Struct RolloutBurnTensors

Source

pub struct RolloutBurnTensors<B: Backend> {
    pub observations: Tensor<B, 2>,
    pub actions: Tensor<B, 1, Int>,
    pub old_log_probs: Tensor<B, 1>,
    pub old_values: Tensor<B, 1>,
    pub advantages: Tensor<B, 1>,
    pub returns: Tensor<B, 1>,
}

Expand description

Bundle of Burn tensors produced by RolloutBatch::to_burn_tensors.

Fields are in the order PPO trainers consume them: policy/value inputs first (observations, actions), then the old policy outputs (log-probs and values used for the importance ratio and value clip), then the GAE outputs (advantages and returns). Generic over the backend B so the same trainer surface works for CPU (NdArray), GPU (Wgpu, Cuda), and Autodiff<_> wrappers.

Fields§

§observations: Tensor<B, 2>

Observations, shape [batch_size, obs_dim], dtype f32.

§actions: Tensor<B, 1, Int>

Discrete actions, shape [batch_size], dtype i64.

§old_log_probs: Tensor<B, 1>

Behavior-policy log-probabilities, shape [batch_size], dtype f32.

§old_values: Tensor<B, 1>

Behavior-policy value estimates V(s_t), shape [batch_size], dtype f32.

§advantages: Tensor<B, 1>

GAE advantages, shape [batch_size], dtype f32.

§returns: Tensor<B, 1>

Value-function targets (advantages + values), shape [batch_size], dtype f32.

RolloutBurnTensors

Struct RolloutBurnTensors Copy item path

Fields§

Trait Implementations§

impl<B: Debug + Backend> Debug for RolloutBurnTensors<B>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Auto Trait Implementations§

impl<B> Freeze for RolloutBurnTensors<B>where <B as BackendTypes>::IntTensorPrimitive: Freeze, <B as BackendTypes>::FloatTensorPrimitive: Freeze, <B as BackendTypes>::QuantizedTensorPrimitive: Freeze,

impl<B> RefUnwindSafe for RolloutBurnTensors<B>where <B as BackendTypes>::IntTensorPrimitive: RefUnwindSafe, <B as BackendTypes>::FloatTensorPrimitive: RefUnwindSafe, <B as BackendTypes>::QuantizedTensorPrimitive: RefUnwindSafe,

impl<B> Send for RolloutBurnTensors<B>

impl<B> Sync for RolloutBurnTensors<B>

impl<B> Unpin for RolloutBurnTensors<B>where <B as BackendTypes>::IntTensorPrimitive: Unpin, <B as BackendTypes>::FloatTensorPrimitive: Unpin, <B as BackendTypes>::QuantizedTensorPrimitive: Unpin,

impl<B> UnsafeUnpin for RolloutBurnTensors<B>where <B as BackendTypes>::IntTensorPrimitive: UnsafeUnpin, <B as BackendTypes>::FloatTensorPrimitive: UnsafeUnpin, <B as BackendTypes>::QuantizedTensorPrimitive: UnsafeUnpin,

impl<B> UnwindSafe for RolloutBurnTensors<B>where <B as BackendTypes>::IntTensorPrimitive: UnwindSafe, <B as BackendTypes>::FloatTensorPrimitive: UnwindSafe, <B as BackendTypes>::QuantizedTensorPrimitive: UnwindSafe,

Blanket Implementations§

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

fn type_id(&self) -> TypeId

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

fn borrow(&self) -> &T

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

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

impl<T> From<T> for T

fn from(t: T) -> T

impl<T> Instrument for T

fn instrument(self, span: Span) -> Instrumented<Self>

fn in_current_span(self) -> Instrumented<Self>

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

fn into(self) -> U

impl<T> IntoEither for T

fn into_either(self, into_left: bool) -> Either<Self, Self>

fn into_either_with<F>(self, into_left: F) -> Either<Self, Self>where F: FnOnce(&Self) -> bool,

impl<T> Pointable for T

const ALIGN: usize

type Init = T

unsafe fn init(init: <T as Pointable>::Init) -> usize

unsafe fn deref<'a>(ptr: usize) -> &'a T

unsafe fn deref_mut<'a>(ptr: usize) -> &'a mut T

unsafe fn drop(ptr: usize)

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

type Error = Infallible

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

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

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

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

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

fn vzip(self) -> V

impl<T> WithSubscriber for T

fn with_subscriber<S>(self, subscriber: S) -> WithDispatch<Self>where S: Into<Dispatch>,

fn with_current_subscriber(self) -> WithDispatch<Self>

Struct RolloutBurnTensors

impl<B> Freeze for RolloutBurnTensors<B>
where <B as BackendTypes>::IntTensorPrimitive: Freeze, <B as BackendTypes>::FloatTensorPrimitive: Freeze, <B as BackendTypes>::QuantizedTensorPrimitive: Freeze,

impl<B> RefUnwindSafe for RolloutBurnTensors<B>
where <B as BackendTypes>::IntTensorPrimitive: RefUnwindSafe, <B as BackendTypes>::FloatTensorPrimitive: RefUnwindSafe, <B as BackendTypes>::QuantizedTensorPrimitive: RefUnwindSafe,

impl<B> Unpin for RolloutBurnTensors<B>
where <B as BackendTypes>::IntTensorPrimitive: Unpin, <B as BackendTypes>::FloatTensorPrimitive: Unpin, <B as BackendTypes>::QuantizedTensorPrimitive: Unpin,

impl<B> UnsafeUnpin for RolloutBurnTensors<B>
where <B as BackendTypes>::IntTensorPrimitive: UnsafeUnpin, <B as BackendTypes>::FloatTensorPrimitive: UnsafeUnpin, <B as BackendTypes>::QuantizedTensorPrimitive: UnsafeUnpin,

impl<B> UnwindSafe for RolloutBurnTensors<B>
where <B as BackendTypes>::IntTensorPrimitive: UnwindSafe, <B as BackendTypes>::FloatTensorPrimitive: UnwindSafe, <B as BackendTypes>::QuantizedTensorPrimitive: UnwindSafe,

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

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

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

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

fn into_either_with<F>(self, into_left: F) -> Either<Self, Self>
where F: FnOnce(&Self) -> bool,

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

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

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

fn with_subscriber<S>(self, subscriber: S) -> WithDispatch<Self>
where S: Into<Dispatch>,