feat(ui): memory reranker model picker with builtin + rerank catalog options - #1781
Merged
Conversation
…options
memory.embedding.rerank_model was a free-text field even though only two
kinds of value make sense: the builtin default the schema ships
(builtin.jina-reranker-v1-tiny-en-q8) and a model the rerank slot
actually serves. A typo'd id doesn't error — Hal0Reranker fail-softs to
fused vector ordering on every recall, silently.
Render it as a dropdown instead: the builtin (labelled as the default)
plus every capability-tagged rerank model from the embed.rerank catalog,
annotated with its gpu/npu backends. A saved id outside both stays
pickable as "(saved)" so the operator can see and move off it. AdvRow's
options override now accepts {value, label} entries alongside plain
strings to carry the annotations.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
thinmintdev
enabled auto-merge (squash)
August 9, 2026 19:24
thinmintdev
added a commit
that referenced
this pull request
Aug 9, 2026
Cut the fourth 1.0 release candidate on the current main tip (295181b), so it includes every PR merged for this candidate — the privileged-seam hardening cluster (#1738/#1740/#1750/#1759, #1585) plus the UI/bench/dashboard polish (#1764/#1766/#1768/#1770/#1781/#1782 and siblings). Folds [Unreleased] into a 1.0.0-rc.4 section (Added / Security / Fixed + the preview-channel Audience / Supported upgrades / Known issues / Operator migrations / Rollback subsections), bumps the version across pyproject/ui/manifest/uv.lock, and refreshes toolbox digests (7 updated, 0 null). Operator migrations flagged: file:// HAL0_RELEASES_URL moves from api.env to root-owned update.conf (#1750); deprecated out-of-list extra_args podman flags now refused at the seam (#1759).
thinmintdev
added a commit
that referenced
this pull request
Aug 9, 2026
Cut the fourth 1.0 release candidate on the current main tip (295181b), so it includes every PR merged for this candidate — the privileged-seam hardening cluster (#1738/#1740/#1750/#1759, #1585) plus the UI/bench/dashboard polish (#1764/#1766/#1768/#1770/#1781/#1782 and siblings). Folds [Unreleased] into a 1.0.0-rc.4 section (Added / Security / Fixed + the preview-channel Audience / Supported upgrades / Known issues / Operator migrations / Rollback subsections), bumps the version across pyproject/ui/manifest/uv.lock, and refreshes toolbox digests (7 updated, 0 null). Operator migrations flagged: file:// HAL0_RELEASES_URL moves from api.env to root-owned update.conf (#1750); deprecated out-of-list extra_args podman flags now refused at the seam (#1759).
thinmintdev
added a commit
that referenced
this pull request
Aug 9, 2026
Cut the fourth 1.0 release candidate on the current main tip (295181b), so it includes every PR merged for this candidate — the privileged-seam hardening cluster (#1738/#1740/#1750/#1759, #1585) plus the UI/bench/dashboard polish (#1764/#1766/#1768/#1770/#1781/#1782 and siblings). Folds [Unreleased] into a 1.0.0-rc.4 section (Added / Security / Fixed + the preview-channel Audience / Supported upgrades / Known issues / Operator migrations / Rollback subsections), bumps the version across pyproject/ui/manifest/uv.lock, and refreshes toolbox digests (7 updated, 0 null). Operator migrations flagged: file:// HAL0_RELEASES_URL moves from api.env to root-owned update.conf (#1750); deprecated out-of-list extra_args podman flags now refused at the seam (#1759).
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What changed
Settings → Memory → Reranker:
embedding.rerank_modelis now a dropdown instead of free text:catalogs.embed.rerankcatalog (GET /api/capabilities), annotated with the backends it runs on (e.g.bge-reranker-v2-m3-q4_k_m · gpu-vulkan / gpu-rocm / cpu; NPU-backed rows annotatenpuautomatically).<id> (saved)so the operator can see and move off a stale value.Free text is deliberately gone for this key: a typo'd id never errors —
Hal0Rerankerfail-softs to fused vector ordering on every recall, silently.AdvRow's per-keyoptionsoverride now accepts{value, label}entries alongside plain strings (backward compatible) to carry the backend annotations.Verification
npm run lint,npm run test:unit(121 passed), newmemory-rerank-model-picker-v3.spec.tse2e (2 passed).🤖 Generated with Claude Code