Get the latest tech news
From Unemployment to Lisp: Running GPT-2 on a Teen's Deep Learning Compiler
[wip] Deep Learning Compiler based on Polyhedral Compiler and Light-weight IRs, and Optimizing Pattern Matcher. - hikettei/Caten
You may still find the token/ms rate slow, but we're not yet at the stage of implementing an AutoScheduler to accelerate kernel performance (as well as GPU support). Once our IR matures enough to handle a wide range of deep learning models, we plan to focus on speeding things up! (If unsure, Emacs or Lem is recommended) Install ISL (Integer Set Library) for the fast kernel generation.
Or read this on Hacker News
