MuhammadEssa
A developer who drew attention for sharing on Reddit his learning experience of writing a neural network from scratch and rewriting it in JAX, and who created the Neural Canvas project.
Timeline (last 90 days)
Developer MuhammadEssa hand-wrote forward propagation, backpropagation, and gradient descent with NumPy, rewrote them with JAX, and ultimately built the interactive MNIST visualization web page Neural Canvas.
The developer's advice to beginners is: start with linear algebra, don't dive straight into advanced architectures, optimizers or high-level frameworks, but at least thoroughly master matrix multiplication and how it represents the operations inside a network.
All Facts (2)
Developer MuhammadEssa hand-wrote forward propagation, backpropagation, and gradient descent with NumPy, rewrote them with JAX, and ultimately built the interactive MNIST visualization web page Neural Canvas.
50%UnverifiedThe developer's advice to beginners is: start with linear algebra, don't dive straight into advanced architectures, optimizers or high-level frameworks, but at least thoroughly master matrix multiplication and how it represents the operations inside a network.
50%