import streamlit as st
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

# Page config
st.set_page_config(page_title="Gemma 3 1B Chat", page_icon="💬")

# Initialize session state
if "messages" not in st.session_state:
    st.session_state.messages = []
if "model" not in st.session_state:
    st.session_state.model = None
if "tokenizer" not in st.session_state:
    st.session_state.tokenizer = None

@st.cache_resource
def load_model():
    """Load the Gemma model and tokenizer"""
    model_name = "google/gemma-2-1b-it"  # Change to "google/gemma-3-1b-it" if available
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModelForCausalLM.from_pretrained(
        model_name,
        torch_dtype=torch.float16,
        device_map="auto"
    )
    return model, tokenizer

# Load model
with st.spinner("Loading Gemma 3 1B model..."):
    if st.session_state.model is None:
        st.session_state.model, st.session_state.tokenizer = load_model()

# Title
st.title("💬 Gemma 3 1B Chat")

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

# Chat input
if prompt := st.chat_input("Type your message here..."):
    # Add user message
    st.session_state.messages.append({"role": "user", "content": prompt})
    with st.chat_message("user"):
        st.markdown(prompt)
    
    # Generate response
    with st.chat_message("assistant"):
        with st.spinner("Thinking..."):
            # Format prompt for chat
            chat_prompt = st.session_state.tokenizer.apply_chat_template(
                [{"role": "user", "content": prompt}],
                tokenize=False,
                add_generation_prompt=True
            )
            
            # Tokenize
            inputs = st.session_state.tokenizer(
                chat_prompt,
                return_tensors="pt"
            ).to(st.session_state.model.device)
            
            # Generate
            with torch.no_grad():
                outputs = st.session_state.model.generate(
                    **inputs,
                    max_new_tokens=512,
                    temperature=0.7,
                    do_sample=True
                )
            
            # Decode response
            response = st.session_state.tokenizer.decode(
                outputs[0][inputs.input_ids.shape[1]:],
                skip_special_tokens=True
            )
            
            st.markdown(response)
    
    # Add assistant response
    st.session_state.messages.append({"role": "assistant", "content": response})

# Sidebar with clear button
with st.sidebar:
    st.header("Settings")
    if st.button("Clear Chat"):
        st.session_state.messages = []
        st.rerun()