eval-fsr-a1-stackexchange-codereview-swe-r378-rf0710-traces
收藏资源简介:
该数据集是一个多轮对话任务执行记录集合,主要用于记录和评估人工智能代理在特定任务中的表现。数据集包含1988个训练样本,每个样本代表一次完整的任务执行过程。核心字段包括:conversations(对话列表,包含每条消息的内容和角色)、agent(执行代理标识)、model和model_provider(使用的模型及提供商)、date(执行日期)、task(任务类型)、episode(任务执行轮次标识)、run_id和trial_name(运行和试验标识)、result(任务执行结果)、verifier_output(验证器输出)、trace_source(数据来源追踪)。数据集适用于对话系统评估、任务导向对话研究、代理行为分析等场景。
This dataset is a collection of multi-turn dialogue task execution records, primarily used to record and evaluate the performance of AI Agents in specific tasks. It contains 1988 training samples, with each sample representing a complete task execution process. The core fields include: conversations (a list of dialogues containing the content and role of each message), agent (execution agent identifier), model and model_provider (the utilized model and its provider), date (execution date), task (task type), episode (task execution round identifier), run_id and trial_name (runtime and trial identifiers), result (task execution result), verifier_output (verifier output), and trace_source (data source tracking). This dataset is applicable to scenarios such as dialogue system evaluation, task-oriented dialogue research, and agent behavior analysis.
- 数据集名称:laion/eval-fsr-a1-stackexchange-codereview-swe-r378-rf0710-traces
- 数据集地址:https://huggingface.co/datasets/laion/eval-fsr-a1-stackexchange-codereview-swe-r378-rf0710-traces
数据集特征
conversations:列表类型,包含以下子字段:content:字符串类型role:字符串类型
agent:字符串类型model:字符串类型model_provider:字符串类型date:字符串类型task:字符串类型episode:字符串类型run_id:字符串类型trial_name:字符串类型result:字符串类型verifier_output:字符串类型trace_source:字符串类型
数据集划分与规模
- 默认配置:
default - 训练集:
- 样本数量:1988
- 字节数:299,240,322
- 总下载大小:232,853,969 字节
- 总数据集大小:299,240,322 字节
数据文件
- 训练集文件路径:
data/train-*(使用通配符匹配多个文件)




