shinigamiRaj/AtharvaVedHindiSanskrit

📖 AtharvaVeda Corpus (Sharma Translation)

Languages: 🇮🇳 Hindi (hi) | 🕉️ Sanskrit (sa)
Dataset size: 1K < n < 10K records

---

Dataset Summary

This dataset contains a curated collection of the AtharvaVeda based on the translation and interpretive work by Dr Ganga Sahay Sharma. It presents verses in original Sanskrit (sa) along with Hindi (hi) rendering, preserving the linguistic and cultural heritage. The dataset is designed for research in Indology, computational linguistics, and dharmic traditions, as well as for training language models on Vedic literature.

Purpose: Facilitate study of Vedic texts, historical analysis, and cross-lingual learning between Sanskrit and Hindi. Suitable for low-resource language tasks, ancient text restoration, and linguistic pattern extraction.

---

Key Features

• Vedic Heritage: Authentic AtharvaVeda text with authoritative translation
• Bilingual Alignment: Sanskrit source with Hindi translation
• Optimized Chunking: Designed for RAG and fine-tuning workflows
• Cultural Preservation: Respectful representation of ancient Indian scriptures

---

Chunking & Preprocessing Parameters

Parameter	Value	Description
Chunk Overlap	150 characters	Maintains semantic continuity between segments
Chunk Size	4 chunks per page	Balanced for context length and granularity
Data Source	Dr Ganga Sahay Sharma's AtharvaVeda translation	Authoritative Hindi-Sanskrit rendering

The chunking strategy preserves natural verse boundaries (suktas and anuvakas) while ensuring optimal performance for downstream tasks like embedding generation and semantic search.

---

Data Structure

Each record in the dataset follows this schema:

{"text": "करता\n \nƔ ं\n, \nȵजस\n \nĒ कार\n \nवेग\n \nवाली\n \nनȰदयां\n \nबहती\n \nहƹ.\n \nहे\n \nसपµ\n! \nउन\nजड़ीबूȰटयǂ\n \nसे\n \nȱवष\n \nसमाËत\n \nहो\n \nजाए.\n (\n२१\n)\nयद¹नौ\n \nसूयƷ\n \nȱवषं\n \nपृȳथȅ ामोषधीषु\n \nयत्.\nकाÊदाȱवषं\n \nकनÊककं\n \nȱनरैÆवैतु\n \nते\n \nȱवषम्..\n (\n२२\n)\nअȷ¹न\n \nमƶ\n, \nसूयµ\n \nमƶ\n, \nपृÇवी\n \nमƶ\n \nतथा\n \nजड़ीबूȰटयǂ\n \nमƶ\n \nजो\n \nȱवष\n \nहै\n, \nवह\n \nतथा\n \nकं दǂ\n \nका\n \nȱवष\nपूणµतया\n \nनȌ\n \nहो\n \nजाए.\n (\n२२\n)\nये\n \nअȷ¹नजा\n \nओषȲधजा\n \nअहीनां\n \nये\n \nअËसुजा\n \nȱवȁ ुत\n \nआबभूवुः.\n******ebook converter DEMO Watermarks*******"}