It was just reposted a few weeks ago, but I really like Andrej Karpathy's "The Unreasonable Effectiveness of Recurrent Neural Networks": http://karpathy.github.

It was just reposted a few weeks ago, but I really like Andrej Karpathy's "The Unreasonable Effectiveness of Recurrent Neural Networks": http://karpathy.github.io/2015/05/21/rnn-effectiveness/ http://karpathy.github.io/2015/05/21/rnn-effectiveness/