You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
After getting the following tatoeba recipes to successfully run:
tatoeba-prepare
tatoeba-train
tatoeba-eval
I have been working on utilizing the fine tuning capabilities on the models built. In this process I noticed that for the tatoeba recipes, SRCLANGS/TRGLANGS is set to a 3 letter language code (ISO-693-3) and all other recipes (including the finetuning recipes) use a two letter language code (ISO 693-1). It is quite confusing to have to switch back and forth, and definitely causes confusion for the Makefiles when they are searching for data files. Is there anyway to make all the recipes use all the same language code for consistency?
The text was updated successfully, but these errors were encountered:
After getting the following tatoeba recipes to successfully run:
I have been working on utilizing the fine tuning capabilities on the models built. In this process I noticed that for the tatoeba recipes, SRCLANGS/TRGLANGS is set to a 3 letter language code (ISO-693-3) and all other recipes (including the finetuning recipes) use a two letter language code (ISO 693-1). It is quite confusing to have to switch back and forth, and definitely causes confusion for the Makefiles when they are searching for data files. Is there anyway to make all the recipes use all the same language code for consistency?
The text was updated successfully, but these errors were encountered: