Use val_precision as keras EarlyStopping criteria

Question

I want my model to stop training if the precision on validation data is not improved after 3 epochs. I compile the model with tf.keras.metrics.Precision() and when I use val