hf_inference.py

import pickle
from transformers import AutoModelForTokenClassification, AutoTokenizer
from transformers import pipeline
text = "A software engineer coded up some software."
#text="software engineer"
#text="cto"
model = AutoModelForTokenClassification.from_pretrained('./jpn202401')
tokenizer = AutoTokenizer.from_pretrained(
   './jpn202401', model_max_length=512)
nlp = pipeline("ner", model=model, tokenizer=tokenizer,
               aggregation_strategy="average")
print(nlp(text))