diff --git a/extensions_built_in/flex2/flex2.py b/extensions_built_in/flex2/flex2.py
index e3234fe1..0f966c4b 100644
--- a/extensions_built_in/flex2/flex2.py
+++ b/extensions_built_in/flex2/flex2.py
@@ -72,6 +72,9 @@ class Flex2(BaseModel):
     def get_train_scheduler():
         return CustomFlowMatchEulerDiscreteScheduler(**scheduler_config)
 
+    def get_bucket_divisibility(self):
+        return 16
+
     def load_model(self):
         dtype = self.torch_dtype
         self.print_and_status_update("Loading Flux2 model")
diff --git a/extensions_built_in/sd_trainer/SDTrainer.py b/extensions_built_in/sd_trainer/SDTrainer.py
index 26bc6abf..cc08cdf6 100644
--- a/extensions_built_in/sd_trainer/SDTrainer.py
+++ b/extensions_built_in/sd_trainer/SDTrainer.py
@@ -198,7 +198,7 @@ class SDTrainer(BaseSDTrainProcess):
         
         if self.train_config.diffusion_feature_extractor_path is not None:
             vae = None
-            if self.model_config.arch != "flux":
+            if self.model_config.arch != "flux" or self.sd.vae.__class__.__name__ == "AutoencoderPixelMixer":
                 vae = self.sd.vae
             self.dfe = load_dfe(self.train_config.diffusion_feature_extractor_path, vae=vae)
             self.dfe.to(self.device_torch)