PyTorch reimplementation of "Deep Hierarchical Planning" RL framework. Features a multi-model architecture with manager-worker policies, world model, and goal autoencoder. Built with Python/PyTorch and includes experiment logging with Weights & Biases.
python machine-learning reinforcement-learning pytorch deeplearning research-paper world-models rl-framework neurips-2022 hierarchical-rl manager-worker goal-autoencoder
-
Updated
Sep 10, 2025 - Python