kami/Maven

Routing heads step 3: train the multi-head model on the warm-started body #202

Closed

claude wants to merge 13 commits from task/661-routing-heads-step-3-train-the-multi-hea into task/660-router-prompt-destination

Author	SHA1	Message	Date
claude	9a333b23d7	Merge master after 199-201 landed (V-661)	2026-08-08 21:05:27 +04:00
kami	663b5c47b9	Merge pull request 'The router prompt has no destination, so the model arm of V-655 names nothing' (#201 ) from task/660-router-prompt-destination into master	2026-08-08 19:03:28 +02:00
kami	45c521e1a6	Merge pull request 'Destination fixture: score Decision.Source, not just the intent' (#200 ) from task/659-destination-fixture into master	2026-08-08 19:03:24 +02:00
kami	e34669a52e	Merge pull request 'Query source is a routing decision made outside the router' (#199 ) from task/655-query-source-is-a-routing-decision-made into master	2026-08-08 19:03:06 +02:00
claude	d434f83c2c	The personal boundary is a guesser, so say so (V-655) CLAUDE.md said naming SourceWorld leaves his notes, his facts and the personal boundary running first. The first two are true and the third is not. The boundary is marked guesses: true, so queryWalk drops it whenever the named destination is not recall. That is deliberate and tested. It is what stops the boundary answering 'кто такой Линус Торвальдс?' with 'не нашла у тебя такой записи'. But it means a destination a model wrote can take the boundary off a turn about him, and the doc claimed the opposite. Flagged as the owner's call rather than changed. Only the utterance leaves the box either way.	2026-08-08 21:02:41 +04:00
claude	c310115fd2	Record the clarify head and the confidence it replaces (V-661)	2026-08-08 20:59:44 +04:00
claude	3024f76e5f	A fourth head asks instead of guessing (V-661) Clarify is not a value of intent. It is a second question over the same pooled vector: can Maven act on this at all. The eight want_clarify fixture cases sat outside every number the heads measured, because a softmax has no clarify class. gen_clarify.py makes the class the corpus lacks. Every existing row was generated FOR an intent, so every one is answerable. The router-prompt agreement filter cannot work here, because routeGrammar has no clarify value and a generated line always agrees with itself. A judge replaces it. The first judge called 24 of 40 answerable rows underspecified. It judged against a generic assistant, one that asks where about lunch. Restating Maven's contract took that to 16 of 60, with all eight fixture cases caught. Three seeds: 7.0 of 8 caught, 2.3 false of 88. The cascade today misses 1 and produces 2. Confidence separates too, 0.851 right against 0.604 wrong.	2026-08-08 20:59:21 +04:00
claude	6bc71553ab	Say that a transport error is not a wrong answer (V-661)	2026-08-08 20:47:32 +04:00
claude	ed1730431c	Distil a slot head and record it beside the other two (V-661) BIO tags had no Maven-domain corpus, which was true of found corpora and false of made ones. A GBNF closed over Maven's five slots plus a substring check gives 2178 spans out of gemma-4-12b at no second call. Three heads over one forward pass: intent 92.8%, destination 82.8%, slot span F1 72.4% over three seeds. The slot head is free. Epoch selection reads the intent dev slice, so it stops the slot head about 4 points early. Recorded rather than fixed.	2026-08-08 20:27:58 +04:00
claude	01e80fce4a	Record a fortnight of usage as a re-runnable baseline (V-661) The 2026-08-07 week of usage was typed by hand and cannot be replayed, so it measured a build and not a change. scripts/usage-run.py drives the same reach from a turns file, which makes the next run a diff. Baseline is master at `beb093a`: 140 turns, p50 1.6s, zero errors. Three defects to move. A parked reminder clarify contaminates 19 later turns and survives a day boundary. Query sources that guess claim six turns they cannot answer, which is the class V-655 removes. And one question was read as a capture. Also records the slot head: gemma distils 2178 spans, three heads score intent 92.8%, destination 82.8%, slot span F1 72.4% over three seeds.	2026-08-08 20:27:27 +04:00
claude	e69f1bd0cf	Fix the floor corpus and re-measure the destination head (V-661) The first 120 floor rows carried one sentence shape, because the generator varies a topic and ambiguity is not a topic. Rotating six shapes takes the floor 3/7 to 6/7 and the destination mean 75.8% to 80.8%. Calendar stays 3/6 at every seed. The possessive agenda rules claim those cases at stage 0 and name nothing, so no label reaches the head. Also corrects the floor-case count in three files: five of the seven are homelab, not six.	2026-08-08 20:10:14 +04:00
claude	f55bedee2e	Train the destination head and beat the teacher (V-661) Step 3 of the routing-heads plan. Intent and destination share one masked mean pool on e5-small. Destination scores 26/33 against 12/33 for the classifier cascade and 24/33 for the cascade with gemma-4-12b, which is the teacher these labels were distilled from. Recall goes 0/15 to 15/15. Two heads, not four, and both cuts are label problems rather than GPU time. Mood describes her own reply state and no dataset maps onto it. BIO slot tags have no Maven-domain corpus. The MASSIVE warm-start from step 2 is worth nothing here either. Stock ties it on intent and leads by a third of a case on destination. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013ptwopxyo3Z2kwFckHkLvN	2026-08-08 19:38:23 +04:00
claude	e470435cf1	Dump the router prompt where the labeler can read it (V-661) The training workspace labels with routeSystem and routeGrammar, and it held its own copies. V-660 changed both. A retyped prompt drifts silently, which is the problem llm/check_prompt_parity.py exists for on the other side. Inert unless MAVEN_DUMP_PROMPT names a directory. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013ptwopxyo3Z2kwFckHkLvN	2026-08-08 18:44:47 +04:00