updates

Author: sayakpaul

tests/pipelines/audioldm2/test_audioldm2.py

@@ -508,9 +508,14 @@ def test_to_dtype(self):
         model_dtypes = {key: component.dtype for key, component in components.items() if hasattr(component, "dtype")}
         self.assertTrue(all(dtype == torch.float16 for dtype in model_dtypes.values()))
 
+    @unittest.skip("Test not supported.")
     def test_sequential_cpu_offload_forward_pass(self):
         pass
 
+    @unittest.skip("Test not supported for now because of the use of `projection_model` in `encode_prompt()`.")
+    def test_encode_prompt_works_in_isolation(self):
+        pass
+
 
 @nightly
 class AudioLDM2PipelineSlowTests(unittest.TestCase):

tests/pipelines/cogview3/test_cogview3plus.py

@@ -231,6 +231,9 @@ def test_attention_slicing_forward_pass(
                 "Attention slicing should not affect the inference results",
             )
 
+    def test_encode_prompt_works_in_isolation(self):
+        return super().test_encode_prompt_works_in_isolation(atol=1e-3, rtol=1e-3)
+
 
 @slow
 @require_torch_accelerator

tests/pipelines/controlnet_xs/test_controlnetxs.py

@@ -334,6 +334,13 @@ def test_to_device(self):
         output_device = pipe(**self.get_dummy_inputs(torch_device))[0]
         self.assertTrue(np.isnan(to_np(output_device)).sum() == 0)
 
+    def test_encode_prompt_works_in_isolation(self):
+        extra_required_param_value_dict = {
+            "device": torch.device(torch_device).type,
+            "do_classifier_free_guidance": self.get_dummy_inputs(device=torch_device).get("guidance_scale", 1.0) > 1.0,
+        }
+        return super().test_encode_prompt_works_in_isolation(extra_required_param_value_dict)
+
 
 @slow
 @require_torch_accelerator

tests/pipelines/i2vgen_xl/test_i2vgenxl.py

@@ -228,6 +228,10 @@ def test_num_videos_per_prompt(self):
 
         assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
 
+    @unittest.skip("Test not supported for now.")
+    def test_encode_prompt_works_in_isolation(self):
+        pass
+
 
 @slow
 @require_torch_accelerator

tests/pipelines/latent_consistency_models/test_latent_consistency_models.py

@@ -213,6 +213,13 @@ def callback_inputs_test(pipe, i, t, callback_kwargs):
         output = pipe(**inputs)[0]
         assert output.abs().sum() == 0
 
+    def test_encode_prompt_works_in_isolation(self):
+        extra_required_param_value_dict = {
+            "device": torch.device(torch_device).type,
+            "do_classifier_free_guidance": self.get_dummy_inputs(device=torch_device).get("guidance_scale", 1.0) > 1.0,
+        }
+        return super().test_encode_prompt_works_in_isolation(extra_required_param_value_dict)
+
 
 @slow
 @require_torch_gpu

tests/pipelines/latent_consistency_models/test_latent_consistency_models_img2img.py

@@ -220,6 +220,13 @@ def callback_inputs_test(pipe, i, t, callback_kwargs):
         output = pipe(**inputs)[0]
         assert output.abs().sum() == 0
 
+    def test_encode_prompt_works_in_isolation(self):
+        extra_required_param_value_dict = {
+            "device": torch.device(torch_device).type,
+            "do_classifier_free_guidance": self.get_dummy_inputs(device=torch_device).get("guidance_scale", 1.0) > 1.0,
+        }
+        return super().test_encode_prompt_works_in_isolation(extra_required_param_value_dict)
+
 
 @slow
 @require_torch_gpu

tests/pipelines/pia/test_pia.py

@@ -437,3 +437,11 @@ def test_xformers_attention_forwardGenerator_pass(self):
 
         max_diff = np.abs(to_np(output_with_offload) - to_np(output_without_offload)).max()
         self.assertLess(max_diff, 1e-4, "XFormers attention should not affect the inference results")
+
+    def test_encode_prompt_works_in_isolation(self):
+        extra_required_param_value_dict = {
+            "device": torch.device(torch_device).type,
+            "num_images_per_prompt": 1,
+            "do_classifier_free_guidance": self.get_dummy_inputs(device=torch_device).get("guidance_scale", 1.0) > 1.0,
+        }
+        return super().test_encode_prompt_works_in_isolation(extra_required_param_value_dict)

tests/pipelines/stable_diffusion_gligen_text_image/test_stable_diffusion_gligen_text_image.py

@@ -207,3 +207,9 @@ def test_attention_slicing_forward_pass(self):
 
     def test_inference_batch_single_identical(self):
         super().test_inference_batch_single_identical(batch_size=3, expected_max_diff=3e-3)
+
+    @unittest.skip(
+        "Test not supported because of the use of `text_encoder` in `get_cross_attention_kwargs_with_grounded()`."
+    )
+    def test_encode_prompt_works_in_isolation(self):
+        pass

tests/pipelines/stable_diffusion_sag/test_stable_diffusion_sag.py

@@ -153,6 +153,13 @@ def test_pipeline_different_schedulers(self):
             # Karras schedulers are not supported
             image = pipeline(**inputs).images[0]
 
+    def test_encode_prompt_works_in_isolation(self):
+        extra_required_param_value_dict = {
+            "device": torch.device(torch_device).type,
+            "do_classifier_free_guidance": self.get_dummy_inputs(device=torch_device).get("guidance_scale", 1.0) > 1.0,
+        }
+        return super().test_encode_prompt_works_in_isolation(extra_required_param_value_dict)
+
 
 @nightly
 @require_torch_gpu

tests/pipelines/stable_unclip/test_stable_unclip_img2img.py

@@ -207,6 +207,10 @@ def test_inference_batch_single_identical(self):
     def test_xformers_attention_forwardGenerator_pass(self):
         self._test_xformers_attention_forwardGenerator_pass(test_max_difference=False)
 
+    @unittest.skip("Test not supported at the moment.")
+    def test_encode_prompt_works_in_isolation(self):
+        pass
+
 
 @nightly
 @require_torch_gpu