Generative AI & LLM – interview notes
1. What is a token?
A chunk of text (often part of a word) that the model reads and writes. Context limits and pricing are counted in tokens.
2. What is an embedding?
A vector of numbers representing meaning. Similar texts have vectors that are close together, which powers semantic search.
3. RAG vs fine-tuning?
RAG retrieves relevant documents at question time and adds them to the prompt – best for changing or private knowledge. Fine-tuning changes the model's weights – best for style, format or a narrow task.
4. What causes hallucinations and how do you reduce them?
The model predicts plausible text, not verified facts. Reduce them with grounding (RAG), asking for citations, lower temperature and evaluation sets.
5. What does temperature control?
Randomness of sampling. Low values give focused, repeatable answers; higher values give more varied, creative output.
Feedback & comments (2)
Helpful, thanks. Please write a follow-up on the advanced topics.
Great summary. It would be nice to add one more example with real data.