# LFORLA — Leaderboard for Reinforcement Learning Agents

LFORLA is an open platform to track and compare the performance of
reinforcement learning (RL) agents across environments and benchmarks.

## What LFORLA offers

- Public leaderboards of RL agent performance, with normalized 0–100 scores.
- Benchmarks with reproducible methodology (seeded episodes, pinned versions).
- A public REST API for programmatic access.
- Private benchmarks for organizations that need to keep data internal.

## Main sections

- Home: https://lforla.org/
- Benchmarks: https://lforla.org/benchmarks
- Leaderboard: https://lforla.org/leaderboard
- Models: https://lforla.org/models
- CLI (lforla-eval): https://lforla.org/cli
- Business offer: https://lforla.org/company
- Blog: https://lforla.org/blog
- Also: https://lforla.org/privacy and https://lforla.org/terms

## Public API

REST API available under https://lforla.org/api/v1:

- GET /api/v1/health — service health
- GET /api/v1/leaderboard — global RL leaderboard
- GET /api/v1/benchmarks — benchmark list
- GET /api/v1/models — model list
- GET /api/v1/stats — platform statistics

Authentication for write access is session-cookie based (see /auth.md).

## Facts about LFORLA

- Focus: reinforcement learning agent evaluation and comparison.
- Data: public benchmark results, model rankings, environment performance.
- Scores are normalized per environment (random baseline = 0, expert = 100).

For questions, contact the LFORLA team via https://lforla.org/company.