Getting different prediction output when using Nyoka to save my xgboost model and PyPMML to load the model for prediction

[Digbijaynk01]

Hi,
I am using nyoka package to generate the PMML file. nyoka version '5.5.0'.
use below pypmml package to load the model in my python code to do prediction.
What is observe there is huge difference in prediction result

Output of one row for cmparision;
PMML model prediction : Probabilti-0 : 0.142825, Probabilti-1 : 0.857175
same row prediction from model directly: Probabilti-0 : 0.516752, Probabilti-1 : 0.483248
I have checked non of feature have null value, exactly same data and same sequence of feature also used for prediction.

Please can you help me check the issue
Thanks
Dibijay