Is it possible to fine tune a Bert model using a small dataset (400 data))

Hello, I’m really new to NLP and sentiment analysis. I found the Hugging face models, which I find very useful for transfer learning.

I have a specific dataset consisting of 400 records in the Spanish language. I need to fine-tune a Bert model in order to be able to predict the sentiment (positive, neutral, negative) of a specific task (medical application). Is it possible? And if so, what are the steps? Is there any example in Python? Thank’s in advance.