[BUG] XGBoost Falling back to prediction using DMatrix due to mismatched devices

I am doing a tabular fit with evaluation_quality and GPU enabled as per the documentation.

I am getting this error: /home/alexko/PycharmProjects/algo-py/.venv/lib/python3.11/site-packages/xgboost/core.py:158: UserWarning: [14:45:25] WARNING: /workspace/src/common/error_msg.cc:58: Falling back to prediction using DMatrix due to mismatched devices. This might lead to higher memory usage and slower performance. XGBoost is running on: cuda:0, while the input data is on: cpu.
#[36m(_ray_fit pid=271196)#[0m Potential solutions:
#[36m(_ray_fit pid=271196)#[0m - Use a data structure that matches the device ordinal in the booster.

I am not sure of any settings where the storage for XGBoost can be changed.

The issue names no repository files, tests, or entry points. Start by reproducing the documented tabular fit with evaluation_quality and GPU enabled, then trace where the XGBoost booster and input data receive their devices. Done means identifying and documenting or fixing the mismatched-device path without the fallback warning.

Written by the indexing model from the issue text.