vh
2b40805b84
Prime (via albok-dev): understand why h2o failed and re-fit the judge. One vLLM on GPU 3 behind two fronts: :8034 with the shim's yes/no floor at 0.801 exactly as the cutover ran (same judge stamp), :8035 with SHIM_NOUL_FLOOR=0 and a distinct stamp (h2o-lightning-4b+noul-floor-0). Both refuse image URLs; the vLLM --allowed-media-domains backstop is now verified (400, no fetch).
h2o-eval (TEMPORARY) — fv-ml1 GPU 3, :8034
UP again since 2026-10-12 1428 PT, round 2 (Prime 1420, via albok-dev): "I'd like to understand why h2o failed, so stand it up on gpu3 and refit it and let's get an understanding." albok-dev runs the investigation and the judge re-fit (thread
01M4S026GWS354WZGXZ5E34F31). No load tests or other traffic while their runs are in. Round 1 (1300–1319) was the pre-cutover validation. Take it down withdocker compose downwhen albok-dev is done.
port front floor judge stamp ( /healthmodel):8034h2o-eval-frontON, 0.801, as the live cutover ran h2o-lightning-4b@b43f759b…, identical to the cutover's:8035h2o-eval-front-nofloorOFF ( SHIM_NOUL_FLOOR=0)h2o-lightning-4b+noul-floor-0@b43f759b…Both fronts sit over ONE vLLM (14,420 MiB, KV 135,070 tokens, util-sized because GPU 3 is otherwise empty). Each
/healthalso reportsbackend.noul_floor {value, source}.Checked at 1428:
/decide/shared200 on both, with the cutover's smoke values reproduced (adult 0.052, personal 0.876).- An https image URL is refused on both: 422 "not accepted here".
- The vLLM backstop answered for the first time. An https image URL posted straight to vLLM got 400 "The URL must be from one of the allowed domains: ['media.invalid']", with no fetch.
h2o-lightning-4b @ b43f759b behind services/h2o-decision-front (copied into conf/front/), for albok-dev's
re-validation of Nemi's sensitivity judge before the GPU 1 cutover (Prime 2026-10-12 via brokkr). Remove right after
the cutover: GPU 3 is for transient work only (Scriberr and MIA use it).
- Weights:
/tank/aimodels/h2o-lightning-4b-b43f759b(copied from the bench kit; model.safetensors sha2562d51561f221e1549d91bd1c38b3512a1435dd32950d454a9f040f1d4d11497b6, tokenizer.json5f9e4d49…, shim git oidfee7807e…, serve_config5e6fef1c…: all equal to Hugging Face at b43f759b). - vLLM exactly as benched:
--max-model-len 40960 --max-num-seqs 16 --gpu-memory-utilization 0.1506, vision limits as the vendor's serve.sh; healthy ~3 min afterup(graph capture), ~14.4 GB on GPU 3. .envon the host:FRONT_TOKEN= intern-decision's token (= vaultalbok/judge-token).- Remove:
cd /opt/docker/compose/h2o-eval && docker compose down(the weights stay for the GPU 1 seat).