davanstrien/similarity-dataset-sc2-8b

---
language:
- en
size_categories: n<1K
dataset_info:
features:
- name: anchor
dtype: string
- name: positive
dtype: string
- name: negative
dtype: string
splits:
- name: train
num_bytes: 1061977
num_examples: 2324
download_size: 488823
dataset_size: 1061977
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
tags:
- synthetic
- distilabel
- sentence-transformers
- DistilSimData
---

<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 similarity-dataset-sc2-8b

This dataset has been created with [distilabel](https://distilabel.argilla.io/) and the pipeline outlined [here](https://github.com/davanstrien/awesome-synthetic-datasets/tree/main/examples/embedding-datasets).
It is designed as a synthetic dataset for training Sentence Transformers models, providing structured examples to help models learn fine-grained semantic distinctions in various domains.

## Dataset Summary

The `similarity-dataset-sc2-8b` was generated to serve as training data for models that need to understand subtle differences and similarities between sentences.
It leverages a custom pipeline for generating positive and negative sentence pairs (positive and negative examples) related to programming tasks, particularly prompts for Python functions.

The dataset is based on [bigcode/self-oss-instruct-sc2-exec-filter-50k](https://huggingface.co/datasets/bigcode/self-oss-instruct-sc2-exec-filter-50k).

## 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/davanstrien/similarity-dataset-sc2-8b/raw/main/pipeline.yaml"
```

or explore the configuration:

```console
distilabel pipeline info --config "https://huggingface.co/datasets/davanstrien/similarity-dataset-sc2-8b/raw/main/pipeline.yaml"
```

## Dataset structure

The examples have the following structure per configuration:

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

```json
{
"anchor": "Write a Python function that checks if an object is a subclass of a given class. The function should safely handle exceptions and return `False` if an exception is raised.",
"generation": "{\"bad\": [\"Write a Python function that generates random numbers based on the input of a cat\u0027s meow. The function should also calculate the average of the generated numbers and return the average as a string.\",\"Write a Python function that takes a string input and translates it into a secret code by replacing each letter with a random number.\"], \n\"good\": [\"Write a Python function that checks if a given object is a subclass of a specified class, handling any exceptions that may occur safely and returning False if an exception is raised.\",\"Write a Python function that determines whether a provided object is a subclass of a given class, and returns False if an exception is encountered during the check.\"]}",
"negative": "Write a Python function that takes a string input and translates it into a secret code by replacing each letter with a random number.",
"positive": "Write a Python function that checks if a given object is a subclass of a specified class, handling any exceptions that may occur safely and returning False if an exception is raised."
}
```

This subset can be loaded as:

```python
from datasets import load_dataset

ds = load_dataset("davanstrien/similarity-dataset-sc2-8b", "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("davanstrien/similarity-dataset-sc2-8b")
```

</details>

原始信息汇总

数据集卡片 for similarity-dataset-sc2-8b

数据集概述

similarity-dataset-sc2-8b 是为需要理解句子之间细微差异和相似性的模型设计的训练数据集。它利用自定义管道生成与编程任务相关的正负句子对（正例和负例），特别是针对Python函数的提示。

该数据集基于 bigcode/self-oss-instruct-sc2-exec-filter-50k。

数据集结构

特征

• anchor: 字符串类型
• positive: 字符串类型
• negative: 字符串类型

分割

• train: 包含 2324 个样本，总大小为 1061977 字节

配置

• default:
  • 数据文件路径: data/train-*

示例

json { "anchor": "Write a Python function that checks if an object is a subclass of a given class. The function should safely handle exceptions and return False if an exception is raised.", "negative": "Write a Python function that takes a string input and translates it into a secret code by replacing each letter with a random number.", "positive": "Write a Python function that checks if a given object is a subclass of a specified class, handling any exceptions that may occur safely and returning False if an exception is raised." }

加载数据集

python from datasets import load_dataset

ds = load_dataset("davanstrien/similarity-dataset-sc2-8b")