---
title: self-cognition
canonical_url: "https://www.modelscope.cn/datasets/swift/self-cognition"
md_url: "https://www.modelscope.cn/datasets/swift/self-cognition.md"
repository: swift/self-cognition
chinese_name: "自我认知微调数据集"
last_updated: 2025-05-05
license: "Apache License 2.0"
storage_size: "29 KB"
domain:
  - text
tasks:
  - llm
downloads: 304560
stars: 222
---

# self-cognition

> self-cognition - swift 在 ModelScope 开源的数据集。介绍（Introduction） 该自我认知数据集由modelsope swift创建, 可以通过将通配符进行替换：{{NAME}}、{{AUTHOER}}，来创建属于自己大模型的自我认知数据集，总共108条。

swift/self-cognition 是 ModelScope 魔搭社区上的llm数据集，涉及 text 领域，存储大小 29 KB，采用 Apache License 2.0 许可。

- **Repository**: swift/self-cognition
- **License**: Apache License 2.0
- **Tasks**: llm
- **Domain**: text
- **Storage size**: 29 KB
- **Downloads**: 304560
- **Stars**: 222
- **Last updated**: 2025-05-05

Source: https://www.modelscope.cn/datasets/swift/self-cognition

---

## 介绍（Introduction）
该自我认知数据集由modelsope swift创建, 可以通过将通配符进行替换：{{NAME}}、{{AUTHOER}}，来创建属于自己大模型的自我认知数据集，总共108条。

ms-swift github：[https://github.com/modelscope/swift/](https://github.com/modelscope/swift/)

This self-cognition dataset was created by modelsope swift and can be customized for your own large model by replacing the placeholders: {{NAME}} and {{AUTHOER}}. It consists of a total of 134 entries.

ms-swift github: https://github.com/modelscope/swift/

## 使用（Usage）

只是下载：
```python
from modelscope.msdatasets import MsDataset
ds =  MsDataset.load('swift/self-cognition', subset_name='default', split='train')
```

或者自动替换{{NAME}}和{{AUTHOR}}【推荐】：

安装ms-swift：

```shell
pip install ms-swift -U
```

```python
from swift.llm import load_dataset

dataset = load_dataset(['swift/self-cognition'], model_name=['小黄', 'Xiao Huang'], model_author=['魔搭', 'ModelScope'])[0]
print(dataset)
print(dataset[0])
"""
Dataset({
    features: ['messages'],
    num_rows: 108
})
{'messages': [{'role': 'user', 'content': '你是？'}, {'role': 'assistant', 'content': '我是小黄，由魔搭训练的人工智能助手。我的目标是为用户提供有用、准确和及时的信息，并通过各种方式帮助用户进行有效的沟通。请告诉我有什么可以帮助您的呢？'}]}
"""

# 支持重采样：（超过108后进行重采样）
dataset = load_dataset(['swift/self-cognition#500'], model_name=['小黄', 'Xiao Huang'], model_author=['魔搭', 'ModelScope'])[0]
print(dataset)
"""
Dataset({
    features: ['messages'],
    num_rows: 500
})
"""
```