The dataloaders have a path parameter, which by default is the current directory. This base path is prepended to each data element upon loading. But what if the data items are already specified with their full paths in the function or data frame, i.e. no base path should be added? Explicitly passing None to the path argument doesn’t work because the loader wants to build a path object and None gives an error. What is the way to bypass that (other than removing the path prefixes in the data frame)?