You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@lucene.apache.org by GitBox <gi...@apache.org> on 2022/05/02 16:26:43 UTC

[GitHub] [lucene] mayya-sharipova commented on pull request #862: LUCENE-9848 Sort HNSW graph neighbors for construction

mayya-sharipova commented on PR #862:
URL: https://github.com/apache/lucene/pull/862#issuecomment-1115093505

   I've run benchmarks on 2 datasets:
   baseline  – main branch
   candidate – this PR:
   
   Indexing time was the same or slightly better.
   Recall slightly increased: 1-3%
   QPS slightly decreased
   
   ---
   **glove-100-angular M:16 efConstruction:100**
   
   
   |             | baseline recall | baseline QPS | candidate recall | candidate QPS |
   | ----------- | --------------: | -----------: | ---------------: | ------------: |
   | n_cands=10  |           0.463 |     3742.506 |            0.476 |      3743.109 |
   | n_cands=20  |           0.529 |     3191.042 |            0.541 |      2954.978 |
   | n_cands=40  |           0.603 |     2576.537 |            0.617 |      2681.529 |
   | n_cands=80  |           0.682 |     1877.982 |            0.696 |      1629.359 |
   | n_cands=120 |           0.721 |     1492.808 |            0.737 |      1447.522 |
   | n_cands=200 |           0.766 |     1099.154 |            0.784 |      1093.390 |
   | n_cands=400 |           0.818 |      672.911 |            0.834 |       667.136 |
   | n_cands=600 |           0.845 |      489.292 |            0.859 |       478.109 |
   | n_cands=800 |           0.862 |      383.402 |            0.875 |       383.488 |
   
   
   ---
   **sift-128-euclidean  M:16 efConstruction:500**
   
   |             | baseline recall | baseline QPS | candidate recall | candidate QPS |
   | ----------- | --------------: | -----------: | ---------------: | ------------: |
   | n_cands=10  |           0.747 |     3891.531 |            0.753 |      3304.833 |
   | n_cands=20  |           0.817 |     3359.364 |            0.819 |      2879.256 |
   | n_cands=40  |           0.889 |     2590.605 |            0.887 |      2280.737 |
   | n_cands=80  |           0.944 |     1798.558 |            0.941 |      1648.794 |
   | n_cands=120 |           0.964 |     1383.721 |            0.963 |      1377.711 |
   | n_cands=200 |           0.983 |      973.862 |            0.981 |       917.995 |
   | n_cands=400 |           0.994 |      586.816 |            0.993 |       548.358 |
   | n_cands=600 |           0.997 |      427.128 |            0.996 |       408.959 |
   | n_cands=800 |           0.998 |      341.178 |            0.997 |       316.840 |
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: issues-unsubscribe@lucene.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@lucene.apache.org
For additional commands, e-mail: issues-help@lucene.apache.org