Tefor
Results
2
comments of
Tefor
In loss.py (line 173 ), gain = torch.ones(7, device=targets.device) → gain = **torch.ones(7, device=targets.device).long()**
> Thank you for your answer, it is indeed a resolution problem, my input image resolution is 128*128, so that when calculating the attention N=H*W=16384, this is too big, I...