From 203724e9d9042eb5555835b2a211b4ce4e98fe02 Mon Sep 17 00:00:00 2001 From: Sayak Paul Date: Fri, 29 Dec 2023 09:38:50 +0530 Subject: [PATCH] [Docs] add note on fp16 in fast diffusion (#6380) add note on fp16 --- docs/source/en/tutorials/fast_diffusion.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/docs/source/en/tutorials/fast_diffusion.md b/docs/source/en/tutorials/fast_diffusion.md index 7392d93547..d81890be52 100644 --- a/docs/source/en/tutorials/fast_diffusion.md +++ b/docs/source/en/tutorials/fast_diffusion.md @@ -96,6 +96,8 @@ bfloat16 reduces the latency from 7.36 seconds to 4.63 seconds: +_(We later ran the experiments in float16 and found out that the recent versions of torchao do not incur numerical problems from float16.)_ + **Why bfloat16?** * Using a reduced numerical precision (such as float16, bfloat16) to run inference doesn’t affect the generation quality but significantly improves latency.