Easy AI Engineer (LLM & Generative AI) questions

34 questions · 5 points each for a correct first answer.

1. 1.What is a tokenllm-basics, tokenizationeasy
2. 2.What text embeddings areembeddingseasy
3. 3.Effect of temperaturesampling, llm-basicseasy
4. 4.What the context window limitsllm-basics, context-managementeasy
5. 5.What RAG doesrageasy
6. 6.What a hallucination isllm-basics, reliabilityeasy
7. 7.Purpose of a system promptprompt-designeasy
8. 8.Comparing two embeddingsembeddings, vector-searcheasy
9. 9.Dot product of two vectorslinear-algebra, embeddingseasy
10. 10.Length of a vectorlinear-algebraeasy
11. 11.Softmax of equal scoressoftmax, probabilityeasy
12. 12.What softmax outputs look likesoftmax, probabilityeasy
13. 13.Temperature close to zerosampling, llm-basicseasy
14. 14.What top-k sampling doessamplingeasy
15. 15.Grouping data with no labelsml-basics, clusteringeasy
16. 16.Why keep a separate test setml-basics, evaluationeasy
17. 17.Spotting overfittingml-basics, overfittingeasy
18. 18.Computing precisionevaluation, ml-basicseasy
19. 19.Why activation functions existdeep-learningeasy
20. 20.A learning rate that is too highdeep-learning, optimizationeasy
21. 21.How rare words get tokenizedtokenizationeasy
22. 22.What LLM pretraining learnsllm-basics, transformerseasy
23. 23.What few-shot prompting isprompt-designeasy
24. 24.Asking the model to reason firstprompt-design, reasoningeasy
25. 25.The required keyword in JSON Schemastructured-output, json-schemaeasy
26. 26.What a tool definition containstool-callingeasy
27. 27.The basic agent loopagents, tool-callingeasy
28. 28.What MCP is formcp, tool-callingeasy
29. 29.What a vector database doesvector-search, rageasy
30. 30.Which words BM25 rewardsbm25, searcheasy
31. 31.What a golden set is forevaluationeasy
32. 32.Handling an HTTP 429 responserate-limits, reliabilityeasy
33. 33.Customer data and a hosted LLMpii, safetyeasy
34. 34.Parsing a JSON reply in Pythonpython, structured-outputeasy