Update app.py

app.py

@@ -44,6 +44,7 @@ yaml=YAML(typ='safe')
 
 use_cuda = torch.cuda.is_available()
 device = torch.device('cuda') if use_cuda else torch.device('cpu')
+device_type = 'cuda' if use_cuda else 'cpu'
 
 ## Load model
 
@@ -133,7 +134,7 @@ def inference(image, audio, video, task_type, instruction):
 
 
 
-    with torch.autocast(device_type='cuda', dtype=torch.float16, enabled=True):
+    with torch.autocast(device_type=device_type, enabled=True):
 
         out = model(image=image, text=text_input, mode='generate', return_dict=True, max_length=max_length, 
                     do_sample=do_sample, num_beams=num_beams)