Keras ModelCheckpoint not saving but EarlyStopping is working fine with the same monitor argument

Question

I\'ve built a model and I\'m using a custom function for validation. The problem is: My custom validation function is saving the validation accuracy in the logs dict, but Keras