What's the differences between vsag and Proxima?

[goodluckcwl]

No description provided.

[wxyucs]

@goodluckcwl thank you for your interesting. I know few about Proxima(it seems not open sourced), I can introduce the vsag for you.

vsag is a vector index library developed by vdb team from antgroup, it initialed at July 2023. vsag is based on HNSW(the most popular ANN algorithm) and DiskANN(the most popular disk-based ANN algorithm). Then we made some performance and functional improvements when we using it in some internal businesses(billion-scale). For now, vsag can perform very high recall searching on HNSW and provides an ease-to-use DiskANN index.

Within the growth of data scale, disk-based index become more important. So we decided to make vsag open-source and developing on GitHub, trying to make some contributions in community.

[nedchu]

@goodluckcwl hi, nice to meet you. Thanks for the nice comments from @wxyucs. I have a few more to add on.

vsag and proxima are both vector search libraries. Here are some reasons why vsag may be a good choice:

1. vsag is competitive in performance. It achieves SOTA in ann-benchmarks (our team is submitting the results).
2. vsag can support large-scale vector search (hundreds of billions of vectors).
3. vsag is deployed in critical services at Ant Group. Our team actively works on vsag every day using GitHub.
4. Our team checks Github regularly. Any question/issue/pr is welcome. You can also chat with us via wechat/dingding group.

Our team highly values the feedback from the community. Please do not hesitate to ask us if you have any other questions.