t5-11b-sharded / handler.py
philschmid's picture
philschmid HF staff
Update handler.py
54d7f3a
from typing import Dict, List, Any
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer, pipeline
import torch
class EndpointHandler:
def __init__(self, path=""):
# load model and processor from path
self.model = AutoModelForSeq2SeqLM.from_pretrained(path, device_map="auto", load_in_8bit=True)
self.tokenizer = AutoTokenizer.from_pretrained(path)
def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
"""
Args:
data (:obj:):
includes the deserialized image file as PIL.Image
"""
# process input
inputs = data.pop("inputs", data)
parameters = data.pop("parameters", None)
# preprocess
input_ids = self.tokenizer(inputs, return_tensors="pt").input_ids
# pass inputs with all kwargs in data
if parameters is not None:
outputs = self.model.generate(input_ids, **parameters)
else:
outputs = self.model.generate(input_ids)
# postprocess the prediction
prediction = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
return [{"generated_text": prediction}]