Recursive neural network implementation in Theano Is there any available recursive neural network implementation in Theano? Theano's deeplearning.net tutorial does not present any recursive neural networks. Most Theano code I've found is CNN, LSTM, GRU, vanilla recurrent neural networks or MLP. I am most interested in implementations for natural language processing.
Example of a recursive neural network:

 A: Here you go https://github.com/ofirnachum/tree_rnn.
This Theano implementation should be roughly what is described in the Kai Sheng Tai paper.
A: Recursive nets are similar to recurrent nets in that they share weights between layers, but are not the same thing at all. Recursive nets are tree-structured, and error backpropagation is through structure, rather than time as in most recurrent nets. One important difference from recurrent nets is that one recursive net can run with a large collection of trees (such as a parsed linguistic corpus) each of which are different (though all are usually binary). Richard Socher has written extensively about recursive nets. See Richard Socher's Web site for papers.
A: In neural network terminology we usually say a network is recurrent rather than recursive. Assuming you mean the same as I do (at least one layer in which nodes feed back into themselves).
That list actually links right to this tutorial on recurrent neural networks and word embeddings. 
Also LSTM and GRU are variants on the RNN topology that have both been proven to work very well in language processing. It's worth learning abotu vanilla RNNs first but also checking out LSTM and GRU if you are in to NLP use cases.
