So, could this be an example of an LLM trained fully on public domain copyright-expired data? Or is this not intended to be the case.