border_tch_agent::sac

Struct Sac

pub struct Sac<E, Q, P, R>where
    Q: SubModel2<Output = Tensor>,
    P: SubModel<Output = (Tensor, Tensor)>,
    Q::Config: DeserializeOwned + Serialize + OutDim + Debug + PartialEq + Clone,
    P::Config: DeserializeOwned + Serialize + OutDim + Debug + PartialEq + Clone,
{ /* private fields */ }

Expand description

Soft actor critic (SAC) agent.

Implementations§

impl<E, Q, P, R> Sac<E, Q, P, R>
where E: Env, Q: SubModel2<Output = Tensor>, P: SubModel<Output = (Tensor, Tensor)>, R: ReplayBufferBase, E::Obs: Into<Q::Input1> + Into<P::Input>, E::Act: Into<Q::Input2>, Q::Input2: From<Tensor>, Q::Config: DeserializeOwned + Serialize + OutDim + Debug + PartialEq + Clone, P::Config: DeserializeOwned + Serialize + OutDim + Debug + PartialEq + Clone, R::Batch: TransitionBatch, <R::Batch as TransitionBatch>::ObsBatch: Into<Q::Input1> + Into<P::Input> + Clone, <R::Batch as TransitionBatch>::ActBatch: Into<Q::Input2> + Into<Tensor>,

pub fn get_policy_net(&self) -> &Actor<P>

Trait Implementations§

impl<E, Q, P, R> Agent<E, R> for Sac<E, Q, P, R>
where E: Env, Q: SubModel2<Output = Tensor>, P: SubModel<Output = (Tensor, Tensor)>, R: ReplayBufferBase, E::Obs: Into<Q::Input1> + Into<P::Input>, E::Act: Into<Q::Input2> + From<Tensor>, Q::Input2: From<Tensor>, Q::Config: DeserializeOwned + Serialize + OutDim + Debug + PartialEq + Clone, P::Config: DeserializeOwned + Serialize + OutDim + Debug + PartialEq + Clone, R::Batch: TransitionBatch, <R::Batch as TransitionBatch>::ObsBatch: Into<Q::Input1> + Into<P::Input> + Clone, <R::Batch as TransitionBatch>::ActBatch: Into<Q::Input2> + Into<Tensor>,

fn train(&mut self)

Set the policy to training mode.

fn eval(&mut self)

Set the policy to evaluation mode.

fn is_train(&self) -> bool

Return if it is in training mode.

fn opt_with_record(&mut self, buffer: &mut R) -> Record

Performs an optimization step and returns some information.

fn save_params<T: AsRef<Path>>(&self, path: T) -> Result<()>

Save the parameters of the agent in the given directory. This method commonly creates a number of files consisting the agent in the directory. For example, the DQN agent in border_tch_agent crate saves two Q-networks corresponding to the original and target networks.

fn load_params<T: AsRef<Path>>(&mut self, path: T) -> Result<()>

Load the parameters of the agent from the given directory.

fn opt(&mut self, buffer: &mut R)

Performs an optimization step. Read more

impl<E, Q, P, R> Configurable<E> for Sac<E, Q, P, R>
where E: Env, Q: SubModel2<Output = Tensor>, P: SubModel<Output = (Tensor, Tensor)>, E::Obs: Into<Q::Input1> + Into<P::Input>, E::Act: Into<Q::Input2> + From<Tensor>, Q::Config: DeserializeOwned + Serialize + OutDim + Debug + PartialEq + Clone, P::Config: DeserializeOwned + Serialize + OutDim + Debug + PartialEq + Clone,

fn build(config: Self::Config) -> Self

Constructs Sac agent.

type Config = SacConfig<Q, P>

Configuration.

fn build_from_path(path: impl AsRef<Path>) -> Result<Self, Error>
where Self: Sized,

Build the object with the configuration in the yaml file of the given path.

impl<E, Q, P, R> Policy<E> for Sac<E, Q, P, R>
where E: Env, Q: SubModel2<Output = Tensor>, P: SubModel<Output = (Tensor, Tensor)>, E::Obs: Into<Q::Input1> + Into<P::Input>, E::Act: Into<Q::Input2> + From<Tensor>, Q::Config: DeserializeOwned + Serialize + OutDim + Debug + PartialEq + Clone, P::Config: DeserializeOwned + Serialize + OutDim + Debug + PartialEq + Clone,

fn sample(&mut self, obs: &E::Obs) -> E::Act

Sample an action given an observation.

Auto Trait Implementations§

impl<E, Q, P, R> Freeze for Sac<E, Q, P, R>
where <P as SubModel>::Config: Sized, <Q as SubModel2>::Config: Sized, P: Freeze,

impl<E, Q, P, R> RefUnwindSafe for Sac<E, Q, P, R>
where <P as SubModel>::Config: Sized, <Q as SubModel2>::Config: Sized, P: RefUnwindSafe, E: RefUnwindSafe, R: RefUnwindSafe, Q: RefUnwindSafe,

impl<E, Q, P, R> Send for Sac<E, Q, P, R>
where <P as SubModel>::Config: Sized, <Q as SubModel2>::Config: Sized, P: Send, E: Send, R: Send, Q: Send,

impl<E, Q, P, R> !Sync for Sac<E, Q, P, R>

impl<E, Q, P, R> Unpin for Sac<E, Q, P, R>
where <P as SubModel>::Config: Sized, <Q as SubModel2>::Config: Sized, P: Unpin, E: Unpin, R: Unpin, Q: Unpin,

impl<E, Q, P, R> UnwindSafe for Sac<E, Q, P, R>
where <P as SubModel>::Config: Sized, <Q as SubModel2>::Config: Sized, P: UnwindSafe, E: UnwindSafe, R: UnwindSafe, Q: UnwindSafe,

Blanket Implementations§

impl<T> Any for T
where T: 'static + ?Sized,

fn type_id(&self) -> TypeId

Gets the TypeId of self. Read more

impl<T> Borrow<T> for T
where T: ?Sized,

fn borrow(&self) -> &T

Immutably borrows from an owned value. Read more

impl<T> BorrowMut<T> for T
where T: ?Sized,

fn borrow_mut(&mut self) -> &mut T

Mutably borrows from an owned value. Read more

impl<T> From<T> for T

fn from(t: T) -> T

Returns the argument unchanged.

impl<T, U> Into<U> for T
where U: From<T>,

fn into(self) -> U

Calls U::from(self).

That is, this conversion is whatever the implementation of From<T> for U chooses to do.

impl<T> Same for T

type Output = T

Should always be Self

impl<T, U> TryFrom<U> for T
where U: Into<T>,

type Error = Infallible

The type returned in the event of a conversion error.

fn try_from(value: U) -> Result<T, <T as TryFrom<U>>::Error>

Performs the conversion.

impl<T, U> TryInto<U> for T
where U: TryFrom<T>,

type Error = <U as TryFrom<T>>::Error

The type returned in the event of a conversion error.

fn try_into(self) -> Result<U, <U as TryFrom<T>>::Error>

Performs the conversion.

impl<V, T> VZip<V> for T
where V: MultiLane<T>,

fn vzip(self) -> V