how to load pickle model in linux system

Question

I have built Xgb model with xgboost package in python. I saved the model using pickle and joblib which works perfectly in my windows 10 system. But it is giving error in AWS instance I am trying to run. The error seems to be related to serialization.

>>> import joblib
>>> joblib.load(xgb_low_lr_fin.sav)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'xgb_low_lr_fin' is not defined
>>> joblib.load("xgb_low_lr_fin.sav")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/ubuntu/anaconda3/lib/python3.7/site-packages/joblib/numpy_pickle.py", line 598, in load
    obj = _unpickle(fobj, filename, mmap_mode)
  File "/home/ubuntu/anaconda3/lib/python3.7/site-packages/joblib/numpy_pickle.py", line 526, in _unpickle
    obj = unpickler.load()
  File "/home/ubuntu/anaconda3/lib/python3.7/pickle.py", line 1085, in load
    dispatch[key[0]](self)
  File "/home/ubuntu/anaconda3/lib/python3.7/site-packages/joblib/numpy_pickle.py", line 339, in load_build
    Unpickler.load_build(self)
  File "/home/ubuntu/anaconda3/lib/python3.7/pickle.py", line 1549, in load_build
    setstate(state)
  File "/home/ubuntu/anaconda3/lib/python3.7/site-packages/xgboost-1.0.0_SNAPSHOT-py3.7.egg/xgboost/core.py", line 1132, in __setstate__
    _LIB.XGBoosterUnserializeFromBuffer(handle, ptr, length))
  File "/home/ubuntu/anaconda3/lib/python3.7/site-packages/xgboost-1.0.0_SNAPSHOT-py3.7.egg/xgboost/core.py", line 189, in _check_call
    raise XGBoostError(py_str(_LIB.XGBGetLastError()))
xgboost.core.XGBoostError: [10:43:02] src/learner.cc:660: Check failed: header == serialisation_header_ (

The above code works perfectly in windows and in mac os.

Answer 1

I had a similar problem when trying to load a pickle file that was pickled on a different machine, and it was related to pickle serialization being different in Python 2 and Python 3 - or maybe even in the different versions of Pickle itself.

Try and check you pickle and python versions and make them match :)

Answer 2

In my case this error was caused by my xgboost versions being slightly different. My model was trained with a CPU implementation of xgboost. After building and installing xgboost with GPU support, the model that was trained on CPU no longer loaded. Re-installing the normal version of xgboost from pypi seemed to resolve my issue.