Chapter 3 of 4 · Google Cloud Generative AI Leader
Techniques to improve gen AI model output
Model limitations, monitoring, prompt engineering, grounding and RAG, and sampling parameters.
- 5topics
- 19 minfast track
- 2diagrams and exhibits
- 20%of the exam
Topics in this chapter
Take them in order, or jump to the one you need.
- 1Foundation Model Limitations and How to Address ThemFoundation model limits such as hallucinations, knowledge cutoff and bias, and the fixes: grounding, RAG, prompt engineering…Diagram4 min
- 2Monitoring and Evaluating Gen AIGoogle-recommended practices for monitoring and evaluating gen AI: model upgrades, KPIs, security patches, versioning…4 min
- 3Prompt Engineering TechniquesPrompt engineering for LLMs: zero-shot, one-shot, few-shot, role prompting, prompt chaining, chain-of-thought and ReAct…Diagram4 min
- 4Grounding and RAGGrounding LLMs in first-party, third-party and world data, how RAG changes output, and Google Cloud's grounding offerings…4 min
- 5Sampling Parameters and SettingsHow token count, temperature, top-p (nucleus sampling), safety settings and output length control gen AI model behavior, with…3 min
About this chapter
What this chapter covers
Foundation models are powerful but imperfect. This section, worth about 20 percent of the exam, covers how to overcome foundation model limitations (limitations, recommended practices, continuous monitoring and evaluation), prompt engineering techniques (zero-shot to ReAct) and grounding techniques (types of grounding, RAG, Google Cloud's grounding offerings and sampling parameters).
The exam wants you to pick the right technique for a stated problem, such as outdated answers, invented facts or inconsistent formatting.
How the topics fit together
- Foundation model limitations and how to address them covers the limitations and the recommended practices.
- Monitoring and evaluating gen AI covers continuous monitoring and evaluation practices.
- Prompt engineering techniques covers zero-shot, one-shot, few-shot, role prompting, prompt chaining, chain-of-thought and ReAct.
- Grounding and RAG covers grounding sources, how RAG changes output, and Google Cloud's grounding offerings.
- Sampling parameters and settings covers token count, temperature, top-p, safety settings and output length.
Match the fix to the cause. Missing or outdated knowledge calls for grounding. Wrong format or style calls for prompting or tuning. High-stakes decisions call for a human in the loop.
How to use these pages
Each topic page has the same shape, so you can skim for what you need.
- What it is: the idea in plain words.
- How it works: the mechanism, with examples and tables.
- What to remember for the exam: the facts to memorise and how questions are phrased.
- Common mistakes: the classic traps.
- Check yourself: three questions to test your understanding.
Ready for more practice?
The full Google Cloud Generative AI Leader question bank has 512 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by Google Cloud. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.