Update app.py

app.py

@@ -9,14 +9,23 @@ from llama_index.llms.huggingface_api import HuggingFaceInferenceAPI
 # 1. SETUP LLM
 hf_token = os.getenv("HF_TOKEN")
 
-# We use the 7B model because it's much faster and reliable for the free tier
+# Create the LLM with the 404 fix
 llm = HuggingFaceInferenceAPI(
     model_name="Qwen/Qwen2.5-7B-Instruct", 
     token=hf_token,
     task="text-generation",
+    provider="together",
     is_function_calling_model=False 
 )
 
+# Create the Agent with a clear instruction (Context)
+agent = ReActAgent.from_tools(
+    tools, 
+    llm=llm, 
+    verbose=True,
+    context="You are a helpful assistant. Always use Thought/Action/Observation format."
+)
+
 # 2. DEFINE YOUR TOOLS
 def get_tokyo_time() -> str:
     """Returns the current time in Tokyo, Japan."""