I have a model which I fine-tuned on Nllb-200-distilled-600M. However, I did expand the vocabulary which I later used to train. After that, the model works. However, when converting the same model to CTranslate2, it no longer works. Why?
I have a model which I fine-tuned on Nllb-200-distilled-600M. However, I did expand the vocabulary which I later used to train. After that, the model works. However, when converting the same model to CTranslate2, it no longer works. Why?