Implement Backprop for a Tiny Network

Quick Overview

This question evaluates understanding of backpropagation, gradient derivation, numerical stability of softmax cross-entropy, and practical implementation skills in both NumPy and PyTorch autograd for a two-layer MLP, including correct batched tensor shapes.

Implement Backprop for a Tiny Network

Company: OpenAI

Role: Machine Learning Engineer

Category: Machine Learning

Difficulty: hard

Interview Round: Onsite

Overview: This question evaluates understanding of backpropagation, gradient derivation, numerical stability of softmax cross-entropy, and practical implementation skills in both NumPy and PyTorch autograd for a two-layer MLP, including correct batched tensor shapes.

|Home/Machine Learning/OpenAI
OpenAI logo
OpenAI
Apr 3, 2026
hardMachine Learning EngineerOnsiteMachine Learning
197
0
Loading...
Loading comments...