Update app.py

app.py

@@ -13,7 +13,7 @@ from utils import PROMPT, INPUT_PROMPT, extract_after_final_answer
 
 # Initialize the HF Inference Client
 # It will automatically pick up the HF_TOKEN environment variable if it's set
-client = InferenceClient()
+client = InferenceClient(timeout=60)
 
 # Define the models we want to use from the Hugging Face Hub
 LLM_MODEL = "Qwen/Qwen2.5-7B-Instruct"
@@ -83,7 +83,7 @@ def start_game():
         initial_event                  # sets state_cur
     )
 
-def play_turn(human_response, loc, player, cur, custom_token, request: gr.Request):
+async def play_turn(human_response, loc, player, cur, custom_token, request: gr.Request):
     """Executes a single turn of the game loop based on user input."""
     status = player.get_attribute()
     if not human_response.strip():
@@ -115,7 +115,7 @@ def play_turn(human_response, loc, player, cur, custom_token, request: gr.Reques
         }
     ]
     try:
-        response = client.chat_completion(
+        response = await client.chat_completion(
             model=LLM_MODEL,
             messages=messages,
             max_tokens=500,
@@ -138,7 +138,7 @@ def play_turn(human_response, loc, player, cur, custom_token, request: gr.Reques
     attribute = player.get_attribute()
 
     # 3. Check for Game Over conditions
-    if attribute['hp'] <= 0 or attribute['mp'] <= 0:
+    if attribute['hp'] <= 0:
         loc = -1
 
     cur = EVENTS[loc] 
@@ -253,4 +253,4 @@ with gr.Blocks() as demo:
     user_input.submit(**turn_config)
 
 if __name__ == "__main__":
-    demo.queue().launch(share=True)
+    demo.queue(max_size=10).launch()