Xgboost 1.7.6 - ValueError: Feature shape mismatch

Hi all,
with new version I am getting the following error
ValueError: Feature shape mismatch, expected: , got
with the UCI US census data
while calling model.predict(X) with a loaded pickle model.

According this issue 7712 it seems to be solved. But I dont’ know why it happens then.
Using a much older version, there it does not happen.

Does someone know a reason and/or solution?
Help is appreciated. Thanks in advance!

Hi all,
after debugging I found the reason, sorry it has been my fault pickle the model with wrong status … now everything is fine with newest XGB library.