Which architecture underlies most modern Large Language Models (LLMs)?
Istqbctgenai Foundations Category
Question · Easy
Spot the AI hallucination
Short
Transformer architecture
GenAI Foundations
Flashcard
What is tokenization in the context of LLMs?
Unit produced when text is split for a model, which may be a whole word, a word fragment or a single character and counts against the context limit.
GenAI foundations
Guess the Term
GenAI Foundations for Testers
1.1 Transformers, tokenization, embeddings & context window
Study guide
Cheat sheet
What does the 'context window' of an LLM refer to?
Read the clue and guess the term in 6 tries.
Game
Tokenization
Which sampling parameter controls the randomness of an LLM's outputs?
Compact system with fewer parameters than its larger counterparts, intended for lightweight, focused tasks.
What does RAG stand for?
Embeddings
A tester uses an LLM to draft test cases from a requirements document. What does it mean for the output to be 'faithful' (grounded) to that document?
Istqbctgenai Evaluation Category
Question · Medium
Written request to a generative system that can carry a role, context, constraints and a requested output format.
Instruction design
1.2 Foundation models, temperature/sampling & prompting basics
Which is the BEST definition of a 'hallucination' in GenAI?
Istqbctgenai Risks Category
Swap tiles to line up three of the same term.
Term Match
Context window
What is 'prompt injection'?
Kind of instruction set fixed at the start of a session that shapes tone, rules and boundaries, usually hidden from the chat participant.
Which is an example of INDIRECT prompt injection?
Question · Hard
KV cache
Which prompt-engineering technique asks the model to reason step-by-step before producing an answer?
Kind of message typed by the person chatting that changes every turn and forms the immediate request.
1.3 RAG & agent architectures testers must understand
What does 'few-shot prompting' mean?
Flip two cards at a time to find matching terms.
Term Pairs
Foundation models
Which is the BEST example of an LLM-as-judge evaluation?
___ prompting has the model draft or sharpen the instructions it will then follow.
Which classical NLP metric is most appropriate for evaluating machine translation?
Temperature vs top-p vs top-k
Which metric is most commonly used for summarization evaluation?
Evaluation dimension such as accuracy, diversity or execution success used to judge generated testware.
Quality Attributes for GenAI
2.1 Faithfulness, factuality, relevance, coherence & fluency
Why should a tester treat 'fluency' and 'correctness' as separate qualities of GenAI output?
Istqbctgenai Quality Attributes Category
Find five hidden terms in a grid of letters.
Term Search
Few-shot prompting
Prompt Engineering
A tester wants an LLM to generate Python Selenium scripts for a regression suite. Which prompt-engineering practice MOST improves the reliability of the generated code?
Istqbctgenai Prompt Engineering Category
Completeness measure: how much of the relevant material a model's output identifies or covers.
Which regulatory framework defines four risk tiers for AI systems (unacceptable, high, limited, minimal)?
Chain-of-thought (CoT) prompting
Which framework provides a voluntary structure of Govern, Map, Measure, Manage for AI risk?
Loading more…