rucJuanLi

I also meet the problem,it used more than 2200 cpu ,I do not know why it use so much cpu.

when I do not add the attention layer,there is not any error.But when I user the attention layer,there is the error.

I think maybe you should use Theano backend.