xgboost.core.XGBoostError: [21:45:44] /root/repo/xgboost/src/gbm/gblinear.cc:119: Check failed: this->updater_:

[sh1ng]

Can't provide end-to-end example.

models.zip

import pickle
import xgboost

model = pickle.load(open("xgboost.pkl", "rb"))

print(model) # removing it solves the issue
model.get_booster().load_model('gblinear_model.bin')

pickle.dump(model, open("model.pkl", 'wb'))

Looks like incorrect internal state initialization

File "/home/sh1ng/dev/.venv/lib/python3.6/site-packages/xgboost/core.py", line 189, in _check_call
    raise XGBoostError(py_str(_LIB.XGBGetLastError()))
xgboost.core.XGBoostError: [21:45:44] /root/repo/xgboost/src/gbm/gblinear.cc:119: Check failed: this->updater_: 
Stack trace:
  [bt] (0) /home/sh1ng/dev/.venv/lib/python3.6/site-packages/xgboost/./lib/libxgboost.so(dmlc::LogMessageFatal::~LogMessageFatal()+0x6a) [0x7f2b3526f7ba]
  [bt] (1) /home/sh1ng/dev/.venv/lib/python3.6/site-packages/xgboost/./lib/libxgboost.so(xgboost::gbm::GBLinear::SaveConfig(xgboost::Json*) const+0x3bf) [0x7f2b3531751f]
  [bt] (2) /home/sh1ng/dev/.venv/lib/python3.6/site-packages/xgboost/./lib/libxgboost.so(xgboost::LearnerImpl::SaveConfig(xgboost::Json*) const+0x346) [0x7f2b35352846]
  [bt] (3) /home/sh1ng/dev/.venv/lib/python3.6/site-packages/xgboost/./lib/libxgboost.so(xgboost::LearnerImpl::Save(dmlc::Stream*) const+0x104) [0x7f2b35346bc4]
  [bt] (4) /home/sh1ng/dev/.venv/lib/python3.6/site-packages/xgboost/./lib/libxgboost.so(XGBoosterSerializeToBuffer+0x9b) [0x7f2b35274e6b]
  [bt] (5) /usr/lib/x86_64-linux-gnu/libffi.so.6(ffi_call_unix64+0x4c) [0x7f2b88fccdae]
  [bt] (6) /usr/lib/x86_64-linux-gnu/libffi.so.6(ffi_call+0x22f) [0x7f2b88fcc71f]
  [bt] (7) /home/sh1ng/dev/.venv/lib/python3.6/lib-dynload/_ctypes.cpython-36m-x86_64-linux-gnu.so(_ctypes_callproc+0x2b4) [0x7f2b891e05c4]
  [bt] (8) /home/sh1ng/dev/.venv/lib/python3.6/lib-dynload/_ctypes.cpython-36m-x86_64-linux-gnu.so(+0x11c33) [0x7f2b891e0c33]

[trivialfis]

@sh1ng Out of curious, do you use glm from XGBoost?

[sh1ng]

Yes, we do.