← Questions

MNIST classifier: per-example loss bounds and adaptive calibration error

0
OpenAIAI/ML designLast asked Nov 2026

An AI/ML design interview question asked in OpenAI's technical screen (coding), last asked Nov 2026. It tests calibration, cross-entropy and numpy.

A classifier on MNIST (10 digit classes) outputs a softmax probability vector p and is trained with cross-entropy (confirm the loss first). The round has four connected tasks.

Sign in to unlock this question

The full prompt, the worked answer and your own canvas or code with the AI tutor. New accounts get 5 credits: each one unlocks a question for good.

Where it's asked

  • OpenAI · Technical screen: coding · Last asked Nov 202660 min · One problem in three to five growing parts filling the hour, usually in CoderPad; ML roles get a Colab notebook with autocomplete off · No AI assistants or autocomplete; documentation is fine

What it tests

calibration · cross-entropy · numpy · testing

Your canvas and the AI tutor open when you unlock this question.