User interface for machine translation, and restful api implementaion.
This repo is an angular project, you can build and setup the frontend following these guidelines.
The server folder were added for the backend implementation, you can build and setup the backend following these guidelines.
To add a new language to the server, you should include your ctranslate2 model under the /models/{source langauge code-target language code}/ folder (i.e /models/en-fr/ for english and frensh), this should also map to the frontend in angular as shown in the frontend guideline.
For sentencepiece tokenoization, it can added like this: /sentencepiece/{language code}.model and /sentencepiece/{language code}.vocab (i.e /sentencepiece/de.model and /sentencepiece/de.vocab to add trained german sentencepiece models)
The util folder will include the preprocess.py for each language pair under their own folder /util/{source langauge code-target language code}/preprocess.py, please look at already existing examples.
The starred folder is where edited/good translation is sent from the users via the frontend, these are tab-limited formatted as {source language sentence {tab} target language sentence}, (i.e /starred/zh-jp.txt is where you would find translations from chinease to japanease added from user's inputs)