Update utils.py

utils.py

@@ -28,7 +28,7 @@ def prepare_inputs(text_input=None, image_input=None, audio_input=None):
     image_processed = Config.processor(images=image_input, return_tensors="pt")
 
     with torch.no_grad():
-      outputs = clip_model(**image_processed)
+      outputs = clip_model(**image_processed.to(Config.device))
       last_hidden_state = outputs.last_hidden_state[:, 1:, :]
       image_embed = projection(last_hidden_state.to(Config.device)).to(torch.float16)