Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-10 06:48:43 +02:00
Code Issues Packages Projects Releases Wiki Activity

reapply offload hooks on diffusers lora load

Browse Source 
Signed-off-by: Vladimir Mandic <mandic00@live.com>
This commit is contained in:
Vladimir Mandic
2026-01-13 10:07:30 +00:00
parent f21a943028
commit e7809b11cd
2 changed files with 3 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
CHANGELOG.md
+2 -1
View File
@@ -1,6 +1,6 @@
# Change Log for SD.Next
## Update for 2025-01-12
## Update for 2025-01-13
- **Models**
- [Qwen-Image-2512](https://huggingface.co/Qwen/Qwen-Image-2512)
@@ -54,6 +54,7 @@
- force model move on offload=none
- kandinsky-5 image and video on non-cuda platforms
- lora loading when using torch without distributed support
- generate slowdown when consequtive lora-diffusers enabled
- google-genai auth, thanks @CalamitousFelicitousness
- reduce triton test verbosity
- improve qwen i2i handling
modules/lora/lora_load.py
+1
View File
@@ -304,6 +304,7 @@ def network_load(names, te_multipliers=None, unet_multipliers=None, dyn_dims=Non
shared.log.error(f'Network load: type=LoRA action=fuse {str(e)}')
if l.debug:
errors.display(e, 'LoRA')
shared.sd_model = sd_models.apply_balanced_offload(shared.sd_model, force=True) # some layers may end up on cpu without hook
if len(l.loaded_networks) > 0 and l.debug:
shared.log.debug(f'Network load: type=LoRA loaded={[n.name for n in l.loaded_networks]} cache={list(lora_cache)} fuse={shared.opts.lora_fuse_native}:{shared.opts.lora_fuse_diffusers}')