up

Author: sayakpaul

src/diffusers/models/transformers/transformer_qwenimage.py

@@ -251,9 +251,10 @@ def _compute_video_freqs(self, frame, height, width, idx=0):
 
         freqs = torch.cat([freqs_frame, freqs_height, freqs_width], dim=-1).reshape(seq_lens, -1)
         freqs = freqs.clone().contiguous()
-        
+
         return freqs
 
+
 class QwenDoubleStreamAttnProcessor2_0:
     """
     Attention processor for Qwen double-stream architecture, matching DoubleStreamLayerMegatron logic. This processor

src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit.py

@@ -258,7 +258,7 @@ def _get_qwen_prompt_embeds(
         template = self.prompt_template_encode
         drop_idx = self.prompt_template_encode_start_idx
         txt = [template.format(e) for e in prompt]
-        
+
         model_inputs = self.processor(
             text=txt,
             images=image,

tests/pipelines/qwenimage/test_qwenimage_edit.py

@@ -13,8 +13,9 @@
 # limitations under the License.
 
 import unittest
-import pytest
+
 import numpy as np
+import pytest
 import torch
 from PIL import Image
 from transformers import Qwen2_5_VLConfig, Qwen2_5_VLForConditionalGeneration, Qwen2Tokenizer, Qwen2VLProcessor
@@ -27,7 +28,7 @@
 )
 from diffusers.utils.testing_utils import enable_full_determinism, torch_device
 
-from ..pipeline_params import TEXT_TO_IMAGE_BATCH_PARAMS, TEXT_TO_IMAGE_IMAGE_PARAMS, TEXT_TO_IMAGE_PARAMS
+from ..pipeline_params import TEXT_TO_IMAGE_PARAMS
 from ..test_pipelines_common import PipelineTesterMixin, to_np
 
 
@@ -57,7 +58,7 @@ class QwenImageEditPipelineFastTests(PipelineTesterMixin, unittest.TestCase):
 
     def get_dummy_components(self):
         tiny_ckpt_id = "hf-internal-testing/tiny-random-Qwen2VLForConditionalGeneration"
-        
+
         torch.manual_seed(0)
         transformer = QwenImageTransformer2DModel(
             patch_size=2,
@@ -162,8 +163,27 @@ def test_inference(self):
         self.assertEqual(generated_image.shape, (3, 32, 32))
 
         expected_slice = torch.tensor(
-            [[0.5637, 0.6341, 0.6001, 0.5620, 0.5794, 0.5498, 0.5757, 0.6389, 0.4174,
-        0.3597, 0.5649, 0.4894, 0.4969, 0.5255, 0.4083, 0.4986]])
+            [
+                [
+                    0.5637,
+                    0.6341,
+                    0.6001,
+                    0.5620,
+                    0.5794,
+                    0.5498,
+                    0.5757,
+                    0.6389,
+                    0.4174,
+                    0.3597,
+                    0.5649,
+                    0.4894,
+                    0.4969,
+                    0.5255,
+                    0.4083,
+                    0.4986,
+                ]
+            ]
+        )
         # fmt: on
 
         generated_slice = generated_image.flatten()
@@ -240,4 +260,4 @@ def test_vae_tiling(self, expected_diff_max: float = 0.2):
 
     @pytest.mark.xfail(condition=True, reason="Preconfigured embeddings need to be revisited.", strict=True)
     def test_encode_prompt_works_in_isolation(self, extra_required_param_value_dict=None, atol=1e-4, rtol=1e-4):
-        super().test_encode_prompt_works_in_isolation(extra_required_param_value_dict, atol, rtol)
+        super().test_encode_prompt_works_in_isolation(extra_required_param_value_dict, atol, rtol)