As shown below ,the left one is wrong when converted to tensor_RT model, the right is good, when I compare those two model ,the output’s shape of left is not specified. Is this a normal or wrong phenomenon?
