Recently I crossed to a situation I can't figure it out why it happened. I applied six predictive models on a specific dataset as training-set and tried to predict the other similar dataset as an unseen test-set. I realised that GRU predicted the other way around compared to LSTM but LSTM performance, whether on Loss graph or extra evaluation by mapping the predicted matrice values satisfying. Also, MSE and MAE are listed in table form Keras. I know in the scripts also there is no significant difference between GRU and LSTM when I used keras. So why GRU's performance in my visualisation evaluation doesn't match with information from keras. Based on the table, even GRU should have predicted better or as good as LSTM.