signed compression

inpho · Apr 26, 2016 · 54d22a2 · 54d22a2
1 parent ce39c73
commit 54d22a2
Show file tree

Hide file tree

Showing 5 changed files with 38 additions and 38 deletions.
diff --git a/vsm/corpus/base.py b/vsm/corpus/base.py
@@ -468,9 +468,9 @@ def __init__(self,
 
         # Integer encoding of a string-type corpus
         if len(self.words) < 2 ** 16:
-            self.dtype = np.uint16
+            self.dtype = np.int16
         else:
-            self.dtype = np.uint32
+            self.dtype = np.int32
 
         self.corpus = np.asarray([self.words_int[unicode(word)] 
                                   for word in self.corpus],

diff --git a/vsm/model/_cgs_update.c b/vsm/model/_cgs_update.c
diff --git a/vsm/model/_cgs_update.pyx b/vsm/model/_cgs_update.pyx
@@ -4,7 +4,7 @@ import cython
 @cython.cdivision(True)
 
 def cgs_update(int itr, 
-               unsigned int [:] corpus,
+               int [:] corpus,
                double [:,:] word_top,
                double [:] inv_top_sums,
                double [:,:] top_doc,
@@ -87,7 +87,7 @@ def cgs_update(int itr,
             mtrand_state[3], mtrand_state[4])
 
 def cgs_update_short(int itr, 
-               unsigned short [:] corpus,
+               short [:] corpus,
                double [:,:] word_top,
                double [:] inv_top_sums,
                double [:,:] top_doc,

diff --git a/vsm/model/ldacgsmulti.py b/vsm/model/ldacgsmulti.py
@@ -194,10 +194,10 @@ def corpus(self, a):
         if self._write_globals:
             global _corpus
             if not '_corpus' in globals():
-                if self.corpus.dtype == 'uint16':
-                    dtype = 'H'
-                elif self.corpus.dtype == 'uint32':
-                    dtype = 'I'
+                if self.corpus.dtype == 'int16':
+                    dtype = 'h'
+                elif self.corpus.dtype == 'int32':
+                    dtype = 'i'
                 else:
                     raise NotImplementedError
 
@@ -405,9 +405,9 @@ def update((docs, doc_indices, mtrand_state, dtype)):
 
     indices = np.array([(j - start) for (i,j) in docs], dtype='i')
 
-    if dtype == np.uint16:
+    if dtype == np.int16:
         update_fn = cgs_update_short
-    elif dtype == np.uint32:
+    elif dtype == np.int32:
         update_fn = cgs_update
     else:
         raise NotImplementedError

diff --git a/vsm/model/ldacgsseq.py b/vsm/model/ldacgsseq.py
@@ -125,9 +125,9 @@ def train(self, n_iterations=100, verbose=1, **kwargs):
         :param kwargs: For compatability with calls to LdaCgsMulti.
         :type kwargs: optional
         """
-        if self.corpus.dtype == np.uint16:
+        if self.corpus.dtype == np.int16:
             update = cgs_update_short
-        elif self.corpus.dtype == np.uint32:
+        elif self.corpus.dtype == np.int32:
             update = cgs_update
         else:
             raise NotImplementedError(