roberta-base-mr / test_marathi_model.py
nipunsadvilkar's picture
Large model with 10% validation set
e307ddc
import sys
import pprint
from transformers import pipeline, AutoTokenizer, RobertaForMaskedLM
tokenizer = AutoTokenizer.from_pretrained("./")
model = RobertaForMaskedLM.from_pretrained("./", from_flax=True)
nlp = pipeline("fill-mask", model=model, tokenizer=tokenizer)
# masked_input = "माझा नाव <mask> आहे"
# masked_input = "वाढदिवसाच्या <mask> शुभेच्छा!"
masked_input = sys.argv[1]
print(masked_input)
pprint.pprint(nlp(masked_input))