app.py

from model import load_model, answer
from components import create_app_layout
import torch 


# Check if a GPU is available; otherwise, use the CPU
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# Load the model and tokenizer
model, tokenizer = load_model()
model.to(device) 

# Define the function for Gradio to call
def gradio_answer_fn(query):
    return answer(model, tokenizer, query)

# Create the app layout
app = create_app_layout(gradio_answer_fn)

# Launch the app
if __name__ == "__main__":
    app.launch(share=True)