metadata

title: Leaderboard Explorer
emoji: 🔍
colorFrom: indigo
colorTo: purple
sdk: docker
hf_oauth: false
pinned: true
license: apache-2.0
app_port: 7860
tags:
  - explorer
short_description: Explore and discover all leaderboards from the HF community

Leaderboard Explorer

An application to explore and discover all leaderboards from the Hugging Face community.

Features

🔍 Browse leaderboards by category
🏷️ Filter by tags (code, math, vision, audio, etc.)
🎯 "Arena only" mode to display only human-evaluated leaderboards
🌓 Dark/Light mode
🔄 Automatic data updates

Data Source

The application fetches data from the leaderboard-explorer/leaderboard_explorer dataset on Hugging Face Hub. This dataset contains the consolidated leaderboard data in the final_leaderboards.json file.

Configuration

Environment Variables

Create a .env file in the server directory with the following variables:

# Hugging Face API token (required)
HF_TOKEN=your_token_here

# Hugging Face dataset repository ID
HF_REPO_ID=OpenEvals/find-a-leaderboard

# Server configuration
API_HOST=0.0.0.0
API_PORT=3002

The HF_TOKEN is required to access private datasets on Hugging Face. You can get your token from your Hugging Face account settings.

The HF_REPO_ID specifies which dataset repository to use. By default, it points to the main leaderboard explorer dataset.

Local Development

Install client dependencies:

cd client
npm install
npm run dev

Install server dependencies:

cd server
poetry install
poetry run dev

Deployment

The application is configured to be deployed on a Hugging Face Space using Docker.

When deploying, make sure to set the HF_TOKEN as a secret in your deployment environment.