AgentSearch/AgentSearchBench-Responses

---
dataset_info:
features:
- name: id
dtype: string
- name: task
dtype: string
- name: agent_id
dtype: string
- name: response
dtype: string
- name: time
dtype: string
splits:
- name: train
num_bytes: 182653139
num_examples: 49040
download_size: 80475361
dataset_size: 182653139
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---

# AgentSearchBench Responses

**AgentSearchBench** is a large-scale benchmark for AI agent search, built from nearly 10,000 real-world agents sourced from the [GPT Store](https://chatgpt.com/gpts), [Google Cloud Marketplace](https://cloud.google.com/marketplace), and [AgentAI Platform](https://agent.ai/).

🌐 [Project Page](https://bingo-w.github.io/AgentSearchBench) • 💻 [Codebase](https://github.com/Bingo-W/AgentSearchBench)

---

## Overview

This repository contains the **raw agent execution responses** collected during the construction of AgentSearchBench. Candidate agents were executed against each task in the validation set, and their outputs were evaluated by an LLM Judge to produce execution-grounded relevance labels.

These responses are released to support reproducibility and to enable research into agent evaluation, output quality analysis, and judge calibration.

---

## Dataset Statistics

| Split | Responses |
|-------|-----------|
| Validation | 60,000+ |

Responses cover single-agent task queries from the validation set.

---

## Data Fields

- `id`: Unique identifier for each response.
- `task`: Task associated with the response.
- `agent_id`: Identifier of the agent associated with the response.
- `response`: Response content
- `latency`: End-to-end latency

---

## Usage

```python
from datasets import load_dataset

ds = load_dataset("AgentSearch/AgentSearchBench-Responses")
```

---

## Related Datasets

| Dataset | Description |
|---------|-------------|
| [AgentSearchBench-Tasks](https://huggingface.co/datasets/AgentSearch/AgentSearchBench-Tasks) | Benchmark tasks: single-agent queries, multi-agent queries, and task descriptions |
| [AgentSearchBench-Agents](https://huggingface.co/datasets/AgentSearch/AgentSearchBench-Agents) | The AgentBase dataset: 9,759 real-world AI agents with metadata |

---

## Citation

```bibtex
@article{}
```

数据集信息：
特征：
- 字段名：id
数据类型：字符串
- 字段名：task
数据类型：字符串
- 字段名：agent_id
数据类型：字符串
- 字段名：response
数据类型：字符串
- 字段名：time
数据类型：字符串
数据集划分：
- 划分名称：train
占用字节数：182653139
样本数量：49040
下载大小：80475361
数据集总大小：182653139
配置项：
- 配置名称：default
数据文件：
- 划分：train
路径：data/train-*

# AgentSearchBench 响应数据集

**AgentSearchBench** 是面向AI智能体（AI Agent）搜索的大规模基准测试集，其数据源涵盖来自[GPT商店](https://chatgpt.com/gpts)、[谷歌云市场](https://cloud.google.com/marketplace)以及[AgentAI平台](https://agent.ai/)的近10000个真实AI智能体。

🌐 [项目主页](https://bingo-w.github.io/AgentSearchBench) • 💻 [代码仓库](https://github.com/Bingo-W/AgentSearchBench)

---

## 数据集概述

本仓库包含AgentSearchBench构建过程中采集的**原始AI智能体执行响应**。候选智能体针对验证集中的每项任务执行推理，其输出由大语言模型评判器（LLM Judge）进行评估，以生成基于执行过程的相关性标签。

本数据集的发布旨在支持研究可复现性，并推动AI智能体评估、输出质量分析以及评判器校准相关研究。

---

## 数据集统计数据

| 数据集划分 | 响应数量 |
|-------|-----------|
| 验证集 | 60000+ |

响应覆盖验证集中的单智能体任务查询。

---

## 数据字段说明

- `id`：每条响应的唯一标识符。
- `task`：该响应对应的任务内容。
- `agent_id`：生成该响应的AI智能体标识符。
- `response`：AI智能体生成的响应文本。
- `latency`：端到端推理延迟。

---

## 使用方法

python
from datasets import load_dataset

ds = load_dataset("AgentSearch/AgentSearchBench-Responses")

---

## 相关数据集

| 数据集名称 | 数据集描述 |
|---------|-------------|
| [AgentSearchBench-Tasks](https://huggingface.co/datasets/AgentSearch/AgentSearchBench-Tasks) | 基准测试任务集：包含单智能体查询、多智能体查询以及任务描述 |
| [AgentSearchBench-Agents](https://huggingface.co/datasets/AgentSearch/AgentSearchBench-Agents) | AgentBase数据集：包含9759个带元数据的真实AI智能体 |

---

## 引用格式

bibtex
@article{}