Use float16 for gpu

app.py

@@ -37,6 +37,7 @@ pipe_dict = {
         "automatic-speech-recognition",
         model=v,
         device="cuda" if torch.cuda.is_available() else "cpu",
+        torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
     )
     for k, v in model_dict.items()
 }