memory: erp-tune-v2 registered in the LiteLLM gateway at operator request

Operator asked for it so he can evaluate the failed tune by hand, overriding my
not-in-the-gateway recommendation. His call.

erp-tune-v1 was DELETED from the config in the same reload rather than
repointed, so the name now 400s cleanly instead of 500ing against a stopped
backend. Deleting rather than repointing is the point: repointing would resolve
a name a consumer already knows to different weights, silently.

The config entry carries the failed-gate table, the long-form truncation (9.9%)
and degeneracy (4.9%) rates, and the rp-length caveat in-line -- so someone
reading the gateway config learns what they are calling without having to find
the althing thread. Fleet verified healthy after the restart.

persistent-memory.md

@@ -115,6 +115,7 @@ _As of 2026-08-26 ~02:10 PDT — **run 2 is trained, merged, coherence-gated and
 - **🔴 GATE VERDICT: FAIL — recorded as FAIL, and the T6/T3 trade is the actual result.** brokkr's full write-up at `4973991`, `research/R47-premium-corpus-gate/run02-gate/RESULT-run02-gate.md`. 3,456 generations, zero API errors, digests verified identical so the run is not void. **Gate 1 T6-must-not-regress: 73.5 → 88.5, +15.0 ✅** (run 1 FAILED this same axis at −3.5). **Gate 2 no-task-regresses->1-item: T3 −12.0, T4 −5.5 ❌.** Gate 3 T5 control 100% ✅. Gate 4 latency 0.11s ✅. T3 reads 88 on both tuned passes — not variance. **Neither run ships; together they price what the abliteration was costing, which is a question neither could answer alone.** Should-help axis worked: diversity **+0.196 (~15x floor)**, confound-tested against attrition (matched-k=8 keeps +0.19, attrition ≈7% of the effect) and against length (story got LONGER 669→727 while rp got SHORTER 137→88, same gain both — a length artifact would have opposite signs). Memorisation none. → `persistent-memory.d/2026-08-26-erp-run2-complete-and-served.md`
 - **⚠ THE FINDING WITH LEGS BEYOND THIS TUNE: an output-stability regression VISIBLE ONLY ON LONG-FORM.** Truncated **0/384 → 38/384**, degenerate **0/384 → 19/384** — and the reasoning battery saw **zero of it on either arm across four passes**, because its answers are short. **A gate composed only of short-answer tasks would have passed this cleanly.** Playbook §4.6.3. brokkr's block-1 rp VOID (degeneracy 12/95 past the preregistered 10% budget) was the guard catching it first, recorded as VOID not reinterpreted.
 - **⚠ PIPPA'S 123-WORD PRODUCT CLIP IS IN THE MIX AND THE DPO PAIRS WOULD INHERIT IT.** Measured 08-26: pippa bot turns max **123 exactly**, 100% at-or-under, **0.00% in the 124-130 band** — a wall, not a preference (every other root crosses its p99 smoothly). The mechanism is the exposure asymmetry: **pippa is 70.3% of bot TURNS but 37.5% of bot WORDS, and length is learned per turn.** Generalises — a clipped root is over-represented in the length signal by exactly the ratio its clipping creates. ⚠ Counter-evidence: the tune landed near pippa's **median 67**, not its **cap 123**, so "learned central tendency" is better supported than "learned the limit". Closing test is the tuned arm's own rp length distribution vs pippa's and bluemoon's. Scripts `/tank/erp-tune/pippa_clip.py`, `clip_share.py`. → `docs/pfi/erp-dpo-stage-prep.md`
+- **🟢 `erp-tune-v2` IS NOW IN THE LITELLM GATEWAY — operator's explicit request, 2026-08-26**, so he can evaluate the failed tune by hand. Reachable at `10.250.50.70:4000` as `erp-tune-v2` on the shared all-agents key (HTTP 200 verified), direct at `:8098`. **`erp-tune-v1` DELETED from the config in the same reload** — it now returns a clean **400** instead of the **500** it had been throwing since its backing container stopped. Config backed up at `/opt/docker/conf/litellm/config.yaml.bak-pre-erpv2-20260826T131938Z`; fleet verified healthy after the restart (summarizer 200, 32 models). ⚠ **The config entry carries the FAILED-gate warning, the long-form truncation/degeneracy rates and the rp-length caveat in-line**, so anyone reading the gateway config learns what they are calling without finding this thread. This supersedes my earlier not-in-the-gateway recommendation — the operator overrode it and that is his call.
 - **🟢 SEATS: `erp-tune-v2` UP on `:8098`, base arm torn down, GPU0 has ~13 GB spare.** The tune stays served (operator asked; it is coherent — the gate verdict is a research result, not a serving fault) and stays **OUT of the LiteLLM gateway**, which now reads as clearly right rather than merely cautious: **a FAILED tune must not be one alias resolution away from a consumer who has not read the thread.**
 - **⏳ SUPERSEDED — the mid-swap state.** Base arm ran 08:54:52Z→09:03:43Z, **zero errors/truncations/empties/degenerates across 1,728 generations**. Swapped to `erp-tune-v2` on `:8098`; **image digest `sha256:4091d5593f77…` VERIFIED IDENTICAL across both arms** (brokkr's requirement — a differing digest voids the run). brokkr is now running the tuned set with `--freeze-from` the base's marker list. **The floors the tuned deltas must clear:** reasoning core **0.5 pt**, diversity overall **0.0125** (rp 0.0153, story 0.0096), story attractor **0.0000**, memorisation signal 0.00%. ⚠ **The rp family froze ZERO markers**, so its attractor metric is structurally 0.0 on both arms and cannot discriminate — rp is measured on the distance axis only; do not quote an rp attractor delta. ⚠ Base's slop attractor: **"Elias" in 92/96 bare-prompt stories** (published figure on the same family is 102/144 — ours is an independent replication at a higher rate, not a novel finding).
 - **⏳ SUPERSEDED — the mid-swap state.** brokkr **withdrew** his both-arms-concurrent requirement (his diversity battery emits the frozen marker list to a *file*, so the arms were never a live dependency; the real requirement is all of ONE arm's passes on ONE instance before the swap). So: **sequential, no fleet seats displaced, operator not woken.** Live now: `erp-eval-base` on `:8099` (`erp-base-stock`, stock bf16 via symlinks into `/tank/aimodels`). **`erp-eval-v2` is DOWN and its container REMOVED**; weights are on disk so it is a cheap restart. Swap back with `/tank/erp-tune/serve-arm.sh tuned` when he reports the base set complete. ⚠ **Both arms go through `serve-arm.sh`**, which pins the vLLM image **by digest** (`sha256:251eba5cc7c1…`) — a version change between arms is a base swap nobody would see — and mounts `/tank/aimodels` for BOTH arms even though only base needs it, because a mount that differs between arms is a difference between arms.