Update app.py

app.py

@@ -49,18 +49,17 @@ _state = {"chat_engine": None}
 def get_chat_engine():
     if _state["chat_engine"] is None:
         llm = HuggingFaceLLM(
-            tokenizer_name="meta-llama/Llama-2-7b-chat-hf",
-            model_name="meta-llama/Llama-2-7b-chat-hf",
-            context_window=3900,
+            tokenizer_name="TinyLlama/TinyLlama-1.1B-Chat-v1.0",
+            model_name="TinyLlama/TinyLlama-1.1B-Chat-v1.0",
+            context_window=2048,
             max_new_tokens=256,
             generate_kwargs={"temperature": 0.2, "do_sample": True},
             device_map="auto",
             model_kwargs={
                 "use_auth_token": os.environ["HF_TOKEN"]
-            },
+                        },
             system_prompt=SYSTEM_PROMPT,
-            query_wrapper_prompt=WRAPPER_PROMPT,
-        )
+            query_wrapper_prompt=WRAPPER_PROMPT,)
         Settings.llm = llm
 
         memory = ChatMemoryBuffer.from_defaults(token_limit=2000)