how to do some pre-processing for the berttopic?

I understand that bert somehow did not need the pre-processing, but I did not want some words identified as a topic because the goal of my topic modeling. How can I achieve this?

The issue names no files, tests, or entry points. Start by locating BERTopic's preprocessing and topic-word selection entry points, then determine where the requested word exclusion behavior should be documented or tested.

Written by the indexing model from the issue text.