app.py

import gradio as gr
import os
import torch
import spaces
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline 
from threading import Thread


# Set an environment variable
HF_TOKEN = os.environ.get("secrets_sindhi_token", None)

# Load the tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("Telugu-LLM-Labs/Indic-gemma-7b-finetuned-sft-Navarasa-2.0")
model = AutoModelForCausalLM.from_pretrained("Telugu-LLM-Labs/Indic-gemma-7b-finetuned-sft-Navarasa-2.0",torch_dtype=torch.float16, device_map="auto")

# Create the text generation pipeline
generator = pipeline("text-generation", model=model, tokenizer=tokenizer)

# Function to generate text
def gemma(message: str, history: list, temperature: float, max_new_tokens: int) -> str:
    """
    Generate a response using the model.
    
    Args:
        message (str): The input message.
        history (list): The conversation history.
        temperature (float): The temperature for generating the response.
        max_new_tokens (int): The maximum number of new tokens to generate.
        
    Returns:
        str: The generated response.
    """
    conversation = []
    for user, assistant in history:
        conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
    conversation.append({"role": "user", "content": message})

    input_prompt = f"""
    ### Instruction:
    You are an AI assistant. Engage in a conversation with the user and provide helpful responses.
    
    ### Input:
    {message}
    
    ### Response:
    """

    inputs = tokenizer([input_prompt], return_tensors="pt").to(model.device)
    
    # Generate response using the model and parameters
    result = generator(input_prompt, max_length=max_new_tokens, temperature=temperature, do_sample=True)

    # Return the output text (clean up if needed)
    return result[0]['generated_text']

# Gradio block for the interface
chatbot = gr.Chatbot(placeholder="پنهنجي مقامي ٻولي ۾ ترغيب ڏيو", height=500)

with gr.Blocks(fill_height=True) as demo:
    # Header
    gr.Markdown("<h1 style='text-align: center; color: #4CAF50;'>سنڌي ٻولي اسسٽنٽ</h1>")
    gr.Markdown("<p style='text-align: center;'>توهان جي ذاتي AI اسسٽنٽ ۾ ڀليڪار! سنڌي ۾ ڪجهه به پڇو..</p>")
    gr.ChatInterface(
        fn=gemma,
        chatbot=chatbot,
        fill_height=True,
        additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=False, render=False),
        additional_inputs=[
            gr.Slider(minimum=0, maximum=1, step=0.1, value=0.95, label="Temperature", render=False),
            gr.Slider(minimum=128, maximum=4096, step=1, value=512, label="Max new tokens", render=False),
        ],
        examples=[
            ["مون کي ڪانءَ جي ڪهاڻي ٻڌاءِ"],
        ],
        cache_examples=False,
    )
    # Footer
    gr.Markdown("""
    <p style='text-align: center; color: #888;'>Made with ❤️ for "سنڌي ٻولي". Powered by Danyal, Yousif & Hajan.</p>
    """)

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