gradli is an autograd engine. the name consists of gradient + li - the swiss have a habit of putting "-li" at the ending of many words to convey smallness (diminutive) and since this project should be a small autograd library - i thought it might be a fun wordplay.
pytorch is nice and all but i like learning by doing so i decided to start this project. i would like to fully understand the challenges of autograd and how such a system's bottlenecks can potentially be optimized. i thought it might be beneficial to build such a system from the ground up.