handler.py

from typing import Dict, List, Any
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import torch
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
class EndpointHandler:
    def __init__(self, path=""):
        self.tokenizer = AutoTokenizer.from_pretrained(path).to(device)
        self.model = AutoModelForSeq2SeqLM.from_pretrained(path).to(device))

    def __call__(self, data: str) -> str:
        inp = self.tokenizer(data, return_tensors="pt")
        for q in inp:
            inp[q] = inp[q].to(device)
        with torch.inference_mode():
            out= model.generate(**inp)
        final_output = tokenizer.batch_decode(out,skip_special_tokens=True)
        return {"translation": final_output[0]}