I noticed that you conducted a few different types of tasks from different data . How similar do the different tasks need to be for the model to be accurate? Could you combine image classification tasks from two different datasets in the same model? What about, for example, a natural language processing task and an image classification task? If these models still need to be somewhat specialized, have you thought about how you might broaden their predictive abilities even further?
I noticed that you conducted a few different types of tasks from different data . How similar do the different tasks need to be for the model to be accurate? Could you combine image classification tasks from two different datasets in the same model? What about, for example, a natural language processing task and an image classification task? If these models still need to be somewhat specialized, have you thought about how you might broaden their predictive abilities even further?