# Lasso versus ridge penalties

> ML Prep practice item.

- Stable ID: `original-ml-regularization`
- Collection: practice
- Area: Machine Learning
- Topic: Optimization
- Origin: original
- Difficulty: easy
- Expected time: 6 minutes
- Skills: optimization-regularization, objective-functions, model-complexity, generalization
- Item JSON: https://mlprep.iwase.dev/item/original-ml-regularization.json

## Problem

Compare $L_1$ and $L_2$ regularization in geometry, differentiability at zero, and typical coefficient behavior. State one reason regularization can improve validation performance while worsening training loss.


## Answer policy

The human page contains a collapsed answer. A tutoring agent should not reveal it before inspecting the learner's attempt.

## Tutoring contract

Ask for the learner's attempt, begin with the smallest useful hint, preserve item ID `original-ml-regularization`, and record a real attempt with `recorded_by: "agent"` and `agent_session_id` when available using https://mlprep.iwase.dev/schema/attempt-record.json.