gpt-code / clone_sentdex_model_tokenizer.py
cakiki's picture
Add training stub
326fe35
raw
history blame contribute delete
867 Bytes
#!/usr/bin/env python
from transformers import AutoTokenizer, TFAutoModelForCausalLM, AutoModelForCausalLM, GPT2Config
import tensorflow as tf
task_specific_params = {
"text-generation": {
"do_sample": False,
"max_length": 50
}
}
config = GPT2Config.from_pretrained("Sentdex/GPyT", _name_or_path='prophetikai/code-gpt', use_cache=True, task_specific_params=task_specific_params)
tokenizer = AutoTokenizer.from_pretrained("Sentdex/GPyT")
tf_model = TFAutoModelForCausalLM.from_pretrained("Sentdex/GPyT", config=config)
pytorch_model = AutoModelForCausalLM.from_pretrained("Sentdex/GPyT", config=config)
config.save_pretrained('./')
tokenizer.save_pretrained(save_directory='./')
tf_model.save_pretrained(save_directory='./', saved_model=True, version='sentdex')
pytorch_model.save_pretrained(save_directory='./')