Unable to assign workers with indices {0} in the parallel pool their own GPUs, so these workers will be unused. When opening the pool, specify a number of workers equal to the number of available GPUs


trainNetwork