[Bug] oneflow.cat one input need return another tensor

[strint]

Summary

torch的cat，单输入，会创建另外一个单输出

>>> import torch
>>> torch.__version__
'1.10.0'
>>> x = torch.randn(2, 5, 5)
>>> y = torch.cat([x], 0)
>>> id(x)
140275845558832
>>> id(y)
140273210832272  # not x

oneflow的cat，单输入，输出直接返回了原输入

>>> import oneflow as torch
>>> x = torch.randn(2, 5, 5)
>>> y = torch.cat([x], 0)
>>> id(x)
140275721802288  
>>> id(y)
140275721802288 # is x

问题代码

class ConcatFunctor {
 public:
  ConcatFunctor() {
    ops_.resize(kMaxInputCount);
    for (int n = 1; n < ops_.size(); ++n) {
      ops_[n] = CHECK_JUST(one::OpBuilder("concat").Input("in", n + 1).Output("out").Build());
    }
  }
  Maybe<Tensor> operator()(const TensorTuple& inputs, const int64_t& dim) const {
    if (inputs.size() == 1) { return inputs.at(0); }  # 问题所在

[Aakifgithub]

Check the torch in few second on click event can be use