CHAPTER 03 / MODEL AND ACTIVITY
What is a model, and how is it developed?
Models and model development
A machine-learning model is a learned mathematical representation used to make predictions or generate outputs. Training adjusts its parameters; inference uses the trained model on new input.
Learning from examples
Parameters are adjustable values within a model. During training, an algorithm changes them to improve performance against an objective using training data. Different model structures suit different tasks; AI includes much more than large language models.
Development is a cycle
Developers prepare data, select a model, train or adapt it and evaluate it. Fine-tuning continues training for a narrower purpose. Evaluation uses examples and measures chosen for the intended task. Good performance on a test does not establish reliable behavior in every new situation.
Using a trained model
Deployment makes a model available for use. During inference, software prepares input, runs computation and returns output. Retrieval can supply information in the input without changing learned parameters. Serving cost, response time and task quality all affect deployment choices.
Keep this connection in mind.
Training changes learned parameters. Inference uses them. Retrieval supplies context; it is not automatically retraining.
- Model-based applications uses Models
- Machine-learning models trained using Training data
- Deployed models served by Serving software
Sources for this introduction
- Google · Machine Learning Glossary ↗
Models, parameters, training, inference, datasets and evaluation.
Draft teaching text, prepared with AI assistance. Larry’s editorial review is pending. Read the editorial approach.
EXPLORE RELATED AREAS