app.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer

# Load the model and tokenizer
model = AutoModelForCausalLM.from_pretrained("Smilyai-labs/Smily-ultra-1")
tokenizer = AutoTokenizer.from_pretrained("Smilyai-labs/Smily-ultra-1")

# Function to generate responses from the model
def chatbot(input_text, chat_history=[]):
    # Encode the new user input, add the chat history as context
    new_user_input_ids = tokenizer.encode(input_text + tokenizer.eos_token, return_tensors='pt')
    
    # Concatenate the chat history and the new user input
    bot_input_ids = new_user_input_ids
    for history in chat_history:
        bot_input_ids = torch.cat([history['input_ids'], bot_input_ids], dim=-1)

    # Generate a response from the model
    chat_history_ids = model.generate(bot_input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id, 
                                      temperature=0.7, top_k=50, top_p=0.95, do_sample=True, 
                                      eos_token_id=tokenizer.eos_token_id)
    
    # Decode the generated response
    bot_response = tokenizer.decode(chat_history_ids[0], skip_special_tokens=True)
    
    # Update the chat history
    chat_history.append({'input_ids': bot_input_ids, 'response': bot_response})
    
    return bot_response, chat_history

# Gradio interface with a more chatbot-like layout
with gr.Blocks() as demo:
    with gr.Column():
        # Create a text box to display conversation history
        chatbot_output = gr.Chatbot(label="Chatbot")
        
        # Create a text box for user input at the bottom
        user_input = gr.Textbox(label="Your Message", placeholder="Type your message here...", show_label=False)
        
        # Create a submit button to trigger the chatbot function
        submit_button = gr.Button("Send")
        
        # Link the button to the chatbot function
        submit_button.click(chatbot, inputs=[user_input, chatbot_output], outputs=[chatbot_output, chatbot_output])

# Launch the Gradio interface
demo.launch()