What does "serving" a model mean?
Options
- Training it on more data
- Storing the training dataset
- Making the trained model available so that other systems can send it data and get predictions (Correct option)
- Printing its metrics
Answer
Making the trained model available so that other systems can send it data and get predictions
Why
Serving is the step that puts a model to work. It can be a web API that answers each request, a job that scores a table overnight, or code inside an app or device.
The serving side has its own concerns: latency, throughput, availability, cost and versions.
Training and serving are separate stages and often run on different machines.
Remember
Serving = putting the model where it can make predictions.