RuntimeError: stack expects each tensor to be equal size, but got [3, 390, 771] at entry 0 and [3, 371, 880] at entry 1

[CHXAdmin]

batch size = 2 will cause this ,when batch size =1 ,it can run

[guoweiyu]

Thanks for your question! If you use different shape images in one batch, you should resize them to the same shape before input to the model