AI Quiz - Perplexity

What does “GPT” stand for in models like GPT-4?

• Generative Pre-Trained Transformer

• General Purpose Transformer

• Generative Predictive Transformer

• Generalized Pre-Trained Tokenizer

Which paper introduced the transformer architecture that powers most modern LLMs?

• Sequence to Sequence Learning with LSTM

• Attention Is All You Need

• Neural Machine Translation by Joint Attention

• Deep Bidirectional Transformers for Language Understanding

In a Retrieval-Augmented Generation (RAG) pipeline, the retriever’s job is to:

• Rank responses from multiple models

• Cache frequently asked questions

• Retrieve external documents relevant to a user query

• Generate contextual embeddings from retrieved data

What’s the main advantage of calling AI APIs through Next.js API routes instead of directly from the client?

• Reduces frontend bundle size

• Prevents token overuse automatically

• Keeps API keys secure and off the client side

• Doubles the response speed

Which of the following best describes the softmax layer in a transformer?

• Balances token frequencies in context

• Converts raw logits into probability distributions

• Normalizes embeddings before training

• Controls gradient clipping

What’s the goal of RLHF in large language models?

• Reduce model size through quantization

• Increase training stability by freezing layers

• Fine-tune model weights for factual accuracy

• Reinforce model outputs that align with human preferences

In LangChain, what does a Chain enable?

• Combining multiple reasoning steps or tools in sequence

• Generating real-time embeddings from prompts

• Storing long-term user memory

• Caching embeddings for faster retrieval

Which of the following companies released an open-weight LLM in 2024?

• Meta

• OpenAI

• Anthropic

• Google DeepMind

What is the key purpose of tokenization in LLMs?

• Compress input data for GPU efficiency

• Identify part-of-speech tags

• Split text into subword units mapped to numerical IDs

• Remove irrelevant punctuation

Which of these metrics best measures semantic similarity between two texts?

• ROUGE

• BLEU

• BERTScore

• Perplexity