Mervyn999/distilabel-instruction-to-preference-dataset

---
size_categories: n<1K
dataset_info:
features:
- name: instruction
dtype: string
- name: generations
sequence: string
- name: generation_models
sequence: string
- name: ratings
sequence: int64
- name: rationales
sequence: string
splits:
- name: train
num_bytes: 31573
num_examples: 10
download_size: 25779
dataset_size: 31573
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
tags:
- synthetic
- distilabel
- rlaif
---

<p align="left">
<a href="https://github.com/argilla-io/distilabel">
<img src="https://raw.githubusercontent.com/argilla-io/distilabel/main/docs/assets/distilabel-badge-light.png" alt="Built with Distilabel" width="200" height="32"/>
</a>
</p>

# Dataset Card for distilabel-instruction-to-preference-dataset

This dataset has been created with [distilabel](https://distilabel.argilla.io/).

## Dataset Summary

This dataset contains a `pipeline.yaml` which can be used to reproduce the pipeline that generated it in distilabel using the `distilabel` CLI:

```console
distilabel pipeline run --config "https://huggingface.co/datasets/Mervyn999/distilabel-instruction-to-preference-dataset/raw/main/pipeline.yaml"
```

or explore the configuration:

```console
distilabel pipeline info --config "https://huggingface.co/datasets/Mervyn999/distilabel-instruction-to-preference-dataset/raw/main/pipeline.yaml"
```

## Dataset structure

The examples have the following structure per configuration:

<details><summary> Configuration: default </summary><hr>

```json
{
"generation_models": [
"microsoft/phi-3-mini-4k-instruct",
"google/gemma-1.1-7b-it"
],
"generations": [
"Let\u0027s denote Robbie\u0027s number of chocolates as R. According to the information given:\n\n1. Arianna has twice as many chocolates as Robbie, so Arianna has 2R chocolates.\n2. Danny has 6 chocolates more than Robbie, so Danny has R + 6 chocolates.\n3. Arianna has 12 more chocolates than Danny, so we can write the equation: 2R = (R + 6) + 12.\n\nNow, let\u0027s solve for R:\n\n2R = R + 6 + 12\n2R = R + 18\n2R - R = 18\nR = 18\n\nNow that we know Robbie has 18 chocolates, we can find out how many chocolates Danny has:\n\nDanny = R + 6\nDanny = 18 + 6\nDanny = 24\n\nDanny has 24 chocolates.\n \u003c|end|\u003e",
"Let X be the number of chocolates Robbie has.\n\nArianna has 2X chocolates.\nDanny has X + 6 chocolates.\n\nAccording to the given information, Arianna has 12 more chocolates than Danny:\n2X = X + 6 + 12 \n2X = X + 18\n\nSolving for X:\nX = 18\n\nTherefore, Danny has X + 6 = 18 + 6 = 24 chocolates.\n\nThus, Danny has **24 chocolates**."
],
"instruction": "Arianna has 12 chocolates more than Danny. Danny has 6 chocolates more than Robbie. Arianna has twice as many chocolates as Robbie has. How many chocolates does Danny have?",
"ratings": [
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
5,
5
],
"rationales": [
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
null,
"The text fully aligns with the instruction, meeting all requirements. It correctly sets up the equations, solves for R, and then uses the value of R to find the number of chocolates Danny has. The solution is clear and concise.",
"The text also fully aligns with the instruction, meeting all requirements. It correctly sets up the equations, solves for X, and then uses the value of X to find the number of chocolates Danny has. The solution is clear and concise, and the use of the variable X is a good way to represent Robbie\u0027s number of chocolates."
]
}
```

This subset can be loaded as:

```python
from datasets import load_dataset

ds = load_dataset("Mervyn999/distilabel-instruction-to-preference-dataset", "default")
```

Or simply as it follows, since there's only one configuration and is named `default`:

```python
from datasets import load_dataset

ds = load_dataset("Mervyn999/distilabel-instruction-to-preference-dataset")
```

</details>

原始信息汇总

数据集概述

数据集基本信息

• 大小分类: n<1K
• 下载大小: 25779字节
• 数据集大小: 31573字节

数据集特征

• instruction: 字符串类型
• generations: 字符串序列
• generation_models: 字符串序列
• ratings: 整数序列（int64）
• rationales: 字符串序列

数据集分割

• 训练集:
  • 示例数量: 10
  • 数据大小: 31573字节

配置

• 默认配置:
  • 数据文件路径: data/train-*

标签

• 合成数据
• distilabel
• rlaif