We’ve just posted a course on the freakcodecamp.org YouTube channel that will teach you how to become an AI researcher.
This course will guide you step-by-step, starting with the fundamentals necessary to understand modern AI, before diving into the fundamentals of Pyrotech. You will then learn about the building blocks of AI, from simple neural networks to the complexities of multi-layered architectures. The course concludes with an in-depth module on Transformers, a key technology underpinning today’s large language models (LLM) and generative AI.
The sections of this course are:
Introduction and Course Overview
Module 1: Basic Mathematics for AI Research
Math Lesson: Functions (Linear, Quadratic, Cubic, Square Root)
Math Lesson: Derivatives (Rates of Change)
Math Lesson: Vectors (Extension, Dot Product, Normalization)
Math Lesson: Gradient (steepest ascent/descent, partial derivative)
Math Lesson: Metrics (Multiplication, Transpose, Identification)
Math Lesson: Probability (Expected Value, Conditional Probability)
Module 2: Pytorch Fundamentals
Get started: Creating pytorch fundamentals and tensors
Pyurech Tutorial: Transforming and Viewing Tensors
PyTorch Lesson: Extruded and Invariant Dimension
PyTorch Tutorial: Indexing and Slicing Tensors
PyTorch Lesson: Special Tensors (Zeros, Ones, Lens Space)
Module 3: Neural Networks
Get started: Coding neural networks from scratch
Neural Networks Lesson: Single Neuron (Weights, Bias, Sum of Weights)
Neural Networks Lesson: Activation Functions (Sigmoid, Relo, TANH)
Neural Networks Lesson: Multilayer Networks and Back Propagation
Module 4: Transformers (For Major Language Models)
Start: Understanding Transformers for LL.M
Transformers Lesson: Attention Mechanism (Query, Key, Value)
Transformers Lesson: Self-Constructed and Formal Self-Constructed
Transformers Lesson: Rotary Positional Embedding (Rope)
Transformers lesson: Multi-head focus
Transformers Lesson: Transformer Block (Feed Forward, Ed and Norm)
Tokenization (for GPT architecture)
The result
View the full course freecodecamp.org YouTube channel (3 hour clock)