tulu-3-sft-personas-instruction-following
收藏魔搭社区2026-04-28 更新2024-11-30 收录
下载链接:
https://modelscope.cn/datasets/LLM-Research/tulu-3-sft-personas-instruction-following
下载链接
链接失效反馈官方服务:
资源简介:
<img src="https://huggingface.co/datasets/allenai/blog-images/resolve/main/tulu-3/Tulu3-logo.png" alt="Tulu3 banner" width="400" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
### Dataset Descriptions
This dataset contains **29980** examples and is synthetically created to enhance model's capabilities to follow instructions precisely and to satisfy user constraints. The constraints are borrowed from the taxonomy in [IFEval dataset](https://arxiv.org/abs/2311.07911).
To generate diverse instructions, we expand the methodology in [Ge et al., 2024](https://arxiv.org/pdf/2406.20094) by using personas. More details and exact prompts used to construct the dataset can be found in our [paper]().
- **Curated by:** Allen Institute for AI
- **Paper:** [TBD]()
- **Repository:** [TBD]()
- **Language(s) (NLP):** English
- **License:** ODC-BY
- **Point of Contact:** [Faeze Brahman](mailto:faezeb@allenai.org)
### Loading
```python
from datasets import load_dataset
dataset = load_dataset("allenai/tulu-3-sft-personas-instruction-following")["train"]
```
### Dataset Structure
Each example in the dataset contains the standard instruction-tuning data points as follow:
- id (str): a unique identifier
- prompt (str): the verifiable instruction which involves satisfying 1 to 3 constraints
- messages (list): message format used for supervised fine-tuning (this contains user prompt and assistant response)
- constraints (list of str): a list of verifiable constraints that need to be satisfied by the assistant response
<img src="https://huggingface.co/datasets/allenai/blog-images/resolve/main/tulu-3/Tulu3-logo.png" alt="Tulu3 横幅" width="400" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
### 数据集描述
本数据集共包含**29980**条样本,为合成生成的数据集,旨在提升模型精准遵循指令、满足用户约束条件的能力。本次采用的约束规则源自[IFEval数据集](https://arxiv.org/abs/2311.07911)的分类体系。
为生成多样化的指令,我们在[Ge等人,2024](https://arxiv.org/pdf/2406.20094)的研究方法基础上进行扩展,引入了人物设定(personas)。有关数据集构建的更多细节与具体提示词,可参见我们的[论文]()。
- **精选出品方:** 艾伦人工智能研究所(Allen Institute for AI)
- **论文:** [待发布]()
- **代码仓库:** [待发布]()
- **自然语言处理(Natural Language Processing,NLP)所用语言:** 英语
- **许可证:** ODC-BY
- **联系方式:** [Faeze Brahman](mailto:faezeb@allenai.org)
### 加载方式
python
from datasets import load_dataset
dataset = load_dataset("allenai/tulu-3-sft-personas-instruction-following")["train"]
### 数据集结构
本数据集的每条样本均遵循标准的指令微调数据格式,具体字段如下:
- `id`(字符串类型):唯一标识符
- `prompt`(字符串类型):需满足1至3项约束的可验证指令
- `messages`(列表类型):用于监督微调(supervised fine-tuning)的消息格式(包含用户提示与助手回复)
- `constraints`(字符串列表):需由助手回复满足的可验证约束列表
提供机构:
maas
创建时间:
2024-11-23
搜集汇总
数据集介绍

背景与挑战
背景概述
该数据集包含29980个合成示例,旨在提升模型精确遵循指令和满足用户约束的能力,约束基于IFEval分类法,并通过角色扩展方法生成多样指令。它由Allen Institute for AI策划,语言为英语,采用ODC-BY许可证。
以上内容由遇见数据集搜集并总结生成



