From 4568631b43f348ea4360b021315a3da8064f3d7b Mon Sep 17 00:00:00 2001
From: Kohya S <52813779+kohya-ss@users.noreply.github.com>
Date: Sat, 13 Sep 2025 20:05:39 +0900
Subject: [PATCH] docs: update README to reflect improved loading speed of
 .safetensors files

---
 README.md | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index 843cf71b..da38a241 100644
--- a/README.md
+++ b/README.md
@@ -13,11 +13,13 @@ For RTX 50 series GPUs, PyTorch 2.8.0 with CUDA 12.8/9 should be used. `requirem
 
 If you are using DeepSpeed, please install DeepSpeed with `pip install deepspeed` (appropriate version is not confirmed yet).
 
-- [FLUX.1 training](#flux1-training)
-- [SD3 training](#sd3-training)
-
 ### Recent Updates
 
+Sep 13, 2025:
+- The loading speed of `.safetensors` files has been improved for SD3, FLUX.1 and Lumina. See [PR #2200](https://github.com/kohya-ss/sd-scripts/pull/2200) for more details.
+    - Model loading can be up to 1.5 times faster.
+    - This is a wide-ranging update, so there may be bugs. Please let us know if you encounter any issues.
+
 Sep 4, 2025:
 - The information about FLUX.1 and SD3/SD3.5 training that was described in the README has been organized and divided into the following documents:
     - [LoRA Training Overview](./docs/train_network.md)