Merge branch 'main' into use-fixture-modular-tests

docs/source/en/api/modular_diffusers/pipeline_blocks.md

@@ -14,8 +14,4 @@
 
 ## AutoPipelineBlocks
 
-[[autodoc]] diffusers.modular_pipelines.modular_pipeline.AutoPipelineBlocks
-
-## ConditionalPipelineBlocks
-
-[[autodoc]] diffusers.modular_pipelines.modular_pipeline.ConditionalPipelineBlocks
+[[autodoc]] diffusers.modular_pipelines.modular_pipeline.AutoPipelineBlocks

docs/source/en/modular_diffusers/auto_pipeline_blocks.md

@@ -121,7 +121,7 @@ from diffusers.modular_pipelines import AutoPipelineBlocks
 
 class AutoImageBlocks(AutoPipelineBlocks):
     # List of sub-block classes to choose from
-    block_classes = [InpaintBlock, ImageToImageBlock, TextToImageBlock]
+    block_classes = [block_inpaint_cls, block_i2i_cls, block_t2i_cls]
     # Names for each block in the same order
     block_names = ["inpaint", "img2img", "text2img"]
     # Trigger inputs that determine which block to run
@@ -129,8 +129,8 @@ class AutoImageBlocks(AutoPipelineBlocks):
     # - "image" triggers img2img workflow (but only if mask is not provided)
     # - if none of above, runs the text2img workflow (default)
     block_trigger_inputs = ["mask", "image", None]
+    # Description is extremely important for AutoPipelineBlocks
 
-    @property
     def description(self):
         return (
             "Pipeline generates images given different types of conditions!\n"
@@ -141,7 +141,7 @@ class AutoImageBlocks(AutoPipelineBlocks):
         )
 ```
 
-It is **very** important to include a `description` to avoid any confusion over how to run a block and what inputs are required. While [`~modular_pipelines.AutoPipelineBlocks`] are convenient, its conditional logic may be difficult to figure out if it isn't properly explained.
+It is **very** important to include a `description` to avoid any confusion over how to run a block and what inputs are required. While [`~modular_pipelines.AutoPipelineBlocks`] are convenient, it's conditional logic may be difficult to figure out if it isn't properly explained.
 
 Create an instance of `AutoImageBlocks`.
 
@@ -152,74 +152,5 @@ auto_blocks = AutoImageBlocks()
 For more complex compositions, such as nested [`~modular_pipelines.AutoPipelineBlocks`] blocks when they're used as sub-blocks in larger pipelines, use the [`~modular_pipelines.SequentialPipelineBlocks.get_execution_blocks`] method to extract the a block that is actually run based on your input.
 
 ```py
-auto_blocks.get_execution_blocks(mask=True)
-```
-
-## ConditionalPipelineBlocks
-
-[`~modular_pipelines.AutoPipelineBlocks`] is a special case of [`~modular_pipelines.ConditionalPipelineBlocks`]. While [`~modular_pipelines.AutoPipelineBlocks`] selects blocks based on whether a trigger input is provided or not, [`~modular_pipelines.ConditionalPipelineBlocks`] is able to select a block based on custom selection logic provided in the `select_block` method.
-
-Here is the same example written using [`~modular_pipelines.ConditionalPipelineBlocks`] directly:
-
-```py
-from diffusers.modular_pipelines import ConditionalPipelineBlocks
-
-class AutoImageBlocks(ConditionalPipelineBlocks):
-    block_classes = [InpaintBlock, ImageToImageBlock, TextToImageBlock]
-    block_names = ["inpaint", "img2img", "text2img"]
-    block_trigger_inputs = ["mask", "image"]
-    default_block_name = "text2img"
-
-    @property
-    def description(self):
-        return (
-            "Pipeline generates images given different types of conditions!\n"
-            + "This is an auto pipeline block that works for text2img, img2img and inpainting tasks.\n"
-            + " - inpaint workflow is run when `mask` is provided.\n"
-            + " - img2img workflow is run when `image` is provided (but only when `mask` is not provided).\n"
-            + " - text2img workflow is run when neither `image` nor `mask` is provided.\n"
-        )
-
-    def select_block(self, mask=None, image=None) -> str | None:
-        if mask is not None:
-            return "inpaint"
-        if image is not None:
-            return "img2img"
-        return None  # falls back to default_block_name ("text2img")
-```
-
-The inputs listed in `block_trigger_inputs` are passed as keyword arguments to `select_block()`. When `select_block` returns `None`, it falls back to `default_block_name`. If `default_block_name` is also `None`, the entire conditional block is skipped — this is useful for optional processing steps that should only run when specific inputs are provided.
-
-## Workflows
-
-Pipelines that contain conditional blocks ([`~modular_pipelines.AutoPipelineBlocks`] or [`~modular_pipelines.ConditionalPipelineBlocks]`) can support multiple workflows — for example, our SDXL modular pipeline supports a dozen workflows all in one pipeline. But this also means it can be confusing for users to know what workflows are supported and how to run them. For pipeline builders, it's useful to be able to extract only the blocks relevant to a specific workflow.
-
-We recommend defining a `_workflow_map` to give each workflow a name and explicitly list the inputs it requires.
-
-```py
-from diffusers.modular_pipelines import SequentialPipelineBlocks
-
-class MyPipelineBlocks(SequentialPipelineBlocks):
-    block_classes = [TextEncoderBlock, AutoImageBlocks, DecodeBlock]
-    block_names = ["text_encoder", "auto_image", "decode"]
-
-    _workflow_map = {
-        "text2image": {"prompt": True},
-        "image2image": {"image": True, "prompt": True},
-        "inpaint": {"mask": True, "image": True, "prompt": True},
-    }
-```
-
-All of our built-in modular pipelines come with pre-defined workflows. The `available_workflows` property lists all supported workflows:
-
-```py
-pipeline_blocks = MyPipelineBlocks()
-pipeline_blocks.available_workflows
-# ['text2image', 'image2image', 'inpaint']
-```
-
-Retrieve a specific workflow with `get_workflow` to inspect and debug a specific block that executes the workflow.
-
-```py
-pipeline_blocks.get_workflow("inpaint")
+auto_blocks.get_execution_blocks("mask")
 ```

src/diffusers/loaders/lora_conversion_utils.py

@@ -856,7 +856,7 @@ def _convert_kohya_flux_lora_to_diffusers(state_dict):
                 )
             state_dict = {k: v for k, v in state_dict.items() if not k.startswith("text_encoders.t5xxl.transformer.")}
 
-        has_diffb = any("diff_b" in k and k.startswith(("lora_unet_", "lora_te_", "lora_te1_")) for k in state_dict)
+        has_diffb = any("diff_b" in k and k.startswith(("lora_unet_", "lora_te_")) for k in state_dict)
         if has_diffb:
             zero_status_diff_b = state_dict_all_zero(state_dict, ".diff_b")
             if zero_status_diff_b:
@@ -895,7 +895,7 @@ def _convert_kohya_flux_lora_to_diffusers(state_dict):
         state_dict = {
             _custom_replace(k, limit_substrings): v
             for k, v in state_dict.items()
-            if k.startswith(("lora_unet_", "lora_te_", "lora_te1_"))
+            if k.startswith(("lora_unet_", "lora_te_"))
         }
 
         if any("text_projection" in k for k in state_dict):

src/diffusers/modular_pipelines/modular_pipeline.py

@@ -1633,14 +1633,7 @@ class ModularPipeline(ConfigMixin, PushToHubMixin):
                 blocks_class_name = self.default_blocks_name
             if blocks_class_name is not None:
                 diffusers_module = importlib.import_module("diffusers")
-                blocks_class = getattr(diffusers_module, blocks_class_name, None)
-                # If the blocks_class is not found or is a base class (e.g. SequentialPipelineBlocks saved by from_blocks_dict) with empty block_classes
-                # fall back to default_blocks_name
-                if blocks_class is None or not blocks_class.block_classes:
-                    blocks_class_name = self.default_blocks_name
-                    blocks_class = getattr(diffusers_module, blocks_class_name)
-
-            if blocks_class is not None:
+                blocks_class = getattr(diffusers_module, blocks_class_name)
                 blocks = blocks_class()
             else:
                 logger.warning(f"`blocks` is `None`, no default blocks class found for {self.__class__.__name__}")

tests/modular_pipelines/flux/test_modular_pipeline_flux.py

@@ -14,7 +14,6 @@
 # limitations under the License.
 
 import random
-import tempfile
 
 import numpy as np
 import PIL
@@ -129,18 +128,16 @@ class TestFluxImg2ImgModularPipelineFast(ModularPipelineTesterMixin):
 
         return inputs
 
-    def test_save_from_pretrained(self):
+    def test_save_from_pretrained(self, tmp_path):
         pipes = []
         base_pipe = self.get_pipeline().to(torch_device)
         pipes.append(base_pipe)
 
-        with tempfile.TemporaryDirectory() as tmpdirname:
-            base_pipe.save_pretrained(tmpdirname)
-
-            pipe = ModularPipeline.from_pretrained(tmpdirname).to(torch_device)
-            pipe.load_components(torch_dtype=torch.float32)
-            pipe.to(torch_device)
-            pipe.image_processor = VaeImageProcessor(vae_scale_factor=2)
+        base_pipe.save_pretrained(tmp_path)
+        pipe = ModularPipeline.from_pretrained(tmp_path).to(torch_device)
+        pipe.load_components(torch_dtype=torch.float32)
+        pipe.to(torch_device)
+        pipe.image_processor = VaeImageProcessor(vae_scale_factor=2)
 
         pipes.append(pipe)
 
@@ -212,18 +209,16 @@ class TestFluxKontextModularPipelineFast(ModularPipelineTesterMixin):
 
         return inputs
 
-    def test_save_from_pretrained(self):
+    def test_save_from_pretrained(self, tmp_path):
         pipes = []
         base_pipe = self.get_pipeline().to(torch_device)
         pipes.append(base_pipe)
 
-        with tempfile.TemporaryDirectory() as tmpdirname:
-            base_pipe.save_pretrained(tmpdirname)
-
-            pipe = ModularPipeline.from_pretrained(tmpdirname).to(torch_device)
-            pipe.load_components(torch_dtype=torch.float32)
-            pipe.to(torch_device)
-            pipe.image_processor = VaeImageProcessor(vae_scale_factor=2)
+        base_pipe.save_pretrained(tmp_path)
+        pipe = ModularPipeline.from_pretrained(tmp_path).to(torch_device)
+        pipe.load_components(torch_dtype=torch.float32)
+        pipe.to(torch_device)
+        pipe.image_processor = VaeImageProcessor(vae_scale_factor=2)
 
         pipes.append(pipe)
 

tests/modular_pipelines/test_modular_pipelines_common.py

@@ -1,7 +1,4 @@
 import gc
-import json
-import os
-import tempfile
 from typing import Callable
 
 import pytest
@@ -330,16 +327,15 @@ class ModularPipelineTesterMixin:
 
         assert torch.abs(image_slices[0] - image_slices[1]).max() < 1e-3
 
-    def test_save_from_pretrained(self):
+    def test_save_from_pretrained(self, tmp_path):
         pipes = []
         base_pipe = self.get_pipeline().to(torch_device)
         pipes.append(base_pipe)
 
-        with tempfile.TemporaryDirectory() as tmpdirname:
-            base_pipe.save_pretrained(tmpdirname)
-            pipe = ModularPipeline.from_pretrained(tmpdirname).to(torch_device)
-            pipe.load_components(torch_dtype=torch.float32)
-            pipe.to(torch_device)
+        base_pipe.save_pretrained(tmp_path)
+        pipe = ModularPipeline.from_pretrained(tmp_path).to(torch_device)
+        pipe.load_components(torch_dtype=torch.float32)
+        pipe.to(torch_device)
 
         pipes.append(pipe)
 
@@ -351,33 +347,6 @@ class ModularPipelineTesterMixin:
 
         assert torch.abs(image_slices[0] - image_slices[1]).max() < 1e-3
 
-    def test_modular_index_consistency(self):
-        pipe = self.get_pipeline()
-        components_spec = pipe._component_specs
-        components = sorted(components_spec.keys())
-
-        with tempfile.TemporaryDirectory() as tmpdir:
-            pipe.save_pretrained(tmpdir)
-            index_file = os.path.join(tmpdir, "modular_model_index.json")
-            assert os.path.exists(index_file)
-
-            with open(index_file) as f:
-                index_contents = json.load(f)
-
-            compulsory_keys = {"_blocks_class_name", "_class_name", "_diffusers_version"}
-            for k in compulsory_keys:
-                assert k in index_contents
-
-            to_check_attrs = {"pretrained_model_name_or_path", "revision", "subfolder"}
-            for component in components:
-                spec = components_spec[component]
-                for attr in to_check_attrs:
-                    if getattr(spec, "pretrained_model_name_or_path", None) is not None:
-                        for attr in to_check_attrs:
-                            assert component in index_contents, f"{component} should be present in index but isn't."
-                            attr_value_from_index = index_contents[component][2][attr]
-                            assert getattr(spec, attr) == attr_value_from_index
-
     def test_workflow_map(self):
         blocks = self.pipeline_blocks_class()
         if blocks._workflow_map is None:
@@ -728,27 +697,3 @@ class TestLoadComponentsSkipBehavior:
 
         # Verify test_component was not loaded
         assert not hasattr(pipe, "test_component") or pipe.test_component is None
-
-
-class TestModularPipelineInitFallback:
-    """Test that ModularPipeline.__init__ falls back to default_blocks_name when
-    _blocks_class_name is a base class (e.g. SequentialPipelineBlocks saved by from_blocks_dict)."""
-
-    def test_init_fallback_when_blocks_class_name_is_base_class(self, tmp_path):
-        # 1. Load pipeline and get a workflow (returns a base SequentialPipelineBlocks)
-        pipe = ModularPipeline.from_pretrained("hf-internal-testing/tiny-stable-diffusion-xl-pipe")
-        t2i_blocks = pipe.blocks.get_workflow("text2image")
-        assert t2i_blocks.__class__.__name__ == "SequentialPipelineBlocks"
-
-        # 2. Use init_pipeline to create a new pipeline from the workflow blocks
-        t2i_pipe = t2i_blocks.init_pipeline("hf-internal-testing/tiny-stable-diffusion-xl-pipe")
-
-        # 3. Save and reload — the saved config will have _blocks_class_name="SequentialPipelineBlocks"
-        save_dir = str(tmp_path / "pipeline")
-        t2i_pipe.save_pretrained(save_dir)
-        loaded_pipe = ModularPipeline.from_pretrained(save_dir)
-
-        # 4. Verify it fell back to default_blocks_name and has correct blocks
-        assert loaded_pipe.__class__.__name__ == pipe.__class__.__name__
-        assert loaded_pipe._blocks.__class__.__name__ == pipe._blocks.__class__.__name__
-        assert len(loaded_pipe._blocks.sub_blocks) == len(pipe._blocks.sub_blocks)

tests/modular_pipelines/test_modular_pipelines_custom_blocks.py

@@ -14,7 +14,6 @@
 
 import json
 import os
-import tempfile
 from collections import deque
 from typing import List
 
@@ -153,25 +152,24 @@ class TestModularCustomBlocks:
         output_prompt = output.values["output_prompt"]
         assert output_prompt.startswith("Modular diffusers + ")
 
-    def test_custom_block_saving_loading(self):
+    def test_custom_block_saving_loading(self, tmp_path):
         custom_block = DummyCustomBlockSimple()
 
-        with tempfile.TemporaryDirectory() as tmpdir:
-            custom_block.save_pretrained(tmpdir)
-            assert any("modular_config.json" in k for k in os.listdir(tmpdir))
+        custom_block.save_pretrained(tmp_path)
+        assert any("modular_config.json" in k for k in os.listdir(tmp_path))
 
-            with open(os.path.join(tmpdir, "modular_config.json"), "r") as f:
-                config = json.load(f)
-            auto_map = config["auto_map"]
-            assert auto_map == {"ModularPipelineBlocks": "test_modular_pipelines_custom_blocks.DummyCustomBlockSimple"}
+        with open(os.path.join(tmp_path, "modular_config.json"), "r") as f:
+            config = json.load(f)
+        auto_map = config["auto_map"]
+        assert auto_map == {"ModularPipelineBlocks": "test_modular_pipelines_custom_blocks.DummyCustomBlockSimple"}
 
-            # For now, the Python script that implements the custom block has to be manually pushed to the Hub.
-            # This is why, we have to separately save the Python script here.
-            code_path = os.path.join(tmpdir, "test_modular_pipelines_custom_blocks.py")
-            with open(code_path, "w") as f:
-                f.write(CODE_STR)
+        # For now, the Python script that implements the custom block has to be manually pushed to the Hub.
+        # This is why, we have to separately save the Python script here.
+        code_path = os.path.join(tmp_path, "test_modular_pipelines_custom_blocks.py")
+        with open(code_path, "w") as f:
+            f.write(CODE_STR)
 
-            loaded_custom_block = ModularPipelineBlocks.from_pretrained(tmpdir, trust_remote_code=True)
+        loaded_custom_block = ModularPipelineBlocks.from_pretrained(tmp_path, trust_remote_code=True)
 
         pipe = loaded_custom_block.init_pipeline()
         prompt = "Diffusers is nice"