Skip to content

Dataset · ankitjh4

Indian government reference documents

ankitjh4/bharat-government-documents

A cleaned corpus of official public Indian documents in parquet format for search and question-answering systems.

What's inside

Contains 64,964 unique normalized texts and 77,526 records from Indian government sources. Each row in parquet format includes extracted text, title, URL, extraction quality score, collection time, and topical tags.

What you can do with it

  • Training and testing RAG systems on legal and reference documents
  • Fine-tuning models for question-answering on public services
Tasks
Question answering, Text retrieval
Size
10K–100K rows
Downloads this month
486
Likes
65

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("ankitjh4/bharat-government-documents")