app.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

# Load the model and tokenizer

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("valeriojob/MedGPT-Llama3.1-8B-BA-v.1")
model = AutoModelForCausalLM.from_pretrained("valeriojob/MedGPT-Llama3.1-8B-BA-v.1")

def respond_to_query(user_input):
    input_ids = tokenizer.encode(user_input + tokenizer.eos_token, return_tensors="pt")

    # Generate a response from the model
    with torch.no_grad():
        response_ids = model.generate(input_ids, max_length=150, num_return_sequences=1)

    response = tokenizer.decode(response_ids[0], skip_special_tokens=True)
    return response

# Create a Gradio interface
iface = gr.Interface(fn=respond_to_query,
                     inputs="text",
                     outputs="text",
                     title="MedGPT Chatbot",
                     description="Ask your medical questions to MedGPT!")

if __name__ == "__main__":
    iface.launch()