Feeding Single channel to Pretrained Resnet

What is best approach to load single channel grayscale data ( type ‘L’), into pretrained model trained on grayscale RGB images (3 channel).

The thought process is I can train larger batch sizes since we only have a single channel (1/3) RGB.

Note: Grayscale RBG images are such that each channel is identical, so we aren’t giving the model different information.