reference/generated/tensordict.TensorDictBase.make_memmap_from_storage

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

TensorDictBase.make_memmap_from_storage

TensorDictBase.make_memmap_from_storage(key: NestedKey, storage: UntypedStorage, shape: Size | Tensor, *, dtype: dtype | None = None, robust_key: bool | None = True) → MemoryMappedTensor

Creates an empty memory-mapped tensor given a storage, a shape and possibly a dtype.

Warning

This method is not lock-safe by design. A memory-mapped TensorDict instance present on multiple nodes will need to be updated using the method memmap_refresh_().

Note

If the storage has a filename associated, it must match the new filename for the file. If it has not a filename associated but the tensordict has an associated path, this will result in an exception.

Parameters:

• key (NestedKey) – the key of the new entry to write. If the key is already present in the tensordict, an exception is raised.

• storage (UntypedStorage) – the storage to use for the new MemoryMappedTensor. Must be a physical memory storage.

• shape (Size or equivalent, torch.Tensor for nested tensors) – the shape of the tensor to write.

Keyword Arguments:

• dtype (dtype, optional) – the dtype of the new tensor.

• robust_key (bool, optional) – if True (default), uses robust key encoding that safely handles keys with path separators and special characters. If False, uses legacy behavior (keys used as-is). If None, uses the default robust behavior.

Returns:

A new memory mapped tensor with the given storage.

Every tensordict class and every tensorclass has this method. For the other methods, see Methods by task.