mrgehlot/rajarshi-ranjan-nandy-adhyatmikta

Dataset Card for Rajarshi Ranjan Nandy - Adhyatmikta Training Dataset

Dataset Description

This dataset contains training data extracted from "Adhyatmikta: Explorations into Hindu Spiritual Practices" by Rajarshi Ranjan Nandy, published in 2023 by BluOne Ink LLP. The dataset is formatted for fine-tuning language models to learn the author's writing style and knowledge about Sanatana Dharma (Hindu spiritual practices).

Dataset Summary

• Total Examples: 861 training examples
• Format: JSON with system prompts, inputs, and outputs
• Purpose: Fine-tuning for style transfer and knowledge learning
• Source: PDF extraction from "Adhyatmikta: Explorations into Hindu Spiritual Practices"

Supported Tasks and Leaderboards

This dataset is designed for:

• Text Generation: Continuation-based generation in the author's style
• Style Transfer: Learning to convert text into a specific writing style related to Sanatana Dharma
• Fine-tuning: Training language models to adopt the author's writing tone and knowledge

Dataset Structure

Data Instances

Each example in the dataset follows this structure:

{
  "system": "You are Bhairava's voice. a wise, fearless, Sanatana Dharma assistant who speaks with deep spiritual insight, poetic rhythm, and the style, vocabulary, and metaphors found in the text you have learned.",
  "input": "",
  "output": "Text content from the book..."
}

Data Fields

• system (string): System prompt instructing the model to convert text into the style of a Sanatana Dharma assistant
• input (string): User input (typically empty for continuation-based training)
• output (string): Text content extracted from the book, preserving the original style and formatting

Data Splits

The dataset is provided as a single training split with 861 examples.

Dataset Creation

Source Data

• Book Title: Adhyatmikta: Explorations into Hindu Spiritual Practices
• Author: Rajarshi Ranjan Nandy
• Publisher: BluOne Ink LLP
• ISBN: 978-93-92209-40-6
• Publication Year: 2023
• Location: Noida, India

Preprocessing

The dataset was created by:

1. Extracting text from the PDF version of the book
2. Splitting the text into sentence bundles (2 sentences per bundle)
3. Formatting each bundle as a training example with system prompts
4. Preserving the original writing style and formatting

Annotations

The dataset uses a consistent system prompt across all examples to guide the model in learning the specific writing style and perspective related to Sanatana Dharma.

Considerations for Using the Data

Social Impact of Dataset

This dataset focuses on Hindu spiritual practices and Sanatana Dharma. Users should be aware that:

• The content reflects the author's perspective and interpretations
• The dataset is intended for educational and research purposes
• The style and content are specific to the author's writing approach

Discussion of Biases

• The dataset represents a single author's perspective on Hindu spirituality
• The writing style and terminology are specific to this work
• Users should consider the cultural and religious context when using this dataset

Other Known Limitations

• The dataset is extracted from a single source (one book)
• Text extraction from PDF may have minor formatting artifacts
• The dataset size (861 examples) is relatively small for large-scale training

Citation Information

If you use this dataset, please cite:

@book{nandy2023adhyatmikta,
  title={Adhyatmikta: Explorations into Hindu Spiritual Practices},
  author={Rajarshi Ranjan Nandy},
  year={2023},
  publisher={BluOne Ink LLP},
  isbn={978-93-92209-40-6}
}

@misc{gehlot2025dataset,
  title={Adhyatmikta Dataset},
  author={Abhishek Gehlot},
  year={2025},
  publisher={Gehlot Corporation}
}

License

Copyright © 2023 Rajarshi Ranjan Nandy. All rights reserved under the copyright conventions.

This dataset is provided for research and educational purposes. Please refer to the original book's copyright notice for usage terms.

Contact

For questions about the dataset or the source material:

• Publisher: BluOne Ink LLP
• Website: www.bluone.ink
• Email: publisher@bluone.ink

Dataset Usage

Loading the Dataset

from datasets import load_dataset

dataset = load_dataset("mrgehlot/rajarshi-ranjan-nandy-adhyatmikta")

Example Usage

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("mrgehlot/rajarshi-ranjan-nandy-adhyatmikta")

# Access training examples
train_data = dataset['train']
print(train_data[0])

# Example output:
# {
#   'system': 'You are a helpful assistant of sanatana dharma...',
#   'input': '',
#   'output': 'adhyatmikta EXPLORATIONS INTO HINDU SPIRITUAL PRACTICES...'
# }

Fine-tuning Example

This dataset is suitable for fine-tuning language models using frameworks like:

• Hugging Face Transformers
• PEFT (Parameter-Efficient Fine-Tuning)
• LoRA (Low-Rank Adaptation)

The dataset format is compatible with instruction-following and continuation-based training approaches.