This blog post was awesome. I am working through a lot of TensorFlow material lately and it was refreshing and atypical to find someone telling you what not to do.
I am capturing it here so I can find it again in the future. If you are interested in TensorFlow, or technical writing, this is worth a read: How Not To Program the TensorFlow Graph
This is why I love trad climbing; it really is the sport for engineers. Every single piece of gear and how it is used gets meticulously engineered and tested. I especially love black diamonds gear as Black Diamond’s testing lab posts a lot about their testing results.
Danijar Hafner has an excellent tutorial on Structuring Your TensorFlow Models. I really liked it so I am posting it here so its always easy to find. He is using Python’s decorators to provide a lazy initialization for the graph components. Its a neat trick. Totally separate from TensorFlow, I am actively working on improving my Python programming and will definitely be using this approach in the future.