feat(embed): static (model2vec) embeddings + extra-languages gating — v0.19.0#7

Merged

anvanster merged 21 commits into

mainfrom

feat/static-embeddings

Jul 1, 2026

Commits on Jun 26, 2026

feat(embed): gated split-identifier words in embed text + distillation plan

anvanster
and
claude
committed
refactor(embed): Embedder trait — swappable backend behind VectorEngine

anvanster
and
claude
committed
feat(embed): StaticEmbedding backend — model2vec-format lookup, no ONNX

anvanster
and
claude
committed
test(embed): throughput example — static vs ONNX
anvanster
committed
docs(embed): record progress + 103x release throughput in the plan

anvanster
and
claude
committed
test(embed): retrieval quality micro-eval — static floor vs BGE

anvanster
and
claude
committed
feat(embed): F16 + SIF-weighted static loading + Jina-Code distill script

anvanster
and
claude
committed
docs(embed): record jina-code-static A/B — 70x faster, micro-eval saturated

anvanster
and
claude
committed
test(embed): real 965-way retrieval eval — static ~65% of BGE (pure semantic)

anvanster
and
claude
committed
test(embed): complete static lever sweep — gap is the contextualization ceiling

anvanster
and
claude
committed
test(embed): hybrid (BM25+semantic) eval — static is ~90% of BGE end-to-end

anvanster
and
claude
committed

Commits on Jun 27, 2026

test(embed): embed_throughput can load a real corpus (CODEGRAPH_THROUGHPUT_CORPUS)

anvanster
and
claude
committed

Commits on Jun 29, 2026

Commits on Jun 30, 2026