Jin
Is post_ln=1 supported for inference with protobuf file? According to the code below, it seems post_ln=1 is not correctly supported. The input is still pre-normalized. ``` template __global__ void ker_norm_layer_resual(T*...
Hi Tianzhi, Thanks for your perfect job. I have a few questions to ask. Do you support to compute the offset mentioned in section 3.3 "Side-refinement" of your paper in...
GPU memory
You mentioned in the paper that the batch size for training is 32. However for my GPU which has 8G I set the batch size to 8 and almost all...
 When training with the code of this repo, my generated images always have white spots. Does anyone have the similar results? And how to avoid it? Thanks