Tensorflow is not for the masses. Despite over 13k stars on Github and being the most trending GitHub repository in November 2015 very few original and working code examples have been released. Why example code? Because TF is really a programming framework, without deep Python and C++ knowledge and being familiar with all twists and turns of libraries and dependencies it is almost impossible to work with.

Of course some Neural Net and machine learning knowledge is also substantial. The very detailed TensorFlow WIKI is geared more towards experts and professionals, but not the average Deep Learning hobby user, which is unfortunate, but maybe will change in the far future.

Intermezzo 2017: The high-level Keras deep learning API makes use of TF and Theaono, therefore nullifying the above criticism. TF remains powerful for low level deep learning approaches, otherwise use Keras. The native TF documentation also improved tremendously.

• TF example - Early collection of TF examples by Aymeric Damien.
• TF@Kaggle - TF in Kaggle contests by Niketan Pansare
• TF WIKI - The TF WIKI with code snippets