Modulesยง
- augmentation
- Image augmentation for visual RL (DrQ-v2 style random shift).
- gae
- kl
- normalization
- packing
- reward_
shaping - Potential-based reward shaping (PBRS) and goal-distance potentials.
- simd_
ops - SIMD-accelerated operations for training hot loops.
- vtrace
- weight_
ops - Weight vector operations for meta-learning and target network updates.