This repo contains a dataset of 1 million PubMed abstracts and corresponding keywords for each of them. You can use this dataset for fine-tuning your models
You can download this dataset from this link: https://drive.google.com/file/d/1CDYeUQn9WZJxzl1KM0nJYdRySn3wgZZu/view?usp=sharing