Knowledge Base Training and Testing Quiz

5 questions Pass: 70% +25 pts (50 with Pro)

Quiz covering Custom Language Models

What you'll be asked

  1. What is the primary purpose of a 'holdout set' when training a custom language model?
  2. When fine-tuning a custom model on a specialized domain, what is a common sign of 'catastrophic forgetting'?
  3. Which of the following is a best practice when preparing a text corpus for domain-specific model training?
  4. You are evaluating your model and notice the training loss is decreasing while the validation loss is increasing. What is the most likely cause?
  5. When implementing a Retrieval-Augmented Generation (RAG) system instead of full model fine-tuning, which architectural constraint are you primarily addressing?

Answers and explanations come after you submit.