When Dynamic Input Shapes will be supported?

Sometimes, our model need dynamic shape input , such a full convolutional network. But it seems that TVM doesn’t support it now.
I hope to know when it will be supported.

Thanks for developers’ nice work !

Execuse me @tqchen