fix: 确保text_encoder_projection_dim在调用时正确初始化

utils.py

@@ -284,6 +284,9 @@ def call(
 
         # 7. Prepare added time ids & embeddings
         add_text_embeds = pooled_prompt_embeds
+        # 确保 text_encoder_projection_dim 被正确初始化
+        if not hasattr(self, 'text_encoder_projection_dim'):
+            self.text_encoder_projection_dim = 128  # 或者其他适当的默认值
         add_time_ids = self._get_add_time_ids(
             original_size, crops_coords_top_left, target_size, dtype=prompt_embeds.dtype
         )
@@ -306,9 +309,6 @@ def call(
         add_text_embeds = add_text_embeds.to(device)
         add_time_ids = add_time_ids.to(device).repeat(batch_size * num_images_per_prompt, 1)
 
-        # 确保 text_encoder_projection_dim 被正确初始化
-        if not hasattr(self, 'text_encoder_projection_dim'):
-            self.text_encoder_projection_dim = 128  # 或者其他适当的默认值
 
         # 8. Denoising loop
         num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)