app.py

import gradio as gr
import requests

# Hugging Face Inference API URL
API_URL = "https://api-inference.huggingface.co/models/Polygl0t/Tucano2-qwen-3.7B-Instruct"

# NOTE: Apne Hugging Face Settings -> Access Tokens mein jaakar ek READ token banayein
# Aur Space ke Settings mein 'HF_TOKEN' naam ke Repository Secret mein use daal dein.
# Agar bina token ke chalana hai, toh kabhi-kabhi rate limit aa sakti hai.
headers = {"Authorization": "Bearer YOUR_HF_TOKEN_HERE"} 

def chat_with_model(user_message):
    payload = {
        "inputs": f"<|im_start|>user\n{user_message}<|im_end|>\n<|im_start||>assistant\n",
        "parameters": {"max_new_tokens": 100, "return_full_text": False}
    }
    
    try:
        response = requests.post(API_URL, headers=headers, json=payload)
        output = response.json()
        
        # Jawab nikalne ke liye
        if isinstance(output, list) and len(output) > 0:
            return output[0].get('generated_text', 'Koi jawab nahi mila.')
        elif isinstance(output, dict) and 'error' in output:
            return f"Error: {output['error']}"
        return str(output)
    except Exception as e:
        return f"Kuch gadbad hui: {str(e)}"

# Gradio UI
demo = gr.Interface(
    fn=chat_with_model, 
    inputs=gr.Textbox(label="Aapka Sawal", placeholder="Yahan likhein..."), 
    outputs=gr.Textbox(label="Model ka Jawab"),
    title="Tucano2 Qwen Chat (Fast API)"
)

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