Model Registries and Versioning

1. What is the purpose of a model registry in AI systems?

To store model weights only To track model identity, provenance, and configuration To automatically update models To generate embeddings

2. Why is it important to version models in a registry?

To increase storage space To prevent accidental changes to models To ensure models can be upgraded continuously To compare outputs from different models easily

3. What happens if a model is upgraded without re-embedding the corpus in the context of embeddings?

Nothing, as the models are compatible The performance may degrade without an error being thrown The system will automatically re-embed the corpus It will create new vectors with no historical link

4. In the context of LyraLearn, what is tracked alongside every stored vector?

The model name only The active embedding-model version The user that queried the model The training data used for the model

Continue: Human-in-the-Loop and Accountability → Review this lesson Retake quiz