Merge branch 'master' of github.com:moses-smt/mosesdecoder

contrib/arrow-pipelines/python/pcl

@@ -1 +1 @@
-Subproject commit 4e51345cd6b405551ba327d91a7d905e204587b1
+Subproject commit fe569f6ad720306e0f354922494eb0203befb165

mert/CderScorer.cpp

@@ -68,8 +68,16 @@ void CderScorer::prepareStatsVector(size_t sid, const string& text, vector<int>&
     const sent_t& ref = m_ref_sentences[rid][sid];
     tmp.clear();
     computeCD(cand, ref, tmp);
-    if (calculateScore(tmp) > max) {
+    int score = calculateScore(tmp);
+    if (rid == 0)
+    {
       stats = tmp;
+      max = score;
+    }
+    else if (score > max) 
+    {
+      stats = tmp;
+      max = score;
     }
   }
 }
@@ -94,8 +102,12 @@ void CderScorer::computeCD(const sent_t& cand, const sent_t& ref,
   vector<int>* row = new vector<int>(I);
 
   // Initialization of first row
-  (*row)[0] = 0;
-  for (int i = 1; i < I; ++i) (*row)[i] = 1;
+  for (int i = 0; i < I; ++i) (*row)[i] = i;
+  
+  // For CDER metric, the initialization is different
+  if (m_allowed_long_jumps) {
+    for (int i = 1; i < I; ++i) (*row)[i] = 1;
+  }
 
   // Calculating costs for next row using costs from the previous row.
   while (++l < L) {

scripts/training/create_count_tables.py

@@ -71,7 +71,7 @@ def create_count_lines(fobj, countobj, countobj_target, prune=0):
 
     i = 0
     original_pos = 0
-    source = ""
+    source = b""
     store_lines = set()
     for line in fobj: