haiku.rag/docs
Yiorgis Gozadinos 2d8b2cb9ed
Expose search.vector_nprobes
The IVF probe count was the one vector search parameter with no
setting, and it is what bounds recall on a large indexed corpus:
partitions grow with the corpus, so a fixed probe count covers less of
it, and vector_refine_factor can only re-score what the probes
returned. Defaults to 20, matching lance, so search is unchanged.

Also corrects the create-index docs, which predate the measurement:
optimize() covers new chunks, so a rebuild is about retraining
centroids rather than reaching unindexed rows.
2026-09-01 15:54:36 +03:00
..
capabilities Finish the comment pass, and escape document fields everywhere Rich renders 2026-08-28 15:34:47 +03:00
configuration Expose search.vector_nprobes 2026-09-01 15:54:36 +03:00
img
stylesheets
apps.md
benchmarks.md Document vector index state and measured cost 2026-09-01 15:19:13 +03:00
changelog.md
chat.md Update caps documentation 2026-08-13 14:08:14 +03:00
cli.md Expose search.vector_nprobes 2026-09-01 15:54:36 +03:00
custom-pipelines.md Linting, doc edits 2026-08-18 12:18:10 +02:00
development.md Delete comments that restate the line below them 2026-08-20 15:22:33 +03:00
index.md Give the docs an architecture page and one extras list 2026-08-20 15:07:06 +03:00
ingester.md Give the docs an architecture page and one extras list 2026-08-20 15:07:06 +03:00
installation.md Give the docs an architecture page and one extras list 2026-08-20 15:07:06 +03:00
mcp.md
overview.md Give the docs an architecture page and one extras list 2026-08-20 15:07:06 +03:00
python.md State what the code does, not what it replaced 2026-08-28 15:13:52 +03:00
remote-processing.md Wire providers.docling_serve.timeout through to the client 2026-08-19 13:30:45 +03:00
tools.md
tuning.md Remove the wix evaluation dataset 2026-08-14 14:46:51 +03:00
tutorial.md