Across framework data loading: Keras to PyTorch and Vice Versa

Sometimes, we need to try things using packages in some other frameworks, like Keras, and data loading is certainly different from PyTorch.

In simple words, I have done quite a few Dataset classes for my work, but now, I want to try them on some Keras package and do not want to spend time to write my_data.load_data(), a function that loads data in Keras, for each of them.
Since we do a lot of work with PyTorch, I wonder if there is a simple way, or future planning, to include across framework data loading.