From 56e384b894914ec9728a768c1ebcf3df093d824f Mon Sep 17 00:00:00 2001 From: lin-bot23 Date: Sat, 11 Jul 2026 06:14:50 +0800 Subject: [PATCH] docs: clarify that --async-offload is enabled by default on Nvidia GPUs The troubleshooting page listed --async-offload as a performance tip but did not mention it is enabled by default on Nvidia GPUs. This caused user confusion, with multiple users asking 'is memory off-loading on by default?' in the Mintlify AI assistant analytics. Source: Mintlify AI Assistant analytics data (2026-07-05/2026-07-04) --- troubleshooting/overview.mdx | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/troubleshooting/overview.mdx b/troubleshooting/overview.mdx index 62b3e7952..de38bc2ea 100644 --- a/troubleshooting/overview.mdx +++ b/troubleshooting/overview.mdx @@ -65,7 +65,7 @@ python main.py --preview-method none python main.py --use-pytorch-cross-attention python main.py --use-flash-attention -# Async weight offloading +# Async weight offloading (enabled by default on Nvidia GPUs) python main.py --async-offload ```