Done it a few months ago but then again if I was working at reddit and in charge of preparing the dataset to feed to the llm, I’d give it access to both a recent one and a snapshot from before July 2023 (or whenever shit hit the fan and we all came to lemmy), most edits would have been made in protest. And AI can figure out which ones by itself