Skip to content

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.

Fix some issues with the model sometimes not getting patched.

comfyanonymous5d43e75e5b94
diff --git a/comfy/model_patcher.py b/comfy/model_patcher.py
index c8009db1a..39d00d82e 100644
--- a/comfy/model_patcher.py
+++ b/comfy/model_patcher.py
@@ -669,6 +669,8 @@ class ModelPatcher:
return memory_freed
def partially_load(self, device_to, extra_memory=0):
+ self.unpatch_model(unpatch_weights=False)
+ self.patch_model(patch_weights=False)
full_load = False
if self.model.model_lowvram == False:
return 0