fixes #518: tiny 10µsec fudge against 0 elapsed

CHANGELOG.txt

@@ -14,6 +14,7 @@ Changes
   - `utils.HAS_PATTERN` flag moved to `utils.has_pattern()`
 * Threadsafe Word2Vec/Doc2Vec finish-check to avoid hang/unending Word2Vec/Doc2Vec training (Gordon Mohr, #571)
 * Tuned `TestWord2VecModel.test_cbow_hs()` against random failures (Gordon Mohr, #531)
+* Prevent ZeroDivisionError when `default_timer()` indicate no elapsed time (Gordon Mohr, #518)
 * Forwards compatibility for NumPy > 1.10 (Matti Lyra, #494, #513)
   - LdaModel and LdaMulticore produce a large number of DeprecationWarnings from
     .inference() because the term ids in each chunk returned from utils.grouper

gensim/models/doc2vec.py

@@ -626,7 +626,7 @@ def scan_vocab(self, documents, progress_per=10000, trim_rule=None):
         document_no = -1
         total_words = 0
         min_reduce = 1
-        interval_start = default_timer()
+        interval_start = default_timer() - 0.00001  # guard against next sample being identical
         interval_count = 0
         vocab = defaultdict(int)
         for document_no, document in enumerate(documents):

gensim/models/word2vec.py

@@ -807,7 +807,7 @@ def job_producer():
             thread.start()
 
         example_count, trained_word_count, raw_word_count = 0, 0, word_count
-        start, next_report = default_timer(), 1.0
+        start, next_report = default_timer() - 0.00001, 1.0
 
         while unfinished_worker_count > 0:
             report = progress_queue.get()  # blocks if workers too slow
@@ -844,7 +844,7 @@ def job_producer():
         elapsed = default_timer() - start
         logger.info(
             "training on %i raw words (%i effective words) took %.1fs, %.0f effective words/s",
-            raw_word_count, trained_word_count, elapsed, trained_word_count / elapsed if elapsed else 0.0)
+            raw_word_count, trained_word_count, elapsed, trained_word_count / elapsed)
         if job_tally < 10 * self.workers:
             logger.warn("under 10 jobs per worker: consider setting a smaller `batch_words' for smoother alpha decay")
 
@@ -968,7 +968,7 @@ def worker_loop():
         self.clear_sims()
         logger.info(
             "scoring %i sentences took %.1fs, %.0f sentences/s",
-            sentence_count, elapsed, sentence_count / elapsed if elapsed else 0.0)
+            sentence_count, elapsed, sentence_count / elapsed)
         return sentence_scores[:sentence_count]
 
     def clear_sims(self):