As for wals algorithm,when I set all weights of the ratings to the same weight 1.0,it turns out that during each iteration the training rmse is all 0,so why it is like this??