I am using BERT model for finetuning for a simple classification task.
However, I want to add words which are very frequent to our domain which are not present in the original vocab.txt.
When I added the words in the ‘unused x’ placeholders in the original vocab file ( vocab.txt and trained using that), it is not recognising the added words.
Can you please guide as to what is the proper way to do it ?
Regards,