Back to Study Plans

25 Problems

Build Micrograd from Scratch

25 NumPy and PyTorch problems covering scalar autograd, backpropagation, neural-network modules, and gradient descent.

7 Easy13 Medium5 Hard

7 sections

Estimate a Scalar Derivative Measure Scalar Expression Partials Check a Product-Chain Gradient Apply a Gradient-Descent Step Derivatives and Gradient Checks Quiz10 questions

Create an Addition Value Node Create a Multiplication Value Node Build a Scalar Expression Graph Trace a Reachable Expression Graph Value Objects and Expression Graphs Quiz10 questions

Differentiate a Tanh Activation Evaluate a Scalar Neuron Backpropagate Through a Scalar Neuron Gradient-Check a Scalar Neuron Neurons and Activation Gradients Quiz10 questions

Apply Local Vector-Jacobian Rules Topologically Sort a Computation DAG Run Reverse-Mode Autodiff Gradient-Check an Autodiff Graph Automatic Reverse-Mode Autodiff Quiz10 questions

Accumulate Shared-Path Gradients Differentiate Exponentials and Powers Accumulation and Complete Scalar Operators Quiz10 questions

Build a Scalar Neuron Module Evaluate a Dense Layer Evaluate a Multi-Layer Perceptron Collect Module Parameters Recursively Neural-Network Modules Quiz10 questions

Calculate Squared-Error Loss Gradients Apply One Full SGD Training Step Train a Tiny Micrograd MLP Losses, Training, and Zeroing Gradients Quiz10 questions

Summary

Total25

Easy7

Medium13

Hard5

Progress0%

Quizzes0/7