Module traits 

Structs

ActionOutput

Action output from a policy.

DQNStepConfig

DQN step configuration.

EvalOutput

Output from evaluating a policy on (obs, actions) pairs.

MLPConfig

Configuration for building an MLP.

PPOStepConfig

PPO step configuration.

SACStepConfig

SAC step configuration.

TD3StepConfig

TD3 step configuration.

TrainMetrics

Training metrics dictionary.

Enums

Activation

Activation function.

Traits

ActorCritic

Actor-Critic policy for on-policy algorithms (PPO, A2C).

ContinuousQFunction

Continuous Q-function for SAC/TD3 (takes obs + action as input).

DeterministicPolicy

Deterministic policy for TD3.

EntropyTuner

Entropy tuning for SAC.

QFunction

Q-value network for off-policy algorithms (DQN).

StochasticPolicy

Continuous stochastic policy for SAC.