Module dqn 

DQN trainer (Burn backend).

After phase 5 of the Burn migration (#82), Burn is the only tensor backend in the workspace. The historical train::dqn_burn parallel sibling has been collapsed back into train::dqn. See the parallel crate::train::ppo module for the same rationale.

Contents

• config — DQNConfig hyperparameters / builder API.
• loss — backend-generic DQN / Double-DQN loss math.
• trainer — DQNTrainerBurn<B, Q, O> that owns the online Q-network module (Burn’s optimizer-consumes-module ownership model) and exposes a train_step that runs Smooth-L1 loss / gradient-step logic.

Re-exports

pub use config::DQNConfig;

pub use loss::compute_dqn_loss;

pub use loss::compute_dqn_loss_double;

pub use loss::compute_loss;

pub use loss::compute_td_target;

pub use loss::compute_td_target_double;

pub use loss::gather_action_q;

pub use loss::huber_per_sample;

pub use trainer::DQNStepStatsBurn;

pub use trainer::DQNTrainerBurn;

Modules

config

DQN configuration and hyperparameters

loss

DQN loss math (Burn backend).

trainer

Burn-backend DQN trainer (phase 3 of the Burn migration, #80).