@the_wilderless there was some mention at some point on twitter about developing "watermarks" for LLM-generated content to avoid contaminating the training data but v unclear how this would work in…

Posters: QiaochuYuan. QiaochuYuan

2 likes recorded by the source corpus.