Update app.py

app.py

@@ -1,165 +1,151 @@
 api_key = "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"
 
-import streamlit as st
-from langchain_groq import ChatGroq
-from langchain_community.utilities import ArxivAPIWrapper, WikipediaAPIWrapper
-from langchain_community.tools import ArxivQueryRun, WikipediaQueryRun, DuckDuckGoSearchRun
-from langchain.agents import initialize_agent, AgentType
 import os
+import gradio as gr
 import requests
-import pandas as pd
+from huggingface_hub import InferenceClient, login
 from dotenv import load_dotenv
+import pandas as pd
 
 # Load environment variables
 load_dotenv()
 
-# Constants for Basic Agent Evaluation
+# Constants
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
+MODEL_NAME = "meta-llama/llama-4-maverick-17b-128e-instruct"
+
+# Initialize the Llama Maverick client
+class MaverickAgent:
+    def __init__(self):
+        try:
+            self.client = InferenceClient(
+                model=MODEL_NAME,
+                token=os.getenv("HUGGINGFACE_TOKEN")
+            )
+            print("MaverickAgent initialized successfully")
+        except Exception as e:
+            print(f"Error initializing MaverickAgent: {e}")
+            raise
+
+    def __call__(self, question: str) -> str:
+        try:
+            print(f"Processing question: {question[:100]}...")
+            
+            # Custom prompt template for the Maverick model
+            prompt = f"""<|begin_of_text|>
+            <|start_header_id|>system<|end_header_id|>
+            You are an AI assistant that provides accurate and concise answers to questions.
+            Be factual and respond with just the answer unless asked to elaborate.
+            <|eot_id|>
+            <|start_header_id|>user<|end_header_id|>
+            {question}
+            <|eot_id|>
+            <|start_header_id|>assistant<|end_header_id|>"""
+            
+            response = self.client.text_generation(
+                prompt,
+                max_new_tokens=256,
+                temperature=0.7,
+                do_sample=True,
+            )
+            
+            # Clean up the response
+            answer = response.split("<|eot_id|>")[0].strip()
+            print(f"Generated answer: {answer[:200]}...")
+            return answer
+        except Exception as e:
+            print(f"Error processing question: {e}")
+            return f"Error: {str(e)}"
+
+# Authentication
+try:
+    login(token=os.getenv("HUGGINGFACE_TOKEN"))
+except Exception as e:
+    print(f"Authentication error: {e}")
+
+def run_and_submit_all(profile: gr.OAuthProfile | None):
+    if not profile:
+        return "Please log in with Hugging Face first.", None
+
+    # Initialize agent
+    try:
+        agent = MaverickAgent()
+    except Exception as e:
+        return f"Agent initialization failed: {e}", None
 
-# Initialize search tools (with warm-up)
-@st.cache_resource
-def load_tools():
-    with st.spinner("Initializing tools (first time may take a few seconds)..."):
-        api_wrapper_arxiv = ArxivAPIWrapper(top_k_results=1, doc_content_chars_max=250)
-        arxiv = ArxivQueryRun(api_wrapper=api_wrapper_arxiv)
-        api_wrapper_wiki = WikipediaAPIWrapper(top_k_results=1, doc_content_chars_max=250)
-        wiki = WikipediaQueryRun(api_wrapper=api_wrapper_wiki)
-        search = DuckDuckGoSearchRun(name="Search")
-        # Warm up tools
-        arxiv.run("machine learning")
-        wiki.run("machine learning")
-        return [search, arxiv, wiki]
-
-tools = load_tools()
-
-# Streamlit app layout
-st.title("Langchain - Chat with Search & Evaluation")
-
-# Sidebar for settings
-st.sidebar.title("Settings")
-api_key = st.sidebar.text_input("Enter your Groq API Key:", type="password")
-
-# Initialize chat messages
-if "messages" not in st.session_state:
-    st.session_state["messages"] = [
-        {"role": "assistant", "content": "Hi, I am a Chatbot who can search the web and evaluate questions. How can I help you?"}
-    ]
-
-# Display chat messages
-for msg in st.session_state.messages:
-    st.chat_message(msg["role"]).write(msg["content"])
-
-# Chat input
-if prompt := st.chat_input(placeholder="What is machine learning?"):
-    st.session_state.messages.append({"role": "user", "content": prompt})
-    st.chat_message("user").write(prompt)
-
-    if not api_key:
-        st.error("Please enter your Groq API key in the sidebar.")
-        st.stop()
-
-    llm = ChatGroq(groq_api_key=api_key, model_name="llama3-70b-8192")
-    search_agent = initialize_agent(tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, handle_parsing_errors=True)
-
-    with st.chat_message("assistant"):
-        response = search_agent.run(st.session_state.messages)
-        st.session_state.messages.append({'role': 'assistant', "content": response})
-        st.write(response)
-
-# Basic Agent Evaluation Section
-st.sidebar.title("Basic Agent Evaluation")
-
-def run_evaluation():
-    """Function to run the evaluation with progress updates"""
-    if not api_key:
-        st.error("Please enter your Groq API key in the sidebar.")
-        return "API key required", pd.DataFrame()
-
-    # Setup progress tracking
-    progress_bar = st.sidebar.progress(0)
-    status_text = st.sidebar.empty()
-    results_container = st.empty()
-    
-    username = "streamlit_user"
-    api_url = DEFAULT_API_URL
-    questions_url = f"{api_url}/questions"
-    submit_url = f"{api_url}/submit"
-    space_id = os.getenv("SPACE_ID", "local")
-    agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main" if space_id != "local" else "local_execution"
-
+    # Fetch questions
     try:
-        # 1. Fetch Questions
-        status_text.text("📡 Fetching questions...")
-        response = requests.get(questions_url, timeout=15)
-        response.raise_for_status()
+        response = requests.get(f"{DEFAULT_API_URL}/questions", timeout=15)
         questions_data = response.json()
-        total_questions = len(questions_data)
-        status_text.text(f"✅ Found {total_questions} questions")
-
         if not questions_data:
-            return "No questions found", pd.DataFrame()
-
-        # 2. Initialize Agent (reuse tools from cache)
-        llm = ChatGroq(groq_api_key=api_key, model_name="llama3-70b-8192")
-        agent = initialize_agent(tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, handle_parsing_errors=True)
-
-        # 3. Process Questions
-        results_log = []
-        answers_payload = []
-        
-        for i, item in enumerate(questions_data):
-            progress = (i + 1) / total_questions
-            progress_bar.progress(progress)
-            status_text.text(f"🔍 Processing question {i+1}/{total_questions}...")
+            return "No questions available.", None
+    except Exception as e:
+        return f"Failed to fetch questions: {e}", None
+
+    # Process questions
+    results = []
+    answers = []
+    for i, item in enumerate(questions_data):
+        task_id = item.get("task_id")
+        question = item.get("question")
+        if not task_id or not question:
+            continue
             
-            task_id = item.get("task_id")
-            question_text = item.get("question")
-            if not task_id or not question_text:
-                continue
-
-            try:
-                submitted_answer = agent.run(question_text)
-                answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
-                results_log.append({"Task ID": task_id, "Question": question_text[:100] + "..." if len(question_text) > 100 else question_text, 
-                                 "Submitted Answer": submitted_answer[:200] + "..." if len(submitted_answer) > 200 else submitted_answer})
-                
-                # Update results table progressively
-                if (i + 1) % 3 == 0 or (i + 1) == total_questions:  # Update every 3 questions or at end
-                    results_container.dataframe(pd.DataFrame(results_log))
-            except Exception as e:
-                results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"❌ Error: {str(e)}"})
-
-        # 4. Submit Answers
-        status_text.text("📤 Submitting answers...")
-        submission_data = {"username": username, "agent_code": agent_code, "answers": answers_payload}
-        response = requests.post(submit_url, json=submission_data, timeout=60)
-        response.raise_for_status()
-        result_data = response.json()
+        try:
+            answer = agent(question)
+            answers.append({"task_id": task_id, "submitted_answer": answer})
+            results.append({
+                "Task ID": task_id,
+                "Question": question[:100] + "..." if len(question) > 100 else question,
+                "Answer": answer[:100] + "..." if len(answer) > 100 else answer
+            })
+        except Exception as e:
+            results.append({
+                "Task ID": task_id,
+                "Question": question,
+                "Answer": f"Error: {str(e)}"
+            })
+
+    # Submit answers
+    try:
+        submission = {
+            "username": profile.username,
+            "agent_code": f"https://huggingface.co/spaces/{os.getenv('SPACE_ID')}",
+            "answers": answers
+        }
+        response = requests.post(f"{DEFAULT_API_URL}/submit", json=submission, timeout=60)
+        result = response.json()
         
-        final_status = (
-            f"✅ Submission Successful!\n"
-            f"📊 Score: {result_data.get('score', 'N/A')}%\n"
-            f"📝 Correct: {result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')}\n"
-            f"💬 Message: {result_data.get('message', 'No message')}"
+        return (
+            f"✅ Submitted {len(answers)} answers\n"
+            f"📊 Score: {result.get('score', 'N/A')}%\n"
+            f"🔢 Correct: {result.get('correct_count', 0)}/{len(answers)}\n"
+            f"🤖 Model: {MODEL_NAME}",
+            pd.DataFrame(results)
         )
-        return final_status, pd.DataFrame(results_log)
-
     except Exception as e:
-        return f"❌ Failed: {str(e)}", pd.DataFrame(results_log if 'results_log' in locals() else [])
-    
-    finally:
-        progress_bar.empty()
-        status_text.empty()
-
-# Evaluation button in sidebar
-if st.sidebar.button("🚀 Run Evaluation & Submit Answers"):
-    with st.spinner("Starting evaluation..."):
-        status, results = run_evaluation()
+        return f"Submission failed: {e}", pd.DataFrame(results)
+
+# Gradio Interface
+with gr.Blocks(theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# 🦙 Llama 4 Maverick Agent")
+    gr.Markdown(f"Using `{MODEL_NAME}` for evaluation")
     
-    st.sidebar.success("Evaluation completed!")
-    st.sidebar.text_area("Results", value=status, height=150)
+    gr.LoginButton()
     
-    if not results.empty:
-        st.subheader("📋 Detailed Results")
-        st.dataframe(results, use_container_width=True)
+    with gr.Row():
+        run_btn = gr.Button("Run Evaluation", variant="primary")
+        
+    with gr.Row():
+        status = gr.Textbox(label="Status", interactive=False)
+        results = gr.DataFrame(label="Results", wrap=True)
+
+    run_btn.click(
+        fn=run_and_submit_all,
+        outputs=[status, results]
+    )
+
+if __name__ == "__main__":
+    demo.launch()
+