LibtorchRLTrajectoryBuffer::TensorBatch Struct Reference

Flattened tensor data assembled from the stored trajectories. More...

#include <LibtorchRLTrajectoryBuffer.h>

Public Member Functions
std::int64_t	size () const
	Return the number of transitions represented by the batch.

Public Attributes
torch::Tensor	observations
	Flattened observation matrix.
torch::Tensor	next_observations
	Flattened next-observation matrix.
torch::Tensor	actions
	Flattened action matrix.
torch::Tensor	log_probabilities
	Flattened action log-probabilities.
torch::Tensor	rewards
	Flattened rewards.
torch::Tensor	value_targets
	Flattened critic targets.
torch::Tensor	advantages
	Flattened advantages.

Detailed Description

Flattened tensor data assembled from the stored trajectories.

Definition at line 53 of file LibtorchRLTrajectoryBuffer.h.

Member Function Documentation

size()

std::int64_t LibtorchRLTrajectoryBuffer::TensorBatch::size ( ) const

inline

Return the number of transitions represented by the batch.

Definition at line 73 of file LibtorchRLTrajectoryBuffer.h.

{ return observations.defined() ? observations.size(0) : 0; }

Referenced by LibtorchRLMiniBatchSampler::sample(), LibtorchDRLControlTrainer::trainController(), and LibtorchRLMiniBatchSampler::validateBatch().

Member Data Documentation

actions

torch::Tensor LibtorchRLTrajectoryBuffer::TensorBatch::actions

Flattened action matrix.

Definition at line 60 of file LibtorchRLTrajectoryBuffer.h.

Referenced by LibtorchRLTrajectoryBuffer::flatten(), LibtorchRLMiniBatchSampler::makeMiniBatch(), and LibtorchRLMiniBatchSampler::validateBatch().

advantages

torch::Tensor LibtorchRLTrajectoryBuffer::TensorBatch::advantages

Flattened advantages.

Definition at line 68 of file LibtorchRLTrajectoryBuffer.h.

Referenced by LibtorchRLTrajectoryBuffer::flatten(), LibtorchRLMiniBatchSampler::makeMiniBatch(), and LibtorchRLMiniBatchSampler::validateBatch().

log_probabilities

torch::Tensor LibtorchRLTrajectoryBuffer::TensorBatch::log_probabilities

Flattened action log-probabilities.

Definition at line 62 of file LibtorchRLTrajectoryBuffer.h.

Referenced by LibtorchRLTrajectoryBuffer::flatten(), LibtorchRLMiniBatchSampler::makeMiniBatch(), and LibtorchRLMiniBatchSampler::validateBatch().

next_observations

torch::Tensor LibtorchRLTrajectoryBuffer::TensorBatch::next_observations

Flattened next-observation matrix.

Definition at line 58 of file LibtorchRLTrajectoryBuffer.h.

Referenced by LibtorchRLTrajectoryBuffer::flatten().

observations

torch::Tensor LibtorchRLTrajectoryBuffer::TensorBatch::observations

Flattened observation matrix.

Definition at line 56 of file LibtorchRLTrajectoryBuffer.h.

Referenced by LibtorchRLTrajectoryBuffer::flatten(), LibtorchRLMiniBatchSampler::makeMiniBatch(), and size().

rewards

torch::Tensor LibtorchRLTrajectoryBuffer::TensorBatch::rewards

Flattened rewards.

Definition at line 64 of file LibtorchRLTrajectoryBuffer.h.

Referenced by LibtorchRLTrajectoryBuffer::flatten().

value_targets

torch::Tensor LibtorchRLTrajectoryBuffer::TensorBatch::value_targets

Flattened critic targets.

Definition at line 66 of file LibtorchRLTrajectoryBuffer.h.

Referenced by LibtorchRLTrajectoryBuffer::flatten(), LibtorchRLMiniBatchSampler::makeMiniBatch(), and LibtorchRLMiniBatchSampler::validateBatch().

---

The documentation for this struct was generated from the following file:

• stochastic_tools/include/libtorch/utils/LibtorchRLTrajectoryBuffer.h