openeuler/hnswlib repository overview

⁠Quick reference

• The official hnswlib docker image.

• Maintained by: openEuler CloudNative SIG⁠.

• Where to get help: openEuler CloudNative SIG⁠, openEuler⁠.

⁠hnswlib | openEuler

Current hnswlib docker images are built on the openEuler⁠. This repository is free to use and exempted from per-user rate limits.

Hnswlib - fast approximate nearest neighbor search. Header-only C++ HNSW implementation with python bindings, insertions and updates.

Hnswlib is lightweight, header-only, and has no dependencies other than C++ 11. It provides interfaces for C++, Python, and external support for Java and R, has full support for incremental index construction and updating the elements, and can work with custom user defined distances (C++).

Learn more about hnswlib on Hnswlib - fast approximate nearest neighbor search⁠.

⁠Supported tags and respective Dockerfile links

The tag of each hnswlib docker image is consist of the version of hnswlib and the version of basic image. The details are as follows

⁠Usage

In this usage, users can select the corresponding {Tag} based on their requirements.

• Pull the openeuler/hnswlib image from docker

  docker pull openeuler/hnswlib:{Tag}
  

  Copy

• Run an hnswlib C++ program

  Create a file named hnsw_example.cpp with the following content:

  #include <hnswlib/hnswlib.h>
  
  int main() {
      int dim = 16;
      int max_elements = 10000;
      int M = 16;
      int ef_construction = 200;
  
      hnswlib::L2Space space(dim);
      hnswlib::HierarchicalNSW<float>* alg_hnsw = new hnswlib::HierarchicalNSW<float>(&space, max_elements, M, ef_construction);
  
      std::mt19937 rng;
      rng.seed(47);
      std::uniform_real_distribution<> distrib_real;
      float* data = new float[dim * max_elements];
      for (int i = 0; i < dim * max_elements; i++) {
          data[i] = distrib_real(rng);
      }
  
      for (int i = 0; i < max_elements; i++) {
          alg_hnsw->addPoint(data + i * dim, i);
      }
  
      float correct = 0;
      for (int i = 0; i < max_elements; i++) {
          std::priority_queue<std::pair<float, hnswlib::labeltype>> result = alg_hnsw->searchKnn(data + i * dim, 1);
          hnswlib::labeltype label = result.top().second;
          if (label == i) correct++;
      }
      float recall = correct / max_elements;
      std::cout << "Recall: " << recall << "\n";
  
      std::string hnsw_path = "hnsw.bin";
      alg_hnsw->saveIndex(hnsw_path);
      delete alg_hnsw;
  
      alg_hnsw = new hnswlib::HierarchicalNSW<float>(&space, hnsw_path);
      correct = 0;
      for (int i = 0; i < max_elements; i++) {
          std::priority_queue<std::pair<float, hnswlib::labeltype>> result = alg_hnsw->searchKnn(data + i * dim, 1);
          hnswlib::labeltype label = result.top().second;
          if (label == i) correct++;
      }
      recall = (float)correct / max_elements;
      std::cout << "Recall of deserialized index: " << recall << "\n";
  
      delete[] data;
      delete alg_hnsw;
      return 0;
  }
  

  Copy

• Compile the program inside the container

  docker run --rm -v $(pwd):/workspace openeuler/hnswlib:{Tag} g++ -std=c++11 -O2 -pthread /workspace/hnsw_example.cpp -o /workspace/hnsw_example
  

  Copy

• Run the compiled program

  docker run --rm -v $(pwd):/workspace openeuler/hnswlib:{Tag} /workspace/hnsw_example
  

  Copy

• Check the container logs

  docker run --rm --name hnswlib-test -v $(pwd):/workspace openeuler/hnswlib:{Tag} /workspace/hnsw_example
  docker logs hnswlib-test
  

  Copy

• Open an interactive shell

  docker run -it --rm -v $(pwd):/workspace openeuler/hnswlib:{Tag} bash
  

  Copy

⁠Question and answering

If you have any questions or want to use some special features, please submit an issue or a pull request on openeuler-docker-images⁠.