Blame for tokenizer_stats.json

{
  "vocab_size": 8192,
  "pre_tokenizer": "' ?[A-Za-z_][A-Za-z_0-9]*| ?[0-9]{1,3}|\\s+(?!\\S)|\\s+| ?[^A-Za-z_0-9\\s]+",
  "own": {
    "avg_tokens": 164.8715,
    "p50": 152,
    "p95": 278
  },
  "roundtrip_failures": 0,
  "gpt2": {
    "avg_tokens": 144.1685,
    "p50": 131,
    "p95": 248
  }
}