wifi-densepose

Commit Graph

Author	SHA1	Message	Date
rUv	1f05456588	feat(ADR-261 M2): multi-bit + large-N ANN scaling study — measured, no crossover (refutes M1 prediction) (#1066 ) * feat(ADR-261): multi-bit (b∈{1,2,4}) quantized HNSW traversal + scaling harness Generalize the SymphonyQG-style quantized-traversal HNSW from 1-bit Hamming to a b-bit-per-dimension code (b ∈ {1,2,4}), mirroring ADR-156 §10's multi-bit RaBitQ scheme (rotate via FHT Pass-2, uniform mid-rise scalar quantizer over [-3,3], ranked by per-dim L1). b=1 is byte-for-byte the original construction (codes in {0,1} ⇒ L1 == Hamming), pinned by one_bit_build_bits_matches_legacy_build. Bytes/node scales linearly: 128-d → 16/32/64 B for b=1/2/4. - hnsw_quantized.rs: QuantizedHnswIndex::build_bits(...,bits,...), bits()/ bytes_per_node() accessors, code-L1 greedy+beam traversal. build(...) kept as the b=1 backward-compatible entry point. +4 tests (multi-bit recall regression, bits clamp, bytes/node, legacy parity). - ann_measure.rs: build_indices_bits / build_quant_bits / run_scaling_study + best_float_op / best_quant_op; scaling_report (#[ignore], --release) and a CI-safe scaling_study_small_is_consistent. - ann_bench.rs: 2-bit and 4-bit quant criterion benches over the shared graph. ruvector lib 151 → 156 passed, 0 failed, 1 ignored (scaling_report). Co-Authored-By: claude-flow <ruv@ruv.net> * docs(adr-261): record M2 multi-bit scaling study — measured, no crossover (refutes M1 prediction) Multi-bit (b∈{1,2,4}) quantized HNSW traversal + N∈{10k,100k,250k} scaling study, measured on this box. No crossover at any (N,b): at 10k more bits help (ratio 0.19→0.48×, b≥2 reaches 0.90 recall) but quant stays slower than float HNSW at equal recall; at 100k/250k quant recall collapses (b=4: 1.0→0.788→0.624, never ≥0.90) while float holds ≥0.92. The predicted large-N crossover moved the wrong way. Published negative with the mechanism explained. ADR-261 §11. Co-Authored-By: claude-flow <ruv@ruv.net> --------- Co-authored-by: ruv <ruvnet@gmail.com>	2026-06-14 10:31:00 -04:00
rUv	f756a8af49	feat(ADR-261): ruvector HNSW graph-ANN (25x measured vs linear) + honest SymphonyQG-direction refutation (#1063 ) * feat(ruvector): real float HNSW + SymphonyQG-style quantized-traversal index (ADR-261) Adds the graph-ANN index the ruvector retrieval path was missing (ADR-156 §5 #1 noted there was no HNSW baseline to measure SymphonyQG against). - hnsw.rs: correct float HNSW (Malkov & Yashunin) — multi-layer NSW graph, ef_construction/ef_search, Algorithm-4 neighbour selection, seeded- deterministic level assignment (SplitMix64, reused from rotation.rs), L2 + cosine, brute-force ground truth, full degenerate-case guards. recall@10 correctness gate >=0.95 vs brute force (L2 + cosine). - hnsw_quantized.rs: SymphonyQG-style variant — same graph, traversal scored by cheap 1-bit Hamming over the RaBitQ Pass-2 rotated sign code, final exact-float rerank. - ann_measure.rs: shared deterministic planted-cluster fixture + recall/QPS measurement (ann_bench_report is the ADR source of truth). Fixes an index-out-of-bounds bug the recall gate caught: insert wired bidirectional edges before pushing the node's own link row. +20 tests, ruvector lib 131->151, 0 failed. Co-Authored-By: claude-flow <ruv@ruv.net> * bench(ruvector): criterion ann_bench for HNSW vs quantized vs linear (ADR-261) Times the same shared ann_measure fixture/indices through criterion so the bench and the report test can never measure different graphs. Co-Authored-By: claude-flow <ruv@ruv.net> * docs(adr-261): graph-ANN index ADR with MEASURED HNSW vs quantized verdict ADR-261 (Accepted): float HNSW ~25x QPS over linear scan at recall >=0.99 (the baseline ADR-156 said was missing). Honest negative: the 1-bit quantized traversal is too coarse to beat float HNSW at equal recall at N=10k (best recall 0.738, no >=0.90 equal-recall point) — the SymphonyQG 3.5-17x is NOT reproduced by our 1-bit construction; expected crossover at large N + a multi-bit code. Caveat: our HNSW + our quant, not SymphonyQG's system — direction tested, not a 1:1 reproduction. ADR-156 §5 #1 + §8 backlog: CLAIMED -> MEASURED-direction-tested. CHANGELOG [Unreleased] entry. Co-Authored-By: claude-flow <ruv@ruv.net>	2026-06-14 02:33:32 -04:00

Author

SHA1

Message

Date

rUv

1f05456588

feat(ADR-261 M2): multi-bit + large-N ANN scaling study — measured, no crossover (refutes M1 prediction) (#1066 )

* feat(ADR-261): multi-bit (b∈{1,2,4}) quantized HNSW traversal + scaling harness

Generalize the SymphonyQG-style quantized-traversal HNSW from 1-bit Hamming to a
b-bit-per-dimension code (b ∈ {1,2,4}), mirroring ADR-156 §10's multi-bit RaBitQ
scheme (rotate via FHT Pass-2, uniform mid-rise scalar quantizer over [-3,3],
ranked by per-dim L1). b=1 is byte-for-byte the original construction (codes in
{0,1} ⇒ L1 == Hamming), pinned by one_bit_build_bits_matches_legacy_build.
Bytes/node scales linearly: 128-d → 16/32/64 B for b=1/2/4.

- hnsw_quantized.rs: QuantizedHnswIndex::build_bits(...,bits,...), bits()/
  bytes_per_node() accessors, code-L1 greedy+beam traversal. build(...) kept as
  the b=1 backward-compatible entry point. +4 tests (multi-bit recall regression,
  bits clamp, bytes/node, legacy parity).
- ann_measure.rs: build_indices_bits / build_quant_bits / run_scaling_study +
  best_float_op / best_quant_op; scaling_report (#[ignore], --release) and a
  CI-safe scaling_study_small_is_consistent.
- ann_bench.rs: 2-bit and 4-bit quant criterion benches over the shared graph.

ruvector lib 151 → 156 passed, 0 failed, 1 ignored (scaling_report).

Co-Authored-By: claude-flow <ruv@ruv.net>

* docs(adr-261): record M2 multi-bit scaling study — measured, no crossover (refutes M1 prediction)

Multi-bit (b∈{1,2,4}) quantized HNSW traversal + N∈{10k,100k,250k} scaling study,
measured on this box. No crossover at any (N,b): at 10k more bits help (ratio
0.19→0.48×, b≥2 reaches 0.90 recall) but quant stays slower than float HNSW at
equal recall; at 100k/250k quant recall collapses (b=4: 1.0→0.788→0.624, never
≥0.90) while float holds ≥0.92. The predicted large-N crossover moved the wrong
way. Published negative with the mechanism explained. ADR-261 §11.

Co-Authored-By: claude-flow <ruv@ruv.net>

---------

Co-authored-by: ruv <ruvnet@gmail.com>

2026-06-14 10:31:00 -04:00

rUv

f756a8af49

feat(ADR-261): ruvector HNSW graph-ANN (25x measured vs linear) + honest SymphonyQG-direction refutation (#1063 )

* feat(ruvector): real float HNSW + SymphonyQG-style quantized-traversal index (ADR-261)

Adds the graph-ANN index the ruvector retrieval path was missing (ADR-156
§5 #1 noted there was no HNSW baseline to measure SymphonyQG against).

- hnsw.rs: correct float HNSW (Malkov & Yashunin) — multi-layer NSW graph,
  ef_construction/ef_search, Algorithm-4 neighbour selection, seeded-
  deterministic level assignment (SplitMix64, reused from rotation.rs),
  L2 + cosine, brute-force ground truth, full degenerate-case guards.
  recall@10 correctness gate >=0.95 vs brute force (L2 + cosine).
- hnsw_quantized.rs: SymphonyQG-style variant — same graph, traversal scored
  by cheap 1-bit Hamming over the RaBitQ Pass-2 rotated sign code, final
  exact-float rerank.
- ann_measure.rs: shared deterministic planted-cluster fixture + recall/QPS
  measurement (ann_bench_report is the ADR source of truth).

Fixes an index-out-of-bounds bug the recall gate caught: insert wired
bidirectional edges before pushing the node's own link row. +20 tests,
ruvector lib 131->151, 0 failed.

Co-Authored-By: claude-flow <ruv@ruv.net>

* bench(ruvector): criterion ann_bench for HNSW vs quantized vs linear (ADR-261)

Times the same shared ann_measure fixture/indices through criterion so the
bench and the report test can never measure different graphs.

Co-Authored-By: claude-flow <ruv@ruv.net>

* docs(adr-261): graph-ANN index ADR with MEASURED HNSW vs quantized verdict

ADR-261 (Accepted): float HNSW ~25x QPS over linear scan at recall >=0.99
(the baseline ADR-156 said was missing). Honest negative: the 1-bit
quantized traversal is too coarse to beat float HNSW at equal recall at
N=10k (best recall 0.738, no >=0.90 equal-recall point) — the SymphonyQG
3.5-17x is NOT reproduced by our 1-bit construction; expected crossover at
large N + a multi-bit code. Caveat: our HNSW + our quant, not SymphonyQG's
system — direction tested, not a 1:1 reproduction.

ADR-156 §5 #1 + §8 backlog: CLAIMED -> MEASURED-direction-tested.
CHANGELOG [Unreleased] entry.

Co-Authored-By: claude-flow <ruv@ruv.net>

2026-06-14 02:33:32 -04:00

2 Commits