models/gpt2.py

from transformers import GPT2Tokenizer, TFGPT2LMHeadModel
import tensorflow as tf

model_name = "gpt2"

def load():
  global model
  global tokenizer
  
  model = TFGPT2LMHeadModel.from_pretrained(model_name)
  tokenizer = GPT2Tokenizer.from_pretrained(model_name)

def generate(input_text):
  # Tokenize the input text
  input_ids = tokenizer.encode(input_text, return_tensors="pt", truncation=True)
  
  # Generate output using the model
  output_ids = model.generate(input_ids, num_beams=3, no_repeat_ngram_size=2, max_new_tokens=200, eos_token_id=tokenizer.eos_token_id)
  
  return tokenizer.decode(output_ids[0], skip_special_tokens=True)