import streamlit as st
from transformers import AutoTokenizer, AutoModelForCausalLM

@st.cache_resource(show_spinner=False)
def load_model():
    model_name = "Alijeff1214/DeutscheLexAI_BGB_2.0"
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModelForCausalLM.from_pretrained(model_name)
    return tokenizer, model

tokenizer, model = load_model()

st.title("DeutscheLexAI_BGB Chat Interface")
st.write("Interact with the fine-tuned Qwen2.5-3B model for German legal texts!")

user_input = st.text_input("Enter your question or prompt:")

if st.button("Generate Response") and user_input:
    # Tokenize and generate response (adjust parameters as needed)
    inputs = tokenizer(user_input, return_tensors="pt")
    outputs = model.generate(**inputs, max_length=500, do_sample=True, temperature=0.7)
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    st.text_area("Model Response:", value=response, height=300)