Update model.py

model.py

@@ -379,8 +379,11 @@ class StreamMultiDiffusion(nn.Module):
             A single string of text prompt.
         """
         question = 'Question: What are in the image? Answer:'
+        print(self.i2t_processor.device, self.i2t_model.device)
         inputs = self.i2t_processor(image, question, return_tensors='pt')
+        print(inputs)
         out = self.i2t_model.generate(**inputs, max_new_tokens=77)
+        print(out[0].device)
         prompt = self.i2t_processor.decode(out[0], skip_special_tokens=True).strip()
         return prompt
 
@@ -450,9 +453,7 @@ class StreamMultiDiffusion(nn.Module):
         flag_changed = False
         if image is not None:
             image_ = image.resize((self.width, self.height))
-            print('111111111111111111111111', self.device, self.pipe.device)
             prompt = self.get_text_prompts(image_) if prompt is None else prompt
-            print('2222222222222222222222', self.device, self.pipe.device)
             negative_prompt = '' if negative_prompt is None else negative_prompt
             embed = self.pipe.encode_prompt(
                 prompt=[prompt],