tokenizer_patch_report.json

{
  "created_at": "2026-05-24T19:24:53.388411Z",
  "base_tokenizer_repo": "vngrs-ai/Kumru-2B",
  "output_dir": "/kaggle/working/kumru-tokenizer-rslm",
  "model_max_length": 32768,
  "vocab_size": 50181,
  "num_added_tokens": 5,
  "core_token_ids": {
    "<|pad|>": 0,
    "<|unknown|>": 1,
    "<|begin_of_text|>": 2,
    "<|end_of_text|>": 3,
    "<|system_instruction|>": 4,
    "<|system_instruction_end|>": 5
  },
  "reasoning_token_ids": {
    "<think>": 50176,
    "</think>": 50177
  },
  "role_token_ids": {
    "<|user|>": 50178,
    "<|assistant|>": 50179,
    "<|end_of_turn|>": 50180
  },
  "model_config_patch": {
    "vocab_size": 50181,
    "pad_token_id": 0,
    "unk_token_id": 1,
    "bos_token_id": 2,
    "eos_token_id": 3
  }
}