StatQuest with Josh Starmer Courses

Transformer Neural Networks, ChatGPT's Foundation, Clearly Explained

Clear explanation of Transformer Neural Networks, the foundation of ChatGPT and modern AI. Covers key concepts like word embedding, self-attention, and encoder-decoder architecture.

• YouTube
• 36 minutes
• Self-Paced
• Free Video

Decoder-Only Transformers, ChatGPT's Specific Transformer, Clearly Explained

Comprehensive explanation of Decoder-Only Transformers used in ChatGPT, covering word embedding, position encoding, masked self-attention, and output generation, with comparisons to normal Transformers.

• YouTube
• 37 minutes
• Self-Paced
• Free Video

Long Short-Term Memory with PyTorch + Lightning

Code an LSTM from scratch, train it with PyTorch Lightning, and compare it with PyTorch’s built-in nn.LSTM while tracking results in TensorBoard.

• YouTube
• 33 minutes
• Self-Paced
• Free Video

Long Short-Term Memory (LSTM), Clearly Explained

An intuitive explanation of LSTM networks, showing how memory paths address vanishing and exploding gradients in long sequences.

• YouTube
• 21 minutes
• Self-Paced
• Free Video

Introduction to Coding Neural Networks with PyTorch and Lightning

Aprende a simplificar el entrenamiento de redes neuronales con PyTorch Lightning, encontrar la tasa de aprendizaje y aprovechar la aceleración por GPU.

• YouTube
• 21 minutes
• Self-Paced
• Free Video

Introduction to PyTorch

Build a simple neural network in PyTorch, visualize its output, and optimize a parameter with backpropagation.

• YouTube
• 23 minutes
• Self-Paced
• Free Video

Neural Networks Part 7 - Cross Entropy Derivatives and Backpropagation

Derive cross-entropy gradients and apply them to backpropagation in a softmax neural network.

• YouTube
• 22 minutes
• Self-Paced
• Free Video

Support Vector Machines in Python from Start to Finish

Build and optimize a support vector machine classifier in Python, handling missing data, encoding categorical features, scaling inputs, and visualizing the final decision boundary.

• YouTube
• 45 minutes
• Self-Paced
• Free Video

Classification Trees in Python from Start to Finish

Build and optimize a heart disease classification tree in Python using missing-data handling, one-hot encoding, and cost-complexity pruning.

• YouTube
• 1 hour 6 minutes
• Self-Paced
• Free Video

How to Calculate P-Values

Learn to calculate p-values from coin-toss probabilities and continuous distributions, including two-sided versus one-sided tests and why one-sided p-values can mislead.

• YouTube
• 25 minutes
• Self-Paced
• Free Video

XGBoost Part 4 - Crazy Cool Optimizations

Explains how XGBoost accelerates training on large datasets through approximate splitting, weighted quantile sketches, sparse-data handling, cache-aware access, and out-of-core computation.

• YouTube
• 24 minutes
• Self-Paced
• Free Video

XGBoost Part 2 - Classification

Explains how XGBoost builds classification trees using similarity scores, gain, cover, pruning, and leaf outputs.

• YouTube
• 25 minutes
• Self-Paced
• Free Video

XGBoost Part 1 - Regression

Learn how XGBoost builds and prunes regularized regression trees, calculates similarity scores and gain, and updates predictions with leaf outputs and a learning rate.

• YouTube
• 26 minutes
• Self-Paced
• Free Video

Support Vector Machines Part 1 - Main Ideas

Explains how support vector machines classify data using maximal margins, soft margins, polynomial and radial basis function kernels, and the kernel trick.

• YouTube
• 21 minutes
• Self-Paced
• Free Video

Regression Trees, Clearly Explained

Explains how regression trees model nonlinear relationships by choosing data-splitting thresholds that minimize squared residuals.

• YouTube
• 23 minutes
• Self-Paced
• Free Video