Upload anytext.py

anytext.py

@@ -1100,7 +1100,7 @@ class AuxiliaryLatentModule(ModelMixin, ConfigMixin):
         masked_img = torch.from_numpy(masked_img.copy()).float().to(device)
         if dtype == torch.float16:
             masked_img = masked_img.half()
-        masked_x = (retrieve_latents(self.config.vae.encode(masked_img[None, ...])) * self.vae.config.scaling_factor).detach()
+        masked_x = (retrieve_latents(self.config.vae.encode(masked_img[None, ...])) * self.config.vae.config.scaling_factor).detach()
         if dtype == torch.float16:
             masked_x = masked_x.half()
         text_info["masked_x"] = torch.cat([masked_x for _ in range(num_images_per_prompt)], dim=0)