Commit graph

140 commits

Author SHA1 Message Date
trvon
e2332a8440
optimizations: throughput optimizations, bug fixes, docs updates and benchmark cleanup 2026-04-26 21:43:00 -04:00
trvon
36152546b0
fix: simeon bump and double lock addressed in search 2026-04-25 16:10:17 -04:00
trvon
84015b39b7
build(ci,tests): cleaning build system and updates related to ci test failures and compat 2026-04-25 14:48:44 -04:00
trvon
0b61c0818a
docs: clean up docs and ci fixes for tests 2026-04-25 00:08:18 -04:00
trvon
1fcb7cf6df
feat: expanding topology instrumentation to expose quality of clusters, cli ux improvements and expanded modifiers 2026-04-23 21:28:35 -04:00
trvon
cc49704791
feat: bench harness data_dir reuse + richer telemetry, doc updates
Topology ablation bench: read YAMS_BENCH_DAEMON_DATA_DIR so multiple
daemon spawns can share a stable data_dir. Lets the topology tuner's
persisted MAB state carry over between bench reps; without this, every
fresh DaemonHarness spawn cold-starts UCB1 and picks the alphabetically-
first arm. Pairs with the TopologyTuner persistence (already on HEAD).

Bench also gains per-surface vector-leg counters (vec_unique_doc_counts,
topology_weak_routing_applied, vector_tier_skip_reason, embedding_status,
vector_db_dim/query_embedding_dim) and a modeOf reducer for categorical
metrics — Phase F1-era richer attribution carried forward.

Docs: AGENTS.md / daemon_architecture / retrieval_precision_optimization /
PROMPT-eng-codex updates from the parallel sessions.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-04-23 20:26:26 -04:00
trvon
d8cb0764d8
feat: address test drift, removing hnsw remants for simeon backend implementation, continued refactor of vector backend 2026-04-22 15:03:36 -04:00
trvon
4617bf4d8f
feat: further removing hnsw references and centralizing on simeon pq implementation 2026-04-22 01:10:57 -04:00
trvon
d669c009d0
feat: removing experiments and unifying compression in embeddings storage. 2026-04-21 18:13:37 -04:00
trvon
bdcff5b636
fix: HNSW and topology rebuilds are now batched in repair. Cleaning up from previous migration work 2026-04-21 13:58:10 -04:00
trvon
6fa8b037b0
feat: removing other graph clustering algorithms for hdbscan with sgc 2026-04-21 10:20:00 -04:00
trvon
c510d5574a
fix: refining communication path in daemon 2026-04-20 14:34:31 -04:00
trvon
3ac9bd68b8
feat: graph clustering improvements for topology system. Checkpoint before experiment with better clustering systems and simeon module bump 2026-04-20 11:31:40 -04:00
trvon
a01bd767ce
feat: bumping simeon, improving graph building, clustering, configuration tuning of said system and updating benchmark files. Additional fixes for daemon readiness and status return underload. 2026-04-20 09:31:45 -04:00
trvon
e4a8639ef6
feat: tuning topology system to work better with simeon via configuration 2026-04-19 20:10:46 -04:00
trvon
8c313fc023
feat: expanding topology system and search for better testing and simeon integration 2026-04-19 09:25:28 -04:00
trvon
d86eb06c3c
fix: enforce resource caps 2026-04-18 10:04:24 -04:00
trvon
d7ab4844a1
fix: small fixes for benchmarks and documentation update 2026-04-17 22:32:04 -04:00
trvon
6eb1635145
fix: improving repair check 2026-04-17 18:47:24 -04:00
trvon
e8ddb31145
fix: address ci source venv and graph repair timeouts 2026-04-17 14:13:35 -04:00
trvon
e2c8ba043b
fix: removing gtest and continued cleanup work of test suite 2026-04-14 12:33:17 -04:00
trvon
b6b6ba9384
feat: additional test clean up and expansion for benchmarks 2026-04-14 11:39:53 -04:00
trvon
a930776655
feat: cleaning up test suite 2026-04-14 10:25:54 -04:00
trvon
f4fffd73a0
feat: instrumentation improvements for benchmarks, shell completition improvements, packaging updates for completions and docs updates 2026-04-13 21:34:18 -04:00
trvon
bb0d33a436
feat: vector optimizations. Most performance gains in search felt on macos 2026-04-12 18:16:10 -04:00
trvon
ef3d925797
feat: continued corectness improvements, ANN / vectordb improvements 2026-04-12 12:53:29 -04:00
trvon
53326b624e
feat: optimizations for multiclient usage, updated benchmarks and ci correctness improvements to improve test runs collection 2026-04-10 13:39:59 -04:00
trvon
8777a27f3c
feat: build improvements, docs updated and optimizations for cli latency 2026-04-09 22:43:02 -04:00
trvon
ce36736e63
feat: search / retrieval optimizations for performance and accuracy 2026-04-09 21:34:43 -04:00
trvon
48a955e793
feat: CLI request handling refactor for unified load and error handling. Continued refactor and improvement of search engine with scoped tuning 2026-04-09 19:38:04 -04:00
trvon
3b5fb8fba8
fix: improvements for laoding handling in UX response 2026-04-09 16:40:04 -04:00
trvon
9a1ed082d4
feat: optimizations for data returns and harden daemon status and normalize MCP/version responses 2026-04-04 17:33:59 -04:00
trvon
e41ec38377
fix: tree-diff benchmark regression accounting and stabilize subtree fast-path measurement 2026-04-03 20:45:09 -04:00
trvon
c182221b48
feat: updating staging 2026-03-27 10:45:36 -04:00
trvon
5715ae95a6
feat: benchmark and validation testing against original system before swap 2026-03-27 09:47:38 -04:00
trvon
c00aded0ab
feat: ann improvements 2026-03-26 13:30:12 -04:00
trvon
dd93cf8fd2
fix: fix cli commands and working on retrieval optimizations 2026-03-26 11:25:28 -04:00
trvon
6874b2220e
feat: TurboQuant search and ANN compression improvements. CLi client regression fixes 2026-03-25 23:59:42 -04:00
trvon
8b393049ba
feat: TurboQuant now supports end-to-end packed-vector storage, reranking, and direct compressed ANN traversal with persisted per-coordinate calibration and validated multi-dimension performance. 2026-03-25 23:00:16 -04:00
trvon
169230645a
feat: TurboQuant integration for re-ranking, starting with benchmarks, then moving to integration 2026-03-25 17:40:35 -04:00
trvon
7609d6355b
fix: continued turboquant work. added additional api's and improved testing / perf tests 2026-03-25 17:19:57 -04:00
trvon
2922835242
fix: cli regression fixes and turboquant improvements for storage retrieval logic 2026-03-25 16:42:26 -04:00
trvon
b7d1b46cc7
turboquant: align all docs and math, add IP scoring prototype task
Benchmark extension:
- turboquant_bench.cpp: measures decoded MAE vs QJL-corrected MAE
- benchmark confirms full QJL is WORSE than decoded-only on random vectors
2026-03-25 16:42:12 -04:00
trvon
7c3b4c0f82
fix(tests): update turboquant_bench docs and fix unused-variable warning 2026-03-25 16:42:04 -04:00
trvon
147f5ca052
feat: moving changelogs to archive, implementing TurboQuant then will integrate in vector engine 2026-03-25 08:15:26 -04:00
trvon
0beaf58b18
feat: test cleanups 2026-03-23 19:43:09 -04:00
trvon
554bf47366
fix: test coverage updates and address protobuf bug in search 2026-03-20 22:15:36 -04:00
trvon
c781640dee
fix: improving daemon repsonsiveness and instrumented the daemon client for improved benchmark tracing 2026-03-20 11:02:52 -04:00
trvon
07805cc60e
fix: fixing semantic drift of readiness system, improving cli routing to reduce churn to proxy socket and improving daemon lifecycle logic 2026-03-20 10:12:42 -04:00
trvon
680cb470bf
feat: search persistence hardening, socket server improvements to reduce churn, vector persistence improvements, and test coverage improvements 2026-03-20 08:39:57 -04:00