Datasets:

munmuni
/

DharmaBench_v1

license: cc-by-4.0
language:
  - bn
  - en
pretty_name: DharmaBench v1 - Bengali Spiritual QA Dataset
task_categories:
  - question-answering
tags:
  - spiritual-qa
  - bengali
  - ramakrishna
  - sarada-devi
  - vivekananda
  - dharma
size_categories:
  - n<100
dataset_info:
  features:
    - name: id
      dtype: large_string
    - name: question
      dtype: large_string
    - name: answer
      dtype: large_string
    - name: speaker
      dtype: large_string
    - name: theme
      dtype: large_string
    - name: emotion
      dtype: large_string
    - name: source_book
      dtype: large_string
    - name: page_no
      dtype: int64
    - name: context
      dtype: large_string
  splits:
    - name: train
      num_bytes: 15749
      num_examples: 50
  download_size: 10243
  dataset_size: 15749
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

DharmaBench v1

DharmaBench v1 is a Bengali spiritual Question-Answering dataset. It contains teachings and quotes from Sri Ramakrishna, Sri Sarada Devi, Swami Vivekananda, and other holy personalities.

Dataset Description

This dataset is created to build AI models that can answer people's life problems with spiritual wisdom from Bengali sources.

Dataset Structure

The dataset has 50 examples with the following columns:

id: Unique ID for each example. Ex: SD_001
question: A person's problem or question in Bengali.
answer: The spiritual answer/quote in Bengali.
speaker: Who said this? Ex: Sarada, Ramakrishna, Vivekananda
theme: The topic. Ex: Mental_Peace, Sadhana, Karma_Yoga
emotion: The emotion of the question. Ex: Anxiety, Sadness, Confusion
source_book: The name of the book it is taken from. Ex: মায়ের কথা

Data Example

id	question	answer	speaker	theme	emotion	source_book
SD_001	মন খুব অশান্ত, শান্তি পাচ্ছি না	মনকে বলো, তুই তো আমার। আমি তো...	Sarada	Mental_Peace	Anxiety	মায়ের কথা

Citation

If you use this dataset, please cite: Munmuni. (2026). DharmaBench v1 [Dataset]. Hugging Face.