app.py

import sys
import subprocess

# implement pip as a subprocess:
subprocess.check_call([sys.executable, '-m', 'pip', 'install','--quiet','sentencepiece==0.1.95'])

import gradio as gr
from transformers import pipeline
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import torch

tokenizer = AutoTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-ar")
model = torch.load("helsinki_fineTuned.pt")
translation_pipeline = pipeline(model)


def translate_gradio(input):
    encode = translation_pipeline.generate(**tokenizer.prepare_seq2seq_batch(input,return_tensors='pt').to('cuda'))
    text_ar = tokenizer.batch_decode(encode,skip_special_tokens=True)[0]
    return text_ar



translate_interface = gr.Interface(fn = translate_gradio,
                                   inputs="text",
                                   outputs="text"                      )
translate_interface.launch(inline = False)