manishiitg/aditi-syn-v2

---
language:
- hi
license: apache-2.0
dataset_info:
features:
- name: messages
list:
- name: content
dtype: string
- name: role
dtype: string
- name: type
dtype: string
splits:
- name: train
num_bytes: 201536052
num_examples: 55450
download_size: 86746197
dataset_size: 201536052
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---

This repo contains synthentic dataset in hindi/hinglish language using to fine tune aditi OOS LLM.

This has different type of data formats

1. TOOLS: teaching hindi/hinglish based function calling
2. RAG/RAG-Complex: teching context based rag for hindi/hinglish
3. CODE: writing code
4. ORCA: reasoning/math questions
5. COT: chain of thought reasoning
6. Prompts: Hindi/Hinglish answers on highly quality curated prompts
7. Instruct: general Q/A on indian context questions
8. Writing: general writing instructions
9. Instruct-Follow: generate system prompts to train model on instructions following
10. Roleplay: roleplaying on indian characters

The entire data generation pipeline can be seen at https://github.com/manishiitg/aditi_dataset

原始信息汇总

数据集概述

基本信息

• 语言: 印地语（Hindi）
• 许可证: Apache-2.0

数据集特征

• 特征名称: messages
  • 子特征:
    • 名称: content
      • 数据类型: 字符串
    • 名称: role
      • 数据类型: 字符串
• 特征名称: type
  • 数据类型: 字符串

数据集分割

• 分割名称: train
  • 数据大小: 201,536,052字节
  • 示例数量: 55,450

下载与数据集大小

• 下载大小: 86,746,197字节
• 数据集大小: 201,536,052字节

配置

• 配置名称: default
  • 数据文件:
    • 分割: train
      • 路径: data/train-*