TorchPQ

TorchPQ is a python library for Approximate Nearest Neighbor Search (ANNS) and Maximum Inner Product Search (MIPS) on GPU using Product Quantization (PQ) algorithm. TorchPQ is implemented mainly with PyTorch, with some extra CUDA kernels to accelerate clustering, indexing and searching.

Install

• make sure you have the latest version of PyTorch installed: https://pytorch.org/get-started/locally/
• install or upgrade to CUDA toolkit 11.0 or greater
• install a version of CuPy library that matches your CUDA toolkit version

pip install cupy-cuda110
pip install cupy-cuda111
pip install cupy-cuda112
...

for a full list of cupy-cuda versions, please go to Installation Guide

• install TorchPQ

pip install torchpq

Quick Start

IVFPQ

InVerted File Product Quantization (IVFPQ) is a type of ANN search algorithm that is designed to do fast and efficient vector search in million, or even billion scale vector sets. check the original paper for more details.

Training

from torchpq.index import IVFPQIndex
import torch

n_data = 1000000 # number of data points
d_vector = 128 # dimentionality / number of features

index = IVFPQIndex(
  d_vector=d_vector,
  n_subvectors=64,
  n_cells=1024,
  initial_size=2048,
  distance="euclidean",
)

trainset = torch.randn(d_vector, n_data, device="cuda:0")
index.train(trainset)

There are some important parameters that need to be explained:

• d_vector: dimentionality of input vectors. there are 2 constraints on d_vector: (1) it needs to be divisible by n_subvectors; (2) it needs to be a multiple of 4.*
• n_subvectors: number of subquantizers, essentially this is the byte size of each quantized vector, 64 byte per vector in the above example.**
• n_cells: number of coarse quantizer clusters
• initial_size: initial capacity assigned to each voronoi cell of coarse quantizer. n_cells * initial_size is the number of vectors that can be stored initially. if any cell has reached its capacity, that cell will be automatically expanded. If you need to add vectors frequently, a larger value for initial_size is recommended.

Remember that the shape of any tensor that contains data points has to be [d_vector, n_data].

* the second constraint could be removed in the future
** actual byte size would be (n_subvectors+9) bytes, 8 bytes for ID and 1 byte for is_empty

Adding new vectors

baseset = torch.randn(d_vector, n_data, device="cuda:0")
ids = torch.arange(n_data, device="cuda")
index.add(baseset, ids=ids)

Each ID in ids needs to be a unique int64 (torch.long) value that identifies a vector in x. if ids is not provided, it will be set to torch.arange(n_data, device="cuda") + previous_max_id

Removing vectors

index.remove(ids=ids)

index.remove(ids=ids) will virtually remove vectors with specified ids from storage. It ignores ids that doesn't exist.

Topk search

index.n_probe = 32
n_query = 10000
queryset = torch.randn(d_vector, n_query, device="cuda:0")
topk_values, topk_ids = index.search(queryset, k=100)

• when distance="inner", topk_values are inner product of queries and topk closest data points.
• when distance="euclidean", topk_values are negative squared L2 distance between queries and topk closest data points.
• when distance="manhattan", topk_values are negative L1 distance between queries and topk closest data points.
• when distance="cosine", topk_values are cosine similarity between queries and topk closest data points.

Encode and Decode

you can use IVFPQ as a vector codec for lossy compression of vectors

code = index.encode(queryset)   # compression
reconstruction = index.decode(code) # reconstruction

Save and Load

Most of the TorchPQ modules are inherited from torch.nn.Module, this means you can save and load them just like a regular pytorch model.

# Save to PATH
torch.save(index.state_dict(), PATH)
# Load from PATH
index.load_state_dict(torch.load(PATH))

Clustering

K-means

from torchpq.clustering import KMeans
import torch

n_data = 1000000 # number of data points
d_vector = 128 # dimentionality / number of features
x = torch.randn(d_vector, n_data, device="cuda")

kmeans = KMeans(n_clusters=4096, distance="euclidean")
labels = kmeans.fit(x)

Notice that the shape of the tensor that contains data points has to be [d_vector, n_data], this is consistant in TorchPQ.

Multiple concurrent K-means

Sometimes, we have multiple independent datasets that need to be clustered, instead of running multiple KMeans sequentianlly, we can perform multiple kmeans concurrently with MultiKMeans

from torchpq.clustering import MultiKMeans
import torch

n_data = 1000000
n_kmeans = 16
d_vector = 64
x = torch.randn(n_kmeans, d_vector, n_data, device="cuda")
kmeans = MultiKMeans(n_clusters=256, distance="euclidean")
labels = kmeans.fit(x)

Prediction with K-means

labels = kmeans.predict(x)

Benchmarks

• K-Means
• Sift1M
• Gist1M
• Sift10M (coming soon)
• Sift100M (coming soon)