tokenizer_meta.json

{
  "vocab_size": 65536,
  "special_tokens": [
    "<pad>",
    "<bos>",
    "<eos>",
    "<unk>",
    "<system>",
    "<user>",
    "<assistant>",
    "<tool>",
    "<tool_call>",
    "<tool_result>",
    "<repo>",
    "<file>",
    "<patch>",
    "<test>",
    "<done>"
  ],
  "special_ids": {
    "<pad>": 0,
    "<bos>": 1,
    "<eos>": 2,
    "<unk>": 3,
    "<system>": 4,
    "<user>": 5,
    "<assistant>": 6,
    "<tool>": 7,
    "<tool_call>": 8,
    "<tool_result>": 9,
    "<repo>": 10,
    "<file>": 11,
    "<patch>": 12,
    "<test>": 13,
    "<done>": 14
  },
  "source": "code-first: Stack v3 + OpenCodeInstruct + SWE-rebench OpenHands"
}