Can i use two dynamic shapes?

AI & Data Science Deep Learning (Training & Inference) TensorRT
, ,
1

for example :
ITensor* shape = addConstant(Dims(4, {1,1,-1,-1}));
// Two of the shapes are variable, but both have the same value.

IShuffleLayer* reshape_layer = addShuffle(*mask); //mask is Dims4 random tensor
reshape_layer->setInput(1, *shape)
//impossible

//but
ITensor* shape = addConstant(Dims(4, {1,1,-1,13}));
//If I put one variable, actually i can’t figure out the shape
IShuffleLayer* reshape_layer = addShuffle(*mask); //mask is Dims4 random tensor
reshape_layer->setInput(1, *shape)
//possible

So, can i use two dynamic shapes? and
How to use it?

Thanks! Have a nice day!

4

Hi @ghdgudfkr

Please refer to the below doc while working with dynamic shapes

Thanks