from transformers import AutoTokenizer, AutoModelForTokenClassification
import numpy as np

# match tag
id2tag = {0:'O', 1:'B_MT', 2:'I_MT'}

# load model & tokenizer
MODEL_NAME = 'MDDDDR/roberta_large_NER'

model = AutoModelForTokenClassification.from_pretrained(MODEL_NAME)
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)

# prepare input
text = 'mental disorder can also contribute to the development of diabetes through various mechanism including increased stress, poor self care behavior, and adverse effect on glucose metabolism.'
tokenized = tokenizer(text, return_tensors='pt')

# forward pass
output = model(**tokenized)

# result
pred = np.argmax(output[0].cpu().detach().numpy(), axis=2)[0][1:-1]

# check pred
for txt, pred in zip(tokenizer.tokenize(text), pred):
    print("{}\t{}".format(id2tag[pred], txt))
    # B_MT	▁mental 
    # B_MT	▁disorder

Framework versions

transformers : 4.39.1
torch : 2.1.0+cu121
datasets : 2.18.0
tokenizers : 0.15.2
numpy : 1.20.0