handler.py

import torch

from typing import Any, Dict
from transformers import AutoModelForCausalLM, AutoTokenizer, AutoConfig
from transformers.models.auto import modeling_auto


class EndpointHandler:
    def __init__(self, path=""):
        print('starting machine')
        config = AutoConfig.from_pretrained("Kowsher/Egol_model", trust_remote_code=True)
        # load model and tokenizer from path
        self.tokenizer = AutoTokenizer.from_pretrained(path, trust_remote_code=True)
        self.model = AutoModelForCausalLM.from_pretrained(
            path, device_map="auto", torch_dtype=torch.float16, config = config, trust_remote_code=True
        )
        self.device = "cuda" if torch.cuda.is_available() else "cpu"

    def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
        # process input
        inputs = data.pop("inputs", data)
        
        parameters = data.pop("parameters", None)

        # preprocess
        print(print("inputs......", inputs))
        inputs = self.tokenizer(inputs, return_tensors="pt").to(self.device)
        



        t=0
        for j in range(len(inputs['token_type_ids'][0])):
            if inputs['input_ids'][0][j]==39 and inputs['input_ids'][0][j+1]== 5584:
                t=0
            if inputs['input_ids'][0][j]==39 and inputs['input_ids'][0][j+1]== 13359:
                t=1
            inputs['token_type_ids'][0][j]=t


        # pass inputs with all kwargs in data
        print("inputs......", inputs)
        if parameters is not None:
            outputs = self.model.generate(**inputs, **parameters)
        else:
            outputs = self.model.generate(**inputs)

        # postprocess the prediction
        prediction = self.tokenizer.decode(outputs[0], skip_special_tokens=True)

        return [{"generated_text": prediction}]