ZTE-AIM/Telecom-Function-Calling-Evaluation
收藏资源简介：
# TFCE(Telecom Function-Calling Evaluation)数据集 ### 数据集摘要 TFCE是一个评估通信领域函数调用能力的数据集，由1800余个函数构成917道Python题目，并应用于通信领域的Simple（简单函数）、Multiple（多函数）、Parallel（并行函数）、Parallel-Multiple（并行多函数）等场景，涉及4G LTE、5G技术与6G探索、无线通信与网络优化、物联网（IoT）与M2M通信、移动通信系统与实施、网络安全与协议等方面的内容。 ### 语言 数据集中question的文本是中文；其他部分的文本是英文。 ## 数据集结构 TECE数据集中的数据按照“question-function-required”的结构；其中，“function”由“name”、“description”、“parameters”组成；“parameters”由“type”和“properties”组成。 按照数据类别分类，TFCE数据集中的数据可分为简单函数、多函数、并行函数、并行多函数4种。 | 数据集类别 | 数据集介绍 | | ----- | -------------------------------------------------------------------------- | | 简单函数 | 简单函数是TFCE数据集中最简单但最常见的格式，包含1个问题、1个函数、1个答案，用户提供一个单一的JSON函数文档，且每次仅调用一个函数。 | | 多函数 | 多函数类别包含1个问题，2-4个函数、1个答案，并根据该问题从多个JSON函数文档中仅调用1个可用函数，其余函数对于该问题均不可用。 | | 并行函数 | 并行函数包含1个问题，1个函数、2-4个答案，根据需要多次调用该函数，因此产生了多个答案。 | | 并行多函数 | 并行多函数包含1个问题，2-4个函数、2-4个答案，可看作是并行函数和多函数两种类别的组合。使用并行多函数时，将根据需要对应调用可用函数及调用次数。 | ## 数据实例 ### 简单函数&并行函数 由于简单函数和并行函数在“question”与“function”的结构上基本相同，仅在“possible\_answer”部分，简单函数为1次调用的结果，而并行函数由于多次调用构成了列表形式。 以下是一个简单函数的举例：通过调用一个函数，根据给定的地理区域（亚洲）、频率范围（1-6 GHz）和服务类型（5G），找到适合该区域5G部署的频段。 ```JSON { "id": "base_simple_3", "question": "在亚洲地区，1-6 GHz频段内，适合5G部署的频段有哪些？", "function": { "name": "find_suitable_bands", "description": "Determines suitable frequency bands for 5G deployment based on regional specifications and frequency range.", "parameters": { "type": "dict", "properties": { "region": { "type": "string", "description": "The geographical region for which suitable bands are sought." }, "frequency_range": { "type": "string", "description": "The frequency range of interest (e.g., below 1 GHz, 1-6 GHz, above 6 GHz)." }, "service_type": { "type": "string", "description": "The type of service (e.g., 5G, IMT-Advanced)." } }, "required": [ "region", "frequency_range", "service_type" ] } } } { "id": "base_simple_3", "ground_truth": { "find_suitable_bands": { "region": "Asia", "frequency_range": "1-6 GHz", "service_type": "5G" } } } ``` ### 多函数&并行多函数 多函数和并行多函数同样在“question”与“function”的结构上基本相同，仅在“possible\_answer”部分，多函数为1次调用的结果，而并行多函数由于多次调用构成了列表形式。 以下是一个并行多函数的举例：通过调用两个函数来分别评估不同HARQ过程并计算28GHz信号在城市环境下的路径损耗。 ```JSON { "id": "base_parallel_multi_55", "question": "在进行毫米波通信测试时，如何评估不同HARQ过程并计算28GHz信号在城市环境下的路径损耗？", "function": [ { "name": "evaluate_v2x_sidelink_harq_procedures", "description": "Evaluates the HARQ procedures for V2X communications, including automatic retransmission without HARQ feedback, HARQ option 1 with NACK reporting, and HARQ option 2 with ACK and NACK reporting.", "parameters": { "type": "dict", "properties": { "automatic_retransmission": { "type": "boolean", "description": "Indicates whether the automatic retransmission procedure without HARQ feedback is properly understood, including the indication of resources for transmission in time and frequency by the SCI and the scheduling of up to three transmissions at the same time." }, "harq_option_1": { "type": "boolean", "description": "Indicates whether the HARQ option 1 procedure with NACK reporting is properly understood, including the reporting of NACK when a packet is not correctly received and the use of a distance range for reporting." }, "harq_option_2": { "type": "boolean", "description": "Indicates whether the HARQ option 2 procedure with ACK and NACK reporting is properly understood, including the reporting of ACK if the packet is correctly received and NACK if the packet is not correctly received, regardless of distance." } }, "required": [ "automatic_retransmission", "harq_option_1", "harq_option_2" ] } }, { "name": "calculate_path_loss", "description": "Calculates the path loss for a given frequency, distance, and environmental conditions in millimeter wave communication.", "parameters": { "type": "dict", "properties": { "frequency": { "type": "float", "description": "The frequency of the signal in GHz." }, "distance": { "type": "float", "description": "The distance between transmitter and receiver in meters." }, "environment": { "type": "string", "description": "The environment type (LOS, NLOS, Urban, Rural)." }, "rain_condition": { "type": "string", "description": "Rain condition (None, Light Rain, Moderate Rain)." }, "vehicle_distance": { "type": "float", "description": "Distance between vehicles in vehicle-to-vehicle communication scenarios.", "default": null }, "vehicle_speed": { "type": "float", "description": "Speed of vehicles in vehicle-to-vehicle communication scenarios.", "default": null } }, "required": [ "frequency", "distance", "environment" ] } } ] } { "id": "base_parallel_multi_55", "ground_truth": { "evaluate_v2x_sidelink_harq_procedures": { "automatic_retransmission": true, "harq_option_1": true, "harq_option_2": true }, "calculate_path_loss": { "frequency": 28, "distance": 200, "environment": "Urban" } } } ``` ## 数据字段 TFCE数据集中的数据按照“问题-函数-答案”的结构，以下为各字段的解释： * question：一道关于通信领域需要调用函数解决的问题。 * function：回答相应问题所需要调用的函数，该部分由函数名称、函数描述、必需参数组成。 * possible\_answer：函数调用的结果，由各个必需函数的结果组成。 ## 数据集组成 | 类型 | Simple | Parallel | Multiple | Parallel-Multiple | | -- | ------ | -------- | -------- | ----------------- | | 数量 | 300 | 166 | 88 | 363 | ### 鸣谢 感谢 [@MeiManlin](https://github.com/MeiManlin) [@ZhangJing](https://github.com/ZhangJing41) [@YuanKejia](https://github.com/KejiaYuan) 对本数据集的贡献。
