Deep Residual Nets (ResNets) from Microsoft Research has become one of the popular deep learning network architecture. Already 800+ citation, given that the paper appeared in 2015.
Recently, I ported all my code from Caffe to Tensorflow. While it is lot easier to deal with caffe but I must say, the control you get with tensorflow is worth the effort. Beware though learning to use Google’s Tensorflow has a steep learning curve. It will be mindnumbing (I think) if you do not understand how neural net’s computations happen. Also useful to read and understand the Batch-Normalization paper.
In case you are looking for a simple explanation (and implementation) on `How Neural Networks Work’ have a look at a post I had written earlier : [here]. You might also be interested in my post on Convolutional Networks, here.
The main difference between tensorflow and caffe is that the computations in tf (tensorflow) are defined as a computational graph. While in caffe you specify the layers in a text file. TF in general can be used to any computations not just neural nets. With its superb support for multi-gpu and also distributed gpu it gives you a lot of teeth ;).
You can download my class to use the Deep Residual Nets here.