We aren't ready for pandas 3.0 yet. In case of `TypeError: Cannot interpret '<StringDtype(na_value=nan)>' as a data type`, downgrade to pandas<=2.3. I'll make a new release requiring pandas<3 for now. Will fix it probably next week.