Ted Underwood: Probably shouldn't admit this, but agentic tools make me better at data cleaning. I never used to connect books/authors to Open Library or Wikidata. If I even thought of it I would 🙄 at work involv…
Ted Underwood: Now, this is not to say Mike is wrong to worry! I bet four out of five people confronted w/ my present task will say "hey, Fable, deduplicate this list of books and web scrape something about their…
Pwnallthethings: Matters a lot *how* they are using LLMs right? Asking cc/codex to identify issues and clean them is not the same as using hallucinated or fully synthetic datasets