dhpo / ghm_loss.pytorch Goto Github PK

ghm_loss.py", line 55, in _custom_loss_grad
return torch.sigmoid(x).detach() - target
RuntimeError: The size of tensor a (2) must match the size of tensor b (208) at non-singleton dimension 1
hi, my input size is [208,2] and target size is [208], and i got error like this. could you please tell me
how to fix this error?

这一步是什么意思呢，没有看到，希望您能有空解释一下。
gd = bin_count * nonempty_bins

请问下class中的alpha是什么意思呢，是类别的权重嘛？（样本数目少的类别权重大）