diff --git a/MANIFEST.in b/MANIFEST.in
index 5fec488..a8927ea 100644
--- a/MANIFEST.in
+++ b/MANIFEST.in
@@ -1 +1,2 @@
-include rlmolecule/policy/preprocessor.json
+include pyproject.toml
+recursive-include rlmolecule *.json
\ No newline at end of file
diff --git a/rlmolecule/policy/data/__init__.py b/rlmolecule/policy/data/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/rlmolecule/policy/preprocessor.json b/rlmolecule/policy/data/preprocessor.json
similarity index 100%
rename from rlmolecule/policy/preprocessor.json
rename to rlmolecule/policy/data/preprocessor.json
diff --git a/rlmolecule/policy/preprocessor.py b/rlmolecule/policy/preprocessor.py
index 81a4c9c..da6d6f4 100644
--- a/rlmolecule/policy/preprocessor.py
+++ b/rlmolecule/policy/preprocessor.py
@@ -1,7 +1,7 @@
 import os
 from typing import Optional
 
-import rdkit
+import rdkit.Chem
 
 # from nfp.preprocessing.mol_preprocessor import MolPreprocessor
 from nfp.preprocessing import MolPreprocessor
@@ -66,7 +66,7 @@ def load_preprocessor(saved_preprocessor_file: Optional[str] = None) -> MolPrepr
 
     if not saved_preprocessor_file:
         saved_preprocessor_file = os.path.join(
-            os.path.dirname(os.path.abspath(__file__)), "preprocessor.json"
+            os.path.dirname(os.path.abspath(__file__)), "data", "preprocessor.json"
         )
 
     preprocessor.from_json(saved_preprocessor_file)
diff --git a/setup.cfg b/setup.cfg
index 053a42b..4b3dcca 100644
--- a/setup.cfg
+++ b/setup.cfg
@@ -11,6 +11,7 @@ classifiers =
 
 [options]
 packages = find:
+include_package_data = true
 install_requires =
   graphenv
   tensorflow