Merge pull request #296 from elyase/patch-2

make use of log_smooth_count
This commit is contained in:
Matthew Honnibal 2016-03-19 06:50:30 +11:00
commit a862edc0e6

View File

@ -129,7 +129,7 @@ def _read_freqs(loc, max_length=100, min_doc_freq=0, min_freq=200):
word = key word = key
smooth_count = counts.smoother(int(freq)) smooth_count = counts.smoother(int(freq))
log_smooth_count = math.log(smooth_count) log_smooth_count = math.log(smooth_count)
probs[word] = math.log(smooth_count) - log_total probs[word] = log_smooth_count - log_total
oov_prob = math.log(counts.smoother(0)) - log_total oov_prob = math.log(counts.smoother(0)) - log_total
return probs, oov_prob return probs, oov_prob