Topics Practice Study

Zenaique

Personalized AI interview prep. Structured questions across every GenAI sub-discipline. Practice, score, master.

Practice

Topics
Practice
Challenges
Dashboard

Browse

Formats
Roles
Companies
Top questions

Product

Pricing
Sign in

© 2026 Zenaique

Formats/Match Pairs

Match Pairs

114 questions

Sort

Questions

Tokenization
Embeddings
Transformer Architecture
Attention Mechanism
Pretraining
Fine-Tuning

Showing 1–25 of 114

← PrevNext →

Character, word, and subword tokenization each have a dominant failure mode. Match them.

Match PairsEasy

Match each similarity metric to its formal definition

Match PairsMedium

Match each contrastive loss to its defining mechanism

Match PairsMedium

Match each use case to the embedding provider that best fits it

Match PairsMedium

Match each embedding provider to a defining feature or pricing fact

Match PairsMedium

Match each embedding storage reduction technique to its compression ratio and quality cost

Match PairsHard

PremiumMatch each positional encoding…

Match PairsMedium

PremiumPair each 2017 transformer…

Match PairsHard

PremiumMatch each modern architecture…

Match PairsHard

Pair each transformer family with its canonical pretraining objective

Match PairsEasy

PremiumMatch each attention variant…

Match PairsHard

Match each transformer family to the kind of attention it uses

Match PairsEasy

Match each sparse attention pattern to its design and exemplar architecture.

Match PairsHard

Match each long-context strategy to what it modifies.

Match PairsHard

Match GPT-2 versus Llama-2 attention design choices to their differences

Match PairsMedium

PremiumMatch Kaplan era and…

Match PairsMedium

PremiumMatch DDP, FSDP, and…

Match PairsMedium

Pair capability changes with the stage that mostly produces them

Match PairsMedium

Match each training stage to its objective and data signal

Match PairsEasy

Match multi-GPU training strategy to what it shards

Match PairsHard

Match each instruction dataset to its on disk row shape.

Match PairsMedium

Match instruction dataset to its source / construction method

Match PairsMedium

Match LoRA rank to task type

Match PairsMedium

Match each PEFT method to the advantage that defines it.

Match PairsMedium

Match PEFT method to the scenario it fits best

Match PairsMedium

Cursor Turbopuffer

Contextual Ai Promptlayer

Neo4j Servicenow

Neptune Ai Perplexity

Braintrust Jane Street

Anthropic Cresta

Figure Ai Ironclad

Alibaba Anthropic

Coinbase Lakera

Mistral AI Tata Digital

Arize Ai Stability Ai

Coinbase Google

Ironclad NVIDIA

Anthropic OpenAI

Anthropic Linkedin

Canva Fiddler Ai

Droom Robust Intelligence

Ironclad Midjourney

Ai4bharat Swiggy