add oom notice

app.py

@@ -111,6 +111,7 @@ no_change_btn = gr.Button()
 enable_btn = gr.Button(interactive=True)
 disable_btn = gr.Button(interactive=False)
 server_error_msg = "**NETWORK ERROR DUE TO HIGH TRAFFIC. PLEASE REGENERATE OR REFRESH THIS PAGE.**"
+server_oom_msg = "**OUT OF GPU MEMORY DETECTED. PLEASE DECREASE THE MAX OUTPUT TOKENS AND REGENERATE.**"
 
 def load_demo_refresh_model_list():
     logging.info(f"load_demo.")
@@ -308,11 +309,7 @@ def http_bot(state, temperature, top_p, max_new_tokens, speaker):
         
             id = str(uuid.uuid4())
             os.makedirs("./demo_audio", exist_ok=True)    
-            try:
-                synthesis(content_unit, style_centroid_embedding, hps, net_g, f"./demo_audio/{id}_temp_audio.wav")
-            except:
-                torch.cuda.empty_cache()
-                synthesis(content_unit, style_centroid_embedding, hps, net_g, f"./demo_audio/{id}_temp_audio.wav")
+            synthesis(content_unit, style_centroid_embedding, hps, net_g, f"./demo_audio/{id}_temp_audio.wav")
             temp_file = f"./demo_audio/{id}_temp_audio.wav"
         except Exception as e:
             os.system("nvidia-smi")
@@ -324,7 +321,7 @@ def http_bot(state, temperature, top_p, max_new_tokens, speaker):
             state.messages[-1][-1] = (output, temp_file)
             yield (state, state.to_gradio_chatbot_public()) + (enable_btn,) * 2
         else:
-            state.messages[-1][-1] = server_error_msg
+            state.messages[-1][-1] = server_oom_msg
             yield (state, state.to_gradio_chatbot_public()) + (enable_btn,) * 2
     else:
         yield (state, state.to_gradio_chatbot_public()) + (enable_btn,) * 2