vh/esh-pfi-infrastructure
1
0
Fork 0
Code Issues 7 Pull Requests Actions Packages Projects Releases Wiki Activity

memory: h2o judge investigation closed — v1 note question was the cause, re-fitted in albok-nemi 0.1.18; eval seat down

Browse Source
This commit is contained in:
vh
2026-10-12 16:20:49 -07:00
parent 2b40805b84
commit 8d9fcd9ffb
2 changed files with 5 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
persistent-memory.md
+1 -1
View File
@@ -169,7 +169,7 @@ Every vLLM seat on a shared card (GPU 0, GPU 1) has byte-pinned KV (`--kv-cache-
### The decision seat `intern-decision` :8033 (Intern-Decision-4B; h2o staged)
- **LIVE: Intern-Decision-4B** (`intern-decision-serve:0.1.3`), again since 2026-10-12 1403. The only production caller is albok-nemi's sensitivity judge (`/decide/shared`, threshold 0.5). Run `scripts/intern-decision-warmup` after an IMAGE change. → `persistent-memory.d/2026-09-30-semif-replaced-by-intern-decision.md`
- **h2o-lightning-4b is STAGED** (`stacks/intern-decision/compose.h2o.yaml` + `switch-to-h2o.sh`, with a GPU 1 free-memory gate; weights `/tank/aimodels/h2o-lightning-4b-b43f759b`; front `services/h2o-decision-front`). Inline images only (`LIGHTNING_IMAGE_URLS=off` and the vLLM `--allowed-media-domains media.invalid` backstop, both verified). **It returns only after albok-dev re-fits the judge for it** (a 4.4 contract revision on an eval seat). **Round 2 eval seat UP since 1428 on GPU 3** (Prime 1420: understand why h2o failed, re-fit): `:8034` floor on (stamp = the cutover's), `:8035` floor off (`h2o-lightning-4b+noul-floor-0`), one vLLM; no load tests while albok-dev runs; down when they're done. Thread `01M4S026GWS354WZGXZ5E34F31`. ⚠ Its seat headroom under 4-way 32k load was 683 MiB.
- **h2o-lightning-4b is STAGED** (`stacks/intern-decision/compose.h2o.yaml` + `switch-to-h2o.sh`, with a GPU 1 free-memory gate; weights `/tank/aimodels/h2o-lightning-4b-b43f759b`; front `services/h2o-decision-front`). Inline images only (`LIGHTNING_IMAGE_URLS=off` and the vLLM `--allowed-media-domains media.invalid` backstop, both verified). **It returns only after albok-dev re-fits the judge for it** (a 4.4 contract revision on an eval seat). **Round 2 (GPU 3 eval seat 1428–1619, Prime 1420) found the cause:** the judge's v1 note question asked two things at once. INC answered it inconsistently and h2o answered it literally. The shim floor is ruled out (identical verdicts floor on and off). Prime ruled "text itself", and albok-nemi 0.1.18 (judge.v2, 4.4 rev 1.15) re-fitted the question. Under v2 the two seats agree (work notes: INC 0/22, h2o 1/22 at 0.52) and fleet/memory adult went 86 → 0 on INC. **The h2o cutover is now a cheap call for Prime**, with the same sequence and the same rollback as this morning. Threads `01M4S026GWS354WZGXZ5E34F31`, `01M4S6MD2GG1WS4FY5WJRP9AMD`. ⚠ Its seat headroom under 4-way 32k load was 683 MiB.
- **Open:** label ~50 real Wyrd/Cicada turns before trusting INC in production; its card makes no contamination claim.