app.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer

# Load the satvikag/chatbot model and tokenizer from Hugging Face
model_name = "satvikag/chatbot"
model = AutoModelForCausalLM.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

# Initialize a cache to store previous responses
cache = {}

# Function to handle user input and generate chatbot response
def chatbot_response(user_input):
    if user_input in cache:
        return cache[user_input]  # Return the cached response if available
    
    # Encode user input
    new_user_input_ids = tokenizer.encode(user_input + tokenizer.eos_token, return_tensors='pt')

    # Get the response from the model
    bot_output = model.generate(new_user_input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id, no_repeat_ngram_size=2)

    # Decode the response and return
    bot_output_text = tokenizer.decode(bot_output[:, new_user_input_ids.shape[-1]:][0], skip_special_tokens=True)

    # Cache the response
    cache[user_input] = bot_output_text
    return bot_output_text

# Create a simple interface using Gradio
iface = gr.Interface(fn=chatbot_response, inputs="text", outputs="text", live=True)

# Launch the chatbot
iface.launch()