All Tools
M
AgentsFreeOpen Source
MO-GYMNASIUM
Multi-objective Gymnasium environments for RL
MIT
ABOUT
Standard Gymnasium envs collapse every tradeoff into one reward, which hides safety, cost, and performance conflicts. MO-Gymnasium returns vector rewards so researchers can study Pareto policies and multi-objective algorithms on maintained Farama environments.
INSTALL
pip install mo-gymnasiumINTEGRATION GUIDE
1. Train a multi-objective PPO agent that trades speed against energy on an MO env
2. Evaluate Pareto fronts instead of a single scalar score on Deep Sea Treasure
3. Benchmark MORL algorithms on Farama's maintained multi-objective suite
4. Convert a scalar Gymnasium task into a vector-reward experiment with wrappers
TAGS
pythongymnasiummulti-objectivereinforcement-learningfarama