facebookresearch/faiss
1
0
Fork 0
mirror of https://github.com/facebookresearch/faiss.git synced 2026-10-12 02:13:36 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix flaky test_hnsw smoke test by increasing efSearch (#5222)

Browse Source 
Summary:
Pull Request resolved: https://github.com/facebookresearch/faiss/pull/5222

test_wheel_smoke::TestIndexFactory::test_hnsw assumes HNSW returns the query vector itself as the nearest neighbor, but the default efSearch (16) is too low to guarantee self-retrieval on 2000 random vectors in d=64. This caused intermittent failures on macOS CI.

Set efSearch=64 so the search explores enough of the graph to find exact matches reliably.

___

overriding_review_checks_triggers_an_audit_and_retroactive_review
Oncall Short Name: faiss

Differential Revision: D105621052

fbshipit-source-id: 80ccd2ec5aecfb88d2fad780449bfb49d5d18db5
This commit is contained in:
Alibek Zhakubayev
2026-05-18 15:44:59 -07:00
committed by meta-codesync[bot]
parent 8d3cc92536
commit 5c92c5c1ee
1 changed files with 1 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/test_wheel_smoke.py
+1
View File
@@ -101,6 +101,7 @@ class TestIndexFactory(unittest.TestCase):
np.random.seed(42)
xb = np.random.random((n, d)).astype("float32")
index = faiss.IndexHNSWFlat(d, 16)
index.hnsw.efSearch = 64
index.add(xb)
assert index.ntotal == n
D, I = index.search(xb[:5], 10)