diff --git a/.travis.yml b/.travis.yml
index abc9aa31..a226139f 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -15,6 +15,8 @@ before_install:
   - sudo apt-get update
   - sudo apt-get install libc6
   - conda install --yes libgcc
+  - sudo apt-get install python-dev
+  - sudo apt-get install python3-dev
 install:
   - conda create --yes -n shorttext-test python=$TRAVIS_PYTHON_VERSION
   - source activate shorttext-test
diff --git a/README.md b/README.md
index a8681f13..d9ee5ee0 100644
--- a/README.md
+++ b/README.md
@@ -114,6 +114,7 @@ If you would like to contribute, feel free to submit the pull requests. You can
 
 ## Possible Future Updates
 
+- [x] Moving the testing word2vec model to AWS S3;
 - [ ] More scalability;
 - [ ] Including BERT models;
 - [ ] Dividing components to other packages;
diff --git a/data/test_w2v_model.bin b/data/test_w2v_model.bin
deleted file mode 100644
index a7b5dd96..00000000
Binary files a/data/test_w2v_model.bin and /dev/null differ
diff --git a/test/test_var_nn_embedded_vec_classifier.py b/test/test_var_nn_embedded_vec_classifier.py
index d594b11f..ca6aca60 100644
--- a/test/test_var_nn_embedded_vec_classifier.py
+++ b/test/test_var_nn_embedded_vec_classifier.py
@@ -11,7 +11,7 @@
 class TestVarNNEmbeddedVecClassifier(unittest.TestCase):
     def setUp(self):
         print("Downloading word-embedding model....")
-        link = "https://github.com/stephenhky/PyShortTextCategorization/blob/master/data/test_w2v_model.bin?raw=true"
+        link = "https://s3.amazonaws.com/shorttext-data-northernvirginia/trainingdata/test_w2v_model.bin"
         filename = "test_w2v_model.bin"
         if not os.path.isfile("test_w2v_model.bin"):
             if sys.version_info[0]==2: