# Diagnose optimization traces

> ML Prep screening item.

- Stable ID: `ml-screen-optimization-advanced-diagnostics`
- Collection: screening
- Area: Machine Learning
- Topic: Optimization
- Origin: original
- Difficulty: hard
- Expected time: 8 minutes
- Skills: objective-functions, optimization-regularization, optimization-diagnostics
- Item JSON: https://mlprep.iwase.dev/item/ml-screen-optimization-advanced-diagnostics.json

## Problem

Diagnose and give one action for each trace.

1. Loss is $10,2,9,1,8$; parameter norms grow and gradients stay large.
2. Loss is stable; the gradient norm is $10^{-8}$; the Hessian is positive definite.
3. Data-fit loss is low, but the total objective is dominated by a very large $L_2$ penalty.


## 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 `ml-screen-optimization-advanced-diagnostics`, and record a real attempt with `recorded_by: "agent"` and `agent_session_id` when available using https://mlprep.iwase.dev/schema/attempt-record.json.