Updating app.py removing llama.cpp

app.py

@@ -1,10 +1,6 @@
 import streamlit as st
 import pandas as pd
-import numpy as np
-from llama_cpp import Llama
-from huggingface_hub import hf_hub_download
-import io
-import os
+from huggingface_hub import InferenceClient
 
 # Page configuration
 st.set_page_config(
@@ -14,58 +10,25 @@ st.set_page_config(
     initial_sidebar_state="expanded"
 )
 
-# Custom CSS for better UI
-st.markdown("""
-    <style>
-    .main {
-        padding: 0rem 1rem;
-    }
-    .stTabs [data-baseweb="tab-list"] {
-        gap: 2px;
-    }
-    </style>
-    """, unsafe_allow_html=True)
-
-# Title and description
 st.title("📊 LLM Data Analyzer")
-st.markdown("""
-    Analyze your CSV/Excel files and chat with an AI assistant powered by Llama 2.
-    This app runs on **free Hugging Face CPU** - response time ~5-10 seconds per query.
-    """)
+st.write("*Analyze data and chat with AI powered by Hugging Face Inference API*")
 
-# Cache model loading to avoid reloading
+# Initialize HF Inference Client
 @st.cache_resource
-def load_llm_model():
-    """Load Llama 2 model from Hugging Face Hub"""
-    st.info("📥 Downloading model (first time only, ~4GB)... This may take 2-3 minutes.")
-    
+def get_hf_client():
+    """Get Hugging Face Inference Client"""
     try:
-        model_path = hf_hub_download(
-            repo_id="TheBloke/Llama-2-7B-Chat-GGUF",
-            filename="llama-2-7b-chat.Q4_K_M.gguf"
-        )
-        
-        llm = Llama(
-            model_path=model_path,
-            n_ctx=2048,
-            n_threads=4,
-            n_gpu_layers=0,  # CPU only (free tier)
-            verbose=False
-        )
-        return llm
+        return InferenceClient()
     except Exception as e:
-        st.error(f"Error loading model: {e}")
+        st.error(f"Error initializing HF client: {e}")
         return None
 
-# Load model
-llm = load_llm_model()
+client = get_hf_client()
 
-if llm is None:
-    st.error("Failed to load model. Please refresh the page.")
+if client is None:
+    st.error("Failed to initialize Hugging Face client")
     st.stop()
 
-st.success("✅ Model loaded successfully!")
-
 # Create tabs
 tab1, tab2, tab3 = st.tabs(["📤 Upload & Analyze", "💬 Chat", "📊 About"])
 
@@ -84,7 +47,6 @@ with tab1:
     if uploaded_file is not None:
         st.success(f"✅ File uploaded: {uploaded_file.name}")
         
-        # Read the file
         try:
             if uploaded_file.name.endswith('.csv'):
                 df = pd.read_csv(uploaded_file)
@@ -118,9 +80,9 @@ with tab1:
             
             if question:
                 with st.spinner("🤔 AI is analyzing your data..."):
-                    # Create prompt
-                    data_summary = df.describe().to_string()
-                    prompt = f"""You are a data analyst expert. You have the following data summary:
+                    try:
+                        data_summary = df.describe().to_string()
+                        prompt = f"""You are a data analyst expert. You have the following data summary:
 
 {data_summary}
 
@@ -128,20 +90,19 @@ Column names: {', '.join(df.columns.tolist())}
 
 User's question: {question}
 
-Please provide a clear, concise analysis based on the data summary. Focus on actionable insights."""
-                    
-                    # Generate response
-                    response = llm(
-                        prompt,
-                        max_tokens=300,
-                        stop=["\n\nUser:", "Question:"],
-                        echo=False,
-                        temperature=0.7
-                    )
-                    
-                    answer = response['choices'][0]['text'].strip()
-                    st.success("✅ Analysis Complete")
-                    st.write(answer)
+Please provide a clear, concise analysis based on the data summary."""
+                        
+                        # Use Hugging Face Inference API
+                        response = client.text_generation(
+                            prompt,
+                            max_new_tokens=300,
+                            temperature=0.7,
+                        )
+                        
+                        st.success("✅ Analysis Complete")
+                        st.write(response)
+                    except Exception as e:
+                        st.error(f"Error analyzing data: {e}")
         
         except Exception as e:
             st.error(f"Error reading file: {e}")
@@ -151,7 +112,7 @@ Please provide a clear, concise analysis based on the data summary. Focus on act
 # ============================================================================
 with tab2:
     st.header("💬 Chat with AI Assistant")
-    st.write("Have a conversation with Llama 2. Ask anything!")
+    st.write("Have a conversation with an AI assistant powered by Hugging Face.")
     
     # Initialize session state for chat history
     if "messages" not in st.session_state:
@@ -176,26 +137,25 @@ with tab2:
         # Generate AI response
         with st.chat_message("assistant"):
             with st.spinner("⏳ Generating response..."):
-                prompt = f"""You are a helpful AI assistant. The user asks: {user_input}
-
-Provide a clear, helpful, and concise response."""
-                
-                response = llm(
-                    prompt,
-                    max_tokens=300,
-                    stop=["\n\nUser:", "User:"],
-                    echo=False,
-                    temperature=0.7
-                )
-                
-                assistant_message = response['choices'][0]['text'].strip()
-                st.markdown(assistant_message)
-                
-                # Add assistant message to history
-                st.session_state.messages.append({
-                    "role": "assistant",
-                    "content": assistant_message
-                })
+                try:
+                    prompt = f"User: {user_input}\n\nAssistant:"
+                    
+                    response = client.text_generation(
+                        prompt,
+                        max_new_tokens=300,
+                        temperature=0.7,
+                    )
+                    
+                    assistant_message = response.strip()
+                    st.markdown(assistant_message)
+                    
+                    # Add assistant message to history
+                    st.session_state.messages.append({
+                        "role": "assistant",
+                        "content": assistant_message
+                    })
+                except Exception as e:
+                    st.error(f"Error generating response: {e}")
 
 # ============================================================================
 # TAB 3: About
@@ -210,42 +170,39 @@ with tab3:
     
     ### 🔧 Technology Stack
     
-    - **Model:** Llama 2 7B (quantized to 4-bit)
-    - **Framework:** Llama.cpp (CPU inference)
-    - **Frontend:** Streamlit
+    - **Framework:** Streamlit
+    - **AI Engine:** Hugging Face Inference API
     - **Hosting:** Hugging Face Spaces (Free Tier)
+    - **Language:** Python
     
-    ### ⚡ Performance
+    ### ⚡ Features
     
-    - **Speed:** ~5-10 tokens per second (free CPU)
-    - **Context:** 2048 tokens max
-    - **Model Size:** 4GB (quantized)
-    - **Hardware:** Free tier CPU
+    1. **Data Analysis**: Upload CSV/Excel and ask questions about your data
+    2. **Chat**: Have conversations with an AI assistant
+    3. **Statistics**: View data summaries and insights
     
-    ### 💡 Use Cases
+    ### 📝 How to Use
     
-    1. **Data Analysis**: Upload CSV/Excel and ask questions
-    2. **Chat**: General conversation with AI
-    3. **Learning**: Understand your data better
+    1. **Upload Data** - Start by uploading a CSV or Excel file
+    2. **Preview** - Review your data and statistics
+    3. **Ask Questions** - Get AI-powered analysis
+    4. **Chat** - Have follow-up conversations
     
-    ### 🚀 Faster Version Available
+    ### 🌐 Powered By
     
-    For **GPU acceleration** (70+ tokens/sec):
-    - Run locally on Apple Silicon Mac using MLX
-    - Upgrade to Hugging Face PRO tier
-    - Deploy on GPU-enabled cloud servers
+    - [Hugging Face](https://huggingface.co/) - AI models and hosting
+    - [Streamlit](https://streamlit.io/) - Web framework
     
-    ### 📝 Tips
+    ### 📖 Quick Tips
     
     - Keep questions focused and specific for best results
-    - First request takes longer (model loading)
+    - Responses may take a few seconds
     - Data is processed locally, not stored on server
     
     ### 🔗 Links
     
-    - [GitHub Repository](#) - Source code
-    - [Hugging Face Hub](#) - Model info
-    - [Llama.cpp](#) - Inference engine
+    - [GitHub Repository](https://github.com/Arif-Badhon/LLM-Data-Analyzer)
+    - [Hugging Face Hub](https://huggingface.co/)
     
     ---