LSTM categorical crossentropy validation accuracy remains constant

Question

I\'ve been doing some LSTM network lately, and I\'d like to predict a one-hot encoded output (2 classes for now, I already tried with binary cross entropy and my problem st