make modules_to_not_convert actually run.

tests/models/testing_utils/quantization.py

@@ -21,11 +21,8 @@ import torch
 from diffusers import BitsAndBytesConfig, GGUFQuantizationConfig, NVIDIAModelOptConfig, QuantoConfig, TorchAoConfig
 from diffusers.utils.import_utils import (
     is_bitsandbytes_available,
-    is_gguf_available,
     is_nvidia_modelopt_available,
     is_optimum_quanto_available,
-    is_torchao_available,
-    is_torchao_version,
 )
 
 from ...testing_utils import (
@@ -59,13 +56,6 @@ if is_bitsandbytes_available():
 if is_optimum_quanto_available():
     from optimum.quanto import QLinear
 
-if is_gguf_available():
-    pass
-
-if is_torchao_available():
-    if is_torchao_version(">=", "0.9.0"):
-        pass
-
 
 class LoRALayer(torch.nn.Module):
     """Wraps a linear layer with LoRA-like adapter - Used for testing purposes only.

tests/models/transformers/test_models_transformer_flux.py

@@ -318,6 +318,10 @@ class TestFluxSingleFile(FluxTransformerTesterConfig, SingleFileTesterMixin):
 class TestFluxTransformerBitsAndBytes(FluxTransformerTesterConfig, BitsAndBytesTesterMixin):
     """BitsAndBytes quantization tests for Flux Transformer."""
 
+    @property
+    def modules_to_not_convert_for_test(self):
+        return ["transformer_blocks.0"]
+
 
 class TestFluxTransformerQuanto(FluxTransformerTesterConfig, QuantoTesterMixin):
     """Quanto quantization tests for Flux Transformer."""
@@ -330,10 +334,18 @@ class TestFluxTransformerQuanto(FluxTransformerTesterConfig, QuantoTesterMixin):
     def pretrained_model_kwargs(self):
         return {}
 
+    @property
+    def modules_to_not_convert_for_test(self):
+        return ["transformer_blocks.0"]
+
 
 class TestFluxTransformerTorchAo(FluxTransformerTesterConfig, TorchAoTesterMixin):
     """TorchAO quantization tests for Flux Transformer."""
 
+    @property
+    def modules_to_not_convert_for_test(self):
+        return ["transformer_blocks.0"]
+
 
 class TestFluxTransformerGGUF(FluxTransformerTesterConfig, GGUFTesterMixin):
     @property
@@ -402,6 +414,10 @@ class TestFluxTransformerGGUFCompile(FluxTransformerTesterConfig, GGUFCompileTes
 class TestFluxTransformerModelOpt(FluxTransformerTesterConfig, ModelOptTesterMixin):
     """ModelOpt quantization tests for Flux Transformer."""
 
+    @property
+    def modules_to_not_convert_for_test(self):
+        return ["transformer_blocks.0"]
+
 
 class TestFluxTransformerModelOptCompile(FluxTransformerTesterConfig, ModelOptCompileTesterMixin):
     """ModelOpt + compile tests for Flux Transformer."""