GRU network is one of LSTM improvements, we have compared their structure in this tutorial.
Understand GRU (Gated Recurrent Unit): Difference Between GRU and LSTM
However, there is a problem: which network we should choose in deep learning?
It means which netwrok has better performance in LSTM and GRU?
As to this question, we can find answer in paper:
An Empirical Exploration of Recurrent Network Architectures
In this paper, authors have compared the performace of GRU and LSTM in some experiments, they found:
The GRU outperformed the LSTM on all tasks with the exception of language modelling
Which means as to language modelling (Use LSTM for NLP), you should choose LSTM, otherwise, GRU is a better choice.