Update pipeline_stable_diffusion_3_ipa.py

pipeline_stable_diffusion_3_ipa.py

@@ -1211,7 +1211,7 @@ class StableDiffusion3Pipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSingle
         print('shape 1: ', stacked_image_embeds.shape)
         print('shape 1a: ', stacked_image_embedsg.shape)
         # 2. Calculate the mean of the stacked embeddings
-        average_image_embed = torch.mean(stacked_image_embeds, dim=1) #.unsqueeze(0)  # Add batch dimension after averaging
+        average_image_embed = torch.mean(stacked_image_embeds, dim=0) #.unsqueeze(0)  # Add batch dimension after averaging
         print('shape 2: ', average_image_embed.shape)
         average_image_embedf = torch.mean(stacked_image_embeds, dim=1).unsqueeze(0)  # Add batch dimension after averaging
         print('shape 2a: ', average_image_embedf.shape)
@@ -1220,7 +1220,9 @@ class StableDiffusion3Pipeline(DiffusionPipeline, SD3LoraLoaderMixin, FromSingle
         print('shape 3: ', zeros_tensor.shape)
         # 4. Concatenate the zeros and the average embedding
         clip_image_embeds = torch.cat([zeros_tensor, average_image_embed], dim=0)
+        clip_image_embeds2 = torch.cat([zeros_tensor, average_image_embedf], dim=0)
         print('shape 4: ', clip_image_embeds.shape)
+        print('shape 4a: ', clip_image_embeds2.shape)
         '''
         #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]
         print('shape 5: ', clip_image_embeds.shape)