changed sim.metric to inner product (dot)

src/DB_config.yml

@@ -9,8 +9,8 @@ faiss:
     bpsv: 8 # bytes per sub vector
 annoy:
   init:
-    smetric: "angular" # similarity metric to be used
-    ntrees: 10 # no. of trees
+    smetric: "dot" # similarity metric to be used
+    ntrees: 500 # no. of trees
     search_k: -1 # search_k is provided in runtime and affects the search performance. A larger value will give more accurate results, but will take longer time to return.
 queue:
   qlen: 100000 # length limit for quesues used

src/vec_index/hfaiss.py

@@ -53,7 +53,7 @@ def spawn (self):
     def init_faiss(self, matrix):
         self.train_data = np.matrix(matrix).astype('float32')
         logging.debug('FAISS init quantizer')
-        self.f_quantizer = faiss.IndexFlatL2(self.dim)
+        self.f_quantizer = faiss.IndexFlatIP(self.dim)
         # Lock index read / wtite until it is built
         with self._lock:
             logging.debug('FAISS init index')