app.py

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

title = "Smart AI ChatBot"
description = "A conversational model capable of intelligently answering questions (DialoGPT)"
examples = [["How are you?"], ["What's the weather like?"]]

# Load DialoGPT model and tokenizer
tokenizer = AutoTokenizer.from_pretrained("microsoft/DialoGPT-large")
model = AutoModelForCausalLM.from_pretrained("microsoft/DialoGPT-large")

# Known question-answer pairs, you can add more as per your requirement
known_questions_answers = {
    "How are you?": "I'm fine, thank you for asking.",
    "What's the weather like?": "The weather is nice today, sunny and warm.",
    "What's your name?": "I am Smart AI ChatBot.",
    "Do you speak English?": "I can understand and respond to English questions.",
}

def predict(input, history=[]):
    response = None
    
    # Check if the input question is in the known question-answer pairs
    if input in known_questions_answers:
        response = known_questions_answers[input]
    else:
        # Tokenize the new user input sentence
        new_user_input_ids = tokenizer.encode(input + tokenizer.eos_token, return_tensors="pt")
        # Append the new user input tokens to the chat history
        bot_input_ids = torch.cat([torch.LongTensor(history), new_user_input_ids], dim=-1)
        # Generate a response
        history = model.generate(
            bot_input_ids, max_length=400, pad_token_id=tokenizer.eos_token_id
        ).tolist()
        # Convert tokens to text, and split the response into lines
        response = tokenizer.decode(history[0], skip_special_tokens=True)
    
    return response, history

def main():
    # You can add logic here to read known question-answer pairs, for example, from a JSON file
    pass

gr.Interface(
    fn=predict,
    title=title,
    description=description,
    examples=examples,
    inputs=["text", "state"],
    outputs=["chatbot", "state"],
    theme="finlaymacklon/boxy_violet",
).launch()