Expand description
§rlx-rl
Compiled flow-map policies and twin critics for offline-to-online RL on RLX.
Re-exports§
pub use buffer::ReplayBuffer;pub use buffer::Transition;pub use dataset::OfflineDataset;pub use env::RlEnv;pub use flow_curriculum::sample_r_t;pub use graph::ActorGraphBundle;pub use graph::CompiledFlowMapAgent;pub use graph::CompiledTwinCritic;pub use graph::CriticGraphBundle;pub use graph::ParamSlot;pub use graph::WeightStore;pub use graph::build_actor_graphs;pub use graph::build_critic_graphs;pub use graph::init_actor_weights;pub use graph::init_critic_weights;pub use guidance::eta_effective;pub use guidance::q_guided_project;pub use policy::EvalConfig;pub use policy::sample_noise;pub use policy::select_action;pub use qgbs::QgbsConfig;pub use qgbs::qgbs_select_action;pub use spec::DistillationType;pub use spec::RlSpec;pub use trainer::FmqTrainer;