in-progressStarted: February 01, 2026
Reinforcement Learning with Gymnasium
#ai#reinforcement-learning#python
I’m currently working through the Gymnasium documentation and implementing classic Reinforcement Learning algorithms from scratch (and with libraries like Stable Baselines3).
Current Focus
- Algorithm: Proximal Policy Optimization (PPO)
- Environment: LunarLander-v2 and CartPole-v1
It is surprisingly challenging to tune the hyperparameters. The agent often gets stuck in local optima. I plan to write a full blog post breakdown of my PPO implementation once I get it consistently solving the harder environments.
