afrizalha/Tumpeng-1-Indonesian

---
license: cc-by-nc-4.0
task_categories:
- text-generation
language:
- id
size_categories:
- 10K<n<100K
---

# Synthetic Indonesian dataset with Llama 3 70B

Tumpeng contains 14.8M words of 48.6K input-output pairs of Indonesian question-answering. It is intended to fine-tune Llama 3 8B, which has limited Indonesian language capabilities, to properly respond in Indonesian.

It is a research preview dataset and not curated for factual accuracy or safety. Use this dataset at your discretion.

# Out of scope use
- Commercial use
- Fine-tuning non-Llama 3 models

# Supported tasks
- General QA across a variety of domains
- Contextual QA (copy-paste a text and ask a question about the contents in the text)
- Multi-turn conversation (alpha; optimized for personal advice)
- Writing (outlines and full-articles)

# Conversational format
This dataset contains experimental multi-turn conversation, with <|user|> and <|assistant|> as the user and LLM headers respectively. For proper formatting, please use the following template:
```
<|user|>
{promt}

<|assistant|>
{response}
```
Alternatively, you can modify the strings in the dataset according to your intended format.

原始信息汇总

数据集概述

基本信息

• 名称: Synthetic Indonesian dataset with Llama 3 70B
• 语言: 印尼语（id）
• 许可: CC-BY-NC-4.0
• 大小: 10K<n<100K

数据内容

• 包含: 14.8M 单词，48.6K 输入-输出对
• 用途: 用于微调 Llama 3 8B 模型，以提升其在印尼语环境下的响应能力

数据特点

• 类型: 研究预览数据集，未经过事实准确性或安全性筛选
• 使用限制: 非商业用途，仅限微调 Llama 3 模型

支持的任务

• 通用领域问答
• 上下文问答
• 多轮对话（优化于个人建议）
• 写作（大纲和完整文章）

格式

• 对话格式: 包含实验性多轮对话，使用 <|user|> 和 <|assistant|> 分别作为用户和语言模型的标识

• 模板:

  <|user|> {promt}

  <|assistant|> {response}