Update app.py

app.py

@@ -23,7 +23,7 @@ download_model()
 # Define OmniAI's chat function (handles two arguments: user message + history)
 def omni_ai_chat(user_message, history):
     try:
-        llm = Llama(model_path=MODEL_PATH, n_ctx=4096)  # Load locally downloaded model
+        llm = Llama(model_path=MODEL_PATH, n_ctx=4096, n_batch=256, flash_attn=True)  # Load locally downloaded model
         response = llm(user_message)  # Process only the latest prompt
         return response["choices"][0]["text"].strip()
     except Exception as e: