From scratch GPT-2 117M for fine-tuning on PG essays (and future datasets)
Fine-tune GPT-2 117M on PG essays