Merge branch 'main' into groupwise-offloading

Author: a-r-r-o-w

examples/dreambooth/train_dreambooth_lora_sana.py

@@ -995,7 +995,8 @@ def main(args):
     if args.enable_npu_flash_attention:
         if is_torch_npu_available():
             logger.info("npu flash attention enabled.")
-            transformer.enable_npu_flash_attention()
+            for block in transformer.transformer_blocks:
+                block.attn2.set_use_npu_flash_attention(True)
         else:
             raise ValueError("npu flash attention requires torch_npu extensions and is supported only on npu device ")
 

src/diffusers/loaders/transformer_flux.py

@@ -177,5 +177,3 @@ def _load_ip_adapter_weights(self, state_dicts, low_cpu_mem_usage=False):
 
         self.encoder_hid_proj = MultiIPAdapterImageProjection(image_projection_layers)
         self.config.encoder_hid_dim_type = "ip_image_proj"
-
-        self.to(dtype=self.dtype, device=self.device)