Toolkit for Machine Learning, Natural Language Processing, and Text Generation, in TensorFlow. This is part of the CASL project: http://casl-project.ai/
Hi,
I wonder if it is possible to evaluate the model on a set of development set during training (to save the best model) based on BLEU score on BPE token ids? Or should we necessarily detokenize and decode back to natural text for bleu score calculation during training?
Hi, I wonder if it is possible to evaluate the model on a set of development set during training (to save the best model) based on BLEU score on BPE token ids? Or should we necessarily detokenize and decode back to natural text for bleu score calculation during training?
Thanks