About Evaluate LRP for your dataset - Clarification on data selection and preprocessing

[zhongmz]

Hello

First and foremost, I would like to express my gratitude to you for this outstanding work

I am interested in evaluating LRP for my dataset, and I have a couple of questions regarding the data selection and preprocessing steps. I would appreciate it if you could provide some clarification on the following points:

1. When selecting the data for evaluation, is it necessary to choose source sentences with equal lengths and target sentences with equal lengths? Or can the sentences have varying lengths?

2. Should the selected data be preprocessed with BPE tokenization, or is it supposed to be the raw test set without any tokenization applied?

Thank you in advance for your assistance! Looking forward to your response.

Best regards,