Merge afe05d68ee into da2bfb5b0a

comfy/model_base.py

@@ -305,7 +305,9 @@ class BaseModel(torch.nn.Module):
                 to_load[k[len(unet_prefix):]] = sd.pop(k)
 
         to_load = self.model_config.process_unet_state_dict(to_load)
-        m, u = self.diffusion_model.load_state_dict(to_load, strict=False)
+        # assign=True will reuse the tensor storage in state dict, this will avoid copy and saving CPU memory
+        # when loading large models with mmap. 
+        m, u = self.diffusion_model.load_state_dict(to_load, strict=False, assign=True)
         if len(m) > 0:
             logging.warning("unet missing: {}".format(m))