{
  "authored_word_counts": {
    "کار": 4,
    "کاربر": 2,
    "بار": 3,
    "زار": 1
  },
  "initial_units": "Unicode characters inside each already-separated word",
  "merge_steps": 3,
  "tie_break": "highest weighted frequency, then lexicographic pair order",
  "no_cross_word_merges": true,
  "unknown_character_policy": "reject",
  "normalization": null,
  "byte_level_BPE": false
}