reference/generated/torchrl.data.replay_buffers.WriterEnsemble

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

WriterEnsemble

class torchrl.data.replay_buffers.WriterEnsemble(*writers)

An ensemble of writers.

This class is designed to work with ReplayBufferEnsemble. It contains the writers but blocks writing with any of them.

Parameters:

writers (sequence of Writer) – the writers to make the composite writer.

Warning

This class does not support writing. To extend one of the replay buffers, simply index the parent ReplayBufferEnsemble object.

add()

Inserts one piece of data at an appropriate index, and returns that index.

extend()

Inserts a series of data points at appropriate indices, and returns a tensor containing the indices.

generations_of(index: int | Tensor) → Tensor

Returns the generation stamp for each physical slot in index.

A slot’s stamp advances once per write it receives, so a single extend that wraps the storage advances a reused slot once per write. Comparing a stamp captured at sampling time against the current stamp tells you whether the slot still holds the data you sampled.

Writers that do not track slot reuse – and writers constructed with track_generations=False, which is the default – report -1 everywhere. Never-written slots also report -1, so -1 means “no usable stamp” rather than “generation zero”.

Parameters:

index (int or torch.Tensor) – dim-0 slot indices. A 1-D tensor is always read as a batch of slot indices; for a storage with ndim > 1, pass a tuple of per-dimension indices (as extend() returns) to identify a single cell – only its dim-0 component is used, since a generation stamps a whole dim-0 slot.

Returns:

int64 stamps shaped like the dim-0 component of index, on index’s device.

Return type:

torch.Tensor