[FEA][Python] Multi-GPU training with DeviceQuantileDMatrix #5583
Description
A helpful feature would be if the newly added DeviceQuantileDMatrix could be extended to Multi-GPU/Multi-node training via dask backend.
{{ message }}