Update app.py

app.py

@@ -9,20 +9,17 @@ 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 faster and more available on the free tier
 llm = HuggingFaceInferenceAPI(
-    # Switching to 7B as it is more stable on the free API
-    model_name="Qwen/Qwen2.5-Coder-7B-Instruct", 
+    model_name="Qwen/Qwen2.5-7B-Instruct", 
     token=hf_token,
-    # Fix for the 404 error: explicitly use text-generation task
-    task="text-generation", 
-    # Use together or auto to ensure the model is found
-    provider="together",
+    task="text-generation",
     is_function_calling_model=False 
 )
 
-# 2. DEFINE TOOLS
+# 2. DEFINE YOUR TOOLS
 def get_tokyo_time() -> str:
-    """Useful for when you need to know the current time in Tokyo, Japan."""
+    """Returns the current time in Tokyo, Japan."""
     tz = pytz.timezone('Asia/Tokyo')
     return f"The current time in Tokyo is {datetime.datetime.now(tz).strftime('%H:%M:%S')}"
 
@@ -36,7 +33,7 @@ tools = [
 ]
 
 # 3. CREATE THE AGENT
-# Ensure you are importing ReActAgent from llama_index.core.agent
+# The system prompt ensures the AI follows the ReAct pattern
 agent = ReActAgent.from_tools(
     tools, 
     llm=llm, 
@@ -46,17 +43,11 @@ agent = ReActAgent.from_tools(
 # 4. GRADIO INTERFACE
 def chat(message, history):
     try:
+        # Use .chat() to maintain memory
         response = agent.chat(message)
         return str(response)
     except Exception as e:
-        # Better error handling for the UI
-        return f"Agent Error: {str(e)}"
+        # This will show you exactly what is failing in the Gradio UI
+        return f"Error: {str(e)}"
 
-demo = gr.ChatInterface(
-    fn=chat, 
-    title="Unit 2: LlamaIndex Agent",
-    description="I can tell you the time in Tokyo or multiply numbers!"
-)
-
-if __name__ == "__main__":
-    demo.launch()
+gr.ChatInterface(chat, title="Unit 2: LlamaIndex Agent").launch()