Change8

v2.1.4

📦 xgboostView on GitHub →
3 features🔧 1 symbols

Summary

Patch 2.1.4 adds scikit-learn 1.6 compatibility, CUDA 12.8 wheel support with Blackwell, and updates for RMM 25.02 logger changes.

Migration Steps

  1. Reinstall XGBoost using the new wheels (e.g., pip install xgboost==2.1.4) to get CUDA 12.8 support.
  2. Verify downloaded artifacts using the provided SHA256 checksums.
  3. If using RMM, ensure you are on RMM 25.02 or later to match logger changes.

✨ New Features

  • Added compatibility with scikit-learn 1.6.
  • Built wheels supporting CUDA 12.8 and Blackwell.
  • Updated integration to work with RMM 25.02 logger changes.

Affected Symbols