noonghunnaandClaude Opus 4.8 90c67dcc9f gemma-4-12b: promote the two vLLM MTP composes to ⚠️ Production w/ caveats ...

vllm/gemma-12b-dual-bf16-mtp + vllm/gemma-12b-single-int8-mtp → status="caveats"
(🧪 → ⚠️). Both cleared the gates:
  - dual-bf16-mtp: rebench-full (verify-full + bench + verify-stress + 8-pack
    94/150 + soak PASS), 256K NIAH overlay-free.
  - single-int8-mtp: bench + 256K NIAH + 8-pack 105/150 + soak PASS (fresh 20x5:
    0/100 silent-empty, 0 MiB growth, 95.1% retention).

CAVEAT (both): the gemma4-unified image is an ephemeral arch-preview tag (0.1.dev)
— pin a digest; promotes to ✅ Production when gemma4_unified ships in a STABLE
vLLM release.

Withheld at 🧪 (deliberate): beellama/gemma-12b-single-q8kxl + llamacpp/...-q8kxl
— no MTP/spec-dec for Gemma-4 on llama.cpp yet (blocked on llama.cpp#23398).

Header Status+Caveats, registry status, and BENCHMARKS markers all flipped
together; test-compose-status-drift + full guard suite 41/41 green.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>

2026-06-04 20:02:44 +00:00

..

fixtures

Revive agentic context-depth benchmark

2026-05-29 18:11:41 +05:00

lib

gemma-4-12b: promote the two vLLM MTP composes to ⚠️ Production w/ caveats

2026-06-04 20:02:44 +00:00

maintenance

fix(tools): resolve profile image pins in audit

2026-05-14 21:41:06 +00:00

tests

gemma-4-12b: normalize gemma-12b slug naming + prune the no-MTP bases

2026-06-04 18:56:59 +00:00

beellama-pin-bump.sh

feat(beellama): release v0.3.0 Q8_K_XL dual composes (experimental) + centralize image pin (#296)

2026-06-02 16:35:48 +05:00

bench-agentic.sh

bench-agentic: anchor TTFT-growth to first warm turn (exclude cold-start)

2026-05-29 18:11:41 +05:00

bench.sh

feat(eval): expose request-level thinking toggles (#196)

2026-05-23 08:24:58 +05:00

detect_nvlink.sh

fix(nvlink): add NVLINK_MODE=pcie_p2p for PCIe P2P without NVLink (#290) (#291)

2026-06-02 02:22:39 +05:00

diagnose-estate.sh

feat(launch): add estate planner orchestration

2026-05-14 17:04:38 +00:00

diagnose-profile.sh

feat(scripts): add diagnose-profile triage

2026-05-14 22:07:00 +00:00

generate-compose.sh

refactor(gemma): rename dual slugs + ladder gemma-bf16-mtp to 131K (#286)

2026-06-01 01:20:46 +05:00

gpu-mode.sh

gpu-mode.sh: align with dual prune — drop dead 27b modes, gemma -> int8 default (#280)

2026-05-31 18:09:36 +05:00

health.sh

Add scripts/health.sh — operational health check for running server

2026-04-28 22:14:04 +00:00

launch.sh

feat(beellama): release v0.3.0 Q8_K_XL dual composes (experimental) + centralize image pin (#296)

2026-06-02 16:35:48 +05:00

power-cap-sweep.sh

Operational robustness (#281): orphan-safe switch.sh · reboot-surviving vLLM · multi-GPU power sweep (#285)

2026-05-31 22:13:52 +05:00

preflight.sh

preflight: resolve ${VAR:-default} model paths in compose-deps check

2026-06-04 19:39:10 +00:00

pull.sh

feat(pull): v0.8.2 STEP V2 — surface pointer + --submit-last/--submit (gh + gh-less, consented, F5 reuse)

2026-05-18 17:37:41 +00:00

quality-test.sh

docs(quality-test): document failure-reason reading; fix drifted Failure-breakdown sample

2026-05-29 01:40:37 +00:00

rebench-full.sh

rebench: add fail-fast verify-full preflight + fix runtime skip set (#306)

2026-06-03 21:00:16 +05:00

rebench-report.py

fix(rebench-report): surface ceiling VRAM margin in verify-stress section (#184)

2026-05-24 20:05:52 +00:00

rebench-runtime.sh

rebench: add fail-fast verify-full preflight + fix runtime skip set (#306)

2026-06-03 21:00:16 +05:00

report.sh

fix(report): detect beellama- containers in report.sh auto-detection

2026-06-01 03:32:03 +00:00

setup.sh

refactor(gemma): rename dual slugs + ladder gemma-bf16-mtp to 131K (#286)

2026-06-01 01:20:46 +05:00

soak-helper.py

docs(soak-test): clarify PASS verdict semantics — closes #140

2026-05-15 22:18:09 +05:00

soak-test.sh

soak-test.sh: add qwen3.6-35b-a3b to container auto-detect glob (#244)

2026-05-28 16:31:15 +05:00

submit-bench.sh

fix(scripts): make submit-bench issue-first

2026-05-13 18:40:43 +00:00

switch.sh

Repoint stale single-card launcher hints off deprecated vLLM composes

2026-06-03 02:28:51 +00:00

update.sh

feat: detect repo drift in preflight + add scripts/update.sh

2026-05-03 12:17:04 +00:00

verify-full.sh

Fix verify-full to accept reasoning_content

2026-05-22 19:15:26 +00:00

verify-stress.sh

fix(verify-stress): three live-caught bugs in ceiling ladder (#199)

2026-05-23 04:42:18 +00:00

verify.sh

fix(verify): drop tail buffer on Genesis check 2 anchor (refines 95b0905)

2026-05-02 11:41:12 +00:00