app.py

import streamlit as st
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
from peft import PeftModel

st.set_page_config(page_title="Hugging Face Chatbot", layout="centered")
st.title("Hugging Face Chatbot with LoRA")

@st.cache_resource
def load_model():
    # Replace this with the actual base model used during LoRA fine-tuning
    base_model_name = "unsloth/Llama-3.2-1B-Instruct"
    
    # Load the base model and tokenizer
    tokenizer = AutoTokenizer.from_pretrained(base_model_name, use_fast=False)
    base_model = AutoModelForCausalLM.from_pretrained(base_model_name, trust_remote_code=True)
    
    # Load the LoRA adapter weights
    # Replace "Grandediw/lora_model_finetuned" with your actual LoRA model repo
    model = PeftModel.from_pretrained(base_model, "Grandediw/lora_model_finetuned")
    
    # Create a pipeline for text generation
    chat_pipeline = pipeline(
        "text-generation",
        model=model,
        tokenizer=tokenizer,
        max_length=64,
        temperature=0.7,
        top_p=0.9,
        pad_token_id=tokenizer.eos_token_id
    )
    return chat_pipeline

chat_pipeline = load_model()

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

# Display chat messages from history
for message in st.session_state.messages:
    with st.chat_message(message["role"]):
        st.markdown(message["content"])

# User input
if prompt := st.chat_input("Ask me anything:"):
    # Display user message
    st.chat_message("user").markdown(prompt)
    st.session_state.messages.append({"role": "user", "content": prompt})

    # Generate response
    with st.spinner("Thinking..."):
        # Generate text with the pipeline
        response = chat_pipeline(prompt)[0]["generated_text"]
        # Remove the prompt from the start if it's included
        if response.startswith(prompt):
            response = response[len(prompt):].strip()

    # Display assistant response
    with st.chat_message("assistant"):
        st.markdown(response)
    st.session_state.messages.append({"role": "assistant", "content": response})