Context manager: build a multi-shard WebDataset pipeline (3 tar files).

! * \brief Allocates storage of size at least `__bytes` bytes. * * The returned storage is aligned to the specified `__alignment` if such

Returns the current checkpoints of the pipelines.

generate sinewaves with given frequencies, add Hann envelope and store in channel-last layout

The layout of tensors in the batch. The "batch dimension" (commonly denoted as N) is not included - a batch of HWC images wi

Natural logorithm. \param arg function argument \return logarithm of \a arg to base e template<typename T> typename enable<expr,T>::type log(T arg) {

Convert a ``PIL.Image`` to a uint8 CHW ``torch.Tensor``. Values are in [0, 255] and the dtype is ``torch.uint8``. No scaling is applied.

Computes reciprocal of the square root of values in `input`. :rtype: TensorList of rsqrt(input). If input is an integer, the result will be float

Sets the shared event handle that marks the readiness of the tensor data. */

* @brief Total number of values in a frame (width * height * channels) */

* @brief Returns the layout of the input at given index */

* Set requested batch size for all outputs */

* @brief Generate block descriptors for given shape. * * @param output_shape - shape to cover with blocks * @param force_variable_size - true

Run one full epoch, return list of per-sample fingerprints.

Run one full epoch and return a list of label values.

Returns the integer labels of a (jpeg, label) batch as a list.

Apply DALI CLAHE using memory-based pipeline for exact input matching

Fully configurable shape generator. Returns a callable which serves as a non-uniform & random shape generator to generate_epoch Usage:

The default ``EvalContext`` for the calling thread.

Returns parallel external source's statistics regarding shared memory consumption. The returned dictionary contains following keys:

Return collected performance metrics.

Get versions at a given index, compatible with provided cuda_version Parameters ---------- idx : int Index of ver

Loads a fragment from memory

Computes natural logarithm (base-e) of values in `input`. :rtype: TensorList of log(input). If input is an integer, the result will be float,

@brief Number of dimensions described by this object; same value as size() */

Next representable value. \param from value to compute next representable value for \param to direction towards which to compute next value \return ne

Computes base to the power of exponents, that is base ** exponent. :rtype: TensorList of pow(base, exponent). Type is calculated based on the typ