ZeroDivisionError in Doc2Vec.build_vocab

[jgc128]

When I call model.build_vocab(docs_iterator) sometimes I got ZeroDivisionError:

Traceback (most recent call last):
  File "patient_notes_doc2vec.py", line 45, in <module>
    model.build_vocab(docs_iterator)
  File "/data2/aromanov/virt_env/theano/lib/python3.4/site-packages/gensim/models/word2vec.py", line 495, in build_vocab
    self.scan_vocab(sentences, trim_rule=trim_rule)  # initial survey
  File "/data2/aromanov/virt_env/theano/lib/python3.4/site-packages/gensim/models/doc2vec.py", line 634, in scan_vocab
    interval_rate = (total_words - interval_count) / (default_timer() - interval_start)
ZeroDivisionError: float division by zero

A few lines above in doc2vec.py interval_start is defined as interval_start = default_timer() and in the for loop we have

interval_rate = (total_words - interval_count) / (default_timer() - interval_start)

Perhaps my system is too fast 😃

[gojomo]

Too fast... or the clock too coarse!

This definitely shouldn't happen; I'll likely add a +0.001 (or similar). Then the rate-estimate may occasionally be off, but the code won't error.

Thanks for the report!

[gojomo]

Also reported by @kikohs in #529.

[tmylk]

@gojomo Pinging this issue. Is it an easy fix that can be include in this month's release?