From d823eb07e9cda7af01047069ffac57124f04d371 Mon Sep 17 00:00:00 2001 From: takuseno Date: Fri, 25 Oct 2024 22:05:08 +0900 Subject: [PATCH] Update readme --- README.md | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/README.md b/README.md index 1f00818..ef79a7e 100644 --- a/README.md +++ b/README.md @@ -38,6 +38,10 @@ copy_tensor_with_assertion(a, b) ``` Since `copy_tensor_with_assertion` does additional GPU-CPU transfer internally, make sure that you switch to `copy_tensor` in your experiments. Otherwise your training loop will be significantly slower. +## Features +- Fast inter-library tensor copy. +- Inter-GPU copy (I believe this is supported with the current implementation. But, not tested yet.) + ## Supported deep learning libraries - PyTorch - Jax