I am aware of the scoring functions provided in the tf version which can be used to score a sentencepair with log-likelihood given a trained model as on the following page
onmt-main \ --config config/my_config.yml --auto_config \ score --features_file newstest2014.en.tok \ --predictions_file newstest2014.en.tok.out
Is there a similar feature available also with opennmt-py. I couldn’t find. Please help.