Search⌘ K
AI Features
Home
Courses
Deep Learning with JAX and Flax

Stochastic Gradient Descent

Explore the implementation and use of stochastic gradient descent with momentum and Nesterov acceleration in JAX and Flax. Understand various optimizers including Noisy SGD, Optimistic Gradient Descent, RMSProp, and Yogi, and learn how to apply them to improve model training and convergence.

We'll cover the following...

SGD implements stochastic gradient descent with support for momentum and Nesterov accelerationNesterov acceleration refers to a method of accelerating the convergence of iterative optimization algorithms commonly used in machine learning.. Momentum makes obtaining optimal model weights faster by accelerating gradient descent in a certain direction.

Gradient function
Gradient function
...