maybe fix lora compile

This commit is contained in:
kijai 2024-12-10 17:30:47 +02:00
parent 6eb9211091
commit 8c590fd5a0

View File

@ -5,6 +5,7 @@ import comfy.sd
import torch
import folder_paths
orig_attention = comfy_attention.optimized_attention
import comfy.model_management as mm
class BaseLoaderKJ:
original_linear = None
@ -155,7 +156,9 @@ def patched_patch_model(self, device_to=None, lowvram_model_memory=0, load_weigh
full_load = True
else:
full_load = False
device_to = mm.get_torch_device()
load_weights = True
if load_weights:
self.load(device_to, lowvram_model_memory=lowvram_model_memory, force_patch_weights=force_patch_weights, full_load=full_load)
for k in self.object_patches: