> [...] for a few publicly funded PDFs is a bit drastic.
It was 4.8 million PDFs, which I think is a bit more than "a few".
In today’s world, this is both a few and fair use. You just need to claim that you’re going to train models for research purposes.
In today’s world, this is both a few and fair use. You just need to claim that you’re going to train models for research purposes.