Getting phrase tables from fast_align outputs

[alvations]

Thanks for the amazingly fast alignment tool, I understand that fast_align gives forward and backwards word alignments.

1. How do we do grow-diag-final-and on the forward an backwards output?
2. How do we get phrase tables from the outputs?

A short tutorial-like answer to the 2 questions would help a lot for users to use fast_align in the MT toolkit pipelines.

[alvations]

Use phrase_based.py from Eliezer : https://github.com/alvations/Eliezer/blob/master/eli/align/phrase_based.py

[alwayshjia]

Use phrase_based.py from Eliezer : https://github.com/alvations/Eliezer/blob/master/eli/align/phrase_based.py

@alvations Hi, the link you mentioned here doesn't exist now. Could you provide the code of phrase_based.py please?

[alvations]

Try this https://github.com/alvations/Eliezer/blob/89c2858c5f058360a583be53577eede955ae852c/eli/align/phrase_based.py there is also a version from nltk at https://www.nltk.org/_modules/nltk/translate/phrase_based.html