Update app.py

app.py

@@ -242,27 +242,27 @@ def inference(image, audio, video, task_type, instruction):
     if task_type == 'Image Captioning':
         text = ['']  
         text_input = tokenizer(text, padding='longest', return_tensors="pt").to(device) 
-        model = model_caption.clone()
+        model = model_caption
     elif task_type == 'Video Captioning':
         text = ['']  
         text_input = tokenizer(text, padding='longest', return_tensors="pt").to(device) 
         model_caption = model_caption.load_state_dict(state_dict_video_caption,strict=False)  
-        model = model_caption.clone()
+        model = model_caption
     elif task_type == 'Audio Captioning':
         text = ['']  
         text_input = tokenizer(text, padding='longest', return_tensors="pt").to(device) 
         model_caption = model_caption.load_state_dict(state_dict_audio_caption,strict=False)  
-        model = model_caption.clone()
+        model = model_caption
     elif task_type == 'Visual Question Answering':
         question = instruction+'?'+special_answer_token
         text_input = tokenizer(question, padding='longest', return_tensors="pt").to(device) 
         model_caption = model_caption.load_state_dict(state_dict_vqa,strict=False)  
-        model = model_caption.clone()
+        model = model_caption
     elif task_type == 'Visual Question Answering':
         question = instruction+'?'+special_answer_token
         text_input = tokenizer(question, padding='longest', return_tensors="pt").to(device) 
         model_caption = model_caption.load_state_dict(state_dict_video_qa,strict=False)  
-        model = model_caption.clone()
+        model = model_caption
     else:
         raise NotImplemented