Currently there are a few things are not convinent:
When we create a dataset we only can create a empty dataset first. Then we can use load function to decide if we load the existing dataset or create a new dataset.
When we use dataset.add_data we always need to set the dataset path. This is not good, because if there is no existing data we don't kown where the dataset path. And also, when we use this function, there is already dataset loaded in program, so why we need to provide a dataset path here. logic error.
Description
Currently there are a few things are not convinent:
Solution
Environment details:
Additional context
Add any other context about the problem here.