Due to the dynamic graph characteristics of PyTorch, there is no way to ... keepdim=True) res.append(correct_k.mul_(100.0 / batch_size)) ...
確定! 回上一頁