Hello. I would like to fine-tune the Turkish pre-trained model on my own dataset. However, the language I will use is different - Azerbaijani. While there are a lot of similarities between the two languages, it is still a different language. So I am wondering how I could set up the fine-tuning so that it adapts to the new language while getting the benefits of transfer learning? The main things that would be changed are:
Letters - there are a few more in Azerbaijani language.
Espeak-ng phonemiser - Azerbaijani language is available there.
I am doing the training in Google Colab, and could not find how to change the above 2. Thanks beforehand!
Hello. I would like to fine-tune the Turkish pre-trained model on my own dataset. However, the language I will use is different - Azerbaijani. While there are a lot of similarities between the two languages, it is still a different language. So I am wondering how I could set up the fine-tuning so that it adapts to the new language while getting the benefits of transfer learning? The main things that would be changed are:
I am doing the training in Google Colab, and could not find how to change the above 2. Thanks beforehand!