Merge branch 'main' into fix-dpmsolvermultistepinversescheduler-for-flowmatching

Author: bhzhang2001

docs/source/en/api/pipelines/qwenimage.md

@@ -120,6 +120,12 @@ The `guidance_scale` parameter in the pipeline is there to support future guidan
   - all
   - __call__
 
+## QwenImageEditInpaintPipeline
+
+[[autodoc]] QwenImageEditInpaintPipeline
+  - all
+  - __call__
+
 ## QwenImaggeControlNetPipeline
   - all
   - __call__

src/diffusers/__init__.py

@@ -494,6 +494,7 @@
             "PixArtSigmaPAGPipeline",
             "PixArtSigmaPipeline",
             "QwenImageControlNetPipeline",
+            "QwenImageEditInpaintPipeline",
             "QwenImageEditPipeline",
             "QwenImageImg2ImgPipeline",
             "QwenImageInpaintPipeline",
@@ -1134,6 +1135,7 @@
             PixArtSigmaPAGPipeline,
             PixArtSigmaPipeline,
             QwenImageControlNetPipeline,
+            QwenImageEditInpaintPipeline,
             QwenImageEditPipeline,
             QwenImageImg2ImgPipeline,
             QwenImageInpaintPipeline,

src/diffusers/models/attention_dispatch.py

@@ -955,12 +955,13 @@ def _native_npu_attention(
     dropout_p: float = 0.0,
     scale: Optional[float] = None,
 ) -> torch.Tensor:
-    return npu_fusion_attention(
+    query, key, value = (x.transpose(1, 2).contiguous() for x in (query, key, value))
+    out = npu_fusion_attention(
         query,
         key,
         value,
-        query.size(2),  # num_heads
-        input_layout="BSND",
+        query.size(1),  # num_heads
+        input_layout="BNSD",
         pse=None,
         scale=1.0 / math.sqrt(query.shape[-1]) if scale is None else scale,
         pre_tockens=65536,
@@ -969,6 +970,8 @@ def _native_npu_attention(
         sync=False,
         inner_precise=0,
     )[0]
+    out = out.transpose(1, 2).contiguous()
+    return out
 
 
 # Reference: https://github.com/pytorch/xla/blob/06c5533de6588f6b90aa1655d9850bcf733b90b4/torch_xla/experimental/custom_kernel.py#L853

src/diffusers/pipelines/__init__.py

@@ -393,6 +393,7 @@
         "QwenImageImg2ImgPipeline",
         "QwenImageInpaintPipeline",
         "QwenImageEditPipeline",
+        "QwenImageEditInpaintPipeline",
         "QwenImageControlNetPipeline",
     ]
 try:
@@ -714,6 +715,7 @@
         from .pixart_alpha import PixArtAlphaPipeline, PixArtSigmaPipeline
         from .qwenimage import (
             QwenImageControlNetPipeline,
+            QwenImageEditInpaintPipeline,
             QwenImageEditPipeline,
             QwenImageImg2ImgPipeline,
             QwenImageInpaintPipeline,

src/diffusers/pipelines/qwenimage/__init__.py

@@ -26,6 +26,7 @@
     _import_structure["pipeline_qwenimage"] = ["QwenImagePipeline"]
     _import_structure["pipeline_qwenimage_controlnet"] = ["QwenImageControlNetPipeline"]
     _import_structure["pipeline_qwenimage_edit"] = ["QwenImageEditPipeline"]
+    _import_structure["pipeline_qwenimage_edit_inpaint"] = ["QwenImageEditInpaintPipeline"]
     _import_structure["pipeline_qwenimage_img2img"] = ["QwenImageImg2ImgPipeline"]
     _import_structure["pipeline_qwenimage_inpaint"] = ["QwenImageInpaintPipeline"]
 
@@ -39,6 +40,7 @@
         from .pipeline_qwenimage import QwenImagePipeline
         from .pipeline_qwenimage_controlnet import QwenImageControlNetPipeline
         from .pipeline_qwenimage_edit import QwenImageEditPipeline
+        from .pipeline_qwenimage_edit_inpaint import QwenImageEditInpaintPipeline
         from .pipeline_qwenimage_img2img import QwenImageImg2ImgPipeline
         from .pipeline_qwenimage_inpaint import QwenImageInpaintPipeline
 else: