Merge branch 'main' into sd2x_ldm_singlefile_fix

Author: DN6

tests/pipelines/controlnet_flux/test_controlnet_flux_img2img.py

@@ -158,30 +158,6 @@ def test_flux_controlnet_different_prompts(self):
 
         assert max_diff > 1e-6
 
-    def test_flux_controlnet_prompt_embeds(self):
-        pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
-        inputs = self.get_dummy_inputs(torch_device)
-
-        output_with_prompt = pipe(**inputs).images[0]
-
-        inputs = self.get_dummy_inputs(torch_device)
-        prompt = inputs.pop("prompt")
-
-        (prompt_embeds, pooled_prompt_embeds, text_ids) = pipe.encode_prompt(
-            prompt,
-            prompt_2=None,
-            device=torch_device,
-            max_sequence_length=inputs["max_sequence_length"],
-        )
-        output_with_embeds = pipe(
-            prompt_embeds=prompt_embeds,
-            pooled_prompt_embeds=pooled_prompt_embeds,
-            **inputs,
-        ).images[0]
-
-        max_diff = np.abs(output_with_prompt - output_with_embeds).max()
-        assert max_diff < 1e-4
-
     def test_fused_qkv_projections(self):
         device = "cpu"  # ensure determinism for the device-dependent torch.Generator
         components = self.get_dummy_components()

tests/pipelines/flux/test_pipeline_flux.py

@@ -136,30 +136,6 @@ def test_flux_different_prompts(self):
         # For some reasons, they don't show large differences
         assert max_diff > 1e-6
 
-    def test_flux_prompt_embeds(self):
-        pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
-        inputs = self.get_dummy_inputs(torch_device)
-
-        output_with_prompt = pipe(**inputs).images[0]
-
-        inputs = self.get_dummy_inputs(torch_device)
-        prompt = inputs.pop("prompt")
-
-        (prompt_embeds, pooled_prompt_embeds, text_ids) = pipe.encode_prompt(
-            prompt,
-            prompt_2=None,
-            device=torch_device,
-            max_sequence_length=inputs["max_sequence_length"],
-        )
-        output_with_embeds = pipe(
-            prompt_embeds=prompt_embeds,
-            pooled_prompt_embeds=pooled_prompt_embeds,
-            **inputs,
-        ).images[0]
-
-        max_diff = np.abs(output_with_prompt - output_with_embeds).max()
-        assert max_diff < 1e-4
-
     def test_fused_qkv_projections(self):
         device = "cpu"  # ensure determinism for the device-dependent torch.Generator
         components = self.get_dummy_components()

tests/pipelines/flux/test_pipeline_flux_control.py

@@ -126,30 +126,6 @@ def test_flux_different_prompts(self):
         # For some reasons, they don't show large differences
         assert max_diff > 1e-6
 
-    def test_flux_prompt_embeds(self):
-        pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
-        inputs = self.get_dummy_inputs(torch_device)
-
-        output_with_prompt = pipe(**inputs).images[0]
-
-        inputs = self.get_dummy_inputs(torch_device)
-        prompt = inputs.pop("prompt")
-
-        (prompt_embeds, pooled_prompt_embeds, text_ids) = pipe.encode_prompt(
-            prompt,
-            prompt_2=None,
-            device=torch_device,
-            max_sequence_length=inputs["max_sequence_length"],
-        )
-        output_with_embeds = pipe(
-            prompt_embeds=prompt_embeds,
-            pooled_prompt_embeds=pooled_prompt_embeds,
-            **inputs,
-        ).images[0]
-
-        max_diff = np.abs(output_with_prompt - output_with_embeds).max()
-        assert max_diff < 1e-4
-
     def test_fused_qkv_projections(self):
         device = "cpu"  # ensure determinism for the device-dependent torch.Generator
         components = self.get_dummy_components()

tests/pipelines/flux/test_pipeline_flux_control_img2img.py

@@ -129,30 +129,6 @@ def test_flux_different_prompts(self):
         # For some reasons, they don't show large differences
         assert max_diff > 1e-6
 
-    def test_flux_prompt_embeds(self):
-        pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
-        inputs = self.get_dummy_inputs(torch_device)
-
-        output_with_prompt = pipe(**inputs).images[0]
-
-        inputs = self.get_dummy_inputs(torch_device)
-        prompt = inputs.pop("prompt")
-
-        (prompt_embeds, pooled_prompt_embeds, text_ids) = pipe.encode_prompt(
-            prompt,
-            prompt_2=None,
-            device=torch_device,
-            max_sequence_length=inputs["max_sequence_length"],
-        )
-        output_with_embeds = pipe(
-            prompt_embeds=prompt_embeds,
-            pooled_prompt_embeds=pooled_prompt_embeds,
-            **inputs,
-        ).images[0]
-
-        max_diff = np.abs(output_with_prompt - output_with_embeds).max()
-        assert max_diff < 1e-4
-
     def test_flux_image_output_shape(self):
         pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
         inputs = self.get_dummy_inputs(torch_device)

tests/pipelines/flux/test_pipeline_flux_control_inpaint.py

@@ -120,46 +120,6 @@ def get_dummy_inputs(self, device, seed=0):
         }
         return inputs
 
-    # def test_flux_different_prompts(self):
-    #     pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
-
-    #     inputs = self.get_dummy_inputs(torch_device)
-    #     output_same_prompt = pipe(**inputs).images[0]
-
-    #     inputs = self.get_dummy_inputs(torch_device)
-    #     inputs["prompt_2"] = "a different prompt"
-    #     output_different_prompts = pipe(**inputs).images[0]
-
-    #     max_diff = np.abs(output_same_prompt - output_different_prompts).max()
-
-    #     # Outputs should be different here
-    #     # For some reasons, they don't show large differences
-    #     assert max_diff > 1e-6
-
-    def test_flux_prompt_embeds(self):
-        pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
-        inputs = self.get_dummy_inputs(torch_device)
-
-        output_with_prompt = pipe(**inputs).images[0]
-
-        inputs = self.get_dummy_inputs(torch_device)
-        prompt = inputs.pop("prompt")
-
-        (prompt_embeds, pooled_prompt_embeds, text_ids) = pipe.encode_prompt(
-            prompt,
-            prompt_2=None,
-            device=torch_device,
-            max_sequence_length=inputs["max_sequence_length"],
-        )
-        output_with_embeds = pipe(
-            prompt_embeds=prompt_embeds,
-            pooled_prompt_embeds=pooled_prompt_embeds,
-            **inputs,
-        ).images[0]
-
-        max_diff = np.abs(output_with_prompt - output_with_embeds).max()
-        assert max_diff < 1e-4
-
     def test_fused_qkv_projections(self):
         device = "cpu"  # ensure determinism for the device-dependent torch.Generator
         components = self.get_dummy_components()

tests/pipelines/flux/test_pipeline_flux_fill.py

@@ -128,30 +128,6 @@ def test_flux_fill_different_prompts(self):
         # For some reasons, they don't show large differences
         assert max_diff > 1e-6
 
-    def test_flux_fill_prompt_embeds(self):
-        pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
-        inputs = self.get_dummy_inputs(torch_device)
-
-        output_with_prompt = pipe(**inputs).images[0]
-
-        inputs = self.get_dummy_inputs(torch_device)
-        prompt = inputs.pop("prompt")
-
-        (prompt_embeds, pooled_prompt_embeds, text_ids) = pipe.encode_prompt(
-            prompt,
-            prompt_2=None,
-            device=torch_device,
-            max_sequence_length=inputs["max_sequence_length"],
-        )
-        output_with_embeds = pipe(
-            prompt_embeds=prompt_embeds,
-            pooled_prompt_embeds=pooled_prompt_embeds,
-            **inputs,
-        ).images[0]
-
-        max_diff = np.abs(output_with_prompt - output_with_embeds).max()
-        assert max_diff < 1e-4
-
     def test_flux_image_output_shape(self):
         pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
         inputs = self.get_dummy_inputs(torch_device)

tests/pipelines/flux/test_pipeline_flux_img2img.py

@@ -126,30 +126,6 @@ def test_flux_different_prompts(self):
         # For some reasons, they don't show large differences
         assert max_diff > 1e-6
 
-    def test_flux_prompt_embeds(self):
-        pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
-        inputs = self.get_dummy_inputs(torch_device)
-
-        output_with_prompt = pipe(**inputs).images[0]
-
-        inputs = self.get_dummy_inputs(torch_device)
-        prompt = inputs.pop("prompt")
-
-        (prompt_embeds, pooled_prompt_embeds, text_ids) = pipe.encode_prompt(
-            prompt,
-            prompt_2=None,
-            device=torch_device,
-            max_sequence_length=inputs["max_sequence_length"],
-        )
-        output_with_embeds = pipe(
-            prompt_embeds=prompt_embeds,
-            pooled_prompt_embeds=pooled_prompt_embeds,
-            **inputs,
-        ).images[0]
-
-        max_diff = np.abs(output_with_prompt - output_with_embeds).max()
-        assert max_diff < 1e-4
-
     def test_flux_image_output_shape(self):
         pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
         inputs = self.get_dummy_inputs(torch_device)

tests/pipelines/flux/test_pipeline_flux_inpaint.py

@@ -128,30 +128,6 @@ def test_flux_inpaint_different_prompts(self):
         # For some reasons, they don't show large differences
         assert max_diff > 1e-6
 
-    def test_flux_inpaint_prompt_embeds(self):
-        pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
-        inputs = self.get_dummy_inputs(torch_device)
-
-        output_with_prompt = pipe(**inputs).images[0]
-
-        inputs = self.get_dummy_inputs(torch_device)
-        prompt = inputs.pop("prompt")
-
-        (prompt_embeds, pooled_prompt_embeds, text_ids) = pipe.encode_prompt(
-            prompt,
-            prompt_2=None,
-            device=torch_device,
-            max_sequence_length=inputs["max_sequence_length"],
-        )
-        output_with_embeds = pipe(
-            prompt_embeds=prompt_embeds,
-            pooled_prompt_embeds=pooled_prompt_embeds,
-            **inputs,
-        ).images[0]
-
-        max_diff = np.abs(output_with_prompt - output_with_embeds).max()
-        assert max_diff < 1e-4
-
     def test_flux_image_output_shape(self):
         pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
         inputs = self.get_dummy_inputs(torch_device)

tests/pipelines/hunyuandit/test_hunyuan_dit.py

@@ -14,7 +14,6 @@
 # limitations under the License.
 
 import gc
-import tempfile
 import unittest
 
 import numpy as np
@@ -128,10 +127,12 @@ def test_inference(self):
         max_diff = np.abs(image_slice.flatten() - expected_slice).max()
         self.assertLessEqual(max_diff, 1e-3)
 
+    @unittest.skip("Not supported.")
     def test_sequential_cpu_offload_forward_pass(self):
         # TODO(YiYi) need to fix later
         pass
 
+    @unittest.skip("Not supported.")
     def test_sequential_offload_forward_pass_twice(self):
         # TODO(YiYi) need to fix later
         pass
@@ -141,99 +142,6 @@ def test_inference_batch_single_identical(self):
             expected_max_diff=1e-3,
         )
 
-    def test_save_load_optional_components(self):
-        components = self.get_dummy_components()
-        pipe = self.pipeline_class(**components)
-        pipe.to(torch_device)
-        pipe.set_progress_bar_config(disable=None)
-
-        inputs = self.get_dummy_inputs(torch_device)
-
-        prompt = inputs["prompt"]
-        generator = inputs["generator"]
-        num_inference_steps = inputs["num_inference_steps"]
-        output_type = inputs["output_type"]
-
-        (
-            prompt_embeds,
-            negative_prompt_embeds,
-            prompt_attention_mask,
-            negative_prompt_attention_mask,
-        ) = pipe.encode_prompt(prompt, device=torch_device, dtype=torch.float32, text_encoder_index=0)
-
-        (
-            prompt_embeds_2,
-            negative_prompt_embeds_2,
-            prompt_attention_mask_2,
-            negative_prompt_attention_mask_2,
-        ) = pipe.encode_prompt(
-            prompt,
-            device=torch_device,
-            dtype=torch.float32,
-            text_encoder_index=1,
-        )
-
-        # inputs with prompt converted to embeddings
-        inputs = {
-            "prompt_embeds": prompt_embeds,
-            "prompt_attention_mask": prompt_attention_mask,
-            "negative_prompt_embeds": negative_prompt_embeds,
-            "negative_prompt_attention_mask": negative_prompt_attention_mask,
-            "prompt_embeds_2": prompt_embeds_2,
-            "prompt_attention_mask_2": prompt_attention_mask_2,
-            "negative_prompt_embeds_2": negative_prompt_embeds_2,
-            "negative_prompt_attention_mask_2": negative_prompt_attention_mask_2,
-            "generator": generator,
-            "num_inference_steps": num_inference_steps,
-            "output_type": output_type,
-            "use_resolution_binning": False,
-        }
-
-        # set all optional components to None
-        for optional_component in pipe._optional_components:
-            setattr(pipe, optional_component, None)
-
-        output = pipe(**inputs)[0]
-
-        with tempfile.TemporaryDirectory() as tmpdir:
-            pipe.save_pretrained(tmpdir)
-            pipe_loaded = self.pipeline_class.from_pretrained(tmpdir)
-            pipe_loaded.to(torch_device)
-            pipe_loaded.set_progress_bar_config(disable=None)
-
-        for optional_component in pipe._optional_components:
-            self.assertTrue(
-                getattr(pipe_loaded, optional_component) is None,
-                f"`{optional_component}` did not stay set to None after loading.",
-            )
-
-        inputs = self.get_dummy_inputs(torch_device)
-
-        generator = inputs["generator"]
-        num_inference_steps = inputs["num_inference_steps"]
-        output_type = inputs["output_type"]
-
-        # inputs with prompt converted to embeddings
-        inputs = {
-            "prompt_embeds": prompt_embeds,
-            "prompt_attention_mask": prompt_attention_mask,
-            "negative_prompt_embeds": negative_prompt_embeds,
-            "negative_prompt_attention_mask": negative_prompt_attention_mask,
-            "prompt_embeds_2": prompt_embeds_2,
-            "prompt_attention_mask_2": prompt_attention_mask_2,
-            "negative_prompt_embeds_2": negative_prompt_embeds_2,
-            "negative_prompt_attention_mask_2": negative_prompt_attention_mask_2,
-            "generator": generator,
-            "num_inference_steps": num_inference_steps,
-            "output_type": output_type,
-            "use_resolution_binning": False,
-        }
-
-        output_loaded = pipe_loaded(**inputs)[0]
-
-        max_diff = np.abs(to_np(output) - to_np(output_loaded)).max()
-        self.assertLess(max_diff, 1e-4)
-
     def test_feed_forward_chunking(self):
         device = "cpu"