altair823-org/kebab
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 40 Wiki Activity
1,125 Commits 7 Branches 40 Tags
403e162ac0153555dbdfe881379d1addf6766fa1
Commit Graph

7 Commits

Author SHA1 Message Date
altair823
e7b58017fd docs(config): v3 재편 도그푸딩 evidence + release notes 
도그푸딩(release 빌드): 사용자 실제 v2 config 변환(값·주석 보존·멱등) +
재색인 0 실증(v2 자동변환·v3 디스크 양 경로 unchanged). v0.28.0 release notes
draft(변경/trade-off/mitigation/upgrade 4단락).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-06-04 13:12:01 +00:00
altair823
d85d7348a5 docs(embed-candle): 도그푸딩 + A1 반증 + MKL 부정결과 증거 기록 
- HOTFIXES + release-notes: candle 전체 도그푸딩 997 docs/23,151 chunks/에러 0 (9.5h)
- A1(taskset -c 0-3) 실서버 반증: 4코어 제한에도 onnxruntime segfault → candle 만이 실 해법
- MKL 가속 부정 결과: 코어 더 쓰나 38~50% 느림 → 미채택, 순수-Rust 유지
- 패리티 2.01e-7 재확인, 성능 트레이드오프 명시

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
 2026-06-02 09:08:12 +00:00
altair823
6ec4e6809f fix(embed-candle): address round-1 review 
- commit track-spec + meta-spec/plan into branch (HIGH: dangling `amends:` ref)
- inline parity evidence (cosine 1.0, max_abs_diff 2.01e-7) into HOTFIXES +
  release notes; drop refs to deleted IMPL_REPORT/SPIKE_REPORT (MEDIUM)
- model guard: reject non-e5-large `model` before the 2GB download so
  model_id() can't mislabel vectors (MEDIUM) + unit test
- parity test now covers BOTH query: and passage: prefixes (MEDIUM)
- guard encodings.first() index; document zero-attention/pooling invariant;
  clarify embed_batch prefixing doc (LOW)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
 2026-06-01 16:54:20 +00:00
altair823
8f7b6ee538 feat(embed): candle 임베딩 provider (NUMA-안전, opt-in) + v0.22.0 
duo-socket NUMA 서버에서 fastembed(onnxruntime)가 intra-op 스레드를 48개로
하드코딩해 NUMA 힙 손상 → double-free 로 ingest 가 죽는 문제를 회피하기 위해,
같은 multilingual-e5-large 모델을 순수 Rust(candle)로 돌리는 opt-in 임베딩
provider 를 추가한다.

- 신규 crate kebab-embed-candle: CandleEmbedder (kebab_core::Embedder).
  hf-hub safetensors → XLMRobertaModel forward → mask mean-pool → L2 → e5
  prefix. candle 의존성 트리를 이 crate 에 격리 (core/config 외 kebab-* 의존 0).
- 스레드 캡: [models.embedding].num_threads + env KEBAB_EMBED_THREADS →
  글로벌 rayon 풀 1회 캡 (NUMA-안전 레버).
- kebab-app::embedder() 가 provider 분기 (fastembed/onnx/"" → 기존 경로 불변,
  candle → CandleEmbedder, 미지값 → 에러).
- Phase 0 스파이크 crate 제거 (production 흡수).
- 버전 0.21.1 → 0.22.0 (신규 config surface, pre-1.0 minor bump).

패리티: cosine_min=1.000000, max abs diff=2.01e-7 (< 1e-5) → embedding_version
유지, 재색인 0. fastembed default 동작/벡터 불변. wire schema 변경 없음.

검증(파일+exit code): clippy -D warnings EXIT=0(warning 0), test EXIT=0
(candle unit 5 + thread_cap rayon=4 + config 68), parity #[ignore] EXIT=0.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
 2026-06-01 14:52:25 +00:00
altair823
166b1404e4 docs(release-notes): correct refusal判정 mechanism + O-2 phrasing 
leader review of writer draft: refusal 판정은 citation marker(`[#번호]`)
유무 기반이며 `<REFUSE>` 특수 마커가 아님. O-2 문구 예시도 실제 rag-v3
규칙으로 정정.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
 2026-05-29 04:58:08 +00:00
altair823
4afcaf96d2 docs(release-notes): add v0.20.2-draft (rag-v3 응답언어 + 검색 품질 eval 인프라) 
v0.20.2 릴리즈 노트 초안 작성. 사용자 영향 4단락 구조로 각 finding 기술.

- Finding #1/O-2: rag-v3 응답언어 자동 매칭 + refusal 언어중립화
- Finding #2: bulk search input schema 확정 (15필드)
- Finding #3: list docs human-readable path 보강
- Finding #7: index_version 두 곳 구분 (vector vs FTS5)
- eval --config facade + 검색 품질 baseline (hybrid hit@3=1.0 / MRR=0.833)
- Finding #4/#5/#6/#8: docs/schema 정비
- version cascade 주의 (rag-v3 → eval compare)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
 2026-05-29 04:54:19 +00:00
altair823
028d9ad4ea docs(release): v0.20.1 release notes draft + spec/plan dogfood cross-link 
#1 (사용자 요청): release notes draft 작성 + spec/plan 의 dogfood
evidence cross-link 보강.

docs/release-notes/v0.20.1-draft.md (신규):
- 4 단락 본문 (한국어 2자 query 지원 + 영어 substring 회귀 + V007→V009
  자동 backfill + ingest 성능 영향).
- Migration cascade table (lexical_index_version, corpus_revision,
  wire schema shape preservation).
- API + dependency 변경 (lindera v3, lindera-ko-dic v3, retired
  short_query_hint helper, 새 facade APIs).
- Breaking changes 명시 (영어 substring 회귀, 첫 부팅 latency, DB/
  binary 크기 증가).
- Upgrade 절차 + Known limitation + 14 dogfood scenario reference.

spec Appendix B (segmentation evidence):
- "Empirical verification (2026-05-28 dogfood — post-merge update)"
  subsection 신규. prior-knowledge 가정 vs 실측 결과 table. Scenario
  1-4 모두 verified 표시. ko-dic 의 '서울특별시' → '[서울, 특별시]'
  분해 증거 명시.

plan Changelog:
- post-implementation entry: 22 commit on branch, S3 blockers, S7
  cascade, S11 sanity regression updates, opus PR review 4 finding
  fixes.
- dogfood evidence entry: 14 scenario verify pass, ko-dic 분해
  evidence, HOTFIXES + spec Appendix B cross-link.

Spec: …spec…md Appendix B
Plan: …plan…md (post-implementation + dogfood evidence Changelog)
Release notes: docs/release-notes/v0.20.1-draft.md
 2026-05-28 13:34:33 +00:00