Merge branch 'main' into automodel-custom-model

Author: DN6

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

@@ -515,6 +515,7 @@
             "QwenImageControlNetPipeline",
             "QwenImageEditInpaintPipeline",
             "QwenImageEditPipeline",
+            "QwenImageEditPlusPipeline",
             "QwenImageImg2ImgPipeline",
             "QwenImageInpaintPipeline",
             "QwenImagePipeline",
@@ -1170,6 +1171,7 @@
             QwenImageControlNetPipeline,
             QwenImageEditInpaintPipeline,
             QwenImageEditPipeline,
+            QwenImageEditPlusPipeline,
             QwenImageImg2ImgPipeline,
             QwenImageInpaintPipeline,
             QwenImagePipeline,

src/diffusers/pipelines/__init__.py

@@ -394,6 +394,7 @@
         "QwenImageImg2ImgPipeline",
         "QwenImageInpaintPipeline",
         "QwenImageEditPipeline",
+        "QwenImageEditPlusPipeline",
         "QwenImageEditInpaintPipeline",
         "QwenImageControlNetInpaintPipeline",
         "QwenImageControlNetPipeline",
@@ -721,6 +722,7 @@
             QwenImageControlNetPipeline,
             QwenImageEditInpaintPipeline,
             QwenImageEditPipeline,
+            QwenImageEditPlusPipeline,
             QwenImageImg2ImgPipeline,
             QwenImageInpaintPipeline,
             QwenImagePipeline,

src/diffusers/pipelines/pipeline_loading_utils.py

@@ -48,10 +48,12 @@
 if is_transformers_available():
     import transformers
     from transformers import PreTrainedModel, PreTrainedTokenizerBase
-    from transformers.utils import FLAX_WEIGHTS_NAME as TRANSFORMERS_FLAX_WEIGHTS_NAME
     from transformers.utils import SAFE_WEIGHTS_NAME as TRANSFORMERS_SAFE_WEIGHTS_NAME
     from transformers.utils import WEIGHTS_NAME as TRANSFORMERS_WEIGHTS_NAME
 
+    if is_transformers_version("<=", "4.56.2"):
+        from transformers.utils import FLAX_WEIGHTS_NAME as TRANSFORMERS_FLAX_WEIGHTS_NAME
+
 if is_accelerate_available():
     import accelerate
     from accelerate import dispatch_model
@@ -112,7 +114,9 @@ def is_safetensors_compatible(filenames, passed_components=None, folder_names=No
     ]
 
     if is_transformers_available():
-        weight_names += [TRANSFORMERS_WEIGHTS_NAME, TRANSFORMERS_SAFE_WEIGHTS_NAME, TRANSFORMERS_FLAX_WEIGHTS_NAME]
+        weight_names += [TRANSFORMERS_WEIGHTS_NAME, TRANSFORMERS_SAFE_WEIGHTS_NAME]
+        if is_transformers_version("<=", "4.56.2"):
+            weight_names += [TRANSFORMERS_FLAX_WEIGHTS_NAME]
 
     # model_pytorch, diffusion_model_pytorch, ...
     weight_prefixes = [w.split(".")[0] for w in weight_names]
@@ -191,7 +195,9 @@ def filter_model_files(filenames):
     ]
 
     if is_transformers_available():
-        weight_names += [TRANSFORMERS_WEIGHTS_NAME, TRANSFORMERS_SAFE_WEIGHTS_NAME, TRANSFORMERS_FLAX_WEIGHTS_NAME]
+        weight_names += [TRANSFORMERS_WEIGHTS_NAME, TRANSFORMERS_SAFE_WEIGHTS_NAME]
+        if is_transformers_version("<=", "4.56.2"):
+            weight_names += [TRANSFORMERS_FLAX_WEIGHTS_NAME]
 
     allowed_extensions = [wn.split(".")[-1] for wn in weight_names]
 
@@ -212,7 +218,9 @@ def variant_compatible_siblings(filenames, variant=None, ignore_patterns=None) -
     ]
 
     if is_transformers_available():
-        weight_names += [TRANSFORMERS_WEIGHTS_NAME, TRANSFORMERS_SAFE_WEIGHTS_NAME, TRANSFORMERS_FLAX_WEIGHTS_NAME]
+        weight_names += [TRANSFORMERS_WEIGHTS_NAME, TRANSFORMERS_SAFE_WEIGHTS_NAME]
+        if is_transformers_version("<=", "4.56.2"):
+            weight_names += [TRANSFORMERS_FLAX_WEIGHTS_NAME]
 
     # model_pytorch, diffusion_model_pytorch, ...
     weight_prefixes = [w.split(".")[0] for w in weight_names]

src/diffusers/pipelines/qwenimage/__init__.py

@@ -28,6 +28,7 @@
     _import_structure["pipeline_qwenimage_controlnet_inpaint"] = ["QwenImageControlNetInpaintPipeline"]
     _import_structure["pipeline_qwenimage_edit"] = ["QwenImageEditPipeline"]
     _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"]
 
@@ -43,6 +44,7 @@
         from .pipeline_qwenimage_controlnet_inpaint import QwenImageControlNetInpaintPipeline
         from .pipeline_qwenimage_edit import QwenImageEditPipeline
         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"