app.py

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

# Load the base model and the fine-tuned model
@st.cache_resource
def load_model():
    base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-Coder-3B-Instruct")
    model = PeftModel.from_pretrained(base_model, "mohamedyd/Natural-Coder-3B-Instruct-V1")
    tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-Coder-3B-Instruct")
    return model, tokenizer

model, tokenizer = load_model()

# Streamlit app
st.title("Natural-Coder-3B-Instruct-V1 Model Interaction")

# Text input for user prompt
user_input = st.text_area("Enter your prompt here:", height=150)

# Button to generate response
if st.button("Generate Response"):
    if user_input:
        # Tokenize the input
        inputs = tokenizer(user_input, return_tensors="pt")
        
        # Generate response
        outputs = model.generate(**inputs, max_length=512, num_return_sequences=1)
        
        # Decode the output
        response = tokenizer.decode(outputs[0], skip_special_tokens=True)
        
        # Display the response
        st.write("Model Response:")
        st.write(response)
    else:
        st.write("Please enter a prompt to generate a response.")