fix use flash attn

main.py

@@ -75,7 +75,7 @@ transcribe_pipeline = pipeline(
     tokenizer=processor.tokenizer,
     feature_extractor=processor.feature_extractor,
     torch_dtype=torch_dtype,
-    device=device,
+    #device=device,
 )
 if device == "cuda":
     transcribe_pipeline.model = torch.compile(transcribe_pipeline.model, mode="max-autotune")