UCL-DARK/sequential-instructions

---
dataset_info:
features:
- name: dataset
dtype: string
- name: instruction
dtype: string
- name: output
dtype: string
- name: generator
dtype: string
splits:
- name: train
num_bytes: 736696
num_examples: 533
download_size: 373739
dataset_size: 736696
license: mit
task_categories:
- question-answering
- text-generation
language:
- en
pretty_name: Sequential Instructions
size_categories:
- n<1K
---
# Sequential Instructions

This is the sequential instructions dataset from [Understanding the Effects of RLHF on LLM Generalisation and Diversity](https://arxiv.org/abs/2310.06452). The dataset is in the `alpaca_eval` format.

For information about how the dataset was generated, see https://github.com/RobertKirk/stanford_alpaca.

The instructions in the dataset generally have a sequence of steps we expect the model to complete all at once. In our work, we found that RLHF models generalise much better to this dataset than SFT models when trained on the AlpacaFarm datasets.

原始信息汇总

数据集概述

基本信息

• 名称: Sequential Instructions
• 语言: 英语 (en)
• 许可证: MIT
• 任务类别:
  • 问答 (question-answering)
  • 文本生成 (text-generation)
• 大小类别: 小于1K (n<1K)

数据集结构

• 特征:
  • dataset: 数据集名称，数据类型为字符串 (string)
  • instruction: 指令，数据类型为字符串 (string)
  • output: 输出结果，数据类型为字符串 (string)
  • generator: 生成器，数据类型为字符串 (string)

数据集大小

• 下载大小: 373739 字节
• 数据集大小: 736696 字节

数据分割

• 训练集:
  • 示例数量: 533
  • 字节数: 736696 字节