Grayscale to 3 channels

[AdamGregor]

Hello, I am trying to load my custom dataset. I have got images with one channel (grayscale) and three channels (color) there. When I try to run training, I recieve a

RuntimeError: stack expects each tensor to be equal size, but got [3, 256, 256] at entry 0 and [1, 256, 256] at entry 2

because of the different shapes. Could anyone please help me how to transform the grayscale images to three channels? Thank you! :)