Struct border_core::DefaultEvaluator

pub struct DefaultEvaluator<E: Env, P: Policy<E>> { /* private fields */ }

A default Evaluator.

This struct runs episodes of the given number of times.

Implementations

impl<E, P> DefaultEvaluator<E, P>where E: Env, P: Policy<E>,

pub fn new(config: &E::Config, seed: i64, n_episodes: usize) -> Result<Self>

Constructs DefaultEvaluator.

config - Configuration of the environment. seed - Random seed, which will be used to create the environment. n_episodes - The number of episodes for evaluation. The evaluator returns the mean value of cumulative reward in each episode.

Trait Implementations

impl<E, P> Evaluator<E, P> for DefaultEvaluator<E, P>where E: Env, P: Policy<E>,

fn evaluate(&mut self, policy: &mut P) -> Result<f32>

Evaluate Policy.