Changed the initialisation of the sampler, hopefully this will work better.

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@376 ec762483-ff6d-05da-a07a-a48fb63a330f

1 files changed, 4 insertions, 2 deletions

diff --git a/gi/pyp-topics/src/pyp-topics.cc b/gi/pyp-topics/src/pyp-topics.cc
index 06499291..4c777f0c 100644
--- a/gi/pyp-topics/src/pyp-topics.cc
+++ b/gi/pyp-topics/src/pyp-topics.cc
@@ -58,7 +58,8 @@ void PYPTopics::sample_corpus(const Corpus& corpus, int samples,
       int new_topic = -1;
       if (freq > frequency_cutoff
           && (!max_contexts_per_document || term_index < max_contexts_per_document)) {
-        new_topic = document_id % m_num_topics;
+        new_topic = sample(document_id, term);
+        //new_topic = document_id % m_num_topics;
 
         // add the new topic to the PYPs
         increment(term, new_topic);
@@ -314,7 +315,8 @@ PYPTopics::F PYPTopics::word_pyps_p0(const Term& term, int topic, int level) con
     Term backoff_term = (*m_backoff)[term];
     if (!m_backoff->is_null(backoff_term)) {
       assert (level < m_backoff->order());
-      p0 = (1.0/(double)m_backoff->terms_at_level(level))*prob(backoff_term, topic, level+1);
+      //p0 = (1.0/(double)m_backoff->terms_at_level(level))*prob(backoff_term, topic, level+1);
+      p0 = prob(backoff_term, topic, level+1);
     }
     else
       p0 = m_term_p0;