moved the spaces decor on bot_comms

app.py

@@ -80,8 +80,6 @@ def gpt_generation(input: str,
     return stream
 
 # Place just input pass and return generation output
-@spaces.GPU(duration=120)
-@contextlib.contextmanager
 def llama_generation(input_text: str,
                      history: list,
                      temperature: float,
@@ -113,18 +111,18 @@ def llama_generation(input_text: str,
     if temperature == 0:
         generate_kwargs["do_sample"] = False
 
-    # Use a lock object to synchronize access to the llama_model
-    lock = threading.Lock()
+    # # Use a lock object to synchronize access to the llama_model
+    # lock = threading.Lock()
 
-    def generate_llama():
-        with lock:
-            # Generate the response using the llama_model
-            response = llama_model.generate(**generate_kwargs)
-            return response
+    # def generate_llama():
+    #     with lock:
+    #         # Generate the response using the llama_model
+    #         response = llama_model.generate(**generate_kwargs)
+    #         return response
 
 
     # start the thread and wait for it to finish
-    thread = threading.Thread(target=generate_llama)
+    thread = threading.Thread(target=llama_model.generate, kwargs=generate_kwargs)
     thread.start()
     thread.join()
     return streamer
@@ -138,6 +136,7 @@ def check_cuda():
 first_time = True
 llm_mode = ""
 
+@spaces.GPU(decoration=120)
 def bot_comms(input_text: str,
               history: list,
               temperature: float,