fix: enabled num_images_per_prompt>1 for lpw_stable_diffusion_xl (community pipeline) (#5807)

* fix: enabled num_images_per_prompt>1 for lpw_stable_diffusion_xl

* style: fixed isort

examples/community/lpw_stable_diffusion_xl.py

@@ -249,6 +249,7 @@ def get_weighted_text_embeddings_sdxl(
     prompt_2: str = None,
     neg_prompt: str = "",
     neg_prompt_2: str = None,
+    num_images_per_prompt: int = 1,
 ):
     """
     This function can process long prompt with weights, no length limitation
@@ -260,6 +261,7 @@ def get_weighted_text_embeddings_sdxl(
         prompt_2 (str)
         neg_prompt (str)
         neg_prompt_2 (str)
+        num_images_per_prompt (int)
     Returns:
         prompt_embeds (torch.Tensor)
         neg_prompt_embeds (torch.Tensor)
@@ -383,6 +385,22 @@ def get_weighted_text_embeddings_sdxl(
     prompt_embeds = torch.cat(embeds, dim=1)
     negative_prompt_embeds = torch.cat(neg_embeds, dim=1)
 
+    bs_embed, seq_len, _ = prompt_embeds.shape
+    # duplicate text embeddings for each generation per prompt, using mps friendly method
+    prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
+    prompt_embeds = prompt_embeds.view(bs_embed * num_images_per_prompt, seq_len, -1)
+
+    seq_len = negative_prompt_embeds.shape[1]
+    negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+    negative_prompt_embeds = negative_prompt_embeds.view(bs_embed * num_images_per_prompt, seq_len, -1)
+
+    pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1).view(
+        bs_embed * num_images_per_prompt, -1
+    )
+    negative_pooled_prompt_embeds = negative_pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1).view(
+        bs_embed * num_images_per_prompt, -1
+    )
+
     return prompt_embeds, negative_prompt_embeds, pooled_prompt_embeds, negative_pooled_prompt_embeds
 
 
@@ -1096,7 +1114,9 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
             negative_prompt_embeds,
             pooled_prompt_embeds,
             negative_pooled_prompt_embeds,
-        ) = get_weighted_text_embeddings_sdxl(pipe=self, prompt=prompt, neg_prompt=negative_prompt)
+        ) = get_weighted_text_embeddings_sdxl(
+            pipe=self, prompt=prompt, neg_prompt=negative_prompt, num_images_per_prompt=num_images_per_prompt
+        )
 
         # 4. Prepare timesteps
         self.scheduler.set_timesteps(num_inference_steps, device=device)