Struct AsyncPolicy 

pub struct AsyncPolicy<P>
where
    P: Policy,
{ /* private fields */ }

An asynchronous policy using an inference server with autobatching.

Implementations

impl<P> AsyncPolicy<P>

where P: Policy + Clone + Send + 'static, <P as Policy>::ActionContext: Clone + Send, <P as Policy>::PolicyState: Send, <P as Policy>::Observation: Clone + Send + Batchable, <P as Policy>::ActionDistribution: Clone + Send + Batchable, <P as Policy>::Action: Clone + Send + Batchable,

pub fn new(autobatch_size: usize, inner_policy: P) -> AsyncPolicy<P>

Create the policy.

Arguments

• autobatch_size - Number of observations to accumulate before running a pass of inference.
• inner_policy - The policy used to take actions.

pub fn increment_agents(&self, num: usize)

Increment the number of agents using the inference server.

pub fn decrement_agents(&self, num: usize)

Decrement the number of agents using the inference server.

Trait Implementations

impl<P> Clone for AsyncPolicy<P>

where P: Clone + Policy,

fn clone(&self) -> AsyncPolicy<P>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<P> Policy for AsyncPolicy<P>

where P: Policy + Send + 'static,

type ActionContext = <P as Policy>::ActionContext

Additional context on the policy’s decision.

type PolicyState = <P as Policy>::PolicyState

The current parameterization of the policy.

type Observation = <P as Policy>::Observation

The observation given as input to the policy.

type ActionDistribution = <P as Policy>::ActionDistribution

The action distribution parameters defining how the action will be sampled.

type Action = <P as Policy>::Action

The action.

fn forward( &mut self, states: <AsyncPolicy<P> as Policy>::Observation, ) -> <AsyncPolicy<P> as Policy>::ActionDistribution

Produces the action distribution from a batch of observations.

fn action( &mut self, states: <AsyncPolicy<P> as Policy>::Observation, deterministic: bool, ) -> (<AsyncPolicy<P> as Policy>::Action, Vec<<AsyncPolicy<P> as Policy>::ActionContext>)

Gives the action from a batch of observations.

fn update(&mut self, update: <AsyncPolicy<P> as Policy>::PolicyState)

Update the policy’s parameters.

fn state(&self) -> <AsyncPolicy<P> as Policy>::PolicyState

Returns the current parameterization.

fn to_device(self, device: &Device) -> AsyncPolicy<P>

Loads the policy on the given device.

fn load_record( self, _record: <<AsyncPolicy<P> as Policy>::PolicyState as PolicyState>::Record, ) -> AsyncPolicy<P>

Loads the policy parameters from a record.