Optimization algorithm

[FernanOrtega]

Hi Garrett, I'm a newbie in Tensorflow and I found you notebook quite useful to understand how to build a LSTM network. Thanks for it 👍

I've a doubt regarding the optimization algorithms in this problem. In your notebook, you used AdaDelta for optimization. I tried others like Adam or RMSProp. I found that AdaDelta, with 50 epochs, is the only algorithm that can converge to a solution since others were moving away from the solution. I didn't try more epochs since I found your notebook yesterday and I hadn't so much time for it but I wish to know if you can explain why AdaDelta is the only algorithm that converges to a relatively good solution.

Thanks in advance.