NeurIPS'24 Exploring the Edges of Latent State Clusters for Goal-Conditioned Reinforcement Learning
PyTorch implementation of Hierarchical Actor Critic (HAC) for OpenAI gym environments
Ein RL-Agent lernt Super Mario Bros nur aus Pixeln und spielt es komplett durch: alle 32 Level geloest (Double DQN + PPO), je 20/20 greedy. Go-Explore, Behavior Cloning und Transfer Learning fuer Hard-Exploration, fuenf gefundene Reward-/Metrik-Fehler, Grad-CAM, Live-Demo, Docker/K8s + CI.
Go-Explore is a research project which applies the novel "Go-Explore" RL search algorithm to coding tasks.
Apple-Silicon Rust NES emulator + RL stack that beat Super Mario Bros end to end — cold boot to the princess, every input receipted — and trains policies graded by the strictest published evaluation protocol.
To associate your repository with the go-explore topic, visit your repo's landing page and select "manage topics."