All pull requests

Pull requests

• Scale the HNSW and PiPNN search check period by the thread count

  CLA Signedmeta-exported
  #5784·alibeklfc opened Oct 26, 2026Contributor·

  ·
  1

• Train ProductQuantizer subquantizers in parallel

  CLA Signedmeta-exported
  #5783·alibeklfc opened Oct 26, 2026Contributor·

  ·
  1

• Use a SIMD masked sum for the fp32 RaBitQ 1-bit estimate

  CLA Signedmeta-exported
  #5782·alibeklfc opened Oct 26, 2026Contributor·

  ·
  1

• Add NEON kernels for fvec_L2sqr_ny / fvec_inner_products_ny

  #5781·shiyu-zhang opened Oct 26, 2026·

  ·
  1

• Fix IndexBinaryMultiHash serialization when a bucket holds all ntotal ids

  CLA Signed
  #5780·ViderStar opened Oct 25, 2026·

  ·
  2

• Use a reservoir for large-k IVF search

  CLA Signed
  #5779·Tickdack opened Oct 25, 2026Contributor·4 of 5 tasks·

  ·

• #5756 :: Forward and translate ID selectors through Refine and PreTransform

  CLA Signed
  #5778·ricardoleal20 opened Oct 25, 2026·

  ·
  1

• Move compiler-specific macros to their own section

  CLA Signed
  #5777·JAicewizard opened Oct 24, 2026Contributor·

  ·
  1

  3

• Add missing bounds checks in reconstruct and ArrayInvertedLists accessors

  CLA Signed
  #5776·Aaryan-Dadu opened Oct 24, 2026·

  ·
  1

  3

• Add remove_ids to IndexShards and IndexReplicas, and add_with_ids to IndexReplicas (#5773)

  CLA Signedmeta-exported
  #5773·mnorris11 opened Oct 23, 2026Contributor·

  ·
  1

• fbcode/faiss/tests/test_fast_scan_distance_to_code.cpp (#5772)

  CLA Signedmeta-exported
  #5772·meta-codesync[bot] opened Oct 23, 2026Bot·

  ·

• Fix OMP exception safety in ProductQuantizer

  CLA Signed
  #5771·thlurte opened Oct 23, 2026·

  ·
  4

• Add BBQ layout to the CAGRA Dataset API

  CLA Signed
  #5769·tarang-jain opened Oct 22, 2026Contributor·

  ·

• Add RISC-V RVV kernels for approx top-k, Hamming top-k and uint16 partitioning

  CLA Signed
  #5767·anthony-zy opened Oct 22, 2026Contributor·

  ·

• Add RISC-V RVV kernels for FastScan

  CLA Signed
  #5765·anthony-zy opened Oct 22, 2026Contributor·

  ·

• Validate codebook size in read_ResidualQuantizer_old and bounds-check IndexBinaryIVF::reconstruct_from_offset

  CLA Signed
  #5764·professor-moody opened Oct 21, 2026·

  ·
  4

• Fix cuVS IVF-Flat workspace sizing after CPU-to-GPU cloning

  CLA Signed
  #5763·nimusp opened Oct 21, 2026·

  ·
  3

• Fix assignment after progressive-dimension Kmeans training

  CLA Signed
  #5761·akshitkhokhani opened Oct 20, 2026·

  ·
  1

• Make HadamardRotation depth configurable (1-3 rounds)

  CLA Signed
  #5760·ranbenbasat opened Oct 20, 2026Contributor·

  ·

• Add MetalIndexBinaryFlat with GPU Hamming search and exact top-k selection

  CLA Signed
  #5759·Evandabest opened Oct 20, 2026Contributor·

  ·

• Fix flat search results when OpenMP provides a smaller team

  CLA Signed
  #5754·MokiMeow opened Oct 17, 2026·

  ·
  1

  2

• Compute the coarse distances in search_preassigned when they are not given

  CLA Signed
  #5748·Kayvan-Zahiri opened Oct 16, 2026Contributor·

  ·
  1

• Make the multi-GPU Python helpers work on the Metal backend

  CLA Signed
  #5747·Kayvan-Zahiri opened Oct 16, 2026Contributor·

  ·
  1

• Add IndexIVFRaBitQSSD: IVF-RaBitQ with codes split across RAM and SSD

  CLA Signed
  #5746·YuexuanXu7 opened Oct 16, 2026·

  ·
  1

• Add dual FMA accumulator to AVX2 scalar-quantizer distance loop

  CLA Signedmeta-exported
  #5745·yfeldblum opened Oct 16, 2026Contributor·

  ·
  1