vjdevane's picture
Update README.md
35bc05b verified
metadata
title: BhashaBench Leaderboard
emoji: πŸ†
colorFrom: blue
colorTo: red
sdk: docker
hf_oauth: true
pinned: true
license: apache-2.0
duplicated_from: open-llm-leaderboard/open_llm_leaderboard
short_description: Evaluating LLMs on BhashaBench tasks
tags:
  - leaderboard
  - modality:text
  - submission:manual
  - test:public
  - judge:function
  - eval:generation
  - language:English, Hindi
  - domain:Ayur, Krishi, Finance, Legal

https://arxiv.org/abs/2510.25409

Open LLM Leaderboard

Modern React interface for comparing Large Language Models (LLMs) in an open and reproducible way.

Features

  • πŸ“Š Interactive table with advanced sorting and filtering
  • πŸ” Semantic model search
  • πŸ“Œ Pin models for comparison
  • πŸ“± Responsive and modern interface
  • 🎨 Dark/Light mode
  • ⚑️ Optimized performance with virtualization

Architecture

The project is split into two main parts:

Frontend (React)

frontend/
β”œβ”€β”€ src/
β”‚   β”œβ”€β”€ components/     # Reusable UI components
β”‚   β”œβ”€β”€ pages/         # Application pages
β”‚   β”œβ”€β”€ hooks/         # Custom React hooks
β”‚   β”œβ”€β”€ context/       # React contexts
β”‚   └── constants/     # Constants and configurations
β”œβ”€β”€ public/            # Static assets
└── server.js          # Express server for production

Backend (FastAPI)

backend/
β”œβ”€β”€ app/
β”‚   β”œβ”€β”€ api/           # API router and endpoints
β”‚   β”‚   └── endpoints/ # Specific API endpoints
β”‚   β”œβ”€β”€ core/          # Core functionality
β”‚   β”œβ”€β”€ config/        # Configuration
β”‚   └── services/      # Business logic services
β”‚       β”œβ”€β”€ leaderboard.py
β”‚       β”œβ”€β”€ models.py
β”‚       β”œβ”€β”€ votes.py
β”‚       └── hf_service.py
└── utils/             # Utility functions

Technologies

Frontend

  • React
  • Material-UI
  • TanStack Table & Virtual
  • Express.js

Backend

  • FastAPI
  • Hugging Face API
  • Docker

Development

The application is containerized using Docker and can be run using:

docker-compose up

Citation

Please cite our benchmark if used in your work:

@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}, 
}