In my data structure the sparsity of the tensors is very hight. On 1M columns I have a sparsity of .99%, i.e. a density of 0.1%. I'm aware that similar indexes for Approximated Nearest Neighbor like FAISS and Annoy work better for dense vectors. How about FLANN performances with sparsity?
In my data structure the sparsity of the tensors is very hight. On 1M columns I have a sparsity of .99%, i.e. a density of 0.1%. I'm aware that similar indexes for Approximated Nearest Neighbor like FAISS and Annoy work better for dense vectors. How about FLANN performances with sparsity?