app.py

import gradio as gr
import requests
# from transformers import pipeline
# pipe = pipeline("text-generation", model="shivanikerai/TinyLlama-1.1B-Chat-v1.0-seo-optimised-title-suggestion-v1.0")
API_URL = "https://api-inference.huggingface.co/models/shivanikerai/TinyLlama-1.1B-Chat-v1.0-seo-optimised-title-suggestion-v1.0"
def query(payload, api_token):
	response = requests.post(API_URL, headers={"Authorization": f"Bearer {api_token}"}, json=payload)
	return response.json()

def my_function(api_token, keywords, product_info):
    B_SYS, E_SYS = "<<SYS>>", "<</SYS>>"
    B_INST, E_INST = "[INST]", "[/INST]"
    B_in, E_in = "[Product Details]", "[/Product Details]"
    B_out, E_out = "[Suggested Titles]", "[/Suggested Titles]"
    prompt = f"""{B_INST} {B_SYS} You are a helpful, respectful and honest assistant for ecommerce product title creation. {E_SYS}
    Create a SEO optimized e-commerce product title for the keywords:{keywords.strip()}
    {B_in}{product_info}{E_in}\n{E_INST}\n\n{B_out}"""
    # predictions = pipe(prompt)
    # output=((predictions[0]['generated_text']).split(B_out)[-1]).strip()
    output = query({
	"inputs": prompt,
    },api_token)
    return (output)

    # Process the inputs (e.g., concatenate strings, perform calculations)
    # result = f"You entered: {input1} and {input2}"
    # return result

# Create the Gradio interface
interface = gr.Interface(fn=my_function, 
                          inputs=["text", "text", "text"], 
                          outputs="text", 
                          title="SEO Optimised Title Suggestion", 
                          description="Enter Keywords and Product Info:")

interface.launch()