Appendix B Token Template of Tool Calling

本文为论文《Kimi K2：面向智能体能力的开源大模型》「论文原文」板块的对应章节，按 arXiv 官方全文逐字保留，层级与「中文全文翻译」板块一一对应。

There are three components in the token structure for tool-calling:

• •

Tool declaration message: defines the list of available tools and the schema of the arguments;

• •

Tool invoking section in assistant message: encodes the model’s request to invoke tools;

• •

Tool result message: encapsulates the invoked tool’s execution result.

The raw tokens of the tool declaration message are formatted as follows:

<|im_begin|> tool_declare <|im_middle|>

Tools

{{ tool declaration content }} <|im_end|>

The blue highlighted marks represent special tokens, and the green part, quoted by brackets, is the tool declaration content. We use TypeScript to express the tool declaration content, since TypeScript is a concise language with a comprehensive type system, able to express the types and constraints of tool parameters with brief text. The code 1 shows an example for two simple tools in JSON format compatible with OpenAI’s chat completion API, as a comparison, the same tools defined in TypeScript (listed in Code 2) is much shorter. To improve compatibility, part of our training data also uses JSON as the tool declaration language, so that 3rd-party frameworks need not additional development to support our tool calling scheme.

Listing 1: Tool definition with JSON in OpenAI compatible API

Listing 2: Tool definition in TypeScript

The token template of the tool invoking section in the model’s response messages is listed as follows:

<tool_call_section_begin|> <|tool_call_begin|> // call_id part functions.{{tool name}}:{{counter}} <|tool_arguments_begin|> {{ json serialized call arguments }} <|tool_call_end|> <|tool_call_begin|> // more tool calls <|tool_call_end|> <|tool_call_section_end|>

As shown in the template, we support parallel tool calling by placing multiple tool calls in a single response turn. Each tool call has a unique call id, formatted as functions.{tool-name}:{counter}, where tool-name is the name of the tool, and counter is an auto-increasing counter of all tool calls starting from 0 in the dialog.

During inference, the model may occasionally generate unexpected tokens, leading to format errors when parsing a tool call. To solve this issue, we developed a constrained decoding module named enforcer, inspired by lm-format-enforcer77 7

https://github.com/noamgat/lm-format-enforcer. When a <tool_call_section_begin|> token is generated, it ensures that the upcoming tool-related tokens follow the predefined template, and the JSON argument string follows the declared schema.

The tool result message is simply a text message encoded with the tool’s call id and the corresponding results.

<|im_begin|> tool <|im_middle|>