Approximate nearest neighbor search with product quantization on GPU in pytorch and cuda
MIT License
Published by DeMoriarty about 3 years ago
Improvements on IVFPQ
added DistributedCellContainer
Published by DeMoriarty about 3 years ago
Small improvements
Published by DeMoriarty over 3 years ago
IVFPQIndex search speed is greatly improved, now at least 2 times faster than before
IVFPQIndex now supports encoding residuals, pass pq_use_residual=True to initializer method in order to toggle residual encoding on. this improves recall rate, espcially for low code sizes. You can see the difference between residual encoding on and off from benchmark results
Added new submodules, such as index, clustering, codec, transform and more
old implementations from v0.1 are moved to torchpq/legacy/
more thorough benchmark results on sift1m dataset
Published by DeMoriarty over 3 years ago