Point Maven at the workstation model: a workstation block, and routing plus replies through llm.Pair #97

2026-08-02T20:42:52+02:00

kami commented

2026-08-02 20:42:52 +02:00

Closes part of Vikunja #485: the silent half of the degradation rule.

workstation config block, dropped when it names no address; health defaults to mavgpud's /health.
modelSeam builds an llm.Pair and the router and replier complete through it.
deploy/mavend.json points at bugmachine.

Not in this PR: the naming half (world questions through PhraseQuery), tick/capture/mail, and the fixture measurement. Follow-up task to come.

Closes part of Vikunja #485: the silent half of the degradation rule. - `workstation` config block, dropped when it names no address; health defaults to mavgpud's /health. - `modelSeam` builds an `llm.Pair` and the router and replier complete through it. - `deploy/mavend.json` points at bugmachine. Not in this PR: the naming half (world questions through PhraseQuery), tick/capture/mail, and the fixture measurement. Follow-up task to come.

kami added 5 commits 2026-08-03 10:12:23 +02:00

config: a workstation block, dropped when it names no address (V-485) edeef19ff0

Health defaults to the supervisor's /health rather than llama-server's,
because mavgpud is what answers 503 while the card is held.

mavend: route and reply through the workstation when its card is free (V-485) 92d5fd580c

modelSeam builds an llm.Pair when a workstation is configured and hands it to
the router and the replier. Both are the silent half of the degradation rule:
the big model is only better there, and he is never told which model answered.
No block, no probe, and the box behaves exactly as it did.

deploy, docs: point homesrv at bugmachine and say what is still unwired (V-485) 2db59d52a7

docs: measure gemma-4-12b on the workstation against the resident model (V-485) 774217199e

Both fixtures, run from homesrv across the LAN with the proxy env stripped.
Routing: 84.4% full / 93.5% intent-only at p50 329ms through the cascade, against
72.7% / 77.9% at p50 0.80-1.04s for Qwen3-1.7B. Talk: 25/27 against 20/27, with
knowledge 9/9. Nudges 15/15. Settles #485's first assumption by measurement.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NJYcaBiuny9UGSpFweQVQ1

docs: record the workstation routing numbers where the router is documented (V-485) 4fae13af75

CLAUDE.md carried only the homesrv figures, which now read as the whole story.
Also points offload.md's order at #490 for the naming half.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NJYcaBiuny9UGSpFweQVQ1

kami force-pushed task/485-run-the-big-model-on-the-workstation-wit from d0cddc01c9 to 4fae13af75

2026-08-03 10:12:23 +02:00

Compare

kami merged commit 0ceeac8df4 into master

2026-08-03 10:13:10 +02:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: kami/Maven#97