from transformers import TFGPT2LMHeadModel, GPT2Tokenizer

model_name = "gpt2"
tokenizer = GPT2Tokenizer.from_pretrained(model_name)
model = TFGPT2LMHeadModel.from_pretrained(model_name)