DLIlab/webarena_paraphrased_instruction_webNinstruction_web
收藏Hugging Face2024-03-26 更新2024-06-11 收录
下载链接:
https://hf-mirror.com/datasets/DLIlab/webarena_paraphrased_instruction_webNinstruction_web
下载链接
链接失效反馈官方服务:
资源简介:
In this case, given_tasks are made through the following steps.
1. make vectorDB with webname, webdescription and instruction in mind2Web.
2. for each website in WebArena, retrieve relevant instructions in mind2web, top_30, with a threshold 0.36(L2 distance)
3. every retrieved instructions(from mind2web) are used as given_task.
---
dataset_info:
features:
- name: given_task
dtype: string
- name: annotation_id
dtype: string
- name: target_web_name
dtype: string
- name: generated_task
sequence: string
splits:
- name: train
num_bytes: 39198
num_examples: 52
download_size: 23401
dataset_size: 39198
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---
提供机构:
DLIlab原始信息汇总
数据集概述
数据集结构
- 特征信息:
given_task:任务描述,数据类型为字符串。annotation_id:注释ID,数据类型为字符串。target_web_name:目标网站名称,数据类型为字符串。generated_task:生成的任务描述,数据类型为字符串序列。
数据集划分
- 训练集:
- 数据大小:39198字节
- 样本数量:52个
数据集大小
- 下载大小:23401字节
- 数据集总大小:39198字节



