build a recurrent neural network for part-of-speech tagging problem
tensorflow(>r0.10) nltk
I use a bidirectional lstm-rnn to finish this task as well as a bucket queue The data is from nltk so if you want to run this model, make sure you have already installed nltk
after training for 10 minutes in a single tesla K20m, this model achieved a peak accuracy about 92%
add multi gpu train