mljet
mljet copied to clipboard
Make more common pydantic model validation for request data
At this point in every backend's implementation, the query model has a data field that has the following type:
data: List[List]
Because the data can be very diverse:
- Arrays of int/double types, arrays of arrays
- Arrays with categorical features
- Tensors
- Dataframes with named columns
- Other
You need to decide which is the most correct and common type to choose. We may also need to dynamically generate validation model data.
This task is subject to careful discussion.
CC: @pacifikus