I get the same error message when I try to load the pickled model through mlflow. The model I am trying to download is from xgboost framework but somehow it is giving me sklearn error. My sklearn is 1.5.2 version but the same error as mentioned above still persists.

Error loading model: 'super' object has no attribute 'sklearn_tags'

channels:

• conda-forge dependencies:
• python=3.9.21
• pip<=24.2
• pip:
  • mlflow==2.19.0
  • numpy==2.0.1
  • pandas==2.2.3
  • psutil==6.1.1
  • scikit-learn==1.5.2
  • scipy==1.13.1
  • xgboost==2.1.3 name: mlflow-env