fix lora loader with torch-no-distributed

Signed-off-by: vladmandic <mandic00@live.com>

CHANGELOG.md

@@ -1,6 +1,6 @@
 # Change Log for SD.Next
 
-## Update for 2025-01-08
+## Update for 2025-01-09
 
 - **Models**
   - [Qwen-Image-2512](https://huggingface.co/Qwen/Qwen-Image-2512)  
@@ -38,8 +38,10 @@
   - switch processing class not restoring params  
   - hidiffusion tracing  
   - add skip_keys to offloading logic, fixes wan frames mismatch, thanks @ryanmeador  
+  - wan 2.2 i2v  
   - force model move on offload=none  
   - kandinsky-5 image and video on non-cuda platforms  
+  - lora loading when using torch without distributed support  
 
 ## Update for 2025-12-26
 

TODO.md

@@ -1,10 +1,5 @@
 # TODO
 
-## Known issues
-
-- kandinsky-image-5 hardcoded cuda: <https://github.com/huggingface/diffusers/pull/12814>
-- peft lora with torch-rocm-windows: <https://github.com/huggingface/peft/pull/2963>
-
 ## Project Board
 
 - <https://github.com/users/vladmandic/projects>

requirements.txt

@@ -36,7 +36,7 @@ fastapi==0.124.4
 rich==14.1.0
 safetensors==0.7.0
 tensordict==0.8.3
-peft==0.18.0
+peft==0.18.1
 httpx==0.28.1
 compel==2.2.1
 torchsde==0.2.6