Merge branch 'main' into qwen-image-edit-controlnet

Author: dimitribarbot

examples/server/README.md

@@ -9,8 +9,8 @@ This guide will show you how to use the [`StableDiffusion3Pipeline`] in a server
 Start by navigating to the `examples/server` folder and installing all of the dependencies.
 
 ```py
-pip install .
-pip install -f requirements.txt
+pip install diffusers
+pip install -r requirements.txt
 ```
 
 Launch the server with the following command.

examples/server/requirements.in

@@ -6,4 +6,5 @@ py-consul
 prometheus_client >= 0.18.0
 prometheus-fastapi-instrumentator >= 7.0.0
 fastapi
-uvicorn
+uvicorn
+accelerate

examples/server/requirements.txt

@@ -39,7 +39,7 @@ fsspec==2024.10.0
     #   torch
 h11==0.14.0
     # via uvicorn
-huggingface-hub==0.26.1
+huggingface-hub==0.35.0
     # via
     #   tokenizers
     #   transformers

src/diffusers/__init__.py

@@ -516,6 +516,7 @@
             "QwenImageEditControlNetPipeline",
             "QwenImageEditInpaintPipeline",
             "QwenImageEditPipeline",
+            "QwenImageEditPlusPipeline",
             "QwenImageImg2ImgPipeline",
             "QwenImageInpaintPipeline",
             "QwenImagePipeline",
@@ -1172,6 +1173,7 @@
             QwenImageEditControlNetPipeline,
             QwenImageEditInpaintPipeline,
             QwenImageEditPipeline,
+            QwenImageEditPlusPipeline,
             QwenImageImg2ImgPipeline,
             QwenImageInpaintPipeline,
             QwenImagePipeline,

src/diffusers/models/attention.py

@@ -241,7 +241,7 @@ def set_use_memory_efficient_attention_xformers(
                             op_fw, op_bw = attention_op
                             dtype, *_ = op_fw.SUPPORTED_DTYPES
                         q = torch.randn((1, 2, 40), device="cuda", dtype=dtype)
-                        _ = xops.memory_efficient_attention(q, q, q)
+                        _ = xops.ops.memory_efficient_attention(q, q, q)
                 except Exception as e:
                     raise e
 

src/diffusers/models/autoencoders/autoencoder_dc.py

@@ -617,7 +617,7 @@ def decode(self, z: torch.Tensor, return_dict: bool = True) -> Union[DecoderOutp
                 returned.
         """
         if self.use_slicing and z.size(0) > 1:
-            decoded_slices = [self._decode(z_slice).sample for z_slice in z.split(1)]
+            decoded_slices = [self._decode(z_slice) for z_slice in z.split(1)]
             decoded = torch.cat(decoded_slices)
         else:
             decoded = self._decode(z)

src/diffusers/pipelines/__init__.py

@@ -394,6 +394,7 @@
         "QwenImageImg2ImgPipeline",
         "QwenImageInpaintPipeline",
         "QwenImageEditPipeline",
+        "QwenImageEditPlusPipeline",
         "QwenImageEditInpaintPipeline",
         "QwenImageEditControlNetPipeline",
         "QwenImageControlNetInpaintPipeline",
@@ -723,6 +724,7 @@
             QwenImageEditControlNetPipeline,
             QwenImageEditInpaintPipeline,
             QwenImageEditPipeline,
+            QwenImageEditPlusPipeline,
             QwenImageImg2ImgPipeline,
             QwenImageInpaintPipeline,
             QwenImagePipeline,

src/diffusers/pipelines/qwenimage/__init__.py

@@ -29,6 +29,7 @@
     _import_structure["pipeline_qwenimage_edit"] = ["QwenImageEditPipeline"]
     _import_structure["pipeline_qwenimage_edit_controlnet"] = ["QwenImageEditControlNetPipeline"]
     _import_structure["pipeline_qwenimage_edit_inpaint"] = ["QwenImageEditInpaintPipeline"]
+    _import_structure["pipeline_qwenimage_edit_plus"] = ["QwenImageEditPlusPipeline"]
     _import_structure["pipeline_qwenimage_img2img"] = ["QwenImageImg2ImgPipeline"]
     _import_structure["pipeline_qwenimage_inpaint"] = ["QwenImageInpaintPipeline"]
 
@@ -45,6 +46,7 @@
         from .pipeline_qwenimage_edit import QwenImageEditPipeline
         from .pipeline_qwenimage_edit_controlnet import QwenImageEditControlNetPipeline
         from .pipeline_qwenimage_edit_inpaint import QwenImageEditInpaintPipeline
+        from .pipeline_qwenimage_edit_plus import QwenImageEditPlusPipeline
         from .pipeline_qwenimage_img2img import QwenImageImg2ImgPipeline
         from .pipeline_qwenimage_inpaint import QwenImageInpaintPipeline
 else:

src/diffusers/pipelines/qwenimage/pipeline_qwenimage_edit.py

@@ -208,7 +208,6 @@ def __init__(
         # QwenImage latents are turned into 2x2 patches and packed. This means the latent width and height has to be divisible
         # by the patch size. So the vae scale factor is multiplied by the patch size to account for this
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor * 2)
-        self.vl_processor = processor
         self.tokenizer_max_length = 1024
 
         self.prompt_template_encode = "<|im_start|>system\nDescribe the key features of the input image (color, shape, size, texture, objects, background), then explain how the user's text instruction should alter or modify the image. Generate a new image that meets the user's requirements while maintaining consistency with the original input where appropriate.<|im_end|>\n<|im_start|>user\n<|vision_start|><|image_pad|><|vision_end|>{}<|im_end|>\n<|im_start|>assistant\n"