app.py

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

# Load model and tokenizer
@st.cache_resource()
def load_model():
    model_name = "deepseek-ai/deepseek-coder-1.3b-instruct"  # Smaller version
    
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModelForCausalLM.from_pretrained(
        model_name, 
        torch_dtype=torch.float16, 
        device_map="auto"  # Automatically assigns to CPU/GPU if available
    )
    
    return model, tokenizer

model, tokenizer = load_model()