Update pipeline_stable_diffusion_3_ipa.py

pipeline_stable_diffusion_3_ipa.py

@@ -1218,8 +1218,9 @@ class StableDiffusion3Pipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSingle
 
         bs_embed, seq_len, _ = clip_image_embeds.shape
         clip_image_embeds = clip_image_embeds.repeat(1, 1, 1)
-        clip_image_embeds = clip_image_embeds.view(2, -1)
-        
+        clip_image_embeds = clip_image_embeds.unsqueeze(0) # Add a dimension at the beginning so now you have [1, 2*seq_len_img, embed_dim_img]
+        clip_image_embeds = clip_image_embeds.view(2, -1, bs_embed) # Reshape to separate the zero embedding and the averaged embedding
+            
         #clip_image_embeds = torch.cat([torch.stack(image_prompt_embeds_list)], dim=0).mean(dim=0)
         
         # 4. Prepare timesteps