Model Training · Level 2 of 5
Training
Adjusting a model using data and a learning objective.
Training changes learned parameters; running an already trained model need not change them.
Example
A network updates its weights after measuring prediction error.
Listen to the definition and example
Audio transcript
Training. Adjusting a model using data and a learning objective. Training changes learned parameters; running an already trained model need not change them. For example: A network updates its weights after measuring prediction error.
Explore this concept
A useful analogy
Practice that changes the model’s adjustable settings.
Why it matters
This helps you read a training loop and diagnose what is changing during learning.
Technical detail
An optimizer updates θ using an objective evaluated on data. Held-out data should guide evaluation rather than leak into fitting.
Common misconception
A model is not necessarily learning new weights whenever it is used.
Start with
Related concepts
Quick recall question
Try answering before looking back at the definition.