How to i get word embeddings for out of vocabulary words using a transformer model?

Question

When i tried to get word embeddings of a sentence using bio_clinical bert, for a sentence of 8 words i am getting 11 token ids(+start and end) because "embeddings"