{
  "num_docs": 20000,
  "seq_length": 1024,
  "tokenizer": "gpt2",
  "output_dir": "data",
  "min_text_length": 50,
  "max_text_length": 10000
}