Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-20 09:38:23 +02:00
Code Issues Packages Projects Releases Wiki Activity

speedup prequant model load

Browse Source 
Signed-off-by: Vladimir Mandic <mandic00@live.com>
This commit is contained in:
Vladimir Mandic
2025-10-08 13:47:34 -04:00
parent 35277a79d3
commit 6907fcd320
2 changed files with 10 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/sd_models.py
+2 -1
View File
@@ -549,11 +549,12 @@ def load_sdnq_model(checkpoint_info, pipeline, diffusers_load_config, op):
continue
module_name = f.replace('_quantization_config.json', '')
quantization_config = shared.readfile(os.path.join(checkpoint_info.path, f), silent=True)
shared.log.debug(f'Load {op}: model="{checkpoint_info.name}" module="{module_name}" prequant=sdnq')
shared.log.debug(f'Load {op}: model="{checkpoint_info.name}" module="{module_name}" direct={shared.opts.diffusers_to_gpu} prequant=sdnq')
module_path = os.path.join(checkpoint_info.path, module_name)
modules[module_name] = sdnq.load_sdnq_model(
model_path=module_path,
quantization_config=quantization_config,
device=devices.device if shared.opts.diffusers_to_gpu else devices.cpu,
)
modules[module_name] = modules[module_name].to(device=devices.device)
sd_model = pipeline.from_pretrained(