diff --git a/couplingSimulationML/ML_PythonC++_Embedding/CouplingDiagram.png b/couplingSimulationML/ML_PythonC++_Embedding/CouplingDiagram.png
index d3cdc83..b4ad2a6 100644
Binary files a/couplingSimulationML/ML_PythonC++_Embedding/CouplingDiagram.png and b/couplingSimulationML/ML_PythonC++_Embedding/CouplingDiagram.png differ
diff --git a/couplingSimulationML/ML_PythonC++_Embedding/README.md b/couplingSimulationML/ML_PythonC++_Embedding/README.md
index 8100756..508777c 100644
--- a/couplingSimulationML/ML_PythonC++_Embedding/README.md
+++ b/couplingSimulationML/ML_PythonC++_Embedding/README.md
@@ -14,7 +14,7 @@ In addition, this code also highlights the advantages of integrating the Python
 
 The test-case demonstrated here aims to capture a modal decomposition using an SVD.  This is representative of a Sci-ML workload. We aim to build this toward a surrogate model using TensorFlow in Python from data generated by a C++ computation. Further details on developing the surrogate model can be found [here](archive/ThetaGPU/Background.md)
 
-Here is how we connected C++ and Python, although we didn't include the neural network training in this demo:
+For today's demo, we didn't include the neural network training. Here is how we connected C++ and Python:
 ![Coupling](CouplingDiagram.png)
 
 For running this mini-app on ThetaGPU, look at the scripts/README within `ML_PythonC++_Embedding/ThetaGPU_OCCA/` subdirectory.