Allows training if model is not modified by "_minimize_model". Adds deprecation warning. #1207
Conversation
gensim/models/word2vec.py
Outdated
| self.model_trimmed_post_training = True | ||
| if not save_syn1 or not save_syn1neg or not save_syn0_lockf: | ||
| self.model_trimmed_post_training = True | ||
| logger.warning("This method would be deprecated in the future. Use KeyedVectors to retain the previously trained word vectors instead.") |
There was a problem hiding this comment.
Yes, I agree. Thanks for your suggestion. I am making this change in the PR.
gensim/models/word2vec.py
Outdated
| @@ -1251,7 +1251,10 @@ def _minimize_model(self, save_syn1 = False, save_syn1neg = False, save_syn0_loc | |||
| del self.syn1neg | |||
| if hasattr(self, 'syn0_lockf') and not save_syn0_lockf: | |||
| del self.syn0_lockf | |||
| self.model_trimmed_post_training = True | |||
| if not save_syn1 or not save_syn1neg or not save_syn0_lockf: | |||
There was a problem hiding this comment.
Do you mean when all 3 are true? Then use if (save_syn1 and save_syn1neg and save_syn0_lockf) and just return. Please move this check to the top of the function.
gensim/models/word2vec.py
Outdated
| self.model_trimmed_post_training = True | ||
| if not save_syn1 or not save_syn1neg or not save_syn0_lockf: | ||
| self.model_trimmed_post_training = True | ||
| import warnings |
There was a problem hiding this comment.
Please move imports to the file header
gensim/models/word2vec.py
Outdated
| if not save_syn1 or not save_syn1neg or not save_syn0_lockf: | ||
| self.model_trimmed_post_training = True | ||
| import warnings | ||
| warnings.warn("This method would be deprecated in the future. Use KeyedVectors to retain the previously trained word vectors instead.") |
There was a problem hiding this comment.
Please move to the top of the function. Better text is "This method would be deprecated in the future. Keep just_word_vectors = model.wv to retain just the KeyedVectors instance for read-only querying of word vectors."
|
@tmylk Thanks for the suggestions. I have updated the code to reflect these changes. |
|
Thanks for the pr |
…recation warning. (piskvorky#1207)
This PR modifies the function
_minimize_modelin word2vec.py by checking if model was modified at all before settingmodel_trimmed_post_training. Also, a deprecation warning has been added for the function_minimize_model. The function would be deprecated in the future as using KeyedVectors is the way to go for word2vec.Maling list discussion.