Mention inconsistencies between the pre-trained models and the model definitions for #351 and #349

docs/index.md

@@ -150,6 +150,11 @@ CMU-CS-16-118, CMU School of Computer Science, Tech. Rep., 2016.
 ```
 
 # Acknowledgements
++ Kenneth Jung noticed that the model definitions are slightly
+  different than the pre-trained models.
+  For more information, see issues
+  [#351](https://github.com/cmusatyalab/openface/issues/351) and
+  [#349](https://github.com/cmusatyalab/openface/issues/349).
 + The fantastic Torch ecosystem and community.
 + [Alfredo Canziani's](https://github.com/Atcold)
   implementation of FaceNet's loss function in
@@ -189,7 +194,7 @@ These portions are noted in the source files and are
 copyright their respective authors with
 the licenses listed.
 
-Project | Modified | License
----|---|---|
+| Project | Modified | License |
+|---|---|---|
 [Atcold/torch-TripletEmbedding](https://github.com/Atcold/torch-TripletEmbedding) | No | MIT
 [facebook/fbnn](https://github.com/facebook/fbnn) | Yes | BSD

docs/models-and-accuracies.md

 # Models and Accuracies
+
+Note: Kenneth Jung noticed that the model definitions are slightly
+different than the pre-trained models.
+For more information, see issues
+[#351](https://github.com/cmusatyalab/openface/issues/351) and
+[#349](https://github.com/cmusatyalab/openface/issues/349).
+
+---
+
 This page overviews different OpenFace neural network models
 and is intended for advanced users.
 

docs/training-new-models.md

 # Training new neural network models
 
+Note: Kenneth Jung noticed that the model definitions are slightly
+different than the pre-trained models.
+For more information, see issues
+[#351](https://github.com/cmusatyalab/openface/issues/351) and
+[#349](https://github.com/cmusatyalab/openface/issues/349).
+
+---
+
 We have also released our deep neural network (DNN)
 training infrastructure to promote an open ecosystem and enable quicker
 bootstrapping for new research and development.