From baef8a6bc153b893652d3e058ab9fb73e5981fa5 Mon Sep 17 00:00:00 2001 From: Liu Liu Date: Thu, 21 Mar 2024 14:10:30 -0400 Subject: [PATCH] Add link to the documentation. --- README.md | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/README.md b/README.md index 6f4290586b9..82c26fac89b 100644 --- a/README.md +++ b/README.md @@ -11,6 +11,10 @@ A high-level language that delegates most of the work to the C runtime seems to Because the heavy-lifting is done in the [libnnc](https://libnnc.org) library, the Swift portion can be light and even automatically generated. At the moment, we have about 3,000 lines of Swift code to run quite a few models on GPU, complete with data feeders, model specifications and optimizers. +## Live Documentation + + + ## Runtime API Currently, s4nnc works better under Linux with CUDA 11, CuDNN and NCCL. The API for s4nnc wraps around [Level-4](https://libnnc.org/api/level-4/) and [Level-5](https://libnnc.org/api/level-5/) C APIs.