Errors when getting dump from the xgboost model exported from treelite about treelite HOT 3 CLOSED

I don't experience this problem

from treelite.

Actually, I do experience this problem now with a different dataset.

In the exported Booster feature_names and feature_types are not set. So, I can't dump the model. It returns this trace:
bst.get_dump(with_stats=True)
File "/home/denper/xgboost/python-package/xgboost/core.py", line 1183, in get_dump
if self.feature_types is None:
AttributeError: 'Booster' object has no attribute 'feature_types'

I checked and indeed feature_types doesn't even exist. I'm not sure how is that even possible.

But if I set those manually after export_as_xgboost, dump works.

from treelite.

export_as_xgboost() will be moved to experimental branch, so that development efforts can be focused to core features.

from treelite.