diff --git a/demo_run.py b/demo_run.py
index ed5b26a..06cb698 100644
--- a/demo_run.py
+++ b/demo_run.py
@@ -40,12 +40,12 @@ def load_data_2_root(data):
 
 
 if __name__ == "__main__":
-    root_name = basedir + "data/root.pkl"
+    root_name = basedir + "/data/root.pkl"
     stopwords = get_stopwords()
     if os.path.exists(root_name):
         root = load_model(root_name)
     else:
-        dict_name = basedir + 'data/dict.txt'
+        dict_name = basedir + '/data/dict.txt'
         word_freq = load_dictionary(dict_name)
         root = TrieNode('*', word_freq)
         save_model(root, root_name)
diff --git a/model.py b/model.py
index 17980a3..4ff227a 100644
--- a/model.py
+++ b/model.py
@@ -81,6 +81,7 @@ def add(self, word):
         length = len(word)
         node = self.root
         if length == 3:
+            word = list(word)
             word[0], word[1], word[2] = word[1], word[2], word[0]
 
             for count, char in enumerate(word):
diff --git a/utils.py b/utils.py
index 1df28b4..f46ccd8 100644
--- a/utils.py
+++ b/utils.py
@@ -15,7 +15,10 @@ def get_stopwords():
 
 
 def generate_ngram(input_list, n):
-    return zip(*[input_list[i:] for i in range(n)])
+    result = []
+    for i in range(1, n+1):
+        result.extend(zip(*[input_list[j:] for j in range(i)]))
+    return result
 
 
 def load_dictionary(filename):