Convenience

default ctor is good

@brief Get the layer instance's name. * * Each layer in a model should have a unique, preferably * human-readable, name. */

@brief Return time in fractional seconds since an epoch. */

Return true if this process is the master process in its trainer. */

Return true if this process is the world master process. */

@brief Returns the reference count. */

Get trainer grid number (0: no primary/secondary grid, 1: part of primary * grid, 2: part of secondary grid). */

@brief Get number of parent layers. */

@brief Current step in the training algorithm * @details Step counts the number of iterations in the training * algorithm's internal state

@brief Clear any internal state in the timer. * @post running() returns @c false; check() and stop() return 0. */

Get the rank of this process in its trainer. */

@brief Get number of child layers. */

@brief Returns weights file_format as string. */

Execute the script. The script is executed directly and is _not_ submitted to a job scheduler. The script file is written before bein

Number of times the training set has been traversed. */

Get which trainer this process is in. */

Get my rank in COMM_WORLD. */

@brief Return an exception-safe, memory-safe copy of this object. */

Convert to an iterable object. Simply returns `obj` if it is alredy iterable. Otherwise returns a 1-tuple containing `obj`. `str`s are treate

Return asynchronous flag for sub-grid parallelism */

@brief Set the layer instance's name. * Each layer in a model should have a unique, preferably * human-readable, name. */

* Construct an OpenCV Mat that refers to data. * No data is copied, this just sets up a cv::Mat header. * @param data The matrix with data to use.

Return the number of processes in a trainer. */