CodeUtility
Making your developer's life easier!

Ai coding challenges

Practise ai with 47 coding challenges, from easy to hard. Every challenge runs in your browser with instant tests, an AI review of your code and community solutions to compare.

Easy · 19 Medium · 24 Hard · 4

Skills you'll practise: Gradient descentBackpropagationChain ruleDecision treesRegressionTransformersA* searchActivationsAdamAlpha-beta pruningAttentionBPE
  • Python · C++ · C#
  • Automated tests
  • AI code review
  • Community solutions
Programming
Programming fundamentalsFrom variables to recursion, for beginners
AI and machine learning
AI Learning HubNeural networks, ML and LLMs from scratch Neural networks from scratchNeurons, backpropagation and optimizers Machine learning from scratchRegression, classification and clustering Transformers and LLMs from scratchTokenizers, attention and text generation Classical AI and reinforcement learningSearch, games and reinforcement learning
226
Challenges
2
Solutions shared
3
Languages

All challenges

Search by topic, language, or difficulty.

47 challenges
Medium

Mini-Batch Gradient Descent

Train a linear model by updating the weights after every small batch of samples instead of the whole dataset.

You'll learn Mini-batchesGradient descent
Neural networksaimachine learning
Python C++ C#
0 solved
No solutions yet
Easy

Gradient Descent with Momentum

Apply momentum updates, which keep part of the previous step so training speeds up along consistent directions.

You'll learn MomentumOptimizers
Neural networksaimachine learning
Python C++ C#
0 solved
No solutions yet
Hard

Multi-Head Self-Attention

Project the input into queries, keys and values, run attention in several heads in parallel, and combine them.

You'll learn Multi-head attentionMatrix projections
aimachine learning
Python C++ C#
0 solved
No solutions yet
Easy

Xavier and He Weight Initialization

Compute the range or standard deviation used to initialize the weights of a layer.

You'll learn Xavier and He initTraining stability
Neural networksaimachine learning
C# Python C++
0 solved
No solutions yet
Hard

Train a Neural Network to Solve XOR

Combine everything: train a 2-2-1 sigmoid network with backpropagation until it learns XOR.

You'll learn Neural network trainingXOR problem
Neural networksaimachine learning
Python C++ C#
0 solved
No solutions yet
Medium

Adam Optimizer

Implement Adam, the default optimizer for deep learning, with moving averages and bias correction.

You'll learn AdamBias correction
Neural networksaimachine learning
Python C++ C#
0 solved
No solutions yet
Medium

Temperature and Top-K Sampling

Choose the next token like a language model does: keep the top k logits, apply temperature, and sample.

You'll learn Text generationSampling
aimachine learning
Python C++ C#
0 solved
No solutions yet
Medium

Train a Neuron with Gradient Descent

Repeat forward pass, gradients and weight updates to teach a sigmoid neuron from examples.

You'll learn Gradient descentTraining loops
Neural networksaimachine learning
Python C++ C#
0 solved
No solutions yet
Medium

Byte Pair Encoding (BPE) Merges

Learn subword merges the way GPT tokenizers do: repeatedly merge the most frequent pair of adjacent symbols.

You'll learn BPESubword tokenization
aimachine learning
Python C++ C#
0 solved
No solutions yet
Medium

Linear Regression with Gradient Descent

Learn the slope and intercept of a line by repeatedly stepping down the mean squared error.

You'll learn Gradient descentRegression
aimachine learning
Python C++ C#
0 solved
No solutions yet
Easy

Categorical Cross-Entropy Loss

Measure the loss of a multi-class classifier from its predicted probabilities and the true classes.

You'll learn Classification lossLogarithms
aimachine learning
Python C++ C#
0 solved
No solutions yet
Easy

One-Hot Encoding

Turn categories into vectors with a single 1, the input format for classifiers and embeddings.

You'll learn One-hot vectorsCategorical data
aimachine learning
Python C++ C#
0 solved
No solutions yet
Easy

Logistic Regression Prediction

Classify samples as 0 or 1 by passing a weighted sum through the sigmoid and comparing it to a threshold.

You'll learn Logistic regressionThresholds
aimachine learning
Python C++ C#
0 solved
No solutions yet
Easy

Cosine Similarity

Measure how similar two vectors are by the angle between them, as search engines do with embeddings.

You'll learn Vector similarityEmbeddings
aimachine learning
Python C++ C#
0 solved
No solutions yet
Easy

Z-Score Standardization

Center a feature at 0 and scale it to a standard deviation of 1.

You'll learn StandardizationMean and std dev
aimachine learning
Python C++ C#
0 solved
No solutions yet
Medium

Simple Linear Regression (Least Squares)

Fit the best straight line through data points with the least squares formulas.

You'll learn Least squaresRegression
aimachine learning
Python C++ C#
0 solved
No solutions yet
Easy

Gini Impurity

Measure how mixed the labels in a group are, the score decision trees use to choose splits.

You'll learn Decision treesImpurity measures
aimachine learning
Python C++ C#
0 solved
No solutions yet
Easy

Min-Max Normalization

Rescale a feature so its smallest value becomes 0 and its largest becomes 1.

You'll learn Feature scalingData preprocessing
aimachine learning
Python C++ C#
0 solved
No solutions yet

Browse by topic

Browse by language