Safety-Gymnasium

The 'Safety-Gymnasium' is a benchmark that evaluates the performance of algorithms in constrained environments, focusing on safety and fairness in automated decision-making processes.

Papers using Safety-Gymnasium (12)

Model-Based Proactive Cost Generation for Learning Safe Policies Offline with Limited Violation Data2026Constrained Policy Optimization for Provably Fair Order Matching2026Lyapunov-Guided Self-Alignment: Test-Time Adaptation for Offline Safe Reinforcement Learning2026Safe Langevin Soft Actor Critic2026Epigraph-Guided Flow Matching for Safe and Performant Offline Reinforcement Learning2026Nightmare Dreamer: Dreaming About Unsafe States And Planning Ahead2026SB-TRPO: Towards Safe Reinforcement Learning with Hard Constraints2025KFCPO: Kronecker-Factored Approximated Constrained Policy Optimization2025Provably Optimal Reinforcement Learning under Safety Filtering2025Optimistic Exploration for Risk-Averse Constrained Reinforcement Learning2025Human-Aligned Skill Discovery: Balancing Behaviour Exploration and Alignment2025Adversarial Constrained Policy Optimization: Improving Constrained Reinforcement Learning by Adapting Budgets2024