How to get the .scores file?

[JZX555]

I'm trying to use Lingeval97 on my NMT system, but it seems need .scores file. I can't find how to get this file in README or the origin paper. How to get the .scores file by NMT system?

[rsennrich]

This will depend on your NMT toolkit. Nematus provides the script score.py for this, but not all toolkits provide an interface to obtain sentence-level scores for a parallel corpus. If they don't, you can check how to get the perplexity for a validation set (which most toolkits support) and see if you can adapt this functionality.