border_core/
generic_replay_buffer.rs

1//! Generic implementation of replay buffers for reinforcement learning.
2//!
3//! This module provides a flexible and efficient implementation of replay buffers
4//! that can handle arbitrary observation and action types. It supports both
5//! standard experience replay and prioritized experience replay (PER).
6//!
7//! # Key Components
8//!
9//! - [`SimpleReplayBuffer`]: A generic replay buffer implementation
10//! - [`GenericTransitionBatch`]: A generic batch structure for transitions
11//! - [`SimpleStepProcessor`]: A processor for converting environment steps to transitions
12//! - [`PerConfig`]: Configuration for prioritized experience replay
13//!
14//! # Features
15//!
16//! - Generic type support for observations and actions
17//! - Efficient batch processing
18//! - Prioritized experience replay with importance sampling
19//! - Configurable weight normalization
20//! - Step processing for non-vectorized environments
21
22mod base;
23mod batch;
24mod config;
25mod step_proc;
26pub use base::{IwScheduler, SimpleReplayBuffer, WeightNormalizer};
27pub use batch::{BatchBase, GenericTransitionBatch};
28pub use config::{PerConfig, SimpleReplayBufferConfig};
29pub use step_proc::{SimpleStepProcessor, SimpleStepProcessorConfig};