fix

Author: a-r-r-o-w

tests/pipelines/wan/test_wan.py

@@ -132,7 +132,6 @@ def test_inference(self):
 
         generated_slice = generated_video.flatten()
         generated_slice = torch.cat([generated_slice[:8], generated_slice[-8:]])
-        print("txt2video:", [round(x, 4) for x in generated_slice.tolist()])
         self.assertTrue(torch.allclose(generated_slice, expected_slice, atol=1e-3))
 
     @unittest.skip("Test not supported")

tests/pipelines/wan/test_wan_image_to_video.py

@@ -149,12 +149,11 @@ def test_inference(self):
         self.assertEqual(generated_video.shape, (9, 3, 16, 16))
 
         # fmt: off
-        expected_slice = torch.tensor([0.4531, 0.4527, 0.4498, 0.4542, 0.4526, 0.4527, 0.4534, 0.4534, 0.5061, 0.5185, 0.5283, 0.5181, 0.5309, 0.5365, 0.5113, 0.5244])
+        expected_slice = torch.tensor([0.4525, 0.4525, 0.4497, 0.4536, 0.452, 0.4529, 0.454, 0.4535, 0.5072, 0.5527, 0.5165, 0.5244, 0.5481, 0.5282, 0.5208, 0.5214])
         # fmt: on
 
         generated_slice = generated_video.flatten()
         generated_slice = torch.cat([generated_slice[:8], generated_slice[-8:]])
-        print("image2video:", [round(x, 4) for x in generated_slice.tolist()])
         self.assertTrue(torch.allclose(generated_slice, expected_slice, atol=1e-3))
 
     @unittest.skip("Test not supported")
@@ -166,7 +165,25 @@ def test_inference_batch_single_identical(self):
         pass
 
 
-class WanFLFToVideoPipelineFastTests(WanImageToVideoPipelineFastTests):
+class WanFLFToVideoPipelineFastTests(PipelineTesterMixin, unittest.TestCase):
+    pipeline_class = WanImageToVideoPipeline
+    params = TEXT_TO_IMAGE_PARAMS - {"cross_attention_kwargs", "height", "width"}
+    batch_params = TEXT_TO_IMAGE_BATCH_PARAMS
+    image_params = TEXT_TO_IMAGE_IMAGE_PARAMS
+    image_latents_params = TEXT_TO_IMAGE_IMAGE_PARAMS
+    required_optional_params = frozenset(
+        [
+            "num_inference_steps",
+            "generator",
+            "latents",
+            "return_dict",
+            "callback_on_step_end",
+            "callback_on_step_end_tensor_inputs",
+        ]
+    )
+    test_xformers_attention = False
+    supports_dduf = False
+
     def get_dummy_components(self):
         torch.manual_seed(0)
         vae = AutoencoderKLWan(
@@ -251,3 +268,32 @@ def get_dummy_inputs(self, device, seed=0):
             "output_type": "pt",
         }
         return inputs
+
+    def test_inference(self):
+        device = "cpu"
+
+        components = self.get_dummy_components()
+        pipe = self.pipeline_class(**components)
+        pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        inputs = self.get_dummy_inputs(device)
+        video = pipe(**inputs).frames
+        generated_video = video[0]
+        self.assertEqual(generated_video.shape, (9, 3, 16, 16))
+
+        # fmt: off
+        expected_slice = torch.tensor([0.4531, 0.4527, 0.4498, 0.4542, 0.4526, 0.4527, 0.4534, 0.4534, 0.5061, 0.5185, 0.5283, 0.5181, 0.5309, 0.5365, 0.5113, 0.5244])
+        # fmt: on
+
+        generated_slice = generated_video.flatten()
+        generated_slice = torch.cat([generated_slice[:8], generated_slice[-8:]])
+        self.assertTrue(torch.allclose(generated_slice, expected_slice, atol=1e-3))
+
+    @unittest.skip("Test not supported")
+    def test_attention_slicing_forward_pass(self):
+        pass
+
+    @unittest.skip("TODO: revisit failing as it requires a very high threshold to pass")
+    def test_inference_batch_single_identical(self):
+        pass

tests/pipelines/wan/test_wan_video_to_video.py

@@ -130,7 +130,6 @@ def test_inference(self):
 
         generated_slice = generated_video.flatten()
         generated_slice = torch.cat([generated_slice[:8], generated_slice[-8:]])
-        print("video2video:", [round(x, 4) for x in generated_slice.tolist()])
         self.assertTrue(torch.allclose(generated_slice, expected_slice, atol=1e-3))
 
     @unittest.skip("Test not supported")