mirror of
https://github.com/rohitg00/ai-engineering-from-scratch.git
synced 2026-10-02 01:54:39 +08:00
omni-moderation-latest returns 13 category booleans, not 8. Update the lesson tl;dr, category list, Key Terms table, Llama Guard vs OpenAI comparison framing, code CATEGORIES docstring, and the moderation-layers SVG to reflect the real taxonomy. Keep the top-level eight in code/main.py but mark it explicitly as a pedagogical collapse.