Struct DqnRunner 

pub struct DqnRunner<E, Enc, Act, B>
where
    E: Environment,
    B: AutodiffBackend,
{ /* private fields */ }

The imperative training runner.

Drives the interaction between an environment and a DQN agent, exposing it as an iterator that yields StepMetrics after every step.

Usage

let mut runner = DqnRunner::new(env, agent, seed);

for step in runner.steps().take(50_000) {
    if step.episode_done {
        println!("Episode {} reward: {}", step.episode, step.episode_reward);
    }
}

Why an iterator?

• You control the loop: add early stopping, custom logging, checkpointing
• bevy-gym can drive the same runner one step per ECS tick
• No callbacks, no closures, no inversion of control

Implementations

impl<E, Enc, Act, B> DqnRunner<E, Enc, Act, B>

where E: Environment, E::Observation: Clone + Send + Sync + 'static, E::Action: Clone + Send + Sync + 'static, Enc: ObservationEncoder<E::Observation, B> + ObservationEncoder<E::Observation, B::InnerBackend>, Act: DiscreteActionMapper<E::Action>, B: AutodiffBackend,

pub fn new(env: E, agent: DqnAgent<E, Enc, Act, B>, seed: u64) -> Self

pub fn steps(&mut self) -> StepIter<'_, E, Enc, Act, B>

Returns an iterator that yields StepMetrics after each environment step.

pub fn agent(&self) -> &DqnAgent<E, Enc, Act, B>

Access the agent for evaluation or inspection.

pub fn env(&self) -> &E

Access the environment directly.