---
configs:
  - config_name: English
    data_files:
      - path: English/test-*
        split: test
  - config_name: Hindi
    data_files:
      - path: Hindi/test-*
        split: test
language:
  - en
  - hi
license: cc-by-4.0
task_categories:
  - multiple-choice
  - question-answering
size_categories:
  - 10K<n<100K
source_datasets:
  - original
pretty_name: test_data
---

<div align="center">
  <img src="https://huggingface.co/bharatgenai/Param-1-2.9B-Instruct/resolve/main/BharatGen%20Logo%20(1).png" width="60%" alt="BharatGen" />
</div>
<hr>
<div align="center">
  <a href="https://arxiv.org/abs/2510.25409" style="margin: 4px;">
    <img alt="Paper" src="https://img.shields.io/badge/arXiv-2510.25409-b31b1b?style=flat" />
  </a>
  <a href="https://creativecommons.org/licenses/by/4.0/" target="_blank" style="margin: 4px;">
    <img alt="License" src="https://img.shields.io/badge/License-CC--BY--4.0-blue.svg" />
  </a>
  <a href="#" target="_blank" style="margin: 4px;">
    <img alt="Blog" src="https://img.shields.io/badge/Blog-Read%20More-brightgreen?style=flat" />
  </a>
</div>

# BhashaBench-Finance (BBF): Benchmarking AI on Indian Financial Knowledge

## Overview
BhashaBench-Finance (BBF) is the first comprehensive benchmark designed to evaluate AI models on Indian financial knowledge and practices. Tailored for India's diverse financial ecosystem, regulatory framework, and economic landscape, BBF draws from 25+ official government and institutional financial exams to assess models' ability to provide accurate, policy-compliant, and contextually relevant financial advice and analysis.

## Key Features

- **Languages**: English and Hindi (with plans for more Indic languages)
- **Exams**: 25+ unique financial government and institutional exams across India
- **Domains**: 30+ financial and allied domains, spanning comprehensive financial knowledge
- **Questions**: 19,433 rigorously validated, exam-based questions
- **Difficulty Levels**: Easy (7,111), Medium (9,348), Hard (2,974)
- **Question Types**: Multiple Choice, Assertion-Reasoning, Match the Column, Rearrange the Sequence, Fill in the Blanks, Reading Comprehension, Essay
- **Focus**: Practical, regulation-aware, India-specific financial knowledge essential for financial professionals and consumers

## Dataset Statistics

| Metric                   | Count                     |
| ------------------------ | ------------------------- |
| Total Questions          | 19,433                    |
| English Questions        | 13,451                    |
| Hindi Questions          | 5,982                     |
| Subject Domains          | 30+                       |
| Government Exams Covered | 25+                       |

## Dataset Structure

### Test Set
The test set consists of the BhashaBench-Finance (BBF) benchmark, which contains approximately 19,433 questions across 2 Indic languages (English and Hindi).  
We will add support for more Indic languages in upcoming versions.

### Subject Domains in BBF

| Subject Domain                                | Count |
|---------------------------------------------|-------|
| Problem Solving                             | 5,686 |
| Mathematics for Finance                     | 4,845 |
| Banking Services                           | 1,171 |
| Governance & Policy                        | 1,064 |
| Language & Communication                   | 946   |
| Corporate Finance & Investment             | 910   |
| Commerce                                   | 863   |
| Accounting                                 | 773   |
| General Knowledge                          | 539   |
| Information Technology Finance             | 490   |
| Economics & Development Studies            | 274   |
| Rural Economics                            | 261   |
| Environmental Finance                      | 168   |
| Taxation & Regulatory Compliance           | 155   |
| Interdisciplinary Finance                  | 153   |
| Data & Analytics in Finance                | 127   |
| History, Sociology & Cultural Studies of Finance | 127 |
| Finance Education                          | 118   |
| Healthcare Economics                       | 114   |
| Science and Technology in Finance          | 101   |
| International Finance & Trade              | 83    |
| Business Management                        | 83    |
| Energy, Infrastructure & Finance           | 82    |
| Behavioral Finance                         | 67    |
| Financial Markets                          | 47    |
| Sports, Media & Finance Linkages           | 45    |
| Marketing Finance                          | 42    |
| Insurance & Risk Management                | 42    |
| Legal Finance                              | 34    |
| Financial Technology                       | 23    |

### Question Distribution

#### By Difficulty Level
| Level  | Count | Percentage |
|--------|-------|------------|
| Easy   | 7,111 | 36.6%      |
| Medium | 9,348 | 48.1%      |
| Hard   | 2,974 | 15.3%      |

#### By Question Type
| Type                    | Count  | Percentage |
|------------------------|--------|------------|
| MCQ                    | 18,019 | 92.7%      |
| Rearrange the sequence | 708    | 3.6%       |
| Fill in the blanks     | 286    | 1.5%       |
| Assertion or Reasoning | 215    | 1.1%       |
| Match the column       | 119    | 0.6%       |
| Reading Comprehension  | 86     | 0.4%       |

## Usage

Since this is a gated dataset, after your request for accessing the dataset is accepted, you can set your HuggingFace token:

```bash
export HF_TOKEN=YOUR_TOKEN_HERE
```

To load the BBF dataset for a Language:

```python
from datasets import load_dataset
language = 'Hindi'
# Use 'test' split for evaluation
split = 'test'
language_data = load_dataset("bharatgenai/BhashaBench-Finance", data_dir=language, split=split, token=True)
print(language_data[0])
```

## Evaluation Results Summary

- **40+ models evaluated**, including GPT-based models, Gemma, Llama, Qwen, and specialized financial AI systems.

- **Language Performance Gap:**
  - Consistent 5-7% performance drop from English to Hindi across most models
  - Best Hindi performance still significantly below English benchmarks

- **Domain Performance Insights:**
  - **Strongest domains (70%+ by top models):**
    - Information Technology Finance (92.24%)
    - Business Management (84.34%)
    - History, Sociology & Cultural Studies of Finance (83.46%)
  - **Most challenging domains (<50% by top models):**
    - Problem Solving (47.12%)
    - Mathematics for Finance (58.04%)
    - Legal Finance (76.47%)

- **Difficulty Level Analysis:**
  - **Easy questions:** Top models achieve 72-73% accuracy
  - **Medium questions:** Performance drops to 59% for best models
  - **Hard questions:** Significant challenge with only 40-41% accuracy

- **Question Type Performance:**
  - **Best:** Fill in the blanks (81.82% top performance)
  - **Moderate:** MCQ (61.7%), Assertion/Reasoning (67.91%)
  - **Challenging:** Rearrange sequence (49.01%), Match column (69.75%)

- **Model Size vs Performance:**
  - Larger models (Qwen3-235B, DeepSeek-v3) significantly outperform smaller models
  - Specialized financial fine-tuning shows mixed results compared to general large models

For detailed results and analysis, please refer to our [blog](#).

## Applications

BBF enables evaluation of AI systems for:
- **Financial advisory services** in Indian context
- **Banking and insurance** customer support systems
- **Regulatory compliance** automation tools
- **Financial education** platforms for Indian markets
- **Investment analysis** systems understanding Indian financial instruments

## Citation

Please cite our benchmark if used in your work:

```bibtex
@misc{devane2025bhashabenchv1comprehensivebenchmark,
      title={BhashaBench V1: A Comprehensive Benchmark for the Quadrant of Indic Domains}, 
      author={Vijay Devane and Mohd Nauman and Bhargav Patel and Aniket Mahendra Wakchoure and Yogeshkumar Sant and Shyam Pawar and Viraj Thakur and Ananya Godse and Sunil Patra and Neha Maurya and Suraj Racha and Nitish Kamal Singh and Ajay Nagpal and Piyush Sawarkar and Kundeshwar Vijayrao Pundalik and Rohit Saluja and Ganesh Ramakrishnan},
      year={2025},
      eprint={2510.25409},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2510.25409}, 
}
```

## License

This dataset is released under the [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/).

## Contact

For any questions or feedback, please contact:
- Vijay Devane (vijay.devane@tihiitb.org)
- Mohd. Nauman (mohd.nauman@tihiitb.org)
- Bhargav Patel (bhargav.patel@tihiitb.org)
- Kundeshwar Pundalik (kundeshwar.pundalik@tihiitb.org)

## Links

- [GitHub Repository 💻](https://github.com/BharatGen-IITB-TIH/BhashaBench)
- [Paper 📄](https://arxiv.org/abs/2510.25409)