Skip to content
This repository was archived by the owner on Nov 21, 2026. It is now read-only.

Docs: say why quantization="PRODUCT" is not recommended - #395

Merged
tae898 merged 2 commits into
mainfrom
pq-docs-10514
Oct 21, 2026
Merged

tae898 merged 2 commits into
mainfrom
pq-docs-10514

Conversation

@tae898

@tae898 tae898 commented Oct 21, 2026

Copy link
Copy Markdown

The docs said "PRODUCT"/PQ is not recommended but not why. This PR states the reason in docs/api/database.md (the quantization and pq_training_limit bullets), docs/guide/vectors.md and the quantization docstring of create_vector_index in core.py: current engine builds score SQL search (vectorNeighbors) candidates with full-precision vectors, so PQ gives no search speedup; the index files are larger than with None because PQ codes are written that nothing reads; and pqTrainingLimit is ignored. Docs and docstring only, no code change.

Closes #394

Checks run locally: uv run --no-project --with pre-commit pre-commit run --files on the three changed files (all applicable hooks passed; the commit hook run passed too); uv run --no-project ... mkdocs build --strict -f bindings/python/mkdocs.yml (passed, unpinned docs deps, not the pinned CI versions). The docstring-only change needs no test run; the code-block tests were not run because no code block changed.

🤖 Generated with Claude Code

tae898 and others added 2 commits October 21, 2026 18:02
Explain the reason in the database API page, the vectors guide and the
create_vector_index docstring: current engine builds score SQL search
candidates with full-precision vectors, write PQ codes that nothing
reads, and ignore pqTrainingLimit (ArcadeDB ArcadeData#10514).

Refs #394

Co-Authored-By: Claude Haiku 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Docs: say why quantization="PRODUCT" is not recommended (ArcadeDB #10514)

1 participant