Update README

README

@@ -6,4 +6,4 @@
 4. Save the dataset splits `train_data.txt`, `test_data.txt`, and `val_data.txt` into `omnigibson/data/3d_front`, and also in the parent directory of 'omnigibson'.
 5. *(Optional)* You may unzip `imitation.zip` to its location and use it for imitation learning training. You should create a directory named 'imitation_data' under the parent directory of 'omnigibson'.
 6. If you finished installing OmniGibson, use python -m omnigibson.create_mesh to preprocess the meshes in the scene.
-7. The imitation dataset provided here is a proof-of-concept example. If you are considering training a high quality imilation policy, please generate yourself.
+7. The imitation dataset provided here is a proof-of-concept example. If you are considering training a high quality imilation policy, please generate yourself (refer to instructions in the README our github repo).