oxicuda_rl/buffer/mod.rs
1//! Experience replay buffers for off-policy RL algorithms.
2//!
3//! Provides three implementations:
4//! * [`crate::buffer::UniformReplayBuffer`] — uniform random sampling (DQN, SAC, TD3)
5//! * [`crate::buffer::PrioritizedReplayBuffer`] — proportional PER with segment
6//! tree sampling (PER-DQN, PER-SAC)
7//! * [`crate::buffer::NStepBuffer`] — n-step return accumulation with discount rollout
8
9pub mod n_step;
10pub mod prioritized;
11pub mod replay;
12
13// Re-exports
14pub use n_step::{NStepBuffer, NStepTransition};
15pub use prioritized::{PrioritizedReplayBuffer, PrioritySample};
16pub use replay::{Transition, UniformReplayBuffer};