In this article

Overview

|
  • Share via
  •  Edit this article

Overview

Do you want to integrate Caten's features into your Common Lisp application?

With caten/apps, you can effortlessly experiment with inference using ready-made implementations of various deep learning models.

caten/apps is a suite of packages that implement pipelines to smoothly perform the following tasks:

  • Automatically download pretrained weights

  • Automatically optimize and compile models for inference

For instance, text generation inference provided by caten/apps.gpt2 can be accomplished in just two lines:

(defparameter *model* (make-gpt2 :gpt2))
(gpt2-generate *model* "What is the answer to life, and the world?")

If you're considering contributing to Caten, we warmly welcome PRs that add new models to caten/apps. If you've implemented a model using Caten, please consider adding it to caten/apps! (Feel free to create a PR!)