app.py

import streamlit as st
from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
from huggingface_hub import login
from threading import Thread
import PyPDF2
import pandas as pd
import torch
import time

# Check if 'peft' is installed
try:
    from peft import PeftModel, PeftConfig
except ImportError:
    raise ImportError(
        "The 'peft' library is required but not installed. "
        "Please install it using: `pip install peft`"
    )

# Set page configuration
st.set_page_config(
    page_title="WizNerd Insp",
    page_icon="🚀",
    layout="centered"
)

# Hardcoded Hugging Face token (replace with your actual token)
HF_TOKEN = "your_hugging_face_token_here"

# Model names
BASE_MODEL_NAME = "google-bert/bert-base-uncased"
MODEL_OPTIONS = {
    "Full Fine-Tuned": "amiguel/instruct_BERT-base-uncased_model",
    "LoRA Adapter": "amiguel/SmolLM2-360M-concise-reasoning-lora",
    "QLoRA Adapter": "amiguel/SmolLM2-360M-concise-reasoning-qlora"  # Hypothetical, adjust if needed
}

# Title with rocket emojis
st.title("🚀 WizNerd Insp 🚀")

# Configure Avatars
USER_AVATAR = "https://raw.githubusercontent.com/achilela/vila_fofoka_analysis/9904d9a0d445ab0488cf7395cb863cce7621d897/USER_AVATAR.png"
BOT_AVATAR = "https://raw.githubusercontent.com/achilela/vila_fofoka_analysis/991f4c6e4e1dc7a8e24876ca5aae5228bcdb4dba/Ataliba_Avatar.jpg"

# Sidebar configuration
with st.sidebar:
    st.header("Model Selection 🤖")
    model_type = st.selectbox("Choose Model Type", list(MODEL_OPTIONS.keys()), index=0)
    selected_model = MODEL_OPTIONS[model_type]
    
    st.header("Upload Documents 📂")
    uploaded_file = st.file_uploader(
        "Choose a PDF or XLSX file",
        type=["pdf", "xlsx"],
        label_visibility="collapsed"
    )

# Initialize chat history
if "messages" not in st.session_state:
    st.session_state.messages = []

# File processing function
@st.cache_data
def process_file(uploaded_file):
    if uploaded_file is None:
        return ""
    
    try:
        if uploaded_file.type == "application/pdf":
            pdf_reader = PyPDF2.PdfReader(uploaded_file)
            return "\n".join([page.extract_text() for page in pdf_reader.pages])
        elif uploaded_file.type == "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet":
            df = pd.read_excel(uploaded_file)
            return df.to_markdown()
    except Exception as e:
        st.error(f"📄 Error processing file: {str(e)}")
        return ""

# Model loading function
@st.cache_resource
def load_model(hf_token, model_type, selected_model):
    try:
        if not hf_token:
            st.error("🔐 Authentication required! Please provide a valid Hugging Face token.")
            return None
        
        login(token=hf_token)
        
        # Load tokenizer
        tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL_NAME, token=hf_token)
        
        # Determine device
        device = "cuda" if torch.cuda.is_available() else "cpu"
        
        # Load model based on type
        if model_type == "Full Fine-Tuned":
            # Load full fine-tuned model directly
            model = AutoModelForCausalLM.from_pretrained(
                selected_model,
                torch_dtype=torch.bfloat16,
                token=hf_token
            ).to(device)
        else:
            # Load base model and apply PEFT adapter
            base_model = AutoModelForCausalLM.from_pretrained(
                BASE_MODEL_NAME,
                torch_dtype=torch.bfloat16,
                token=hf_token
            ).to(device)
            model = PeftModel.from_pretrained(
                base_model,
                selected_model,
                torch_dtype=torch.bfloat16,
                is_trainable=False,  # Inference mode
                token=hf_token
            ).to(device)
        
        return model, tokenizer
        
    except Exception as e:
        st.error(f"🤖 Model loading failed: {str(e)}")
        return None

# Generation function with KV caching
def generate_with_kv_cache(prompt, file_context, model, tokenizer, use_cache=True):
    full_prompt = f"Analyze this context:\n{file_context}\n\nQuestion: {prompt}\nAnswer:"
    
    streamer = TextIteratorStreamer(
        tokenizer, 
        skip_prompt=True, 
        skip_special_tokens=True
    )
    
    inputs = tokenizer(full_prompt, return_tensors="pt").to(model.device)
    
    generation_kwargs = {
        "input_ids": inputs["input_ids"],
        "attention_mask": inputs["attention_mask"],
        "max_new_tokens": 1024,
        "temperature": 0.7,
        "top_p": 0.9,
        "repetition_penalty": 1.1,
        "do_sample": True,
        "use_cache": use_cache,
        "streamer": streamer
    }
    
    Thread(target=model.generate, kwargs=generation_kwargs).start()
    return streamer

# Display chat messages
for message in st.session_state.messages:
    try:
        avatar = USER_AVATAR if message["role"] == "user" else BOT_AVATAR
        with st.chat_message(message["role"], avatar=avatar):
            st.markdown(message["content"])
    except:
        with st.chat_message(message["role"]):
            st.markdown(message["content"])

# Chat input handling
if prompt := st.chat_input("Ask your inspection question..."):
    # Load model if not already loaded or if model type changed
    if "model" not in st.session_state or st.session_state.get("model_type") != model_type:
        model_data = load_model(HF_TOKEN, model_type, selected_model)
        if model_data is None:
            st.error("Failed to load model. Please check your token and try again.")
            st.stop()
            
        st.session_state.model, st.session_state.tokenizer = model_data
        st.session_state.model_type = model_type
    
    model = st.session_state.model
    tokenizer = st.session_state.tokenizer
    
    # Add user message
    with st.chat_message("user", avatar=USER_AVATAR):
        st.markdown(prompt)
    st.session_state.messages.append({"role": "user", "content": prompt})

    # Process file
    file_context = process_file(uploaded_file)
    
    # Generate response with KV caching
    if model and tokenizer:
        try:
            with st.chat_message("assistant", avatar=BOT_AVATAR):
                start_time = time.time()
                streamer = generate_with_kv_cache(prompt, file_context, model, tokenizer, use_cache=True)
                
                response_container = st.empty()
                full_response = ""
                
                for chunk in streamer:
                    cleaned_chunk = chunk.replace("<think>", "").replace("</think>", "").strip()
                    full_response += cleaned_chunk + " "
                    response_container.markdown(full_response + "▌", unsafe_allow_html=True)
                
                # Calculate performance metrics
                end_time = time.time()
                input_tokens = len(tokenizer(prompt)["input_ids"])
                output_tokens = len(tokenizer(full_response)["input_ids"])
                speed = output_tokens / (end_time - start_time)
                
                # Calculate costs (hypothetical pricing model)
                input_cost = (input_tokens / 1000000) * 5  # $5 per million input tokens
                output_cost = (output_tokens / 1000000) * 15  # $15 per million output tokens
                total_cost_usd = input_cost + output_cost
                total_cost_aoa = total_cost_usd * 1160  # Convert to AOA (Angolan Kwanza)
                
                # Display metrics
                st.caption(
                    f"🔑 Input Tokens: {input_tokens} | Output Tokens: {output_tokens} | "
                    f"🕒 Speed: {speed:.1f}t/s | 💰 Cost (USD): ${total_cost_usd:.4f} | "
                    f"💵 Cost (AOA): {total_cost_aoa:.4f}"
                )
                
                response_container.markdown(full_response)
                st.session_state.messages.append({"role": "assistant", "content": full_response})
                
        except Exception as e:
            st.error(f"⚡ Generation error: {str(e)}")
    else:
        st.error("🤖 Model not loaded!")