finBERT on large dataset

Hi. I have a quite large dataset with three column: date, stock name and headline of a news related to the stock. I would like to apply finBERT to the headline in order to obtain a sentimental score, which will be used to evaluate a sentimental analysis.
The problem is that the dataset has ~1 million rows, which makes this task impossible. Hence, my question is: does anyone know a possible way to approach to this problem? What can I do in order to run this all model on my dataset?