handler.py

from typing import Dict, List, Any
import guidance
from transformers import AutoTokenizer, AutoTokenizer, AutoModelForCausalLM, AutoConfig
import torch

class EndpointHandler():
    def __init__(self, path=""):
        # Preload all the elements you are going to need at inference.
        name = "mosaicml/mpt-30b-instruct"

        config = AutoConfig.from_pretrained(name, trust_remote_code=True)
        config.attn_config["attn_impl"] = "triton"
        config.init_device = "cuda:0"  # For fast initialization directly on GPU!

        model = AutoModelForCausalLM.from_pretrained(
            name, config=config, torch_dtype=torch.bfloat16, trust_remote_code=True  # Load model weights in bfloat16
        )

        # model = AutoModelForCausalLM.from_pretrained("mosaicml/mpt-30b-chat", trust_remote_code=True)
        tokenizer = AutoTokenizer.from_pretrained("EleutherAI/gpt-neox-20b")
        guidance.llm = guidance.llms.Transformers(model=model, tokenizer=tokenizer)

    def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
        """
       data args:
            inputs (:obj: `str` | `PIL.Image` | `np.array`)
            kwargs
      Return:
            A :obj:`list` | `dict`: will be serialized and returned
        """

        prompt = data.pop("prompt",data)
        guidance_prompt = guidance(prompt)

        out = guidance_prompt()

        return out.text