GenAI-LLMs Quiz for CS Advocates

What is the purpose of embeddings in Large Language Models (LLMs)?

• Capture semantic relationships between words

• Assign weights to input tokens

• Fine-tune on specific domains

• Predict missing tokens

What is the function of the attention mechanism in LLMs?

• Learn embeddings during pre-training

• Retrieve relevant information from knowledge bases

• Break down raw sequences of text into tokens

• Allow the model to focus on relevant parts of the input sequence

What is the process of training an LLM on a large, unlabeled corpus called?

• Tokenization

• Pre-training

• Transfer learning

• Fine-tuning

What technique can be used to customize LLMs for specific users or situations?

• Adaptive Decoding

• Fine-tuning

• Prompt Engineering

• Control Codes

What is the main risk associated with LLMs related to data security and privacy?

• Risk of data breaches and exposure of sensitive customer information

• Limited understanding of prompts

• Operational challenges affecting efficiency

• Hallucinations and errors leading to misinformation

What is the purpose of the Transformer architecture in LLMs?

• Generate educational materials

• Automate clinical documentation

• Handle routine customer inquiries

• Maintain the encoder-decoder model

What is the concept of Transfer Learning in LLMs?

• Generate educational materials

• Automate clinical documentation

• Leverage pre-trained models to improve performance on downstream tasks

• Handle routine customer inquiries

What is the term used when an LLM produces a response that is not related to the input prompt?

• Vectors

• Embedding

• Semantic Chunks

• Hallucination

How can LLM malfunctions be addressed effectively?

• Regularly audit and monitor LLM interactions

• Define roles, responsibilities, and escalation procedures

• Set up monitoring tools to detect anomalies

• Update and fine-tune the LLM

What is the significance of pre-training in Large Language Models (LLMs)?

• Generate educational materials

• Automate clinical documentation

• Enhance model performance on downstream tasks

• Handle routine customer inquiries