chatbot_ui.py

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

# --- Model and Tokenizer Loading ---
MODEL_PATH = "01/medical_model_rl/final"
TOKENIZER_PATH = "01/medical_model_rl/final"

print("Loading model and tokenizer...")

try:
    tokenizer = AutoTokenizer.from_pretrained(TOKENIZER_PATH, padding_side='left')
    model = AutoModelForCausalLM.from_pretrained(MODEL_PATH)

    model.resize_token_embeddings(len(tokenizer))

    if tokenizer.pad_token is None:
        tokenizer.pad_token = tokenizer.eos_token

    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    model.to(device)
    model.eval()

    print(colored("Model loaded successfully.", "green"))

except Exception as e:
    print(colored(f"Error loading model: {e}", "red"))
    model = None
    tokenizer = None

# --- Chatbot Inference Function ---
def medical_chatbot(message, history):
    """
    Generates a response from the medical chatbot model.
    """
    if not model or not tokenizer:
        return "Error: Model is not loaded. Please check the console for errors."

    try:
        # Format the prompt
        full_prompt = f"Question: {message}\n\nAnswer:"

        # Tokenize the input
        inputs = tokenizer(full_prompt, return_tensors="pt", truncation=True).to(device)

        # Generate a response
        with torch.no_grad():
            output_sequences = model.generate(
                input_ids=inputs["input_ids"],
                attention_mask=inputs["attention_mask"],
                max_length=128,
                do_sample=True,
                top_k=50,
                top_p=0.95,
                num_return_sequences=1,
                pad_token_id=tokenizer.eos_token_id,
            )
        
        # Decode the response
        response_text = tokenizer.decode(output_sequences[0], skip_special_tokens=True)
        
        # Extract only the answer part
        answer = response_text.split("Answer:")[-1].strip()

        return answer

    except Exception as e:
        print(colored(f"An error occurred during inference: {e}", "red"))
        return "Sorry, I encountered an error. Please try again."

# --- Gradio UI ---
chatbot_interface = gr.ChatInterface(
    fn=medical_chatbot,
    title="Medical Chatbot",
    description="Ask any medical question, and the AI will try to answer.",
    examples=[
        ["What are the symptoms of diabetes?"],
        ["How does metformin work?"],
        ["What is the difference between a virus and a bacteria?"],
    ],
    theme="soft",
).launch(share=True)