llama-server holds 7.9GB RSS for a 1.1GB model, and its startup log goes nowhere #105

2026-08-03T22:01:56+02:00

claude commented

2026-08-03 22:01:56 +02:00

Closes Vikunja #496.

Acceptance criteria and quality gate are in TASK.md on this branch.
Review the review, not the diff — leave comments and the agent will apply them via task start 496.

Closes Vikunja #496. Acceptance criteria and quality gate are in `TASK.md` on this branch. Review the review, not the diff — leave comments and the agent will apply them via `task start 496`.

claude added 1 commit 2026-08-03 22:01:56 +02:00

memory: record the recall topic veto's real price (V-496) 9e25f18a3e

#496 asked to skip the veto when the question and the hit are in
different scripts, so an English question stops losing a Russian note.
Measured first: the fixture has no cross-language case, and en-hard-024
is an English question against an English note. Both proposed fixes are
no-ops.

What the veto actually does on the fixture, with the real embedder: it
costs en-hard-024 and buys ru-silent-029. Pass count is 22/32 either
way; false recall is 0/5 with it and 1/5 without. The two cases are one
lexical class, so no rule cheap enough for RecallAllowed separates them.

Accepts the loss and pins both sides in a test.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

kami merged commit 1c2541f7d6 into master

2026-08-03 22:04:29 +02:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: kami/Maven#105