Done, you've completed the course
Thanks for taking this course. Six modules and a surprising amount of LLM fundamentals later, we hope “safety layer” has shifted from a term to a working attitude.
Among other things, you've learned to:
- Explain in one sentence what an LLM is, and where tokens, embeddings and attention come from.
- Distinguish pre-training, SFT and RLHF, and know why RAG is often safer than a “medical model”.
- Compare the players (ChatGPT, Claude, Gemini) with Mistral as the EU route and DeepSeek as a sovereignty case.
- Build clinical prompts following role + context + task + format + uncertainty, providing source instead of asking source.
- Recognise hallucinations, GDPR traffic-light and automation bias, and apply the four rules of thumb.
Want to leave a quick review?
Your experience helps fellow cardiologists and trainees. Reviews appear (anonymously or with first name) on the course landing page. Two minutes.
Thank you!
Your review has been sent. Now grab the assignment book to deepen your new knowledge.
Next step: the assignment book
12 extra exercises at cardiology level: differential diagnosis, MDT preparation, guideline summary, patient information and hallucination hunt. You do them at your own pace, tick off what you've done, and can keep notes in the page itself (stored on your device).