bind cuda

app.py

@@ -19,15 +19,15 @@ CHUNK_LENGTH_S = 15
 FILE_LIMIT_MB = 1000
 YT_LENGTH_LIMIT_S = 3600  # limit to 1 hour YouTube files
 # device setting
-if torch.cuda.is_available():
-    torch_dtype = torch.bfloat16
-    device = "cuda:0"
-    model_kwargs = {'attn_implementation': 'sdpa'}
-else:
-    torch_dtype = torch.float32
-    device = "cpu"
-    model_kwargs = {}
-
+# if torch.cuda.is_available():
+#     torch_dtype = torch.bfloat16
+#     device = "cuda:0"
+#     model_kwargs = {'attn_implementation': 'sdpa'}
+# else:
+#     torch_dtype = torch.float32
+#     device = "cpu"
+#     model_kwargs = {}
+device = "cuda"
 # define the pipeline
 pipe = pipeline(
     model=MODEL_NAME,
@@ -167,4 +167,4 @@ yt_transcribe = gr.Interface(
 with demo:
     gr.TabbedInterface([mf_transcribe, file_transcribe, yt_transcribe], ["Microphone", "Audio file", "YouTube"])
 
-demo.launch()
+demo.queue(api_open=False, default_concurrency_limit=40).launch(show_api=False, show_error=True)