Search code, repositories, users, issues, pull requests...
The project's goal is to democratize AI compute such that the systems of the future are neither controlled by rich countries nor big tech.
We build on Finite Field theory, a branch of mathematics I studied at Yale University. The main idea is to use the properties of prime numbers to multiply several array elements at the same time. It's like GPU hardware parallelization but we replace the GPU kernels with prime numbers.
Finite Field Assembly merely applies these obscure topics in mathematics to a mainstream problem - making AI compute available to everyone. We are financially supported by our users on substack
✔ Recursive computing support : not cache-aware vectorization, not parallelization, but performing a calculation inside a calculation inside another calculation.
