Hi, recently I raised an issue regarding Pyspark XGB Ranker and got response very quickly, while I got the hotfix branch and was going to test it, I could not build from source successfully although I strictly followed the installation guidance, my bash script is here,
"
git clone --recursive https://github.com/wbo4958/xgboost.git
cd xgboost
git checkout ranker
mkdir build
cd build
cmake …
make -j$(nproc)
cd …/python-package
pip install -e . # or equivalently python setup.py develop
"
However, I got the error when I import xgboost
Traceback (most recent call last):
from xgboost.spark import SparkXGBRanker
File “/home/…/xgboost/python-package/xgboost/spark/init.py”, line 9, in
from .estimator import (
File “/home/…/xgboost/python-package/xgboost/spark/estimator.py”, line 9, in
from .core import ( # type: ignore
File “/home/…/xgboost/python-package/xgboost/spark/core.py”, line 54, in
from .params import (
File “/home/…/xgboost/python-package/xgboost/spark/params.py”, line 7, in
class HasArbitraryParamsDict(Params):
File “/home/…/xgboost/python-package/xgboost/spark/params.py”, line 14, in HasArbitraryParamsDict
arbitrary_params_dict: Param[dict] = Param(
TypeError: ‘type’ object is not subscriptable
Could anyone help with this issue? Please correct me if you see anything inappropriate! Thanks!