Small Models and Distillation

1. What defines a Small Language Model (SLM)?

Models in the 1-14B parameter range Models that require distributed GPUs Any model less than 1 billion parameters Models that can only run in the cloud

2. What is the main purpose of knowledge distillation?

To increase the size of the student model To reduce the training time of large models To allow small models to match the performance of larger models To improve the general capabilities of the teacher model

3. Which of the following is NOT a reason to use a small local model?

Cost at volume Latency and availability Higher performance on all tasks Privacy and sovereignty

4. What technique is used when a system tries a small model first and escalates to a larger one if needed?

Model clustering Hybrid routing Cascade modeling Sequential processing

Continue: Data Is the Architecture → Review this lesson Retake quiz