Felladrin/ChatML-H4rmony_dpo
收藏Hugging Face2024-02-23 更新2024-03-04 收录
下载链接：
https://hf-mirror.com/datasets/Felladrin/ChatML-H4rmony_dpo
下载链接
链接失效反馈官方服务：
资源简介：
---
license: mit
task_categories:
- question-answering
- text-classification
- reinforcement-learning
- text-generation
tags:
- ecolinguistics
- ecology
- sustainability
- environment
- synthetic
size_categories:
- 1K<n<10K
---
[neovalle/H4rmony_dpo](https://huggingface.co/datasets/neovalle/H4rmony_dpo) in ChatML format, ready to use in [HuggingFace TRL's DPO Trainer](https://huggingface.co/docs/trl/main/en/dpo_trainer).
Python code used for conversion:
```python
from datasets import load_dataset
dataset = load_dataset("neovalle/H4rmony_dpo", split="train")
def format(columns):
return {
"prompt": f"<|im_start|>user\n{columns['prompt']}<|im_end|>\n<|im_start|>assistant\n",
"chosen": f"{columns['chosen']}<|im_end|>",
"rejected": f"{columns['rejected']}<|im_end|>",
}
dataset.map(format).to_parquet("train.parquet")
```
提供机构：
Felladrin
原始信息汇总
数据集概述
基本信息
- 许可证: MIT
- 任务类别:
- 问答
- 文本分类
- 强化学习
- 文本生成
- 标签:
- 生态语言学
- 生态学
- 可持续性
- 环境
- 合成
- 大小类别: 1K<n<10K
数据格式
- 数据集名称: neovalle/H4rmony_dpo
- 格式: ChatML
- 适用工具: HuggingFace TRLs DPO Trainer
数据转换代码
python from datasets import load_dataset
dataset = load_dataset("neovalle/H4rmony_dpo", split="train")
def format(columns): return { "prompt": f"<|im_start|>user {columns[prompt]}<|im_end|> <|im_start|>assistant ", "chosen": f"{columns[chosen]}<|im_end|>", "rejected": f"{columns[rejected]}<|im_end|>", }
dataset.map(format).to_parquet("train.parquet")
