avidale / dependency-paraphraser Goto Github PK

Hi there!
Trying to use dependency-paraphraser for English. As the word2vec model I use 'word2vec-google-news-300' and load it like this:

import dependency_paraphraser.udpipe
import gensim.downloader

w2v_model = gensim.downloader.load('word2vec-google-news-300')

path = 'english-ewt-ud-2.5-191206.udpipe
pipe = dependency_paraphraser.udpipe.Model(path)
projector = dependency_paraphraser.udpipe.en_udpipe_projector

text = 'in April 2012 they released the videoclip for a new single entitled Giorgio Mastrota.'
for i in range(3):
print(dependency_paraphraser.udpipe.paraphrase(text, pipe, w2v=w2v_model, projector=projector, tree_temperature=1))

I get the following error:

Traceback (most recent call last):
File "dependency_rewriter.py", line 34, in
print(dependency_paraphraser.udpipe.paraphrase(text, pipe, w2v=w2v_model, projector=projector, tree_temperature=1))
File "/home/.local/lib/python3.7/site-packages/dependency_paraphraser/udpipe.py", line 90, in paraphrase
toks, w2v=w2v, morph_vocab=morph, min_sim=min_sim, p_rep=p_rep,
File "/home/.local/lib/python3.7/site-packages/dependency_paraphraser/synonyms.py", line 48, in replace_synonyms
result.append(token.text)
AttributeError: 'Token' object has no attribute 'text'

How could I fix it? Thanks in advance for answer!

For example: 'где ближайший магазин продуктов от меня' - question for a smart assistant
The paraphraser returns: '' (empty string)

Привет!
Запускаю первый пример из ридми.
Сталкиваюсь с такой ошибкой:

/usr/local/lib/python3.9/dist-packages/sklearn/base.py:318: UserWarning: Trying to unpickle estimator DictVectorizer from version 0.22.2.post1 when using version 1.2.2. This might lead to breaking code or invalid results. Use at your own risk. For more info please refer to:
https://scikit-learn.org/stable/model_persistence.html#security-maintainability-limitations
  warnings.warn(
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
[<ipython-input-2-d8a1d21065ce>](https://localhost:8080/#) in <cell line: 3>()
      1 text = 'каждый охотник желает знать где сидит фазан'
      2 
----> 3 import dependency_paraphraser.natasha
      4 import random
      5 random.seed(42)

1 frames
[/usr/local/lib/python3.9/dist-packages/dependency_paraphraser/pretrained_projectors.py](https://localhost:8080/#) in <module>
      6 
      7 with open(os.path.join(models_path, 'natasha_projector.pkl'), 'rb') as f:
----> 8     natasha_projector = pickle.load(f)
      9 
     10 with open(os.path.join(models_path, 'en_udpipe_projector.pkl'), 'rb') as f:

AttributeError: Can't get attribute 'PolynomialFeatures' on <module 'sklearn.preprocessing._data' from '/usr/local/lib/python3.9/dist-packages/sklearn/preprocessing/_data.py'>

Пока не пробовал делать ничего с версиями.
По всей видимости что-то изменилось в аттрибутах sklearn?