diff --git a/pytorch_lightning/trainer/ignored_warnings.py b/pytorch_lightning/trainer/ignored_warnings.py
new file mode 100644
index 0000000000..9260720ec3
--- /dev/null
+++ b/pytorch_lightning/trainer/ignored_warnings.py
@@ -0,0 +1,14 @@
+import warnings
+
+
+def ignore_scalar_return_in_dp():
+    # Users get confused by this warning so we silence it
+    m_1 = """
+    Was asked to gather along dimension 0, but all
+    input tensors were scalars; will instead unsqueeze
+    and return a vector.
+    """
+    warnings.filterwarnings('ignore', message=m_1)
+
+
+ignore_scalar_return_in_dp()
diff --git a/pytorch_lightning/trainer/trainer.py b/pytorch_lightning/trainer/trainer.py
index 6f8bd6005f..375d332841 100644
--- a/pytorch_lightning/trainer/trainer.py
+++ b/pytorch_lightning/trainer/trainer.py
@@ -22,6 +22,7 @@ from pytorch_lightning.pt_overrides.override_data_parallel import (
 from pytorch_lightning.callbacks import GradientAccumulationScheduler
 from pytorch_lightning.utilities.debugging import MisconfigurationException
 import pdb
+from pytorch_lightning.trainer import ignored_warnings
 
 try:
     from apex import amp