefined name 'reload'
./a05_HierarchicalAttentionNetwork/p1_HierarchicalAttention_predict.py:5:1: F821 undefined name 'reload'
./a05_HierarchicalAttentionNetwork/p1_HierarchicalAttention_train.py:5:1: F821 undefined name 'reload'
./a06_Seq2seqWithAttention/a1_seq2seq_attention_predict.py:5:1: F821 undefined name 'reload'
./a06_Seq2seqWithAttention/a1_seq2seq_attention_train.py:5:1: F821 undefined name 'reload'
./a07_Transformer/a2_predict.py:5:1: F821 undefined name 'reload'
./a07_Transformer/a2_train.py:5:1: F821 undefined name 'reload'
./a08_EntityNetwork/a3_predict.py:5:1: F821 undefined name 'reload'
./a08_EntityNetwork/a3_train.py:5:1: F821 undefined name 'reload'
./aa1_data_util/2_predict_zhihu_get_question_representation.py:3:1: F821 undefined name 'reload'
./aa1_data_util/3_process_zhihu_question_topic_relation.py:3:1: F821 undefined name 'reload'
./aa4_TextCNN_with_RCNN/p72_TextCNN_with_RCNN_train.py:5:1: F821 undefined name 'reload'
./aa5_BiLstmTextRelation/p9_BiLstmTextRelation_train.py:5:1: F821 undefined name 'reload'
./aa6_TwoCNNTextRelation/p9_twoCNNTextRelation_train.py:5:1: F821 undefined name 'reload'

Do you have any experiments that compare w2v_256（ZHIHU provide） with w2v_100（trained by yourself）？

2.average vectors, to get representation of the sentence

self.sentence_embeddings = tf.reduce_mean(sentence_embeddings, axis=1) # [None,self.embed_size]

When training, we pad all sentences as max sentence length, then tf.reduce_mean(sentence_embeddings, axis=1) means sum/(max sentence length). How could it know the length of each sentence?

Thanks!

I am wondering that the word2vec package is the user defined package, it is not include int the project

conda install word2vec

Thank you for sharing these wonderful codes!

Just one issue, it seems in a03 TextRNN, the result part (self.predictions) does not support multi-label classification. There is a lot to change if I want to adapt it to a multi-label classification task. Also the inference part is mostly about single label classification.

Best wishes,
Hang

@brightmart
Dear Mr.brightmart,
Hi,

While I run train a03_TextRNN with google_news_wor22vec.bin and a text file with my documents + labels, I've got these errors :

How can I solve this issue?

cache_path: cache_vocabulary_label_pik/rnn_word_voabulary.pik file_exists: False
create vocabulary. word2vec_model_path: GoogleNews-vectors-negative300.bin
rnn_model.vocab_size: 3000001
create_voabulary_label_sorted.started.traning_data_path: train-zhihu4-only-title-all.txt
length of list_label: 146
label: 8476641588870267502 count_value: 3
label: 3738968195649774859 count_value: 3
label: -3517637179126242000 count_value: 3
label: 810067918938531886 count_value: 2
label: 7476760589625268543 count_value: 2
label: 4313812860434517324 count_value: 2
label: 1462130073299421617 count_value: 2
label: -8377411942628634656 count_value: 2
label: -7046289575185911002 count_value: 2
label: -6259864339809244567 count_value: 2
count top10: 23
create_voabulary_label_sorted.ended.len of vocabulary_label: 146
load_data.started...
load_data_multilabel_new.training_data_path: train-zhihu4-only-title-all.txt
0 x0: w18476 w4454 w1674 w6 w25 w474 w1333 w1467 w863 w6 w4430 w11 w813 w4463 w863 w6 w4430 w111
0 x1: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
ys_index:
0 y: -6270130442784051389 ;ys_mulithot_list: 107
1 x1: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
ys_index:
1 y: 1945786109636206690 ;ys_mulithot_list: 70
ys_index:
2 y: 7792886053889220161 ;ys_mulithot_list: 22
ys_index:
3 y: 465065448523711562 ;ys_mulithot_list: 49
number_examples: 164
load_data.ended...
start padding & transform to one hot...
trainX[0]: [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
end padding & transform to one hot...

**

Traceback (most recent call last):
File "p8_TextRNN_train.py", line 184, in
tf.app.run()
File "/home/eslami/anaconda3/lib/python3.5/site-packages/tensorflow/python/platform/app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "p8_TextRNN_train.py", line 68, in main
vocab_size, FLAGS.embed_size, FLAGS.is_training)
File "/home/eslami/Downloads/all-kind-text_classification-master/a03_TextRNN/p8_TextRNN_model.py", line 33, in init
self.instantiate_weights()
File "/home/eslami/Downloads/all-kind-text_classification-master/a03_TextRNN/p8_TextRNN_model.py", line 45, in instantiate_weights
self.Embedding = tf.get_variable("Embedding",shape=[self.vocab_size, self.embed_size],initializer=self.initializer) #[vocab_size,embed_size] tf.random_uniform([self.vocab_size, self.embed_size],-1.0,1.0)
File "/home/eslami/anaconda3/lib/python3.5/site-packages/tensorflow/python/ops/variable_scope.py", line 1065, in get_variable
use_resource=use_resource, custom_getter=custom_getter)
File "/home/eslami/anaconda3/lib/python3.5/site-packages/tensorflow/python/ops/variable_scope.py", line 962, in get_variable
use_resource=use_resource, custom_getter=custom_getter)
File "/home/eslami/anaconda3/lib/python3.5/site-packages/tensorflow/python/ops/variable_scope.py", line 367, in get_variable
validate_shape=validate_shape, use_resource=use_resource)
File "/home/eslami/anaconda3/lib/python3.5/site-packages/tensorflow/python/ops/variable_scope.py", line 352, in _true_getter
use_resource=use_resource)
File "/home/eslami/anaconda3/lib/python3.5/site-packages/tensorflow/python/ops/variable_scope.py", line 664, in _get_single_variable
name, "".join(traceback.format_list(tb))))
ValueError: Variable Embedding already exists, disallowed. Did you mean to set reuse=True in VarScope? Originally defined at:

File "/home/eslami/Downloads/all-kind-text_classification-master/a03_TextRNN/p8_TextRNN_model.py", line 45, in instantiate_weights
self.Embedding = tf.get_variable("Embedding",shape=[self.vocab_size, self.embed_size],initializer=self.initializer) #[vocab_size,embed_size] tf.random_uniform([self.vocab_size, self.embed_size],-1.0,1.0)
File "/home/eslami/Downloads/all-kind-text_classification-master/a03_TextRNN/p8_TextRNN_model.py", line 33, in init
self.instantiate_weights()
File "/home/eslami/Downloads/all-kind-text_classification-master/a03_TextRNN/p8_TextRNN_model.py", line 123, in test
textRNN=TextRNN(num_classes, learning_rate, batch_size, decay_steps, decay_rate,sequence_length,vocab_size,embed_size,is_training)

**

Hi @brightmart
When i run the fasttext train script on windows or centos machine, I've got these errors "pickle.UnpicklingError: could not find MARK", it puzzled me a few days,please help me

D:\Anaconda3\python.exe D:/text_classification/a01_FastText/p6_fastTextB_train_multilabel.py
started...
ended...
D:\Anaconda3\lib\site-packages\gensim\utils.py:862: UserWarning: detected Windows; aliasing chunkize to chunkize_serial
warnings.warn("detected Windows; aliasing chunkize to chunkize_serial")
curses is not supported on this machine (please install/reinstall curses for an optimal experience)
Traceback (most recent call last):
File "D:/text_classification/a01_FastText/p6_fastTextB_train_multilabel.py", line 192, in
tf.app.run()
File "D:\Anaconda3\lib\site-packages\tensorflow\python\platform\app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "D:/text_classification/a01_FastText/p6_fastTextB_train_multilabel.py", line 38, in main
vocabulary_word2index, vocabulary_index2word = create_voabulary()
File "D:\text_classification\aa1_data_util\data_util_zhihu.py", line 31, in create_voabulary
model = Word2Vec.load(word2vec_model_path)
File "D:\Anaconda3\lib\site-packages\gensim\models\word2vec.py", line 1483, in load
model = super(Word2Vec, cls).load(*args, **kwargs)
File "D:\Anaconda3\lib\site-packages\gensim\utils.py", line 282, in load
cache_path: cache_vocabulary_label_pik/_word_voabulary.pik file_exists: False
create vocabulary. word2vec_model_path: ../data/zhihu-word2vec-title-desc.bin-100
obj = unpickle(fname)
File "D:\Anaconda3\lib\site-packages\gensim\utils.py", line 935, in unpickle
return _pickle.load(f, encoding='latin1')
_pickle.UnpicklingError: could not find MARK

No such file or directory: '..test-zhihu-forpredict-title-desc-v6.txt'

Thank you a lot for your sharing.

I find that in your implementation of Hierarchical Attention Network (HAN), the sentences are separated through setting an equal sentence length. This is however not the true sentence length in the data.

I wonder if it is easy to change this to using a sentence parser to find the sentences? How would be the difference in performance?

Please kindly let me know if you have any idea on parsing the real sentences based on your HAN code. Many thanks!

@brightmart
Hi Mr.Brightmart
I want to run fast text, but in first step for running train - during running train I've got below error :

Actually I have 2 datasets (one for economic_news and one for lifestyle -- totally there are 14000 documents) in both of them at the end of each document with label labels have been defined.
I attached them
train-zhihu4-only-title-all.txt
Also I made some changes on these 3 program :

data_util_zhihu.txt
p5_fastTextB_model.txt
p5_fastTextB_train.txt

interpreter report no module name this when i run fasttext.sorry, i realize it is a python package

Traceback (most recent call last):
File "p5_fastTextB_train.py", line 163, in
tf.app.run()
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/platform/app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "p5_fastTextB_train.py", line 76, in main
fast_text=fastText(FLAGS.label_size, FLAGS.learning_rate, FLAGS.batch_size, FLAGS.decay_steps, FLAGS.decay_rate,FLAGS.num_sampled,FLAGS.sentence_len,vocab_size,FLAGS.embed_size,FLAGS.is_training)
File "/home/defy/text_classification-master/a01_FastText/p5_fastTextB_model.py", line 29, in init
self.instantiate_weights()
File "/home/defy/text_classification-master/a01_FastText/p5_fastTextB_model.py", line 42, in instantiate_weights
self.Embedding = tf.get_variable("Embedding", [self.vocab_size, self.embed_size])
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/variable_scope.py", line 1049, in get_variable
use_resource=use_resource, custom_getter=custom_getter)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/variable_scope.py", line 948, in get_variable
use_resource=use_resource, custom_getter=custom_getter)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/variable_scope.py", line 356, in get_variable
validate_shape=validate_shape, use_resource=use_resource)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/variable_scope.py", line 341, in _true_getter
use_resource=use_resource)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/variable_scope.py", line 653, in _get_single_variable
name, "".join(traceback.format_list(tb))))
ValueError: Variable Embedding already exists, disallowed. Did you mean to set reuse=True in VarScope? Originally defined at:

File "/home/defy/text_classification-master/a01_FastText/p5_fastTextB_model.py", line 42, in instantiate_weights
self.Embedding = tf.get_variable("Embedding", [self.vocab_size, self.embed_size])
File "/home/defy/text_classification-master/a01_FastText/p5_fastTextB_model.py", line 29, in init
self.instantiate_weights()
File "/home/defy/text_classification-master/a01_FastText/p5_fastTextB_model.py", line 104, in test
fastText=fastTextB(num_classes, learning_rate, batch_size, decay_steps, decay_rate,5,sequence_length,vocab_size,embed_size,is_training)

python 2.7 ,tensorlfow 1.1, run p5_fastTextB_model, but get this error @brightmart

Hello.

Is there any chance to get some reference to papers (or any other documents) to the TextRNN model?

Thanks in advance.

Hi bright,

nice job on these baselines. I saw you said you did some ensemble work, so could you give some words about?

I didn't find a good way to do a traditional bagging or boosting, and I tried the hard voting, but results' bad. :-(

Multiple models (a02_TextCNN, a04_TextRCNN) have training and validation accuracy fixed at 0.5, while training loss/validation loss stably drops over time (dramatically, from 10^8-9 digits down to 1 to 2 digits.)

Is it normal or something wrong?

One of models, aa6_TwoCNNTextRelation has training accuracy fluctuate (above 0.5).

Thx.

I can't find the word2vec.py anywhere.

I followed the steps in issue#3 and still having an issue with the download.
@brightmart Can you please provide steps to download and/or to run with another dataset?

As a noob,I find it hard to debug your programs.There are too many problems...
Dear author,would you please provide even one complete code that could run directly?

Besides,I would really appreciate it if you explain how to make personal files similar with 'zhihu-word2vec-title-desc.bin-100'.

Can not find this file under the folder.

When I meet a mult-class and imbalance dataset.Add 'bias' variable to l2_loss looks like better.

你好，我想请问一下怎么把我自己的英文语料库的格式改成和你的数据一样的格式？

Issue when trying to run the p7_TextCNN_predict.py after training

ModuleNotFoundError: No module named 'a02_TextCNN'

Question: can you get it running with clean anaconda and Python 3.6 and additional such word2vec, etc. installs ? I think that there are a some code changes needed in the files.

i tried to run the sample. but i found the code" import word2vec" in data_util_zhihu.py couldn't work. i want to know where can i download the word2vec, thank you very mcuh.

1.get emebedding of words in the sentence

    sentence_embeddings = tf.nn.embedding_lookup(self.Embedding,self.sentence)  #  [None,self.sentence_len,self.embed_size]

2.average vectors, to get representation of the sentence

    self.sentence_embeddings = tf.reduce_mean(sentence_embeddings, axis=1)  # [None,self.embed_size]

Since the length of sentences is variable, do we need a mask tensor for geting the average?
For example, tf.reduce_sum(tf.multiply(sentence_embeddings,mask), axis=1) / tf.reduce_sum(mask, axis=1)

I'm getting this issue when I run training on a08 entity network and a06 seq2seq models.

Can I get or train this file?

zhihu-word2vec-title-desc.bin-100

Also, do you have sample datasets compatible with these models?

• the final loss is defined as the sum of cross_entropy_loss and L2-loss to penalty large variables, where a hyperparameter l2_lambda is used to balance these two items. In the algorithm, as the hyperparameter l2_lambda is directly set to 0.0001, would it be too small(or too large) under some circumstances such that one in the two items loses its contribution to the whole loss?
• in general, is there any pratical methods to guide me to set values of hyperparameters, e.g. l2_lambda
  @brightmart

There is no file named "test-zhihu-forpredict-title-desc-v6.txt",when i run the p1_HierarchicalAttention_predict.py in the Hierarchical Attention Network. Also i have tried to use the test-zhihu6-title-desc.txt instead, but there will be an error. Can you give me some advice? @brightmart