AI Framework

JAX

Google JAX is a machine learning framework for transforming numerical functions, which Google refers to as XLA (Accelerated Linear Algebra) that combines modified vers...

Tags:

Google JAX is a machine learning framework for transforming numerical functions, which Google refers to as XLA (Accelerated Linear Algebra) that combines modified versions of Autograd (gradient function automatically obtained through function differentiation) and TensorFlow. The design of this framework follows the structure and workflow of NumPy as much as possible, and works collaboratively with various existing frameworks such as TensorFlow and PyTorch.
The main functions of JAX include:
Grad: Automatic differentiation
Jit: compile
Vmap: automatic vectorization
Pmap: SPMD programming

data statistics

Relevant Navigation

No comments

No comments...