vijayaiitk / nlp-text-classification-model Goto Github PK

Hi,

I'm using your code here, based on your article in Medium. Because of gensim's update from 3.8 to 4 (which was a major revision) I received some errors.

The first was that model.wv.index2word has been moved to model.wv.index_to_key. So I changed the code accordingly to get rid of this error.

#w2v = dict(zip(model.wv.index2word, model.wv.syn0))  #combination of word and its vector
w2v = dict(zip(model.wv.index_to_key, model.wv.syn0))  #combination of word and its vector

However, this then triggered a new error that I haven't been able to figure out.

AttributeError      Traceback (most recent call last)
<ipython-input-54-04b58a6eb72e> in <module>
      6 #if min_count=2 means if the word is present less than 2 times across all the documents then we shouldn't consider it
      7 
----> 8 w2v = dict(zip(model.wv.index_to_key, model.wv.syn0))  #combination of word and its vector
      9 #w2v = dict(zip(model.wv.index2word, model.wv.syn0))  #combination of word and its vector
     10 
AttributeError: 'KeyedVectors' object has no attribute 'syn0'

Checking the web, the recommended solution for a object has no attribute 'syn0' is to change from model.syn0 to model.wv.syn0 ... but of course this is what the code already is.

Do you have any idea how I can correct this code?

Thanks!

Hello,

During my study on this NLP modelling I realized that @vijayaiitk should have mentioned the versions for gensim and some other modules. Because the deprecated versions do not work. For example gensim 3.8.3 should be added to the requirements otherwise visitors who try will encounter repetitive issues related to version.

If you can add requirements I would appreciate it,
Thanks

if my target has 3 classes: positive negative and neutral, will your algorithm work or will I need to modify something?