Train on GPU, Run on CPU


#1

I’m using xgboost 0.81 in python 3.6

Is it possible to train on GPU (desktop) and run on CPU (laptop)?
I ask because when I run my GPU-trained joblib/pickled files on CPU, the results are wacky.

Thanks!


#2

My code seems to running fine now but getting the following warning. Should I worry? I fitted the models using GPUs on my desktop but running it off CPU (no GPU in laptop).

d:\build\xgboost\xgboost-0.81.git\src\learner.cc:377: Parameter ‘predictor’ will be set to ‘cpu_predictor’ since XGBoots wasn’t compiled with GPU support.