1
I've learning about PyTorch sparse tensors : https://pytorch.org/docs/stable/sparse.html From the docs (https://pytorch.org/docs/stable/sparse.html) : "Torch supports sparse tensors in COO(rdinate) format, which can efficiently store and process tensors for which the majority of elements are zeros." Is one of the intended use of sparse tensors instead of regular PyTorch tensors to decrease training time ?
python machine-learning deep-learning pytorch
share | improve this question
asked Nov 24 '18 at 23:08
blue-sky blue-sky
17.5k 84 274 504