app.py

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
import gradio as gr

# Load model directly
tokenizer = AutoTokenizer.from_pretrained("MTSAIR/multi_verse_model")
model = AutoModelForCausalLM.from_pretrained("MTSAIR/multi_verse_model")

def greet(name):
    #i want to get same result res = pipe(name, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
    #but using tokenizer and model
    input_ids = tokenizer.encode(name, return_tensors='pt')
    res = model.generate(input_ids, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
    generated = tokenizer.decode(res[0], skip_special_tokens=True)
    return generated

iface = gr.Interface(fn=greet, inputs="text", outputs="text")
iface.launch()