# Gradient Descent Practice

> ML preparation set.

- Stable ID: `gradient-descent-practice`
- Area: machine-learning
- Kind: practice
- Timebox: 25 minutes

Compute deterministic and stochastic updates and diagnose step-size behavior.

## Instructions

1. Write the gradient at the current iterate before updating.

## Ordered items

1. [Gradient descent on a quadratic](https://mlprep.iwase.dev/item/original-ml-gd/?set=gradient-descent-practice) — `original-ml-gd` (7 min)
2. [Full gradient versus stochastic gradient](https://mlprep.iwase.dev/item/original-ml-sgd/?set=gradient-descent-practice) — `original-ml-sgd` (7 min)
3. [Linear regression objective and gradient](https://mlprep.iwase.dev/item/original-ml-regression/?set=gradient-descent-practice) — `original-ml-regression` (8 min)

## Completion

Explain learning-rate and gradient-variance effects.