Update app.py

app.py

@@ -2,6 +2,25 @@ import streamlit as st
 
 st.markdown("""
 
+# Imperative?
+Torch is more popular since its imperative which is often easier to understand.
+What does imperative mean with respect to programming?
+The open source community is amazing at solving for gaps.  Competition is healthy since frameworks evolve.
+Theano, Keras -> TF and Keras -> Pytorch and GPU libs.
+Early was great, but hard to teach since they define computational graph up front.
+Hard to debug.
+Pytorch was strongest to use normal python.  Everything you know works on GPU when necessary which causes a huge leap in teaching.
+Pytorch is not easy for newcomers or researchers.  API however writes NN in three lines of code using progressive API's
+Fast AI library was critical.
+Are we still limited by python?  Is Swift an alternative?  Torch or TF?  Hard to say.
+    - Vision for tensorflow was good.
+    - Pytorch features interactive computation
+    - TF eager is tentimes slower than pytorch - caution might be dead end.
+    - Look at the code, develop your own perspective.
+    - SWIFT could be layer above GPU and be great.
+
+
+
 # GPU Setup Instructions for NVIDIA GPU RTX 3070+
 
 1. Upgrade pip: ```python.exe -m pip install --upgrade pip```